SEARCH

SEARCH BY CITATION

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References

Biofilm formation is commonly described as a developmental process regulated by environmental cues. In the current study we present a mechanistic model to explain regulation of Pseudomonas fluorescens biofilm formation by the environmentally relevant signal inorganic phosphate (Pi). We show that activation of the Pho regulon, the major pathway for adaptation to phosphate limitation, results in conditional expression of a c-di-GMP phosphodiesterase referred to as RapA. Genetic analysis indicated that RapA is an inhibitor of biofilm formation and required for loss of biofilm formation in response to limiting Pi. Our results suggest that RapA lowers the level of c-di-GMP, which in turn inhibits the secretion of LapA, a large adhesion required for biofilm formation by P. fluorescens. The ability of c-di-GMP to modulate protein secretion is a novel finding and further extends the biological influence of c-di-GMP beyond that of regulating exopolysaccharide synthesis and motility. Interestingly, Pho regulon expression does not impinge on the rate of attachment to a surface but rather inhibits the transition of cells to a more stable interaction with the surface. We hypothesize that Pho regulon expression confers a surface-sensing mode on P. fluorescens and suggest this strategy may be broadly applicable to other bacteria.


Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References

The decision to change one's lifestyle is never to be taken lightly. This is as true for bacteria as it is for humans. Biofilms are one such bacterial lifestyle, and can be simply defined as a community of bacteria living attached to a surface (Davey and O'Toole, 2000; Mah et al., 2003). Current research suggests that the decision to commit to a surface attached lifestyle versus maintenance of a free swimming existence is highly regulated by environmental cues (Jackson et al., 2002; Stanley and Lazazzera, 2004; Banin et al., 2005). However, little is known about the exact identity of the signals or the mechanisms by which they are sensed and integrated with pathways for biofilm formation.

The Pho regulon constitutes the major pathway for adaptation to phosphate starvation and is defined by the set of genes whose transcription is modulated by the response regulator PhoB (Wanner, 1996). The activity of PhoB is linked to the external concentration of Pi through its interactions with the sensor histidine kinase PhoR and the Pst system. In Pi-sufficient conditions the Pst system inhibits PhoR from phosphorylating and activating PhoB. However, in Pi-limiting conditions PhoR is directed to phosphorylate PhoB, thereby promoting its interaction with Pho promoters. Typically members of the Pho regulon function in processes relevant to phosphate assimilation or metabolism; however, increasing evidence has been provided demonstrating that this system has been usurped to control broader cellular phenomena in a variety of bacteria (Gonin et al., 2000; Slater et al., 2003).

Research into the molecular mechanisms of biofilm formation by Pseudomonas aureofaciens uncovered a role for inorganic phosphate (Pi) as an important extracellular signal (Monds et al., 2001). Specifically, Pho regulon expression in response to limiting concentrations of Pi resulted in inhibition of biofilm formation. Genetic evidence was presented showing that: (i) mutations to phoB suppressed the loss of biofilm formation in response to Pi-limiting environments; and (ii) constitutive activation of the Pho regulon by mutation of the pstSCAB-phoU operon resulted in loss of biofilm formation in Pi-sufficient environments (Monds et al., 2001). This research did not provide further insight as to the stage of biofilm formation that is inhibited by Pho expression, or identify the members of the Pho regulon that are responsible for eliciting the response. In short, the signal and the upper hierarchical regulator are known, but little is known about the mechanistic basis of biofilm inhibition in response to Pi-limiting conditions.

Pseudomonas fluorescens, like P. aureofaciens, is primarily a soil-based bacterium that is often found colonizing the rhizosphere (Compeau et al., 1988; Silby and Levy, 2004). One focus of our lab has been to understand the molecular basis of biofilm formation by P. fluorescens (O'Toole and Kolter, 1998). Previous research identified the Lap system as a critical component of biofilm formation by P. fluorescens. LapA is a large adhesin that is transported to the outer membrane by an ABC transporter LapEBC (Hinsa et al., 2003). For P. fluorescens, cells initially attach to a surface by a single pole; attachment of this nature is commonly referred to as ‘reversible’ because attachment is typically unstable, with cells often returning to the bulk phase. Cells that commit to a more stable surface existence are seen to attach to the surface via their longitudinal axis. This phenomenon is referred to as ‘irreversible attachment’. LapA was shown to be an important determinant in the ability of cells to transition from reversible to irreversible attachment (Hinsa et al., 2003).

Previously we have shown that the Pho regulon is conserved in P. fluorescens Pf0-1 (Monds et al., 2006) and were interested in exploring whether Pi was a signal utilized by P. fluorescens to regulate biofilm formation. In this report we demonstrate that, similar to P. aureofaciens, limiting levels of Pi inhibit biofilm formation via activation of the Pho regulon. Furthermore we have identified and characterized Pho-regulated activities responsible for integrating this physiologically relevant environmental signal with central pathways for biofilm formation and present a mechanistic model to explain regulation of P. fluorescens biofilm formation by Pi.

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References

Pho regulon expression inhibits biofilm formation by P. fluorescens

We first assessed whether Pi concentrations modulated P. fluorescens biofilms in a Pho regulon-dependent fashion. Similar to P. aureofaciens, biofilm formation by P. fluorescens was severely reduced when grown in Pi-limiting medium (K10Tπ) compared with Pi-sufficient medium (K10T-1) (Fig. 1A and B). Furthermore, deletion of phoB, which encodes the response regulator required for Pho regulon activation, suppressed the loss of biofilm formation in Pi-limiting conditions (Fig. 1A and B). Inhibition of Pho regulon expression was confirmed by the inability of ΔphoB to express Pho-regulated alkaline phosphatases under Pi-limiting conditions (Fig. 1A). Disruption of the Pst system leads to constitutive expression of the Pho regulon irrespective of the extracellular Pi concentration, which is indicated by the expression of Pho-regulated alkaline phosphatase activities in Pi-sufficient conditions (Fig. 1A). As was the case with P. aureofaciens, the pst mutant of P. fluorescens was unable to form biofilms, even in Pi-sufficient conditions (Fig. 1A and B).

image

Figure 1. Pho regulon expression inhibits biofilm formation. A. Alkaline phosphatase activity (AP) and static biofilm phenotypes (Bfm) for strains grown in high Pi (K10T-1) and low Pi (K10Tπ) conditions. Biofilms are visualized as a ring of stained bacteria attached to the well of a microtiter dish at the air/liquid interface. The degree of blue coloration indicates the level of phosphatase activity, which is known to a surrogate marker for Pho regulon induction. B. Shear-dependent biofilm phenotypes for strains grown in high Pi (K10T-1) and low Pi (K10Tπ) conditions. Fluorescent images were acquired of attached bacteria after staining with the fluorescent dye Syto-9. C. Comparison of the initial rates of attachment for Pf0-1 (WT) and Δpst. Shown is the total number of cells that attached to a set surface area (136 μM × 102 μM) over a 5 min period when grown in Pi-sufficient conditions. D. Propensity of strains to transition from reversible to irreversible attachment. Cells showing movement around a single pole were defined as reversibly attached (REV), whereas cells lying still and flat on their longitudinal axis were defined as irreversibly attached (IRREV). Both Δpst and the lapA mutant showed defects in the transition to irreversible attachment relative to the WT.

Download figure to PowerPoint

We performed a further experiment to demonstrate that improper activation of PhoB is necessary and sufficient to explain the biofilm phenotype of the pst mutant. Deletion of ΔphoB in the Δpst background restored biofilm formation, with the pst phoB double mutant displaying identical phenotypes to that of the phoB single mutant (Fig. 1A). In regard to biofilm formation, phoB is epistatic to pst. Taken together, these results confirm that, similar to P. aureofaciens, Pho regulon expression in response to Pi-limiting environments negatively regulates biofilm formation by P. fluorescens Pf0-1.

Pho regulon expression prevents stable surface attachment

For many bacteria, commitment to a surface involves two stages: (i) transition from the planktonic phase to reversible surface attachment; and (ii) transition from a reversibly attached state to a more stable association with the surface referred to as irreversible attachment. We utilized quantitative time-lapse microscopy to determine whether Pho expression inhibits biofilm formation at either of these two stages.

First, we assessed the initial attachment rates of the wild type (WT) and pst mutant. We inoculated Pi-sufficient medium (K10T-1) with equal numbers of bacteria and acquired a series of 5 min time-lapse movies for each strain attaching to a polystyrene surface. The number of cells that interacted with the surface was recorded over the entire 5 min and used as a relative measure of a given strain's propensity for surface attachment. This analysis indicated that both WT and Δpst initiate contact with the surface to similar degrees (Fig. 1C).

Next we asked whether Pho regulon expression inhibits transition from reversible to irreversible attachment. To do this, we took a series of 1 min time-lapse movies and recorded the fate of each cell attached to the surface, with respect to reversible and irreversible attachment. Specifically, cells that were lying down on their longitudinal axis and remained static for the duration of the movie were defined as irreversibly attached. Conversely, cells that showed movement around a single pole at any time during the movie were classified as reversibly attached. For the WT 97% of cells were irreversibly attached under these conditions (Fig. 1D). In contrast, only 34% of the Δpst cells remained irreversibly attached, with 66% demonstrating reversible attachment. As a control we also included a lapA mutant, which is known to be defective in transition from reversible to irreversible attachment (Hinsa et al., 2003). In accordance with this description, only 16% of the lapA mutant cells were irreversibly attached (Fig. 1D). From this analysis it is clear that Pho regulon expression does not impinge on the propensity of a given cell to initiate interactions with a surface. However, Pho regulon expression does decrease the ability of cells to commit to that surface through the formation of more stable interactions.

Transcriptional regulation of the lap genes by PhoB

The Lap system is composed of an ABC transporter (LapEBC) that is thought to transport a large adhesin (LapA) to the outer membrane (Hinsa et al., 2003). This system is absolutely required for biofilm formation in every condition we have tested. Given that both the pst and lapA mutants have defects in the transition to irreversible attachment, we hypothesized that the activated-PhoB might repress lap gene transcription.

As a first measure, we searched for Pho boxes in the promoter regions of lapA and lapEBC, which are divergently transcribed (Fig. 2A). The Pho box constitutes the binding motif for activated PhoB and was first described in Escherichia coli (Fig. 2B). We have shown that this sequence is functionally conserved in P. fluorescens (Monds et al., 2006). Analysis of putative promoter regions led to the identification of degenerate Pho boxes for both lapA and lapEBC (Fig. 2A and B). The lapA Pho box has five mismatches to the consensus, whereas the putative lapE Pho box has four mismatches, with the direct repeats separated by a 5 bp linker region as opposed to the usual 4 bp (Fig. 2B). Interestingly, both Pho boxes appeared to overlap −35 consensus sequences, providing a simple model whereby PhoB could repress transcription at these loci by competing with RNA polymerase for binding at the promoter (Fig. 2B).

image

Figure 2. Pho regulon-dependent regulation of lap gene transcription. A. Schematic of lapA-lapEBC genomic context, including relative positions of putative Pho boxes. B. Comparison of putative lapEBC and lapA Pho boxes to the E. coli consensus and known Pho boxes for P. fluorescens Pf0-1 Pho-regulated genes, phoX and phoD. Matches to consensus are underlined and putative −35 sequences are shaded. C. Quantification of Pho regulon-dependent regulation of lap gene transcription by qRT-PCR. Relative gene expression of lapA, lapE, lapB, lapC and the positive control phoD was compared for different strains grown in Pi-sufficient conditions.

Download figure to PowerPoint

To further test this model we assessed the impact of Pho regulon expression on lapA and lapE transcription using merodiploid luciferase fusions. For this analysis we compared the WT to the pst mutant when grown in Pi-sufficient conditions. Furthermore, we also included the ΔpstΔphoB strain to allow confirmation that any impact on lap gene transcription was dependent on PhoB. This analysis showed that lapA transcription was only marginally reduced in the pst mutant (615 ± 23 RLU/OD600) relative to the WT (726 ± 12 RLU/OD600). Conversely lapE transcription was reduced by 73% in the pst mutant (500 ± 38 RLU/OD600) relative to WT (1849 ± 54 RLU/OD600). Importantly, this reduction was dependent on PhoB, as indicated by the pst phoB mutant (1748 ± 76 RLU/OD600) showing similar levels of transcription to the WT.

