ABSTRACT
Atypical enteropathogenic Escherichia coli (aEPEC) causes endemic diarrhea, diarrheal outbreaks, and persistent diarrhea in humans, but the mechanism by which aEPEC causes disease is incompletely understood. Virulence regulators and their associated regulons, which often include adhesins, play key roles in the expression of virulence factors in enteric pathogenic bacteria. In this study we identified a transcriptional regulator, RalR, in the rabbit-specific aEPEC strain, E22 (O103:H2) and examined its involvement in the regulation of virulence. Microarray analysis and quantitative real-time reverse transcription-PCR demonstrated that RalR enhances the expression of a number of genes encoding virulence-associated factors, including the Ral fimbria, the Aap dispersin, and its associated transport system, and downregulates several housekeeping genes, including fliC. These observations were confirmed by proteomic analysis of secreted and heat-extracted surface-associated proteins and by adherence and motility assays. To investigate the mechanism of RalR-mediated activation, we focused on its most highly upregulated target operons, ralCDEFGHI and aap. By using primer extension, electrophoretic mobility shift assay, and mutational analysis, we identified the promoter and operator sequences for these two operons. By employing promoter-lacZ reporter systems, we demonstrated that RalR activates the expression of its target genes by binding to one or more 8-bp palindromic sequences (with the consensus of TGTGCACA) located immediately upstream of the promoter core regions. Importantly, we also demonstrated that RalR is essential for virulence since infection of rabbits with E22 carrying a knockout mutation in the ralR gene completely abolished its ability to cause disease.
INTRODUCTION
Enteropathogenic Escherichia coli (EPEC) is among the most important diarrheal pathogens infecting children and is also a major cause of persistent diarrhea and diarrhea-associated mortality (1, 2). The hallmark of EPEC pathogenicity is colonization of the intestine accompanied by the formation of characteristic “attaching-and-effacing” (A/E) lesions on the surface of intestinal epithelial cells (3, 4). A 35-kb chromosomal pathogenicity island, termed the locus of enterocyte effacement (LEE), encodes the genetic determinants required for A/E lesion formation (5–8).
EPEC can be further categorized into two subgroups, typical EPEC (tEPEC) and atypical EPEC (aEPEC) (9), which differ from each other in terms of their genetic characteristics, serotypes, and virulence factors (10, 11). By definition, all EPEC strains carry the LEE, but tEPEC strains also carry an EPEC adherence factor plasmid (pEAF), which encodes a type IV-like bundle-forming pilus (Bfp) that facilitates the adherence of tEPEC cells to the intestinal mucosa and allows them to form microcolonies on epithelial cells in vitro and in vivo (11). Apart from Bfp, pEAF also encodes two transcriptional activators, PerA and PerC. The former activates transcription of bfpA (the gene for the major structural subunit of Bfp), while the latter stimulates the transcription of several LEE-encoded genes, including those for a type III secretion system (T3SS) (12). In contrast, the EAF plasmid is absent from aEPEC.
The epidemiology of EPEC infection has shifted during the past 20 years. Recent data suggest that aEPEC is more prevalent than tEPEC in most developing countries and is also associated with childhood diarrhea in developed countries. Several clinical and epidemiological studies have shown that aEPEC causes diarrhea in areas where these organisms are endemic, diarrheal outbreaks, and persistent diarrhea (10, 13–15), but the mechanism by which aEPEC causes diarrhea is incompletely understood. Importantly, although aEPEC strains lack the EAF plasmid, they are able to cause disease, unlike tEPEC strains, which become markedly attenuated when they are “cured” of this plasmid (16). These observations suggest that aEPEC must produce colonization factors and regulators that compensate for the absence of Bfp, PerA, and PerC, but the identities of these virulence factors are yet to be determined.
In contrast to tEPEC, which is found exclusively in humans, some aEPEC strains are important pathogens of animals. For example aEPEC has been isolated from diarrheic food production animals, such as cattle, sheep, goats, pigs, poultry, and rabbits and from companion animals (dogs and cats) (reviewed in reference 17). One such group of aEPEC that has been well investigated is rabbit-specific aEPEC (REPEC), which is a leading cause of diarrhea in young rabbits (18). REPEC causes an illness in rabbits that closely resembles that caused by EPEC in humans in terms of its clinical and pathological features (19, 20). Thus, REPEC infection of rabbits serves as a valuable model of human infection with EPEC and has been used to establish the contribution of LEE-encoded proteins to virulence (19, 21–23).
We recently reported that a non-LEE-encoded, AraC-like regulatory protein, RegR, controls the expression of a series of accessory adhesins that significantly enhance the virulence of the prototypical REPEC strain E22 (O103:H2) (24). These surface-located factors include a fimbria (SefABCD), an autotransporter adhesin (AdcA/Tsh), and a serine protease (EspC). We also showed that RegR, like the AraC-like virulence regulators, RegA of Citrobacter rodentium and ToxT of Vibrio cholerae (25, 26), requires a gut-associated environmental signal, bicarbonate ions, which are abundant in intestinal secretions (27), to exert its effect on gene expression. Because of this, RegR allows REPEC to sense when it has entered the small intestine and then respond by activating a suite of virulence-related genes.
The regR gene is located on an ∼62-kb, E22-specific genetic island which also includes a fimbrial operon that is highly homologous (91% identity) to the ral operon (ralCDEFGHI) that we identified in REPEC strain 83/39 (O15:H−) (24, 28). The ralCDEFGHI locus encodes fimbriae that play an essential role in mediating REPEC adherence to the rabbit intestinal mucosa prior to A/E lesion formation (28, 29). Interestingly, the ral operon of E22 is not a member of the RegR regulon, indicating that another virulence regulatory mechanism is involved in colonization by this strain (24).
In E. coli pathogens, the expression of fimbrial operons is usually subject to tight transcriptional control by regulatory protein(s) (30). In the case of ral, this mechanism of transcriptional regulation is unknown. In the present study, we identified the ralR gene from REPEC strain E22 and demonstrated that RalR controls the expression of a number of genes, including the ral operon and other putative virulence factors.
