Autogenous Transcriptional Regulation of the regA Gene, Encoding an AraC-Like, Essential Virulence Regulator in Citrobacter rodentium

Abstract

We identified several promoters responsible for the expression of regA, which encodes a global virulence regulator in Citrobacter rodentium. Expression of some of the promoters was strongly autoactivated by RegA in conjunction with bicarbonate. Biochemical and mutational analyses were used to determine the consensus sequence of the RegA-binding sites.

Infection of mice with Citrobacter rodentium is used as a small-animal model of infection with the attaching and effacing pathogens enteropathogenic Escherichia coli and enterohemorrhagic E. coli (4, 13, 20). All three pathogens carry the locus for enterocyte effacement pathogenicity island (LEE PAI), which is required for the intimate attachment of bacteria to intestinal epithelial cells and the induction of attaching and effacing lesions (3, 6, 16). We have previously identified a virulence regulon (the RegA regulon) that C. rodentium requires to colonize the mouse intestine (9, 24). The master regulator of this regulon, the RegA protein, is a member of the AraC family of regulators (8) and is closely related to the virulence regulators Rns (from enterotoxigenic E. coli) (7), AggR (from enteroaggregative E. coli) (15), and ToxT (from Vibrio cholerae) (5, 10). RegA, in conjunction with its cofactor, bicarbonate ions, activates the expression of the LEE PAI via the stimulation of the grlRA operon and of the divergently transcribed adcA and kfc operons (18, 23). Given the central role of RegA in the virulence of C. rodentium, we decided to examine the transcriptional regulation of the regA gene itself.

Expression of regA from upstream promoters.

The regA gene is located on the C. rodentium chromosome, downstream of the adcA gene and an uncharacterized open reading frame (hereinafter termed regB) (Fig. 1 A). The close proximity and head-to-tail orientation of these three genes point to the possibility of cotranscription.

FIG. 1.

Transcriptional analysis of the adcA-regB-regA gene cluster of C. rodentium by reverse transcription (RT)-PCR and assay of promoter-lacZ fusions. (A) A schematic map of the genetic region including regA and three upstream genes is shown by large arrows (not to scale; intergenic regions are 208 and 212 bp for adcA-regB and regB-regA, respectively). The locations of various oligonucleotide primers used for RT-PCR analysis are indicated by small arrows below the map. Their sequences are as follows: AT1, 5′ TATTGAGCGGTCAGCCTTCG-3′; AT2, 5′-CTGATCTGTAAGCAAGCGCAC-3′; AT3, 5′-GAAGCTCCCGGTGTTACTAAG-3′; AT4, 5′-TCATAAGGAACTCCGGAGCC-3′; AT15, 5′-CGGTCACGGTTAAACAGTCGCTG-3′; and AT16, 5′-CCAGCCATCGCCATGACGATTAC-3′. cDNA (dashed line) obtained by using total RNA of C. rodentium as a template and AT4 as a primer was used as a template in PCRs involving primer pairs AT1/AT2, AT3/AT4, and AT15/AT16. The sizes of the PCR products are indicated. (B) Samples were analyzed on a 1% agarose gel. The PCR primer pairs are shown above the gel. Lanes: M, marker (100-bp ladder; New England Biolabs); g, control experiment (carried out using C. rodentium genomic DNA as a template); −, negative control experiment (carried out using total RNA as a template); and +, amplicon obtained using AT4-generated cDNA as a template. (C) β-Galactosidase activities of derivatives of E. coli MC4100. Cells carrying the plasmid pMU2385 (negative control) or the regB-lacZ or regA-lacZ plasmid were grown in Luria broth at 37°C. The β-galactosidase activities shown are the averages of results from three independent experiments, and error bars represent standard deviation. Units are those defined by Miller (11).

To determine whether the entire adcA-regB-regA region is transcribed as a single unit from the adcA promoter, total RNA from wild-type C. rodentium strain ICC169 (13) was extracted and purified. The cDNA generated using a primer that binds to the regA gene was assayed by PCR using primer pairs designed to amplify the adcA coding region and the adcA-regB and regB-regA intergenic regions (Fig. 1A). Amplicons were produced for all primer pairs, demonstrating the existence of a polycistronic transcript that spans the entire region between the adcA and regA genes (Fig. 1B).

