Recycling of Shiga Toxin 2 Genes in Sorbitol-Fermenting Enterohemorrhagic Escherichia coli O157:NM

Abstract

Using colony blot hybridization with stx₂ and eae probes and agglutination in anti-O157 lipopolysaccharide serum, we isolated stx₂-positive and eae-positive sorbitol-fermenting (SF) enterohemorrhagic Escherichia coli (EHEC) O157:NM (nonmotile) strains from initial stool specimens and stx-negative and eae-positive SF E. coli O157:NM strains from follow-up specimens (collected 3 to 8 days later) from three children. The stx-negative isolates from each patient shared with the corresponding stx₂-positive isolates fliC_H7, non-stx virulence traits, and multilocus sequence types, which indicates that they arose from the stx₂-positive strains by loss of stx₂ during infection. Analysis of the integrity of the yecE gene, a possible stx phage integration site in EHEC O157, in the consecutive stx₂-positive and stx-negative isolates demonstrated that yecE was occupied in stx₂-positive but intact in stx-negative strains. It was possible to infect and lysogenize the stx-negative E. coli O157 strains in vitro using an stx₂-harboring bacteriophage from one of the SF EHEC O157:NM isolates. The acquisition of the stx₂-containing phage resulted in the occupation of yecE and production of biologically active Shiga toxin 2. We conclude that the yecE gene in SF E. coli O157:NM is a hot spot for excision and integration of Shiga toxin 2-encoding bacteriophages. SF EHEC O157:NM strains and their stx-negative derivatives thus represent a highly dynamic system that can convert in both directions by the loss and gain of stx₂-harboring phages. The ability to recycle stx₂, a critical virulence trait, makes SF E. coli O157:NM strains ephemeral EHEC that can exist as stx-negative variants during certain phases of their life cycle.

Sorbitol-fermenting (SF) enterohemorrhagic Escherichia coli (EHEC) O157:NM strains (nonmotile) were first identified in 1988 in Germany (28) and have now emerged as a cause of human diseases, including life-threatening hemolytic-uremic syndrome (HUS), in additional countries (4, 7, 8, 17, 20, 22, 37). In Germany, SF EHEC O157:NM strains account for >10% of sporadic cases of HUS (20, 22), and after EHEC O157:H7, they are the second most common cause of HUS (20, 22). They have also caused several outbreaks in Germany (5, 41, 42) and the United Kingdom (3, 18).

The genes encoding Shiga toxins (Stx) (44), the major virulence factors of EHEC, are located in temperate lambdoid bacteriophages which are integrated in the host genome during lysogenic growth (2, 25, 29, 35, 48, 53). A lytic phage cycle can be induced in EHEC strains using sublethal doses of UV light (35) or subtherapeutic concentrations of various antibiotics (23, 33, 35, 45, 47, 58). The existence of stx genes in a multitude of E. coli serotypes (2, 55) is attributable to transduction with stx-converting phages (2, 25, 45). Transduction in vivo has been demonstrated in several animal models including mice (1), sheep (14), and insects (39).

Both loss and transfer of the stx gene appear to occur during human infection and can lead to a change in the pathotype of the infecting strain (10, 21, 34). We have demonstrated the excretion of stx₂-positive EHEC O157:NM in the initial stool specimen (collected 8 days after the onset of prodromal diarrhea) of an HUS patient, followed by the excretion of stx-negative E. coli O157:NM in the follow-up stool specimen collected 3 days later (34). Comparison of stx gene losses in SF EHEC O157:NM and non-SF EHEC O157:H7 isolates showed a significantly higher proportion of stx-negative strains (12.7% versus 0.8%) among SF E. coli O157:NM (21). This loss of stx genes has important diagnostic implications if laboratories rely on toxin detection to find these organisms late in illness (34). Furthermore, the loss of stx complicates epidemiological investigations (11) and might even influence the outcome of the disease (21).

One of integration sites for stx₂-harboring bacteriophages in the chromosome of SF EHEC O157:NM is the yecE gene (11). In this study of the dynamic processes in the evolution of the genome in SF EHEC O157:NM, we compared the integrity of the phage integration site in stx₂-positive SF EHEC O157:NM strains and their stx-negative derivatives and investigated the ability of stx₂ phages from SF EHEC O157 patients’ isolates to lysogenize stx-negative variants of SF EHEC O157:NM.

MATERIALS AND METHODS

Bacterial strains.