MATERIALS AND METHODS
Bacterial strains, plasmids, and growth conditions.Bacterial strains and plasmids used in the present study are listed in Table 1. Unless stated otherwise, bacteria were grown at 37°C in Luria-Bertani broth (LB) or on Luria agar (LA) plates supplemented with the appropriate antibiotics at the following concentrations: ampicillin, 100 μg/ml; chloramphenicol, 25 μg/ml; kanamycin, 50 μg/ml; and trimethoprim, 40 μg/ml. Primers used here are listed in Table 2.
Strains and plasmids used in this study
PCR primers used in this study
Construction of EPEC E22ΔralR and E22Δ2 nonpolar mutant strains.The λ Red recombinase system was utilized to construct nonpolar deletion mutants of REPEC E22 (33). Phusion high-fidelity DNA polymerase (Finnzymes), which generates blunt-ended fragments, and the primer pairs RalR-F/RalRkan-R and RalR-R/RalRkan-F were used to amplify the DNA sequences flanking the ralR gene from E22 genomic DNA, and primers pKD4F and pKD4R were used to amplify the kanamycin (kan) resistance gene from plasmid pKD4. The products of these three PCRs (100 ng each) served as templates in overlapping extension PCR (36) using Platinum Taq DNA polymerase (Invitrogen) and the primers RalR-F and RalR-R to generate a DNA fragment carrying a kan resistance gene flanked by ∼500-bp regions up- and downstream of ralR. This DNA fragment was cloned into pGEM-T Easy (Promega); the recombinant plasmid (pYS21) was introduced into E. coli K-12 TOP10 cells (Invitrogen), and the insert was then confirmed by sequencing. pYS21 was then used as a template in a PCR with primer pair RalR-F/RalR-R to amplify the linear allelic replacement DNA fragment, which was introduced into strain E22 expressing λ Red recombinase from plasmid pKD46. The resultant E22ΔralR mutant was confirmed by PCR using two primer pairs in which one primer flanked the targeted region and the other primer acted within the kan gene (RalR-F1/pKD4seqR and RalR-R1/pKD4seqF). Strain E22Δ2 was generated as previously described (24).
Construction of a trans-complementing plasmid, pYS23.For complementation, a 1.7-kb fragment containing ralR and ∼300-bp flanking sequences was amplified from genomic DNA of REPEC strain E22 by using the PCR primer pair RalR-F1/RalR-R1 and cloned into pGEM-T-Easy to yield pGEM-T Easy+ralR (pYS22), which was then confirmed by sequencing. The fragment was excised by EagI digestion and ligated into the EagI site of pACYC184 to generate the plasmid pYS23 (pACYC184+ralR).
Antisense EPEC E22 microarrays.An antisense oligonucleotide microarray was custom designed using the Agilent eArray platform (Agilent Technologies). The array contained open reading frames (ORFs), representing all gene predictions, for the E. coli strain E22 genome available at the National Center for Biotechnology Information (GenBank accession number AAJV00000000). Each ORF was represented by at least three different oligonucleotides.
RNA isolation and labeling.E. coli strains E22Δ2(pYS23) and E22Δ2(pACYC184) were cultivated in LB overnight at 37°C. Quadruplicate cultures of a 1 in 100 dilution were grown to an optical density at 600 nm (OD600) of 0.85. One volume of cells was incubated with 2 volumes of RNAprotect solution and processed according to the manufacturer's instructions (Qiagen). Cell lysis and RNA preparation were carried out using a FastRNA Pro Blue kit (Qbiogene, Inc.). After a 10-min treatment with RNase-free DNase I (Qiagen), the RNA was further purified utilizing the RNeasy MiniElute kit (Qiagen). A total of 5 μg of RNA was labeled with either Cy5-ULS or Cy3-ULS as described in the Kreatech ULS labeling manual (Kreatech Diagnostics). RNA quality and degree of labeling were determined by using an Agilent 2100 bioanalyzer and an ND-1000 spectrophotometer (NanoDrop Technologies). A dye swap was performed for two of the four cultures to minimize the effects of any labeling artifacts.
Fragmentation, microarray hybridization, scanning, and analysis.Fragmentation, hybridization, and scanning were performed at the Australian Genome Research Facility, Ltd., Melbourne, Australia. Normalization and data analysis were performed using the limma package in Bioconductor (37). Genes were considered differentially expressed if they showed an average change of ≥1.5-fold with an adjusted P value of ≤0.05.
Quantitative real-time RT-PCR (QRT-PCR).To verify global gene expression data, bacterial cultures of wild type and an isogenic ΔralR mutant of E22 were cultivated in LB overnight at 37°C. Triplicate cultures of a 1 in 100 dilution were grown to an OD600 of 0.85 to 0.88, and 1 ml of each culture was combined and treated with RNAprotect solution according to the manufacturer's instructions. About 1 μg of RNA (isolated as described above) was further treated with DNase I (2.7 Kunitz units) for 1 h at 37°C, followed by inactivation of the enzyme with 2 μl of 25 mM EDTA and heating for 5 min at 65°C. Half of the reaction mix (10 μl) was subjected to reverse transcription using SuperScript III reverse transcriptase as instructed by the manufacturer (Invitrogen). The lack of residual genomic DNA in each sample was verified prior to quantification of gene transcription. The primer pairs ralC-Frt/ralC-Rrt, aap-Frt/aap-Rrt, aatp-Frt/aatp-Rrt, and ICCrrsBFrt/ICCrrsBRrt were used to amplify ralC, aap, aatP, and the16S rRNA gene, respectively. The Brilliant II SYBR green QPCR Master Mix was used to perform PCR in real time, as recommended by the manufacturer (Agilent Technologies). Reactions were performed in triplicate and contained 500 nM concentrations of each primer in a total volume of 20 μl. Amplification and detection of specific products were performed with a CFX96 real-time PCT detection system and a C1000 thermal cycler (Bio-Rad Laboratories), using the following protocol: 95°C for 30 s, followed by 40 cycles of 95°C for 5 s and 55°C for 20 s. The data were analyzed by using CFX Manager Version 2.0 software (Bio-Rad Laboratories). Obtained threshold cycle (CT) values of the target genes were normalized to the 16S rRNA gene and expressed as the fold difference by calculating the 2ΔΔCT.
