|Home | About | Journals | Submit | Contact Us | Français|
The locus for enterocyte effacement (LEE) is the virulence hallmark of the attaching-and-effacing (A/E) intestinal pathogens, namely, enteropathogenic Escherichia coli, enterohemorrhagic E. coli, and Citrobacter rodentium. The LEE carries more than 40 genes that are arranged in several operons, e.g., LEE1 to LEE5. Expression of the various transcriptional units is subject to xenogeneic silencing by the histone-like protein H-NS. The LEE1-encoded regulator, Ler, plays a key role in relieving this repression at several major LEE promoters, including LEE2 to LEE5. To achieve appropriate intracellular concentrations of Ler in different environments, A/E pathogens have evolved a sophisticated regulatory network to control ler expression. For example, the LEE-encoded GrlA and GrlR proteins work as activator and antiactivator, respectively, of ler transcription. Thus, control of the transcriptional activities of the LEE1 (ler) promoter and the grlRA operon determines the rate of transcription of all of the LEE-encoded virulence factors. To date, only a single promoter has been identified for the grlRA operon. In this study, we showed that the non-LEE-encoded AraC-like regulatory protein RegA of C. rodentium directly stimulates transcription of the grlRA promoter by binding to an upstream region in the presence of bicarbonate ions. In addition, in vivo and in vitro transcription assays revealed a σ70 promoter that is specifically responsible for transcription of grlA. Expression from this promoter was strongly repressed by H-NS and its paralog StpA but was activated by Ler. DNase I footprinting demonstrated that Ler binds to a region upstream of the grlA promoter, whereas H-NS interacts specifically with a region extending from the grlA core promoter into its coding sequence. Together, these findings provide new insights into the environmental regulation and differential expressions of the grlR and grlA genes of C. rodentium.
Citrobacter rodentium causes transmissible colonic hyperplasia and diarrhea in mice (34). Like the human diarrheagenic pathogens enteropathogenic Escherichia coli (EPEC) and enterohemorrhagic E. coli (EHEC), C. rodentium induces attaching-and-effacing (A/E) lesions in the intestinal epithelium of its host (40, 49). All three of these enteric pathogens possess a pathogenicity island known as the locus for enterocyte effacement (LEE), which is responsible for the A/E phenotype (13, 18, 27, 36, 43). So far, all of the LEE-encoded virulence factors investigated in C. rodentium play roles in virulence equivalent to the roles played by those from EPEC and EHEC (14). Furthermore, the regulatory networks controlling transcription of the LEE are broadly similar in the three pathogens (37, 63). For these reasons, infection of mice with C. rodentium has been used as a convenient small animal model to investigate the molecular and cellular pathogenesis of EPEC and EHEC and the regulation of LEE expression by these organisms (14, 40).
The LEE comprises 41 open reading frames, most of which are clustered into five operons: LEE1, LEE2, LEE3, LEE5, and LEE4 (13, 18). These transcriptional units encode a type III secretion system (T3SS), translocator and effector proteins secreted by this system, intimin (an outer membrane protein), and its type III secreted translocated receptor, Tir (13, 18, 38, 48, 54). Transcription of these operons is controlled by a number of general and specific regulators encoded on the chromosomal backbone and the LEE itself, respectively (6, 17, 20, 21, 23, 38, 44, 51, 52, 57). As with many other horizontally acquired virulence operons in Gram-negative pathogens, the expression of each of the LEE operons is subject to xenogeneic silencing by the global regulator H-NS (histone-like nucleoid structuring protein) (16, 19, 56). This form of repression is achieved through the binding of H-NS to curved AT-rich regions located mainly in promoter regions of horizontally acquired genes, preventing the formation of promoter open complexes or inhibiting the elongation of RNA polymerase (41, 50). In the case of the LEE, the Ler protein (the LEE-encoded regulator), which is encoded by the first gene of the LEE1 operon, antagonizes H-NS-mediated silencing (6) and thus activates transcription of all the major promoters from LEE2 to LEE5 but not that of LEE1 itself (2, 17, 20, 38, 48, 54).
Expression of the LEE1 promoter, which is an important checkpoint of LEE expression overall, is delicately regulated. This promoter is negatively regulated by Hha and H-NS and also negatively regulated in a concentration-dependent manner by Ler (2, 51, 57) but is positively regulated by the integration host factor (IHF) and quorum sensing (20, 55). In addition, the LEE1 promoter is activated by the GrlA protein (global regulator of LEE activator), which is encoded on a transcriptional unit located between LEE1 and LEE2 (1, 14, 46). The grlA gene is positioned downstream of an open reading frame named grlR (encoding GrlR [global regulator of LEE repressor]) within the same transcriptional unit. In contrast to GrlA, the GrlR protein exerts a negative effect on LEE1 expression (1, 14, 32), by forming heterodimers with GrlA (11, 28).
The grlRA operon is said to be transcribed from a single σ70 promoter (designated the grlRA promoter here), whose expression is directly stimulated by Ler (1). In this study, we carried out in vivo and in vitro transcriptional analysis of the grlA gene and identified a highly regulated σ70 promoter (the grlA promoter) immediately upstream of the GrlA coding sequence.
