SEARCH

SEARCH BY CITATION

Keywords:

  • c-di-GMP;
  • diguanylate cyclase;
  • phosphodiesterase;
  • biofilm formation;
  • adhesion factors;
  • curli;
  • oxygen sensing

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. References
  8. Supporting Information

In bacteria, intracellular amounts of the signal molecule cyclic di-GMP (c-di-GMP) are determined by biosynthetic enzymes, or diguanylate cyclases (DGCs), and degradative enzymes, or c-di-GMP phosphodiesterases (c-PDEs). In Escherichia coli, the production of curli fibers, an important adhesion factor, responds to c-di-GMP. The yddV-dos operon, which encodes a DGC and a c-PDE acting as a protein complex, is highly expressed at a low growth temperature and in the stationary phase, i.e. conditions that also stimulate curli production. We show that perturbations in the balance between YddV and Dos, obtained either through inactivation of the yddV gene or through overproduction of either YddV or Dos, strongly affect curli production. Both YddV and Dos proteins regulate the transcription of the csgBAC operon, which encodes curli structural subunits, while not affecting the expression of the regulatory operon csgDEFG. Consistent with the role of both YddV and Dos proteins as oxygen sensors, their effects on csgBAC gene expression were dramatically reduced in cells grown under anoxic conditions. Our results show that the yddV-dos operon plays an important role in the expression of curli-encoding genes in aerobically growing E. coli, and suggest that YddV and Dos, through their opposite activities, might finely tune curli production in response to oxygen availability.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. References
  8. Supporting Information

Most bacteria are able to switch between two different ‘lifestyles’: single cells (planktonic mode) and biofilm, i.e. a sessile microbial community. Biofilm and planktonic cells can significantly differ in their physiology, morphology and gene expression pattern. An extensive description of the molecular mechanisms regulating biofilm-related cellular processes can be found in a recent review by Karatan & Watnick (2009). Biofilms are characterized by a lower susceptibility to antibiotics compared with planktonic cells (Anderl et al., 2000; Harrison et al., 2009). The tolerance of biofilm cells to antibiotics is mediated by different mechanisms, such as reduced penetration into bacterial cells and induction of a dormant physiological state, poorly sensitive to antibiotics, in a significant fraction of the biofilm population (reviewed in Lewis, 2008). In addition to tolerance to antibiotics, biofilms show reduced sensitivity to the host immune defenses. These properties contribute to the chronic nature of most biofilm-mediated infections. Finally, some biofilm determinants can also act as virulence factors: for instance, the production of fimbrial structures, such as curli fibers or type 1 fimbriae, allows the adherence of Escherichia coli of eukaryotic cells and promotes its internalization (Gophna et al., 2001; Wright et al., 2007).

