Isolation and Genetic Characterization of a Coinfection of Non-O157 Shiga Toxin-Producing Escherichia coli

Abstract

A coinfection of O177:NM and O55:H7 Shiga toxin-producing Escherichia coli (STEC) was identified for a child with acute bloody diarrhea and hemolytic uremic syndrome by using culture and serotype-specific molecular reagents. The profile of O157-related genetic islands revealed that the O55:H7 isolate was highly similar to O157 STEC whereas the O177:NM isolate lacked several fimbrial O islands and non-locus-of-enterocyte-effacement effector determinants. However, both STEC serotypes are known to cause serious disease, and the significant repertoire of virulence determinants in both strains made it impossible to determine their individual contributions to the clinical symptoms.

In August 2006 in New Brunswick, Canada, a 4-year-old child was hospitalized with acute bloody diarrhea and hemolytic uremic syndrome (HUS). These are the characteristic clinical manifestations associated with Shiga toxin-producing Escherichia coli (STEC) (6), otherwise known as verocytotoxin-producing E. coli. In North America, strains of serogroup O157 that do not ferment sorbitol are the most frequently isolated STEC associated with HUS, partly because these strains can be distinguished from the benign E. coli flora present in clinical stool samples using differential media containing sorbitol. In contrast, most non-O157 STEC and some O157 strains do ferment sorbitol and therefore cannot be distinguished from the nonpathogenic fecal flora using this culture-based technique (6). Stool cultures collected from the hospitalized patient 3 to 5 days after onset of the first symptoms and 1 day after progression to bloody diarrhea did not yield O157 STEC, nor did they yield other diarrheal pathogens, such as Shigella or Campylobacter. Using recently developed molecular tools, clinical samples were screened for STEC, and both O177:NM and O55:H7 strains were ultimately recovered. Both of these serotypes have previously been associated with bloody diarrhea and HUS (5), and further genetic characterization was performed to attempt to identify the virulence potential of the strains individually to determine which strain(s) resulted in the clinical symptoms.

The presence of Shiga toxin in stool samples was first detected with the Premier enzyme immunoassay (Meridan Diagnostics, Inc.). To isolate STEC strains, stool samples were enriched by separate culturing in MacConkey broth or brain heart infusion broth and then plated onto Sorbitol-MacConkey (SMAC) or MacConkey (MAC) plates. From the SMAC plates, a sorbitol-nonfermenting (SNF) strain was isolated and characterized using Light Upon eXtension (LUX) real-time PCR targeting the toxin determinants stx₁ and stx₂, molecular O serogrouping using the gnd locus, and subtyping of the locus of enterocyte effacement (LEE) using allele-specific LUX primers for espZ-β, -γ1, -γ2, and -ɛ (2, 3). SNF STEC strains commonly belong to serogroup O157, and to rapidly screen for this serogroup, a LUX assay was developed to target the lpfA locus (strain EDL 933 coding sequence Z5225 of O-island 154 [OI-154]), previously observed exclusively in O157:H7 and O157:NM strains (12, 13). LUX oligonucleotides were 5′-GGCACCCGCACCAGTATAATC and 5′-cggctAACTCTAACGATGCTGCCG, with the penultimate nucleotide labeled with an Alexa-546 fluorophore and the lowercase nucleotides at the 5′ end being those required for LUX primer hairpin formation but not present in the target sequence. The SNF strain was identified as O177 through gnd-based molecular serogrouping and confirmed as O177:NM with traditional serotyping (Table 1). Accordingly, this isolate did not carry lpfA_OI-154 or the espZ-γ1 allele associated with O157 but instead carried the espZ-β1 allele.

TABLE 1.

Genetic profile of a sorbitol-negative Escherichia coli strain isolated on SMAC medium compared to that of genetic determinants amplified from an enrichment stool culture from the same patient^a

Sample	Serogroup (gnd allele analysis)	Serotype	Presence of gene			espZ allele	Presence of lpfA OI-154c
			stx1	stx2	fliCH7
SMAC isolate (06-5121)	O177	O177:NM	−	+	−	β1	−
Stool enrichment	NAb	NA	−	+	+	γ1	+

^aMolecular and serotyping tests are described in the text.