Primer extension.Primer extension was performed as described previously (38). Briefly, total cellular RNA was purified from the test E. coli strains MC4100(pYS23+pYS26) and MC4100(pYS23+pYS27), and the control strain MC4100(pYS23+pMU2385). Cells were grown to mid-log phase (OD600 = 0.8), and RNA was isolated by using the FastRNA Pro kit. The primers RalC-PE-R and Aap-PE-R were labeled at their 5′ ends with 32P by using T4 polynucleotide kinase and [γ-32P]ATP. The labeled primer was coprecipitated with 5 μg of total RNA. Hybridization was carried out at 45°C for 15 min in 10 μl of Tris-EDTA 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 dithiothreitol [DTT], 2 mM deoxynucleoside triphosphates [dNTPs], 1 U of avian myeloblastosis virus reverse transcriptase/ml) and were carried out at 42°C for 60 min. Samples were then ethanol precipitated and analyzed on a sequencing gel. A GA ladder was made according to the method of Maxam and Gilbert (39) to sequence ralC and aap fragments that were generated by PCR using the primer pairs RalC-PE-F/32P-RalC-PE-R and Aap-PE-F/32P-Aap-PE-R, respectively.
Construction of lacZ fusion plasmids.To construct the translational-fusion plasmid, a 0.35-kb ralR fragment was amplified from E22 genomic DNA by using the primer pair RalR-PstIF-t/RalRBamHIR-t. This was then cloned into pGEM-T Easy to yield pYS38. The fragment was excised from vector pYS38 by PstI/BamHI digestion and inserted into the single-copy-number translational-fusion vector pMU2386 to generate plasmid pYS39.
The ralC-lacZ and aap-lacZ transcriptional-fusion plasmids were constructed by PCR amplification of the regulatory region of ralC and aap by using the primer pairs RalC-BamHI-F/RalC-BglII-R and Aap-BamHI-F/Aap-HindIII-R, respectively. The resulting PCR fragments were cloned into pCR2.1-TOPO to yield pYS24 and pYS25, and these fragments were sequenced. The fragments were then excised from the pCR2.1-TOPO derivatives by BamHI/BglII or BamHI/HindIII digestion and cloned into the same sites of the single-copy transcriptional-fusion vector pMU2385 to create the ralC-lacZ (pYS26) and aap-lacZ (pYS27) transcriptional fusions.
ralC1-lacZ (pYS35) carrying a mutated RalR-Box1 was generated by using a multiple-step overlapping PCR method. The first PCR involved the use of the primer pairs RalC-BamHI-F/RalC-Bind1-R and RalC-BglII-R/RalC-Bind1-F, template pYS24, and Phusion high-fidelity DNA polymerase. The two resulting PCR fragments (100 ng each) were mixed together with Platinum Taq DNA polymerase master mix. After seven cycles of extension, the primers RalC-BamHI-F and RalC-BglII-R were added to the mixture. The sample was then subjected to 30 cycles of PCR, which resulted in the formation of a 1.2-kb DNA fragment. This fragment was cloned into pGEM-T Easy to generate pYS28, and the mutation was confirmed by sequencing. The mutant ralC fragment was excised by BamHI/BglII digestion and cloned into the same sites of the single-copy transcriptional fusion vector pMU2385 to create ralC1-lacZ (pYS35). The ralC2-lacZ (pYS36, RalR-Box2) and aap1-lacZ (pYS37, RalR-Box3) mutations were constructed in the same manner, except that the primer pairs RalC-BamHI-F/RalC-Bind2-R and RalC-BglII-R/RalC-Bind2-F with template pYS24 (for ralC2-lacZ) and Aap-BamHI-F/Aap-Bind-R and Aap-HindIII-R/Aap-Bind-F with template pYS25 (for aap1-lacZ) were used in the Phusion PCR, and the resultant fragments generated from the Platinum Taq PCR were cloned into pGEM-T Easy to generate pYS29 and pYS34, respectively.
β-Galactosidase assay.Bacteria were grown to mid-log phase (OD600 ∼0.6). The β-galactosidase activity was assayed as described by Miller (40), and the specific activity was expressed in Miller units (MU). The data shown are the results of at least three independent assays.
Expression and purification of MBP::RalR.The coding sequence of ralR was amplified from E22 genomic DNA by using the primer pair RalR-BamHIF/RalR-PstIR and cloned into pGEM-T Easy to obtain plasmid pYS30. The ralR gene was excised by BamHI/PstI digestion and inserted into expression vector pMAL-c2x (New England BioLabs) for N-terminal fusion to malE. The resulting vector, pYS31, was transformed into E. coli strain BL21(DE3). Overnight cultures of transformants were diluted 1 in 100 in fresh LB and grown at 30°C and 200 rpm to OD600 of 0.9. Induction of gene expression was carried out for 19 h at 16°C by the addition of IPTG (isopropyl-β-d-thiogalactopyranoside) to a final concentration of 0.2 mM. Afterward, bacterial cells were harvested (15 min, 3,000 × g, 4°C) and disrupted by the addition of lysozyme (100 μg/ml) and subsequent sonication in column buffer (20 mM Tris-HCl [pH 7.4], 1 M NaCl, 1 mM EDTA). Purification of RalR was achieved through binding of the fusion protein to an amylose resin as recommended by the manufacturer (New England BioLabs). All steps were carried out at 4°C. The concentration and purity of eluted MBP::RalR was determined by using a ND-1000 spectrophotometer, as well as by SDS-PAGE of the fusion protein.
Electrophoretic mobility shift assay (EMSA).Labeling of DNA fragments with 32P was performed as follows. The primers RalC-EMSA-F, Aap-EMSA-R, and RalC-EMSAc-F were labeled at their 5′ ends by using [γ-32P]ATP and T4 polynucleotide kinase. DNA fragments to be analyzed for RalR binding were generated by PCR using the primer pairs 32P-RalC-EMSA-F/RalC-EMSA-R (for fragment ralCprom), Aap-EMSA-F/32P-Aap-EMSA-R (for fragment aapprom), and 32P-RalC-EMSAc-F/RalC-EMSAc-R (for the control fragment) (Table 2), with E22 genomic DNA as a template. EMSA was carried out as published previously (27). Briefly, each fragment was incubated with various amounts of purified MBP::RalR protein at 25°C for 30 min in binding buffer [10 mM Tris-HCl (pH 7.4), 100 mM KCl, 0.1 mM DTT, 0.01% (vol/vol) Triton X-100, 1 mM EDTA, and 100 μg of bovine serum albumin/ml, 5 ng of poly(dI-dC)/μl, 10% (vol/vol) glycerol]. DNA and DNA-protein complexes were then separated on 5% native polyacrylamide gels (37.5:1) for ∼12 h at 10 V/cm and 4°C.
