Enteropathogenic
Escherichia coli (EPEC) is a prominent cause of diarrhea worldwide, especially among young children (
28,
32,
41). In developing countries, EPEC is responsible for endemic infantile diarrhea and is estimated to cause the deaths of several hundred thousand children each year (
32,
41). EPEC employs a large number of determinants to colonize the intestine and produces characteristic attaching and effacing (A/E) lesions in the intestinal mucosa (
8,
20). The genetic determinants required for the production of A/E lesions are located on a pathogenicity island called the locus of enterocyte effacement (LEE), which encodes a type III protein secretion system, an outer-membrane protein adhesin (called intimin and encoded by the
eae gene), and a translocated intimin receptor (Tir), as well as other type III secreted proteins (
8,
14). Many EPEC strains also carry an adherence factor plasmid (pEAF) that encodes bundle-forming pili (Bfp), which promote bacterial adherence to epithelial cells and are essential for virulence (
7,
25,
39).
Carriage of the
bfpA gene, which encodes the major structural pilin subunit, is used to classify EPEC into two major subgroups, known as typical (Bfp positive) and atypical (Bfp negative) EPEC (
19,
41). Typical EPEC bacteria adhere to HEp-2 cells in a localized pattern, whereas atypical EPEC, if they adhere to HEp-2 cells at all, do so in a variety of patterns, termed localized-like adherence, diffuse adherence, and aggregative adherence (
33,
41). Despite their lack of Bfp, the results of epidemiological, clinical, and volunteer studies indicate that atypical EPEC are able to cause diarrhea (
25,
33,
41).
Given that, as a group, atypical EPEC lack Bfp and display variable patterns of adherence to HEp-2 cells, we hypothesized that atypical EPEC strains carry novel adhesin(s) responsible for these phenotypes. Other than intimin, however, only one adhesin has so far been described in an atypical EPEC strain. This is a novel afimbrial adhesin called the locus for diffuse adherence (LDA), which was present in an atypical EPEC strain (O26:H11) isolated from an infant with diarrhea (
36). However, the prevalence of LDA in other atypical EPEC strains is low (
36). The aim of this study was to identify the determinants of atypical EPEC strain E128012 (O114:H2) which allow this strain to adhere to HEp-2 cells. Originally isolated from an infant with sporadic diarrhea in Bangladesh, E128012 shows localized-like adherence to HEp-2 cells and, when fed to volunteers, caused diarrhea of severity similar to that caused by a typical EPEC strain, E2348/69 (
25). Our results indicated that atypical EPEC strain E128012 requires an intact
pst-phoU operon to adhere to HEp-2 cells and, moreover, that
Citrobacter rodentium strain ICC169, an A/E pathogen of mice that is used as a model of infections with A/E strains of
E. coli, requires the same operon for virulence.
MATERIALS AND METHODS
Bacterial strains, media, and culture conditions.
The bacterial strains and plasmids used in this study are listed in Table
1. Strains were maintained on Luria-Bertani (LB) medium and grown overnight at 37°C with shaking unless otherwise stated. Where necessary, the following antibiotics were used at the indicated concentrations per milliliter: ampicillin (Amp; 100 μg), kanamycin (Kan; 50 μg), tetracycline (Tet; 12.5 μg), chloramphenicol (Cam; 25 μg), and nalidixic acid (Nal; 50 μg). 5-Bromo-4-chloro-3-indolyl-phosphate (XP) was used at a final concentration of 50 μg/ml, together with 0.2% (wt/vol) glucose, to detect alkaline phosphatase activity. To grow bacteria in known concentrations of phosphate, minimal medium containing 121 salts (
40), 0.2% (wt/vol) glucose, 0.01 mM Casamino Acids, and 0.01 mM thiamine was made without added phosphate, after which various amounts of KH
2PO
4 were added to a final concentration of 6.5 mM for high-phosphate medium or 65 μM for low-phosphate medium, respectively. The final concentration of phosphate was determined by flow injection analysis (
11).
To compare the growth kinetics of the bacterial strains used in this study in different media, overnight cultures of the test strains, grown in LB, were diluted 1 in 50 and allowed to grow at 37°C with shaking in LB, minimal essential medium, or high- or low-phosphate medium in a Klett flask. Absorbance was measured at regular time intervals by using a Klett-Summerson colorimeter (Klett Manufacturing Co., Inc., Brooklyn, NY).
Recombinant DNA techniques.
