Classification of perA sequences and their correlation with autoaggregation in typical enteropathogenic Escherichia coli isolates collected in Japan and Thailand


Mariko Iida, Graduate School of Health Care Sciences, Tokyo Medical and Dental University, 1-5-45 Yushima, Bunkyo-ku, Tokyo, 113-8510, Japan.
Tel: +81-3-5803-4515; fax: +81-3-5803-5375; email:


Enteropathogenic Escherichia coli (EPEC) strains produce a bundle-forming pilus (BFP) that mediates localized adherence (LA) to intestinal epithelial cells. The major structural subunit of the BFP is bundlin, which is encoded by the bfpA gene located on a large EAF plasmid. The perA gene has been shown to activate genes within the bfp operon. We analyzed perA gene polymorphism among typical (eae- and bfpA- positive) EPEC strains isolated from healthy and diarrheal persons in Japan (n= 27) and Thailand (n= 26) during the period 1995 to 2007 and compared this with virulence and phenotypic characteristics. Eight genotypes of perA were identified by heteroduplex mobility assay (HMA). The strains isolated in Thailand showed strong autoaggregation and had an intact perA, while most of those isolated in Japan showed weak or no autoaggregation, and had a truncated perA due to frameshift mutation. The degree of autoaggregation was well correlated with adherence to HEp-2 cells, contact hemolysis and BFP expression. Our results showed that functional deficiency due to frameshift mutation and subsequent nonsense mutation in perA reduced BFP expression in typical EPEC strains isolated in Japan.

List of Abbreviations: 

amino acid




bundle-forming pilus


brain heart infusion




diffuse adhesion


DNA Data Bank of Japan


Dulbecco's modified Eagle medium


EPEC adherence factor


ethylenediaminetetraacetic acid


Enteropathogenic Escherichia coli


fetal bovine serum


heteroduplex mobility assay


immunoglobulin G


localized adherence


locus of enterocyte effacement


National Institute of Infectious Diseases




optical density


phosphate-buffered saline


polymerase chain reaction


red blood cells


standard deviation


scanning electron microscopy


translocated intimin receptor


type III secretion system


yttrium aluminium garnet

Enteropathogenic Escherichia coli (EPEC) causes diarrhea which represents a major health problem among infants, particularly in developing countries (1). EPEC produces localized adherence (LA) to HEp-2 cell monolayers and characteristic attaching-and-effacing (A/E) lesions on intestinal epithelial cells (2–5). The A/E phenotype is encoded by a cluster of genes including the eae gene located on the locus of enterocyte effacement (LEE), a ∼35 kb pathogenicity island in the E. coli chromosome. LA is caused primarily by type IV fimbriae known as a bundle-forming pilus (BFP) which is encoded by a cluster of 14 bfp genes located on a large virulence plasmid called the EPEC adherence factor (EAF) plasmid (6–10). The first gene of the cluster, bfpA, encodes bundlin, the major structural subunit of BFP. BFP is also involved in bacteria-bacteria interaction and subsequent autoaggregation (11). In addition, the bfpF gene, which encodes a putative nucleotide-binding protein, is required for the dispersal phase of EPEC autoaggregation (12–14). N-acetyllactosamine is presumed to be essential for the BFP receptor on epithelial cells (15). Studies on adult volunteers have demonstrated that intimin, the EAF plasmid and BFP are essential virulence determinants of EPEC (13, 16, 17).

Recently EPEC strains have been classified as typical or atypical. Typical EPEC strains possess both the eae gene and EAF plasmid, whereas atypical EPEC strains do not possess the EAF plasmid (18). Recent studies have suggested that bfp-defective strains become less virulent (19, 20) and Tennant et al. have reported that atypical EPEC expresses functional type I pili instead of BFP (21). Most of the EPEC strains isolated in Japan are atypical EPEC (22, 23). In addition to other bfp genes, the EAF plasmid contains the perA, B, and C (also called bfpT, V, and W) genes (24–26). It has been demonstrated that perA and perC are important for full expression of the bfpA and LEE genes (25). In addition, perA activation is assisted by perC (27). The perC homologue (pch) is found in atypical EPEC strains (28). Though polymorphism of the perA gene (29) has been reported elsewhere, such polymorphism has not been seen in EPEC isolates in Japan.