The transition from planktonic cells to biofilm is regulated by environmental and physiological cues, relayed to the bacterial cell by signal molecules such as cyclic di-GMP (c-di-GMP). Intracellular levels of c-di-GMP are regulated by two classes of isoenzymes: diguanylate cyclases (DGCs, c-di-GMP biosynthetic enzymes), also termed GGDEF proteins from the conserved amino acid sequence in their catalytic site, and c-di-GMP phosphodiesterases (c-PDEs), which degrade c-di-GMP (Cotter & Stibitz, 2007). While in Gram-negative bacteria genes encoding DGC and c-PDE proteins are present in high numbers, they are much less conserved in Gram-positive bacteria (Galperin, 2004), where c-di-GMP does not appear to play a significant role in biofilm-related cell processes (Holland et al., 2008). In contrast, c-di-GMP promotes biofilm formation in Gram-negative bacteria by stimulating the production of adhesion factors, while repressing cell motility (Kader et al., 2006; Méndez-Ortiz et al., 2006; Weber et al., 2006; Jonas et al., 2008), by acting on c-di-GMP-responsive proteins (reviewed in Hengge, 2009). In addition, c-di-GMP can directly affect virulence factor production in pathogenic bacteria (Kulasakara et al., 2006; Hammer & Bassler, 2009). In E. coli and other enterobacteria, the production of curli fibers and cellulose, which form an extracellular matrix promoting biofilm formation (Römling et al., 1998a; Prigent-Combaret et al., 2000; Gualdi et al., 2008), is strongly stimulated by c-di-GMP (Zogaj et al., 2001; Kader et al., 2006; Weber et al., 2006). At least six different genes encoding c-di-GMP-related proteins are involved in curli gene regulation (Sommerfeldt et al., 2009). The expression of several DGC-encoding genes, as well as curli-encoding genes, is controlled by σS, an alternative σ factor mainly active under slow growth conditions and in response to cellular stresses. The yddV-dos operon is the most expressed among σS-dependent genes encoding enzymes related to c-di-GMP metabolism (Weber et al., 2006; Sommerfeldt et al., 2009). The yddV-dos operon encodes, respectively, a protein with DGC activity and a c-PDE that can degrade c-di-GMP to pGpG (the noncyclic form of diguanylic acid), not known to function as a signal molecule (Schmidt et al., 2005). Dos stands for direct oxygen sensor, because the Dos protein is complexed to a heme prosthetic group that can bind O2, CO and nitric oxide (NO) (Delgado-Nixon et al., 2000). A recent publication (Tuckerman et al., 2009) has reported that YddV is also a heme-binding oxygen sensor, and that YddV and Dos interact to form a stable protein complex. Although it has been reported that YddV overexpression can stimulate biofilm formation (Méndez-Ortiz et al., 2006), the targets of yddV-dependent biofilm induction have not yet been identified. In this work, we have investigated the role of the yddV-dos operon in the regulation of curli production.

Materials and methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. References
  8. Supporting Information

Bacterial strains and growth conditions

The bacterial strains used in this work are listed in Table 1. Escherichia coli MG1655 mutant derivatives were constructed using either the λ Red technique (Datsenko & Wanner, 2000) or by bacteriophage P1 transduction (Miller, 1972). The primers used for gene inactivation and for confirmation of target gene disruption by PCR are listed in Supporting Information, Table S1. Bacteria were grown in M9Glu/sup medium (Brombacher et al., 2006), a glucose-based medium supplemented with 2.5% Luria–Bertani medium as a source of amino acids and vitamins. For growth under anoxic conditions, liquid cultures were grown with no shaking in 12-mL glass tubes filled to the top; these conditions are sufficient for the full induction of genes responding to anaerobiosis (Landini et al., 1994). Antibiotics were used at the following concentrations: ampicillin, 100 μg mL−1; chloramphenicol, 50 μg mL−1; tetracycline, 25 μg mL−1; and kanamycin, 50 μg mL−1. For Congo red-binding assays, bacteria grown overnight in a microtiter plate were spotted, using a replicator, on Congo red-supplemented medium (CR medium), composed of 1.5% agar, 1% Casamino acids, 0.15% yeast extract, 0.005% MgSO4 and 0.0005% MnCl2, to which 0.004% Congo red and 0.002% Coomassie blue were added after autoclaving. Bacteria were grown for 20 h at 30 °C; phenotypes were better detectable after 24–48 h of additional incubation at 4 °C. Surface adhesion assays in polystyrene microtiter plates were performed as described (Dorel et al., 1999).

Table 1. Escherichia coli strains and plasmids used in this work
Escherichia coliRelevant genotype or characteristicsReferences or sources
Strains
 MG1655K-12, Fλrph-1Blattner et al. (1997)
 EB1.3MG1655 rpoStetPrigent-Combaret et al. (2001)
 PHL856MG1655 csgA-uidAkanGualdi et al. (2008)
 AM75MG1655 ΔcsgDcatThis work
 AM95MG1655 yddVΔ9311383catThis work
Plasmids
 pGEM-T EasyControl vector allowing direct cloning of PCR products, ampicillin resistancePromega
 pGEM-YddVWTThe yddV gene cloned as a PCR product into the pGEM-T Easy vectorThis work
 pGEM-YddVGGAAFyddV allele carrying the mutation resulting in the GGDEF[RIGHTWARDS ARROW]GGAAF change in the YddV DGC catalytic siteThis work
 pGEM-DosWTThe dos gene cloned as a PCR product into the pGEM-T Easy vectorThis work
 pGEM-DosAAAdos allele carrying the mutation resulting in the EAL[RIGHTWARDS ARROW]AAA change in the Dos c-PDE catalytic siteThis work

