Involvement of the Cpx signal transduction pathway of E. coli in biofilm formation

Authors

  • Corinne Dorel,

    Corresponding author
    1. Laboratoire de Génétique Moléculaire des Microorganismes et des Interactions Cellulaires, CNRS UMR 5577, Institut National des Sciences Appliquées de Lyon, 20 avenue Albert Einstein, 69621 Villeurbanne, Cedex, France
    Search for more papers by this author
  • Olivier Vidal,

    1. Laboratoire de Génétique Moléculaire des Microorganismes et des Interactions Cellulaires, CNRS UMR 5577, Institut National des Sciences Appliquées de Lyon, 20 avenue Albert Einstein, 69621 Villeurbanne, Cedex, France
    Search for more papers by this author
  • Claire Prigent-Combaret,

    1. Laboratoire de Génétique Moléculaire des Microorganismes et des Interactions Cellulaires, CNRS UMR 5577, Institut National des Sciences Appliquées de Lyon, 20 avenue Albert Einstein, 69621 Villeurbanne, Cedex, France
    Search for more papers by this author
  • Isabelle Vallet,

    1. Laboratoire de Génétique Moléculaire des Microorganismes et des Interactions Cellulaires, CNRS UMR 5577, Institut National des Sciences Appliquées de Lyon, 20 avenue Albert Einstein, 69621 Villeurbanne, Cedex, France
    Search for more papers by this author
  • Philippe Lejeune

    1. Laboratoire de Génétique Moléculaire des Microorganismes et des Interactions Cellulaires, CNRS UMR 5577, Institut National des Sciences Appliquées de Lyon, 20 avenue Albert Einstein, 69621 Villeurbanne, Cedex, France
    Search for more papers by this author

*Corresponding author. Tel.: +33 472 43 80 88; Fax: +33 472 43 87 14, E-mail address: dorel@insa.insa-lyon.fr

Abstract

In a genetic screening directed to identify genes involved in biofilm formation, mutations in the cpxA gene were found to reduce biofilm formation by affecting microbial adherence to solid surfaces. This effect was detected in Escherichia coli K12 as well as in E. coli strains isolated from patients with catheter-related bacteremia. We show that the negative effect of the cpxA mutation on biofilm formation results from a decreased transcription of the curlin encoding csgA gene. The effect of the cpxA mutation could not be observed in cpxR mutants, suggesting that they affect the same regulatory pathway. The cpxA101 mutation abolishes cpxA phosphatase activity and results in the accumulation of phosphorylated CpxR. Features of the strain carrying the cpxA101 mutation are a reduced ability to form biofilm and low levels of csgA transcription. Our results indicate that the cpxA gene increases the levels of csgA transcription by dephosphorylation of CpxR, which acts as a negative regulator at csgA. Thus, we propose the existence of a new signal transduction pathway involved in the adherence process in addition to the EnvZ-OmpR two-component system.

1Introduction

Biofilm formation on solid surfaces is a very common phenomenon among bacteria with important economic and medical consequences. A strain of Escherichia coli K-12 able to form a thick biofilm on inert surfaces such as glass or polystyrene has previously been isolated from a continuous culture in minimal media [1]. Increased adhesion to solid surfaces by this mutant is the result of the overproduction of curli, a particular class of pili. A single point mutation resulting in the replacement of a leucine by an arginine residue at position 43 in the OmpR protein is responsible for this phenotype. The presence of the ompR234 allele significantly increases the expression of the curlin encoding csgA gene. Loss of adherence properties of several strains of adherent E. coli is caused by transduction of knockout mutations in either the csgA or ompR gene [1]. Curli are therefore essential for biofilm formation and the EnvZ-OmpR two-component system regulates the synthesis of curlin, their major component. A similar regulation was demonstrated in Salmonella[2,3]. Besides osmolarity, curli expression in Salmonella and E. coli strains, including K12, avian E. coli and different clinical types, is highly regulated by environmental conditions such as temperature and the growth phase. Biogenesis of curli is restricted to temperatures below 30°C and the stationary growth phase [3–6]. Thus, it appears that curli production is tightly regulated by complex environmental signals and we used a genetic approach to characterize new systems at the molecular level required for biofilm formation on inert surfaces. Our results allow us to identify the Cpx signal transduction pathway as an important partner in the adherence process beside the EnvZ-OmpR pathway.

