Molecular Characterization Reveals Similar Virulence Gene Content in Unrelated Clonal Groups of Escherichia coli of Serogroup O174 (OX3)

Abstract

Most severe illnesses that are attributed to Shiga toxin-producing Escherichia coli are caused by isolates that also carry a pathogenicity island called the locus of enterocyte effacement (LEE). However, many cases of severe disease are associated with LEE-negative strains. We characterized the virulence gene content and the evolutionary relationships of Escherichia coli isolates of serogroup O174 (formerly OX3), strains of which have been implicated in cases of hemorrhagic colitis and hemolytic uremic syndrome. A total of 56 isolates from humans, farm animals, and food were subjected to multilocus virulence gene profiling (MVGP), and a subset of 16 isolates was subjected to multilocus sequence analysis (MLSA). The MLSA revealed that the O174 isolates fall into four separate evolutionary clusters within the E. coli phylogeny and are related to a diverse array of clonal groups, including enteropathogenic E. coli 2 (EPEC 2), enterohemorrhagic E. coli 2 (EHEC 2), and EHEC-O121. Of the 15 genes that we surveyed with MVGP, only 6 are common in the O174 strains. The different clonal groups within the O174 serogroup appear to have independently acquired and maintained similar sets of genes that include the Shiga toxins (stx₁ and stx₂) and two adhesins (saa and iha). The absence of certain O island (OI) genes, such as those found on OI-122, is consistent with the notion that certain pathogenicity islands act cooperatively with the LEE island.

Shiga toxin (Stx)-producing Escherichia coli (STEC) strains are food-borne bacterial pathogens that can cause mild to severe gastrointestinal disease in humans. STEC is defined by the elaboration of Stx, a bacteriophage-encoded cytotoxin. In addition, many STEC strains harbor the locus of enterocyte effacement (LEE), a pathogenicity island that encodes the adhesin intimin (encoded by the eae gene). Pathogens that harbor the LEE island elicit a particular pathology called attaching-and-effacing lesions (known as A/E pathogens). The subset of STEC strains that are also A/E pathogens is often called enterohemorrhagic E. coli (EHEC), and these strains appear to account for most cases of severe disease (12, 16). However, many sporadic cases and small outbreaks of STEC disease, including hemolytic uremic syndrome (HUS), have been caused by LEE-negative strains (15, 24). The adherence mechanism of STEC strains that lack eae is still unclear.

While investigating the adherence mechanism of a LEE-negative strain of serotype O113:H21, Paton et al. (23) uncovered a novel adhesin (the STEC autoagglutinating adhesin [Saa]). The gene encoding Saa (saa) has since been detected in at least 10 other LEE-negative serotypes; however, it is not clear whether saa is a virulence determinant (12). We have begun to investigate the gene content of LEE-negative strains and the distribution of saa; we chose to examine strains of serogroup O174 (formerly OX3 [29]), which are LEE-negative STEC strains that are routinely isolated in moderate numbers from human clinical samples, food, and farm animals. Serogroup O174 isolates accounted for 3.7 to 5.2% of the STEC strains isolated from ruminants in three studies in Europe (2, 4, 5) and were the predominant STEC serogroup isolated from cattle and food in France (26). Serogroup O174 is less commonly associated with disease in humans, accounting for about 1 to 2% of the STEC isolates from patients in Germany (3) and Denmark (8). Strains of serogroup O174 have been associated with cases of HUS (6, 15, 30), bacteremia (20), and sudden infant death syndrome (22).

Little is known about the evolutionary relationships among the strains of serogroup O174 and between these strains and other pathogenic clones of E. coli; nor is much known about the virulence mechanisms. In addition to saa and the well-characterized virulence determinants eae and stx, we were curious about the presence or absence of certain “O island” (OI) genes, those that were identified on pathogenicity islands in the genomes of E. coli O157:H7 (10, 25), and their potential contribution to STEC-induced disease. We have applied a multilocus virulence gene profiling assay (MVGP) (31) that detects 15 genes found in pathogenic E. coli. In addition to the genes discussed above, the genes in the MVGP include plasmid-borne genes (bfpA, ehxA, and toxB), two genes that are homologous to genes on the high-pathogenicity island (HPI) of Yersinia spp., as well as genes on OI-122, OI-48, and OI-115. The distribution of virulence genes was then examined within a phylogenetic framework that was constructed through multilocus sequence analysis (MLSA). We report here the gene content and the evolutionary relationships of O174 strains to each other and to other pathogenic clones of E. coli.