In EPEC, the type III secretion system (TTSS) mediates the delivery of a protein known as translocated intimin receptor (Tir) (30, 31). TTSS-positive strains have been shown to cause hemolysis after adhesion to sheep red blood cells (RBC) (contact hemolysis) (32), and a contact hemolysis assay is considered to be a convenient method of detecting the TTSS in E. coli.

Variants of bfpA, which are clusters of 2 main clades are widely known (33). We previously identified 5 types of bfpA genotype (34) using a heteroduplex mobility assay (HMA). Although comparisons of phenotypic activities among these variants have been attempted, there are few detailed reports on this. In this study, we examined typical EPEC strains isolated from diarrheal and healthy persons for polymorphism of the bfpA and perA genes, presence or absence of BFP-related genes, and such virulence-associated characteristics as autoaggregation, adherence to HEp-2 cells and contact hemolysis.


Bacterial strains

The nucleotide primer sets eaek1/eaek4 and bfpAks/bfpAkcomas were used for PCR to amplify and identify eae and bfpA genes, respectively (Table 1). A total of 53 typical EPEC strains (eae+bfpA+) isolated in Japan (27 strains) and Thailand (26 strains) from healthy humans and patients with diarrhea, and 2 reference EPEC strains, E2348/69 (O127a: H6) (17) and 886L (O111: H2), were used in this study. In addition, the KI1924 and KI1455 strains, neither of which has the eae nor bfpA gene, were used as negative controls.

Table 1.  Oligonucleotide primers used in this study
PrimerTarget gene5′- Sequence -3′Temperature (C)CycleReference
  1. PCR was performed for 25 cycles for detection of bfpA and for 35 cycles for DNA sequencing.

  2. bfpA primer set for HMA; nucleotides 154 to 366 of bfpA gene (582bp).

  3. §perA primer set for HMA; nucleotides 277 to 511 of perA gene (825bp).

  4. eae primer set for HMA; nucleotides 95 to 528 of eae gene (2820bp).

perAkas2§perATAACCCTGTCTACGATGCTC5525This study
perAmas1perATCATTTCGAGTGCTCATTGC5535This study
perCkas1perCTTACAAGCCCCATTTTCTTA5525This study


The O and H serotypes were determined with antisera kits (Denka-Seiken, Tokyo, Japan) and H8-antisera (Statens Serum Institut, Copenhagen, Denmark).

PCR assay

Detection of eae and BFP-related genes (bfpA, bfpF, perA, perC, and pchA) was performed by PCR using specific primers for amplification. The specific primers used in this study are shown in Table 1. The DNA template was prepared by suspension of a bacterial culture grown overnight on an antibiotic medium 3 agar plate (Difco, BD, Sparks, MD, USA) with 100 μl of distilled water, followed by boiling for 10 min. PCR assays were performed in 25 μl of a reaction mixture consisting of PCR buffer (20 mM Tris-HCl pH 8.4, 50 mM KCl, and 1.5 mM MgCl2), 0.1 mM dNTPs, 0.1 μM of each primer, 1 unit/0.2 μl of Taq polymerase (Promega Corporation, Madison, WI, USA) and 2 μl of template DNA. The reactions were run in a DNA thermal cycler 9600 (Roche Molecular Biochemicals, Indianapolis, IN, USA) for 25 cycles of denaturation (94 C for 30 sec), annealing (50 C or 55 C for 1 min), and extension (72 C for 1.5 min), with a final extension at 72 C for 10 min. PCR products were electrophoresed on a 13% polyacrylamide gel electrophoresis system and visualized with ethidium bromide under ultraviolet light.

Heteroduplex mobility assay (HMA)

The typing of eae and bfpA was performed by HMA as previously described (34, 35). HMA is a convenient way of determining the similarity of sequences from their heteroduplex mobility in polyacrylamide gel electrophoresis (36). Amplicons obtained from the bfpA-PCR and perA-PCR were subjected to HMA. An appropriate amount of amplicons was mixed with 2 μl of the amplicons from a reference strain, 2 μl of 50 mM EDTA [pH 8.0], and sterile distilled water added to 10 μl. The mixture was denatured at 94 C for 5 min, re-annealed at 72 C for 3 min and at 50 C for 1 hr. The heteroduplexes were electrophoresed on a 10% polyacrylamide gel, containing 5% stacking gel, in Tris-glycine buffer without SDS.

