Transcription of the ehx Enterohemolysin Gene Is Positively Regulated by GrlA, a Global Regulator Encoded within the Locus of Enterocyte Effacement in Enterohemorrhagic Escherichia coli

Abstract

The pathogenicity island termed locus of enterocyte effacement (LEE) encodes a type 3 protein secretion system, whose function is required for full virulence of enterohemorrhagic Escherichia coli (EHEC). GrlR and GrlA are LEE-encoded negative and positive regulators, respectively, for controlling transcription of the ler gene, which encodes a central activator of LEE gene expression. We previously reported that the GrlR-GrlA regulatory system controls not only the LEE genes but also flagellar gene expression in EHEC (S. Iyoda et al., J. Bacteriol. 188:5682-5692, 2006). In order to further explore virulence-related genes under the control of the GrlR-GrlA regulatory system, we characterized a grlR-deleted EHEC O157 strain, which was found to have high and low levels of expression of LEE and flagellar genes, respectively. We report here that the grlR deletion significantly induced enterohemolysin (Ehx) activity of EHEC O157 on plates containing defibrinated sheep erythrocytes. Ehx levels were not induced in the grlR grlA double mutant strain but increased markedly by overexpression of GrlA even in the ler mutant, indicating that GrlA is responsible for this regulation. Ehx of the EHEC O157 Sakai strain is encoded by the ehxCABD genes, which are carried on the large plasmid pO157. The expression of ehxC fused with FLAG tag or a promoterless lacZ gene on pO157 was significantly induced under conditions in which GrlA was overproduced. These results together suggest that GrlA acts as a positive regulator for the ehx transcription in EHEC.

Enterohemorrhagic Escherichia coli (EHEC) strains cause hemorrhagic colitis and hemolytic-uremic syndrome (HUS) in humans. Although the most relevant virulence factor for disease development is Shiga toxin, there are other virulence factors associated with the incidence of HUS (reviewed in references 14, 25, and 34). The initial attachment of EHEC to the host intestinal epithelial cells requires cooperative activities of several virulence factors. The locus of enterocyte effacement (LEE), a pathogenicity island encompassing 41 genes is commonly found on the chromosomes of enteropathogenic E. coli, EHEC, and Citrobacter rodentium (30). LEE encodes the adhesion factor intimin, intimin receptor (Tir), a type 3 secretion system (T3SS), and regulators to control expression of those genes (reviewed in references 14, 25, 31, and 34). Expression of the LEE genes is under the control of several regulatory elements (8, 15, 17, 18, 22, 23, 33, 37, 42-46, 48, 49, 51, 57). At least three regulators, Ler, GrlR, and GrlA, are encoded within the LEE. Ler acts as a central activator for expression of most of the LEE genes (31). The grlR and grlA genes comprise a probable operon, which is under the positive control of Ler (2, 13). GrlR and GrlA are negative and positive regulators, respectively, for controlling ler transcription (12, 27). GrlR interacts with GrlA and thereby inhibits GrlA function, suggesting that GrlR acts as an anti-GrlA factor (10, 23, 24). In the grlR-deleted mutant strain, free GrlA activates ler transcription, which in turn induces expression of all LEE genes, including the grlR operon. Therefore, a positive regulatory loop governs expression of LEE genes (2). Intracellular protein levels of GrlR are regulated in a growth-phase-dependent manner; GrlR levels are minimal as the EHEC cells enter the stationary phase when LEE gene expression is induced (23).

We have recently shown that the grlR deletion resulted in a nonmotile phenotype in the EHEC O157 Sakai strain (20). This phenotype was not observed in the grlR grlA double mutant strain but was markedly enhanced when GrlA was overexpressed even in the ler mutant strain. Transcription of the master flagellar operon (flhD) was reduced in the grlR mutant background, indicating that GrlA acts as a negative regulator of the flagellar regulon (20). Consistent with these results, constitutive expression of the flhD operon abrogated adhesion of EHEC to cultured epithelial cells (20). Therefore, GrlA-dependent repression of the flagellar regulon can be considered to be important for the efficient adhesion of EHEC to host cells.