MATERIALS AND METHODS

A total of 56 strains of serogroup O174 were isolated from clinical samples and food samples and from animal feces (Table 1). Isolates were stored in 15% glycerol-LB medium and frozen at −70°C until plated on LB agar. The H type of nonmotile strains was determined by amplification of the fliC gene, following the protocol described by Reid et al. (28), and by subsequent restriction fragment analysis of the amplicon. There were a total of seven fragment profiles among our samples that were identical to profiles of known serotypes and one strain for which the H antigen could not be classified by this approach.

TABLE 1.

E. coli isolates of serogroup O174 and corresponding molecular virulence gene profiles

Accession no.	Serotypea	Locality	Host	Clinical sourceb	Virulence gene detectedc
					saa	ehx	toxB	stx1	stx2	Z4326	invG	spaP	iha	ureA	terC	eae	irp2	fyuA
TW09346	O174:H8	Denmark	Human	HC	+	+		+					+	+	+
TW07727	O174:H8	Germany	Sheep		+	+		+	+				+
TW07728	O174:H8	Germany	Sheep		+	+		+	+				+
TW07729	O174:H8	Germany	Sheep		+	+		+	+				+
TW07730	O174:H8	Germany	Sheep		+	+		+	+				+
TW07732	O174:H8	Germany	Sheep		+			+	+				+
TW07736	O174:H8	Germany	Sheep		+			+	+				+
TW07739	O174:H8	Germany	Sheep		+			+	+				+
TW07746	O174:H8	Germany	Sheep		+			+	+				+
TW07747	O174:H8	Germany	Sheep		+			+	+				+
TW07748	O174:H8	Germany	Sheep		+			+	+				+
TW07749	O174:H8	Germany	Sheep		+			+	+				+
TW08572	O174:[H8]	Germany	Human	D	+			+	+				+
TW08581	O174:[H8]	Norway	Sheep		+			+	+				+
TW09335	O174:H8	Denmark	Caprine		+			+	+				+
TW09336	O174:H8	Denmark	Caprine		+			+	+				+
TW09337	O174:H8	Denmark	Caprine		+			+	+				+
TW09341	O174:[H8]	Denmark	Human		+			+	+				+
TW09342	O174:[H8]	Denmark	Human	D	+			+	+				+
TW09345	O174:H8	Denmark	Human	D	+			+	+				+
TW09348	O174:H8	Denmark	Ovine		+	+		+	+				+
TW08580	O174:H8	Germany	Human	D	+			+					+
TW08582	O174:[H8]	Norway	Sheep		+			+					+
TW09344	O174:H8	Denmark	Human	D	+			+					+
TW07856	O174:H21	Australia	Human	SIDS					+			+	+					+
TW05188	O174:[H21]	Canada	Cow		+	+			+			+	+
TW05257	O174:H21	Canada	Cow		+	+			+			+	+		+
TW08575	O174:H21	Germany	Human	D	+	+			+			+	+
TW09347	O174:H21	Denmark	Human	HC	+	+			+			+	+
TW00120	O174:H21	United States	Food						+			+	+
TW00121	O174:H21	United States	Food						+			+	+
TW05179	O174:[H21]	Canada	Cow						+			+	+
TW05180	O174:[H21]	Canada	Cow						+			+	+
TW05181	O174:H21	Canada	Cow						+			+	+
TW05242	O174:[H21]	Canada	Cow						+			+	+
TW05246	O174:H21	Canada	Cow						+			+	+
TW05277	O174:[H21]	Canada	Cow						+			+	+
TW07946	O174:[H21]	United States	Human						+			+	+
TW08583	O174:H21	Germany	Human	D					+			+	+
TW08584	O174:[H21]	Germany	Beef						+			+	+
TW08866	O174:H21	United States	Human						+			+	+
TW05183	O174:[H21]	Canada	Cow									+	+
TW05189	O174:[H21]	Canada	Cow									+	+
TW08573	O174:H2	Germany	Human		+	+			+			+
TW08574	O174:H2	Germany	Human	HC	+	+		+	+			+			+
TW08576	O174:H2	Germany	Human	HC	+	+		+	+			+
TW08579	O174:H2	Germany	Cattle		+	+		+	+			+
TW09350	O174:H2	Denmark	Ovine		+	+		+	+			+
TW05252	O174:H16	Canada	Cow						+			+	+
TW08578	O174:H16	Germany	Human	A								+
TW08585	O174:H19	Uganda	Cow		+	+		+	+			+	+
TW05199	O174:[H19]	Canada	Cow									+
TW08577	O174:H28	Germany	Food		+	+			+			+
TW08867	O174:H28	United States	Human						+			+	+
TW05276	O174:[H25]	Canada	Cow									+
TW05247	O174:UT	Canada	Cow									+