2Materials and methods

2.1Bacterial strains, plasmids and media

Bacterial strains and plasmids used in this study are listed in Table 1. The bacteria were grown at 30°C in complete Luria-Bertani (LB) medium or in minimal M63 medium supplemented with mannitol (0.2%) as a carbon source. Antibiotics were used at the following concentrations: ampicillin, 100 μg ml−1; chloramphenicol, 20 μg ml−1; kanamycin, 25 μg ml−1; tetracycline, 10 μg ml−1.

Table 1. E. coli K12 strains and plasmids used
  1. aompR234 was moved in CLC280 [11] by the linked transposon malA::kan and the presence of the ompR234 allele was detected by CFA red colony screening.

StrainDescriptionSource or reference
D355F Llac3350 galK2 galT22 recD1014 rpl179 (rrnD-rrnE)Laboratory collection
MC4100araD139, Δ(argF-lac)U169, rpsL150, relA1, flbB5301, deoC1, ptsF25, rbsRLaboratory collection
PHL644MC4100 malA-kan ompR234[1]
PHL696PHL644 cpxA::lacZ (Mu dX)This study
PHL744MC4100 malT::Tn10 ompR234[1]
PHL852PHL744 csgA::uidA-kanThis study
PHL881E. coli isolated from a percutaneous trans-hepatic catheter, 71-years old female patientEdouard Herriot Hospital (Lyon)
PHL885E. coli isolated from a urethral catheter, 64-years old female patientEdouard Herriot Hospital (Lyon)
PHL904PHL644 cpxA::camThis study
PHL907PHL644 cpxR::spcThis study
PHL910MC4100 malA-kan ompR234 cpxA101aThis study
PHL911PHL881 cpxA::camThis study
PHL912PHL885 cpxA::camThis study
PHL919MC4100malT54::Tn10 ompR234 cpxA::cam csgA::uidA-kanThis study
PHL920MC4100 malT54::Tn10 ompR234 cpxA::lacZ (Mu dX)csgA::uidA-kanThis study
PHL972PHL907 csgA::uidA-kanThis study
PHL988PHL910 csgA::uidA-kanThis study
PHL1049MC4100 malA-kan ompR234 cpxA::cam Δ(pta-ackA) zej223::Tn10This study
PHL1056PHL644 ara74::cam/pBAD18This study
PHL1057PHL644 ara74::cam/pND18This study
PHL1058PHL852 ara74::cam csgA::uidA-kan/pBAD18This study
PHL1059PHL852 ara74::cm csgA::uidA-kan/pND18This study
pCSG4pUC19 with 3.5-kb HindIII fragment carrying the csgBA operon[6]
pKKcpxRpKK232 with 0.7-kb NcoI-HindIII fragment carrying the PCR-amplified MC4100 cpxR gene (primers: 5′-TATTTAAACCATGGATAAAATC-3′, 5′-CTATCATGAAGCTTAAACCATC-3′)This study
pND10cpxR gene in pAMPTs[11]
pND18nlpE gene in pBAD18[11]
pNoa70pBR322 with a 20-kb BamHI fragment harboring a genomic fragment (12 kb) with the truncated cpxA gene fusioned to the lacZ gene (8 kb) from the Mu dXThis study
pOK101cpxA gene in pINIII[16]
pOV874pUC19 carrying uidA-kan fusion inserted at the unique ClaI site of the csgA geneThis study

2.2Genetic methods

Random mutagenesis with phage Mu dX was performed as described by Baker et al. [7]. Phage P1 vir was used for transductions, which were carried out following the procedure described by Miller [8]. The pta-ackA deletion was moved as described in [11]. The 7.25-kb PvuII fragment containing the csgA-uidA fusion of pOV874 [1] was used to transform the recD csgA+ D355 strain. White colonies were selected on CFA medium and the fusion was then moved by P1 vir transduction in different genetic backgrounds.

