Ferric Citrate Regulator FecR Is Translocated across the Bacterial Inner Membrane via a Unique Twin-Arginine Transport-Dependent Mechanism

Iron is essential for nearly all living organisms due to its role in metabolic processes and as a cofactor for many enzymes. The FecRI signal transduction pathway regulates citrate-mediated iron import in many Gram-negative bacteria, including Escherichia coli. The interactions of FecR with the outer membrane protein FecA and cytoplasmic anti-sigma factor FecI have been extensively studied. However, the mechanism by which FecR inserts into the membrane has not previously been reported. In this study, we demonstrate that the targeting of FecR to the cytoplasmic membrane is dependent on the Tat system. As such, FecR represents a new class of bitopic Tat-dependent membrane proteins with an internal twin-arginine signal sequence.

ABSTRACT

In Escherichia coli, citrate-mediated iron transport is a key nonheme pathway for the acquisition of iron. Binding of ferric citrate to the outer membrane protein FecA induces a signal cascade that ultimately activates the cytoplasmic sigma factor FecI, resulting in transcription of the fecABCDE ferric citrate transport genes. Central to this process is signal transduction mediated by the inner membrane protein FecR. FecR spans the inner membrane through a single transmembrane helix, which is flanked by cytoplasm- and periplasm-orientated moieties at the N and C termini. The transmembrane helix of FecR resembles a twin-arginine signal sequence, and the substitution of the paired arginine residues of the consensus motif decouples the FecR-FecI signal cascade, rendering the cells unable to activate transcription of the fec operon when grown on ferric citrate. Furthermore, the fusion of beta-lactamase C-terminal to the FecR transmembrane helix results in translocation of the C-terminal domain that is dependent on the twin-arginine translocation (Tat) system. Our findings demonstrate that FecR belongs to a select group of bitopic inner membrane proteins that contain an internal twin-arginine signal sequence.

IMPORTANCE Iron is essential for nearly all living organisms due to its role in metabolic processes and as a cofactor for many enzymes. The FecRI signal transduction pathway regulates citrate-mediated iron import in many Gram-negative bacteria, including Escherichia coli. The interactions of FecR with the outer membrane protein FecA and cytoplasmic anti-sigma factor FecI have been extensively studied. However, the mechanism by which FecR inserts into the membrane has not previously been reported. In this study, we demonstrate that the targeting of FecR to the cytoplasmic membrane is dependent on the Tat system. As such, FecR represents a new class of bitopic Tat-dependent membrane proteins with an internal twin-arginine signal sequence.

INTRODUCTION

Iron is an essential element for virtually all organisms. Despite this, the toxicity, availability, and solubility of iron present major challenges for bacteria, which require specialized iron transport systems. Escherichia coli K-12 strains encode two main nonheme pathways for the acquisition of iron from the environment. The first of these systems involves sequestering iron via the siderophores enterobactin and ferrichrome that compete for Fe³⁺ bound to host proteins (1), whereas the second system involves ferric citrate uptake (2). Ferric citrate transport is mediated via the Fec system, which is composed of a TonB-dependent outer membrane protein (FecA), a periplasmic-binding protein (FecB), and inner membrane proteins (FecCD) and ATPase (FecE) (3). The fecABCDE transport genes are induced upon detection of ferric citrate via the FecR-FecI signal cascade. FecR is an inner membrane protein with both periplasmic and cytoplasmic globular domains on either side of a single transmembrane helix (4). The binding of ferric citrate to FecA induces a conformational change, which is detected by the periplasmic C-terminal domain of FecR (Fig. 1A). The signal is transduced across the membrane, whereupon the cytoplasmic N-terminal face of FecR activates and releases the sigma factor FecI, which in turn recruits RNA polymerase to the fec operon (5, 6). Signal transduction across the inner membrane is central to this process, and a number of key residues on the periplasmic and cytoplasmic domains of FecR have been identified as essential for interactions with FecA and FecI, respectively (7, 8). However, the mechanism by which FecR is inserted into the cytoplasmic membrane has not previously been reported.

FIG 1.