To examine if there are any internal promoters in this region, the adcA-regB and regB-regA intergenic regions were cloned into the single-copy lacZ fusion plasmid pMU2385 (21) to produce the regB-lacZ and regA-lacZ transcriptional fusion plasmids, respectively. E. coli strain MC4100 (1) was transformed with each of these plasmids. Cells were grown to mid-log phase, and promoter activity was measured by a β-galactosidase assay. Significant promoter activity was detected for both the adcA-regB and regB-regA intergenic regions (Fig. 1C).

Identification of transcriptional start sites for regB and regA.

Transcriptional start sites associated with the regB and regA promoters were determined by using primer extension analysis. Briefly, E. coli MC4100 strains (lacking adcA, regB, and regA) were transformed with either the regB-lacZ or regA-lacZ plasmid and grown to mid-log phase, after which total RNA was extracted and primers AT2 and AT40 (Fig. 2 B and C) were used to probe the transcriptional start sites of the regB and regA genes, respectively.

FIG. 2.

Determination of the start sites of transcription of the regB and regA promoters by primer extension analysis. (A) Derivatives of E. coli MC4100 carrying regB-lacZ, mutant regB-lacZ, regA-lacZ, or mutant regA-lacZ were grown in Luria broth at 37°C to mid-exponential phase. Total cellular RNA isolated from each of the strains was hybridized with ³²P-labeled primer AT2 or AT40. Lanes: 1, GA ladders from the respective promoter sequences; 2, reaction with RNA from MC4100 carrying either wild-type (WT) regB-lacZ or WT regA-lacZ; and 3, reaction with RNA from MC4100 carrying either mutant regB-lacZ or mutant regA-lacZ. Extension products from regB-lacZ and regA-lacZ are indicated by the labels P_B and P_A, respectively. (B and C) The sequences of the promoter regions of regB (B) and regA (C) are shown. The transcriptional start sites are indicated with angled arrows. The putative −35 and −10 regions of the various promoters are overlined, and the TGN motif of the putative P_A promoter is marked with asterisks. Base changes in the −10 regions of the regB and regA promoters are shown below the sequences.

The assay with the regB-lacZ plasmid generated one major extension signal (P_B), mapping 41 bp upstream of the putative regB start codon (Fig. 2A). This transcriptional start site is associated with a near-perfect putative −10 sequence (5′-TATAGT-3′, with 5 of 6 possible matches to the σ⁷⁰ consensus sequence), an optimal spacer of 17 bp, and a poor putative −35 sequence (5′-TGGTGT-3′, with 2 of 6 possible matches) (Fig. 2B). The assay also produced a number of minor signals that are shorter than P_B. To find out whether these shorter RNA species were degradation products of the major RNA signal (P_B), the putative −10 sequence was mutagenized from 5′-TATAGT-3′ to 5′-AAAAGA-3′. Primer extension analysis of the mutant regB-lacZ plasmid showed that knocking out this promoter resulted in complete removal of all extension products, indicating that P_B is the sole promoter of regB and that the shorter RNA species are likely to be those degraded from P_B.

The primer extension assay involving the regA-lacZ plasmid identified a product mapping 23 bp upstream of the regA start codon (Fig. 2A). The putative σ⁷⁰ promoter (P_A) associated with this product has a near-perfect −35 sequence (5′-TTTACA-3′, with 5 of 6 matches) and an extended −10 motif (5′-TGAAATTAT-3′) separated by 17 bp (Fig. 2C). To confirm the identity of the putative P_A promoter, the extended −10 region was mutagenized from 5′-TGAAATTAT-3′ to 5′-ACAAAAAGA-3′. The primer extension assay showed that the mutation completely ablated the transcription activity of the P_A promoter (Fig. 2A).

RegA regulates the expression of the regB but not the regA promoter.