^aH antigens listed in square brackets were determined by PCR amplification and restriction fragment length polymorphism analysis of the fliC gene.

^bClinical information is provided where known. Abbreviations: D, diarrhea; HC, hemorrhagic colitis; SIDS, sudden infant death syndrome; A, asymptomatic; HUS, hemolytic uremic syndrome.

^cThe presence of a gene is indicated by a + sign.

A protocol described in detail by Tarr et al. (31) was used to prepare the template for genetic analysis and to conduct the MVGP assay. A total of three single colonies were analyzed for every isolate. The MVGP assay determines the presence or absence of 15 genes by PCR and includes the well-characterized virulence genes in STEC (stx₁, stx₂, and eae), 4 plasmid-borne genes (saa, ehxA, toxB, and bfpA) (18, 21, 33, 35), and 8 genes found on chromosomal islands of E. coli (10, 25). This last group of genes includes the OI-115 spaP and invG genes (which are homologous to genes in the Inv-Spa system of Salmonella spp.), a homologue of the Shigella spp. enterotoxin 2 gene (Z4326), two genes found in E. coli O26:H11 strains that are homologous to irp2 and fyu2 on the HPI of Yersinia spp. (13), and three genes (iha, terC, and ureA) from the tellurite resistance-and-adherence-conferring island (TAI) (32).

We chose 16 isolates of serogroup O174 to represent six different serotypes for the MLSA. Multiple strains for a serotype were chosen so that each differed by host, fragment profile, and/or motility. Approximately 500 bp of each of six housekeeping genes (clpX, fadD, icdA, lysP, mdh, and uidA) was sequenced for each isolate. The MLSA protocol can be downloaded from http://www.shigatox.net/stec/mlst-new/index.html. The sequences were aligned with 62 other E. coli/Shigella sequences and with two Escherichia albertii isolates that served as an outgroup (11) by using Clustal X (34). Information about the other strains used with the MLSA can be found at http://www.shigatox.net/cgi-bin/stec/index. MEGA software (17) was used to construct a gene tree from nucleotide sequences following the procedures described by Tarr et al. (31). Strains were partitioned into clonal groups (CGs) based on a composite analysis of sequences, using the BURST algorithm (www.MLST.net) (9) and bootstrap analysis. The BURST analysis was conducted with a set of 217 strains; the result for a subset of 78 E. coli/Shigella strains is shown here.

RESULTS

MLSA.

We sequenced segments of six housekeeping genes for a total of 3,240 base pairs of nucleotide sequence for 78 E. coli/Shigella strains and 2,652 bp for E. albertii (which lacks the uidA gene). There were a total of 249 polymorphic nucleotide sites in our sample of E. coli. The O174 isolates fell into four clusters in the neighbor-joining tree, and the BURST analysis identified four CGs within the E. coli phylogeny (Fig. 1).