(A) Schematic representation of FecR-mediated signal transduction. Binding of ferric citrate to the outer membrane protein FecA initiates a signal that is transmitted across the cytoplasmic membrane by FecR. (B) Sequence alignment of the putative internal Tat motif of FecR (residues 68 to 109) with the N-terminal Tat motif of TorA (residues 1 to 52). The Tat motif is boxed in red, and the TorA cleavage site is boxed in orange. The region of hydrophobicity/membrane-spanning region are boxed in green. Positively charged residues adjacent to the FecR c-region are indicated by arrows, and putative Sec avoidance (which corresponds to the WebLogo shown in panel D) is boxed in purple. (C) Consensus Tat sequence motif of predicted FecR eggNOG-derived orthologues, excluding the six sequences that do not contain a consensus twin-arginine motif. Amino acid positions relative to the twin-arginine motif are denoted below. (D) Sequence alignment consensus logo of the c-region of the predicted FecR eggNOG-derived orthologues. Charged amino acid residues are colored blue.

Targeting of proteins to the bacterial cytoplasmic membrane occurs via the action of secretory (Sec) machinery, YidC insertase, or the twin-arginine translocation (Tat) translocation pathway (9, 10). While most Tat substrates are soluble proteins released into the periplasm, a few substrates remain anchored in the cytoplasmic membrane, usually by an uncleaved N-terminal signal peptide or a single C-terminal transmembrane helix (11–13). Whereas Sec exported proteins fold postexport, the Tat system exports proteins that have folded and matured in the cytoplasm. Tat substrates often contain complex cofactors and may coexport bound partner proteins, or they may be utilized by bacteria in extreme environments (14–16). However, some Tat substrates also include a number of monomeric, cofactorless proteins; for example, some halophilic archaea secrete the majority of their proteins via the Tat system, which may be an adaptive response to the fast-folding kinetics of proteins in a highly saline environment (17, 18).

Substrates are addressed to the Tat pathway via N-terminal signal peptides with a distinctive tripartite structure consisting of a basic n-region containing a conserved S/T-R-R-X-F-L-K “Tat motif,” a hydrophobic h-region, and a polar c-region harboring the signal peptidase cleavage site (19). They frequently also contain one or more positive charges in the c-region that are not required for Tat transport but act as a Sec avoidance motif (20, 21). Recently, the polytopic Rieske protein of Streptomyces coelicolor was shown to be an unusual Tat substrate because it utilizes the Sec machinery for the insertion of its first two transmembrane helices and has an internal Tat signal sequence that forms the third transmembrane domain and that mediates export of the folded cofactor-containing domain across the membrane (22, 23). Since this initial study, other families of polytopic inner membrane proteins that are simultaneously targeted to the Sec and Tat pathways have recently been described (21).

Here, we demonstrate that FecR uses an internal Tat-targeting sequence for export of the 22-kDa C-terminal domain to the periplasm, while leaving a 9-kDa N-terminal domain in the cytoplasm. This is the first example of a class of bitopic Tat-dependent membrane proteins with an internal twin-arginine signal sequence.

RESULTS

An internal Tat motif is conserved among FecR orthologues.

Previous studies have shown that E. coli K-12 FecR spans the cytoplasmic membrane (4), but the mechanism by which it is inserted into the bilayer has not been described. FecR does not contain a classical signal sequence within its cytoplasmically orientated N-terminal domain (residues 1 to 75). However, a twin-arginine motif with a good match to the Tat consensus sequence immediately precedes a region of hydrophobicity that corresponds to the transmembrane domain (Fig. 1B). Both features are conserved characteristics of Tat signal peptides. Similar to other Tat-dependent inner membrane proteins, FecR contains no predicted cleavage site (Fig. 1B). The amino acid sequences of putative FecR orthologues derived from the eggNOG database (24) were aligned to assess the conservation of this putative Tat-targeting sequence (see Fig. S1 in the supplemental material). Out of the 95 predicted orthologues analyzed, 93.7% contained an internal twin-arginine motif. Those without Tat motifs were among those that displayed the lowest sequence amino acid similarity to E. coli K-12 FecR (Table S1) and clustered independently from the other orthologues (Table S1 and Fig. S2A), suggesting that these may not represent true FecR orthologues. Sequence motifs were generated using WebLogo.3 (25) with and without these outliers (Fig. 1C and S2B, respectively). These alignments show clear conservation of the twin arginines, indicating that they may be required for function. Furthermore, high frequencies of serine/threonine (−1 position), leucine (+3 position), and lysine (+4 position) residues were also noted, which are hallmarks of Tat signal sequences (26). FecR also contains two basic residues adjacent to the c-region that are known to act as Sec avoidance motifs (Fig. 1B) (20, 21). This feature was also conserved among putative FecR orthologues (Fig. 1D).