Preparation of surface-associated proteins.Bacterial strains were cultivated in LB overnight at 37°C, diluted 1 in 100 in 10 ml of LB, and then grown at 37°C to late-log phase (OD600 = 2.0). Cells were harvested by centrifugation at 3,000 × g for 10 min, resuspended in 160 μl of phosphate-buffered saline (PBS; pH 7.4), vortexed at high speed for 1 min and subsequently incubated at 60°C for 30 min with intermittent vortexing. The samples were then pelleted by centrifugation at 3,000 × g for 10 min, and the supernatant was transferred to a fresh tube, where it was mixed with NuPAGE lithium dodecyl sulfate sample reducing buffer and boiled at 100°C for 5 min. The samples were then separated by SDS-PAGE using 4 to 12% Bis-Tris NuPAGE gels (Invitrogen), and the separated proteins were stained with Coomassie brilliant blue R250.
Preparation of secreted proteins.Bacterial strains were cultivated in LB overnight at 37°C, diluted 1 in 100 in 10 ml of LB, and then grown at 37°C to late-log phase (OD600 = 2.0). Cells were harvested by centrifugation (5,000 × g, 10 min), and the supernatant, containing the extracellular proteins, was passed through a 0.20-μm-pore-size filter (Sartorius). Proteins in the supernatants were precipitated with 20% (vol/vol) trichloroacetic acid on ice for 1 h, washed in 25% (vol/vol) acetone, separated by SDS-PAGE using 4 to 12% Bis-Tris NuPAGE gels, and stained with Coomassie brilliant blue R250.
Protein identification.Bands of interest separated by PAGE were carefully excised and subjected to tandem mass spectrometry at the Walter and Eliza Hall Institute for Medical Research, Proteomics Laboratory, Melbourne, Australia.
Motility assay.Overnight cultures grown in LB were standardized to an OD600 of 1.0, and 2 μl of culture was stabbed into LA plates, containing 0.3% (wt/vol) agar, by using a sterile pipette tip. The plates were incubated for 5 h at 37°C, after which time the diameter of the swimming zone around the inoculation site was measured.
Assay for bacterial adhesion to HEp-2 cells.The Center for Vaccine Development method was used to determine the pattern of bacterial adherence to HEp-2 epithelial cells (41). Briefly, bacteria were grown in Penassay broth (Oxoid) without shaking at 37°C overnight. HEp-2 cells were passaged at 37°C in air containing 5% CO2 in Dulbecco modified Eagle medium supplemented with 5% (vol/vol) fetal calf serum and 0.5% (wt/vol) d-mannose. HEp-2 monolayers that were ∼70% confluent on 12-mm-diameter glass coverslips were incubated with 10 μl of a washed bacterial suspension for 3 h. The coverslips were then washed three times with PBS, and the cells were fixed with Giemsa buffer (Sørensen buffer, 34 mM KH2PO4, 33 mM Na2HPO4, 0.005% sodium azide [pH 6.8]) and 100% methanol. Cells were stained with 10% Giemsa stain before examination by using light microscopy.
Infection of rabbits.For in vivo assays of virulence, 4- to 5-week-old New Zealand White rabbits were inoculated with a wild-type REPEC strain (E22) or its isogenic ralR mutant. Rabbits were examined daily for clinical signs of illness, including weight loss and evidence of diarrhea, as described previously (28). Fecal shedding of the infecting strains was determined by culture of rectal swabs on MacConkey agar supplemented with rifampin (50 μg/ml) to distinguish them from the rabbits' microbiota. Animals were euthanized when they lost >15% of their body weight or demonstrated severe diarrhea. All experimental procedures using animals were approved by the University of Melbourne Animal Experimentation and Ethics Committee and performed in accordance with the guidelines for animal experimentation of the Australian National Health and Medical Research Council.
Statistical analysis.All analyses of quantitative data were performed by using Student t test. A two-tailed P value of <0.05 was taken to indicate statistical significance.
Microarray data accession number.The microarray data have been submitted to the Gene Expression Omnibus (http://www.ncbi.nlm.nih.gov/geo) under accession number GSE41093.
RESULTS
In silico identification of a putative virulence regulator of the ral operon.Immediately upstream of the ral operon of E22 is an ORF, EcE22_5293, which is in the opposite orientation to the genes encoding the structural subunits of the Ral fimbria (GenBank accession number AAJV02000028.1). Analysis using the Pfam database domain search (Wellcome Trust Sanger Institute, United Kingdom) revealed that the predicted protein of EcE22_5293 contains a helix-turn-helix (HTH; Pfam PF00165) region at the N-terminal end that is part of an AraC-type DNA-binding domain. This raised the possibility that the predicted protein product may be a regulator which controls expression of the virulence-fimbrial operon, ralCDEFGHI. However, unlike most other AraC-like regulators whose HTH DNA-binding motif is located at the C-terminal region, this putative protein carries a HTH DNA-binding motif in its N-terminal region. The C-terminal region of this putative regulatory protein contains a Gyr-I motif (Pfam PF06445), which is the small molecule-binding domain of a number of transcriptional activators (42).
EcE22_5293 is transcriptionally and translationally active in vivo.The fact that the putative regulator is encoded by an intact ORF suggested that this protein is expressed in vivo. To investigate this, we generated a construct where the promoter region of EcE22_5293, from the 11th codon to a position ∼350 bp upstream of the predicted translational start site, was fused in frame with the lacZ structural gene on the single-copy-number translational-fusion vector pMU2386 (35). The resulting pMU2386 derivative, pYS39, along with the control plasmid (pMU2386), were each transformed into REPEC strain E22, and β-galactosidase assays were carried out following the growth of the transformants in LB. Very low levels of β-galactosidase activity (2.10 ± 0.01 MU [mean ± the SD of three independent assays]) were detected for the E22 control; however, the pYS39 translational fusion produced 103.91 ± 2.21 MU. The results show that both the transcriptional and translational machineries of EcE22_5293 were active, and the ORF was named ralR in keeping with the nomenclature of the previously characterized ral operon (28).