^bNA, not applicable.

^c−, real-time PCR amplification signal did not cross cycle threshold; +, real-time cycle threshold exceeded.

Template DNA prepared during stool enrichments was also characterized by real-time PCR (Table 1). Enrichments were cultured in BHI broth for 16 to 18 h at 37°C with rotation at 200 rpm. Bulk DNA was purified from 0.5 ml of broth culture by using a Puregene genomic DNA purification kit (Gentra Systems, Minneapolis, MN), and the same broth was used for cytotoxicity assays to detect Shiga toxin. The stool enrichment contained the stx₂ toxin determinant, and cytotoxicity could be neutralized in cell culture by Stx2 antisera. Additionally, traits characteristic of O157 STEC, including the espZ-γ1 allele, the H7 allele of fliC (14), and lpfA_OI-154, were detected in the stool enrichment, which conflicted with the observed genetic profile for the O177:NM strain isolated from the same patient (Table 1).

Because molecular characterization of the enriched stool preparation suggested the presence of a second pathogenic E. coli strain in the clinical stool sample, we endeavored to isolate additional STEC strains. Enrichment broths were plated on MAC medium, individual colonies of standard appearance on MAC plates were isolated, and boiled-cell templates were prepared from each and pooled into groups of 10. Each template pool was examined for stx determinants using real-time PCR, allowing for a rapid initial screen of a large number of isolates but without the requirement of performing a large multitude of PCR. From the pools that amplified stx, the individual templates were screened using the same real-time methods and four individual STEC strains were identified. These each carried stx₂, espZ-γ1, and lpfA_OI-154 (a profile identical to that of the master stool culture) (Table 1), and all four were subsequently serotyped as O55:H7.

The O177:NM and O55:H7 isolates were toxic to Vero cells in the cytotoxicity assay, and DNA cloning and sequencing of the stx₂ loci were performed. Both strains carried an stx₂ locus that encoded A and B toxin subunits identical to the Stx2d variant produced by O91:H21 strain B2F1 (GenBank accession number AF479829) and identical to the reference Stx2d3 variant produced by O157:H7 strain 7279 (accession number X61283), with the exception of a single valine-to-isoleucine substitution in the A subunit. DNA cloning and sequencing showed that the O177:NM strain carried a second stx₂ locus that encoded an Stx2c variant identical to that of O157:H7 strain FLY16 (accession number AB015057) except for a single threonine-to-alanine substitution in the A subunit.

The isolation of two non-O157 strains of STEC from the same patient provided an opportunity to conduct comparative genetic analyses of the two strains to ascertain which strain was most relevant to the disease symptoms. The genomic sequence is complete for two O157:H7 strains, and comparative genomic analyses with nonpathogenic strains identified O islands unique within these data sets to O157:H7 STEC (4, 9). This has allowed the identification and characterization of putative virulence determinants other than the prototypical LEE pathogenicity island. For example, the non-LEE effector nleB encoded in OI-122 was observed to be important in outbreak-associated STEC strains and was essential to induce virulence phenotypes with an animal model (16). We screened for multiple O islands using PCR targeting individual coding sequences, and the O55:H7 strain isolated in this study encoded all selected markers for the putative virulence determinants OI-36, OI-71, and OI-122 (with the exception of the nleA locus in OI-71), as well as the fimbrial islands OI-01, OI-47, OI-141, and OI-154 (Table 2). The O177:NM isolate encoded incomplete OI-36, OI-,71, and OI-122 regions and lacked all the screened fimbrial islands.

TABLE 2.