In the present study, we explored other virulence-related genes under the control of the GrlR-GrlA regulatory system by further characterizing the grlR mutant phenotype. We found that the grlR mutant showed a hyperhemolytic phenotype. Beutin et al. (5) found a novel hemolytic activity of EHEC called enterohemolysin (designated EHEC-Hly [39] or Ehx [3]) produced by a high proportion of Shiga-toxin-producing E. coli strains. The phenotype of Ehx is in contrast to that of alpha-hemolysin (Hly), an important virulence factor of extraintestinal E. coli (54), in spite of the high sequence homology shared between them (41). EHEC O157 strains producing Ehx are not hemolytic on standard blood agar plates but produce small and turbid hemolytic zones on plates containing defibrinated sheep erythrocytes and calcium, which is not required for Hly assay on blood agar plates (5).

Ehx of EHEC O157 is encoded on a large plasmid (designated pO157) on which genes encoding several potential virulence factors have been identified (7, 28, 50). The gene organization of Ehx is the same as that of Hly in that each consists of a single operon containing four genes, hly/ehxCABD (39, 55). hlyA (ehxA) is the structural gene of Hly (Ehx). HlyC (EhxC) is a modifying factor that converts the inactive hemolysin into an active form by the addition of a fatty acid group (19). The specific secretion machinery for HlyA (EhxA) is encoded by hlyB (ehxB) and hlyD (ehxD) (53). Expression of Hly is under the control of several regulators (1, 16, 36).

Here, we provide evidence that the LEE-encoded regulator GrlA activates transcription of ehx genes. We propose a regulation model in which the GrlR-GrlA regulatory system not only controls the expression of the LEE and flagellar gene expression but is also required for expression of virulence-related genes, including Ehx, in EHEC O157.

MATERIALS AND METHODS

Bacterial strains, plasmids, and media.

The bacterial strains and plasmids used in this study are summarized in Table 1. SKI-5142 is a Lac-negative derivative of the EHEC O157:H7 Sakai strain (23) and was used as the wild-type strain in this study. SKI-5143, -5152, -5153, and -5154 are the ler, grlR, grlA, and grlR grlA mutant derivatives of SKI-5142, respectively, as described previously (23). Bacteria were grown in L broth (LB) or tryptic soy broth, and ampicillin (100 μg/ml), chloramphenicol (25 μg/ml), and kanamycin (50 μg/ml) were added as required. Motility agar plates were prepared as described previously (20). Enterohemolysin test (EHT) agar plates (Kyokuto, Japan) containing 5% defibrinated sheep red blood cells and 10 mM CaCl₂ were used to examine Ehx activity. A one-step inactivation method (11) was used to construct SKI-5159, -5169 through -5174, -5211, -5221, and -5231 through -5233 as described previously (20, 22, 23). PCR primer sets (shown in Table S1 in the supplemental material or as described previously [23]) to insert Δ(escN)::kan, Δ(ehxA)::kan, Δ(ehxCABD)::kan, Δ(grlR)::cat, Δ(grlA)::cat, Δ(grlR-grlA)::cat, Δ(grlR)::kan, Δ(grlA)::kan, and Δ(grlR-grlA)::kan deletions into each strain were ESCN-P1 and ESCN-P2, EHXA-P1 and EHXA-P2, EHXCABD-P1 and EHXCABD-P2, ORF10P5 and ORF10P6, ORF11P5 and ORF11P6, ORF10P5 and ORF11P6, ORF10-P7 and ORF10-P8, ORF11-P7 and ORF11-P8, and ORF10-P7 and ORF11-P8, respectively. A modified Lambda red recombinase-mediated method (52) was used for constructing ehxA-FLAG and ehxC-FLAG fusion strains, designated SKI-5210 (using primers EHXA-FG-F and EHXA-FG-R) and SKI-5220 (EHXC-FG-F and EHXC-FG-R), respectively. The transcriptional ehxC′-′lacZ fusion on pO157 in the SKI-5142 strain was constructed as follows. To construct the ehxC′-′lacZ strain, SKI-5230, a promoterless lacZ gene with a ribosome binding site flanked by 45 bp of homology to the 5′ terminus of ehxC, was amplified from pRL124 (29) with primers EHXC-LACZf1 and LACZ-FRTr1, and the cat gene flanked by 45 bp of homology to the 3′ terminus of ehxC was amplified from pKD3 with LACZ-FRTf1 and EHXC-FRTr1 primers. These PCR products were purified, mixed, and used as templates for a second PCR with primers EHXC-LACZf1 and EHXC-FRTr1. The resulting PCR products containing ehxC′-′lacZ-cat-′ehxC were purified and used to replace the ehxC gene on pO157.