RalR controls transcription of virulence-associated genes.To identify genes whose expression is controlled by RalR, we carried out a microarray analysis comparing a RalR− strain to a RalR+ strain. Located on the same genetic island is a gene encoding another transcriptional regulator, RegR (24), which shares significant homology within its HTH DNA-binding motif with RalR. To avoid any complications arising from possible redundancy and interplay between the two regulators in the microarray, we knocked out both loci in the E22 genome. The resulting E22 ΔregR ΔralR strain was designated E22Δ2 (24). Total RNA from both the control strain, E22Δ2(pACYC184) (RalR−), and the ralR-complemented test strain, E22Δ2(+pYS23) (RalR+), were fluorescently labeled and hybridized to a custom-made microarray chip (Agilent Technologies) covering all predicted ORFs of the E22 genome.
The microarray data showed that 500 genes represented by 1,203 probes were differentially expressed (adjusted P value of <0.05; see Table S1 in the supplemental material). Seventeen genes organized in four separate operons were upregulated (>1.5-fold). The most significantly upregulated gene (12-fold) was ralC, the first gene of the ralCDEFGHI cluster (28) (Table 3), confirming our hypothesis that RalR is responsible for transcriptional regulation of this fimbrial operon. Another operon found to be upregulated by RalR was a homolog of the aat secretion system (aatPABCD) of enteroaggregative E. coli (EAEC) (55 to 41% similarity) (43). The other three operons upregulated in the RalR+ background included trpEDCBA, mtr, and degP, which are involved in tryptophan biosynthesis, tryptophan transport, and periplasmic-protein quality control, respectively.
Summary of genes strongly differentially regulated by RalR, identified by analyzing microarray dataa
RalR downregulated the expression of nine genes >1.5-fold (Table 3), which included glcDEFGB, glcA, fliC, tnaA, and tnaB, which encoded members of the glycolate and glyoxylate degradation II pathway, the associated glycolate permease, flagellin, tryptophanase and the low-affinity tryptophan permease, respectively.
The substrate of the Aat transport system of EAEC is dispersin (Aap). Adjacent to the aat operon of the E22 genome is an ORF whose predicted protein shares similarity (31%) to Aap from EAEC. This ORF is not annotated in the E22 genome and thus was absent from our E22 array chip. To determine whether expression of aap is regulated by RalR, we performed QRT-PCR. Total cellular RNA was isolated from the wild type and an isogenic ΔralR mutant of E22 grown in LB. After the PCR amplifications, the levels of expression from each sample were normalized to the reference 16S rRNA gene, and the relative abundance of aap transcripts from the RalR+ and RalR− strains was assessed. An 18.9-fold activation of aap transcription by RalR was detected. In addition, we performed QRT-PCR assays on the first genes of the ral (ralC) and aat (aatP) operons. RalR-mediated activation was found to be 7.4-fold for ralC and 3.1-fold for aatP. These results confirmed the observation from the microarray analysis that the ral and aat operons are both upregulated by RalR (Table 3).
Mapping of the start sites of transcription of the ralC and aap promoters.We next characterized the promoters of ralC and aap, the two most highly upregulated gene targets of RalR. To map the start site of transcription of ralC and aap, 32P-labeled primers (RalC-PE-R for ralC and Aap-PE-R for aap) were hybridized with total RNA isolated from the E. coli test and control strains (Materials and Methods). After extension using reverse transcriptase in the presence of dNTPs, the samples were analyzed on sequencing gels. The start site of ralC transcription was mapped to a cytosine residue located 61 nucleotides (nt) upstream of the start codon for RalC (Fig. 1). Based on this start site, a putative ralC promoter region was identified. Its putative core elements included a good −10 region (TAAAAT), a poor −35 region (ATGCAG), and an 18-bp spacer (Fig. 1B). The start site of transcription of aap was mapped to a guanine residue located 47 nt upstream of the start codon of Aap (Fig. 2). The putative aap promoter deduced from the primer extension results included a perfect −10 region (TATAAT), a reasonable −35 region (TTGCGA), and a spacer of 18 bp (Fig. 2B).
ralC regulatory region. (A) Nucleotide sequence of the ralC regulatory region. The numbering at the left of the sequence is relative to the transcriptional start site of ralC, which is indicated by an angled arrow. The putative −10 and −35 regions are indicated and underlined. The putative start codon is underlined with a dashed line. The putative RalR-binding sites, Box1 and Box2, are underlined with a double line and are labeled. (B) The start site of transcription of the ralC promoter was mapped by primer extension using RNA isolated from E. coli MC4100 strains containing pYS23 (RalR+) with either pMU2385 (control, lane 1) or pYS26 (ralC-lacZ, lane 2). The positions corresponding to the primer and the extension product are marked with “P” and “E,” respectively. The nucleotide sequence corresponding to a region of the GA ladder is also shown.
aap regulatory region. (A) Nucleotide sequence of the aap regulatory region. The numbering at the left of the sequence is relative to the transcriptional start site of aap, which is indicated by an angled arrow. The putative −10 and −35 regions are indicated and underlined. The putative start codon is underlined with a dashed line. The putative RalR-binding site, Box3, is underlined with a double line and is labeled. (B) The start site of transcription of the aap promoter was mapped by primer extension using RNA isolated from E. coli MC4100 strains containing pYS23 (RalR+) with either pMU2385 (control, lane 1) or pYS27 (aap-lacZ, lane 2). The positions corresponding to the primer and the extension product are marked with “P” and “E,” respectively. The nucleotide sequence corresponding to a region of the GA ladder is also shown.
Analysis of RalR-mediated activation of the ralC and aap promoters.To further investigate RalR-mediated activation of ralC and aap, we carried out transcriptional analyses using a lacZ reporter system. Promoter-lacZ transcriptional fusions were constructed by separately cloning the promoter fragments of ralC and aap, encompassing positions −795 to +433 (relative to the transcriptional start site of ralC) and positions −454 to +99 (relative to the transcriptional start site of aap), in front of the lacZ structural gene of the single-copy vector, pMU2385. The resulting plasmids, ralC-lacZ (pYS26) and aap-lacZ (pYS27), were each transformed into E. coli MC4100 derivatives containing either the pYS23 plasmid (RalR+) or the control vector pACYC184 (RalR−). Strains were grown to mid-log phase, and the promoter activity was measured by assaying the β-galactosidase activity. Very low levels of β-galactosidase activities (9 MU) were detected for the ralC-lacZ construct in the RalR− background. However, the ralC promoter produced 173 MU of β-galactosidase in the RalR+ background, representing 19-fold activation by RalR (Fig. 3A).