Plasmid construction

The plasmids used in this work are listed in Table 1. For the construction of the pGEM-T Easy (http://www.promega.com/tbs/tm042/tm042.pdf) plasmid derivatives, either the yddV or the dos genes were amplified by PCR from E. coli MG1655 genomic DNA and the PCR product was cloned into the plasmid, producing pGEM-YddVWT and pGEM-DosWT, respectively. The pGEM-YddVGGAAF and pGEM-DosAAA plasmids were obtained by three-step PCR mutagenesis (Li & Shapiro, 1993) using the primers listed in Table S1. All constructs were verified by sequencing.

Gene expression studies

Quantitative real-time PCR (qRT-PCR) for the determination of the relative expression levels was performed on cultures grown at 30 °C in M9Glu/sup medium. Samples were taken in the early (OD600 nm=0.25) and late (OD600 nm=0.7) exponential phase and in the stationary phase (OD600 nm≈2.5) for cultures grown aerobically, and in the stationary phase (OD600 nm≈1.6) for cultures grown under anoxic conditions. RNA extraction, reverse transcription and cDNA amplification steps were performed as described (Gualdi et al., 2007), using 16S RNA as the reference gene.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. References
  8. Supporting Information

Partial deletion of the yddV and dos genes

We investigated the possibility that the yddV-dos operon might affect the production of curli fibers, a major biofilm determinant in E. coli (Prigent-Combaret et al., 2000, 2001), by constructing two mutants in either the yddV or the dos genes. In order to evaluate more precisely the contribution of c-di-GMP synthesis and turnover toward YddV and Dos protein activities, our mutagenesis strategy targeted exclusively the region of the gene encoding the domains involved in c-di-GMP metabolism, allowing the production of truncated YddV and Dos proteins carrying functional heme-binding and sensor domains. Because yddV and dos are part of the same transcriptional unit (Méndez-Ortiz et al., 2006), insertions of antibiotic resistance cassettes into the yddV gene can result in transcription termination, thus preventing dos transcription. However, in the AM95 (yddVΔ9311383cat) mutant, replacement of the distal part of the yddV gene by the chloramphenicol acetyl-transferase (cat) gene, placed in the same orientation, results in semi-constitutive transcription of the dos gene from the cat promoter, as determined by qRT-PCR (data not shown). Because YddV and Dos constitute a highly expressed protein complex possessing both DGC and PDE activity (Sommerfeldt et al., 2009; Tuckerman et al., 2009), the production of truncated forms of either YddV or Dos should result in the formation of mutant YddV–Dos protein complexes unbalanced either towards accumulation or towards degradation of c-di-GMP. However, we found that mutants in the dos gene showed phenotypic instability at the level of cell aggregation in liquid culture and Congo red binding, suggesting that the dos mutant strain might accumulate spontaneous mutations suppressing the dos defect. Thus, the dos mutant strain was not investigated any further, and we focused on the yddV mutant AM95 and on MG1655 derivatives overexpressing either the YddV or the Dos proteins from multicopy plasmids.