Routine DNA manipulations were performed by using standard techniques (
1,
35), with the buffers and instructions supplied by the manufacturers of the kits and reagents used. Genomic and plasmid DNA were isolated by using the cetyltrimethylammonium bromide method (
1) and a Wizard plus SV DNA purification system (Promega, Madison, WI), respectively. PCR amplifications were performed using Vent proofreading DNA polymerase (New England Biolabs, Ipswich, MA) or high-fidelity Platinum
Taq DNA polymerase (Invitrogen, Carlsbad, CA). Synthetic oligonucleotides for PCR and sequencing (Table
2) were obtained from GeneWorks Pty., Ltd. (Hindmarsh, South Australia, Australia).
Transposon mutagenesis and Southern hybridization.
Tn
phoA was introduced into atypical EPEC strain E128012 on the suicide plasmid pRT733 by conjugation, as described previously (
38). Forty-eight blue colonies were selected on LB agar containing Kan and XP and tested for loss of adherence to HEp-2 cells. A 2.8-kb BglII fragment from pRT733 that spans the BamHI site of Tn
phoA was labeled with digoxigenin by the random primer method (Roche Diagnostics, Mannheim, Germany) and used in Southern blotting to determine whether mutants contained one or more transposon insertions. The insertion site of Tn
phoA in each nonadherent mutant was determined by using inverse PCR to amplify the sequences flanking the transposon (
29). Briefly, genomic DNA of each nonadherent mutant was digested with BamHI and EcoRV, the resulting BamHI 5′ overhangs were filled in using the Klenow fragment of DNA polymerase I (New England Biolabs), and the products were recircularized by self-ligation. The unknown region was amplified by PCR using primers T1 and T2 for the region upstream of the transposon and primers T3 and T4 for the region downstream of the transposon (Table
2). DNA sequencing was performed by using an ABI Prism BigDye Terminator cycle sequencing kit, version 3.1 (Applied Biosystems). Reaction mixtures were analyzed at the Australian Genome Research Facility (Parkville, Victoria, Australia), and sequences were edited and assembled in contiguous sequences by using the Sequencher program (Gene Codes, Ann Arbor, MI). BLAST searches and sequence analyses were conducted using databases at the National Center for Biotechnology Information (
http://www.ncbi.nlm.nih.gov/BLAST/ ) and the CLUSTAL W (
http://clustalw.genome.jp ) websites.
Construction of nonpolar pstCA and phoB mutants.
Knockout mutations were constructed in
E. coli E128012 and
C. rodentium by using overlapping extension PCR (
9) and the “gene gorging” technique described by Herring et al. (
18). First, ∼0.6 kb of DNA flanking the target genes was amplified by using primer pairs pstCAF/pstCAKanR and pstCAKanF/pstCAR for
pstCA and phoBF/phoBKanR and phoBKanF/phoBR for
phoB. The fragment length polymorphism (FLP) recombinase target (FRT)-flanked Kan resistance (Kan
r) gene from pKD4 (
12) was amplified by using primers pKD4F and pKD4R. This product, together with each pair of amplified flanking regions, was used as the template in a PCR using primer pairs pstCAISceIF/pstCAISceIR (
pstCA) and phoBISceIF/phoBISceIR (
phoB) (Table
2). The I-
SceI-flanked PCR products were cloned into pGEM-T Easy to yield the donor plasmids required for gene gorging. These plasmids and pACBSR, which carries the λ Red recombinase genes and the gene for I-
SceI under an arabinose-inducible promoter, were cointroduced into electrocompetent
E. coli E128012 or
C. rodentium cells. Mutants were selected on LB plates supplemented with Kan. All mutations were confirmed by PCR using primers flanking the targeted region and primers within the Kan
r gene. When required, the Kan
r gene was excised by using the FRT sites that flank the Kan
r gene and FLP helper plasmid pCP20 (
10). E128012 and ICC169
pstCA phoB::
kan double mutants were achieved by the introduction of Δ
phoB::
kan by allelic exchange in the
pstCA mutant strains. The Kan
r gene was excised accordingly.
Construction of trans-complementing plasmids.
Wild-type pstCA was amplified from E128012 genomic DNA by using primers pstCAcF and pst4′. The resultant 1.8-kb, gel-purified, blunt-end PCR product was ligated with SmaI-linearized pBSII. This plasmid, designated pAC1, was digested with BamHI and EcoRV to release the insert, which was then ligated to BamHI- and EcoRV-digested pACYC184 to give pAC2, which carried pstCA behind the Tetr promoter of pACYC184.