Expression of β-galactosidase by various ralC-lacZ and aap-lacZ fusions in E. coli strains MC4100 and E22. The numbering of the ralC and aap fragments is relative to the start sites of transcription of ralC and aap, respectively. The promoter activities of the various constructs are shown as specific activities of β-galactosidase (Miller units), which are mean values from three independent assays, with variation of <15%. Fold activation (Fold act.) is the specific activity of β-galactosidase of the RalR+ strain divided by that of the RalR− strain. (A) Analysis of ralC-lacZ and aap-lacZ was performed in E. coli RalR− [MC4100(pACYC184)] and RalR+ [MC4100(pYS23)] strains. (B) Analysis of ralC-lacZ and aap-lacZ was performed in E22 RalR− [E22ΔralR(pACYC184)] and RalR+ [E22ΔralR(pYS23)] strains. (C) Analysis of ralC1-lacZ (RalR-Box1 mutation), ralC2-lacZ (RalR-Box2 mutation), and aap1-lacZs (RalR-Box3 mutation) was performed in E. coli RalR− [MC4100(pACYC184)] and RalR+ [MC4100(pYS23)] strains.
As for the aap-lacZ construct, the basal-level promoter activity was 46 MU in the RalR− background and 4,810 MU in the presence of RalR, representing a 91-fold increase in expression (Fig. 3A).
Transcriptional analysis of the ralC-lacZ and aap-lacZ constructs was also carried out in native REPEC E22 using RalR− [E22ΔralR(pACYC184)] and RalR+ [E22ΔralR(pYS23)] strains. As shown in Fig. 3B, similar regulatory patterns to those observed in E. coli strain MC4100 were seen in the E22 derivatives. Together, these data clearly demonstrate that the expression of ralC and aap is strongly activated by RalR.
Identification of the operator sites within the regulatory regions of ralC and aap.To locate possible operator sequences within the regulatory regions of ralC and aap, we identified three 8-bp palindromic sequences, TGTGCACA (RalR-Box1) and TTTGCACA (RalR-Box2), which are centered at −66/−65 and −47/−46, respectively, relative to the start site of ralC transcription (Fig. 1B), and AGTGCACA (RalR-Box3), centered at −57/-56 relative to the start site of aap (Fig. 2B). To test whether these regions are involved in the control of ralC and aap by RalR, we changed the sequences of the three sites to AACCACTG (Fig. 3C). The three constructs—ralC1-lacZ (pYS35, RalR-Box1), ralC2-lacZ (pYS36, RalR-Box2), and aap1-lacZ (pYS37, RalR-Box3)—containing the mutated regions were assessed for their ability to be activated by RalR. The data in Fig. 3C showed that all three boxes were important for RalR-mediated activation. In the case of the ralC promoter, scrambling RalR-Box1 and RalR-Box2 reduced the RalR-mediated activation from 19-fold to 5- and 2-fold, respectively. As for the aap promoter, the base changes within the putative operator site (RalR-Box3) completely destroyed the ability of RalR to activate its transcription (Fig. 3C).
RalR binds directly to the ralC and aap promoter region.To test whether RalR interacts directly with the ralC and aap regulatory regions, we performed an EMSA. For this assay, we used a purified fusion protein (MBP::RalR), because RalR, like most other AraC-like regulatory proteins, is highly insoluble (44). Three 32P-labeled DNA fragments were also used in the EMSA. These included (i) a 349-bp ralC promoter fragment (ralCprom) that spanned between positions −319 and +30, relative to the start site of transcription; (ii) a 279-bp aap promoter fragment (aapprom), which spanned between positions −271 and +8 relative to the start site of transcription, and (iii) a control fragment which spanned between positions +555 and +888 within the ralC coding sequence. After incubation of each of these DNA fragments with various amounts of MBP::RalR, the samples were analyzed using native PAGE. The data (Fig. 4) showed that MBP::RalR was able to shift ralCprom and aapprom fragments to form protein-DNA complexes but failed to cause any gel retardation for the control fragment, indicating that MBP::RalR can bind specifically to the ralC and aap promoter regions.
EMSA analysis of the binding of recombinant RalR protein to the ralC and aap regulatory regions. The 32P-labeled PCR fragments containing the ralC regulatory region ralCprom (A), the aap regulatory region aapprom (B), and a control fragment (C) were mixed with 0, 9.4, 18.8, 37.5, 75, 150, or 300 nM MBP::RalR protein for ralCprom and 0, 18.8, 37.5, 75, 150, or 300 nM MBP::RalR protein for aapprom and the control fragment. After incubation at 30°C for 20 min, the samples were analyzed on native polyacrylamide gels. The unbound DNA (free DNA) and protein-DNA complexes are marked as arrowed “C”s.
Effect of RalR on the expression of surface-associated and secreted proteins.To identify surface-associated and secreted proteins whose synthesis is controlled by RalR, we carried out proteomic analyses of the surface-associated and secretome fractions of E22 derivatives using SDS-PAGE. To facilitate visualization of proteins, we compared protein expression by the ralR mutant, E22ΔralR+pACYC184 (RalR−), and the ralR-complemented strain, E22ΔralR(pYS23) (RalR+). Comparison of the protein profiles revealed two bands in the surface-associated protein fraction (∼29 and ∼ 21 kDa) and one band in the secretome (∼13 kDa), which were present in the RalR+ strain and absent from the RalR− mutant (Fig. 5). These proteins were excised and analyzed by using tandem mass spectrometry. The ∼29- and ∼21-kDa surface-associated proteins were identified as the full-length and truncated RalG (EcE22_5298) proteins, respectively. RalG is the major fimbrial subunit of Ral fimbriae. The ∼14-kDa secretome protein was identified as the dispersin protein, Aap.
Effects of RalR on the expression of surface-associated and secreted proteins by E. coli E22. RalR+ [E22ΔralR(pYS23)] and RalR− [E22ΔralR(pACYC184)] strains of E22 were grown in LB. Proteins were separated by SDS-PAGE and stained with Coomassie brilliant blue R250. Three protein bands (two from the surface-associated fraction and one from the secretome fraction) which were present in the RalR+ strain but not in the RalR− strain (indicated by asterisks) were excised and analyzed by tandem mass spectrometry. The identities of these proteins are shown at the right of the gels.