To further substantiate these results we used real-time quantitative RT-PCR to assess the levels of lapA and lapE transcripts in the WT, Δpst and ΔpstΔphoB strains when grown in the same conditions as were used for luciferase fusion experiments. In agreement with transcriptional fusion analysis, lapA showed similar levels of transcript for all three strains, whereas lapE transcription was reduced by ∼80% in the pst mutant relative to the WT (Fig. 2C). Once again this reduction in lapE transcription was PhoB-dependent. To extend our analysis, we also assessed the impact of Pho expression on the transcript levels of lapB and lapC, which are in a putative operon with lapE. Levels of lapB were seen to be decreased by 37% in the pst mutant relative to the WT, whereas transcript levels of lapC were reduced by 28% (Fig. 2C). In both cases, reductions in transcription were dependent on the presence of PhoB. Transcript levels of phoD served as a positive control for Pho regulon expression. As expected phoD expression was induced in Δpst relative to both WT and ΔpstΔphoB strains (Fig. 2C). In summary, both luciferase transcriptional fusions and qRT-PCR data support Pho-dependent inhibition of lapEBC transcription, but do not provide support for Pho-dependent inhibition of lapA transcription.

Reductions in lapEBC transcription are not sufficient to account for Pho-dependent reductions in biofilm formation

We sought to test whether Pho-dependent reductions in lapEBC transcription were necessary and sufficient to account for loss of biofilm formation in response to limiting Pi. To do this we replaced the right arm of both putative Pho boxes with PacI restriction sites, generating the strain lapΔbox (Fig. 3A). We have shown previously that such perturbations are sufficient to prevent PhoB-dependent activation of phoD transcription, likely due to prevention of productive PhoB binding at the promoter (Monds et al., 2006). By leaving the putative −35 sequence intact we hoped to create a situation whereby RNAP could transcribe the lapEBC genes unhindered by Pho regulon-mediated repression (Fig. 3A). If our model was correct, and if reductions in lapEBC are sufficient to account for Pho regulation of biofilm formation, we predicted that lapΔbox would show recovery of biofilm formation in low Pi conditions relative to the WT.

image

Figure 3. Functional assessment of the contribution of inhibition of lapEBC transcription to Pho-dependent inhibition of biofilm formation. A. Strategy for testing the model of direct repression of lap transcription by PhoB. We predicted that disruption of putative PhoB binding sites without also disrupting −35 sequences should prevent occlusion of RNA Pol from the promoter, relieving repression of lapE transcription due to Pho regulon expression. B. Effect of Pho Box deletions on lapE transcription. Luciferase activity is shown for merodiploid luciferase fusions in WT and lapΔbox strains when grown in both high Pi and low Pi conditions. Disruption of putative Pho boxes resulted in Pho-independent repression of lapE transcription. C. Quantification of biofilm formation by WT and lapΔbox grown in both high Pi and low Pi conditions. Despite severe reductions in lapE transcription, the lapΔbox mutant showed only a small reduction in biofilm formation when grown in high Pi conditions. Both strains showed inhibition of biofilm formation as a consequence of growth in low Pi conditions. D. Ability of lapEBC expression to rescue biofilm formation in Pi-limiting conditions. WT harbouring either the vector (pMQ71B) or the lapEBC expressing plasmid (pMQ-lapEBC) were tested for biofilm formation in both high and low Pi conditions. Expression of lapEBC was induced by addition of arabinose to 0.02% w/v. Expression of lapEBC from a heterologous promoter could not rescue inhibition of biofilm formation in low Pi.

Download figure to PowerPoint

Interestingly, disruption of the putative Pho boxes for both lapA and lapE led to general inhibition of lapE transcription, even under Pi-sufficient conditions (Fig. 3B). We assessed this by generating luciferase fusions analogous to those used previously for measurement of lapE transcription. Comparison of WT and lapΔbox fusion strains in high and low Pi conditions indicated that lapE transcription was approximately threefold lower in the lapΔbox strain when grown in high Pi conditions than in the WT grown under low Pi conditions. Based on our strategy, we had expected to see no major impact of the Pho box mutations on lapE transcription rates in Pi-sufficient conditions. This result, while providing no general support for the direct interactions of PhoB with the putative Pho boxes, did provide an alternative way to test whether reductions in lapEBC transcription could account for the Pho-dependent loss of biofilm formation. We reasoned that if reductions in lapEBC transcription were sufficient to explain the decreased biofilm formation in low Pi, the biofilm formed by the lapΔbox strain in high Pi would be equivalent to, or worse than, the reduced biofilm formed by the WT in low Pi conditions. The logical framework for this reasoning is based on our data showing that the promoter mutations in the lapΔbox strain reduce lapE expression to levels lower than that seen due to growth of the WT in low Pi. If a positive correlation does exist between lapEBC transcription and biofilm formation we should see significant decreases in biofilm formation by the lapΔbox strain. This was not what we observed. In fact the lapΔbox strain formed biofilms only slightly reduced to that of the WT in high Pi (Fig. 3C). Furthermore, lapΔbox showed similar reductions in biofilm formation in response to low Pi environments. Ultimately these experiments suggest that very little lapEBC transcript may be required to support robust biofilm formation and that Pho-mediated reductions in lapEBC transcription are not likely to be sufficient to account for Pho-mediated reductions in biofilm formation.

To further address the role of lapEBC transcriptional repression in Pho-dependent modulation of biofilm formation, we assessed the ability of lapEBC to restore biofilm formation in low Pi conditions when expressed constitutively in trans. For this experiment we cloned the entire lapEBC operon into pMQ71B, placing it under transcriptional control of the PBAD promoter rather than its native Pho-regulated promoter. This construct, referred to as pMQ-lapEBC, was demonstrated to express functional LapEBC by its ability to complement mutations in lapE, B and C when provided in trans (data not shown). With the functionality of pMQ-lapEBC confirmed, we went on to test whether pMQ-lapEBC could suppress loss of WT biofilm formation in response to Pi-limiting conditions. WT harbouring pMQ-lapEBC demonstrated reductions in biofilm formation in Pi-limiting conditions that were indistinguishable to that of the WT harbouring the empty vector, indicating that lapEBC, uncoupled from Pho regulation, is unable to suppress Pho-dependent inhibition of biofilm formation (Fig. 3D). Concordant with these results, pMQ-lapEBC was also unable to restore biofilm formation to pst mutant in Pi-sufficient conditions (data not shown). We further extended this experimental strategy by inserting the PBAD promoter (without araC) upstream of lapEBC on the chromosome and testing this strain for suppression of Pi-dependent reductions in biofilm formation. We could not detect any restoration of biofilm formation in Pi-limiting media relative to the WT (data not shown).

In summary, three independent experimental approaches indicated that Pho regulon-mediated repression of lapEBC transcription is not sufficient to account for inhibition of biofilm formation in response to Pi-limiting environments.

Bioinformatic-based search for candidate Pho biofilm genes

As an alternative strategy for identification of the downstream effectors of Pho-mediated biofilm inhibition, we took advantage of Pho box conservation in P. fluorescens to identify other potential Pho regulon genes (Monds et al., 2006). From these, we hoped to identify candidate genes with predicted functions that may be applicable to biofilm formation (e.g. polysaccharide synthetic genes, regulators). We searched the P. fluorescens genome (http://genome.ornl.gov/microbial/pflu/) for putative Pho boxes using the consensus sequence CTGTCAT NNNN CTGTCAT, as defined from E. coli (Wanner, 1996). Additional criteria included: allowance of up to three mismatches with the consensus and positioning of the motif within 200 bp of the annotated start codon as well as in the same 5′ to 3′ direction as the open reading frame (ORF). In total, we recovered 43 genes that matched these criteria (Table 1). Within this group, we recovered the known Pho genes pstS, phoD and phoB, providing validation for our approach.

Table 1.  List of genes with putative Pho boxes.
Gene no.MotifMismatchesDistance to geneGene nameGene description
Pfl0388CaGTCATATCGCTGTCAc2155Hypothetical protein
Pfl0796gTGTCAcACCCCTtTCAT341phoDTat pathway signal sequence domain protein
Pfl0799tTGgCATCCTTtTGTCAT315Hypothetical PKHD-type hydroxylase
Pfl1031CgGTtATTTTTCTGTCAT2141Hypothetical protein
Pfl1031CTGTCgTCACCCcGaCAT3180Conserved domain protein
Pfl1424gTGcCAcTGCCCTGTCAT3130Hypothetical protein
Pfl1424tcGTCATCAACCTGcCAT3152Hypothetical protein
Pfl1475CTGaCgcACACCTGTCAT354Hypothetical protein
Pfl1513CTGTCAaTAATtTGTCAg353SAM-dependent methytransferases
Pfl1678CaGTCATCATTCTGTCgT273rapAEAL domain/GGDEF domain protein
Pfl1678tcGTCATCAAACaGTCAT384rapAEAL domain/GGDEF domain protein
Pfl1686CcGaCATCAAAgTGTCAT389afuAABC-type Fe3+ transport system
Pfl1687tTGTCATCTCTCgGTCAT297phnFTranscriptional regulator, GntR family
Pfl1841tTtaCATGGACCTGTCAT360Hypothetical protein
Pfl2024aTGTCATTTTGCTGaCAc379ugdUDP-glucose dehydrogenase
Pfl2182CTGTCtTTTTCCTGTCca356oppAABC-type oligopeptide transport system
Pfl2182CTtTttTTCGCCTGTCAT3134oppAABC-type oligopeptide transport system
Pfl2264CaGTCATTCGGCTGTgAa396Predicted sigma 24 homologue
Pfl2299tTGaCAgGCTGCTGTCAT396qorQuinone oxidoreductase
Pfl2518tTtTCATGGAGtTGTCAT35cbpADnaJ-class molecular chaperone
Pfl2555CTtTCATCACAtTGTCAT2176Hypothetical protein
Pfl2854CcGTCAcCGTCCTGTCcT367glnQABC-type polar amino acid transport system
Pfl3024CTGTtATCGAGCgtTCAT362Hypothetical protein
Pfl3132CTGTCgaCCACCTGTCAg3107Hypothetical protein
Pfl3246CTGTCATGACACTGTCAT1167Hypothetical protein
Pfl3247gcGTCATCAGGCTGTCAT267gppAExopolyphosphatase
Pfl3640CTGTCAaGCCCCTGaCAg3162ppiBPeptidyl-prolyl cis-trans isomerase B
Pfl3715CgGTCATCTATtTGTCAT263Hypothetical protein
Pfl3851CTGTCATGCAACTGTCAc142Predicted phosphatase
Pfl3851tcGTCAcTGGCCTGTCAT353Hypothetical protein
Pfl3854CTGTCAgGTCGCTGcCAg356Hypothetical protein
Pfl3854tTGTaATCTCTCTGTCAg367Hypothetical protein
Pfl4318acGTCATTTTGCTGTCAc384corACation transporter
Pfl4318CTGTCAcACAAgcGTCAT373corACation transporter
Pfl4358CTGTCATTTGGCctgCAT352Hypothetical protein
Pfl4465CTGTCAcCGTCCTGTCAc2162Transcriptional regulator, AraC family
Pfl4543CTGTCATCGAACTGTggc365mhpCHydrolase, alpha/beta fold family
Pfl4543tctTCATCAAACTGTCAT376mhpCHydrolase, alpha/beta fold family
Pfl4849CcGTCAaAGCGCTGTaAT375ispH4-hydroxy-3-methylbut-2-enyl diphosphate reductase
Pfl4866CTGTCgcTGGGCTGTCAa3127Hypothetical protein
Pfl4919CaGTCATCGGGtTGTCAT2143Hypothetical protein/ predicted flavoprotein
Pfl5388tgGTCATGAAGtTGTCAT337Hypothetical protein
Pfl5465CTGTCATACGACcGTCAT12hemBDelta-aminolevulinic acid dehydratase
Pfl5605tTGTCATTTAACTGTCtT2122Hypothetical protein
Pfl5606CTGTCAcACGACTGcaAT354phoBPhosphate regulon transcription factor
Pfl5614tTGTCATCTTTCTGTCAT123pstCPeripheral membrane protein C
Pfl5615CTtTCATCCAAtTGTCAT279pstSPhosphate ABC transporter
Pfl5615tTGTCATATTTCaGcCAT368pstSPhosphate ABC transporter