Previously, we reported that RegA regulates adcA expression (23). To determine if RegA also plays a role in the expression of regB and regA, we transformed either the wild-type RegA⁺ C. rodentium strain ICC169 or the isogenic RegA⁻ strain EMH1 (9) with the regB-lacZ and regA-lacZ transcriptional fusions. Cells were grown to mid-log phase in the absence or presence of 45 mM bicarbonate, and promoter activity was measured by assaying for β-galactosidase activity. The data in Table 1 show that the regA-lacZ fusion was not regulated by RegA because the levels of transcription were essentially the same in the RegA⁻ and RegA⁺ backgrounds. In contrast, transcription of the regB-lacZ fusion was activated 3.5- and 79-fold by RegA in the absence and presence of bicarbonate, respectively.

TABLE 1.

Effects of RegA and bicarbonate on expression of regB and regA

lacZ fusion plasmid	β-Galactosidase activitya of strain:
	EMH1 (RegA−)		ICC169 (RegA+)
	−NaHCO3	+NaHCO3	−NaHCO3	+NaHCO3
pMU2385 (control)	2	2	2	2
regB-lacZ plasmid	51	82	176 (3.5)b	4026 (79)
regA-lacZ plasmid	1,995	1,766	1,434	1,624

^aβ-Galactosidase assays were carried out with C. rodentium EMH1 derivatives containing either plasmid pACYC184 (control) or pEH6 (pACYC184::regA) after growth of the cells in Luria broth in the absence (−) or presence (+) of 45 Mm NaHCO₃. The β-galactosidase activity shown (expressed in Miller units) is the average of results from three independent experiments, with the standard deviation below 15%.

^bNumbers in parentheses indicate the level of activation (n-fold) by RegA (i.e., the β-galactosidase activity of the RegA⁺ strain divided by that of the RegA⁻ control).

RegA binds directly to the regB regulatory region.

We next performed an electrophoretic mobility shift assay (EMSA) to determine whether RegA binds specifically to the regB promoter region. A hybrid protein in which the C terminus of RegA is fused with the N terminus of the maltose-binding protein (RegA-MBP) from E. coli was used (23). Three DNA fragments (designated S1, S2, and S3), which cover the entire regB regulatory region, were labeled with ³²P and used as DNA-binding targets (Fig. 3 A). Various concentrations of RegA-MBP were incubated with each of the DNA fragments in the presence of 45 mM bicarbonate, and the samples were then analyzed on native polyacrylamide gels.

FIG. 3.

Identification of the RegA-binding site in the regB regulatory region by EMSA and in-gel DNase I footprinting. (A) Schematic of the regB promoter region showing the location of promoter P_B. The three DNA fragments (S1, S2, and S3) were generated by PCR. The coordinates of each fragment relative to the transcriptional start site of P_B are shown. (B) In the EMSA, the following concentrations of RegA-MBP were used: lanes 1, 0 nM; lanes 2, 10 nM; lanes 3, 20 nM; lanes 4, 40 nM; and lanes 5, 80 nM. Unbound DNA is labeled F. The RegA-MBP-S2 complex is labeled R. (C) In-gel DNase I footprinting analysis of the 5′-end-labeled noncoding strand. The region protected by RegA is indicated by a vertical line. Hypersensitive positions are indicated with blue arrows, and protected positions are labeled with red arrows. (D) Data from a densitometric analysis (using ImageGauge v4.23 [FujiFilm]) comparing the free-DNA lane (red) to the RegA-DNA complex (blue) are shown. H and P represent hypersensitive and protected sites, respectively.

As shown in Fig. 3B, RegA-MBP was unable to bind to fragment S3 under the assay conditions, as no protein-DNA complex was detected. As for fragment S1, no protein-DNA complex was formed at or below RegA-MBP concentrations of 40 nM. Although RegA-MBP partially shifted the S1 fragment at a concentration of 80 nM, the complexes formed were smeared and unstable, indicating a weak binding affinity of RegA-MBP for S1. In contrast, strong binding of RegA-MBP to S2 was detected, with a distinct protein-DNA complex formed at a RegA-MBP concentration of 10 nM. Higher protein concentrations resulted in a complete shift of the S2 DNA fragment by RegA-MBP, indicating that a RegA-binding site resides within this fragment.