Three SF EHEC O157:NM strains and their corresponding stx-negative derivatives were isolated from stool specimens collected 8 to 11 days after the onset of diarrhea and in subsequent stool specimens from the same patients collected 3 to 8 days later, respectively. All patients had HUS. Pair A1-A2 has been partially described previously (34). Pairs B1-B2 and C1-C2 are from a 15-month-old boy and a 20-month-old girl, respectively. In each pair, isolate 1 is stx₂ positive and isolate 2 is stx negative. The stx₂-positive SF EHEC O157:NM strain from the initial stool specimen from patient B (B1; 258/98) has been described previously (11). Strains were isolated as described previously (20, 21, 34). Briefly, stool specimens were inoculated into Hajna broth and grown, and the broth was then enriched for E. coli O157 using immunomagnetic separation with Dynabeads coated with anti-E. coli O157 antibody (Dynal, Oslo, Norway). Beads with attached E. coli O157 were cultured on sorbitol-MacConkey (SMAC) agar (Oxoid, Basingstoke, United Kingdom). Overnight cultures harvested into saline were screened for stx₁, stx₂, rfb_O157, and eae by PCR (19, 34). SF E. coli O157 was isolated from PCR-positive cultures using colony blot hybridization with stx₂ and eae probes and slide agglutination with antiserum specific for the E. coli O157 lipopolysaccharide (21, 34).

PCR.

PCR analyses were performed as described previously (11, 48). stx₂, rfb_O157, and sfpA (a specific marker for SF E. coli O157:NM) (19) were detected using primer pairs LP43 and LP44, O157-F and O157-R, and sfpA-U and sfpA-L, respectively (19, 20); stx₂ genes were subtyped using PCR with primers GK3 and GK4 and restriction of the amplicon with HaeIII (20). Genes encoding other toxins (cdt-V and EHEC-hlyA) and adhesins (eae, efa1, lpf_O157-OI141, and lpf_O157-OI154), other plasmid-carried genes (katP, espP, and etpD), and terE (used as a marker for the gene cluster encoding tellurite resistance) (24, 38) were detected using established protocols (9, 12, 26, 27, 49, 51). The eae genes were subtyped (56), and the genotype of the fliC gene, encoding the structural subunit of the H antigen, was determined as described previously (49, 57). The intact or occupied status of yecE was tested using primers EC10 and EC11 (16). The integration of an stx₂-carrying bacteriophage homologous to phage Φ258₃₂₀ (originating from SF EHEC O157:NM strain 258/98) (11) in yecE in wild-type EHEC strains and lysogens was identified by PCR with primers Int-258₃₂₀ and EC11, which connect the integrase gene of phage Φ258₃₂₀ with yecE (11).

Multilocus sequence typing.

Internal fragments of seven housekeeping genes (adk, fumC, gyrB, icd, mdh, purA, and recA) were analyzed as described previously (54), except for use of a newly designed forward primer for icd (5′-CGATTATCCCTTACATTGAAG-3′), which yielded more reliable sequencing results. Alleles and sequence types were assigned in accordance with the E. coli multilocus sequence typing website (http://web.mpiib-berlin.mpg.de/mlst/dbs/Ecoli).

Induction of stx₂-converting phages from SF EHEC O157:NM and transduction experiments.

The stx₂-converting bacteriophage Φ258₃₂₀ was induced from SF EHEC O157:NM strain B1 (258/98) (11) using 0.5 μg/ml of mitomycin C (Sigma-Aldrich, Deisenhofen, Germany) (45). One hundred microliters of the phage lysate was mixed with 100 μl of the log-phase culture (l0⁶ to 10⁷ CFU) of an stx-negative SF E. coli O157:NM strain (A2, B2, or C2) and 125 μl of 0.1 M CaCl₂ solution and incubated for 2 h at 37°C without shaking. After that, the mixture was transferred into 4 ml of Luria-Bertani (LB) broth and incubated for 48 h at 37°C and 180 rpm. Tenfold dilutions of the culture were inoculated on LB agar, and overnight growths harvested into 1 ml of saline were screened for stx₂ by PCR with primers LP43 and LP44. The PCR-positive cultures were restreaked on LB agar, and plates containing ∼200 well-separated colonies were used to isolate stx₂-containing colonies (putative lysogens) by colony blot hybridization with the stx₂ probe (20, 34). The stx₂-positive colonies were subcultured three times on LB agar and tested by PCR for stx₂ after the third passage to identify stable lysogens (45); the lysogens were checked for the maintenance of stx₂ at 3-month intervals for 1 year. E. coli strain C600, which can be lysogenized with phage Φ258₃₂₀ (11), was used as a positive control in the transduction experiments.