The wild-type pstCA gene was amplified from C. rodentium genomic DNA (pstCACR) using primers CrpstCAcF and CrpstCAcR. The resultant purified 1.9-kb PCR product was ligated with pGEM-T Easy vector and then linearized by digestion with NcoI, and the 5′ overhangs were filled in as described above. The resultant fragment was then digested with SalI, gel purified, and cloned into EcoRV- and SalI-digested pACYC184 to yield pAC4, which possessed wild-type pstCACR behind the Tetr promoter of pACYC184.
Quantitative real-time RT-PCR.
Overnight cultures of
E. coli and
C. rodentium strains were inoculated 1:50 in LB and grown to an optical density at 600 nm of 0.6. Ten milliliters of culture was incubated with 20 ml of RNAprotect solution (Qiagen, Valencia, CA) at room temperature for 10 min, after which cells were pelleted and RNA was purified by using a FastRNA pro blue kit (Qbiogene, Inc., Carlsbad, CA). The samples were treated with DNase I before further purification using an RNeasy MinElute kit (Qiagen). Real-time PCR was performed with an MxPro-Mx3005P multiplex quantitative PCR system (Agilent Technologies, Santa Clara, CA). First-strand cDNA synthesis was performed with 5 μg of total RNA, SuperScript II reverse transcriptase (Invitrogen), and random primers (Invitrogen) according to the manufacturer's recommendations. Each 25-μl reaction mixture contained 10 ng cDNA, 300 nM of each specific primer (Table
2), and 12.5 μl 2× SYBR green master mix (Applied Biosystems, Foster City, CA). All reverse transcription-PCR (RT-PCR) data were normalized with the results for the housekeeping gene
rpoD, and the relative expression ratio of the target gene was calculated as described by Pfaffl (
30).
Adherence of bacteria to cultured epithelial cells.
HEp-2 cell adherence assays were performed as previously described (
33). Cells were examined by using bright-field microscopy for characteristic patterns of adherence and photographed with a Leica DC2000 digital camera (Leica Microsystems AG, Wetzlar, Germany). Quantitative bacterial adherence to HEp-2 cells was expressed as the number of cells with five or more attached bacteria as a percentage of the total number of cells counted. Each assay was performed in triplicate, with at least 100 cells counted for each bacterial strain.
To determine the ability of atypical EPEC strain E128012 and its derivatives to adhere to polarized cells, T84 cells of human intestinal origin were grown in a 1:1 mixture of Dulbecco's modified Eagle's medium and Ham's F-12 medium containing 5% fetal calf serum in 5% CO2 at 37°C. For cell adherence assays, T84 cells were seeded in 24-well plates at a density of 7.5 × 104 cells per well and were used when just confluent (7 to 10 days). Before infection with E. coli, the growth medium was replaced with medium containing 0.5% fetal calf serum and 0.5% d-mannose. Overnight cultures of E. coli were diluted 1 in 33 in LB, grown to early log phase at 37°C, and then added to the T84 cells at a multiplicity of infection of 100:1. Bacteria and monolayers were incubated for 3 h at 37°C in 5% CO2, after which nonadherent bacteria were removed by washing in phosphate-buffered saline (PBS) and the numbers of attached bacteria were determined by lysing the T84 cells in 100 μl of 1.0% Triton X-100 (Sigma Chemical Co., St. Louis, MO) and enumerating the bacteria on LB agar.
Alkaline phosphatase.
Alkaline phosphatase activity was determined as described by Brickman and Beckwith (
6). Briefly, the optical densities at 600 nm of cultures grown overnight in high- or low-phosphate medium were recorded and a measured amount of the culture was centrifuged to pellet the bacteria. The bacteria were then resuspended in 1 M Tris HCl, pH 8.0, and permeabilized with 0.1% sodium dodecyl sulfate and chloroform. The alkaline phosphatase activity was determined by using
p-nitrophenol phosphate and was expressed in Miller units as the mean and standard deviation (SD) of the results of at least three separate assays.
Infection of mice.
