Shiga Toxin 2e-Producing Escherichia coli Isolates from Humans and Pigs Differ in Their Virulence Profiles and Interactions with Intestinal Epithelial Cells

Abstract

Thirteen Escherichia coli strains harboring stx_2e were isolated from 11,056 human stools. This frequency corresponded to the presence of the stx_2e allele in 1.7% of all Shiga toxin-producing E. coli (STEC) strains. The strains harboring stx_2e were associated with mild diarrhea (n = 9) or asymptomatic infections (n = 4). Because STEC isolates possessing stx_2e are porcine pathogens, we compared the human STEC isolates with stx_2e-harboring E. coli isolated from piglets with edema disease and postweaning diarrhea. All pig isolates possessed the gene encoding the F18 adhesin, and the majority possessed adhesin involved in diffuse adherence; these adhesins were absent from all the human STEC isolates. In contrast, the high-pathogenicity island encoding an iron uptake system was found only in human isolates. Host-specific patterns of interaction with intestinal epithelial cells were observed. All human isolates adhered to human intestinal epithelial cell lines T84 and HCT-8 but not to pig intestinal epithelial cell line IPEC-J2. In contrast, the pig isolates completely lysed human epithelial cells but not IPEC-J2 cells, to which most of them adhered. Our data demonstrate that E. coli isolates producing Shiga toxin 2e have imported specific virulence and fitness determinants which allow them to adapt to the specific hosts in which they cause various forms of disease.

Shiga toxin (Stx)-producing Escherichia coli (STEC) isolates, which cause diarrhea and hemolytic-uremic syndrome (HUS) in humans (19, 26, 50), generally cause minimal or no injury in their animal host reservoirs (26). The only naturally occurring diseases in animals caused by STEC are swollen head syndrome in chickens (44) and edema disease in piglets (20). Edema disease is characterized by vascular necrosis, edema, and neurological signs and can be fatal (20). Although the exact mechanisms that lead to edema disease are unknown, Stx2e and adherence-mediating virulence factors such as the F18 adhesin, F4 fimbriae, and adhesin involved in diffuse adherence (AIDA) seem to be common among strains isolated from diseased pigs (21, 35). In one study, the presence of Stx2e in the erythrocyte fraction was strongly associated with clinical disease (30). stx_2e is the most frequent stx₂ variant found in fecal samples from pigs (14), and it was the second most common stx₂ variant in environmental STEC isolates (54). In the latter study, the stx_2e variant was found not only in STEC strains isolated from pig samples but also in isolates from a dairy cattle herd, suggesting that such strains spread from pigs to cattle (54).

Stx2e-producing E. coli strains have also occasionally been isolated from humans (5, 15, 40, 52). The majority of the patients had uncomplicated diarrhea (5, 15, 40), and some had HUS (52). However, the frequency with which Stx2e-producing STEC strains occur in humans, their virulence factors, their mechanisms of interaction with the human host, their reservoir(s), and their mode(s) of transmission are poorly understood.

Here, we compared the putative virulence genes in Stx2e-producing E. coli strains isolated from humans and diseased pigs in order to assess the extent to which they are related. We also analyzed the interactions of the two groups of organisms with homologous and heterologous intestinal epithelial cells in vitro in a search for characteristics that might be related to adaptation in the host.

MATERIALS AND METHODS

Bacterial strains.

After screening 11,056 stools (9,206 from patients with diarrhea or HUS and 1,850 from asymptomatic individuals), we isolated 13 E. coli strains containing the stx_2e gene. These isolates were obtained from patients with uncomplicated diarrhea (n = 9) or from asymptomatic carriers (n = 4) and were recovered at the Institute of Hygiene and Microbiology, University of Würzburg, Würzburg, Germany, and the Institute of Hygiene, University Hospital Münster, Münster, Germany, during routine diagnostic examinations and epidemiological investigations between January 1997 and December 2003. The procedures used for STEC isolation from stools have been described previously (15). Briefly, enriched primary stool cultures were screened using PCRs for stx and eae genes, and STEC strains were isolated from PCR-positive stools using colony blot hybridization with digoxigenin-labeled stx probes (15). The 13 human isolates showed no geographical or temporal linkage. A subset of these strains was investigated for stx_2e transcription in a previous study (57). Twelve porcine STEC strains harboring stx_2e were isolated from German piglets with edema disease or postweaning diarrhea (35), while one strain (strain E57) was isolated from a pig with diarrhea in Canada (27).

stx subtyping.

The isolated strains were tested for stx₁ and stx₂ using the primer pairs KS7-KS8 (stxB₁ and stxB_1c) and LP43-LP44 (stxA₂ and stxA₂ variants) (Table 1). stx₁ and stx_1c were distinguished by HhaI restriction of the KS7-KS8 PCR products (16, 56). The strategy used to distinguish stx₂ from its variants has been described previously (15). Briefly, STEC strains positive in the PCR with primers LP43 and LP44 were subjected to PCR with primers GK3 and GK4 (Table 1), and the amplification products were digested with HaeIII (New England Biolabs, Frankfurt, Germany) to differentiate between stx₂ and stx_2c (15). Isolates in which amplification products could not be elicited with primers GK3 and GK4 were tested further for the presence of the stx_2d (41) and stx_2e (55) genes using primers VT2-cm and VT2-f and primers FK1 and FK2, respectively (Table 1). Strains positive in the PCR with primers FK1 and FK2, which target the stxB_2e subunit gene (15), were confirmed to contain the stxA_2e subunit gene using the PCR with primers FK9 and FK10 (13) (Table 1). The presence of stx_2e in PCR-positive isolates was confirmed by Southern blot hybridization with digoxigenin-labeled stxA_2e and stxB_2e probes derived from stx_2e-harboring human isolate VUB-EH60 (40) by PCRs with primer pairs FK9-FK10 and FK1-FK2 (Table 1), respectively. Moreover, the identity of stx_2e genes was verified by nucleotide sequence analysis performed as described previously (57).