Periplasmic translocation of the FecR C-terminal domain is Tat dependent.

To determine whether the membrane integration of FecR is dependent on the Tat export machinery, we constructed a fusion protein that could be deployed as a reporter for periplasmic translocation of the C-terminal domain (Fig. 2A). The reporter FecR-BlaM was constructed by fusing the N-terminal domain and membrane-spanning region (amino acid residues 1 to 115) of E. coli FecR to β-lactamase (BlaM). This reporter was expressed under the transcriptional control of an arabinose-inducible promoter in a tat⁺ E. coli K-12 strain, 10β. β-Lactamase fusions serve as ideal reporters for periplasmic export because they must be trafficked beyond the bacterial inner membrane to effectively protect the cell from β-lactam antibiotics. Furthermore, many β-lactamases can be translocated across the inner membrane via either Sec or Tat systems and are used as a reporter for both pathways (27, 28).

FIG 2.

Mutation of twin-arginine residues inhibits periplasmic translocation of a FecR-BlaM fusion. (A) Domain architecture of the FecR-BlaM reporter. (B and C) The ampicillin (AMP) sensitivities of the FecR-BlaM reporter with RR-to-AA and RR-to-KK substitutions (E. coli strain, 10β) (B) and the FecR-BlaM reporter expressed in a Tat-deficient strain (E. coli strains HS3018-A and HS3018-A ΔtatABC) (C) were determined using M.I.C. Evaluator strips. Representative images of three biological replicates are shown.

We assessed resistance to the β-lactam antibiotic ampicillin using M.I.C. Evaluator strips and by spotting serial dilutions on ampicillin plates (Fig. S3). Table 1 and Fig. 2B demonstrate that cells producing the FecR-BlaM reporter grew to a concentration of >256 μg ml⁻¹, indicating effective translocation of the β-lactamase to the periplasm. The removal of arabinose from the plate rendered this strain fully sensitive to ampicillin. To test whether the twin-arginine motif was important for recognition of FecR by the Tat pathway, we constructed substitutions of the arginine pair to twin alanine (radical) or twin lysine (conservative). Cells producing the mutated fusion proteins demonstrated dramatically increased sensitivity to ampicillin (MIC, 16 μg ml⁻¹), suggesting that almost no β-lactamase had now been translocated to the periplasm. To further explore the Tat dependence of this fusion, the FecR-BlaM reporter was expressed in the Tat-deficient strain, HS3018-A ΔtatABC. Consistent with the notion that the insertion of the FecR transmembrane helix was Tat dependent, this strain demonstrated markedly increased ampicillin sensitivity relative to the isogenic wild type (Fig. 2C). Translation and membrane localization of the fusion proteins were not negatively affected by mutation of the twin-arginine motif or deletion of the Tat system (Fig. 3B and C), suggesting that ampicillin sensitivity was the consequence of ineffective β-lactamase translocation.

TABLE 1.

Effect of amino acid substitutions and a functional Tat system on periplasmic translocation of the FecR-BlaM fusion and the ability to support growth on ampicillin^a

Strain	FecR-BlaM fusion wild type or variant	Mean ampicillin MIC (μg ml−1)
10β	Wild type	>256
	R79A R80A	16 ± 0.0
	R79K R80K	16 ± 0.0
HS3018-A	Wild type	>256
HS3018-A ΔtatABC	Wild type	2 ± 0.0

^aDetermination of the MIC was performed in triplicate, and representative images are shown in Fig. 2.

FIG 3.