Effect of RalR on motility.The microarray analysis indicated that the fliC gene was downregulated in the RalR+ background. To test this finding, we compared the motility of an E22ΔralR mutant to that of the wild-type strain and found that the ralR-mutant strain was less motile than the wild type, indicating that RalR can act as a repressor of flagellar gene expression (Fig. 6). We also complemented the ralR mutant strain with a plasmid expressing RalR (pYS23) and observed that the complemented strain had decreased motility, again suggesting that RalR has a repressive role in motility (Fig. 6).
Inhibitory effect of RalR on the motility of E. coli E22. Motility plates were inoculated with wild-type E22 (E22+pACYC184), ralR mutant [E22ΔralR(pACYC184)], and trans-complemented strain [E22ΔralR(pYS23)]. After 5 h of incubation at 37°C, the diameters of the swimming zones around the inoculation sites were measured. Values are means and the standard deviations for three independent experiments (**, P < 0.0005; *, P < 0.005).
RalR mediates adherence through transcriptional control of the ral locus.The ral fimbrial operon of REPEC strain 83/39 plays a key role in the adherence of REPEC by mediating bacterial attachment to the rabbit intestinal mucosa (29). To determine whether RalR controls the transcription of the genes involved in adherence, we conducted HEp-2 cell adherence assays in the presence of mannose, comparing the adherence profile of wild-type strain E22 to a RalR− strain (E22ΔralR). As shown in Fig. 7A, the wild-type strain exhibited a strong adherence phenotype. In contrast, the E22ΔralR mutant (Fig. 7B) displayed a markedly attenuated phenotype with no adherence to HEp-2 cells observed. The adherence phenotype, however, was restored to E22ΔralR when it was trans-complemented with ralR (pYS23), confirming that RalR controls adherence of REPEC strain E22 by controlling transcription of the ral fimbrial operon (Fig. 7C).
Adherence of E. coli E22 with or without RalR to HEp-2 cells. (A) RalR+ strain (wild-type E22) showing extensive adherence; (B) RalR− strain (E22ΔralR) showing no adherence; (C) E22ΔralR trans-complemented strain [E22ΔralR(pYS23)]. Arrows point to adherent bacteria.
RalR is a virulence determinant of REPEC strain E22.To investigate the effect of RalR on the virulence of EPEC, we infected infant rabbits with wild-type REPEC strain E22 and the ralR mutant (E22ΔralR). After the inoculation of six rabbits with 5 × 105 CFU of wild-type E22, large numbers of REPEC were shed in stools, such that >105 CFU were recovered from rectal swabs from all rabbits within 2 days of inoculation (Fig. 8A). Loss of body weight began within 48 h of infection and continued until the animals were euthanized (Fig. 8B). All rabbits developed clinical illness characterized by diarrhea with weight loss requiring euthanasia. Five rabbits were euthanized on day 5, and one was euthanized on day 7. These results were consistent with previous reports of infection of rabbits with E22 (19). After inoculation with 8 × 105 CFU of the ralR mutant strain, the challenge strain was only detected on day 1 after inoculation in two of the six rabbits.
Effect of RalR on virulence of E. coli E22 for rabbits. Six infant rabbits were inoculated with 5 × 105 CFU wild-type E22 or 8 × 105 CFU of its isogenic ralR mutant (E22ΔralR). Some rabbits in a group were euthanized due to a loss of body weight of >15% or the presence of severe diarrhea, and the surviving numbers of rabbits are shown adjacent to the data points. (A) Colonization of rabbits by REPEC strain E22 and the ralR mutant as measured by quantitative culture of rectal swabs. The data are the geometric mean CFU per swab. The detection limit of the culture method is indicated by the horizontal line. (B) Body weight of rabbits. Some rabbits in a group were euthanized due to a body weight loss of >15% or severe diarrhea, and the surviving number of rabbits is shown adjacent to the data point. Values are means ± the standard deviations for each group expressed as a percentage of the weight on the day of infection (day 0).
Body weight is another sensitive indicator of illness in the REPEC/rabbit infection model (23, 45) and, during the course of infection, rabbits infected with the mutant strain showed weight gain, in contrast to those infected with the wild type (Fig. 8B). Taken together, these results indicated that RalR is an essential virulence determinant of REPEC strain E22.
DISCUSSION
In this study, we showed that a rabbit-specific aEPEC (REPEC) strain, E22, contains a non-LEE-encoded virulence regulon whose expression is controlled by the RalR protein. The RalR regulon includes the regulatory gene ralR and its gene targets which code for homologs of established virulence determinants, including the Ral fimbria, the Aat secretion system, and its substrate Aap (28, 43, 46). The ralR allele appears to be widespread among REPEC strains since 8 of the 10 REPEC strains tested in our collection (47) harbored this gene (data not shown). In addition, 8% of the aEPEC clinical isolates from humans that we previously characterized (48) also carry the ralR allele (data not shown).
Studies with EAEC have shown that the Aat protein-secretion system is responsible for exporting the Aap dispersin protein out of the bacterial cell (43). Aap's role in EAEC pathogenesis is associated with colonization, since the data suggest that Aap could neutralize the bacterial cell surface charge so that the positively charged aggregative adherence fimbriae are free to extend out from the bacterium and adhere to the colonic mucosa (43, 46, 49). The aat, aap, and aaf operons are coregulated by the AraC-like transcriptional activator AggR in EAEC (50–52). Interestingly, the prototypic enterotoxigenic E. coli (ETEC) strain H10407 also possesses an aat-like locus that is colocated with a gene which encodes an extracytoplasmic protein, CexE, that is proposed to play a role similar to that of Aap of EAEC (53). CexE expression is under the control of the AraC-like virulence regulator CfaD/Rns, which also regulates the expression of the CFA/I colonization pilus (54). Our data indicate that RalR may play a similar role in colonization and virulence to that of AggR of EAEC and CfaD/Rns of ETEC. This hypothesis is supported by our in vitro and in vivo data which demonstrated that RalR (i) is required by E22 to adhere to HEp-2 cells, (ii) is responsible for the abundant presence of products of the ral and aat operons, either on the bacterial cell surface or in the secretome, and (iii) is required for full virulence of its natural host.