TABLE 1.

PCR primers and conditions used in this study

Primer	Sequence (5′-3′)	Target(s)	PCR conditionsa			Size of PCR product (bp)	Reference	Positive controlb
			Denaturation	Annealing	Extension
KS7	CCCGGATCCATGAAAAAAACATTATTAATAGC	stxB1	94°C, 30 s	52°C, 60 s	72°C, 40 s	285	15	EDL933
KS8	CCCGAATTCAGCTATTCTGAGTCAACG	stxB1c
LP43	ATCCTATTCCCGGGAGTTTACG	stxA2 and variants	94°C, 30 s	57°C, 60 s	72°C, 60 s	584	15	EDL933
LP44	GCGTCATCGTATACACAGGAGC
GK3	ATGAAGAAGATGTTTATG	stxB2, stxB2c	94°C, 30 s	52°C, 60 s	72°C, 40 s	260	15	EDL933
GK4	TCAGTCATTATTAAACTG
VT2-cm	AAGAAGATATTTGTAGCGG	stxB2d	94°C, 30 s	55°C, 60 s	72°C, 60 s	256	41	EH250
VT2-f	TAAACTGCACTTCAGCAAAT
FK1	CCCGGATCCAAGAAGATGTTTATAG	stxB2e	94°C, 30 s	55°C, 60 s	72°C, 40 s	280	15	VUB-EH60
FK2	CCCGAATTCTCAGTTAAACTTCACC
FK9	CCCGGATCCATGAAGTGTATATTGTTA	stxA2e	94°C, 30 s	52°C, 60 s	72°C, 60 s	260	13	VUB-EH60
FK10	CCCGAATTCAGCACAATCCGCCGCCAT
SK1	CCCGAATTCGGCACAAGCATAAGC	eae	94°C, 30 s	52°C, 60 s	72°C, 60 s	863	15	EDL933
SK2	CCCGGATCCGTCTCGCCAGTATTCG	Conserved
Iha-I	CAGTTCAGTTTCGCATTCACC	iha	94°C, 30 s	56°C, 60 s	72°C, 90 s	1,305	46	4797/97
Iha-II	GTATGGCTCTGATGCGATG
SAADF	CGTGATGAACAGGCTATTGC	saa	94°C, 30 s	52°C, 60 s	72°C, 40 s	119	16	3937/97c
SAADR	ATGGACATGCCTGTGGCAAC
E643f	TATCAGGCCAATCAAAACAG	efa-1d	94°C, 30 s	50°C, 60 s	72°C, 60 s	974	22	493/89
E1598r	AGACACTGGTAAATTTCGC
E5242f	TAAGCGAGCCCTGATAAGCA	efa-2d	94°C, 30 s	55°C, 60 s	72°C, 60 s	630	22	493/89
E5854r	CGTGTTGCTTGCCTTTGC
E7044f	TGTCTAACTGGATTGTATGGC	efa-3d	94°C, 30 s	56°C, 60 s	72°C, 60 s	685	22	493/89
E7710r	ATGTTGTTCCCGGCCCAGT
sfpA-U	AGCCAAGGCCAAGGGATTATTA	sfpA	94°C, 30 s	59°C, 60 s	72°C, 60 s	440	17	493/89
sfpA-L	TTAGCAACAGCAGTGAAGTCTC
HlyA1	GGTGCAGCAGAAAAAGTTGTAG	EHEC hlyA	94°C, 30 s	57°C, 60 s	72°C, 90 s	1,550	45	EDL933
HlyA4	TCTCGCCTGATAGTGTTTGGTA
esp-A	AAACAGCAGGCACTTGAACG	espP	94°C, 30 s	56°C, 60 s	72°C, 150 s	1,830	56	EDL933
esp-B	GGAGTCGTCAGTCAGTAGAT
D1	CGTCAGGAGGATGTTCAG	etpD	94°C, 30 s	56°C, 60 s	72°C, 70 s	1,062	56	EDL933
D13R	CGACTGCACCTGTTCCTGATTA
wkat-B	CTTCCTGTTCTGATTCTTCTGG	katP	94°C, 30 s	56°C, 60 s	72°C, 150 s	2,125	56	EDL933
wkat-F	AACTTATTTCTCGCATCATCC
Cdt I-f	TGGTGAGAATCGGAACTG	cdt-IA	94°C, 30 s	51°C, 60 s	72°C, 60 s	418	6	6468/62
Cdt I-r	CATTCCATCAGGTTTGTC
Cdt II-f	AATCCCTATCCCTGAACC	cdt-IIA	94°C, 30 s	52°C, 60 s	72°C, 60 s	542	6	9142/88
Cdt II-r	GTTCTATTGGCTGTGGTG
Cdt III-f	AAACAGGACGGTAATAATGACTAATA	cdt-III	94°C, 30 s	54°C, 60 s	72°C, 180 s	2,230	6	1404
Cdt III-r	GTGATCTCCTTCCATGAAAATATAGT	Complete
c338f	AGCATTAAATAAAAGCACGA	cdt-VAe	94°C, 30 s	52°C, 60 s	72°C, 60 s	1,329	23	493/89
c2135r	TACTTGCTGTGGTCTGCTAT
c1309f	AGCACCCGCAGTATCTTTGA	cdt-VBe	94°C, 30 s	52°C, 60 s	72°C, 60 s	1,363	23	493/89
c2166r	AGCCTCTTTTATCGTCTGGA
P105	GTCAACGAACATTAGATTAT	cdt-VCe	94°C, 30 s	49°C, 60 s	72°C, 60 s	748	23	493/89
c2767r	ATGGTCATGCTTTGTTATAT
RTsubAF	CGAATGTTTTTCTTGCTCCAG	subA	94°C, 30 s	53°C, 60 s	72°C, 60 s	220	38	3706/02f
RTsubAR	ACACTGCTGACAGGATGATAAG
espI-I	ATGGACAGAGTGGAGACAG	espI	94°C, 30 s	52°C, 60 s	72°C, 60 s	560	46	4797/97
espI-II	GCCACCTTTATTCTCACCA
Irp2 FP	AAGGATTCGCTGTTACCGGAC	irp2g	94°C, 30 s	60°C, 60 s	72°C, 60 s	280	25	5720/96
Irp2 RP	TCGTCGGGCAGCGTTTCTTCT
FyuA f	TGATTAACCCCGCGACGGGAA	fyuAg	94°C, 30 s	63°C, 60 s	72°C, 90 s	880	25	5720/96
FyuA r	CGCAGTAGGCACGATGTTGTA
UN19	CTGGGTGACATTATTGCTTGG	orfAh	94°C, 60 s	64°C, 60 s	72°C, 90 s	370	35	2787
UN20	TTTGCTTGTGCGGTAGACTG
UN21	TGAAAACATTAAGGGCTCG	orfBi	94°C, 60 s	64°C, 60 s	72°C, 90 s	450	35	2787
UN22	CCGGAAACATTGACCATACC
UN23	CAGTTTATCAATCAGCTCGGG	orfBj	94°C, 60 s	64°C, 60 s	72°C, 90 s	543	35	2787
UN24	CCACCGTTCCGTTATCCTC
fedA1	GTGAAAAGACTAGTGTTTATTTC	fedA	94°C, 60 s	56°C, 60 s	72°C, 60 s	230	35	2787
fedA2	CTTGTAAGTAACCGCGTAAGC
TerF1	TTACAATCCGGACAAAACA	terF	94°C, 30 s	55°C, 60 s	72°C, 60 s	244	51	EDL933
TerF2	CAATGACAACGGTGATCG
UreC-f	TCTAACGCCACAACCTGTAC	ureC	94°C, 60 s	60°C, 60 s	72°C, 60 s	398	32	EDL933
UreC-r	GAGGAAGGCAGAATATTGGG