A survey of the remaining Pho regulon candidates for indications of functions pertinent to biofilm formation identified Pfl1678 (Fig. 4A). Pfl1678 was predicted to encode a protein with an N-terminal EAL domain, as well as a C-terminal GGDEF domain (Fig. 4B). These domains have been implicated in the regulation of surface-associated behaviours via their direct modulation of c-di-GMP levels. The Pfl1678 promoter has two, well-conserved, overlapping Pho boxes, providing good support for its membership to the Pho regulon (Fig. 4A). To confirm this observation we performed RT-PCR comparing the levels of Pfl1678 transcript present in the WT after growth in high and low Pi conditions. Similar to the Pho-regulated phoD-control, much higher levels of Pfl1678 transcript were detected in low Pi conditions relative to growth in high Pi. In fact transcription of Pfl1678 was negligible when sufficient Pi was present in the medium (Fig. 4C). In addition, we also assessed Pfl1678 transcription in the WT relative to the pst mutant when grown in Pi-sufficient conditions. The pst mutant is constitutive for Pho regulon expression and should express Pfl1678 and the phoD-control in Pi-sufficient conditions. Indeed, this is what we observed (Fig. 4C). Both RT-PCR experiments utilized rplU transcript level as a Pho-independent control for cDNA synthesis. In each case the levels of rplU were comparable across experimental conditions (Fig. 4C). Taken together, these results indicate that Pfl1678 is a member of the Pho regulon, such that it is specifically expressed under conditions of Pi limitation.

image

Figure 4. Characterization of rapA as a Pho-regulated biofilm gene. A. Tandem overlapping Pho boxes located within RapA promoter. Bold type denotes direct repeats, brackets delineate separate Pho box motifs and arrow indicates the start codon for the rapA ORF. B. Schematic of RapA and the location of the predicted EAL and GGDEF domains. C. Analysis of rapA expression by RT-PCR. In the WT background rapA is only expressed in Pi-limiting conditions. RapA is also expressed in the pst mutant in high Pi, which is constitutive for Pho regulon expression. phoD is a known Pho regulon gene and served as a positive control, whereas rplU expression levels served as a Pho-independent control. D. Contribution of rapA to inhibition of biofilm formation in response to Pi limitation. Biofilm formation was assessed for WT, ΔrapA and ΔphoB in both high and low Pi. E. Contribution of rapA to the Δpst biofilm defect. Biofilms phenotypes were assessed for strains grown in Pi-sufficient conditions. ΔpstΔrapA::HRB-rapA expresses rapA in single copy from the Tn7 att site under control of its native promoter. ΔpstΔrapA::HRB2 is the vector only control and shows the same phenotype as the ΔpstΔrapA strain.

Download figure to PowerPoint

Pfl1678 is required for Pho regulon-mediated inhibition of biofilm formation

A clean deletion of Pfl1678 was constructed in the Pf0-1 strain background and assessed for biofilm formation in Pi-limiting conditions relative to the WT and the phoB mutant (Fig. 4D). In these conditions, the WT is inhibited for biofilm formation, whereas the phoB mutant is recalcitrant to the effects of Pi limitation, forming a robust biofilm. The Pfl1678 mutant was partially rescued for biofilm formation in Pi-limiting conditions, showing more biofilm than the WT, but less than that of the phoB mutant (Fig. 4D). Importantly, loss of Pfl1678 did not confer enhanced biofilm formation in Pi-sufficient conditions, suggesting that the biofilm rescue observed in Pi-limiting conditions is specific for Pi-dependent pathways. This result is consistent with the fact that Pfl1678 is only expressed in Pi-limiting conditions. Given this phenotype we refer to Pfl1678 as rapA (regulator of adherence by phosphate).

To further establish rapA as a major determinant of Pho-dependent biofilm inhibition, we investigated whether rapA was also required for mediating loss of biofilm formation in the pst mutant. Consistent with this hypothesis, an unmarked deletion of rapA in the pst mutant background partially rescued biofilm formation (∼70%) relative to the WT strain grown in Pi-sufficient conditions (Fig. 4E). Loss of rapA was both necessary and sufficient to account for the rescue of Δpst biofilm formation as providing rapA in trans at single copy, under control of its native promoter, was able to completely restore the loss of biofilm phenotype associated with the pst single mutant (Fig. 4E).

RapA is required to promote transition from reversible to irreversible attachment

The hallmark of the Δpst biofilm phenotype is a defect in transition from reversible to irreversible attachment. We were interested in understanding to what degree RapA activity mediated this early attachment defect. To address this question we quantified the relative proportion of reversible to irreversible attachment exhibited by the ΔpstΔrapA strain and compared this with the WT and Δpst strain. ΔpstΔrapA showed much higher levels of irreversible attachment (80% ± 1.2%) compared with the pst mutant (34% ± 3.5%). Importantly, ΔpstΔrapA still exhibited a defect in transition to irreversible attachment relative to the WT (97% ± 0.8%), which is consistent with results indicating that ΔpstΔrapA does not form biofilms to the same degree as the WT (Fig. 4E).

RapA is a phosphodiesterase that cleaves c-di-GMP to pGpG

Sequence analysis indicated that RapA has both a conserved EAL and a GGDEF domain (Fig. 4B). Previous work has shown these domains to be involved in the synthesis and degradation of c-di-GMP, a novel intracellular signalling molecule in bacteria (D'Argenio and Miller, 2004; Camilli and Bassler, 2006; Romling and Amikam, 2006). Specifically, proteins with the GGDEF domain have been shown to have diguanylate cyclase (DGC) activity, catalysing the synthesis of c-di-GMP from 2 GTP molecules, whereas proteins with the EAL domain have been shown to act as phosphodiesterases (PDE) that specifically cleave c-di-GMP. To assess these activities, RapA was overexpressed as a recombinant protein with an N-terminal His tag, and purified using a nickel affinity resin.

We assessed the ability of RapA to synthesize c-di-GMP from α-32P-GTP using conditions previously reported to support DGC activity (Paul et al., 2004). PleD, a well-characterized DGC, was utilized as a positive control to confirm that our assay allowed detection of c-di-GMP synthetic activity. Although PleD showed strong DGC activity, we were unable to observe any evidence of RapA DGC activity (Fig. 5A). We utilized 32P-labelled c-di-GMP, generated from DGC reactions performed with purified PleD, as substrate for PDE assays. Using published reaction conditions for PDE activity (Christen et al., 2005), we assessed whether RapA was capable of degrading c-di-GMP. As a control we included snake venom phosphodiesterase (SV-PDE), which has been shown to cleave c-di-GMP efficiently into GMP and we also included CC3396, a PDE shown to preferentially cleave c-di-GMP to the linear form, pGpG (Christen et al., 2005). Interestingly, these assays showed that RapA could indeed degrade c-di-GMP in a concentration-dependent fashion, but that similar to CC3396, RapA favours cleavage of c-di-GMP to the linear form, pGpG (Fig. 5B).

image

Figure 5. A. Assessment of the ability of RapA to synthesize c-di-GMP (DGC activity). PleD* served as a positive control the synthesis of c-di-GMP. Purified RapA (400 nM final concentration) did not exhibit DGC activity. B. Assessment of the ability RapA to cleave c-di-GMP (PDE activity). c-di-GMP synthesized with PleD* was used as substrate for PDE reactions. PDE reactions were performed with increasing concentrations (100, 200 and 400 nM) of RapA and the EAL domain mutant RapA-AAL. SV-PDE served as a control for complete cleavage of c-di-GMP to GMP, whereas CC3396 served as a control for linearization of c-di-GMP to yield pGpG.

Download figure to PowerPoint

Work in other systems has shown that mutation of the glutamic acid residue in the EAL domain to an alanine is sufficient to abolish PDE activity (Bobrov et al., 2005; Tamayo et al., 2005). We engineered this same mutation in RapA and purified the protein (RapA-AAL) in a similar fashion to WT RapA. Although RapA-AAL was stable, it no longer possessed the ability to cleave c-di-GMP (Fig. 5B). Therefore, as with other systems, the EAL domain, and more specifically the glutamic acid residue, is required for RapA PDE activity.

rapA-dependent modulation of c-di-GMP in vivo

In vitro studies suggested RapA serves to reduce the intracellular level of c-di-GMP under conditions of Pi limitation. We decided to test this prediction, by measuring the in vivo levels of c-di-GMP for WT and the rapA mutant when grown in Pi-limiting medium. For this analysis cells were grown for 6 h in Pi-limiting medium before whole cell labelling with sodium dihydrogen [32P]-orthophosphate. c-di-GMP was visualized by resolving whole cell acid extracts by two-dimensional thin-layer chromatography (2D-TLC) and subsequent analysis by autoradiography (Fig. 6A). c-di-GMP was identified based on TLC profiles published previously (Tischler and Camilli, 2004; Hickman et al., 2005). Visual inspection of TLC plates indicated increased levels of c-di-GMP in the rapA mutant relative to the WT (Fig. 6A). Quantification of this difference showed the rapA mutant to have 2.2-fold more intracellular c-di-GMP than the WT (Fig. 6B). This result agrees with in vitro studies, and further supports the role of RapA as a c-di-GMP PDE.

image

Figure 6. rapA is required for modulation of c-di-GMP in vivo. A. Autoradiographs of whole cell acid extracts separated by 2D-TLC. Cells were grown in Pi-limiting media for 6 h before labelling with sodium dihydrogen [32P]-orthophosphate for a further 6 h followed by acid extraction. This process was carried out for both WT and the rapA single mutant. Arrows indicate the position of c-di-GMP. B. Quantification of c-di-GMP levels from replicate experiments performed as described above. Standard error is shown for three or four replicates. rapA has 2.2-fold more c-di-GMP than the WT and this difference is statistically significant for the criteria P < 0.05.

Download figure to PowerPoint

The PDE activity of RapA is required for inhibition of biofilm formation

As a means to investigate the relationship between the biofilm inhibitory properties of RapA and its biochemical activities, we assessed whether the EAL-dependent PDE activity of RapA is required for repression of biofilm formation. For this experiment we cloned both the His tagged WT rapA (His-rapA) and the rapA EAL domain mutant (His-rapA-AAL), which were used for biochemical studies, into a Tn7 construct modified to place rapA under control of its native promoter. In this way, our analysis of His-rapA and His-rapA-AAL function directly mimicked that used previously for trans complementation studies (Fig. 4E). Importantly, addition of an N-terminal His tag did not impinge on the ability of WT rapA to complement a rapA null mutation. However, consistent with the PDE activity of RapA being required for biofilm repression, His-rapA-AAL was unable to restore loss of biofilm formation to the ΔpstΔrapA strain when expressed in trans (Fig. 7A). In fact, His-RapA-AAL showed complete loss of function within the resolution of our assay. Loss of complementation was not due to instability of His-RapA-AAL as we were able to detect full-length His-RapA-AAL in clarified cell extracts using the penta-His antibody (Fig. 7B).

image

Figure 7. The EAL domain of RapA is required for inhibition of biofilm formation. A. Complementation analysis of biofilm formation by the ΔpstΔrapA strain. Biofilm phenotypes are shown for each strain grown in Pi-sufficient conditions. ΔpstΔrapA::HRB-His-rapA expresses His-tagged rapA in single copy from the Tn7 att site under control of its native promoter. HRB-His-rapA-AAL differs from HRB-His-rapA in that the glutamic acid of the EAL domain has been changed to an alanine. ΔpstΔrapA::HRB2 is the vector only control and shows the same phenotype as the ΔpstΔrapA strain. B. Strains used for the above complementation analysis were analysed by Western Blot using an anti-His antibody. Both His-RapA and His-RapA-AAL are expressed and stable in P. fluorescens. Pf0-1::HRB2 served as a negative control, whereas purified His-RapA was used as a positive control for Western Blot detection.

Download figure to PowerPoint

RapA inhibits secretion and localization of the adhesin LapA

The Lap system plays a central role in biofilm formation, and is the only component known that is unconditionally required for biofilm formation in all environments tested to date. Our analysis had shown that lapA expression was not affected by Pho regulon expression at the level of transcription (Fig. 2C). Furthermore, although lapEBC transcription was reduced in a Pho-dependent fashion, the effect of this reduction on biofilm formation could not functionally account for Pi-mediated inhibition of biofilm formation (Figs 2 and 3). Given these data, we explored the possibility that Pho regulon expression was affecting LapA function via a post-transcriptional mechanism, either at the level of translation, secretion and/or localization to the outer membrane.