FIG. 1.

Gene phylogeny inferred by a neighbor-joining algorithm, from six housekeeping gene sequences. The number at each branching point indicates the percentage of trees out of 1,000 bootstrap replications that support a particular node. CGs for the O174 strains, as determined by BURST and bootstrapping, are delineated by boxes. The scale at the bottom shows the branch length that corresponds to 2 nucleotide substitutions per 100 nucleotide sites.

The O174:H21 strains formed CG 34, which also contained an O91:H21 strain. The strains most closely related to this group were two nonmotile O174 isolates that were typed as O174:[H19] and O174:[H25] (the H antigens listed in square brackets were determined by PCR amplification and restriction fragment length polymorphism analysis of the fliC gene). All the strains were contained within a cluster that also included enteropathogenic E. coli 2 (EPEC 2) strains, an additional O174:H19 strain, and an O104:H21 isolate. The O174:H8 isolates formed CG 19, which was closely related to CG 30, a cluster of two isolates that contained an O113:H21 strain and a nonmotile O174:[H19] strain. These two clonal groups, CG 19 and CG 30, fell within a cluster that also contained two enterohemorrhagic E. coli (EHEC) O121:H19 strains and two Shigella strains. The O174:H2 isolates formed a cluster (CG 41) that was most closely related to strains of the EHEC 2 pathotype. The O174:H16 and O174:H28 isolates formed a group (CG 21) that was not closely related to other CGs.

MVGP.

The O174 isolates collectively carried 9 of the 15 genes that we included in the assay, and 6 of the genes were common among the O174 isolates (Table 1). With the exception of saa, all genes detected in O174 strains are also found in EHEC strains (Table 2). All O174 isolates were eae negative. Of the 56 strains tested, 50 (88%) carried one or both types of the Stx gene. The most common loci were iha (82%), stx₂ (81%), and saa (61%). All but four strains (7%) had either saa or iha or both, and the adhesins were each found in five serotypes.

TABLE 2.

Unique serotype x profile combinations for E. coli isolates

Serotypea	Accession no.	No. of isolates	No. of human isolates	Virulence gene detectedb
				bfpA	saa	ehx	toxB	stx1	stx2	Z4326	invG	spaP	iha	ureA	terC	eae	irp2	fyuA
O174:H21	TW07856	1	1						+			+	+					+
O174:H21	TW05257	1	0		+	+			+			+	+		+
O174:H21	TW08575	3	2		+	+			+			+	+
O174:H21	TW08584	12	2						+			+	+
O174:H21	TW05189	2	0									+	+
O174:H8	TW09346	1	1		+	+		+					+	+	+
O174:H8	TW09348	5	0		+	+		+	+				+
O174:H8	TW08581	15	4		+			+	+				+
O174:H8	TW08582	3	2		+			+					+
O174:H2	TW08574	1	1		+	+		+	+			+			+
O174:H2	TW08576	3	1		+	+		+	+			+
O174:H2	TW08573	1	1		+	+			+			+
O174:H16	TW05252	1	0						+			+	+
O174:H16	TW08578	1	1									+
O174:H19	TW08585	1	0		+	+		+	+			+	+
O174:[H19]	TW05199	1	0									+
O174:[H25]	TW52760	1	0									+
O174:H28	TW08577	1	0		+	+			+			+
O174:H28	TW08867	1	1						+			+	+
O113:H21	TW01391	1	1		+	+		+	+			+	+
O91:H21	TW01393	1	1		+	+			+			+	+
O121:H19	TW08023	13	13			+	+		+	+				+	+	+
O111:H8	TW07926	2	2			+		+	+	+		+	+	+	+	+
O26:H11	TW07936	1	1			+	+	+		+		+	+	+	+	+	+	+
O157:H7	TW08264	3	3			+	+	+	+	+	+	+	+	+	+	+
O111:H2	TW00185	2	2	+						+		+				+
O142:H6	TW03118	1	1	+						+						+
O55:H6	TW00588	2	2	+						+						+

^aH antigens listed in square brackets were determined by PCR amplification and restriction fragment length polymorphism analysis of the fliC gene.