^aAll PCRs included 30 cycles, followed by a final extension of 5 min at 72°C.

^bThe PCR positive control strains were the strains described in the references unless indicated otherwise.

^csaa⁺ E. coli O91:NM (16).

^dThree different regions of efa1 were targeted to detect the whole gene (9,996 bp) (22).

^eThe presence of three open reading frames encoding cytolethal distending toxin V was investigated.

^fsubA⁺ E. coli O113:H21 from our collection (H. Karch, unpublished).

^gMarkers for the HPI. The presence of additional HPI genes, their links, and the insertion site of HPI were determined previously (25).

^horfA encodes a 45-kDa protein which is required to modify AIDA-I to adhere to target cells (3).

ⁱThe primer amplifies a fragment from the coding region for AIDA-I (35).

^jThe primer amplifies a fragment from the coding region for AIDA^C (35).

PCR.

PCRs were performed with a Biometra TGradient 96 cycler (Biometra GmbH, Göttingen, Germany) (16, 48). The PCR primers, target sequences, conditions, and positive controls are shown in Table 1. E. coli C600 was used as a negative control. The specificity of PCR products was confirmed by analyzing the sequences of representative amplicons (6, 48).

Southern blot hybridization.

Southern blot hybridization of plasmid DNA with digoxigenin-labeled enterohemorrhagic E. coli (EHEC) hlyA, katP, espP, and etpD probes was performed as described previously (58).

Phenotypic methods.

Isolates were serotyped using antisera against E. coli O antigens 1 to 181 and H antigens 1 to 56 (42). Stx production was tested using a commercial latex agglutination assay (verotoxin-producing E. coli reverse passive latex agglutination; Denka Seiken Co., Ltd., Tokyo, Japan). Fermentation of sorbitol was detected on sorbitol MacConkey (SMAC) agar plates after overnight incubation. The enterohemolytic phenotype was investigated on enterohemolysin agar containing 5% defibrinated and washed sheep erythrocytes and 10 mM CaCl₂ (45). Resistance to tellurite was determined from the ability of isolates to grow on cefixime-tellurite (CT)-SMAC agar (Oxoid, Hampshire, United Kingdom) (7). Urease activity was examined in urea degradation broth (Heipha) after 24 h of incubation at 37°C (9, 17).

Cell cultures.

The T84 cell line (human colonic carcinoma epithelial cells; ATCC CCL-248) and the HCT-8 cell line (human ileocecal adenocarcinoma cells; ATCC CCL-244) were used. The culture medium for T84 cells contained a 1:1 mixture of Dulbecco's modified Eagle medium and Ham's F-12 medium (Cambrex Bioscience, Verviers, Belgium) supplemented with 10% (vol/vol) fetal calf serum (FCS) (Cambrex). HCT-8 cells were grown in RPMI 1640 (Cambrex) supplemented with 10% FCS, 2 mM l-glutamine, and 1 mM sodium pyruvate (Cambrex). The IPEC-J2 cell line (4) from jejunal epithelial cells of a neonatal piglet was maintained in a 1:1 mixture of Dulbecco's modified Eagle medium and Ham's F-12 medium (Cambrex) supplemented with 5% FCS. All cell cultures were grown at 37°C in 5% CO₂ until they reached confluence, and then they were subcultured using a 0.1% trypsin-EDTA solution (Cambrex).