To facilitate quantification of LapA protein levels we constructed a vector that allowed us to introduce sequence encoding a 3× HA tag into the lapA gene of P. fluorescens Pf0-1. Allelic exchange using pMQ-lapA-HA led to a 3HA tag being positioned towards the C-terminus of the LapA protein, downstream of the last internal repeat of LapA. WT strains expressing LapA-HA showed normal levels of biofilm formation indicating that LapA-HA is functional (data not shown). We next confirmed that export of LapA-HA was still dependent on the LapEBC transporter. Consistent with this, LapA-HA could not be detected in the either the cell-associated (CA) or supernatant fractions of the lapB mutant (Fig. 8A). We routinely use the CA fraction as a measure of the amount of LapA that is tethered to the outer membrane. LapA was also seen to accumulate in the cytoplasm of the lapB mutant, consistent with a block in secretion (Fig. 8A). Together these results validated the use of LapA-HA as an appropriate surrogate for functional analysis of WT LapA.

image

Figure 8. Pho regulon expression impacts secretion and localization of LapA and requires rapA. A. The LapEBC transporter is required for LapA-HA secretion. Western Blot detection of LapA-HA for cytoplasmic (Cyto), cell-associated (CA) and supernatant (Sup) fractions prepared from both WT and a lapB mutant grown in Pi-sufficient conditions. B. Western Blot detection of Lap-HA for cytoplasmic (Cyto), cell-associated (CA) and supernatant (Sup) fractions prepared from the WT grown in high and low Pi conditions. C. Western Blot detection of Lap-HA for cytoplasmic (Cyto), cell-associated (CA) and supernatant (Sup) fractions prepared from WT, Δpst,ΔpstΔphoB and ΔpstΔrapA strains grown in Pi-sufficient conditions. D. Quantification of LapA-HA levels derived from replicate experiments performed as shown in (C). LapA-HA levels for mutants are shown as a percentage of WT levels. Standard error is shown for three or four replicates.

Download figure to PowerPoint

The impact of Pho regulon expression on LapA translation, secretion and localization to the outer membrane was first investigated by assessing the level of LapA-HA in cytoplasmic, CA and supernatant fractions of the WT grown in both high and low Pi conditions (Fig. 8B). We observed that growth in Pi-limiting conditions resulted in accumulation of LapA-HA in the cytoplasmic fraction (130% ± 12% WT level) as well as a marked decrease in the amount of CA LapA-HA (22% ± 3% WT level) relative to growth in Pi-sufficient conditions. These results are consistent with a Pi-dependent block in LapA secretion rather than repression of LapA translation in low Pi. Interestingly, we also observed that Pi-limiting environments resulted in higher levels of LapA-HA in the supernatant fraction (115% ± 6% WT level), suggesting that relocalization of LapA-HA from the CA fraction to the supernatant had also occurred as a consequence of phosphate starvation (Fig. 8B).

To test the genetic requirements for modulation of LapA secretion and localization by the Pho regulon, we compared LapA-HA levels in cytoplasmic, CA and supernatant fractions for the WT, Δpst and ΔpstΔphoB strains when grown in Pi-sufficient conditions (Fig. 8C and D). Consistent with a block in LapA secretion, the pst mutant showed a 76% reduction in CA LapA-HA relative to the WT, which was concomitant with a 56% increase in the cytoplasmic levels of LapA-HA relative to WT. To confirm that the effects on LapA secretion in the pst mutant are dependent on Pho regulon expression we carried out the same analysis for the pst phoB mutant. Our analysis showed that loss of Pho regulon expression in the pst mutant, due to deletion of phoB, restored levels of LapA-HA in the CA and cytoplasmic fractions to levels similar to that of the WT. In contrast to our analysis of the WT in high and low Pi (Fig. 8B), the pst mutant showed a 44% decrease in the level of LapA-HA in the supernatant fraction. A result that further supports inhibition of secretion, but provides no direct evidence for relocalization of LapA.

We assessed a role for RapA in mediating Pho-dependent decreases in LapA secretion by measuring levels of LapA-HA in cytoplasmic, CA and supernatant fractions of a pst rapA mutant. Deletion of rapA in the pst mutant background was seen to partially restore LapA secretion to levels demonstrated by the WT. Most instructive were the CA fractions, in which the level of LapA-HA increased from 24% to 61% of the WT level in the ΔpstΔrapA strain. This difference was statistically significant with a P-value of 0.003. Concomitant with increases in CA LapA-HA, we also observed partial restoration of WT LapA-HA levels in the cytoplasmic and supernatant fractions of the ΔpstΔrapA strain (Fig. 8C and D). Therefore, rapA is necessary, but not sufficient, to account for Pho-dependent effects on LapA secretion.

Modulation of LapA secretion by overexpression of PDE activities

The link between PDE activity and LapA secretion was further explored by looking at the effects of overexpressing rapA or a heterologous PDE in the WT background. We wondered whether modulating c-di-GMP levels in the absence of Pi-starvation would be sufficient to reduce biofilm formation as well as secretion of LapA. For this analysis we utilized pBB-rapA, which has rapA under transcriptional control of the PLAC promoter and pJN2133, which expresses PA2133 under control of the arabinose-inducible PBAD promoter. PA2133 is a PDE from P. aeruginosa that has been shown to reduce cellular pools of c-di-GMP in vivo (Hickman et al., 2005). We first demonstrated that overexpressing rapA or PA2133 led to severe reductions in biofilm formation by Pf0-1 when grown in Pi-sufficient media (Fig. 9A). Concomitant with loss of biofilm formation was a marked reduction of LapA in both CA and supernatant fractions when either rapA or PA2133 was overexpressed (Fig. 9B). This result mimics that seen for the pst mutant grown in Pi-sufficient conditions (Fig. 8C and D), except that we did not observe accumulation of LapA in the cytoplasmic fraction, as was the case in the pst strain (Fig. 9B).

image

Figure 9. Overexpression of PDE activities. A. Effect on P. fluorescens Pf0-1 biofilm formation by overexpression of RapA (pBB-rapA) and PA2133 (pJN2133). B. Effect of overexpressing RapA and PA2133 on LapA secretion/localization. Shown is Western Blot detection of LapA-HA for cytoplasmic (Cyto), cell-associated (CA) and supernatant (Sup) fractions prepared from WT harbouring no plasmid, pBB-rapA or pJN2133.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References

The work presented here extends our knowledge of c-di-GMP signalling and regulation of surface-associated behaviours in two major areas. We present the first mechanism linking a relevant environmental signal to control of c-di-GMP levels in the cell. Furthermore, we demonstrate a novel role for c-di-GMP in regulating secretion of an adhesin required for biofilm formation, which further extends the biological roles of c-di-GMP beyond regulation of exopolysaccharide production and motility.

The impetus for this study derived from a desire to understand the molecular mechanism by which Pho regulon expression elicits its negative control on biofilm formation. We have approached answering this question by identifying and characterizing the involvement of candidate Pho-regulated genes in Pi-dependent biofilm formation. We initially investigated a role for Pho regulon expression in regulating transcription of the lapA and lapEBC, which are known to be required for biofilm formation by P. fluorescens. Although lapEBC transcription was repressed in a PhoB-dependent fashion, the magnitude of this repression, in of itself, was unable to functionally account for the inhibition of biofilm formation due to Pho regulon expression.

Subsequently, bioinformatic analysis identified rapA as a putative Pho-regulated gene based on the presence of tandem overlapping Pho boxes upstream of the rapA ORF. A potential role for RapA in biofilm formation was suggested by the fact that RapA was predicted to have an N-terminal EAL domain as well as a C-terminal GGDEF domain. Proteins with these domains have been shown to regulate surface-associated behaviours in a range of bacteria, most likely through direct effects on the metabolism of c-di-GMP, which is a novel intracellular signalling molecule in bacteria (Aldridge et al., 2003; Simm et al., 2004; Tischler and Camilli, 2004; Bobrov et al., 2005; Hickman et al., 2005; Kader et al., 2006; Ryan et al., 2006; Thormann et al., 2006).

Analysis of rapA expression confirmed that it was a member of the Pho regulon; its transcription being dependent on activation of the response regulator PhoB. Subsequent genetic analysis indicated that rapA was indeed required for Pho-dependent inhibition of biofilm formation. Two lines of evidence support this conclusion: (i) deletion of rapA partially suppresses loss of biofilm formation in Pi-limiting conditions; and (ii) deletion of rapA in the pst mutant restores ∼70% of the biofilm formation seen for a WT strain grown in Pi-sufficient conditions. The pst mutant is constitutive for Pho regulon expression and is therefore unable to form biofilms in part due to the inappropriate expression of rapA.

It is common for bacterial genomes to encode multiple GGDEF/EAL domain proteins (Galperin et al., 2001). P. fluorescens has 43 ORFs predicted to encode either an EAL and/or a GGDEF domain. An important question that must be addressed is, ‘How are the activities of such a large number of proteins coordinated to effectively modulate the levels of c-di-GMP?’ Assuming a reasonable fraction of EAL/GGDEF domains are catalytically active, it seems likely that the c-di-GMP modulating activity of EAL/GGDEF domain proteins will be tightly regulated, either at the level of gene expression, allosteric regulation, or control of spatial dynamics within the cell. Here we show that RapA abundance is regulated by the extracellular concentration of Pi through the coupling of rapA transcription to expression of the Pho regulon, the major pathway for adaptation to growth in Pi-limiting environments. In this way, environments that are low in Pi are sensed by PhoR, leading to specific induction of rapA transcription via activation of the response regulator PhoB. PhoB∼P is likely to directly activate transcription of RapA due to the presence of tandem overlapping Pho boxes in the rapA promoter.

Putative orthologues of RapA could be identified in P. fluorescens Pf-5 (92% aa identity), P. fluorescens SBW25 (87%), P. syringae DC3000 (81%), P. aeruginosa PA14 (76%), P. aeruginosa PAO1 (75%) and P. putida KT2440 (67%). In all the above cases we were able to locate overlapping Pho boxes with similar matches to the E. coli consensus as was seen for P. fluorescens Pf0-1 (data not shown). These data indicate the potential for conserved regulation of rapA orthologues by Pi; however, empirical studies will be required to address this further, as well as determine whether rapA-dependent inhibition of biofilm formation is a common trait among Pseudomonads. Outside of the genus Pseudomonas the identity for putative orthologues drops to below 45%, which combined with the widespread occurrence of EAL and GGDEF domain-containing proteins, suggests that in these cases inference of evolutionary relationships with RapA may be inappropriate.

Proteins containing GGDEF domains have been shown to catalyse the synthesis of c-di-GMP from GTP (Ryjenkov et al., 2005), whereas proteins containing EAL domains are known to express PDE activity, specifically cleaving c-di-GMP (Schmidt et al., 2005; Tamayo et al., 2005). This simple correlation is complicated by the occurrence of many dual domain proteins, such as RapA. Of the limited number studied, none have demonstrated both activities (Christen et al., 2005; Kazmierczak et al., 2006). In a continuance of this trend, RapA was shown to be a PDE that can cleave c-di-GMP, but is unable to synthesize c-di-GMP from GTP. Interestingly, RapA effectively cleaved a single phosphodiester bond of c-di-GMP to give the linear form, pGpG, but was seemingly unable to subsequently cleave pGpG to give GMP. Similar substrate specificity has also been demonstrated for other EAL domain proteins, such as VieA (Tamayo et al., 2005), FimX (Kazmierczak et al., 2006) and CC3396 (Christen et al., 2005). The EAL domain of RapA is required for its PDE activity as substituting the glutamic acid residue of the EAL domain for an alanine residue completely abolished RapA activity. This mutation also rendered RapA unable to complement a rapA null mutation, indicating that the PDE activity of RapA is necessary and sufficient to explain its biofilm inhibitory properties. Furthermore, we used whole cell labelling experiments to demonstrate that, consistent with an in vivo role for RapA as a PDE, rapA mutants have over twofold higher levels of intracellular c-di-GMP than the WT when grown in Pi-limiting medium. Recently, Schmidt et al. (2005) proposed that conservation of regions in the EAL domain, in addition to the EAL residues, was predictive of PDE activity. Interestingly, RapA, although an active PDE, does not show conservation within some of these proposed regions, suggesting that more empirical studies are needed to validate sequence-based activity predictions. Similar observations were recently made for FimX (Kazmierczak et al., 2006). In summary, in vitro and in vivo biochemical data, together with genetic data, strongly suggest that RapA acts specifically to reduce the intracellular levels of c-di-GMP and that reduction in levels of c-di-GMP have an inhibitory effect on biofilm formation.

Although catalytically inactive, the GGDEF domain of RapA may play a regulatory role. This possibility is suggested by recent experiments in Caulobacter crescentus which have shown that the GGDEF domain of CC3396 can bind GTP, but instead of catalysing c-di-GMP synthesis regulates the activity of the associated EAL domain (Christen et al., 2005). Addition of GTP was observed to enhance the PDE activity of CC3396 40-fold, and this stimulation of activity required the GGDEF domain. We are currently testing whether the GGDEF domain of RapA can allosterically regulate PDE activity in a manner similar to that described for CC3396. Of the 22 predicted EAL domain-containing proteins in the Pf0-1 genome, 17 also encode a GGDEF domain, suggesting that allosteric control may play an important role in c-di-GMP metabolism.