Characterization of O-island genetic content for non-O157 STEC strains isolated from the same patient with hemolytic uremic syndrome^a

Strain	Serotype	Presence of stx1	stx2 variant	Presence of eae	espZ allele	Presence of gene												Presence of fimbrial OIb
						OI-71			OI-122					OI-36				1	47	141	154
						nleA	nleF	nleH1	efa	pagC	sen	nleB	nleE	nleB2	nleC	nleH2	nleD
06-5231	O55:H7	−	stx2d	+	γ1	−	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+
06-5121	O177:NM	−	stx2d, stx2c	+	β1	+	+	−	−	−	+	+	+	−	−	−	+	−	−	−	−

^aThe methods utilized were previously described by Karmali et al. (7) and by B. K. Coombes and M. A. Karmali (unpublished).

^bFimbrial OI coding sequences screened using PCR and Southern hybridization: for OI-01, Z0024; for OI-47, Z1528, Z1530, Z1531, Z1533, Z1534, Z1535, Z1536, Z1537, Z1538, Z1539, Z1540/41, Z1542, Z1548, and Z1556; for OI-141, Z4965, Z4966, and Z4971; for OI-154, Z5220, Z5221, and Z5225 (lpfA).

The distribution and modular nature of OI-122 coding sequences among STEC serotypes was previously reported (7, 16), and an incomplete OI-122 was observed in all strains except for those of STEC serogroups O157, O121, and O111, including the frequent absence of efa (EDL 933 coding sequence Z4332). The O177:NM isolate described in this study lacked this determinant, as well as segments of other putative virulence O islands. In contrast, the genetic profile of the O55:H7 STEC strain isolated in this study was identical to that for O157:H7, since all screened OI determinants were present in the O55:H7 strain except for a single nle locus and both these serotypes encode the espZ-γ1 allele (3). Serotype O55:H7 is a close genetic relative of O157 STEC and has also been associated with severe human disease (15).

The precise contribution of the O islands to STEC virulence is not fully understood, and a complete OI-122 is not required for HUS (16). It is therefore not possible to accurately speculate on which of the two STEC serotypes contributed to the disease of this patient, if not both, since both O55:H7 and O177:NM have previously been associated with bloody diarrhea and HUS (5). Additionally, the Stx2d variant encoded by both the O55:H7 and O177:NM strains and the Stx2c variant encoded by the O177:NM strain have both been associated with HUS (10). If STEC pathogenicity for humans could be measured as the sum of virulence determinants, then the O55:H7 isolate encoded the larger arsenal. However, the O177:NM strain also encoded stx₂, the LEE pathogenicity island, and significant components of OI-122.

The clinical significance associated with non-O157 STEC is likely understated due to the difficulties associated with microbial isolation. In this study, we implemented several recently developed molecular methods to comprehensively screen and isolate two non-O157 serotypes from clinical stool samples. To our knowledge, this is the first observation of more than one non-O157 STEC serotype isolated concurrently from a single patient. Simultaneous infections of O157 and non-O157 STEC have previously been observed (11), and serological assays have identified mixed STEC infections through immunological responses to O157 and other STEC serogroups (8). Additionally, during an outbreak of O111:NM STEC wherein a HUS patient had antibodies to O111, an O145 STEC strain was isolated from clinical samples, suggesting that both serogroups were involved in the infection (1). Although the O111 strain was not isolated from this patient, the outbreak-associated etiology of this strain and the positive serological diagnosis provided evidence that the O111 STEC strain was the significant pathogen. Otherwise, the routine identification of all STEC serogroups is not currently conducted in the majority of public health microbiology laboratories, but the methods utilized in this study allowed for enhanced detection capabilities beyond serogroup O157 and indicated the possibility of isolating multiple serotypes. Further comparative genetic analyses of the virulence gene dosing effect and comprehensive testing of clinical samples will be required to determine the importance and frequency of non-O157 STEC infections in regard to their potential as human pathogens.

Nucleotide sequence accession numbers.

DNA sequences of the O55:H7 stx_2d and O177:NM stx_2c loci were deposited in GenBank under accession numbers EF584538 and EU086525, respectively.