Nucleotide sequence and phylogenetic analysis

The PCR product of each HMA-type strain was purified using an AutoSeq G-50 PCR Purification Kit (GE Healthcare, Buckinghamshire, UK) in accordance with the manufacturer's recommendations. Purified PCR fragments were sequenced with an ABI Prism 3100 DNA sequencer (Applied Biosystems, Carlsbad, CA, USA). Amino acid sequence data were aligned and phylogenetic trees were produced using the CLC sequence viewer (CLC bio, Aarhus, Denmark).

Counting of autoaggregated mass of EPEC

Bacterial strains were grown overnight in brain heart infusion (BHI; BBL, Sparks, MD, USA) broth at 30 C. Overnight cultures were diluted 1:250 into 20 ml of Dulbecco's modified Eagle medium (DMEM) F-12 (Gibco, Carlsbad, CA, USA) and shaken at 250 rpm for 3 hr in 50-ml conical polypropylene tubes at 37 C. Cell mass numbers were counted with a Multisizer 3 system (Coulter Scientific Instruments, Inc, Fullerton, CA, USA) fitted with a 30 or 50 μm aperture.

Morphological observation of autoaggregates

A drop of autoaggregated culture was placed on a five-window microscope slide (Sekisui Chemical, Tokyo, Japan), and each culture was examined with the naked eye and with phase-contrast microscopy at a magnification of ×400. Categories were determined by comparison of the size of aggregates.

Quantitation of autoaggregation

To determine categories of autoaggregation, two equivalent 10 ml samples were removed from each culture. The OD600 of the first sample was measured immediately using a spectrophotometer and the second sample was kept for 30 min at 4 C for precipitation. The supernatant containing the aggregate was mixed for 30 sec on a vortex mixer and trypsinized for 5 min at 4 C before measurement of OD600. The autoaggregation index was calculated by subtracting the OD600 of the first sample from that of the second, dividing the result by the OD600 of the first sample, and multiplying by 100.

Electron microscopy

Suspensions of autoaggregates were placed on silane-coated glass slides, fixed in 2.5% glutaraldehyde and then postfixed in 1% osmium tetroxide in 0.1 M PBS. The slides were then dehydrated in a graded series of ethanol and dried in a critical point drying apparatus HCP-2 (Hitachi Ltd., Tokyo, Japan.) with liquid CO2. Next, they were spatter-coated with platinum using a E102 system (Hitachi Ltd., Tokyo, Japan.) and examined using a S-4500 scanning electron microscope (Hitachi Ltd., Tokyo, Japan) and an yttrium aluminium garnet (YAG) backscattered detector (Hitachi Ltd., Tokyo, Japan).

HEp-2 cell adherence assay

HEp-2 cells that had been maintained in DMEM supplemented with 10% fetal bovine serum (FBS; Gibco) were plated onto cover slips in 24-well microtiter plates (Corning) at a density of 105 cells/ml and then incubated at 37 C for 16 hr in the presence of 5% CO2. After washing the HEp-2 cells three times in DMEM without FBS, 107 bacterial cells were inoculated into each well or slide, which contained FBS-free DMEM, and were incubated for 1 hr at 37 C in the presence of 5% CO2. The cells were then washed three times with phosphate-buffered saline (PBS), fresh medium was added, and they were incubated for another 3hr. The monolayers were then washed three times with PBS and fixed with 4% paraformaldehyde in PBS for 30 min and then stained with Giemsa solution for 45 min. The coverslips were examined using a light microscope to determine the adherence of the strains. Categories (+++ to −) were assigned through comparison of the size of the microcolonies.

Contact hemolysis assay

Contact hemolysis assay was performed as previously described with slight modification (32). Bacterial strains were grown overnight in BHI broth at 30 C. The cultures were then diluted 1:100 into 5 ml of DMEM and shaken at 250 rpm for 100 min in 15-ml conical polypropylene tubes at 37 C. Next, the bacterial aggregates were centrifuged at 2,000 × g for 15 min. The pellets were resuspended into 5 ml of DMEM, 50 μl of which was placed onto 96-well microtiter plates and monitored for viable bacteria at an optical wavelength of 600 nm. 50 μl of a 25% sheep RBC (Nippon Biotest Laboratories Inc., Tokyo, Japan)-DMEM suspension was added and this was centrifuged at 1,000 × g for 15 min to form a close EPEC-RBC contact. After 2 hr of incubation at 37 C, bacterium-RBC pellets were gently resuspended to facilitate the release of hemoglobin. Cells were centrifuged at 1,000 × g for 15 min, and the supernatant was monitored for released hemoglobin at an optical wavelength of 550 nm. Similarly-treated uninfected RBCs were used as a spectrophotometric zero. The hemolysis ratio was calculated using E2348/69 as a standard.