How do decreases in c-di-GMP levels inhibit biofilm formation? The majority of research currently available demonstrates a role for c-di-GMP in regulation of exopolysaccharide production and/or motility (Bomchil et al., 2003; Simm et al., 2004; Hickman et al., 2005; Lim et al., 2006). Increased levels of c-di-GMP are usually correlated with increased exopolysaccharide production and increased biofilm formation, whereas lower levels of c-di-GMP are seen to increase motility. Indeed these results were the basis of the generalized description of c-di-GMP as a modulator of the transition from sessile to motile lifestyles (Simm et al., 2004). Contrary to other systems we could find no evidence for rapA-dependent effects on exopolysaccharide production or flagellar-mediated motility (data not shown). Instead we describe a novel role for c-di-GMP in regulating protein secretion. Specifically, we show that Pho regulon expression inhibits secretion of the adhesin LapA, and that rapA is required for this inhibition. This finding supports other research suggesting that the range of biological processes regulated by c-di-GMP is not confined to regulation of polysaccharide synthesis and motility (Aldridge et al., 2003; Kader et al., 2006; Kazmierczak et al., 2006).

LapA is a large outer membrane adhesin that plays a pivotal role in the biofilm formation of P. fluorescens. Secretion of LapA via the LapEBC transporter is required for the transition from reversible to irreversible attachment (Hinsa et al., 2003). This is manifest in the inability of lapA mutants to commit to more stable and prolonged interactions with the surface. Importantly, expression of the Pho regulon was shown to result in a similar attachment defect to a lapA mutant, which is consistent with the observed defects in LapA secretion. In fact the LapA secretion profiles correlate directly with a given strains propensity to effectively make the transition to a more stable association with the surface. For instance, deletion of rapA in the pst mutant background led to restoration of both LapA secretion and transitioning to irreversible attachment, but in both cases restoration was not to that of WT levels. In general, rapA-dependent regulation of transition to irreversible attachment suggests that c-di-GMP is also not restricted to regulation of biofilm formation at the stage of maturation and development of tertiary structure.

In addition to a role for Pho regulon expression in inhibition of LapA secretion, we also obtained evidence suggesting that Pho regulon expression regulates localization of LapA to the outer membrane. Specifically we observed that the WT accumulates more LapA in the supernatant under Pi-limiting conditions relative to growth in Pi-sufficient conditions. This result suggests that Pho regulon induction directs release of LapA from the cell surface, in addition to inhibiting further secretion of LapA to the outer membrane. Interestingly, we did not see accumulation of LapA in the supernatant fraction of the pst mutant when grown in Pi-sufficient conditions, as was observed for the WT grown in Pi-limiting conditions. Genetic data support the conclusion that Pho regulon expression is necessary and sufficient to explain inhibition of WT biofilm formation in Pi-limiting conditions, as well as the biofilm defect of the pst mutant grown in Pi-sufficient conditions. Therefore, the differences in LapA localization we observed are more likely a consequence of differences in the temporal kinetics of Pho regulon induction between the two experiments. In the pst mutant Pho regulon expression is constitutive irrespective of Pi concentration. However, in the case of the WT grown in low Pi conditions, cells must transition from a ‘Pho-inactive’ to a ‘Pho-activated’ state because a certain amount of growth is required to deplete Pi in media to levels required for Pho regulon induction. This raises the intriguing possibility that the Pho regulon may impact two different aspects of LapA biology in a temporally regulated fashion. We are currently investigating this hypothesis by examining the temporal dynamics of Pho regulon induction in relation to secretion and localization of LapA to the outer membrane.

Interestingly, Pho-dependent expression of rapA cannot completely account for the biofilm inhibitory effects of Pho regulon expression. This is consistent with the fact that RapA activity is also unable to account for the total inhibition of LapA secretion conferred by Pho regulon expression. Together, these data suggest that other Pho-regulated genes play a role in repressing biofilm formation and may do so by also contributing to the inhibition of LapA secretion. One possibility is that other Pho biofilm genes also play roles in c-di-GMP metabolism, either directly affecting synthesis/degradation of this nucleotide or by increasing the realized activity of RapA. Support for this idea comes from demonstration that overexpressing rapA, or a heterologous PDE, led to complete inhibition of biofilm formation and was sufficient to markedly reduce LapA secretion to the outer membrane, independent of Pi concentration. However, we are wary of placing too much stock in overexpression experiments, which may be eliciting many secondary effects in the cell due to non-physiological modulation of c-di-GMP pools (Kazmierczak et al., 2006; Weber et al., 2006). Indeed, overexpression of PDE activities led to slight reductions in cytoplasmic levels of LapA, a result inconsistent with experiments performed under physiologically relevant conditions. To further address this question we are currently pursing the identity of the other Pho regulon genes required for inhibition of biofilm formation in response to environments limited for Pi.

We do not yet know the mechanisms by which LapA secretion is modulated in response to c-di-GMP levels. In the case of Acetobacter xylinum, c-di-GMP has been shown to act as an allosteric activator of cellulose synthase activity (Ross et al., 1990). Potentially, c-di-GMP is acting via a post-transcriptional mechanism to directly inhibit the ABC transporter, LapEBC. However, it is also likely that other genes will be involved. Recently, lapD was also shown to be required for efficient secretion of LapA to the cell surface in P. fluorescens (Hinsa and O'Toole, 2006), a phenotype similar to that which we describe here for rapA mutants. We do not yet understand the relationship of lapD to rapA in regard to regulation of LapA secretion, but the fact that lapD also possesses putative EAL and GDDEF domains suggests the involvement of c-di-GMP metabolism.

Inorganic phosphate is the preferred source of phosphate for bacteria; however, its concentration in soil is typically very low (< 2 μM) (Paul and Clark, 1989). Therefore, Pi is likely to be a limiting resource for P. fluorescens, which is commonly found inhabiting soil environments. Furthermore, alternative phosphate sources, such as phosphate minerals and organic phosphate are often adsorbed to surfaces (Paul and Clark, 1989). Therefore, the coupling of regulatory systems for adaptation to Pi-limiting environments with mechanisms for modulation of surface attachment would seem advantageous to P. fluorescens. We suggest that Pho regulon expression, due to Pi-limiting environments leads to the manifestation of a ‘surface-sensing mode’. In this mode, P. fluorescens initiates attachment to the surface in a polar fashion, but does not commit to that surface unless Pho regulon expression is subsequently repressed, thereby allowing LapA to be secreted to the outer membrane. The products of Pho regulon genes are often involved in assimilation of alternative sources of phosphate (i.e. alkaline phosphatase) (Wanner, 1996). Pho regulon expression therefore serves to generate Pi, a process facilitated by the diffusion constraints imposed by proximity of the bacterium to a surface. If P. fluorescens attaches to a surface where the local environment contains suitable phosphate sources, more stable attachment to that surface is then facilitated by repression of the Pho regulon. If however, the local environment does not contain suitable phosphate sources, P. fluorescens will not repress Pho expression and ultimately detach from the surface to seek out alternative environments to colonize. Interestingly, Agrobacterium tumefaciens has also been reported to regulate biofilm formation in response to the concentration of Pi (Danhorn et al., 2004). However, in this case expression of the Pho regulon enhances biofilm formation. The selective forces responsible for this difference are unknown, but may relate to the fact that, in contrast to P. fluorescens, A. tumefaciens is a plant pathogen. For A. tumefaciens biofilm formation may only serve as a prelude to infection rather than form the basis of a long-term lifestyle commitment. In any respect, this work suggests that Pi may be a generally important signal utilized by soil-bacteria to regulate biofilm formation.

Experimental procedures

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References

Strains and media

Strains and plasmids used in this study are listed in Table 2. P. fluorescens and E. coli were routinely cultured in Lysogeny Broth (LB) unless stated otherwise and were grown at 30°C and 37°C respectively. K10T-based media were prepared as previously described (Monds et al., 2006). K10Tπ media were used for Pi-limiting conditions and contained no added Pi, giving a Pi concentration of 0.14 mM (Monds et al., 2006). K10T-1 was utilized as Pi-sufficient conditions and consisted of K10T medium amended with 1 mM K2HPO4. TSP salts were prepared as described previously (Monds et al., 2006).

Table 2.  Bacterial strains and plasmids.
Strain or plasmidGenotype or descriptionReference
Escherichia coli
 DH5αsupE44 ΔlacU169(φ80lacZΔM15) hsdR17 thi-1 relA1 recA1Hanahan (1983)
 ER2566Strain for expression from T7 promotersNEB
 S17-1(λpir)thi pro hsdR- hsdM recA RP4-2::TcMu-Km::Tn7Simon et al. (1983)
Pseudomonas fluorescens
 Pf0-1Wild typeCompeau et al. (1988)
 lapBPf0-1::pMQ89-lapB; Gm(r)This study
 lapAPf0-1::pKO-lapA; Tc(r)This study
 ΔpstPf0-1 with deletion of pstSCAB-phoU; Gm(r)Monds et al. (2006)
 ΔphoBPf0-1 with deletion of phoB; Gm(r)Monds et al. (2006)
 ΔpstΔphoBΔpst with unmarked deletion of phoB; Gm(r)Monds et al. (2006)
 ΔrapAPf0-1 with unmarked deletion of rapAThis study
 ΔpstΔrapAΔpst with unmarked deletion of rapA; Gm(r)This study
 PF-001Pf0-1::pUC-PlapA-luc; Km(r)This study
 PF-002Δpst::pUC-PlapA-luc; Gm(r), Km(r)This study
 PF-003ΔpstΔphoB::pUC-PlapA-luc; Gm(r), Km(r)This study
 PF-004Pf0-1::pUC-PlapE-luc; Km(r)This study
 PF-005Δpst::pUC-PlapE-luc; Gm(r), Km(r)This study
 PF-006ΔpstΔphoB::pUC-PlapE-luc; Gm(r), Km(r)This study
 PF-007Pf0-1::pUC-PlapE-Δbox; Km(r)This study
 PF-008ΔpstΔrapA::HRB-rapA; Gm(r), Km(r)This study
 PF-009ΔpstΔrapA::HRB2; Gm(r), Km(r)This study
 PF-010ΔpstΔrapA::HRB-rapA-AAL; Gm(r), Km(r)This study
 PF-011ΔpstΔrapA::HRB-His-rapA; Gm(r), Km(r)This study
 PF-012ΔpstΔrapA::HRB-His-rapA-AAL; Gm(r), Km(r)This study
 PF-013Pf0-1 expressing lapA-HAThis study
 PF-014ΔlapB expressing lapA-HAThis study
 PF-015Δpst expressing lapA-HAThis study
 PF-016ΔpstΔphoB expressing lapA-HAThis study
 PF-017ΔpstΔrapA expressing lapA-HAThis study
 lapΔboxPf0-1 with PacI substituted Pho boxes within lapA and lapE promotersThis study
Plasmids
 pBBRMCS-2Broad-host range cloning vector; Km(r)Kovach et al. (1995)
 pBB-rapApBBRMCS-2 with rapA under control of Plac; Km(r)This study
 pET15bProtein expression vector; Ap(r)Novagen
 pET-rapApET15b derived rapA expression vector; Ap(r)This study
 pET-rapA-AALpET15b derived rapA-AAL expression vector; Ap(r)This study
 pEX-lap-ΔboxpMQ30 based vector for lapΔbox allele replacement; Gm(r)This study
 pHRB2pUC18T-Tn7T containing Km gene from pZE21-MCS2; Km(r)This study
 pHRB-rapApHRB2 expressing rapA; Km(r)This study
 pHRB-rapA-AALpHRB2-expressing rapA-AAL; Km(r)This study
 pHRB3pUC18T-Tn7T containing Km gene from pUC-lucK; Km(r)This study
 pHRB-PrapApHRB3 with rapA promoter preceding MCS; Km(r)This study
 pHRB-His-rapApHRB-PrapA expressing His-rapA from pET-rapA; Km(r)This study
 pHRB-His-rapA-AALpHRB-PrapA expressing His-rapA-AAL from pET-rapA-AAL; Km(r)This study
 pJN2133Expression plasmid for PA2133Hickman et al. (2005)
 pKO3Pseudomonas integration vector; MCS, oriT, lacZ′, Tc(r)Monds et al. (2006)
 pKO-lapASingle cross-over knockout vector for lapA derived from pKO3This study
 pKO-rapA-FSpKO3 derived rapA KO vector; Tc(r)This study
 pMQ30sacB aacC1 ColE1 lacZalpha oriT CEN4 ARSH6 URA3Shanks et al. (2006)
 pMQ71BpMQ71 modified to remove aacC1; Km(r)This study
 pMQ83-p30-TcsacB tetR ColE1 lacZalpha oriT CEN4 ARSH6 URA3Shanks et al. (2006)
 pMQ-lapA-HApMQ83-p30-Tc based vector to construct lapA-HA tagThis study
 pMQ-lapEBCpMQ-71B expressing lapEBC operonThis study
 pUC18T-Tn7TMini-Tn7 vector; Ap(r)Choi et al. (2005)
 pUC-lucKVector for construction of luciferase transcriptional fusions; Km(r)Monds et al. (2006)
 pUC-PlapA-lucpUC-lucK with PlapA luciferase fusion; Km(r)This study
 pUC-PlapE-lucpUC-lucK with PlapE luciferase fusion; Km(r)This study
 pUC-PlapE-ΔboxpUC-lucK with PlapE luciferase fusion with mutated Pho box; Km(r)This study
 pUX-BF13Tn7 helper plasmid encoding Tn7 transposition functions; Ap(r)Zuber et al. (2003)
 pZE21-MCS2Source of Km gene for pHRB2 constructionLutz and Bujard (1997)

Antibiotics were used at the following concentrations, unless otherwise stated: ampicillin (Ap), 100 μg ml−1; kanamycin (Km), 50 μg ml−1; tetracycline (Tc), 10–15 μg ml−1 (E. coli) and 30 μg ml−1 (P. fluorescens); gentamicin (Gm), 30 μg ml−1; chloramphenicol (Cm), 20 μg ml−1.