Effects of the yddV and dos mutations on Congo red binding and biofilm formation

To determine the possible effects of mutations in the yddV gene on curli production, we performed Congo red-binding assays using CR medium. Curli fibers bind with very Congo red high affinity, due to their β-amyloid structure (Olsén et al., 1989; Chapman et al., 2002). Congo red can bind, albeit with a lower affinity, other cell surface-exposed structures, such as the extracellular polysaccharides cellulose and poly-N-acetylglucosamine (Jones et al., 1999; Zogaj et al., 2001); however, in E. coli MG1655, due to the low production of extracellular polysaccharides, the red phenotype on CR medium is totally dependent on curli production (Gualdi et al., 2008). Indeed, a mutant carrying a null mutation in the csgA gene, encoding the main curli structural subunit, displays a white phenotype on CR medium (Fig. 1a). The yddVΔ9311383cat mutation resulted in a clear, albeit partial, loss of the red phenotype on CR medium, indicative of a reduction in curli production. To further confirm the effects of the mutation in the yddV gene, we cloned either the yddV or the dos genes into the pGEM-T Easy vector, under the control of the lac promoter, producing the pGEM-YddVWT and pGEM-DosWT plasmids (Table 1). In addition, we constructed plasmids carrying mutant alleles of either gene (pGEM-YddVGGAAF and pGEM-DosAAA, Table 1), in which the coding sequence for the amino acids responsible for either DGC activity (in the YddV protein) or c-PDE activity (in the Dos protein) had been altered. The substitution of GGDEF motif into the DGC catalytic site to GGAAF results in a drastic loss (>90%) of DGC activity (De et al., 2008; Antoniani et al., 2010). In the Dos protein, the glutamic acid and leucine in the EAL motif were changed to alanine residues, giving rise to the DosAAA mutant; mutations affecting the EAL motif abolish c-PDE activity (Kirillina et al., 2004; Bassis & Visick, 2010). Transformation of the yddV mutant AM95 strain with pGEM-YddVWT, but not with pGEM-YddVGGAAF, restored the red phenotype on CR medium (Fig. 1b), indicating that YddV can affect the CR phenotype in a manner dependent on its DGC activity. Transformation of MG1655 with the pGEM-DosWT plasmid (Fig. 1c) resulted in a white CR phenotype, consistent with a negative role of Dos in curli production. In contrast, no effects were observed on the CR phenotype in the MG1655 strain harboring the pGEM-DosAAA plasmid, carrying the mutant Dos protein impaired in its c-PDE activity.

image

Figure 1.  (a) Congo red phenotype of MG1655 (WT), PHL856 (csgA) and AM95 (yddV). (b) AM95 (yddV) strain transformed with either pGEM-YddVWT or pGEM-YddVGGAAF. (c) MG1655 strain transformed with either pGEM-DosWT or pGEM-DosAAA.

Download figure to PowerPoint

In E. coli MG1655, curli fibers are the main determinant for adhesion to abiotic surfaces (Prigent-Combaret et al., 2000). Thus, we confirmed the results of Congo red-binding assays by biofilm formation experiments on polystyrene microtiter plates (Fig. 2). Consistent with the pivotal role of curli in adhesion to abiotic surfaces, biofilm formation on microtiter plates was reduced by about 10-fold by the inactivation of the csgA gene, encoding the major curli subunit (Fig. 2), as well as by growth at 37 °C (data not shown), the temperature at which curli fibers are not produced in most enterobacteria (Römling et al., 1998a). Inactivation of the yddV gene resulted in a c. 3.5-fold reduction in biofilm formation. Overexpression of YddVWT, but not of the YddVGGAAF protein, results in strong biofilm stimulation, in agreement with CR phenotypes (Fig. 1b). Overexpression of the Dos protein mimicked the effects of the yddV mutation, resulting in decreased biofilm production; however, no effect was detected for overexpression of the Dos mutant protein impaired in c-PDE activity (Fig. 2). Thus, the results of Congo red binding studies and biofilm formation experiments strongly support the hypothesis that the YddV and Dos proteins control curli production through the modulation of intracellular c-di-GMP concentrations.

image

Figure 2.  Surface adhesion on polystyrene microtiter plates by strains: MG1655 (WT), PHL856 (csgA), AM95 (yddV) and MG1655 transformed with pGEM-YddVWT, pGEM-YddVGGAAF, pGEM-DosWT and pGEM-DosAAA. The relative adhesion value was set to 1 for MG1655 (indicated by the dotted line); the actual adhesion unit value for MG1655 was 3.1. Results are the average of three independent experiments, and SEs are shown.