2.3Visualization and quantification of biofilm formation

Cells were grown at 30°C in 24- or 96-well polystyrene plates containing respectively 2 or 0.2 ml of M63 minimal medium (mannitol 0.2%) in each well. Biofilm formation was visualized after 24 h of culturing as follows: planktonic cells were discarded and the biofilm which had developed on the bottom of the plate was washed twice with M63 before drying for 1 h at 80°C. A drop of crystal violet was added for a few min in each well followed by extensive washes with M63, this allowed us to visualize the surface-attached bacteria. The thickness of biofilm was quantified in a 24-well plate. For each well, the two washes were pooled with the initial supernatant and referred to as swimming cells. The biofilm was recovered in 1 ml of M63 by scraping and pipetting up and down. The number of surface-attached and swimming bacteria was estimated from the optical density at 600 nm (OD600) to give the adherence percentage corresponding to each bacterial strain. A minimum of three independent assays were performed and averages were calculated.

2.4Enzyme assay

β-Glucuronidase activity was determined spectrophotometrically [9] on toluenized cultures after different times of growth and expressed as U mg−1 of bacterial dry weight, where U=nmol of product liberated per min. A minimum of four independent assays were performed. A representative experiment is shown for kinetics and the mean value with S.D. in other cases.

3Results and discussion

3.1Isolation of a Mu dX mutant that is defective in adherence to polystyrene

Mu dX phage insertion in the proper orientation in a transcriptional unit generates a fusion with the lacZ reporter gene and a mutation in the interrupted gene. A Mu dX phage stock was used to mutagenize the adherent MC4100 ompR234 strain (PHL644). Mutants were selected on ampicillin and chloramphenicol and screened for biofilm formation ability at 30°C in 96-well plates. After 24 h of culturing without agitation, each plate was inverted several times to remove the supernatant, rinsed twice and dried before crystal violet staining. The non-adherent phenotype of the mutants isolated from this first screening was then confirmed in 24-well plates and quantified. PHL696 shows a reduced ability to adhere when compared to PHL644 (Fig. 1A).

Figure 1.

The biofilm formation phenotype of cpxA mutants on polystyrene. After 24 h of culturing in mannitol containing M63 medium (2 g l−1) at 30°C in 24-well plates, biofilms were visualized by crystal violet staining. From a replicated plate, the adherence percentage corresponding to each bacterial strain was determined as described in Section 2. The total growth in each well reached 1.5 OD600 U. (A) MC4100 strains of E. coli. (B) Medical isolates: PHL881 and PHL885 were isolated respectively from percutaneous trans-hepatic and urethral catheters.

3.2Identification of the cpxA gene as the integration site of Mu dX in strain PHL696

The chromosomal DNA of the mutant was extracted, digested with BamHI and cloned in pBR322. We took advantage of the lacZ presence in Mu dX to recover the 12-kb adjacent genomic fragment in pNoa70 after plate screening in the presence of X-Gal. This DNA fragment was partially sequenced from the Mu S end after subcloning in pBC (Stratagene). From this, we were able to conclude that in PHL696, the Mu dX has been inserted in the codon corresponding to the 168th amino acid of the cpxA gene product. This gene encodes an inner membrane sensor of 458 amino acids with two membrane-spanning domains, a periplasmic sensor domain and a catalytic cytoplasmic domain [10]. In PHL696, the insertion is located in the beginning of the second transmembrane domain coding region. To confirm the negative effect of the cpxA mutation on biofilm formation, we tested the adherence properties of another cpxA mutant in a ompR234 background (PHL904). In this strain, the cpxA gene is disrupted by a chloramphenicol resistance gene in the codon corresponding to the 192nd amino acid of the protein at the end of the second transmembrane domain [11]. PHL904 has a poor capacity to form biofilm and reproduced the phenotype of PHL696. We were able to restore its full adherence properties by transformation with the plasmid pOK101 carrying the wild-type cpxA gene (Fig. 1A). Therefore, the Mu dX or chloramphenicol resistance gene insertion in the cpxA gene is responsible for the adherence defect to polystyrene.