We also showed that expression of GrlA is influenced by RegA, an essential virulence-regulating protein of C. rodentium (25, 31). RegA belongs to the AraC family of transcriptional regulators and activates the expression of a number of putative virulence genes of C. rodentium, encoding various surface proteins (25). In addition, RegA inhibits transcription of more than 40 housekeeping genes involved in maintaining normal cellular functions, such as amino acid and carbohydrate biosynthesis and uptake (63). Importantly, RegA requires an environmental signal, bicarbonate, that is found in the gut, to exert its regulatory effect on gene expression (61, 62). By using microarray analysis, we previously demonstrated that transcription of the grlRA operon of C. rodentium is significantly stimulated by RegA in the presence of bicarbonate (63). In this study, we performed a molecular analysis of grlRA transcription and showed that RegA directly activates expression of the grlRA promoter by binding to an upstream region of grlRA when bicarbonate is present.
The bacterial strains and plasmids used in this work are listed in Table Table1.1. Oligonucleotides used in this study are listed in Table Table2.2. Bacteria were grown at 37°C in Luria-Bertani both (LB) or in M9 minimal medium (47). For solid media, 1.5% (wt/vol) agar was added. Where appropriate, media were supplemented with antibiotics at the following concentrations: ampicillin, 100 μg/ml; chloramphenicol, 10 μg/ml; kanamycin, 50 μg/ml; trimethoprim, 40 μg/ml in LB and 10 μg/ml in M9 minimal medium. Restriction enzymes and chemicals were purchased commercially. Purified E. coli RNA polymerase holoenzyme was purchased from the USB Corporation.
Standard recombinant DNA procedures as described by Sambrook and Russell (47) were used. Plasmids were purified using the Wizard Plus SV Minipreps DNA purification system (Promega). DNA was sequenced by using a model 377 DNA sequencer and ABI Big Dye terminators (Perkin-Elmer Corporation). PCRs were carried out using PCR master mix from Promega. A TOPO TA cloning kit (Invitrogen) was routinely used for cloning and sequencing of PCR fragments.
The lacZ transcriptional fusions used in this study were constructed by PCR amplification of DNA fragments which span the regulatory regions of the genes, grlR and grlA, by using C. rodentium ICC169 chromosomal DNA as the template and the primers listed in Table Table2.2. Each of the PCR fragments was cloned into TOPO TA cloning vector pCR2.1-TOPO and sequenced. The fragments were then excised from the TOPO TA derivatives and cloned into the appropriate sites of the single-copy plasmid pMU2385 to create lacZ transcriptional fusions. The forward primers MT125, MT156, MT155, and MT154 were used with reverse primer MT124 to generate the fusions grlRA-1, -2, -3, and -4, respectively. Fusions grlA-1 and -2 were constructed using reverse primer MT158 with forward primers MT143 and MT157, respectively.
Cells were grown to mid-log phase (optical density at 600 nm [OD600] = 0.6), and β-galactosidase activity was assayed as described by Miller (39). Specific activity was expressed in units as described in reference 39. The data shown are the results of at least three independent assays.
Overlapping extension PCR was used to generate a DNA fragment carrying a chloramphenicol resistance (Cmr) gene flanked by regions upstream and downstream of C. rodentium (ICC169) ler (9). First, primer pairs ler.F/lerCm.R and lerCm.F/ler.R were used to amplify DNA flanking the region to be deleted from the chromosome of C. rodentium, and primers for priming sites 1 and 2 were used to amplify the Cmr gene from plasmid pKD3 (12). The products of these three PCRs (100 ng each) served as the template in overlapping extension PCR using primers ler.F and ler.R to generate a linear construct, which was cloned into pCR2.1-TOPO, introduced into E. coli K-12 TOP10 cells and confirmed by sequencing. The pCR2.1-TOPO construct was used as a template in a PCR with primer pair ler.F/ler.R to amplify the linear allelic replacement DNA fragment, which was introduced into the C. rodentium strain EMH1, expressing λ Red recombinase from plasmid pKD46 (12). The Δler mutation in the generated C. rodentium strain, EMH8, was confirmed by PCR and sequencing.
A wild-type copy of the ler gene was amplified from genomic DNA of C. rodentium strain ICC169 by using primers MT137 and MT136. A 0.5-kb fragment encompassing ler and its ~100-bp flanking sequences was cloned into the NheI and BamHI sites of pBR322 behind the tet promoter to create plasmid pMTLer. For the overexpression and purification of Ler, the ler coding region was PCR amplified using primers CrlerNdeF and CrlerXhoR and cloned into the NdeI and XhoI sites of pET30b (Novagen) to generate plasmid pDH275, which expressed Ler as a fusion protein tagged with six histidine residues at its C terminus (Ler-His).