Analysis of bfpA gene expression by reverse transcription (RT)-PCR

RT-PCR was used to analyse the transcriptional expression of the bfpA gene indicating expression of the bundlin. Overnight bacteria cultures were diluted 1:100 in DMEM F-12 broth and grown to the mid-logarithmic phase (OD600= 0.5) at 37 C with shaking. Cultures were pelleted by centrifugation at 13,000 × g for 10 min, and RNA was isolated using TRIzol® Reagent (Invitrogen, Faraday Avenue, CA, USA) according to the manufacturer's instructions. Total RNA (1 μg) and 60 μM of random hexamers (Roche, Mannheim, Germany) were incubated for 10 min at 65 C, immediately cooled on ice and then reverse transcribed in a final volume of 20 μl-containing 1 mM of deoxynucleotide mix, 20 U of RNase inhibitor, Transcriptor RT reaction Buffer 1× and 10 U of Transcriptor reverse transcriptase (Roche)-that was reacted for 30 min at 55 C. PCR amplification of cDNA was performed with an initial denaturation step of 5 min at 94 C, followed by 19 cycles of 30 sec at 94 C, 1 min at 55 C and 1.5 min at 72 C, and finishing with one cycle of 10 min at 72 C, using primer sets for the bfpA gene (Table 1). The number of PCR cycles used came within the linearity range for PCR amplification and constitutive expression of 16S rRNA assessed from the same cDNA preparation was used as a standard. Samples (10 μl) of each PCR product were separated by electrophoresis in 2.0% agarose and visualized by ethidium bromide staining. The bands of the bfpA gene were confirmed visually and results were standardized with the 16S rRNA band density.


The aggregate cultures which had been incubated in DMEM to induce BFP expression were used. After centrifugation at 10,000 × g for 1min, the supernatant solutions were removed and the resulting bacterial cell pellets were resuspended in 1 ml of cell lysis buffer (10 mM Tris-HCl [pH 8.0], 1 mM EDTA, 1% SDS). The optical density of a portion of these samples was measured on a spectrophotometer at 595 nm. Aliquots (75 μl or less) of these samples were mixed with 20 μl of 4× sodium dodecyl sulfate-polyacrylamide gel electrophoresis loading buffer (Invitrogen) and then adjusted to a total volume of 100 μl with additional cell lysis buffer such that the resulting gel samples were derived from roughly equivalent densities of bacteria. Five microliters of each gel sample were loaded per lane of a sodium dodecyl sulfate-12.5% polyacrylamide gel. After electrophoretic separation, the protein in the gel was electrotransferred to a polyvinylidene difluoride membrane. The membranes were blocked with 5% (wt/vol) nonfat dry milk in PBS (pH 8.0) plus 0.05% (vol/vol) Tween 20. Primary, affinity-purified rabbit α1 bundlin antisera (37) were used at a dilution of 1:2,000 in PBS plus 5% nonfat dry milk and 0.05% Tween 20. Bands were detected with alkaline phosphatase conjugated goat anti-rabbit IgG antibodies (Promega) at a dilution of 1:4,000 and enhanced Western blue stabilized substrate for alkaline phosphatase reagents (Promega). Band images were obtained with an image scanner.

Nucleotide sequence accession numbers

The sequences for bfpA and perA were submitted to DDBJ and given the accession numbers AB364243 and AB364244, and AB523678 to AB523702, respectively.

Pulse-field gel electrophoresis (PFGE)

Genomic DNA of the EPEC strains was prepared in agarose plugs that had been treated with lysozyme and pronase K using a Gene Path reagent kit (Bio-Rad, Tokyo, Japan) according to the manufacturer's recommendations. The DNA in agarose plugs was digested with 20 U of the restriction endonuclease XbaI (Roch Diagnostics, Tokyo, Japan). The DNA fragments generated were then separated through a 1% agarose gel in Tris-borate-EDTA buffer at 14 C in a CHEF-DR II (Bio-Rad Laboratories Inc., Hercules, CA) with the following electrophoresis conditions: initial switch time of 2.2 sec, final switch time of 54.2 sec, 6 V/cm, at an angle of 120° for 19 hr. The resulting profiles were scanned and saved in the TIFF format to be analyzed using Bio-Numerics software (version 3.0; Applied Maths, Kortrijk, Belgium). Similarity was determined using the Dice coefficient, and clustering was based on the unweighted pair group method with arithmetic averages (UPGMA) with a band position tolerance of 1%. PFGE patterns of the strain were classified as independent clusters with similarity of 80%.