To ensure that this cpxA effect on biofilm formation is a general phenomenon, we introduced by transduction the cpxA mutation in two E. coli medical isolates: PHL881 was isolated from the percutaneous trans-hepatic catheter of a patient with cholecystisis and PHL885 was isolated from the urine of a patient with a urethral catheter-related infection. In these two strains, the cpxA mutation causes an important polystyrene adherence defect, showing that the cpxA gene is a key adherence factor for E. coli species (Fig. 1B). It is noticeable that Salmonella typhi mutated in the homologous cpxA gene was found to be defective in epithelial cell adherence [12]. CpxA could be an adherence key sensor widely used in enterobacteria.

3.3Involvement of cpxA in adherence via regulation of expression of the curlin gene

Adherence of both E. coli K12 and medical isolates depends on curli production [1]. In order to address the possible regulation of the curli operon by cpxA, we used a csgA::uidA transcriptional fusion which enables us to estimate the synthesis rate of curlin in different cpxA backgrounds. The cpxA insertion mutation decreased csgA::uidA transcription approximately 8-fold in minimal medium (Fig. 2A). Similar results were obtained with the cpxA::lacZ (Mu dX) mutation (data not shown). Transformation of cpxA mutant PHL919 with the plasmid pOK101 carrying the wild-type cpxA gene allowed us to restore the same expression level as in the cpxA+ strain (PHL852) and even more during the early exponential growth phase. This last effect probably results from the larger amount of CpxA produced from the multicopy plasmid pBR322. Therefore, the adherence defect in minimal medium of cpxA mutant is likely to result from a drastic reduction of csgA expression. If this was the case, we would predict that providing csgBA genes on a multicopy plasmid in a cpxA mutant would restore the wild-type adherence phenotype. Indeed, cpxA strain PHL904 harboring the pCSG4 plasmid gives 54.7±4.0% of adherent cells versus 23.8±3.3% in the control strain PHL904 carrying the pUC19. This result strengthens our hypothesis.

Figure 2.

Medium dependent inhibition of the transcription of the curlin gene by cpxA mutations. Comparison of the csgA::uidA expression level in wt and cpxA mutants. (A) Growth in minimal medium. Cells were grown in mannitol containing M63 medium (2 g l−1) in a shaker at 30°C. Filled circle: PHL852 (cpxA+), filled triangle: PHL919 (cpxA::cm), open triangle: PHL919+pOK101 (cpxA+). The cpxA::lacZ (MudX) mutant gives the same csgA::uidA expression pattern as the PHL919 strain. (B) Growth in LB. Same symbols as above and open square: PHL920 (cpxA::lacZ (MudX)).

3.4The adherence defect observed in cpxA mutants depends on accumulation of CpxR-P