(A to C) Cell localization of FecR-Bla-His reporter and FecR-His with RR-to-KK substitutions (10β) (A and B) or FecR-Bla-His expressed in a Tat-deficient strain (HS3018-A ΔtatABC) (C). Soluble and membrane fractions were resolved by SDS-PAGE and transferred to nitrocellulose membranes probed with an anti-6×His antibody. All gels were imaged prior to transfer to determine the total protein loaded in each well (shown in lower images). (A) Soluble fractions. Lane 1, FecR-Bla-His; lane 2, FecR-Bla-His R79/80A; lane 3, FecR-His; lane 4, FecR-His R79/80K. (B) Membrane fractions. Lane 1, FecR-Bla-His; lane 2, FecR-Bla-His R79/80A; lane 3, FecR-His; lane 4, FecR-His R79/80K. (C) Lane 1, HS3018-A FecR-Bla-His, soluble fraction; lane 2, HS3018-A Δtat FecR-Bla-His, soluble fraction; lane 3, HS3018-A FecR-Bla-His, membrane fraction; lane 4, HS3018-A Δtat FecR, FecR-Bla-His, membrane fraction. (D) Crude cell extracts were washed with either 0.2 M Na₂CO₃ or 4 M urea prior to membrane sedimentation. Lane 1, HS3018-A FecR-Bla-His, membrane fraction, with Na₂CO₃; lane 2, HS3018-A Δtat FecR-Bla-His, membrane fraction, with Na₂CO₃; lane 3, HS3018-A FecR-Bla-His, membrane fraction, with urea; lane 4, HS3018-A Δtat FecR-Bla-His, membrane fraction, with urea. wt, wild type.

Mutation of the twin-arginine motif does not prevent membrane interaction.

Next, we determined whether substitutions of the paired arginine residues R79 and R80 (R79/80) influenced localization of the FecR and the FecR-BlaM reporter (Fig. 3A and B). A comparison of the relative proportion of FecR and FecR-BlaM in the soluble (Fig. 3A) and membrane (Fig. 3B) fractions revealed that the R79/80 substitutions did not prevent membrane association of the proteins. We also expressed the FecR-BlaM reporter in the wild-type (HS3018-A) and tat mutant (HS3018-A ΔtatABC) strains (Fig. 3C). Similarly, FecR-BlaM localized to the membrane in the absence of a functional Tat system. To determine whether FecR-BlaM was fully integrated into the bilayer, membranes of the wild-type and tat strains producing FecR-BlaM were washed with either 0.2 M Na₂CO₃ or 4 M urea (which can displace peripheral membrane proteins by disrupting ionic interactions and disrupting hydrogen bonding). Figure 3D shows that while carbonate washing had little effect on the membrane localization of FecR-BlaM in either strain, urea washing displaced FecR-BlaM from the membrane fraction of the tat mutant strain but not the wild type. These results confirm that FecR-BlaM behaves like an integral membrane protein in the wild-type strain but is not correctly membrane integrated in the absence of the Tat machinery.

Mutation of the twin-arginine residues results in downregulation of the fecABCDE operon.

No marked difference in growth rates was observed between HS3018-A and the tat mutant when grown with ferric citrate as a sole iron source (Fig. S4), which is consistent with observations made by Ize et al. (29). Previous characterization of the Fec system in E. coli has been performed using strains with an aroB mutation (E. coli strain AA93), which are unable to synthesize the siderophore enterobactin (which may compensate for the lack of Fec-mediated ferric iron import) (5, 30). Although these studies have been performed using different genetic backgrounds, these data indicate that some strains of E. coli K-12 can acquire ferric citrate in the absence of functional FecR.

To explore whether the fecR RR-to-KK mutation influenced the recognition of ferric citrate under iron-limiting conditions, transcription of the fecABCDE genes was determined by quantitative PCR (qPCR) analysis. A strain deficient in the fecR gene (E. coli strain BW25113) was complemented in trans with either wild-type fecR or the fecR R79/80K substitution; these were grown in medium supplemented with 2′,2′-dipyridyl and 1 mM sodium citrate. Cells were harvested at early stationary phase, and we observed no difference in the final optical densities between the strains tested, suggesting that the fecR mutant could grow using ferric citrate as a sole iron source.

We observed statistically higher expression of fecABCD (but not fecE) in the fecR mutant relative to the fecR R79/80K mutant, indicating that there is a low level of transcription of the fec genes in this strain. Crucially, we observed significantly increased expression of the fecABCDE operon in the strain complemented with wild-type fecR relative to the fecR R79/80K and fecR mutants (Fig. 4). This is consistent with the notion that the arginine-to-lysine mutation prevents periplasmic translocation of the C-terminal domain, which abrogates FecR binding to citrate-loaded FecA and decouples the FecIR signal cascade.

FIG 4.

Relative gene expression of the fecABCDE operon and fecR under iron-limiting conditions. Quantitative real-time PCR of a fecR mutant (strain BW25113) expressing either wild-type fecR or the fecR R79/80K substitution and an empty plasmid control was performed on RNA/cDNA extracted from cell cultures grown in medium supplemented with 2′,2′-dipyridyl and 1 mM sodium citrate. Error bars represent the standard deviation of mean values derived from three biological replicates.