TABLE 1.

Bacterial strains and plasmids used in this study

Strain or plasmid	Genotype or relevant phenotype	Source or reference
Strains
SKI-5142	EHEC O157:H7 Sakai Δ(lacZYA)	23
SKI-5143	SKI-5142 Δ(ler)::kan	23
SKI-5152	SKI-5142 Δ(grlR)::cat	23
SKI-5153	SKI-5142 Δ(grlA)::cat	23
SKI-5154	SKI-5142 Δ(grlR-grlA)::cat	23
SKI-5159	SKI-5143 Δ(grlR)::cat	This study
SKI-5169	SKI-5142 Δ(escN)::kan	This study
SKI-5170	SKI-5152 Δ(escN)::kan	This study
SKI-5171	SKI-5142 Δ(ehxA)::kan	This study
SKI-5172	SKI-5142 Δ(ehxCABD)::kan	This study
SKI-5173	SKI-5152 Δ(ehxA)::kan	This study
SKI-5174	SKI-5152 Δ(ehxCABD)::kan	This study
SKI-5210	SKI-5142 ehxA-FLAG; Kmr	This study
SKI-5211	SKI-5210 Δ(grlR)::cat	This study
SKI-5220	SKI-5142 ehxC-FLAG; Kmr	This study
SKI-5221	SKI-5220 Δ(grlR)::cat	This study
SKI-5230	SKI-5142 ehxC-lacZ; Cmr	This study
SKI-5231	SKI-5230 Δ(grlR)::kan	This study
SKI-5232	SKI-5230Δ(grlA)::kan	This study
SKI-5233	SKI-5230Δ(grlR-grlA)::kan	This study
EDL933	EHEC O157:H7, isolated in 1982, United States	STEC Center
93-111	EHEC O157:H7, isolated in 1993, United States	STEC Center
OK-1	EHEC O157:H7, isolated in 1996, Japan	STEC Center
86-24	EHEC O157:H7, isolated in 1986, United States	STEC Center
2886-75	EHEC O157:H7, isolated in 1975, United States	STEC Center
493/89	EHEC O157:H−, isolated in 1989, Germany, sorbitol fermenting	STEC Center
E32511	EHEC O157:H−	STEC Center
G5101	EHEC O157:H7, isolated in 1995, United States	STEC Center
Plasmids
pGEM-self	Self-ligated pGEM-T Easy vector (Promega)	23
pGEMEXCABD	pGEM-T Easy-ehxCABD	This study
pACYC184	Cloning vector; Cmr Tcr	9
pACGA	pACYC184-grlA	20
pBR322C	pBR322; Aps Cmr Tcr	22
pBRCGA	pBR322C-grlA	This study
pMW119	pSC101-ori, cloning vector; Apr	Nippon Gene
pMW119GA	pMW119-grlA	This study
pRL124	pMB1-ori, promoter cloning vector, Apr	29
pRLLER	pRL124-ler promoter region	23
pRLEHXC1	pRL124-ehxC promoter region	This study
pRLEHXC2	pRL124-ehxC promoter region	This study
pET22b(+)	T7 promoter expression vector, His tag	Novagen
pET22bGA	pET22b(+)-grlA	This study

Construction of plasmids.

pACGA and pRLLER were described previously (20, 23). A 7.44-kbp PCR product containing ehxCABD amplified with the EHXCABD-BHI-F and EHXCABD-STU-R primer set was inserted into the pGEM-T Easy vector (Promega) to yield pGEME-XCABD. Plasmids pBRCGA and pMW119GA were constructed by cloning grlA amplified with ORF11BHI and ORF11STU primers (23) into the BamHI-PvuII sites of pBR322C (22) and BamHI-SmaI sites of pMW119 (Nippon Gene), respectively. Plasmids pRLEHXC1 and pRLEHXC2 were constructed by cloning the putative promoter region (−564 to +22 and −258 to +22, respectively, relative to the ehxC initiation codon) amplified with primers EHXCP-KPN1 and EHXCP-ERI1 and EHXCP-KPN2 and EHXCP-ERI1, respectively, into the KpnI-EcoRI sites of the promoter cloning vector, pRL124 (29). A 0.46-kbp PCR product containing grlA, amplified with the ORF11-NDEI and ORF11-XHOI-22B primers, was digested with NdeI and XhoI and inserted into the corresponding sites of pET22b(+) (Novagen) to yield pET22bGA.