Expression and purification of MBP::RegA (RegA fused to the maltose binding protein [MBP]) was performed as described previously (62). To overexpress H-NS, a DNA fragment containing the coding sequence of H-NS was amplified by PCR using primer pair MT162/MT163 and chromosomal DNA of E. coli strain MC4100 as template. The H-NS fragment was cloned into TOPO TA and sequenced. The NdeI-XhoI H-NS fragment was then excised from the TOPO TA derivative and cloned into the same sites of plasmid pET22b (Novagen), where H-NS was expressed as a fusion protein with six histidine residues tagged at its C terminus (H-NS-His). The H-NS-His fusion protein was purified by nickel affinity chromatography as described by Smyth et al. (53).
For overexpression of Ler-His, E. coli expression strain BL21(DE3) containing plasmid pDH275 was induced for 1 h with 0.1 mM IPTG (isopropyl-β-d-thiogalactopyranoside) in 30 ml of Terrific broth (47). Ler-His was purified using immobilized metal affinity chromatography (Novagen) according to the manufacturer's instructions with minor alterations. Pelleted cultures were resuspended in 20 ml 1× binding buffer (Novagen) containing lysozyme and Triton X-100 at final concentrations of 100 μg/ml and 0.1%, respectively. Cells were then disrupted by sonication, and after centrifugation (10,000 × g with an SS34 rotor [Sorvall] at 4°C for 30 min), the clarified supernatant was batch bound with 0.5 ml settled Ni-nitrilotriacetic acid (NTA) resin (Qiagen) using a rotating wheel at 4°C for 2 h. The slurry was poured into a column and washed with 20 ml of 1× binding buffer containing 0.5% Tween 20, followed by 20 ml of wash buffer (Novagen) containing 0.5% Tween 20, and finally with wash buffer alone. Protein was eluted in 1-ml fractions by using 1× elution buffer (Novagen) containing 300 mM imidazole. The concentration of purified Ler-His protein was estimated by comparison with known concentrations of bovine serum albumin (BSA) on a Coomassie blue-stained polyacrylamide gel.
32P-labeled DNA fragments to be used in the electrophoretic mobility shift assay (EMSA) were generated as follows. Oligonucleotide primers were labeled with 32P at their 5′ ends by using [γ-32P]ATP and T4 polynucleotide kinase (New England Biolabs). The DNA fragments for analysis of RegA binding to the grlRA promoter region were amplified by PCR using primer pairs MT132/MT131 (for fragment A), MT130/MT129 (for fragment B), MT130/MT160 (for fragment C), and MT159/MT129 (for fragment D), with plasmid pMTgrlRA carrying the entire grlRA regulatory region as the template. The DNA fragments for analysis of RegA binding to the grlA promoter region were amplified by PCR using primer pairs MT147/MT164 and MT165/MT166 (control). EMSA was carried out as described previously (64). Briefly, end-labeled fragments were incubated with various amounts of purified MBP::RegA protein and 45 mM NaHCO3 at 37°C for 20 min in the binding buffer [10 mM Tris-HCl (pH 7.4), 50 mM KCl, 1 mM dithiothreitol (DTT), 100 μg/ml BSA, 5 ng/μl poly(dI-dC)]. Glycerol was added to a final concentration of 6.5% (wt/vol). The samples were analyzed by electrophoresis on 5% native polyacrylamide gels.
Runoff transcription assays were performed by using a method based on the standard single-round conditions described by Igarashi and Ishihama (26). The reaction mixtures contained linear DNA template (approximately 300 ng) and 1 unit of RNA polymerase. The samples were incubated at 37°C for 25 min in a total volume of 35 μl of transcription buffer (50 mM Tris-HCl [pH 7.8], 50 mM NaCl, 3 mM magnesium acetate, 0.1 mM EDTA, 0.1 mM DTT, 25 μg/ml BSA). Following the incubation, 15 μl of start solution (containing 1× transcription buffer with heparin [0.67 mg/ml], ATP, CTP, and GTP [0.53 mM each], UTP [0.053 mM], and [α-32P]UTP [3 μCi]) was added to initiate RNA synthesis. Transcription was allowed to proceed for 5 min before the reaction was terminated by the addition of phenol. Each sample was precipitated with ethanol and a portion of the precipitate was analyzed on a 6% sequencing gel next to a GA ladder. The ladder, which served as a molecular weight standard, was made by using the Maxam and Gilbert method (47) to sequence a grlRA fragment that was generated by PCR using primers [32P]MT147 and MT148.
Primer extension was performed as described previously (64). Briefly, total cellular RNA was purified from E. coli BSN29 and BSN29 containing the grlA-1 plasmid. Cells were grown to mid-log phase (OD600 = 0.6), and RNA molecules were isolated using the FastRNA Pro kit (Q-Biogene). Primer MT161 was labeled at its 5′ end with [γ-32P]ATP and T4 polynucleotide kinase and coprecipitated with 10 μg of total RNA. Hybridization was carried out at 45°C for 15 min in 10 μl of Tris-EDTA (TE) buffer containing 150 mM KCl. Primer extension reactions were started by the addition of 24 μl of extension solution (20 mM Tris-HCl [pH 8.4], 10 mM MgCl2, 10 mM DTT, 2 mM deoxynucleoside triphosphates [dNTPs], 1 U/ml AMV reverse transcriptase) and were carried out at 42°C for 60 min. Samples were then precipitated and analyzed on a sequencing gel. A GA ladder was made by using the Maxam and Gilbert method (47) to sequence a grlA fragment that was generated by PCR using primers MT157 and [32P]MT161.