Four- to five-week-old male C57BL/6 mice were bred, housed, and maintained in the Department of Microbiology and Immunology animal facility at the University of Melbourne. Animals in this facility are certified free of infection with C. rodentium and other common bacterial, viral, and parasitic infections of laboratory mice. For single-strain infections of mice, each of nine mice per group was inoculated by oral gavage with approximately 2 × 109 CFU of an overnight culture of a test strain of C. rodentium in 200 μl of PBS. Control animals received 200 μl of sterile PBS. Fecal samples were recovered aseptically for up to 20 days after inoculation, and the number of viable C. rodentium bacteria per gram of stool was determined by plating onto selective medium. The limit of detection was 100 CFU/g feces.
For mixed-strain infections, five mice were inoculated perorally with approximately 10
9 CFU of a mutant or complemented mutant strain together with an approximately equal number of wild-type
C. rodentium cells in 200 μl of PBS. Mice were killed 7 days after infection; their colons excised; and the contents removed, serially diluted, and spread on two LB agar plates containing appropriate antibiotics to determine the proportion of wild-type
C. rodentium bacteria to mutant or complemented mutant bacteria. The ability of the mutant or trans-complemented mutant to compete with the wild-type strain was determined for three to five mice and expressed as the competitive index (CI), which was the proportion of mutant or complemented mutant to wild-type bacteria recovered from animals divided by the proportion of the mutant or complemented mutant to wild-type bacteria in the inoculum (
17). Mutants with a CI of less than 0.5 were considered to be attenuated.
Colonic hyperplasia.
At days 6, 10, 14, and 20 after infection, one mouse from each single-strain-infected group was killed and 4 cm of the colon, beginning at the anal verge, was excised. The contents were removed, and the colon was weighed and fixed in 10% (wt/vol) neutral buffered formalin or 10% (wt/vol) glutaraldehyde for histological examination. Formalin-fixed sections were stained with hematoxylin and eosin as described previously (
17) and photographed using a Leica DC2000 digital camera. The crypt heights of well-oriented sections were measured by micrometry, with at least 10 measurements taken in the distal colon of each mouse. Glutaraldehyde-fixed sections were processed and examined by transmission electron microscopy as described previously (
34).
Stool water content.
The water content of feces in the distal colon of mice infected with
C. rodentium was determined as described by Guttman et al. (
15,
16). Briefly, 7 days after inoculation with PBS or a test strain of
C. rodentium, mice were killed, and the distal 3.5 cm of the large intestines were excised. The contents of the excised intestines were removed and weighed immediately and again after drying at 37°C for 48 h. The difference between the wet and the dry weights was used to calculate the percentage of water in the gut contents.
Statistical analyses.
Statistical analyses were performed using the Instat and Prism software packages (GraphPad Software, San Diego, CA). A two-tailed P value of <0.05 was taken to indicate statistical significance.
Nucleotide sequence accession numbers.
The complete sequences of the pstSCAB-phoU operon and the phoB genes of EPEC strain E128012 and C. rodentium strain ICC169 have been deposited in the GenBank database under accession numbers FJ377883, FJ393267, FJ415986, and FJ415987.
DISCUSSION
The Pho regulon is a global regulatory network that bacteria use to manage phosphate acquisition and metabolism (
23). At the core of the regulon is a two-component system which activates or inhibits transcription, comprising PhoR, an inner-membrane histidine kinase sensor protein, and PhoB, a response regulator that is a DNA-binding protein (
4,
44). In
E. coli, the Pho regulon comprises at least 47 genes (
23), although this is likely to be an underestimate given that as many as 400 genes in
E. coli respond to environmental concentrations of phosphate (
42).
A key component of the Pho regulon is the Pst system, which captures periplasmic inorganic phosphate and transports it into the cytosol. Pst comprises four elements: PstS, a periplasmic protein that binds inorganic phosphate; PstC and PstA, which form an inner membrane channel for phosphate transport; and PstB, a permease that provides the energy needed to transport phosphate (
23). The Pst system also regulates the entire Pho regulon by preventing the activation of PhoB in phosphate-rich environments. Thus, in
E. coli, mutations in the Pst system lead to constitutive expression of the Pho regulon regardless of phosphate concentration. Although PhoB is normally activated by PhoR, it is subject to cross-regulation by other sensor proteins in response to a variety of environmental signals other than phosphate (
23). At least six such histidine kinases, QseC, ArcB, CreC, KpdD, BaeS, and EnvZ, can activate PhoB in the absence of PhoR (
45). The Pho regulon is also interrelated with the stress response (
23).