Western blotting.

Bacteria were grown in LB with or without antibiotics at 37°C with shaking until they reached an optical density at 600 nm of 0.8. Proteins in whole-cell lysates were analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and Western blotting as described previously (21-23). Western blotting was performed with monoclonal anti-FLAG (clone H2; Sigma), polyclonal anti-H7 (FliC) (22), or anti-EspB (22) antibodies to detect EhxC/A-FLAG, flagellin, or EspB, respectively, in whole-cell lysates as described previously. All assays were performed in duplicate or triplicate and were repeated at least three times.

Assay of β-galactosidase activity.

β-Galactosidase activity was assayed as described previously (21, 26). Bacteria grown in LB or Dulbecco's modified Eagle's medium (Invitrogen) with or without antibiotics at 37°C with shaking were harvested at an optical density at 600 nm of 0.8. All assays were performed in duplicate or triplicate and were repeated at least three times.

Motility assay.

Motility of bacterial cells was measured by the spreading of colonies on motility agar plates as described previously (20).

Ehx assay.

Ehx activity on EHT plates was measured as follows. Three microliters of bacteria cultured in LB without shaking at 37°C was spotted onto an EHT plate. After incubation for 18 h at 37°C with 5% CO₂, hemolytic activity on the EHT plate was determined by comparing the sizes of clear lysis zones observed as outer circles and the clearness of inner circles where bacteria were grown. Quantitative hemolytic activity was measured by the following method using defibrinated sheep red blood cells: EHEC O157 strains grown in 1.5 ml LB with shaking at 37°C were harvested at an optical density at 600 nm of 1.0. Bacterial cells were washed once with phosphate-buffered saline (PBS) and suspended in 900 μl of PBS. Seventy microliters of washed bacteria was mixed with an equal volume of erythrocyte solution (containing 5% defibrinated sheep red blood cells and 10 mM CaCl₂ in tryptic soy broth) in a 96-well plate, and centrifuged at 1,000 × g for 5 min. The plate was incubated for 18 h at 37°C with 5% CO₂. Cells were repelleted by centrifugation, and the supernatants were monitored for hemoglobin release by measuring the optical density at 540 nm. Supernatants from bacterium-free erythrocyte solution were used to provide a baseline level of hemolysis. The measured value at an optical density at 540 nm was corrected by dividing by the number of bacterial cells (see below), and the relative activity of hemolysis was calculated when the corrected activity of wild-type cells (SKI-5142) was 100%. The numbers of bacterial cells in each well were counted after plating serially diluted mixtures onto LB agar plates.

RNA preparation and RT-PCR analysis.

Bacteria grown in LB with or without antibiotics at 37°C with shaking were harvested at an optical density at 600 nm of 0.8. Total RNA was extracted from these cells using Isogen (Nippon Gene) or a Qiagen RNeasy kit (Qiagen) according to the manufacturer's instructions. A 1.0-μg amount of total RNA was used to synthesize cDNA with Transcriptor first-strand cDNA synthesis kit (Roche) as described by the manufacturer. The resulting cDNA was used for reverse transcription (RT)-PCR with gene-specific primer sets to amplify ehxD (primers EHXD-RT-F and EHXD-RT-R), ehxC (EHXC-RT-F and EHXC-RT-R), and 16S rRNA genes (16S-RT-F and 16S-RT-R) (see Table S1 in the supplemental material). Band intensities of PCR products were measured by using the NIH ImageJ program (developed at the U.S. National Institutes of Health and available at http://rsb.info.nih.gov/nih-image/) and are shown as percentages of intensity with standard deviations of each band (wild type = 100%).

RESULTS

The grlR mutant shows a hyperhemolytic phenotype on optimized blood agar plates.