Static biofilm assay and quantification

Biofilm assays were carried out essentially as described (O'Toole and Kolter, 1998), except that K10Tπ was used for Pi-limiting conditions and K10T-1 was used as Pi-sufficient conditions. Quantification of biofilms was done as follows. Crystal violet was solubilized from biofilms by adding 150 μl of a solution containing 30% methanol and 10% acetic acid to individual wells of the microtiter dish. From here, 125 μl was transferred to a fresh flat bottom microtiter plate and the absorbance read at 550 nM on a Spectra Max M2 microplate reader (Molecular Devices, Sunnyvale, CA). Readings from no less than 10 wells per strain were used to calculate the average and standard error.

Fluorescent imaging of biofilms

Biofilms were grown in a 6 well polystyrene tray (BD Biosciences, Rockville, MD) filled with 5 ml of K10T-1. Biofilm assays were inoculated in a 1:50 dilution and incubated for 16 h at room temperature (RT) gently shaking. To visualize biofilms, Syto-9 (Molecular Probes, Carlsbad, CA) staining was carried as described by Kadouri and O'Toole (2005), except cells were rinsed using 1× TSP salts.

Enzyme assays

Qualitative alkaline phosphatase assays were pre-formed as previously described (Monds et al., 2006). Luciferase assays were performed as described previously (Monds et al., 2006), except that cells were grown, non-shaking, in 6 well polystyrene trays (BD Biosciences, Rockville, MD) for 6 h. This was done in an effort to analyse transcription of genes in conditions that more closely reflect the environment in which biofilm phenotypes are assessed.

Construction of luciferase fusions

Luciferase fusions were constructed for lapA and lapE by individually cloning 1–1.5 kb fragments, from directly upstream of the annotated start site, as NcoI/BglII fragments into pUC-lucK. The NcoI site of pUC-lucK overlaps the ATG start codon of luciferase such that fusions to luciferase included the native RBS for each gene. lapA and lapE fusions are referred to as pUC-PlapA-luc and pUC-PlapE-luc respectively. The Pho box mutant derivative, pUC-PlapE-Δbox, was constructed with the same primers as for pUC-PlapE-luc except that pEX-lapΔbox was used as template for the PCR. These constructs were used to generate merodiploid fusions as described previously (Monds et al., 2006).

RT-PCR

Qualitative RT-PCR was performed as described previously (Monds et al., 2006). Real-time RT-PCR was performed as described previously (Kuchma et al., 2005). Growth conditions used for harvesting of RNA were the same as used for analysis of luciferase fusions. The primers used were as follows: rapA-RT-F, 5′-AAG TGC CGG TCA ATC TTC ACG TAG; rapA-RT-R, 5′-TGC TTC AGG ATT TCG GCA TTC CAC; phoD-RT-F, 5′-ATA TTC ACC GAT GCG TTT GCA GGC; phoD-RT-R, 5′-ATC ACC TTC CAC TGG GCT TTC GAG; lapE-RT-F, ACT CGG AAA CCA ACT TCC TCA CG; lapE-RT-R, 5′-ATT CGT TGT TGT GAC CGT TCT GGC; lapA-RT-F2, 5′-AAC CTC ACG TAC TTC ATC ACC GAC; lapA-RT-R2, 5′-AGC TCC AGT TGT TGT TCT TGA GGG; lapB-RT-F, 5′-ACC CTC CTG ATC TTC ATT GTC ATC GC; lapB-RT-R, 5′-TTT CAG CGT TGT TGA CCT TCA CCG; lapC-RT-F, 5′-CGT CAA CAA GAT GCT GGT CAA CAC; lapC-RT-F-R, 5′-TCT GTT CCA GCT TGG CTT TGA GAC; rplU-RT-F, 5′-ATA CGC TTC ATG TGG TGC TTA CGG; rplU-RT-R, 5′-AGC AAT ACA AGG TCG CTG AAG GTG.

Quantitative time-lapse microscopy

(i) Initial attachment: overnight cultures of strains were back diluted 1:50 into K10T-1 and grown to exponential phase (∼OD600 = 0.6). These cultures were then used to inoculate 5 ml of K10T-1 in a 6 well polystyrene tray. The inocula were normalized to 100 μl/OD600 = 0.6 to standardize the number of cells added to each well. After 2 min equilibration time at RT, a single time-lapse movie of 5 min duration was captured for each strain by phase-contrast microscopy using a DM IRB inverted microscope equipped with a cooled charged-couple-device digital camera (Leica Microsystems, Wetzlar, Germany). For analysis, the total number of bacteria that attached to the surface over the entire 5 min of the movie was recorded. This process was repeated 10 times for each strain and an average and standard error calculated.

(ii) Transition to irreversible attachment: overnight cultures of strains were back diluted 1:50 into K10T-1 and grown to exponential phase (∼OD600 = 0.6). These cultures were then used to inoculate 5 ml of K10T-1 in a 6 well polystyrene tray with a 1:50 dilution. Trays were incubated at 30°C for 30 min before a series of 1 min time-lapse movies were taken for each strain. For analysis, the fate of each cell adhered to the surface was recorded over the duration of the movie. Cells that stayed firmly adhered by their longitudinal axis were deemed to be ‘irreversibly’ attached, whereas, cells that moved about a single pole were deemed to be ‘reversibly’ attached. The percentage of bacteria within these two classes was calculated as a fraction of the total number of bacteria adhered to the surface at t = 0. A minimum of seven independent movies were used to calculate an average and standard error for each strain.

Construction of lapA-E Pho box mutant

Yeast cloning methods as described by Shanks et al. (2006) were utilized to create the construct to replace the right arm of the putative lapA-E Pho boxes with PacI restriction sites. Three PCR products were generated with the primer pairs: Pho KI-F1, CGT TGT AAA ACG ACG GCC AGT GCC AAG CTT GCA TGC CTG CA GCA CCA GGT TTT CGA AGT TG; Pho KI-R1, TGT GAC ATT ACT GGG GCG ATT GTT TAG GAT GGC ACC CAG A TTAATTAA TAG TTT GGC ATA AAG TTA ATA GCG and Pho KI-F2TCT GGG TGC CAT CCT AAA CA; Pho KI-R2, TAT GTC GGA AAG TTT AAA ACC GG and Pho KI-F3, CAG CGT GTG ATC CAG TTC CGG TTT TAA ACT TTC CGA CAT A TTAATTAA TCC GGT TGC CAT CTT CCG AC, Pho KI-R3, ACA CAG GAA ACA GCT ATG ACC ATG ATT ACG AAT TCG AGC TC TAG GAC TTG GAC TCC AGG TC. The sequence overlap present within these PCR products was used to facilitate homologous recombination with pMQ30, generating pEX-lapΔbox. Allelic exchange was performed as described previously (Monds et al., 2006). The strain with mutated versions of the putative lap Pho boxes is referred to as lapΔbox.

Construction of rapA mutants

An unmarked deletion of rapA was constructed as follows. First, a ∼1 kb fragment upstream of the rapA ORF was amplified by PCR with the primers RapA-pEX-F1-XhoI, ACA TTT CTC GAG TGA ACG ATG AAG GCA GCG AC and RapA-pEX-R1-EcoRI, TAA GCA GAA TTC TGC CAT GTC CGC CAA CTG AC. This fragment extended 24 bp into the rapA ORF and incorporated an XhoI site followed by a stop codon. PCR conditions were as described previously, except that Phusion polymerase (New England Biolabs, Beverly, MA) was used in place of Taq polymerase. A second fragment was amplified starting at bp 780 of the 1790 bp rapA ORF and extending ∼1 kb downstream, an XhoI site was also incorporated into the forward primer. Both amplicons were restricted with XhoI, ligated and used as the template for PCR using primers nested inside the borders of the ligated fragment. The resultant ∼2 kb fragment was gel purified, and cloned into the vector pKO3 using the restriction enzymes EcoRI and BamHI, resulting in the plasmid pKO-rapA-FS. Pf0-1 was then transformed with pKO-rap-FS as described previously (Monds et al., 2006). Merodiploids, were recovered as Tc-resistant colonies, and the correct point of recombination subsequently verified by PCR.

Cycloserine enrichment was used in conjunction with merodiploids to generate the rapA unmarked deletion. The procedure was modified from existing methodology (X.-X. Zhang, pers. comm.). Briefly, overnight cultures of merodiploids, grown without selection, were back-diluted 1:50 into 5 ml of LB supplemented with Tc. After incubating shaking for 1 h, the culture was amended with cycloserine to a final concentration of 1.5 mg ml−1 and incubated shaking for another 4 h. Then 2 ml of cells was pelleted by centrifugation, washed twice with and resuspended in 1 ml of 1× M63 salts. The cell suspension was then serial diluted and plated for single colonies on LB + Xgal. Bacteria that underwent recombination to lose the vector were identified as white colonies, whereas those that had not undergone recombination at this locus remained blue due to the presence of the lacZ gene on pKO3. White colonies were transferred in replica to a fresh LB + Xgal plate as well as a Tc plate. White, Tc-sensitive clones were then screened for the presence of the mutation using colony PCR. Candidate rapA mutants were sequenced to verify the mutation, the end result being a truncation in rapA conferring expression of only the first 8 aa. This procedure was repeated in the Δpst background to create the pst rapA double mutant.

rapA complementation constructs

(i) Construction of pHRB2. To generate a Km-marked Tn7, the Km gene from pZE21-MCS2 was amplified by PCR using the primers ZE21-Km-F-V, 5′-TAG TTT GAT ATC CGT CGG AAT TGC CAG CTG G and ZE21-Km-R-V, 5′-TAA TGC GAT ATC CCA GAG TCC CGC TCA GAA G. The resultant product was cloned into pUC18T-mTn7T using the unique EcoRV site in the MCS, generating the plasmid pHRB2.

(ii) Tn7 single copy construct. WT rapA was amplified, inclusive of its native promoter, using the primers pBB-rapA-F-EcoRI, AAG GCT GAA TTC AAC CGT CAG CCT CAA TGA AG and pBB-rapA-R-BamHI, TTA GAT GGA TCC TTA CTA AGT CGT GGA GGG AAG. The resultant 2.2 kb fragment was then cloned into pHRB2 using the EcoRI and BamHI restriction sites, generating pHRB-rapA.

(iii) Over-expression construct. The rapA ORF inclusive of RBS was amplified using the primer set pBB-rap(lac)-F, 5′-TTA GAT CTC GAG TAT CGC CAG AGA TTG CCA TGA C; pBB-rap(lac)-R, 5′-AAG GCT GAA TTC GCA TTT GAA GAG GTC AGA ACT GC and cloned into pBBRMCS-2 using the EcoRI/XhoI restriction sites, generating the plasmid pBB-Plac-rapA. This placed rapA under transcriptional control of the Plac promoter rather than its Pho-regulated native promoter.

Bioinformatics

We first searched the P. fluorescens genome for sites that match the defined E. coli Pho box sequence (CTGTCATnnnnCTGTCAT). This sequence is two 7-mers separated by four nucleotides. The search was done using ‘find sequence’ in Gene Inspector (Textco BioSoftware, West Lebanon, NH). Allowing varying amounts of mismatch in the 14 defined nucleotides we found 2800 occurrences with four mismatches, 270 with three mismatches, 33 with two mismatches, and six with one mismatch, 0 with no mismatches. Using all motifs having less than four mismatches (a total of 305 occurrences), we then filtered the list to choose only those occurrences that were no more than 200 nucleotides away from the start of a gene and in the same orientation as the gene. These sequences are candidates for serving a regulatory role. There were 43 motifs that matched these criteria (Table 1).