^bThe presence of a gene is indicated by a + sign.

There were four profiles for serotype O174:H8, which were characterized by the presence of saa, stx₁, and iha; these genes occurred in all 24 of the isolates that we examined. A total of 20 isolates (83%) also carried a copy of the stx₂ gene, and 6 strains were positive for ehxA. Interestingly, one isolate that was positive for ehxA also carried both terC and ureA but was negative for stx₂; this profile occurred only once in our sample, and the isolate was implicated in a case of hemorrhagic colitis.

Isolates of serotype O174:H21 were positive for spaP and iha; the majority of isolates also carried a copy of the stx₂ gene (89%), while a minority harbored copies of the saa and ehxA (21%) genes. A single isolate was positive for terC, and one was positive for fyuA. A total of three profiles were seen with O174:H2, and these were characterized by saa, ehx, stx₂, and spaP. The stx₁ gene was present in 4 of 5 isolates, and terC was present in one isolate. Our representation of other serotypes was too small to characterize the group.

DISCUSSION

A key finding of this study is that the O174 strains are similar to each other in their gene content, even though the isolates fall into distantly related CGs. Thus, the particular set of genes that characterizes the O174 serogroup has been acquired and maintained in multiple, independent lineages of E. coli. Previous studies have shown evidence of parallel evolution for other CGs of E. coli. Reid et al. (27) found that each EHEC and EPEC pathotype had arisen twice, and Tarr et al. (31) found that strains of serotype O121:H19 also represented the independent emergence of a third EHEC clone. Studies such as these that examine the gene content of serotypes within a phylogenetic context can identify additional genetic elements that may be associated with particular E. coli pathotypes and that may be functionally coordinated.

Recent evidence suggests that OI-122 is associated with the LEE island, as the two islands often cooccur in E. coli strains (19), and genes on the two islands may even act cooperatively (7). We screened for one gene on OI-122 (Z4326) that encodes a homologue of the Shigella enterotoxin 2. The single gene appears to be a good marker for a complete or nearly complete OI-122, as it is present in all STEC strains that have two or more of the four genes that were screened by Karmali et al. (14). We found no evidence of O-122 among the O174 isolates, all of which are also eae negative. Interestingly, we find the eae-Z4326 association not only in the STEC strains that we assayed but also in EPEC 1 and EPEC 2 strains as well (Table 2); our results suggest that OI-122 is associated not with STEC pathogens as suggested by Karmali et al. (14) but with A/E pathogens in general, including some EPEC isolates (however, not all EPEC isolates contain OI-122; see the article by Afset et al. [1]).

Another island of interest in STEC pathogenesis is the TAI, which includes iha and operons for urease activity (the ure genes) and tellurite resistance (ter) genes (32). Although iha appears to be broadly distributed in E. coli, our preliminary data suggest that the island carrying the ter and ure genes is found primarily in EHEC pathogens. The presence of terC in three O174 serotypes suggests that the island has been acquired, but its rarity suggests that it has not been maintained in the serogroup and is not an important component of this STEC pathotype. It is unclear whether iha was acquired by these O174 serotypes as part of the TAI, as it has also been found on the pO113 plasmid (23).

The genes that underlie the virulence of LEE-negative STEC, particularly the genes involved in colonization of the bowel, remain largely unknown. The eae and saa genes appear to have a mutually exclusive distribution throughout the STEC strains, suggesting that Saa can serve as an alternative adhesin in STEC pathogenesis. The Saa protein may be involved in colonization, since mutating saa or curing the large plasmid in O113:H21 resulted in reduced adherence (but did not abolish it [23]). The role of iha in LEE-negative STEC adherence bears further study as well. Unfortunately, studies of STEC often lump all LEE-negative strains into one category and do not screen for saa, so it is unclear whether saa is a risk factor for severe diarrheal disease. Virulent gene screening of STEC strains should include other adhesins, especially saa, so that we can better assess the extent to which the genes are risk factors for hemorrhagic colitis and HUS.