The above autoaggregation and contact hemolysis experiments were repeated three times. Results were expressed as mean ± SD. Statistical analysis was performed using Welch's t-test with correction for multiple testing. P values < 0.02 were taken as significant.



Fifty-three typical EPEC strains were classified into 20 serotypes (Table 2). Most of the strains belonging to EPEC classical serotypes (17) were isolates from Thailand (16/26 strains; 61.5%), whereas only one out of 27 strains isolated in Japan belonged to classical serotypes, though this strain (O142:H6) was isolated from someone who had traveled to the Philippines. The strains which were isolated in Japan were distributed in O153 and O157 serogroups. There were no common serotypes between those from Thailand and Japan.

Table 2.  Comparison of bacterial HMA-types, serotypes, autoaggregation index, and gene possession.
bfpA-typeperA-typeSerotypeNo. of strainsAuto-aggregationCell AdherencebfpFperCOriginCircumstance
  1. Classical serotypes of EPEC. A traveler to the Philippines. §Truncated sequence of perA.


Typing of bfpA by HMA and DNA sequencing

We previously reported 5 HMA-bfpA types (34). In this study, we identified a new type, HMA-bfpA type 6 (Fig. 1). All the strains of this type were isolates from Thailand (Table 2). Most strains isolated in Japan were bfpA types 1, 4 and 5, while, those isolated in Thailand were bfpA type 2, 3 and 6. Several serotypes could be assigned to each bfpA type.

Figure 1.

Alignment of amino acid sequences encoded by the bfpA gene. Invariant amino acids are indicated by “.”, variant amino acids by alphabet, and gaps by “-”.

Typing of perA by HMA

The perA genes were classified as 8 HMA-types (Table 2). Most strains isolated in Japan were perA types A and B, whereas those isolated in Thailand were perA types C to H. Although perA variation was more complex than bfpA variation, each perA genotype corresponded to a main bfpA type.

DNA sequencing of bfpA and perA

Amplicons of the bfpA gene (including new HMA-type) and perA gene were sequenced. PCR amplification was performed with whole coding region primers (Table 1). Figure 5 shows the phylogenetic tree of the perA sequences of our strains and those reported by Lacher et al. (29). The perA genotypes were clustered into four major groups, α, β, γ and δ, as described (29). Most of the isolates from Japan were in the β cluster. In this study, the new perA sequence types, β3.2, β3.3 and β3.4 were identified (Fig. 2). HMA typing produced similar results to those of sequence typing in the polymorphism analysis on bfpA and perA.

Figure 5.

Electrophoresis of RT–PCR product of bfpA gene and 16S rRNA (Top). Immunoblot of bundlin protein from EPEC whole-cell extracts (Bottom). The arrow indicates the position of bundlin (molecular mass of ca. 20 kDa). The extracts are from the following strains: lane 1, KI1857 (HMA-bfpA type 1); lane 2, E2348/69 (HMA-bfpA type 2a); lane 3, KI1923 (HMA-bfpA type 2a); lane 4, KI1318 (HMA-bfpA type 2b); lane 5, KI1540 (HMA-bfpA type 3); lane 6, KI1709 (HMA-bfpA type 1); lane 7, KI1776 (HMA-bfpA type 4a); lane 8, KI1860 (HMA-bfpA type 4b); lane 9, KI1333 (HMA-bfpA type 5); lane 10, negative control, KI1455 (bfpA-negative), respectively.

Figure 2.

Phylogenetic tree of perA encoding region genes obtained in this study. Names of strains, perA sequence types, perA HMA-types, and autoaggregation phenotypes are shown. The scale bar indicates the number of nucleotide replacements per site.