Enzyme purification

(i) RapA and RapA-AAL. The rapA ORF was amplified with the primer set rapA-pET-F-NdeI, 5′-AAA TGA CAT ATG ACC ACG ACC GAA CAG CTG AG; rapA-pET-R-XhoI, 5′-TAG TTA CTC GAG CAT TTG AAG AGG TCA GAA CTG CTC G and then cloned into pET15b using NdeI and XhoI restriction sites, generating pET-rapA. The EAL domain mutant of rapA was constructed using overlap extension PCR as described previously (Sambrook and Russell, 2001). First round primer sets were rapA-AAL-F1, 5′-ATG ACC ACG ACC GAA CAG CTG AGT GC; rapA-AAL-R1, 5′-CGG GTC AGG GCT GCG TAG CCG AGG ATG CGC CGT and rapA-AAL-F2, 5′-CGC AGC CCT GAC CCG CGG CCC GT; rapA-AAL-R2, 5′-CAT TTG AAG AGG TCA GAA CTG CTC GAT TGT C. The second round set was the same as for amplification of the WT rapA ORF as listed above. Subsequent cloning steps were the same as outlined for construction of pET-rapA and resulted in generation of pET-rapA-AAL, which expresses RapA with an alanine in place of the conserved glutamic acid of the EAL domain. To purify RapA and derivatives, ER2566 (New England Biolabs, Beverley, MA) harbouring pET-rapA was grown o/n in LB + Ap and then back diluted 1:100 into 1 l of fresh LB + Ap and incubated shaking at 37°C. At an OD600 of 0.6, expression of RapA was induced by addition of isopropyl-beta-D-thiogalactopyranosideI to a final concentration of 0.4 mM and incubation o/n at 12°C. Cells were harvested at 4000 g for 10 min and resuspended in binding buffer (20 mM sodium phosphate pH 7.4, 0.5 M NaCl, 20 mM imidazole). EDTA-free protease inhibitor (Roche, Indianapolis, IN) was added as per the manufacturers' instructions and the cells lysed by French Press. The lysate was centrifuged at 13 000 g for 30 min and the soluble fraction recovered by decanting the supernatant. Genomic DNA was subsequently sheared by passing the supernatant through a 25G needle and then loaded on to a 5 ml HisTrap FF column (GE Healthcare, Piscataway, NJ) attached to a BioLogic LP low-pressure chromatography system (Bio-Rad, Hercules, CA). The column was washed with binding buffer before His-RapA was eluted using an imidazole gradient as per the manufacturers' instructions. Fractions containing RapA were pooled and concentrated using an Amicon Ultracel 10 000 molecular weight cut-off (MWCO) column (Millipore, Billerica, MA) and then dialysed against 75 mM Tris-HCl pH 8.0, 250 mM NaCl, 25 mM KCl, 10 mM MgCl2, 30% glycerol using a Slide-A-Lyzer 10 000 MWCO dialysis cassette (Pierce, Rockford, IL). Protein concentrations were determined using the BCA protein assay kit (Pierce, Rockford, IL). His-RapA-AAL was purified in an identical fashion.

(ii) PleD* and CC3396. Purification of PleD* was performed as described previously (Paul et al., 2004). Purification of CC3396 was performed as described previously (Christen et al., 2005).

Assessment of His-RapA and His-RapA-AAL function and stability

(i) Construction of pHRB-PrapA. In order to facilitate direct cloning of His-tagged rapA from pET15b-based vectors, an alternative Km-marked Tn7 was constructed. The Km gene from pUC-lucK was amplified with the primers Km-luc-F2, 5′-CGA TTC AGG CCT GGT ATG AG; Km-luc-R2, 5′-TCT AGC AAG CTG CAA GAT CC and cloned into pUC18T-Tn7T using the EcoRV restriction site, generating pHRB3. No restriction sites were incorporated in the primer because the Phusion polymerase (New England Biolabs, Beverly, MA) generates blunt ends during amplification. pHRB3 was further modified by insertion of the rapA promoter between the KpnI/EcoRI restriction sites, generating pHRB-PrapA. The rapA promoter was amplified using the primers pBB-rapA-KpnI, 5′-AAG GCT GGT ACC AAC CGT CAG CCT CAA TGA and pBB-rapA-NcoI-EcoRI, 5′-TTA GAT GAA TTC CCA TGG CAA TCT CTG GCG ATA. pHRB-PrapA was designed to contain unique NcoI, BamHI sites in the MCS, and were positioned such that the ATG of the NcoI site overlapped the start site for the rapA ORF.

(ii) Expression and detection of His-rapA and His-rapA-AAL. The ∼1.8 kb fragment containing rapA or rapA-AAL was excised from pET-rapA and pET-rapA-AAL, respectively, using the restriction enzymes NcoI and BamHI. This fragment was then cloned into pHRB-PrapA to generate either pHRB-PrapA-rapA or pHRB-PrapA-rapA-AAL. The NcoI site of pET15b is positioned at the start site of the N-terminal His ORF, such that cloning into pHRB-PrapA generates a seamless fusion with the rapA promoter, mimicking the native genomic context. ΔpstΔrapA was transformed separately with pHRB-PrapA-rapA and pHRB-PrapA-rapA-AAL as described previously (Monds et al., 2006).

To detect His tagged proteins, cultures were grown, shaking, in 50 ml of K10T-1 for 6 h and clarified cell extracts prepared as described by Caiazza and O'Toole (2004). His epitopes were detected by Western blot analysis using the penta-His antibody (Qiagen, Valencia, CA) essentially following the manufacturer's instructions.

Diguanylate cyclase assays

Diguanylate cyclase assays were performed essentially as described by Paul et al. (2004), with some modification. Briefly, purified proteins were added to reaction buffer consisting of 75 mM Tris pH 8.0, 250 mM NaCl, 25 mM KCl, 10 mM MgCl2 and 0.2 mM of a 1:8 mix of GTP (Sigma-Aldrich, St Louis, MO) and [α-32P]-GTP (3000 Ci mMol−1; GE Healthcare), in a total reaction volume of 50 μl. Reactions were incubated for 30 min at RT before termination of the reaction by addition of 10 μl of 0.5 M EDTA pH 8.0.

Synthesis of c-di-GMP

Purified PleD* was used to synthesize c-di-GMP as substrate for PDE assays. For this, the standard DGC assay was scaled up 10-fold. In addition, reactions were processed as described by Tamayo et al. (2005) to hydrolyse non-cyclized phosphate groups and remove proteins. This involved treating reactions with calf intestinal phosphatase (Roche, Indianapolis, IN) for 30 min at 37°C, followed by passage through a Microcon YM-10 10K MWCO column (Millipore, Billerica, MA) by centrifugation at 13 000 g for 30 min. Products were analysed by TLC as described below and the concentration of c-di-GMP substrate calculated.

Phosphodiesterase assays

Phosphodiesterase assays were performed essentially as described by Christen et al. (2005). Briefly, purified proteins were added to reaction buffer consisting of 75 mM Tris pH 8.0, 250 mM NaCl, 25 mM KCl, 10 mM MgCl2. Reactions were initiated by addition of ∼40 ρM of c-di-GMP substrate. Reactions were incubated at RT for 30 min before termination by addition of 10 μl of 0.5 M EDTA pH 8.0.

Thin-layer chromatography

Reaction products from in vitro assays were mixed with an equal volume of running buffer consisting of 1:1.5 (v/v) saturated NH4SO2 and 1.5 M KH2PO4, pH 3.6 before 6 μl of the reaction was spotted and dried onto Cellulose PEI TLC plates (Selecto Scientific, Suwanee, GA) in 4 × 1.5 μl aliquots. Plates were developed in running buffer, air-dried and exposed to a storage phosphor screen (GE Healthcare, Piscataway, NJ). After an o/n exposure phosphor screens were read on a Storm 860 (Molecular Devices, Sunnyvale, CA) and analysed using Image Quant software v5.1 (Molecular Devices, Sunnyvale, CA).

In vivo quantification of c-di-GMP

Whole cell 32P-orthophosphate-labelling, acid extraction and 2D-TLC analysis of nucleotides were performed as reported (Hickman et al., 2005), with the following modifications. Cells were grown in Pi-limiting media for 6 h before labelling with sodium dihydrogen [32P]-orthophosphate, then incubated for a further 6 h followed by acid extraction. This process was carried out for both WT and the rapA single mutant. TLC plates were developed in running buffer, air-dried and exposed to a storage phosphor screen (GE Healthcare, Piscataway, NJ). After an o/n exposure phosphor screens were read on a Storm 860 (Molecular Devices, Sunnyvale, CA) and analysed using Image Quant software v5.1 (Molecular Devices, Sunnyvale, CA). The percentage of label incorporated into c-di-GMP was normalized to total 32P labelling and expressed as percentage c-di-GMP.

Construction of HA-tagged lapA

A vector for lapA-HA allelic exchange was constructed using the yeast cloning vector pMQ83-p30-Tc and associated yeast cloning techniques as described by Shanks et al. (2006). The following primers were used to amplify two approximately 1 kb fragments of lapA: lapA-HA-F1B, 5′-AAG CTT GCA TGC CT G CAG GTC GAC TCT AGA GGA TCC CCG CAC CGC TGA CCA ACT CGA TC; lapA-HA-R1, 5′-AAC ATC GTA TGG GTA GAG CAG GCT GCC GGT GAT GCT GAC AG and lapA-HA-F2, 5′-GAC GTT CCA GAT TAC GCT GCT ACC TCG GTA ACC GAA GGC CAG; lapA-HA-R2, 5′-CAG CTA TGA CCA TGA TTA CGA ATT CGA GCT CGG TAC CCG TTG GCA TCC GGC ACG AAG C. The HA tag was amplified using pRMQS95 + mltR-3HA as a template with the primers HA tag-F, 5′-CTG TCA GCA TCA CCG GCA GC CTG CTC TAC CCA TAC GAT GTT CCT G; HA tag-R, 5′-CTG GCC TTC GGT TAC CGA GGT AGC AGC GTA ATC TGG AAC GTC ATA TG. All three PCR products as well as SmaI-digested pMQ83-p30-Tc were used to transform Saccharomyces cerevisiae. The PCR primers were designed such that all three products can recombine by homologous recombination in only one orientation with the linearized vector. Recircularization of the vector through homologous recombination is required for plasmid maintenance, which can be selected by plating cells in URA-deficient medium. Plasmids from yeast transformants were recovered and subsequently used to transform E. coli. Routine methods were employed to verify the authenticity of the recombinant plasmid, which was referred to as pMQ-lapA-HA. Allelic replacement techniques were then used to move the HA tag into P. fluorescens Pf0-1 and derivatives. The final result is a LapA with 3HA epitopes inserted after residue 4093 (12278 bp) in the full-length protein of 5211 aa.

LapA secretion assays

Samples were prepared as reported (Hinsa et al., 2003) with the following modifications. Bacterial cultures were grown overnight in LB, and then subcultured into 30 ml of fresh LB at a 1:100 dilution and grown shaking at 200 rpm. For assaying secretion in high versus low Pi, subculturing was performed in K10T-1 and K10Tπ respectively. After 6 h of incubation, cells were harvested by centrifugation (4000 g, 10 min). Fifteen millilitres of the resulting supernatant was passed through a 0.22 μM filter (Millipore, Billerica, MA) to remove residual cells, and then concentrated 30-fold in 30 000 MWCO filter column (Millipore, Billerica, MA). The final volume was normalized to the optical density of the source culture, yielding the supernatant fraction. Cell pellets were resuspended in 300 μl of resuspension buffer (Tris-HCl, 20 mM, pH 8 plus 10 mM MgCl2) transferred to a microcentrifuge tube, vortexed for 10 s, and then centrifuged again (14 000 g, 4 min). The resulting supernatant was removed and the final volume normalized to optical density, yielding the CA fraction. The remaining pellet was resuspended in 2 ml of resuspension buffer plus 1× Complete, EDTA-free protease inhibitors (Roche, Indianapolis, IN) and a Thermo Spectronic French press minicell was used to lyse the cells by processing twice at 20 000 psi. Lysates were centrifuged (14 000 g, 15 min) to pellet unbroken cells and the resulting supernatants designated cytoplasmic fractions. Relative protein concentrations for cytoplasmic fractions were determined using the BCA assay as per the manufacturer's instructions (Pierce, Rockford, IL). All steps were performed on ice or at 4°C.