Download figure to PowerPoint

Effect of the yddV and dos mutations on curli gene expression

The regulation of adhesion factors' production by DGCs can take place at different levels, such as allosteric activation, as in the stimulation of cellulose biosynthesis by AdrA (Zogaj et al., 2001), or gene regulation, such as in the transcription regulation of the csgDEFG operon by YdaM and YegE (Sommerfeldt et al., 2009). We tested the possibility that the yddV gene might affect the CR phenotype and adhesion to polystyrene through gene expression regulation of the curli-encoding operons. Curli production and assembly is mediated by two divergent operons; csgDEFG encodes the transport and assembly proteins and the CsgD regulator, which in turn activates the csgBAC operon, encoding curli structural subunits (Römling et al., 1998b). Because curli genes are subject to growth phase-dependent regulation mediated by the rpoS gene (Römling et al., 1998b), we assessed the effects of the yddV mutation at different growth stages: early exponential phase (OD600 nm=0.25), late exponential phase (OD600 nm=0.7) and stationary phase (overnight cultures, OD600 nm≥2.5). Transcription levels of the csgB and csgD genes in M9Glu/sup medium at 30 °C were determined by qRT-PCR (Table 2). Interestingly, the expression of csgD and csgB follows different kinetics: while csgB is only induced in the late stationary phase, csgD transcription levels are very similar both in the exponential and in the stationary phase. A different timing between csgD and csgB transcription in E. coli MG1655 has already been reported (Prigent-Combaret et al., 2001). Although the lack of stationary-phase-dependent-activation of the csgD gene might appear to be surprising, rpoS-dependent gene expression during the exponential phase is rather common (Dong et al., 2008); indeed, the expression of both csgB and csgD is totally abolished in the rpoS-deficient EB1.3 mutant derivative of MG1655 (data not shown). yddV inactivation caused a drastic decrease in csgB expression (c. 400-fold reduction, Table 2), while showed a much more reduced effect on csgD transcription (c. 2.5-fold), suggesting that the YddV protein specifically regulates the transcription of the csgBAC operon. Overexpression of either the YddV or the Dos protein confirmed this result, showing csgBAC upregulation by YddV and downregulation by Dos, in a manner dependent on their DGC and c-PDE activities, respectively (Table 3). The observation that YddV regulates csgBAC transcription, which is also dependent on the CsgD protein, may suggest that c-di-GMP synthesis by YddV might trigger CsgD activity as a transcription regulator. To test this hypothesis, we studied the effect of the yddV mutation on the expression of adrA, a CsgD-dependent gene involved in the regulation of cellulose production (Zogaj et al., 2001): as shown in Table 2, adrA transcript levels were not significantly affected by yddV inactivation, suggesting that the CsgD protein can function as a transcription activator in the yddV mutant strain AM95.

Table 2.   Relative expression of csgB and csgD genes in MG1655 vs. AM95 (yddVcat)
GenescsgBcsgDadrA
StrainsMG1655 (WT)AM95 (yddVcat)MG1655 (WT)AM95 (yddVcat)MG1655 (WT)AM95 (yddVcat)
  • *

    ΔCt between the gene of interest and the 16S gene was arbitrarily set at 1 for MG1655 in the early exponential growth phase (OD600 nm=0.25) for csgB and csgD genes, and in the stationary phase for adrA. The actual ΔCt values were: csgD=15.0; csgB=21.7; adrA=22.4. ΔCt between the gene of interest and the 16S gene for different growth phases and for mutant strains are expressed as relative values. Values are the average of two independent experiments performed in duplicate. ND, not determined.

Growth conditions
 Early exponential (OD600 nm=0.25)1*0.71*0.6NDND
 Late exponential (OD600 nm=0.7)0.80.91.50.7NDND
 Stationary (OD600 nm≥2.5)3910.91.40.61*0.74
 Stationary, anoxic (OD600 nm≥1.6)57.222.41.61.4NDND
Table 3.   Relative expression of csgB and csgD genes in response to either YddV or Dos overexpression
StrainscsgB expression (aerobic conditions)csgB expression (anoxic conditions)csgD expression (aerobic conditions)
  • *

    ΔCt between the gene of interest and the 16S gene was arbitrarily set at 1 for MG1655/pGEM-T under aerobic conditions. Actual ΔCt values in MG1655/pGEM-T: csgB=15.9; csgD=14.6. Values are the average of two independent experiments performed in duplicate.