During further studies characterizing the grlR mutant strain of EHEC O157 Sakai, we noticed that the grlR mutant showed a hyperhemolytic phenotype on an optimized hemolysin agar plate (EHT plate; see Materials and Methods). Although most EHEC O157 strains have only weak hemolytic activity, unlike Hly-mediated hemolysis (5), the grlR mutant of O157 Sakai created large and clear lysis zones on EHT agar plates containing defibrinated sheep erythrocytes; clear outer circles represented hemolytic activity, and the inner circles showed bacterial growth on the EHT plate (Fig. 1A). The effect of the grlR deletion on LEE and flagellar gene expression was shown to be mediated through GrlA function (20, 23). To know whether GrlA is also responsible for the hyperhemolytic phenotype of the grlR mutant strain, we examined the hemolytic phenotype of grlA and grlR grlA mutant strains. Since the hemolytic activity of these strains was comparable to that of the wild-type strain (Fig. 1A), it appears that constitutively active GrlA in a GrlR-depleted strain is responsible for the hyperhemolytic phenotype, as seen in the regulation of LEE and flagellar gene expression (20, 23). We also performed a quantitative assay of hemolytic activity as described in Materials and Methods. Hemolytic activity of the grlR mutant was almost five times higher than those of the wild-type, grlA, and grlR grlA strains (Fig. 1B). As described previously, Ehx is encoded by the ehxA gene, which belongs to the ehxC operon (comprising ehxCABD genes) located on the large plasmid, pO157 (39). We therefore constructed ehxA and ehxCABD mutants with or without a grlR deletion and examined their hemolytic phenotypes. In these strains, no clear lysis zones were observed and the clearness of inner circles where bacteria were grown was lower than that of the wild-type strain, even when grlR was deleted. These results suggest that both ehxA and ehxCABD deletions completely abolished the hemolytic phenotype of the EHEC O157 Sakai strain on EHT plates (Fig. 1C) (data not shown). Complementation of the ehxCABD mutant with a plasmid carrying ehxCABD (pGEMEHXCABD) restored the enterohemolytic phenotype (Fig. 1C), indicating that Ehx is indispensable for the GrlA-dependent hyperhemolytic phenotype. We further examined the effect of T3SS on the hyperhemolytic phenotype of the grlR mutant because LEE-encoded type 3 translocator proteins such as EspB and EspD localized at the tip of the needle structure of T3SS has been hypothesized to have pore-forming activity in the host cell membrane (14). Therefore, high expression levels of LEE-encoded EspB and EspD in the grlR mutant may contribute to the hyperhemolytic phenotype of the grlR mutant. To rule out this possibility, we constructed a deletion mutant of escN, which encodes ATPase of the T3SS and its activity is indispensable for the secretion of LEE-encoded translocator and effector proteins. A defect of T3SS did not affect the hyperhemolytic phenotype of the grlR mutant on EHT agar plates (Fig. 1D). These results indicate that Ehx expression is under the control of the GrlR-GrlA system, and upregulation of LEE-encoded T3SS is not involved in the hyperhemolytic phenotype of the grlR mutant on the EHT plate. Since the GrlA-dependent repression of the flagellar regulon occurs without Ler (20), the central activator for LEE gene expression (32), we examined whether Ler is involved in the GrlA-dependent upregulation of Ehx expression. For this purpose, we compared the hemolytic phenotype of the ler mutant with the ler grlR double mutant strain (Fig. 1E) and the ler mutant carrying a control vector with the same strain harboring the plasmid pACGA (carrying full-length grlA) (Fig. 1E). A hyperhemolytic phenotype was observed in the ler grlR mutant strain and the ler strain with a GrlA-expressing plasmid, indicating that GrlA can induce Ehx activity without Ler function.

FIG. 1.

Ehx expression is under the control of the GrlR-GrlA regulatory system. Shown are the hemolytic phenotypes of EHEC O157 Sakai on an EHT plate. (A) Wild-type (SKI-5142) and grlR (SKI-5152), grlA (SKI-5153), and grlR grlA (SKI-5154) strains. (C, left panel) ehxA (SKI-5171) and ehxA grlR (SKI-5173) strains. (C, right panel) ehxCABD (SKI-5172) with control vector (pGEM-self) and ehxCABD (SKI-5172) with ehxCABD-positive plasmid (pGEMEHXCABD). (D) escN (SKI-5169) and escN grlR (SKI-5170) strains. (E, left panel) ler (SKI-5143) and ler grlR (SKI-5159) strains. (E, right panel) Strain SKI-5143 with vector (pACYC184) or the grlA+ (pACGA) strain. The inner and outer circles shown on each photograph indicate grown bacteria with or without Ehx activity and hemolytic lysis zones created by the activity of Ehx, respectively. (B) Relative hemolytic activity (wild type = 100%) of the same strains used in panel A.