Interaction of stx_2e-harboring E. coli with intestinal epithelial cells.

For the adherence assay, cells were grown on coverslips in six-well plates (Corning Inc., Corning, N.Y.) which were seeded with 1 × 10⁶ T84 or HCT-8 cells/well or 2.5 × 10⁵ IPEC-J2 cells/well; the plates were incubated at 37°C with 5% CO₂ until the cultures were semiconfluent. One hundred fifty microliters of a bacterial overnight culture in Luria-Bertani broth (8 × 10⁷ to 1 × 10⁸ CFU) was added to the cells and allowed to attach for 5 h. The cells were then thoroughly washed with phosphate-buffered saline (Cambrex), fixed with 70% ethanol, and stained with 10% Giemsa stain (Merck). The adherence assay was performed in parallel in the absence and presence of 0.5% (wt/vol) d-mannose (Roth, Karlsruhe, Germany) in the growth medium. For quantitative analysis, the numbers of bacteria attached to one cell were determined, and the results were scored as follows: ++++, >100 bacteria attached; +++, 50 to 100 bacteria attached; ++, 10 to 50 bacteria attached; +, 1 to 10 bacteria attached; −, no bacteria attached. Enteropathogenic E. coli strain 2348/69 (O127:H6) (33) and E. coli K-12 strain C600 were used as positive and negative controls, respectively. To ensure that the bacteria interacted specifically with the intestinal epithelial cells, the assays with all strains were also performed in wells without cells. To test the effects of culture supernatants on the intestinal epithelial cells, strains were grown with aeration (180 rpm) in Luria-Bertani broth overnight, the bacterial cells were removed by centrifugation (8,000 rpm, 15 min), and the supernatants were filter sterilized (pore size, 0.22 μm; Schleicher & Schuell GmbH, Dassel, Germany). The presence of Stx2e was verified by the latex agglutination assay as described above.

RESULTS

Frequency of stx_2e-containing E. coli in human stools.

A total of 747 STEC strains were isolated from 11,056 stools from patients with HUS or diarrhea or asymptomatic individuals. The 13 stx_2e-harboring STEC strains isolated during the period studied thus accounted for 1.7% of all STEC isolates and were in 0.12% of all stool samples investigated. All STEC strains harboring stx_2e were isolated from patients with mild diarrhea (n = 9) or from asymptomatic carriers (n = 4). None was associated with HUS.

Diagnostic characteristics of stx_2e-harboring strains.

Table 2 compares the serotypes of human STEC isolates containing stx_2e with those of stx_2e-positive STEC isolates from pigs. The human isolates belonged to none of the serotypes associated with edema disease in piglets, and more than one-half were nontypeable with antisera against currently known E. coli O antigens, suggesting that they might represent new serotypes. All but one human strain and all but three porcine strains produced Stx2e, as demonstrated by a commercial latex agglutination assay. All isolates fermented sorbitol on SMAC agar within 24 h, and none grew on CT-SMAC agar. This is consistent with the absence in all of the strains of terF (Table 2), which is used as a marker for the ter cluster that encodes tellurite resistance (7, 51). Similarly, in accordance with the absence of the EHEC hlyA gene in all 26 strains (Table 2), none displayed an enterohemolytic phenotype on enterohemolysin agar. None of the strains investigated possessed the ureC gene (Table 2), a marker for the ure gene cluster (18, 32), and accordingly none of them produced urease.

TABLE 2.

Distribution of putative virulence genes and other genes investigated among stx_2e-harboring E. coli strains isolated from humans and pigs

Gene	Predicted product or markera	Human isolates (n = 13)b		Porcine isolates (n = 13)c
		No. positive	No. negative	No. positive	No. negative
stx2e	Stx2e	13	0	13	0
eae	Intimin	0	13	0	13
iha	Iha	0	13	0	13
efa1	Efa1	0	13	0	13
saa	Saa	0	13	0	13
sfpA	SfpA	0	13	0	13
fedA	Major subunit of F18	0	13	13	0
orfA	AIDA-MP	0	13	13	0
orfB	AIDA	0	13	11	2
espI	Serine protease EspI	0	13	10	3
espP	Serine protease EspP	0	13	0	13
EHEC hlyA	EHEC hemolysin	0	13	0	13
etpD	Type II secretion system	0	13	0	13
katP	Catalase peroxidase	0	13	0	13
cdtd	CDT	0	13	0	13
subA	Subtilase cytotoxin A subunit	0	13	0	13
irp2	HPI	5	8	0	13
fyu A	HPI	5	8	0	13
terF	Tellurite resistance	0	13	0	13
ureC	Urease	0	13	0	13

^aIha, iron-regulated gene A homologue adhesin; Efa1, EHEC factor for adherence; Saa, Shiga toxin-producing E. coli autoagglutinating adhesin; SfpA, major pillin subunit of sorbitol-fermenting STEC O157 plasmid-encoded fimbriae (Sfp); AIDA-MP, AIDA modifying protein (3); F18, fimbrial adhesin; CDT, cytolethal distending toxin.

^bThe serotypes were O8:H10 (one strain), O8:H19 (one strain), O8:H- (one strain), O8:HNT (two strains), ONT:H10 (one strain), ONT:H19 (two strains), and ONT:H- (five strains).