The DNA fragments used for footprinting analysis were amplified by PCR using primer pairs [32P]MT130/MT160, [32P]MT147/MT148, and MT147/[32P]MT148 (for a Ler-His binding assay) and MT157/[32P]MT161 (for an H-NS-His binding assay) with plasmid pMTgrlRA as the template. For the Ler-His DNA binding control experiment, primer pair rpoDForw/[32P]rpoDRev was used to amplify a 304-bp DNA fragment corresponding to nucleotides 1200 to 1503 of the coding sequence of rpoD, using C. rodentium genomic DNA as the template. The labeled fragments were then incubated with various amounts of Ler-His or H-NS-His in 25 μl of binding buffer (10 mM Tris-HCl [pH 8.0], 50 mM NaCl, 2 mM DTT, 5% glycerol, 0.5 mg/ml BSA, 2 mM CaCl2, 1 mM MgCl2) for 15 min, after which each sample was treated with 0.5 U of DNase I (New England Biolabs) at room temperature for 30 s. The reactions were terminated by the addition of phenol. Samples were precipitated with ethanol and analyzed on a sequencing gel.
Microarray analysis of gene expression by C. rodentium has shown that transcription of the grlR and grlA genes is activated 5-fold by RegA in the presence of bicarbonate (63). To determine if the upregulation we observed in the native host was due to the direct interaction of RegA with the grlRA promoter, we performed β-galactosidase analysis. A series of transcriptional fusions to the lacZ reporter gene, encompassing different lengths of the grlRA regulatory region, were constructed using the single-copy plasmid pMU2385 (Table (Table1).1). These grlRA-lacZ fusions were designated grlRA-1, -2, -3, and -4, respectively. All four grlRA-lacZ fusions contained a common 3′ end at position +263 relative to the start site of grlRA transcription (1) but carried variable 5′ ends at positions −497, −391, −222, and −72, respectively (Fig. (Fig.11 and and22 A). Each of the constructs (grlRA-1, -2, -3, and -4) and the control plasmid pMU2385 was transformed into E. coli K-12 strain MC4100, which carries either plasmid pACYC184 (RegA− control) or pEH6 (pACYC184 expressing RegA). The expression levels of β-galactosidase by the various constructs in the RegA− and RegA+ backgrounds were assessed by measuring β-galactosidase activity following the growth of the MC4100 derivatives in LB medium in the absence or presence of 45 mM NaHCO3.
Constructs grlRA-1 and grlRA-2 exhibited similar expression patterns (Fig. (Fig.2B).2B). In the RegA− host, approximately 230 U of β-galactosidase activity was produced by grlRA-lacZ with or without NaHCO3. In the RegA+ host, the levels of expression of grlRA-1 and grlRA-2 increased marginally, ~1.6-fold, in the absence of NaHCO3, but in the presence of NaHCO3, the grlRA promoter activities from grlRA-1 and grlRA-2 showed a more pronounced enhancement (to 600 U of β-galactosidase activity), resulting in 3.0- and 3.6-fold-increased activations, respectively (Fig. (Fig.2B2B).
Relative to grlRA-1 and grlRA-2, much stronger expression (approximately 2,000 U of β-galactosidase activity) from grlRA-3 was seen in both the RegA− and RegA+ backgrounds regardless of the presence of NaHCO3 (Fig. (Fig.2B).2B). Like that on grlRA-3, the grlRA promoter carried on grlRA-4 was also highly expressed (about 2,500 U), and this construct also was not activated by RegA. β-Galactosidase analysis of the four constructs in isogenic native host strains (RegA− and RegA+) of C. rodentium showed patterns of regulation and expression similar to those observed in E. coli MC4100 (Fig. (Fig.2C2C).
Taken together, these results indicate that (i) transcription of the grlRA promoter is activated by RegA, (ii) NaHCO3 facilitates the RegA-mediated activation of grlRA transcription, (iii) the region between positions −497 and −222 contains a cis-acting element responsible for RegA activation, and (iv) the region between position −391 and −72 includes a cis-acting element involved in transcriptional repression of the grlRA promoter.
To determine if RegA binds directly to the promoter region of grlRA, an electrophoretic mobility shift assay (EMSA) was performed using purified MBP::RegA (62). Initially, two DNA fragments which extended from positions −780 to −390 and from −398 to +43 relative to the transcription start site of grlRA (Fig. (Fig.11 and and3)3) were amplified by PCR with 32P-labeled primers (see Materials and Methods). Following incubation of each fragment with various amounts of MBP::RegA and 45 mM NaHCO3 at 37°C for 20 min, the samples were analyzed on a native polyacrylamide gel. At concentrations of 140 and 280 nM, MBP::RegA was able to shift and form a stable complex with the fragment from position −398 to +43 (Fig. (Fig.3,3, panel B) but failed to bind to the fragment from position −780 to −390 (Fig. (Fig.3,3, panel A). These results suggested that the binding site for RegA was between positions −398 and +43.