Expression of Ehx is positively regulated by GrlA.

To examine the effect of GrlA on the expression level of Ehx, we constructed translational fusions of ehxA and ehxC with a FLAG tag at the 3′ terminus of each gene on pO157 and examined the expression levels of EhxC and EhxA using anti-FLAG monoclonal antibody. The levels of expression of both EhxC-FLAG and EhxA-FLAG were markedly induced in the grlR-deleted strain in comparison to those in the wild-type strain (Fig. 2A and B), although the intensities of nonspecifically cross-reacted bands did not change. Furthermore, complementation of the ehxC-FLAG and ehxA-FLAG strain with a GrlA-expressing plasmid, pACGA or pBRCGA, induced high-level expression of EhxC-FLAG and EhxA-FLAG compared to those of the same strain transformed with a control vector (pACYC184 or pBR322C, respectively), confirming that GrlA upregulates expression levels of EhxC and EhxA (Fig. 2B). We further confirmed expression levels of LEE (as EspB protein levels) and flagella (as expression levels of FliC). EspB and FliC were activated and repressed, respectively, in the same protein sample of whole-cell cultures (Fig. 2B). These results together suggest that GrlA upregulates Ehx and LEE levels but simultaneously downregulates flagellar gene expression.

FIG. 2.

Expression of EhxA-FLAG and EhxC-FLAG. The amounts of EhxA-FLAG (A) and EhxC-FLAG, EspB, and FliC (B) in whole-cell lysates were examined by Western blotting using monoclonal anti-FLAG and polyclonal anti-EspB and anti-FliC antibodies, respectively. Wild-type (SKI-5142), grlR (SKI-5152), SKI-5142/ pACYC184, pACGA (pACYC184-grlA⁺), pBR322C, and pBRGA (pMR322C-grlA⁺) strains were used. The positions of nonspecifically reacting protein bands served as loading control are indicated.

Transcription of the ehxC operon is positively regulated by GrlA.

We next examined whether GrlA-dependent upregulation of Ehx expression affects transcription of the ehxC operon. To this end, we cloned the putative promoter region of ehxC into the promoter-probe vector, pRL124, to construct a transcriptional fusion of ehxC with a promoterless lacZ gene. β- Galactosidase activity mediated by the ehxC promoter (on plasmids pRLEHXC1 and pRLEHXC2) was higher than that of the promoterless control vector (pRL124) in the wild-type strain (<50 Miller units), indicating that both plasmids contained a functional ehxC promoter. Because the activity of the ehxC promoter on pRLEHXC2 was approximately twofold higher than that of pRLEHXC1 (Fig. 3A), the sequences between −564 and −259 relative to the translational start site of EhxC may contain cis elements necessary for negative regulation of ehxC transcription. Activities of the ehxC promoter (on pRLEHXC1 and pRLEHXC2) in a grlR mutant were approximately twofold higher than those in the wild-type strain (Fig. 3A). This twofold induction was also observed when the wild-type strain harboring pRLEHXC1 or pRLEHXC2 was transformed with a multicopy plasmid carrying the grlA gene (pACGA) (Fig. 3A). As a positive control, we examined the promoter activity of ler on pRL124 (pRLLER) in the same strains. β-Galactosidase activity mediated by the ler promoter in a grlR-deleted strain was 2.4-fold higher than that in the wild-type strain and was induced 3.2-fold by the introduction of plasmid pACGA (Fig. 3A).

FIG. 3.

Transcriptional regulation of the ehxC operon. (A) β-Galactosidase activity from ehxC and ler promoters on the promoter-probe vector pRL124 in the following strains cultured in LB: wild type (SKI-5142), grlR (SKI-5152), wild type with pACYC184 or pACYC184-grlA⁺ (pACGA). (B) β-Galactosidase activity from ehxC-lacZ transcriptional fusion constructed on pO157 in the following strains cultured in LB and DMEM: the wild type (SKI-5142), grlR (SKI-5152), grlA (SKI-5153), and grlR grlA (SKI-5154) mutants; and the wild type with vector (pMW119) or grlA⁺ (pMW119GA). The error bars indicate standard deviations.