To further resolve the RegA-binding site associated with the regB regulatory region, DNA from fragment S2 was used for in-gel DNase I footprinting as described previously (23). In this experiment, we first used DNase I to digest free DNA and the protein-DNA complex immobilized in a polyacrylamide gel. The samples were then analyzed on a 7% DNA sequencing gel. This assay revealed a number of sites that were protected from or hypersensitive to DNase I cleavage in the presence of bound RegA-MBP compared to free DNA (Fig. 3C). Densitometry analysis (using ImageGauge v4.23 [FujiFilm]) of this footprint showed a clear difference between the intensities of a number of bands from free DNA (red) and those of the RegA-DNA complex (blue) (Fig. 3D). The area protected by RegA-MBP extends from position −34 to −76 relative to the start site of transcription of the regB promoter P_B.

Identification of RegA-binding sites.

The helix-turn-helix (HTH) DNA-binding motif of RegA exhibits close homology to those of the AggR and Rns proteins (Fig. 4 A). Moreover, AggR and Rns have similar DNA-binding sequences (12). To test if AggR and Rns can activate the transcription of RegA-regulated promoters, derivatives of pACYC184 (2) producing AggR (pAT-AggR) and Rns (pAT-Rns) were each introduced into C. rodentium strain EMH1 (ΔregA) carrying a promoter-lacZ fusion plasmid known to be regulated by RegA (the adcA-lacZ, regB-lacZ, or kfcC-lacZ plasmid) (22). Assays of β-galactosidase activity showed that both AggR and Rns were able to activate transcription from the three promoters in the presence of 45 mM NaHCO₃ (Fig. 4B). These results suggest that the three regulatory proteins recognize similar DNA-binding sites.

FIG. 4.

Identification of RegA-binding sites by comparative and mutational analyses. (A) Alignment of helix-turn-helix (HTH) motifs of AggR, Rns, and RegA using the PS01124 profile defined in PROSITE (19). The numbers indicate the positions of amino acid residues within each of the proteins. Identical and conserved amino acids are indicated with asterisks and colons, respectively. (B) C. rodentium EMH1 (RegA⁻) strains carrying various promoter-lacZ fusions with or without the pACYC184 derivatives pAT-AggR (AggR⁺) and pAT-Rns (Rns⁺) were grown in Luria broth in the presence of 45 mM NaHCO₃. Promoter activities were measured by a β-galactosidase assay. The levels of activation (n-fold) by the various regulators are shown in parentheses above the error bars. The result shown is the average of results from three independent experiments, and error bars represent standard deviation. (C) Rns-binding site logo as defined by Munson et al. (14). The putative RegA-binding sites of the kfcC, adcA, and regB regulatory regions are shown, together with the coordinates relative to the transcriptional start site from the kfcC, adcA, or regB (P_B) promoter. The 10 bases shown in uppercase letters are those deleted in the kfcC, adcA, and regB promoter mutants. The consensus of the three RegA-binding sites is represented by the binding site logo. (D) Effects of deletions of putative RegA-binding sites on the activation of kfcC, adcA, and regB. C. rodentium EMH1 strains containing either pACYC184 (RegA⁻) or pEH6 (RegA⁺) (9) were transformed with kfcC-lacZ, adcA-lacZ, or regB-lacZ. The positions of deletions in the mutants are indicated above the downward-pointing arrowheads, and angled arrows identify the transcriptional start sites. These C. rodentium derivatives were grown in Luria broth at 37°C in the presence of 45 mM NaHCO₃. The numbers in parentheses represent the levels of activation (n-fold) by RegA (expressed as the ratio of the β-galactosidase activity of the RegA⁺ strain to that of the RegA⁻ strain). The β-galactosidase activity shown is the mean of results from three independent experiments, with standard deviations below 15%. (E) The putative RegA-binding site sequence is boxed, and the highly conserved bases are in boldface. The base substitutions in the five regB mutants are shown below the sequence. Each of the mutant fragments was cloned into plasmid pMU3285 to form lacZ transcriptional fusions (regB-lacZmut-1 to regB-lacZmut-5). β-Galactosidase assays were carried out with C. rodentium strain EMH1 carrying pEH6 (RegA⁺) following the growth of cells under conditions inducing RegA activation. The result shown is the average of results from three independent experiments, and error bars represent standard deviation.