MG1655/pGEM-T1*0.381*
MG1655/pYddVWT31.20.322.1
MG1655/pYddVGGAAF2.30.451.6
MG1655/pDosWT0.060.34ND
MG1655/pDosAAA1.040.37ND

Both the YddV and the Dos protein require binding of their heme prosthetic groups to O2, or alternatively to NO, in order to trigger either DGC or c-PDE activity (Taguchi et al., 2004; Tuckerman et al., 2009). Thus, we measured csgB and csgD expression levels in bacteria grown in oxygen limitation, comparing MG1655 with its yddVΔ9311383cat mutant derivative. Growth under anoxic conditions did not affect csgD transcript levels, while reducing csgB expression by c. 7-fold; yddV inactivation resulted only in a c. 2.5-fold reduction in csgB transcript levels, vs. the c. 400-fold reduction in aerobic growth (Table 2), suggesting that YddV-dependent regulation of the csgBAC operon is bypassed under oxygen-limiting conditions. Consistent with this observation, no effect on csgBAC expression by either YddV or Dos overexpression could be detected in MG1655 grown in oxygen limitation (Table 3).

Growth-phase dependent regulation of the yddV-dos operon

Our results clearly indicate that a functional yddV gene is required for csgBAC, but not csgDEFG, expression (Table 2), suggesting that the YddV protein acts downstream of CsgD in the regulatory cascade leading to curli production. It is thus possible that the CsgD protein might activate the transcription of the yddV-dos operon and, in turn, YddV might trigger csgBAC expression in the stationary phase of growth. However, cotranscription of the yddV and the dos genes also raises the question of how the opposite activities of the YddV and Dos proteins are modulated. We investigated the possibility that the yddV-dos transcript might be processed in the stationary phase of growth, resulting in the accumulation of the YddV protein, with consequent activation of csgBAC expression. To address these questions, we determined both yddV and dos transcripts at different growth stages, and we tested the possible dependence of yddV-dos transcription on the CsgD protein by comparing MG1655 with its csgD mutant derivative AM75. In addition, because transcription of the yddV-dos operon is controlled by the rpoS gene (Weber et al., 2006; Sommerfeldt et al., 2009), which also regulates curli-encoding genes (Römling et al., 1998b), we also determined gene expression kinetics of the yddV-dos operon in the rpoS mutant derivative EB1.3. As shown in Fig. 3, transcription of the yddV gene was induced in an rpoS-dependent manner in the late exponential phase, reaching maximal induction in overnight cultures; in contrast, csgD inactivation did not affect yddV expression. Transcription of the yddV and of the dos genes followed a very similar pattern (Fig. 3) and the overall ratio between yddV and dos transcripts remained constant in different growth phases, suggesting that neither yddV nor dos is subject to specific regulation at the level of mRNA processing, at least under the conditions tested.

image

Figure 3.  Relative expression levels of the yddV gene in strains MG1655 (WT), EB1.3 (rpoS) and AM75 (csgD), and of the dos gene in strain MG1655, as measured by real-time PCR experiments. Expression values in MG1655 in the early exponential growth phase (OD600 nm=0.25; light gray bars) (corresponding to a ΔCt relative to 16S rRNA=16.3 for yddV, and 15.8 for dos) were set to 1. The other samples were taken in late exponential phase (OD600 nm=0.7; gray bars) and stationary phase (OD600 nm≥2.5; black bars). Data are the average of two independent experiments, each performed in triplicate, with very similar results.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. References
  8. Supporting Information