DISCUSSION

In this study, we have addressed the membrane integration pathway for the bitopic membrane protein FecR. Analysis of the transmembrane domain of FecR homologues demonstrates that it is preceded by a conserved twin-arginine motif and that several positive charges are located close to the C-terminal end, a feature which is known to act as a Sec avoidance motif. Consistent with this, replacement of the C-terminal extracellular domain of FecR with beta-lactamase resulted in beta-lactamase translocation that was dependent on both the twin arginines and the Tat pathway. The Tat system is known to integrate several classes of membrane proteins, including monotopic proteins that are anchored by a single N- or C-terminal transmembrane domain, and polytopic proteins, where only the final transmembrane domain is Tat dependent (11, 13, 21, 22, 31). FecR constitutes a new class of bitopic Tat-dependent membrane proteins with an internal, uncleaved twin-arginine signal sequence that separates two globular domains.

The polytopic Rieske protein of Streptomyces coelicolor, which requires the concerted action of Sec and Tat pathways for membrane integration, contains more than one transmembrane helix and has an odd number of transmembrane helices before the twin-arginine residues. Given that FecR only contains a single helix, it seems unlikely that it is targeted to the membrane by a similar dual-action mechanism. Our data indicate that FecR associates with the membrane in the absence of a functional Tat system (Fig. 3), which could suggest cooperation with another pathway for its insertion. However, a study by Gray et al. demonstrated that FecR membrane localization was unperturbed in a yidC mutant (10). The E. coli Tat substrate SufI and some thylakoid proteins have been shown to bind to the membrane before interaction with the Tat or in the absence of functional Tat machinery (32–35). Although our data clearly demonstrate that FecR is a Tat substrate, we do not rule out the possibility that it targets to the membrane via another pathway.

The mechanism by which the Tat system recognizes this internal signal sequence is unclear, but it should be noted that the related thylakoid Tat system is capable of translocating the substrate protein pOE17 even after deliberate fusion of a large polypeptide domain N-terminal to the Tat signal peptide (36). This indicates that the integration of bitopic proteins is likely to be a common feature of the Tat pathway from different organisms.

Interestingly, some complex Tat substrates have signal peptides that contain greatly extended n-regions prior to the twin-arginine motif (37). Such extensions are almost invariably found on substrates that bind redox cofactors and/or partner proteins prior to export, and they appear to serve as binding sites for dedicated chaperones that coordinate folding and assembly (38–42). FecR is distinct from these Tat substrates since it does not contain any redox cofactor, and its signal sequence n-region is considerably longer than other Tat signal peptide n-regions. It is not clear why FecR should be a Tat substrate, although feasibly, it may be energetically favorable for FecR to fold in the cytoplasm prior to transport through the Tat machinery. Alternatively, it is conceivable that the FecR N-terminal domain binds FecI, driving cytoplasmic folding before its integration into the membrane. In conclusion, FecR joins an expanding list of inner membrane proteins that contain a non-N-terminal Tat signal sequence.

MATERIALS AND METHODS

Strains and plasmids.

The bacterial strains and plasmids used in this study are listed in Tables S2 and S3, respectively in the supplemental material. The ampicillin resistance cassette of plasmid pEC415 was exchanged for a kanamycin cassette by Gibson Assembly (NEBuilder HiFi assembly master mix; NEB) using the primers pET28akanF/R (pET28a as the template) and pEC415kanF/R (pEC415 as the template) to generate the plasmid pEC415K.

Plasmid pECfecR-blaM was constructed by Gibson Assembly using the primers fecRF/fecRR and pEC415KfecRF/pEC415KfecRR with E. coli genomic DNA and pEC415K as the template. Amino acid residues 79 and 80 were mutated from arginine to alanine or lysine using the primers fecR R7980A F/fecR R7980 R and fecR R7980K F/fecR R7980 R, respectively.

5′-phosphorylated primers fecR-blaM His F and R were used to introduce a C-terminal His tag into the products of fecR-blaM and fecR-R7980K-blaM using pECfecR-blaM and pECfecR-R7980K-blaM as the templates, generating plasmids pECfecR-blaM-His and pECfecR-R7980K-blaM-His, respectively. Similarly, primers fecR His F and R were used to generate C-terminally His-tagged product of fecR using pECfecR and pECfecR R7980K as the templates to generate pECfecR-His and pECfecR R7980K-His, respectively.