^cThe serotypes were O138:K81 (one strain), O139:K12 (one strain), O139:K82 (eight strains), O141:K45 (one strain), O141:K85 (one strain), and O149:K91 (one strain).

^dThe presence of cdt-I, cdt-II, cdt-III, and cdt-V alleles (6) was investigated.

Distribution of virulence genes.

Genes encoding various adhesins, such as intimin, the iron-regulated gene A homologue adhesin (Iha) (49), EHEC factor for adherence (Efa1) (22, 34), STEC autoagglutinating adhesin (Saa) (37), and Sfp fimbriae (17), which are frequently found in STEC strains harboring stx₂ and the variants stx_2c and stx_2d (15, 16, 17, 22), were not found in any of the 26 E. coli strains harboring the stx_2e allele (Table 2). All of the porcine isolates, but none of the human isolates, possessed fedA, which encodes the major subunit of F18 fimbrial adhesin (35). Most of the porcine isolates (11 of 13) contained the orfB gene, which encodes AIDA (2) (Table 1). The orfA gene encoding a 45-kDa protein, which is required to modify AIDA so that it adheres to target cells (3), was present in all porcine isolates (Table 2). Ten of the 13 porcine isolates but none of the human isolates contained espI (Table 2), which is located on a pathogenicity island termed the locus of proteolysis activity which is inserted into selC of locus of enterocyte effacement-negative stx_2d-harboring STEC strains (46). espI encodes a novel serine protease (EspI) which cleaves swine pepsin A and human apolipoprotein A-I (46). In contrast, the espP gene encoding plasmid-encoded serine protease (EspP) in E. coli O157:H7, as well as the other plasmid-borne genes of STEC strains, such as EHEC hlyA, katP, and etpD (48, 58), were absent from all 26 strains investigated (Table 2). Similarly, various alleles encoding cytolethal distending toxin (cdt-I, cdt-II, cdt-III, and cdt-V), some of which were previously identified in a subset of eae-negative STEC strains from patients (6), were absent from all human and porcine stx_2e-harboring E. coli isolates (Table 2). Also, none of the strains investigated possessed the subA gene encoding the A subunit of the subtilase cytotoxin (38).

HPI is present in human STEC isolates that harbor stx_2e.

Five of the 13 human E. coli isolates that possessed stx_2e, but none of the corresponding porcine isolates, contained irp2 and fyuA (Table 2), which are components of an iron uptake-mediating gene cluster located on the high-pathogenicity island (HPI) (25). This prompted us to investigate whether a complete HPI is present in these five human isolates. Moreover, we compared HPIs of stx_2e-harboring E. coli strains with previously characterized HPIs in STEC isolates belonging to serogroups O26 and O128 and Yersinia pestis (25). To do this, all five stx_2e-harboring E. coli strains were subjected to 14 additional PCRs which target the other HPI genes or link consecutive genes (25). The results of the HPI analysis of these strains and a comparison of the HPIs of stx_2e-harboring STEC isolates with the HPIs of other STEC strains and Y. pestis are summarized in Table 3. Each of the five stx_2e-containing STEC isolates yielded amplicons that were of the same size as those detectable in the positive control STEC and Y. pestis strains in each of the PCRs targeting single HPI genes or links of the genes that constitute the siderophore yersiniabactin biosynthetic cluster (ybtS, ybtQ, ybtA, irp2, irp1, ybtU, ybtT, and ybtE) and the fyuA gene encoding the yersiniabactin receptor (Table 3) (PCRs IV to X and XII to VIII). Moreoever, similar to HPI of Y. pestis, but unlike HPIs of the STEC strains belonging to serogroups O26 and O128, the HPI in each of the five stx_2e-harboring E. coli strains contained the insertion element IS100 (PCR XI) (Table 3). The sizes of the amplicons elicited from the integrase gene (int) (PCR III) (Table 3) in four of the five stx_2e-containing STEC strains were identical to the size of the amplicon elicited from STEC O26 strain 5720/96 (Table 3), which was previously shown to possess a truncated int gene (25). In contrast, one remaining strain possessing stx_2e (24059/97) yielded an int amplicon that was the same size as the amplicons of STEC O128 strain 3172/87 and Y. pestis strain (Table 3), both of which contain an intact int gene (25).

TABLE 3.

PCR analysis of HPIs in stx_2e-harboring E. coli strains of human origin and comparison with HPIs of other STEC strains and Y. pestis

Strain or speciesa	Serogroup	Results of PCRs targeting HPI genes or their linksb
		PCR I, asnT-int(1)	PCR II, asnT-int(2)	PCR III, int	PCR IV, ybtS	PCR V, ybtQ	PCR VI, ybtA	PCR VII, irp2	PCR VIII, irp1	PCR IX, ybtE-fyuA	PCR X, fyuA	PCR XI, IS100	PCR XII, int-ybtS	PCR XIII, ybtS-ybtQ	PCR IV, ybtQ-ybtA	PCR XV, ybtA-irp2	PCR XVI, irp2-irp1	PCR XVII, irp1-ybtT	PCR XVIII, ybtT-fyuA
5720/96c	O26	900	1,100	900	160	800	230	280	240	360	780	−	800	2,800	2,800	1,300	300	1,700d	2,500
3172/97c	O128	1,200	1,400	1,200	160	800	230	280	240	360	780	−	800	2,800	2,800	1,300	300	1,700d	2,500
Y. pestisc	NAg	1,255	1,500	1,203	160	797	233	286	237	359	780	100	830	2,797	2,805	1,340	300	1,762	2,518
24059/97	ONT	+e	+e	+e	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+
3357/98	ONT	+f	+f	+f	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+
665/00	O8	+f	+f	+f	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+
E01/233	ONT	+f	+f	+f	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+
E02/25	ONT	+f	+f	+f	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+
E. coli C600	NAg	−	−	−	−	−	−	−	−	−	−	−	−	−	−	−	−	−	−

^aE. coli strains 5720/96 and 3172/97 and Y. pestis strain KIM6 were analyzed previously to determine their HPI structures (25) and were used as positive controls in this study; E. coli C600 was a negative control. The other five strains are human STEC isolates containing stx_2e.