As pRL124 is a multicopy plasmid, the effect of GrlA on the transcription levels of the ehxC operon may have been underestimated. To test this idea, we constructed a transcriptional lacZ fusion with endogenous ehxC on pO157 by a modified Lambda red recombinase-mediated recombination method as described in Materials and Methods. The expression of the pO157-encoded ehxC-lacZ fusion was induced more than fourfold in the grlR mutant cultured in LB, compared to that in the wild-type strain (Fig. 3B). When cells were grown in DMEM, basal activities of the ehxC-lacZ strain were slightly higher than in LB, and the activity in the grlR strain was 2.2-fold higher than that in the wild-type strain. grlA and grlR grlA mutants showed slightly lower activities than those in the wild-type strain cultured in both media (Fig. 3B). Consistent with these results, the strain with a low-copy-number plasmid carrying grlA (pMW119GA) increased the transcription level of ehxC more than threefold in LB culture, compared to the wild-type strain carrying a control vector (Fig. 3B). These results suggest that GrlA upregulates the transcription of ehxC, which leads to the hyperhemolytic phenotype of EHEC O157.

To further confirm these results, we performed semiquantitative RT-PCR analysis to measure transcription of ehxC and ehxD in the different mutant backgrounds. The intensities of PCR-amplified bands of ehxC were slightly induced in a grlR strain compared to that in wild-type, grlA, and grlR grlA strains (Fig. 4). Consistent with this, the amounts of reverse-transcribed ehxD in the absence of GrlR were 2.3-fold higher than those in other strains (Fig. 4). These results confirm that transcription of the ehxC operon is positively regulated by GrlA.

FIG. 4.

Semiquantitative RT-PCR analysis of ehxC and ehxD transcripts. (A) Transcriptional levels of ehxC (253 bp), ehxD (240 bp), and 16S rRNA (476 bp) were measured by semiquantitative RT-PCR using purified total RNA of the following strains: the wild type (SKI-5142) and grlR (SKI-5152), grlA (SKI-5153), grlR grlA (SKI-5154), and ehxCABD (SKI-5172) mutants. Control experiments without using reverse transcriptase (−RT) with total RNA of SKI-5152 are indicated in the upper left two panels. Data are representative of one of at least three independent experiments. (B) Relative band intensity of the RT-PCR products in panel A analyzed with the NIH ImageJ program (wild type = 100%). Data are means of at least three independent experiments with standard deviations.

GrlA-dependent regulation of Ehx is universal in EHEC O157 strains.

We examined whether GrlA-dependent upregulation of Ehx activity is universally applicable in other EHEC O157 strains. For this purpose, we used EHEC O157 reference strains (obtained from the STEC Center at Michigan State University) and complemented these strains with pACYC184 or pACGA to see whether the GrlA-dependent hyperhemolyic activity can be observed in these strains. We confirmed that Ehx activities of all of the strains tested were induced by the presence of a GrlA-expressing plasmid, although the effect in strain E32511 was not as prominent as in the other strains (Fig. 5). The sorbitol-fermenting O157 strain, 493/89, carrying the control vector did not produce any Ehx and had a phenotype similar to the ehxA mutant of the Sakai strain as described previously (6). However, the grlA-expressing plasmid significantly induced the phenotype, suggesting that Ehx is not defective but steady-state expression levels are somehow quite low in that strain. These results suggest that the GrlA-dependent hyperhemolytic phenotype is universal in EHEC O157 strains.

FIG. 5.

GrlA-dependent hyperhemolytic phenotype in various EHEC O157 strains. Shown are the hemolytic phenotypes of EHEC strains from the STEC Center with the control vector (pACYC184) or the grlA⁺ plasmid (pACGA) on EHT plates.

DISCUSSION

Previous studies have shown that GrlA upregulates LEE gene expression via activation of ler transcription (2, 12). Our recent study showed that GrlA also acts as a negative regulator of the flagellar genes without Ler function (20). In addition to these observations, we demonstrate here that Ehx expression is under positive control by GrlA. This GrlA-dependent upregulation of Ehx also occurred in the ler mutant strain, indicating that GrlA can induce Ehx expression without Ler function, as shown for the GrlA-dependent flagellar gene repression. As the grlA mutant did not show significant reduction of Ehx activity compared to that of the wild-type strain, even though we used ehxC-lacZ transcriptional fusion constructed on pO157, hyperexpression of Ehx can be observed only when the intracellular concentration of GrlR is low (23).