Autoaggregation assay

All except 4 strains showed autoaggregation (Table 2). Since aggregates of various sizes were observed, we defined the extent of autoaggregation according to 4 categories (+++ to –) (Fig. 3b). Those in category +++ (n= 30) were huge aggregates clearly visible with the naked eye, category ++ (n = 4) aggregates of medium thickness, and category + (n= 17) small, weak aggregates (Fig. 3b). Particle measurements were also carried out on the autoaggregates in each category and a different peak was observed for each one (Fig. 3a). When morphological changes were investigated by scanning electron microscopy, we observed microcolony structures at 3 hr post inoculation. Microcolonies in category +++ were intricately intertwined, whereas in category +, they were barely visible (Fig. 3c). The rate of aggregation was quantitated by measuring the turbidity with reference to the E2348/69 strain using the representative strain of each category (Fig. 3e). Significant differences were observed among categories (P < 0.02).

Figure 3.

Phenotypic characteristics of typical EPEC strains. Representative EPEC strains (E2348/69, KI1776, KI1860 and KI1709) are shown. (a) Size characterization of autoaggregates determined using a Coulter Multisizer 3 counter. (b) Autoaggregation index determined by observation with a phase contrast microscope (×400). Category +++: huge aggregates which could be seen clearly by the naked eye. Category ++: medium aggregates with some thickness. Category +: small, thin aggregates. Scale bars represent 20 μm. (c) SEM images of microcolonies. Scale bars indicate 1.5 μm. (d) Adherence of EPEC strains to HEp-2 cells (Giemsa stain). Different sizes of adherent EPEC microcolonies were observed. Category +++: large microcolonies. Category ++: medium microcolonies. Category +: small, thin microcolonies. Scale bars indicate 50 μm. (e) Degree of autoaggregation. Each experiment was performed 3 times in duplicate. The error bars indicate the standard deviations from the means (P < 0.02, Welch's t-test). There were significant differences in autoaggregation among categories.

Adherence to HEp-2 cells

Adherence to HEp-2 cells has been used to identify EPEC (5, 38). In this regard, LA is a qualitative adherence pattern consisting of compact microcolonies on the surface of epithelial cells. The results of the HEp-2 cell adherence assay were similar to those of the autoaggregation assay for all bfpA genotypes (Fig. 3d). The negative autoaggregation strain KI1218 showed diffuse adherence (DA) (Table 2).

Comparison of bfpA-types by autoaggregation

All strains belonging to bfpA types 2, 3 and 6 were in category +++. As for bfpA type 1 strains, 3 strains were in category ++ and 2 strains in category +. In most of the type 4 strains autoaggregation was weak or there was none, but one strain with the serotype O157:H45 showed autoaggregation of category ++ (Table 2). All strains negative for autoaggregation were the bfpA type 4a (Table 2). Most of the strains showing weak or no autoaggregation were isolates from Japan.

Contact hemolysis

We examined the hemolytic activity of the representative strains in each bfpA-genotype. Figure 4 shows the percentage hemolytic activity for EPEC in each autoaggregation category relative to that of the E2348/69 strain. There were significant differences in hemolysis among categories (P < 0.02).

Figure 4.

Contact hemolysis assay. As far as category +++ is concerned, representative EPEC strains (E2348/69, KI1318, KI1458 and KI1540) are shown. Each experiment was performed 3 times in duplicate. The error bars indicate the standard deviations from the means (P < 0.02, Welch's t-test). The degree of contact hemolysis was significantly different among categories.

Transcription and expression of bfpA gene

Selected EPEC strains were examined if they produced detectable bundlin. The prototype EPEC strain E2348/69 served as a positive control. To identify bundlin, polyclonal antiserum (37) was used to probe whole-cell extracts from each of the EPEC strains. Antisera were affinity purified after conjugation of purified soluble α1 bundlin (37). Bundlin protein was readily detected in extracts from type α (HMA-type 2), type β5 (HMA-type 3) and some type β7.1 (HMA-type 4a) strains which showed strong autoaggregation, and from type β8 (HMA-type 1 and type β7.1 (HMA-type 4a) which showed moderate autoaggregation. Bundlin was not detected in strains showing weak or no autoaggregation (Fig. 5). Transcriptional expression of the bfpA gene in the EPEC strains was also analysed by semi-quantitative RT–PCR. Electrophoresis of RT–PCR product of the bfpA gene and 16S rRNA is shown in Figure 5. Results of RT-PCR confirmed those of the Western blotting.

Possession of the BFP-related genes

We next examined strains by PCR for possession of the BFP-related genes bfpF and perC which are necessary for biosynthesis of bfpA (Table 2). Nearly all strains possessed both genes but 2 had neither of them. These 2 strains had the perC homologue (pch) instead and did not show any autoaggregation activity (data not shown).