To localize the RegA binding region more precisely, two sections from the −398-to-+43 fragment (positions −398 to −165 and −186 to +43) were analyzed separately. EMSA showed that, in the presence of bicarbonate, MBP::RegA was able to bind to the −398-to-−165 fragment but not to the −186-to-+43 fragment (Fig. (Fig.3,3, panels C and D, respectively). We also examined the effect of bicarbonate on the binding affinity of MBP::RegA for fragment −398 to −165. The results (Fig. (Fig.3,3, panel C) showed that, in the presence of bicarbonate, MBP::RegA was able to completely shift the DNA fragment from positions −398 to −165, forming a discrete protein-DNA complex at an MBP::RegA concentration of 280 nM. In the absence of bicarbonate, by contrast, only a small proportion of the DNA fragment was bound by MBP::RegA at the same protein concentration. The protein-DNA complex that did form appeared to be smeared, indicating weaker DNA binding by MBP::RegA in the absence of bicarbonate. The requirement for bicarbonate for efficient RegA binding and the RegA binding region located by EMSA agreed with the results obtained from the analysis of the various grlRA fragments in the β-galactosidase assay (see above).
β-Galactosidase analysis of the grlRA regulatory region indicated the presence of a cis-acting element involved in repression of the grlRA promoter (see above). To test if H-NS and its paralog StpA negatively regulate grlRA transcription, the four plasmids containing the various grlRA-lacZ transcriptional fusions (grlRA-1, -2, -3, and -4) were each transformed into isogenic E. coli K-12 strains MC4100 (wild type), PD145 (hns), and BSN29 (hns stpA), after which β-galactosidase activity was determined for each of the transformants grown in LB at 37°C.
In the case of grlRA-1, a major difference with regard to grlRA expression in the three different host strains was seen (Fig. (Fig.4).4). In MC4100 (wild type), grlRA-1 expressed 200 units of β-galactosidase, which increased 7- and 40-fold, to 1,400 and 7,800 U, in PD145 (hns) and BSN29 (hns stpA), respectively. Similar regulatory effects by H-NS and StpA were also observed for grlRA-2. By contrast, the difference in grlRA expression in the three hosts was far less for grlRA-3, for which, due to stronger expression of the grlRA promoter in MC4100 (1,800 U), the observed increase in expression was only 1.6-fold in PD145 (2,800 U) and 1.9-fold in BSN29 (3,500 U). As with grlRA-4, the H-NS- and StpA-mediated effects on grlRA expression were almost negated, as essentially the same levels of β-galactosidase were measured in the three different backgrounds. The most likely explanation for the increased transcriptional levels of grlRA-3 and -4 in the MC4100 background is the removal of the H-NS- and StpA binding sites.
The data presented here clearly demonstrated the involvement of H-NS and StpA in the repression control of the grlRA promoter and agree with the in vitro H-NS binding data reported by Barba et al. (1). By using the four grlRA-lacZ fusions, we were able to localize the region responsible for H-NS- and StpA-mediated repression to between positions −391 and −72.
To determine if Ler bound directly to the promoter region of grlRA, a DNase I footprinting assay was performed using purified Ler-His (see Materials and Methods). A 234-bp DNA fragment containing the grlRA fragment, which extended from positions −398 to −165 (relative to the transcription start site of grlRA) (Fig. (Fig.1),1), was labeled at the 5′ end of the coding strand. Following incubation in various concentrations of Ler-His (ranging from 125 nM to 1.5 μM), the reaction mixes were treated with DNase I. After ethanol precipitation, the samples were analyzed on a sequencing gel. Weak protection of DNA was seen at the Ler-His concentrations of 125 and 500 nM (Fig. (Fig.55 A). At a higher concentration of 1.5 μM, however, the region between positions −367 and −228 (relative to the transcription start site of grlRA) was strongly protected (Fig. (Fig.5A).5A). In contrast, under the same conditions, Ler-His was unable to bind to the DNA fragment of the C. rodentium housekeeping gene rpoD (Fig. (Fig.5B5B).
Although grlR and grlA have previously been shown to transcribe as a single unit (1), this operon contains a short intergenic region of 67 bp between the stop codon of grlR and the start codon of grlA (Fig. (Fig.1).1). To determine if the region upstream of the coding sequence of grlA harbored a promoter(s), we made two constructs in which the regions between positions −399 and +157 (grlA-1) and positions −107 and +157 (grlA-2) relative to the start site of grlA translation were each fused with the lacZ structural gene on plasmid pMU2385 (Table (Table22 and Fig. Fig.66 A). Constructs grlA-1 and grlA-2 along with the control plasmid, pMU2385, were each introduced into E. coli strain MC4100. In vivo analysis showed that grlA-1 produced very low levels of β-galactosidase (2.3 U), but grlA-2, which contained a shorter grlA fragment, exhibited significant promoter activity (33 U).