Despite extensive research on phosphate uptake and phosphate-related gene regulation in bacteria, evidence of the contribution of the Pho regulon to virulence gene expression has emerged only recently (reviewed in reference
23). For example, mutations in
pst genes can interfere with the expression of virulence-associated type III protein secretion systems of
Edwardsiella tarda and
Salmonella enterica (
3,
26,
31) and diminish the virulence of avian pathogenic
E. coli for chickens (
22). In addition,
pstS mutants of porcine EPEC show reduced adherence to piglet ileal explants (
2), and a
phoB mutant of
Vibrio cholerae showed reduced ability to colonize the rabbit small intestine (
43). Given that the intestine contains high concentrations of phosphate, these observations suggest that stimuli other than phosphate concentrations are responsible for the reduced virulence of some Pho regulon mutants. Among the Pho-regulated systems that may be relevant in this regard are responses to changes in pH and other environmental stimuli; the expression of surface components, including adhesins; and the capacity to form biofilms (reviewed in reference
23).
In this study, we used Tn
phoA mutagenesis to identify adhesins of
E. coli E128012, an atypical EPEC strain of proven pathogenicity (
25) that adheres to HEp-2 cells in a localized-like pattern. In all six PhoA-positive, nonadherent mutants that we identified, Tn
phoA had inserted into the
pst operon. The results of subsequent sequence and deletion analysis and trans-complementation studies confirmed that strain E128012 requires
pstCA to adhere to HEp-2 and T84 epithelial cells. In addition, by showing that adhesion could be restored to a
pstCA mutant of E128012 by inactivating
phoB, we established that the contribution of
pstCA to bacterial adherence is exerted via the Pho regulon. Similar results were obtained in mouse infection studies with site-directed
pstCA and
phoB mutants of
C. rodentium, thus establishing the role of the Pho regulon in the virulence of some A/E strains of enterobacteria. In addition, our finding that
pst mutants of EPEC and
C. rodentium grew equally well in high- and low-phosphate medium indicated that phosphate starvation was not responsible for the attenuation of these strains and suggested that, as with some other enteric pathogens, Pho regulation in these bacteria may be effected via cross-regulation by signals other than phosphate concentration (
23).
Recently, Ferreira and Spira reported that the
pst operon enhances the adhesion of
E. coli LRT9, a typical EPEC strain, to cultured epithelial cells (
13). They concluded that the reduced cell adherence of a
pst mutant of LTR9 was not mediated via the Pho regulon, because adherence was not restored by mutating
phoB. They also found that the
pst mutant showed reduced expression of the principal adhesins of typical EPEC, namely, Bfp and intimin, partly as a consequence of reduced expression of the
per operon, which positively regulates the expression of
bfp and several LEE-encoded genes, including
eae. Our findings are in broad agreement with those of Ferreira and Spira regarding the requirement by EPEC for an intact
pst operon to adhere to epithelial cells and that signaling through
pst is probably unrelated to phosphate concentrations, but there are several important points of difference. First, atypical EPEC and
C. rodentium lack the adherence factor plasmid, pEAF, and hence do not express Bfp or Per. Second, attenuation of the
pst mutants investigated in this study was clearly mediated through the Pho regulon, because in
pst mutants of both EPEC and
C. rodentium, the wild-type phenotype was restored to the mutants after inactivation of
phoB. Third, the attenuation of a
pstCA mutant of
C. rodentium for mice was evidently not mediated via reduced expression of LEE-encoded genes, given that the expression of two key LEE genes, namely
ler and
eae, was normal in the
pstCA mutant and that the mutant evoked A/E lesions indistinguishable in extent and severity from those induced by the parent strain in the colons of mice (Fig.
4). The observation that mice infected with a
pst mutant of
C. rodentium developed A/E lesions but not colonic hyperplasia shows that these two pathological outcomes are not interdependent. This confirms our previous observations that a prerequisite of colonic hyperplasia is extensive colonization of the colon by
C. rodentium (
21,
24) and suggests that any situation which reduces colonization is likely to affect hyperplasia.
In conclusion, we have shown that pst genes acting through the Pho regulon are required by atypical EPEC to adhere to epithelial cells and by C. rodentium to colonize the mouse intestine and cause diarrhea. Although we did not achieve our original aim, namely, to identify novel adhesins of atypical EPEC, our findings indicate that adherence of atypical EPEC and C. rodentium is mediated by one or more adhesins that are negatively regulated either by PhoB itself or by PhoB-regulated genes. We are currently using microarray analysis to identify downstream genes that EPEC and C. rodentium require for adherence.