CpxA is the sensor protein of a two-component signal transduction pathway [13] with CpxR as cognate response regulator. CpxR activates extracytoplasmic stress-induced genes such as degP, dsbA, rotA, cpxP, etc. [11,14,15]. Target gene activation depends on the phosphorylation degree of the CpxR protein regulated by the phosphatase kinase activities of CpxA [16]. To determine the role of CpxR in biofilm formation, different mutants were constructed. Fig. 4A shows that the non-adherent phenotype is not observed in absence of the cpxR gene. In a similar way, Danese et al. [11] observed a ‘phenotype’ (degP fusion activation) specifically in cpxA mutant but not in cpxR mutant. These authors have shown that in the cpxA mutant, CpxR-P accumulates via phosphorylation by acetyl phosphate, resulting in transcriptional activation of the degP fusion. This phenomenon occurs only in glucose-supplemented minimal medium but not in LB. The non-adherent phenotype of the cpxA mutants grown in mannitol-supplemented M63 could result from the same process. To test this hypothesis, we have first checked the csgA-uidA fusion expression in LB. Curlin gene transcription is indeed not affected in cpxA mutants grown in rich media (Fig. 2B). Moreover, deletion of the two genes responsible for acetyl phosphate synthesis abolishes the non-adherent phenotype (Fig. 3). These results show that bacterial adherence depends on phosphorylation of CpxR by acetyl phosphate in the cpxA mutant.

Figure 4.

The cpxA101 mutation confers both the non-adherent phenotype and reduced curlin gene transcription. Strains were grown in M63 media with 0.2% mannitol as a carbon source. (A) Biofilm formation estimated from the adherence percentage (see Section 2). The total growth in each well reached 1.5 OD600 U. (B) Curlin gene fusion expression (OD600=1.5).

Figure 3.

The non-adherent phenotype results from CpxR phosphorylation by acetyl phosphate in the absence of CpxA. Strains were grown in M63 media with 0.2% mannitol as a carbon source. The total growth in each well reached 1.5 OD600 U.

We introduced the ompR234 mutation in the cpxA101 mutant lacking phosphatase activity. Strains carrying this mutation accumulate phosphorylated CpxR. Strain PHL910, as well as the cpxA mutant, loses the ability to form thick biofilms (Figs. 1A and 4A). We conclude that the high phosphorylation level of CpxR inhibits the adherence process. To test if this inhibition occurs via curlin gene transcription, we assayed the csgA-uidA fusion in different genetic backgrounds. Curlin fusion expression is indeed extremely reduced in the cpxA101 mutant when it slightly increases in the cpxR mutant (Fig. 4B), suggesting that CpxR-P could be responsible for the repression of the csgA-uidA expression. According to this hypothesis, presence of cpxR on a multicopy plasmid leads to a decrease of the curlin gene fusion expression (data not shown). Our results clearly demonstrate that Cpx pathway activation has a negative effect in the process of biofilm formation resulting in repression of curlin gene transcription. To test the physiological importance of this phenomenon, we have investigated the adherence properties and the curlin gene fusion expression in bacteria overproducing the NlpE protein. This outer membrane protein is known to activate the Cpx pathway when synthesized from a multicopy plasmid such as pND18 [11]. Fig. 5 shows as expected that activation of the Cpx pathway by NlpE causes an adherence defect and a decrease of the csgA-uidA gene fusion expression. Expression of P-pilus subunits in the absence of its own chaperone PapD activates the Cpx pathway that upregulates protease and chaperone proteins [19]. Soto and Hultgren [20] proposed that in response to bacterial attachment, pilus subunits accumulate in the periplasm, triggering Cpx system activation that in turn facilitates pilus biogenesis. Our results could be included in this model as a clue, indicating that beyond degradation and refolding of unassembled subunits, activation of the Cpx pathway following the attachment could directly inhibit the curli biogenesis.

Figure 5.

Cpx pathway induction by NlpE leads to both reduced adherence and reduced transcription of the curlin gene. Strains transformed with pND18 overexpress nlpE. pBAD18 is the control for pND18. Strains were grown in M63 media containing 50μg ml−1 ampicillin, 0.2% mannitol as a carbon source and arabinose 0.4% as inductor. (A) Biofilm formation estimated from the adherence percentage (see Section 2). The total growth in each well reached 1 OD600 U. (B) Curlin gene fusion expression.

Acknowledgements

We thank Thomas Silhavy and members of his lab for gifts of strains. We thank Sylvie Reverchon and Guy Condemine for helpfull discussions. This work was partially supported by a Grant from the French Defense Ministry (96-048/DRET).

Ancillary