In this work, we have shown that the production of curli fibers, a major adhesion factor in E. coli, is affected by the yddV-dos operon, encoding a YddV–Dos protein complex involved in c-di-GMP biosynthesis and turnover. Control of curli production by yddV-dos takes place at the level of transcription regulation of the csgBAC operon, encoding curli structural subunits (Table 2), and is mediated by the DGC and PDE activities of YddV and Dos (Fig. 1b and c; Table 3). In contrast, the YddV–Dos protein complex does not strongly influence csgDEFG expression, nor does it affect the expression of the CsgD-dependent adrA gene, encoding a positive effector for cellulose biosynthesis (Table 2). Regulation of the csgBAC operon, but not of csgDEFG, has already been described for another DGC, the product of the yeaP gene (Sommerfeldt et al., 2009). Thus, in E. coli, the production of curli and cellulose involves DGC and c-PDE proteins at various levels (summarized in Fig. 4). Indeed, csgDEFG transcription is regulated by the DGC YdaM (Weber et al., 2006) and by the PDEs YciR and YhjH (Pesavento et al., 2008), csgBAC transcription by YeaP and by the YddV–Dos complex (Sommerfeldt et al., 2009; this work), while the AdrA protein activates cellulose production (Zogaj et al., 2001). The involvement of such a large number of c-di-GMP-related proteins might depend on the need to relay different environmental signals to the activation of the csgD regulon. Indeed, curli and cellulose production responds to a variety of environmental cues, including low temperature, slow growth and low osmolarity (Römling et al., 1998b). In addition, devoted DGCs and PDEs can trigger the expression of individual CsgD-dependent genes (e.g. csgBAC) in response to specific environmental signals, thus altering the relative expression of genes belonging to the CsgD regulon and, in particular, the balance between curli and cellulose production.

image

Figure 4.  Model summarizing gene expression regulation of curli- and cellulose-related genes by DGC and c-PDE proteins. Proteins with DGC activity are indicated by open ellipses; c-PDEs are shown as dark squares. The CsgD protein, which activates both the csgBAC operon encoding curli structural subunits and adrA, encoding a DGC acting as a positive effector of cellulose biosynthesis, is indicated in the gray ellipse. See text for further details.

Download figure to PowerPoint

Depending on the prevalence of either its DGC or its c-PDE activities, the YddV–Dos complex can either activate or repress csgBAC expression. In the bacterial cell, this could be achieved by changing the relative intracellular concentrations of either YddV or Dos proteins, for instance through post-transcriptional regulation. However, our results seem to rule out the possible regulation of either yddV or dos at the level of mRNA stability (Fig. 3), suggesting that YddV and Dos might be regulated either in their protein stability level or through modulation of their enzymatic activities. Both DGC activity by YddV and c-PDE activity by Dos are inhibited in the absence of oxygen; however, YddV and Dos respond differently to oxygen concentrations, which can thus affect the overall balance between DGC and c-PDE activities in the YddV–Dos protein complex (Tuckerman et al., 2009). Thus, oxygen availability in the bacterial cell might function as an environmental signal for the modulation of intracellular c-di-GMP concentrations via the YddV–Dos complex.

Oxygen tension is known to regulate curli production in Salmonella (Gerstel & Römling, 2001). It is conceivable that oxygen sensing can be important for curli expression in relation to biofilm growth: indeed, with the exception of cells in the external layers, bacteria growing in biofilms are exposed to a gradient in oxygen availability that leads to a switch to anaerobic metabolism in the innermost biofilm layers (Borriello et al., 2004; Rani et al., 2007). The growth of E. coli MG1655 in oxygen limitation results in a sevenfold decrease in csgBAC, but not csgDEFG, expression (Table 2); under these growth conditions, overexpression of neither YddV nor Dos affects csgBAC expression (Table 3), consistent with the inhibition of both DGC and c-PDE activities of the YddV and Dos proteins in the absence of oxygen. This result would suggest that in the innermost biofilm layers, oxygen limitation might lead to reduced csgBAC expression and curli production, without, however, affecting the expression of csgDEFG and of other genes belonging to the CsgD regulon. We conclude that relay of oxygen sensing to curli production is mediated by c-di-GMP signalling involving the YddV–Dos complex.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. References
  8. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. References
  8. Supporting Information

Table S1. Primers used in this work.

Please note: Wiley-Blackwell is not responsible for the content or functionality of any supporting materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.

FilenameFormatSizeDescription
FIM_702_sm_table-s1.doc48KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.