^bThe numbering of the PCRs and the sequences targeted are the same as described by Karch et al. (25). int is the integrase gene; ybtS, ybtQ, ybtA, irp2, irp1, ybtE, and ybtT are components of the siderophore yersiniabactin biosynthetic gene cluster; fyuA encodes yersiniabactin receptor; and IS100 is an insertion element. PCR results: +, amplicon of the same size as that elicited from the positive control strains was obtained, unless indicated otherwise; −, no PCR product was obtained.

^cThe values for PCRs I to XVIII indicate the sizes (in bp) of amplicons obtained in the PCRs with the positive control strains. −, no PCR product was obtained. The sizes of the Y. pestis PCR products are the sizes in the previously published HPI sequence (25).

^dThe 1,700-bp amplicon indicates that the ybtU gene located between irp1 and ybtT is present and intact.

^e+, amplicons that were the same size as those from Y. pestis and E. coli strain 3172/97 were obtained.

^f+, amplicons that were the same size as those from E. coli strain 5720/96 were obtained.

^gNA, not available.

The integration site of HPI in stx_2e-harboring STEC strains was investigated using PCRs (25) linking the int gene of HPI with three different tRNA loci (asnT, asnU, and asnV). These sites are used by HPI to integrate into the chromosomes of pathogenic yersiniae (10). Amplicons that were 900 and 1,100 bp long, similar to those in STEC O26 strain 5720/96, were obtained from four strains in two different PCRs connecting asnT with the int gene (PCRs I and II) (Table 3). In these two PCRs, the remaining stx_2e-positive strain yielded amplicons that were 1,200 and 1,500 bp long and were similar to those in Y. pestis and STEC strain 3172/97 (Table 3). asnU-int and asnV-int PCRs were negative for all strains investigated (data not shown). These findings demonstrate that in STEC strains harboring stx_2e HPI is located near asnT, similar to the location in STEC O26 and O128 and Y. pestis (25). These PCR analyses also confirmed that four of the five strains harboring stx_2e (3357/98, 665/00, E01/233, and E02/25), like STEC O26 strain 5720/96, possess an HPI with a truncated int gene. In contrast, the remaining stx_2e-positive strain, 24059/97, like STEC O128 strain 3172/97 and Y. pestis, contains an HPI with an intact int gene (Table 3). Together, these data demonstrate that each of the five STEC human isolates harboring stx_2e possesses a complete HPI whose structure is closer to that of Y. pestis than to that of STEC strains belonging to serogroups O26 and O128.

Interaction of stx_2e-containing STEC with cultured intestinal epithelial cells.

The known STEC adhesins are absent from E. coli strains containing stx_2e (Table 2). Therefore, we investigated whether stx_2e-containing STEC can adhere to intestinal epithelial cells in vitro. As shown in Table 4, all human strains adhered with various intensities to human cell lines T84 and HCT-8. The presence of d-mannose in the culture medium did not inhibit the adherence of most of these strains; the only exception was strain 3096/00 (Table 4), which adhered more strongly to T84 cells in the absence than in the presence of 0.5% d-mannose. This suggests that an as-yet-unidentified adhesin(s), different from type 1 pili, plays a role in the adherence of human STEC strains harboring stx_2e to human intestinal epithelial cells. However, none of the human stx_2e-containing STEC strains adhered to pig intestinal epithelial cell line IPEC-J2 (Table 4).

TABLE 4.

Interaction of stx_2e-harboring STEC strains from humans and pigs with intestinal epithelial cells from homologous and heterologous hosts

Strain	Serotype	Source	Adherence to cell linea:
			T84	HCT-8	IPEC-J2
2771/97	ONT:H-	Human	+++	++	−
3054/97	O8:HNT	Human	++	+++	−
3583/97	ONT:H-	Human	+	+++	−
24059/97	ONT:H10	Human	++	+++	−
24066/97	ONT:H-	Human	+	+	−
26725/97	ONT:H-	Human	+	+++	−
3229/98	O8:H-	Human	+	++	−
3357/98	ONT:H19	Human	+	++	−
3615/99	O8:H10	Human	++	+++	−
665/00	O8:H19	Human	+++	+	−
3096/00	O8:HNT	Human	++	++	−
E01/233	ONT:H-	Human	+	++	−
E02/25	ONT:H19	Human	++	+++	−
S103G	O141:K45	Pig	CLb	CL	+
S105G	O139:K12	Pig	80%c	CL	+
S115G	O139:K82	Pig	CL	CL	++
S116G	O139:K82	Pig	CL	CL	++
S123G	O139:K82	Pig	CL	CL	++
S125G	O139:K82	Pig	CL	CL	++
S126G	O139:K82	Pig	CL	CL	−
S128G	O139:K82	Pig	90%c	CL	++++
S130G	O149:K91	Pig	90%c	CL	+
S131G	O139:K82	Pig	CL	CL	+
S132G	O139:K82	Pig	CL	CL	+++
S138G	O139:K82	Pig	CL	CL	+++
E57	O138	Pig	CL	CL	++
2348/69	O127:H6	Human EPEC (positive control)d	++	++++	−
C600	Not available	Laboratory strain (negative control)	−	+	−

^aAfter 5 h of incubation of bacteria with cells in the presence of 0.5% d-mannose, the adherence was quantified based on the number of bacteria attached to one cell, as follows: ++++, >100 bacteria; +++, 50 to 100 bacteria; ++, 10 to 50 bacteria; +, 1 to 10 bacteria; −, no bacteria attached.