In contrast to other AraC-like virulence regulators such as PerA, CfaD/Rns, and AggR, whose DNA-binding domains are located at their C termini, the putative HTH DNA-binding motif of RalR is present at the N-terminal region, whereas its C terminus contains a putative small molecule-binding domain. Furthermore, whereas the AraC-like virulence transcriptional activators upregulate gene expression by overcoming transcriptional silencing mediated by the histone-like protein, H-NS (55), the transcription of the RalR regulon genes, aap and ralC, is not subject to the repression control by H-NS (data not shown), indicating that RalR acts differently from other AraC-like virulence regulators of intestinal pathogens.
The mechanism of RalR-mediated activation was analyzed by a number of experiments using the promoter regions of the ralC and aap genes. The data from primer extension assays indicated that both ralC and aap are driven by a single σ70 promoter. Although the ralC and aap promoters are both strongly upregulated by RalR, the latter exhibited a stronger basal level of expression in the absence of RalR (Fig. 3B). The difference in promoter activities is reflected in their respective core sequences in that the aap promoter contains a well-conserved −35 region (TTGCGA versus the consensus TTGACA), a perfect −10 region (TATAAT), and an 18-bp spacer (versus the ideal 17-bp spacer), whereas the ralC promoter comprises an almost unrecognizable −35 region (ATGCAG), a conserved −10 region (TAAAAT), and an 18-bp spacer. The data from EMSA showed that RalR binds specifically to the regulatory regions of ralC and aap (Fig. 4). Mutational experiments identified the core operators as 8-bp palindromes with the consensus sequence of TGTGCACA (Fig. 3C). The relevant distance between these operator sites and the promoter core sequences of ralC and aap is consistent with the hypothesis that RalR functions as a class I activator which stimulates transcription by recruiting RNA polymerase to its target promoters through interacting with the α subunit of RNA polymerase (56, 57).
Our microarray analysis also showed that RalR downregulated the expression of fliC, which encodes flagellin, the major component of the E22 flagellum. This result was verified in a motility assay. Motility plays a role in the initial phases of intestinal infection by Salmonella and E. coli, after which motility is downregulated in favor of a specific cell invasion or signal transduction program (reviewed in reference 58). Although the mode of action of RalR at the fliC promoter is unknown, downregulation of flagellum production, once REPEC has reached a favorable niche, may be required for successful colonization and intimate adherence, which are mediated by the ral locus and the LEE genes, respectively.
In addition to its ability to control virulence gene expression, RalR also regulated the expression of several housekeeping operons. RalR significantly upregulated the transcription of the degP gene encoding a periplasmic protease (Table 3). DegP is a major player in periplasmic protein quality control, and it degrades damaged or misfolded proteins, generally cleaving substrates between paired hydrophobic residues (59, 60). Although the mechanism of activation by RalR is unknown, the increased expression of the DegP protein in strain E22 was accompanied by increased production of RalG and Aap. It is possible that DegP is required to remove the accumulated proteins from the periplasm, thus, DegP may also be responsible for the truncated version of RalG that we observed in the surface-associated protein fraction of E22 (Fig. 5).
The trpEDCBA and mtr operons encoding proteins involved in tryptophan biosynthesis and uptake, respectively, are also significantly upregulated by RalR, whereas the tnaAB operon, which encodes a tryptophanase responsible for tryptophan degradation and utilization and a low-affinity tryptophan transport protein, is strongly downregulated by RalR (Table 3). In E. coli, the transcription of the trpEDCBA, mtr, and tnaAB operons is tightly controlled by endogenous levels of tryptophan via TrpR-mediated repression and/or a transcriptional termination mechanism known as attenuation (61, 62). Low levels of tryptophan cause an increased expression of trpEDCBA and mtr and reduced expression of tnaAB and vice versa (61–63). Further analysis using EMSA and promoter-lacZ fusions indicated that RalR neither binds to the regulatory regions of these operons nor directly regulates their expression (data not shown). Therefore, the most likely scenario is that upregulation of trpEDCBA and mtr and downregulation of tnaAB are triggered by RalR-mediated depletion of endogenous tryptophan pool. Indeed, RalR strongly activates the transcription of two genes encoding the surface-associated proteins, RalG and Aap (Fig. 5), which contain significantly more tryptophan, 2.4 to 2.6% versus 1.5%, which is the average of the complete E. coli proteome (63). Tryptophan residues are known to help translocation of surface proteins through cell membranes and serve as anchors on the periplasmic side of the membrane (64).
The glcDEFGBA operon, whose products are involved in the transport and utilization of glycolate, was also downregulated (Table 3). Transcription of the glcDEFGBA operon is activated by GlcC (65), but the mode of action of RalR at the glcD promoter is unknown. Downregulation of the expression of this operon by RalR possibly helps the pathogen conserve energy for the production of virulence factors without having any adverse effect on bacterial growth and replication within its host.
Taken together, the data from this and previous studies indicate that the presence of the LEE PAI in rabbit-specific aEPEC strain E22 is necessary but not sufficient for bacterial virulence (19, 24, 66–68). Like other A/E pathogens, such as tEPEC and C. rodentium, E22 needs to produce a series of non-LEE-encoded virulence factors for full virulence (16, 69–71). In E22, expression of these additional virulence proteins is subject to positive control by two distinct regulators, RalR and RegR, via different mechanisms. Employing more than one regulator for the control of non-LEE-encoded virulence factors may conceivably allow E22 to utilize a variety of environmental cues to achieve precise regulation of specific genes in different environmental niches. Whether RalR contributes to the virulence of aEPEC strains of human origin remains to be determined.
ACKNOWLEDGMENTS
This study was supported by grants from the Australian National Health and Medical Research Council (NHMRC), the NHMRC Independent Research Institutes Infrastructure Support Scheme, and the Victorian State Government Operational Infrastructure Support Program to the Murdoch Childrens Research Institute. M.T. and Y.S. are recipients of an NHMRC Peter Doherty Australian Biomedical Fellowship.
FOOTNOTES
- Received 6 June 2013.
- Returned for modification 29 July 2013.
- Accepted 24 August 2013.
- Accepted manuscript posted online 3 September 2013.
Supplemental material for this article may be found at http://dx.doi.org/10.1128/IAI.00710-13.
- Copyright © 2013, American Society for Microbiology. All Rights Reserved.