Growth conditions.

E. coli strains were cultured in LB broth or agar (Merck Millipore) at 37°C and supplemented, when required, with 50 μg ml⁻¹ kanamycin and 100 μg ml⁻¹ ampicillin.

The susceptibility of the E. coli strains harboring FecR-BlaM reporter to ampicillin was investigated by determining the MIC that prevented growth. Overnight cultures of each strain were diluted to an optical density at 600 nm (OD₆₀₀) of 0.5, and a bacterial lawn was grown on LB plates (supplemented with kanamycin) by swabbing. M.I.C. Evaluator strips (Oxoid) were placed on the plates, which were grown overnight at 37°C. Three independent replicates were performed, and representative images are shown in Fig. 2.

For transcript and growth kinetic analyses, cells were grown at 37°C in nutrient broth (Merck) supplemented with 50 μM 2,2′-dipyridyl and 1 mM citrate.

Membrane extractions.

E. coli cells harboring the plasmid pECfecR-blaM were grown overnight at 37°C in medium supplemented with 0.2% (wt/vol) l-arabinose and 50 μg ml⁻¹ kanamycin. Cells were pelleted by centrifugation (3,200 × g) and resuspended in 20 mM Tris-HCl (pH 7.5) and 200 mM NaCl. Cells were lysed using a FastPrep homogenizer (MPBio), and unlysed cells and large cell debris were removed by centrifugation (7,000 × g). The resulting clarified lysate was pelleted by ultracentrifugation (1 h at 150,000 × g) to separate membrane and soluble fractions. Membrane pellets were resuspended in 50 mM Tris HCl (pH 7.5), 5 mM MgCl₂, and 10% (vol/vol) glycerol.

For the membrane interaction assays, cells were resuspended in 50 mM Tris HCl (pH 7.5) and 10% (vol/vol) glycerol and lysed as described above. Crude lysate was treated with either 0.2 M Na₂CO₃ or 4 M urea for 1 h at 4°C, followed by ultracentrifugation at 150,000 × g. Membrane pellets were resuspended in 50 mM Tris HCl (pH 7.5), 5 mM MgCl₂, and 10% (vol/vol) glycerol.

Immunoblot analysis.

The total protein concentration of each sample was quantified using a Bradford assay and normalized to equal concentrations. Proteins were resolved by SDS-PAGE with Mini-PROTEAN TGX stain-free gels (Bio-Rad) and transferred to nitrocellulose membranes using the iBlot 2 dry blotting system (Thermo Fisher). Prior to transfer, gels were imaged using the Bio-Rad ChemiDoc MP imaging system to determine the total protein content loaded in each well. The primary antibody, mouse anti-6×His (used at a 1:10,000 dilution; Invitrogen, UK), was suspended in phosphate-buffered saline (PBS) and 0.1% (vol/vol) Tween 20 and incubated with the membrane for 1 h. Membranes were washed three times with PBS and incubated for 45 min with a secondary goat anti-mouse IgG IRDye680 antibody (both at 1:10,000 dilution; Li-Cor Biosciences, UK). Fluorescent signal was detected using the Odyssey detection system (Li-Cor Biosciences, UK).

qPCR.

Cells were grown to early stationary phase. All cells were harvested and stored in RNAlater (Ambion) at 4°C overnight. Cells were sedimented by centrifugation, and RNA was extracted using the Monarch total RNA miniprep kit (NEB), according to the manufacturer’s instructions. The resulting RNA was used as the template for reverse transcription and conversion into cDNA using SuperScript IV reverse transcriptase (Invitrogen). qPCR was performed on the cDNA using Power SYBR green (Thermo), according to the manufacturer’s instructions, with the appropriate primers at 10 pmol (see Table S4). Amplification was performed using an ABI Prism 7500 real-time PCR system, and fluorescence data were processed using the SDS software (ABI). Relative gene expression was determined using gyrA and rpoS as controls. Three independent biological replicates were performed for each strain and growth condition.

Bioinformatic analysis.

Sequences of theoretical FecR orthologues were retrieved from the eggNOG database of orthologous groups and functional annotations (24). Sequence alignments were constructed using ClustalW and ESPript 3.0 (43). Phylogenetic trees were generated using the Interactive Tree of Life software (44). WebLogo sequence motifs were generated using WebLogo3 (25).