Previous work by Munson's group has determined the consensus of the Rns-binding sites (Fig. 4C) (14). A sequence alignment using the Rns consensus sequence identified a putative RegA-binding site immediately upstream of the adcA, kfcC, and regB promoter core sequences (Fig. 4C). Deletion of 10 bases (one helix turn) from each of the putative RegA-binding sites resulted in severe loss of RegA-mediated activation of the respective promoters (Fig. 4D), confirming the importance of these sequences in RegA binding.

We next made a number of double- or triple-base changes within or outside the putative regB operator (Fig. 4E). While the mutant regB-lacZmut-1, which contains base changes outside the putative operator, exhibited the same level of transcriptional activity as the wild-type regB-lacZ fusion upon activation by RegA, all the other four mutants (regB-lacZmut-2 to regB-lacZmut-5), which carry base changes within the RegA-binding site, had a marked reduction in the levels of transcriptional activation by RegA (to 10 to 35% of the wild-type regB-lacZ level) (Fig. 4E). These results are consistent with those from the deletion mutagenesis and DNase I footprinting analyses of the regB promoter and provide further confirmation of the identity of the regB operator.

Closing remarks.

Intestinal colonization is an essential step in the pathogenesis of C. rodentium, and the RegA protein plays an essential role in this process (9). Because RegA regulates the expression of more than 60 operons, including the LEE PAI and a number of non-LEE virulence genes of C. rodentium (18, 24), the transcriptional regulation of the regA gene itself in response to different environmental conditions must be important in governing the pathogenic process. Indeed, three independent upstream regions are responsible for regA transcription, two of which (the adcA and regB regulatory regions) are autoactivated by RegA in response to the environmental factor bicarbonate.

The regB gene encodes a short protein of 166 amino acids which exhibits sequence homology to domain 4 of the σ⁷⁰ subunit of E. coli RNA polymerase. Although the regB gene is translationally active (data not shown), the function of the RegB protein is yet to be determined. Transcription from the regB promoter is subject to negative regulation by the H-NS protein (data not shown). The RegA protein functions in a way similar to that of its homologs, ToxT, Rns, and AggR, which allow transcription of their target promoters by relieving the inhibitory effect of H-NS. We recently reported that the N-terminal arm of RegA, which comprises the first 16 amino acid residues, is responsible for modulating the response to bicarbonate (22). It has been postulated that, in the absence of bicarbonate, the N-terminal arm wraps around the C-terminal DNA-binding domain, interfering with its ability to bind to DNA targets. Bicarbonate can alter the conformation of the RegA protein, mitigating the autoinhibition and enhancing the binding affinity of RegA for its DNA target (22). This cofactor-mediated modulation of RegA activity is similar to the light switch mechanism described previously for the AraC protein (see reference 17 for a review). Using the consensus sequence of the Rns-binding sites, we have identified putative RegA-binding sites which are positioned immediately upstream of the adcA, kfcC, and regB promoters. The predictions are also in agreement with the data from EMSA and DNase I footprinting (23) (Fig. 3C). Mutational analysis confirmed the operator status of these putative RegA-binding sites.

In summary, data presented in this study show that the expression of regA is highly regulated and environmentally responsive. This regulation is achieved via two positive feedback loops, with the RegA protein acting directly at the adcA and regB regulatory regions. While the constitutive regA promoter allows the constant presence of low levels of RegA in the cell, the strong activation of the promoters associated with the adcA and regB genes ensures a rapid enhancement of cellular levels of RegA in response to a gut-associated chemical signal, bicarbonate, and thereby leads to the expression of RegA-regulated virulence factors within C. rodentium at the site of infection in a timely fashion.