Samples were resolved by SDS-PAGE in 5% polyacrylamide gels (Bio-Rad, Hercules, CA) and Western blotting was performed as described previously (Caiazza and O'Toole, 2004). Blots were probed for LapA-HA using an anti-HA antibody (QED Bioscience, San Diego, CA). A horseradish peroxidase-conjugated secondary antibody (goat anti-rabbit; MP Biomedicals, Solon, OH) was used for chemiluminescent detection of bound ligands. Detection was carried out using Visualizer EC reagents (Upstate, Billerica, MA) and exposure to film (Blue XB-1; Kodak, Rochester, NY). Film images of Western blots were digitized using an HP scanjet 3970 scanner and densitometry performed on digital images using Image J software (http://rsb.info.nih.gov/ij/).

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References

We thank C. Harwood and U. Jenal for the gift of plasmids. This work was supported by T32 GM08704 predoctoral fellowship to P.D.N. and NSF9984521 to G.A.O.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  • Aldridge, P., Paul, R., Goymer, P., Rainey, P., and Jenal, U. (2003) Role of the GGDEF regulator PleD in polar development of Caulobacter crescentus. Mol Microbiol 47: 16951708.
  • Banin, E., Vasil, M.L., and Greenberg, E.P. (2005) Iron and Pseudomonas aeruginosa biofilm formation. Proc Natl Acad Sci USA 102: 1107611081.
  • Bobrov, A.G., Kirillina, O., and Perry, R.D. (2005) The phosphodiesterase activity of the HmsP EAL domain is required for negative regulation of biofilm formation in Yersinia pestis. FEMS Microbiol Lett 247: 123130.
  • Bomchil, N., Watnick, P., and Kolter, R. (2003) Identification and characterization of a Vibrio cholerae gene, mbaA, involved in maintenance of biofilm architecture. J Bacteriol 185: 13841390.
  • Caiazza, N.C., and O'Toole, G.A. (2004) SadB is required for the transition from reversible to irreversible attachment during biofilm formation by Pseudomonas aeruginosa PA14. J Bacteriol 186: 44764485.
  • Camilli, A., and Bassler, B.L. (2006) Bacterial small-molecule signaling pathways. Science 311: 11131116.
  • Choi, K.H., Gaynor, J.B., White, K.G., Lopez, C., Bosio, C.M., Karkhoff-Schweizer, R.R., and Schweizer, H.P. (2005) A Tn7-based broad-range bacterial cloning and expression system. Nature Methods 2: 443448.
  • Christen, M., Christen, B., Folcher, M., Schauerte, A., and Jenal, U. (2005) Identification and characterization of a cyclic di-GMP-specific phosphodiesterase and its allosteric control by GTP. J Biol Chem 280: 3082930837.
  • Compeau, G.B., Kilstrup, M., Barilla, K., Jochimsen, B., and Levy, S.B. (1988) Survival of rifampin-resistant mutants of Pseudomonas fluorescens and Pseudomonas putida in soil systems. Appl Environ Microbiol 54: 24322438.
  • D'Argenio, D.A., and Miller, S.I. (2004) Cyclic di-GMP as a bacterial second messenger. Microbiology 150: 24972502.
  • Danhorn, T., Hentzer, M., Givskov, M., Parsek, M.R., and Fuqua, C. (2004) Phosphorus limitation enhances biofilm formation of the plant pathogen Agrobacterium tumefaciens through the PhoR-PhoB regulatory system. J Bacteriol 186: 44924501.
  • Davey, M.E., and O'Toole, G.A. (2000) Microbial biofilms: from ecology to molecular genetics. Microbiol Mol Biol Rev 64: 847867.
  • Galperin, M.Y., Nikolskaya, A.N., and Koonin, E.V. (2001) Novel domains of the prokaryotic two-component signal transduction systems. FEMS Microbiol Lett 203: 1121.
  • Gonin, M., Quardokus, E.M., O'Donnol, D., Maddock, J., and Brun, Y.V. (2000) Regulation of stalk elongation by phosphate in Caulobacter crescentus. J Bacteriol 182: 337347.
  • Hanahan, D. (1983) Studies on transformation of Escherichia coli with plasmids. J Mol Biol 166: 557580.
  • Hickman, J.W., Tifrea, D.F., and Harwood, C.S. (2005) A chemosensory system that regulates biofilm formation through modulation of cyclic diguanylate levels. Proc Natl Acad Sci USA 102: 1442214427.
  • Hinsa, S.M., and O'Toole, G.A. (2006) Biofilm formation by Pseudomonas fluorescens WCS365: a role for LapD. Microbiology 152: 13751383.
  • Hinsa, S.M., Espinosa-Urgel, M., Ramos, J.L., and O'Toole, G.A. (2003) Transition from reversible to irreversible attachment during biofilm formation by Pseudomonas fluorescens WCS365 requires an ABC transporter and a large secreted protein. Mol Microbiol 49: 905918.
  • Jackson, D.W., Simecka, J.W., and Romeo, T. (2002) Catabolite repression of Escherichia coli biofilm formation. J Bacteriol 184: 34063410.
  • Kader, A., Simm, R., Gerstel, U., Morr, M., and Romling, U. (2006) Hierarchical involvement of various GGDEF domain proteins in rdar morphotype development of Salmonella enterica serovar Typhimurium. Mol Microbiol 60: 602616.
  • Kadouri, D., and O'Toole, G.A. (2005) Susceptibility of biofilms to Bdellovibrio bacteriovorus attack. Appl Environ Microbiol 71: 40444051.
  • Kazmierczak, B.I., Lebron, M.B., and Murray, T.S. (2006) Analysis of FimX, a phosphodiesterase that governs twitching motility in Pseudomonas aeruginosa. Mol Microbiol 60: 10261043.
  • Kovach, M.E., Elzer, P.H., Hill, D.S., Robertson, G.T., Farris, M.A., Roop, R.M., 2nd, and Peterson, K.M. (1995) Four new derivatives of the broad-host-range cloning vector pBBR1MCS, carrying different antibiotic-resistance cassettes. Gene 166: 175176.
  • Kuchma, S.L., Connolly, J.P., and O'Toole, G.A. (2005) A three-component regulatory system regulates biofilm maturation and type III secretion in Pseudomonas aeruginosa. J Bacteriol 187: 14411454.
  • Lim, B., Beyhan, S., Meir, J., and Yildiz, F.H. (2006) Cyclic-diGMP signal transduction systems in Vibrio cholerae: modulation of rugosity and biofilm formation. Mol Microbiol 60: 331348.
  • Lutz, R., and Bujard, H. (1997) Independent and tight regulation of transcriptional units in Escherichia coli via the LacR/O, the TetR/O and AraC/I1–I2 regulatory elements. Nucleic Acids Res 25: 12031210.
  • Mah, T.F., Pitts, B., Pellock, B., Walker, G.C., Stewart, P.S., and O'Toole, G.A. (2003) A genetic basis for Pseudomonas aeruginosa biofilm antibiotic resistance. Nature 426: 306310.
  • Monds, R.D., Silby, M.W., and Mahanty, H.K. (2001) Expression of the Pho regulon negatively regulates biofilm formation by Pseudomonas aureofaciens PA147-2. Mol Microbiol 42: 415426.
  • Monds, R.D., Newell, P.D., Schwartzman, J.A., and O'Toole, G.A. (2006) Conservation of the Pho regulon in Pseudomonas fluorescens Pf0-1. Appl Eviron Microbiol 72: 19101924.
  • O'Toole, G.A., and Kolter, R. (1998) Initiation of biofilm formation in Pseudomonas fluorescens WCS365 proceeds via multiple, convergent signalling pathways: a genetic analysis. Mol Microbiol 28: 449461.
  • Paul, E.A., and Clark, F.E. (1989) Soil Microbiology and Biochemistry. San Diego, CA: Academic Press.
  • Paul, R., Weiser, S., Amiot, N.C., Chan, C., Schirmer, T., Giese, B., and Jenal, U. (2004) Cell cycle-dependent dynamic localization of a bacterial response regulator with a novel di-guanylate cyclase output domain. Gene Dev 18: 715727.
  • Romling, U., and Amikam, D. (2006) Cyclic di-GMP as a second messenger. Curr Opin Microbiol 9: 218228.
  • Ross, P., Mayer, R., Weinhouse, H., Amikam, D., Huggirat, Y., Benziman, M., et al. (1990) The cyclic diguanylic acid regulatory system of cellulose synthesis in Acetobacter xylinum. J Biol Chem 265: 1893318943.
  • Ryan, R.P., Fouhy, Y., Lucey, J.F., Crossman, L.C., Spiro, S., He, Y.W., et al. (2006) Cell-cell signaling in Xanthomonas campestris involves an HD-GYP domain protein that functions in cyclic di-GMP turnover. Proc Natl Acad Sci USA 103: 67126717.
  • Ryjenkov, D.A., Tarutina, M., Moskvin, O.V., and Gomelsky, M. (2005) Cyclic diguanylate is a ubiquitous signaling molecule in bacteria: insights into biochemistry of the GGDEF protein domain. J Bacteriol 187: 17921798.
  • Sambrook, J., and Russell, D.W. (2001) Molecular Cloning: A Laboratory Manual. Woodbury, NY: Cold Spring Harbor Laboratory Press.
  • Schmidt, A.J., Ryjenkov, D.A., and Gomelsky, M. (2005) The ubiquitous protein domain EAL is a cyclic diguanylate-specific phosphodiesterase: enzymatically active and inactive EAL domains. J Bacteriol 187: 47744781.
  • Shanks, R.M., Caiazza, N.C., Hinsa, S.M., Toutain, C.M., and O'Toole, G.A. (2006) Saccharomyces cerevisiae-based molecular tool kit for manipulation of genes from gram-negative bacteria. Appl Environ Microbiol 72: 50275036.
  • Silby, M.W., and Levy, S.B. (2004) Use of in vivo expression technology to identify genes important in growth and survival of Pseudomonas fluorescens Pf0-1 in soil: discovery of expressed sequences with novel genetic organization. J Bacteriol 186: 74117419.
  • Simm, R., Morr, M., Kader, A., Nimtz, M., and Romling, U. (2004) GGDEF and EAL domains inversely regulate cyclic di-GMP levels and transition from sessility to motility. Mol Microbiol 53: 11231134.
  • Simon, R., Priefer, U., and Puhler, A. (1983) A broad host range system for in vivo genetic engineering: transposon mutagenesis in Gram-negative bacteria. Biotechnology 1: 784791.
  • Slater, H., Crow, M., Everson, L., and Salmond, G.P. (2003) Phosphate availability regulates biosynthesis of two antibiotics, prodigiosin and carbapenem. Serratia via both quorum-sensing-dependent and -independent pathways. Mol Microbiol 47: 303320.
  • Stanley, N.R., and Lazazzera, B.A. (2004) Environmental signals and regulatory pathways that influence biofilm formation. Mol Microbiol 52: 917924.
  • Tamayo, R., Tischler, A.D., and Camilli, A. (2005) The EAL domain protein VieA is a cyclic diguanylate phosphodiesterase. J Biol Chem 280: 3332433330.
  • Thormann, K.M., Duttler, S., Saville, R.M., Hyodo, M., Shukla, S., Hayakawa, Y., and Spormann, A.M. (2006) Control of formation and cellular detachment from Shewanella oneidensis MR-1 biofilms by cyclic di-GMP. J Bacteriol 188: 26812691.
  • Tischler, A.D., and Camilli, A. (2004) Cyclic diguanylate (c-di-GMP) regulates Vibrio cholerae biofilm formation. Mol Microbiol 53: 857869.
  • Wanner, B.L. (1996) Phosphorous assimilation and control of the phosphate regulon. In Eschericihia coli and Salmonella: Cellular and Molecular Biology. Neidhardt, F.C., Curtiss, R., III, Ingraham, J.L., Lin, E.C.C., Low, K.B., Magasanik, B., et al. (eds). Washington, DC: American Society for Microbiology Press, pp. 13571381.
  • Weber, H., Pesavento, C., Possling, A., Tischendorf, G., and Hengge, R. (2006) Cyclic-di-GMP-mediated signalling within the sigma network of Escherichia coli. Mol Microbiol 62: 10141034.
  • Zuber, S., Carruthers, F., Keel, C., Mattart, A., Blumer, C., Pessi, G., et al. (2003) GacS sensor domains pertinent to the regulation of exoproduct formation and to the biocontrol potential of Pseudomonas fluorescens CHA0. Mol Plant Microbe Interact 16: 634644.