Comparison of the upstream regulatory regions of ler, ehxC, and flhD, putative target genes of GrlA, did not reveal any obvious common regulatory sequences (data not shown). Additionally, a previous study did not show direct binding of GrlA to the promoter region of the ler gene (2). We also purified GrlA-His₆ recombinant proteins (overproduced from the plasmid pET22bGA) and examined their DNA binding to the transcriptional regulatory region of several possible target genes. However, we have not observed specific DNA binding activity of GrlA (data not shown), implying that the GrlA protein might have been inactivated during the purification process. Alternatively, GrlA may require another factor(s) for the direct activation of ehxC transcription.

Hly expression has shown to be negatively regulated by several regulators, including nucleoid proteins such as H-NS and Hha (16, 36). If GrlA does not directly activate transcription of ehxC as discussed above, these nucleoid proteins may be possible targets of GrlA. Like several other transcriptional regulators counteracting H-NS-mediated silencing at each promoter (reviewed in reference 35), GrlA may overcome the repressive effect of nucleoid proteins to induce ehx transcription. Further study to investigate roles of nucleoid proteins and GrlA on the regulation of ehxC transcription is now in progress in our laboratory.

We have shown that the hyperhemolytic phenotype was observed in the ler grlR mutant but not in the grlA grlR mutant. These results indicate that GrlA is indispensable for the hyperhemolytic phenotype of the grlR mutant. Although the transcription of the grlR operon, consisting of grlR and grlA, has been shown to be under the control of Ler (2, 13), basal GrlA activity or expression may be induced in the ler grlR mutant. The mechanisms of Ler-independent expression and/or activation of GrlA in the grlR mutant remain to be elucidated.

We have shown that constitutive expression of the flagellar regulon abrogated adhesion of EHEC to HeLa cells (20). Although the molecular mechanism of this regulation remains to be elucidated, these results suggest the existence of a fine-tuning system that prevents simultaneous expression of flagella and LEE genes. Given that bacterial flagella have been shown to stimulate the host innate immune system (4, 58), avoidance of coexpression of both sets of genes is an important strategy for the efficient infection of host cells by EHEC. However, the ehxCABD deletion and the constitutive expression of EhxCABD by introducing a multicopy plasmid carrying ehxCABD did not affect the expression levels of LEE and flagellar genes as well as the adhesion of bacteria to cultured epithelial cells (S. Iyoda and Y. Lu, unpublished data). Accordingly, the biological significance of these coregulations remains to be clear.

The central regulators HilA and SsrB for T3SS encoded on Salmonella pathogenicity islands 1 and 2 (SPI-1 and SPI-2), respectively, are also required for activation or repression of several non-SPI-1 and -SPI-2 genes, respectively (47, 56). These are examples of virulence-related genes that are coordinately or exclusively regulated by the same regulator. Further studies of the molecular regulatory mechanisms of GrlA-dependent gene expression as well as identification of the GrlA regulon will be necessary to elucidate the biological significance of coordinate or reciprocal regulation of EHEC virulence gene expression.

The role of Hly in the pathogenesis of extraintestinal E. coli infection has been well documented (54). However, the involvement of Ehx in human diseases caused by EHEC infection is still unclear, although ehx genes are frequently found in EHEC strains (39, 40). One plausible explanation for the role of Ehx in EHEC pathogenesis is to acquire hemoglobin as a source of iron released from erythrocytes, for better EHEC growth in the host. Previous studies suggest a link between Ehx and disease development by EHEC infection. For example, 19 out of 20 sera sampled from HUS patients reacted with Ehx, compared with only 1 serum of 20 from age-matched controls (39); 16 out of 18 eae (the LEE gene encoding intimin)-positive EHEC O111:H⁻ strains from patients with HUS were positive for Ehx production, while only 4 of 18 from patients with diarrhea had this phenotype (40). GrlA-dependent upregulation of Ehx expression was observed in various EHEC O157 strains, as shown in Fig. 5. Furthermore, using the same strain sets, we also confirmed that bacterial motility mediated by flagella is repressed by the presence of multiple copies of GrlA in all strains tested, except 493/89 and E32511, both of which were originally nonmotile (data not shown). As shown in a previous study, the presence of Ehx was highly associated with that of the eae gene (38). Therefore, control of Ehx and flagellar expression by the GrlR-GrlA regulatory system may also be universal in most LEE-positive EHEC strains other than those of serogroup O157.