^bCL, complete lysis of cells after 5 h of incubation with bacteria.

^cPercentage of cells which underwent lysis after 5 h of incubation with bacteria.

^dEPEC, enteropathogenic E. coli.

In contrast to human strains, which lysed none of the intestinal epithelial cell lines investigated, most pig isolates completely lysed human intestinal epithelial cells during a 5-h incubation (Table 4). However, with an equally long incubation period, the pig isolates did not lyse IPEC-J2 cells (Table 4), but 12 of the 13 strains adhered (Table 4). Representative patterns of the interaction of human and pig STEC strains with intestinal epithelial cells from homologous and heterologous hosts are shown in Fig. 1. No lysis of human intestinal epithelial cells was observed with sterile-filtered culture supernatants of the 26 strains, 22 of which contained Stx2e as determined by the latex agglutination assay.

FIG. 1.

Interaction of human and pig stx_2e-containing STEC isolates with intestinal epithelial cells from homologous and heterologous hosts. (A, D, and G) Control (untreated) cells of cell lines T84 (human), HCT-8 (human), and IPEC-J2 (pig), respectively. (B) Adherence of human strain 24059/97 to T84 cells. (E) Adherence of human strain 3583/97 to HCT-8 cells. (H) Lack of adherence of human strain E02/25 to IPEC-J2 cells. (C and F) Lysis of T84 and HCT-8 cells by pig strains S103G and S115G, respectively. (I) Adherence of pig strain S128G to IPEC-J2 cells. No adherence was observed with any of the strains tested in wells without intestinal epithelial cells, indicating that the adherence is cell dependent.

DISCUSSION

STEC ecology is complex and only partially understood. This is due to the marked heterogeneity of STEC strains. Epidemiological studies and molecular profiling indicate that most STEC infections in humans are food-borne and that the source of the pathogen is a nonhuman reservoir (26). Recent studies indicate that besides ruminants, swine also harbor STEC capable of causing human illness (11, 14, 43). A study conducted in the United States showed that 13% and 6% of STEC strains isolated from swine feces during a farm survey possessed the stx₁ and stx₂ genes, respectively (14). These genes are typically found in human STEC strains; however, the majority (80%) of these strains harbored the stx_2e variant (14) known to cause edema disease in weaned pigs (20). Although strains causing pig edema disease have been extensively characterized (1, 21, 35), to our knowledge this is the first detailed analysis of phenotypic and molecular characteristics of human stx_2e-containing STEC isolates. Both serotyping and molecular profiling demonstrated that Stx2e-producing STEC strains that cause human diseases are different from the strains that cause edema disease in pigs. Moreover, the two groups vary in their interactions with intestinal epithelial cells.

Most laboratories do not routinely screen for Stx2e-producing STEC in the way that they screen for E. coli O157:H7. The former strains would be overlooked on sorbitol MacConkey agar because all isolates investigated ferment sorbitol. They also do not grow on CT-SMAC agar, a medium routinely used to isolate E. coli O157:H7 (26, 50), and this is due to their tellurite sensitivity, as demonstrated in this study. Furthermore, EHEC hlyA, the structural gene encoding EHEC hemolysin (45), was not present in any of the stx_2e-harboring strains investigated. As a result, none of the strains showed the characteristic enterohemolytic phenotype. Although EHEC hemolysin production is a useful marker for the detection of STEC (9, 48, 53), it cannot be used to detect strains producing Stx2e. In addition, all Stx2e-producing STEC investigated lacked ureC, which we used as a marker for the ure operon (18, 24). It has recently been reported (32) that the presence of ureC distinguishes STEC strains belonging to the major serogroups associated with human diseases (O157, O26, O103, O111, and O145) from diarrheagenic E. coli belonging to other pathogroups (32). On the basis of this finding, ureC has been recommended as a target in the screening for such STEC strains (32). Our data, on the other hand, demonstrate that STEC strains harboring stx_2e would be missed by this screening procedure because of the absence of ureC. Thus, because of the poor repertoire of diagnostically useful phenotypic markers in stx_2e-positive STEC, the detection of the stx_2e gene with PCR, as used in this study, followed by colony blot hybridization, should be superior to the culture methods for screening primary stool cultures. The PCR approach enabled us to show that the frequency with which strains possessing the stx_2e allele occur in human stools is very low (0.12%) and that such strains can be present in stools of asymptomatic subjects. This raises a question about the etiological role of these strains in human diseases. In this study, we were unable to identify stx_2e-positive STEC strains in association with bloody diarrhea or HUS, although such strains have been isolated from an HUS patient by other workers (52). The absence of data on the anti-O157 lipopolysaccharide antibody response in the HUS patient of Thomas et al. (52), however, does not allow exclusion of a coinfection with E. coli O157:H7, which might have caused the HUS.