To test if Ler was able to activate the putative grlA promoter, we introduced plasmid pMTLer (pBR322 expressing Ler; Table Table2)2) or the control plasmid pBR322 into the MC4100 derivatives containing grlA-1 or grlA-2. While introduction of pBR322 into these strains had no effect on grlA-1 or grlA-2 expression (data not shown), the presence of pMTLer resulted in a 9-fold increase in transcriptional activity (20 U) from grlA-1 but not from grlA-2 (Fig. (Fig.6B).6B). Analysis of the effect of Ler on grlA expression in isogenic strains (Ler+ and Ler−) of C. rodentium showed regulatory effects of Ler on grlA expression similar to those seen in E. coli (Fig. (Fig.6C6C).
To test if RegA played a role in the control of the putative grlA promoter, we performed a β-galactosidase assay using E. coli MC4100 derivatives, which contained grlA-1 or grlA-2 with either pEH6 (pACYC184 expressing RegA) or pACYC184 (control). Expression of RegA caused a 3-fold increase in transcription from grlA-1, but not from grlA-2, and this RegA-mediated stimulation of grlA expression was dependent on the presence of bicarbonate (45 mM) in the culture medium (data not shown).
In summary, our results demonstrated the presence of significant promoter activity specific for the grlA gene, which was upregulated markedly by Ler and less so by RegA. By using two lacZ fusions, we were able to localize the promoter activity to a region downstream of the −107 position relative to the start codon of grlA and the Ler binding region to somewhere between positions −339 and −107. Moreover, the apparent difference in transcriptional activity between grlA-1 and grlA-2 indicated that the putative grlA promoter was subject to negative control by one or more repressors.
A single-round in vitro transcription experiment was performed to map the start site of grlA transcription. In this assay, two linear grlA fragments encompassing positions 1155 to 1353 and 1155 to 1422, were amplified using primer pairs MT146/MT144 and MT146/MT145, respectively, and used as DNA templates (Fig. (Fig.1).1). In the presence of E. coli σ70 RNA polymerase, each PCR yielded a single transcript, indicating the presence of only one promoter in the DNA fragments used (Fig. (Fig.77 A). Transcription from the templates from positions 1155 to 1353 and 1155 to 1422 produced 61/62-nucleotide and 130/131-nucleotide transcripts, respectively. The size difference of the two transcripts matched the differences in length of the two DNA templates at their 3′ ends. Based on these data, the start site of transcription for grlA was mapped to either the thymine or the adenine residue at position 1292 or 1293, which is 24 or 23 bp upstream of the putative start codon for the GrlA protein, respectively (Fig. (Fig.11).
To confirm this finding, primer extension analysis was carried out using total RNA from E. coli BSN29 (control) and BSN29 containing plasmid grlA-1. As the grlA promoter is a weak promoter, BSN29 (hns stpA) was chosen as the host strain to maximize transcription under nonrepressive conditions (by H-NS and StpA regulation of grlA; see below). The initiation of transcription of grlA was probed by using primer MT161, located 60 bp downstream of the translational start site of grlA (Fig. (Fig.7B).7B). This result agreed with that obtained from in vitro runoff transcription assay.
Inspection of the sequence immediately upstream of the start site of transcription (the adenine residue [A] is designated position +1) revealed the presence of a putative −10 (GTTAAT) region and a TGn motif (Fig. (Fig.1).1). Eighteen nucleotides upstream of the −10 sequence is a potential −35 region (TAAATA).
To determine if RegA bound directly to the promoter region of grlA, an EMSA was performed using purified MBP::RegA (62). Two DNA fragments, which extended from positions −364 to +61 and from positions +79 to +407, relative to the transcription start site of grlA (Fig. (Fig.1),1), were amplified by PCR with 32P-labeled primers (see Materials and Methods). At concentrations of 140 and 280 nM, MBP::RegA was able to shift the −364-to-+61 fragment (Fig. (Fig.8,8, panel A) but not the +79-to-+407 fragment (Fig. (Fig.8,8, panel B). In addition, MBP::RegA formed a stable complex with the −364-to-+61 fragment at a concentration of 280 nM. These results mapped the binding site for RegA to between positions −364 and +61.
To provide additional evidence for the involvement of Ler in grlA activation, a DNase I footprinting assay was performed using purified Ler-His (see Materials and Methods). A 296-bp DNA fragment containing the grlA fragment, which extended from −364 to −69 (relative to the transcription start site of grlA; Fig. Fig.1),1), was labeled in separate reactions at the 5′ ends of both the coding and noncoding strands. Following incubation in various concentrations of Ler-His (from 125 nM to 1.5 μM), the reaction mixes were treated with DNase I. After ethanol precipitation, the samples were analyzed on a sequencing gel. At Ler-His concentrations of 500 nM and 1.5 μM, the region between −320 and −114 (relative to the transcription start site of grlA) of the coding strand and the corresponding region on the noncoding strand (between −127 and −283) were protected, demonstrating that the grlA promoter region contains sequences recognized by Ler (Fig. (Fig.9).9). A Ler-dependent DNase I-hypersensitive site was observed at position −208 on the noncoding strand, suggesting that the binding of Ler to the grlA regulatory region induced a structural change in the DNA that resulted in increased cleavage by DNase I.