Correlation between amino-acid sequence of perA and autoaggregation phenotype

The perA nucleotide sequences were converted into amino acid sequences as shown in Figure 6, with the amino acid sequences of α8 type (KI 2001) at the top. Completed perA amino-acid sequences were 274 aa in size. Strains showing marked aggregation had an intact perA sequence with exception of the strains of sequence type α1.4. Most of the strains isolated in Japan which showed weak or no aggregation had truncated perA amino-acid sequences (61 aa to 118 aa) due to a frame shift mutation in perA. The amino acid sequence of α5.1, β4.2 and β3.2 were identical to those of α5.3, β4.3, and and β3.3, respectively.

Figure 6.

Alignment of amino acid sequences in full-length perA gene. Invariant amino acids are indicated by “.”, variant amino acids by letters of the alphabet, and gaps by “-”. Stop codons are boxed.


The genetic similarity of the strains which were isolated in Japan was evaluated using PFGE. They were classified into six PFGE types. Serotype O157:H45 strains were classified into two types (Fig. 7). The strains which showed weak autoaggregation were distributed over wide area of Japan.

Figure 7.

Geographical location of the cases in which six subtypes of EPEC strains were isolated in Japan. PFGE cluster analysis compared to phylogenetic ancestry, virulence profile, phylogenetic group, and source of typical EPEC strains from Japan with or without diarrhea. Dendrogram of representative XbaI-digested PFGE profiles of the EPEC strains.


On examining the phenotypic characteristics of the various EPEC strains, we found that aggregates of the strains isolated in Japan were smaller and weaker than those of strains isolated in Thailand. Further, when we examined adherence to HEp-2 cells, the results were similar to those of the autoaggregation assay. The EPEC strains which showed strong autoaggregation also showed a greater degree of contact hemolysis. It seemed that the contact hemolysis would be promoted by the presence of BFPs, which would facilitate more effective adherence, so we tested these strains for the bfpA gene expression by RT-PCR and for BFP expression by performing Western blotting. mRNA of the bfpA gene and BFPs were not detected in strains which showed weak or no autoaggregation. The EPEC strain, which showed weak aggregation and pili-like structure in Figure 3c, but not BFP (Fig. 5), might have been expressing the type I pili. While, it remains to be seen why the strain with truncated perA sequences showed strong autoaggregation. We observed frame shift mutant of perA even in the E2348/69 strain which changed to weak phenotype, so these region of perA might be liable to mutation.

We also tested the EPEC strains for the presence of BFP-related genes such as bfpF and perC, and detected them in most strains. We then converted the perA genes into amino acid sequences and found that the amino acid sequences of some of the perA genotypes had been truncated by a frame shift mutation of the perA gene. Strains with a truncated perA gene showed weak or no autoaggregation and decreased HEp-2 cell adherence (Table 2). We did not find truncated amino acid sequences in bfpA genes. This study showed that most of the typical EPEC strains isolated in Japan did not express BFP, and it appeared that a truncated perA gene was connected with inhibition of BFP expression (Fig. 5).

We performed PFGE analysis to show molecular typing of EPEC strains isolated from Japan (Fig. 7). There were no relationship between PFGE profiles and bfpA polymorphism.

According to the recent studies, the prevalence of atypical EPEC has continued to increase not only in developed but also in developing countries (39). In Japan, most EPEC isolates have been classified as atypical EPEC, and even the supposedly typical EPEC strains from Japan used in this study could in fact be atypical EPEC, although bfpA genes were detected with PCR. As comparable results were obtained with HMA and DNA sequencing for bfpA and perA genes, this shows that genotyping by HMA was a useful method for classifying these genes. The distributions of bfpA and perA genotypes differed between the EPEC isolates from Japan and those from Thailand.

A study of global polymorphisms of virulence genes and their phenotypic characteristics would yield more significant information on the pathogenesis of EPEC.


We are grateful to Michael Donnenberg (University of Maryland School of Medicine) for providing alpha bundlin antisera. We also thank Sunao Iyoda (National Institute of Infectious Diseases: NIID) and Yan Lu (NIID) for assistance with HEp-2 adherence assay, and Shizuko Ichinose (Tokyo Medical and Dental University) for assistance with the electron microscopy. Hidemasa Izumiya (NIID) kindly provided the EPEC reference strain. This study was supported by grants-in-aid for Food and Chemical Safety from the Ministry of Health, Labor, and Welfare of Japan.