It is well established that STEC strains produce factors other than Stx that are potentially injurious to the human host (6, 8, 22, 23, 28, 34, 37, 38, 49). Intimin, the best-characterized STEC adhesin (33, 59), mediates attaching and effacing lesions in vitro and in animal models (33), but there are several other putative adherence factors, including Iha (49), Saa (37), and Efa1 (34), which also mediate adherence in vitro. However, all of these factors are absent from the stx_2e-containing strains investigated here. Notably, we observed different patterns when the strains interacted with intestinal epithelial cells. Whereas stx_2e-containing STEC strains from humans adhered to human epithelial cells, they did not adhere to pig intestinal epithelial cells. Although the molecular basis of this phenomenon is not known, differences in receptor-binding capacities of intestinal epithelial cells from different hosts are likely to be involved. Moreover, porcine strains were able to lyse human, but not porcine, intestinal epithelial cells. The cell lysis is not attributable to Stx2e because of its rapid occurrence (after 5 h) and because T84 cells do not express Gb3 and Gb4, the receptors for Stx2e (29, 53, 55). Moreover, sterile culture supernatants containing Stx2e displayed no visible lysis of the intestinal epithelial cells used in this study. The factors determining the interaction of Stx2e-producing STEC strains with intestinal epithelial cells from the homologous and heterologous species are not known, but it is likely that the differences in the interaction are due to differences in the molecular mechanisms involved. We have initiated transposon mutagenesis experiments to determine which genes are involved in the ability of porcine strains to cause the cell lysis.

It is noteworthy that the plasmid-encoded cytolysin EHEC hemolysin and cytolethal distending toxin, a potent toxin produced by a subset of STEC strains associated with human disease (6, 23, 39) and by Stx2f-producing STEC strains found in pigeons (31), are absent from STEC strains producing Stx2e. Furthermore, in addition to having no EHEC hlyA, all Stx2e-producing strains also lack other plasmid-borne genes, such as katP, espP, and etpD, which are usually present in STEC strains harboring stx₁ and stx₂ (9, 48, 58) and their variants (stx_1c, stx_2c, or stx_2d) (16, 48). Our finding that known putative virulence determinants of STEC strains pathogenic for humans are absent from Stx2e-producing human isolates extends a previous observation by our group (13). Although we found a close relatedness between one human and four porcine E. coli O101 strains by DNA fingerprinting, the virulence factors typically found in porcine STEC (i.e., heat-stable and heat-labile enterotoxins and F107 fimbriae) were absent from the human isolate (13). Moreover, this isolate also lacked virulence factors (eae and EHEC hemolysin) typical of STEC pathogenic for humans (13). This indicated that the pathogenicity of the human Stx2e-producing E. coli O101 strain might involve different mechanisms. Taken together, these data demonstrate that the mechanisms of pathogenicity of Stx2e-producing STEC strains associated with human diseases warrant further investigation.

Although swine are a potential reservoir of STEC strains that cause human illness (11, 14, 43), in an analysis of 11,056 stool samples from humans we were unable to detect the Stx2e-producing strains belonging to serogroups O138, O139, and O141 which are associated with edema disease in piglets (1, 35). Furthermore, a detailed characterization of the Stx2e-producing strains isolated from humans showed that they lack virulence factors, such as AIDA and F18 adhesins, that are frequently found in Stx2e-producing strains associated with pig edema disease (1, 21, 35). Therefore, it is unlikely that the Stx2e-producing STEC strains that cause pig edema disease are human pathogens. Moreover, only some of the serotypes identified among the Stx2e-producing STEC strains from humans in this study and in a study by Beutin et al. (O43:H30, O60:H4, O91:H21, Ont:H10, and Ont:H19) (5) have been found among the stx_2e-harboring STEC strains isolated from healthy pigs (14). These data indicate that there may be additional, as-yet-unknown reservoirs of STEC strains harboring stx_2e and that the extent to which these porcine strains play a role in the epidemiology of human infections needs further investigation. The majority of Stx2e-producing STEC strains isolated from humans in this study failed to agglutinate with O antisera currently available for serotyping of E. coli. Therefore, the development of diagnostic sera against such strains, which might represent novel O serogroups, would improve laboratory diagnosis of them and thus increase our understanding of their epidemiology.

The presence of the HPI of pathogenic yersiniae in a subset of the human stx_2e-containing STEC strains is noteworthy. This island confers virulence in highly pathogenic Yersinia species. HPI is also widely distributed among other Enterobacteriaceae (36), especially extraintestinal pathogenic E. coli strains that cause bacteremia and urosepsis in humans (47) and septicemia in poultry (12), and it contributes to the virulence of such strains (47). Recently, HPI was also found in certain serotypes of STEC pathogenic to humans, and it has been hypothesized that HPI can contribute to the fitness of such strains in diverse environments under iron limitation conditions (25). HPI contains a P4-like integrase (int) gene at the 5′ end and the fyuA gene encoding the receptor for yersiniabactin and pesticin at the 3′ end of the HPI core. A cluster of genes encoding the siderophore yersiniabactin is located between int and fyuA (25). Moreover, HPI in Y. pestis contains the insertion element IS100 upstream of fyuA (25). Our examination of the structure of HPI, identified in the five human stx_2e-containing STEC strains, showed that each of these strains contained a complete HPI structurally similar to HPI in Y. pestis. However, three of the five HPI-positive STEC strains in this study were isolated from patients with diarrhea, and two were isolated from asymptomatic carriers, making it impossible to speculate on the putative contribution of HPI to the pathogenicity of such strains.

In conclusion, Stx2e-producing E. coli strains, although having stx_2e in common, appear to have independently imported and exchanged virulence determinants. This has led to differences in the pathogenicity profiles and forms of disease, suggesting that there has been specific host adaptation.