β-Galactosidase analysis of grlA-1 and grlA-2 indicated that grlA expression was subject to repression control (see above). As the DNA sequence surrounding the grlA promoter region is highly AT rich and is predicted by the bend.it program (http://hydra.icgeb.trieste.it/dna/index.php) to be highly curved (Fig. (Fig.1010 A), we hypothesized that H-NS and StpA are responsible for the repression. To test this, grlA-1 and grlA-2, were analyzed in strains MC4100 (wild type), PD145 (hns), and BSN29 (hns stpA). Relative to the promoter activities in MC4100 (wild type), both constructs showed enhanced grlA expression in PD145 (hns) (Fig. 10B). In BSN29 (hns stpA), further increases in grlA transcription (compared to the promoter activities in the MC4100) were observed for both grlA-1 and grlA-2. These findings confirmed the involvement of H-NS and StpA in the negative regulation of grlA expression. Furthermore, the fact that deletion of the upstream sequence in the grlA-2 construct resulted in a large increase in promoter activity in E. coli MC4100 points to the partial removal of the H-NS and StpA binding sites.
We next carried out a DNase I footprinting assay to determine the H-NS binding site for the grlA promoter. A 189-bp DNA fragment, which extended from −84 to +105 (relative to the transcription start site of grlA; Fig. Fig.1),1), was labeled at the 5′ end of the noncoding strand. The DNA fragment was incubated with various amounts of purified H-NS-His (from 100 nM to 1.6 μM) and treated with DNase I. Samples were then analyzed on a sequencing gel. At H-NS-His concentrations of 800 nM and 1.6 μM, the region between positions −48 and +73 (relative to the transcription start site of grlA; Fig. Fig.1)1) of the noncoding strand was protected, demonstrating that H-NS binds specifically to the grlA promoter region (Fig. 10C). An H-NS-dependent DNase I-hypersensitive site was observed at position −54 on the noncoding strand.
The regulatory network which controls the expression of the virulence genes of A/E pathogens is complex. Much of this regulation centers on controlling the expression of ler (from the LEE1 promoter), which encodes the master regulator of the LEE. In C. rodentium, the expression of the LEE1 promoter can be repressed by H-NS and Ler but is activated by GrlA (Fig. (Fig.11)11) (1, 14). In contrast to how it affects LEE1, Ler is responsible for the positive control of the LEE2 to LEE5 operons and activation of transcription of the grlRA operon. The reciprocal stimulation of transcription of the LEE1 and grlRA operons by GrlA and Ler, respectively, leads to the formation of a positive feedback regulatory loop (1).
Three lines of evidence indicate that RegA activates transcription of the grlRA promoter: (i) microarray analysis showed that a RegA+ C. rodentium strain produced five times more grlRA mRNA than an isogenic regA mutant (63), (ii) β-galactosidase assays using grlRA-lacZ transcriptional fusions demonstrated bicarbonate-mediated activation by RegA in E. coli K-12 and C. rodentium, and (iii) purified RegA::MBP fusion protein was able to shift a DNA fragment containing the grlRA promoter and its immediate upstream region. It appears that the binding sites for RegA, Ler, H-NS, and StpA overlap within this region in both C. rodentium (this study) and E. coli (1). Like Ler and other AraC-like virulence regulators, such as PerA (from EPEC) (22), Rns (from enterotoxigenic E. coli) (7), ToxT (from Vibrio cholerae) (4), and AggR (from enteroaggregative E. coli) (42), RegA acts as an antirepressor to overcome H-NS-mediated gene silencing (19, 56). However, the degree of RegA-mediated activation of different operons varies considerably (25). For example, RegA can stimulate transcription of the adcA and kfc operons by more than 50-fold and activate the expression of the gene encoding a dispersin-like factor by about 130-fold (25). By contrast, activation by RegA at the grlRA promoter is only 3- to 5-fold. Although we have not yet been able to identify a consensus sequence for RegA binding, data from DNase I footprinting experiments of the adcA and kfcC promoter regions demonstrated that RegA binds to AT-rich sequences and that the interaction of RegA with this region causes structural distortion of DNA (62). In the case of the grlRA promoter, although MBP::RegA bound specifically to a region between positions −398 and −165 relative to the start site of transcription of the grlRA promoter in the presence of bicarbonate (Fig. (Fig.3C),3C), a DNase I protection assay did not reveal a clear RegA footprint (data not shown), probably due to relatively weak binding of RegA to this region. Notwithstanding the moderate degree of RegA-mediated activation of the grlRA promoter itself, the contribution of RegA to the grlRA-ler regulatory circuit allows the LEE to respond to the gut-associated environmental factor bicarbonate.
Prior to this study, only a single promoter located upstream of the grlR structural gene had been identified for the expression of the entire grlRA operon. Transcription of this promoter is activated by Ler via binding to its upstream region (1). If the two proteins are synthesized from the same mRNA transcript, we would expect to find a fixed ratio between the activator (GrlA) and antiactivator (GrlR). In this work, however, we demonstrated the presence of a separate σ70 promoter specifically responsible for grlA transcription. The transcriptional start site of grlA was mapped to 23 bp upstream of the grlA translational start site (Fig. (Fig.1).1). The short intergenic region between the coding sequences of grlR and grlA (67 bp) is just long enough to harbor the components of the grlA promoter. The deduced grlA core promoter contains a putative −10 hexamer (GTTAAT) and −35 hexamer (TAAATA) and an 18-bp spacer. However, this promoter also possesses an extended −10 (TGn) motif, and its spacer is extremely AT rich (AT/GC ratio of 8:1). The TGn motif probably compensates for a poor −35 region, while the AT richness contributes to the promoter strength (5, 30, 33). As for grlRA, a number of other bacterial operons possess internal promoters for supplementary expression of downstream genes (35, 58, 59). For example, the E. coli rplKAJL-rpoBC operon, which codes for ribosomal proteins L11, L1, L10, and L7/L12, as well as the β and β′ subunits of RNA polymerase, contains two major upstream promoters, rplKp and rplJp, and two minor downstream promoters, rplLp and rpoBp (45). These internal promoters permit differential expressions of the various genes within the operon.
In an E. coli background, BSN29 (hns stpA), the grlA promoter exhibited moderately strong activity (Fig. 10B). In the wild-type E. coli strain, MC4100, however, expression of the grlA promoter was repressed more than 70-fold by H-NS and StpA. The region surrounding the grlA promoter is AT rich (68%) and highly curved (Fig. 10A) and was shown to interact specifically with H-NS in vitro. In vivo analysis showed that both H-NS and StpA proteins inhibited transcription from the grlA promoter. In E. coli MC4100, H-NS- and StpA-mediated transcriptional silencing of the grlA promoter was partially reversed by Ler. Although the transcriptional activity of the grlA promoter is relatively moderate, even when activated, it could nevertheless play a role in adjusting the ratio of GrlA to GrlR under different growth or environmental conditions. This is likely to be a factor in LEE regulation, as the ratio of intracellular levels of the activator and antiactivator will influence the dynamic operations of the Ler-GrlA regulatory loop.
Ler, a homolog of H-NS, has previously been shown to bind cooperatively to extended regions of several LEE promoters (1, 2, 24, 54). Like H-NS, Ler does not recognize a specific DNA sequence (e.g., DNA palindromes), as is the case with other classical regulators (e.g., cyclic AMP receptor protein [CRP] and AraC) but binds to AT-rich regions around its target promoters. Our DNase I protection assay (Fig. (Fig.9)9) showed that Ler bound to a large section of the grlA promoter region spanning about 200 nucleotides between positions −320 and −114 in the coding strand and −283 and −127 in the noncoding strand. These results agree with those observed in an EMSA in which Ler was shown to bind to the same grlA fragment (1). This region has been proposed as a downstream binding site for Ler-mediated activation of the grlRA promoter (1). Given that the newly identified grlA promoter is located a short distance downstream of this sequence, it is likely that it is also directly responsible for Ler-mediated expression of grlA. Expression of the grlA promoter was also upregulated by RegA in the presence of bicarbonate. Both in vivo deletion analysis and EMSA demonstrated a direct role of RegA in grlA activation.
A DNase I protection assay showed that Ler bound to the same grlRA promoter region as that determined for RegA by EMSA (Fig. (Fig.1).1). Further analysis indicated that RegA and Ler acted independently rather than synergistically to activate transcription of the grlRA promoter (data not shown). The findings of this study together with published data have allowed us to develop a model for the induction of the LEE positive regulatory loop of C. rodentium (Fig. (Fig.11).11). When C. rodentium is growing in environments outside its host where the temperature is generally lower than 37°C, transcription of the LEE1, grlRA, and grlA promoters is silenced by H-NS and StpA. Upon entering the intestinal lumen, C. rodentium encounters bicarbonate, which enhances RegA binding to DNA, leading to the expression of GrlA through RegA-mediated activation of the grlRA and grlA promoters. GrlA then activates transcription of the LEE1 promoter, resulting in enhanced expression of the Ler protein. Ler further induces GrlA production, accelerating the rate of the synthesis of both Ler and GrlA, and activates the transcription of other promoters (LEE2-LEE5) on the LEE pathogenicity island that are essential for efficient assembly of the T3SS and virulence. Once optimal levels of Ler and GrlA are reached, Ler negatively regulates its own expression and, hence, that of GrlA by repressing the LEE1 operon. Elucidation of the mechanics of this finely tuned system provides fascinating new insights into the subtle elegance and complexity of virulence gene regulation in A/E enterobacteria.
Work in our laboratory is supported by research grants from the Australian National Health and Medical Research Council and the Australian Research Council. M. Tauschek is supported by a Peter Doherty Fellowship of the Australian National Health and Medical Research Council, D. Hocking is the recipient of a Melbourne Research Scholarship, and A. Tan is the recipient of an Australian Postgraduate Award.
Published ahead of print on 14 May 2010.