Phenotypes.

Sorbitol fermentation was detected on SMAC after overnight incubation. β-d-Glucuronidase activity was determined on Difco nutrient agar with 4-methylumbelliferyl β-d-glucuronide (Becton Dickinson, Heidelberg, Germany). Production of EHEC hemolysin was detected on enterohemolysin agar (Sifin, Berlin, Germany) and resistance to tellurite on cefixime-tellurite-SMAC (9). Stx production was investigated using a latex agglutination assay (Verotox-F; Denka Seiken Co., Ltd., Tokyo, Japan). Stx cytotoxicity titers in culture supernatants were assessed using the Vero cell assay (30). Production of cytolethal distending toxin V (CDT-V) was tested using the Chinese hamster ovary cell assay (26).

RESULTS

Characteristics of stx₂-positive and stx-negative SF E. coli O157:NM strains isolated from sequential stool specimens.

All three initial EHEC O157:NM strains isolated from stool specimens from the three patients and all three subsequent stx-negative isolates shared putative virulence genes encoding non-Stx toxins (EHEC hemolysin and CDT-V) and various adhesins (Table 1). Moreover, all six strains displayed an identical combination of plasmid-carried genes (presence of EHEC-hlyA, etpD, and sfpA and absence of katP and espP). Each of the strains contained rfb_O157 and fliC_H7 but lacked the terE gene and, accordingly, failed to grow on cefixime-tellurite-SMAC. All strains fermented sorbitol, produced β-d-glucuronidase, and secreted CDT-V in similar amounts, but all failed to express the enterohemolytic phenotype (Table 1). Moreover, they shared identical sequence types (ST11), indicating identical clonal backgrounds (Table 1). The finding that the stx₂-positive and stx-negative intrapatient isolates share stx-independent characteristics suggested that the latter organisms descended from the former ones by the loss of stx₂ during infection.

TABLE 1.

Molecular and phenotypic traits of consecutive stx₂-positive and stx-negative SF E. coli O157 isolates from patients A, B, and C

Strain	Molecular traita														Phenotypeb				STc
	rfbO157	fliCH7	stx2	EHEC hlyA	cdt-V	eae	efa1d	sfpA	lpfAO157-OI141	lpfAO157-OI154	terE	katP	espP	etpD	SF/ GUD	EHEC-Hly	CDT	Ter
A1	+	+	+	+	+	γ	+	+	+	+	−	−	−	+	+/+	−	1:16	−	11
A2	+	+	−	+	+	γ	+	+	+	+	−	−	−	+	+/+	−	1:8	−	11
B1	+	+	+	+	+	γ	+	+	+	+	−	−	−	+	+/+	−	1:8	−	11
B2	+	+	−	+	+	γ	+	+	+	+	−	−	−	+	+/+	−	1:8	−	11
C1	+	+	+	+	+	γ	+	+	+	+	−	−	−	+	+/+	−	1:8	−	11
C2	+	+	−	+	+	γ	+	+	+	+	−	−	−	+	+/+	−	1:4	−	11

^aThe presence of the genes was determined by PCR as described in Materials and Methods; +, presence of the gene; −, absence of the gene.

^bPhenotypes were determined as described in Materials and Methods. SF, sorbitol fermentation; GUD, production of β-d-glucuronidase; EHEC-Hly, production of EHEC hemolysin; CDT, production of CDT-V (titers determined using Chinese hamster ovary cells); Te^r, tellurite resistance. +, presence of the phenotype; −, absence of the phenotype.

^cST, sequence type.

^dA complete efa1 gene (ca. 10 kb) was identified in all strains using PCRs targeting the 5′, middle, and 3′ regions of the gene (27).

Integrity of yecE in stx₂-positive and stx-negative SF E. coli O157:NM strains.

yecE was occupied in each of the stx₂-positive strains of the original infection (A1, B1, and C1) (Table 2; Fig. 1, lanes 1 to 3) and was intact in all three stx-negative strains (A2, B2, and C2) (Table 2; Fig. 1, lanes 4 to 6).

TABLE 2.

Presence of stx₂ genes and occupation of yecE with an stx₂-harboring phage in SF E. coli O157:NM parental stx₂-positive strains, their stx-negative derivatives, and lysogens

Patient's isolate or lysogen	Result of PCR targeting:			Stx2 titer (latex agglutination assay)c	Stx cytotoxicity titer (Vero cell assay)
	stx2	yecEa	yecE-intΦ258320b
A1	+	−	+	1:64	1:128
A2	−	+	−	<1:2	<1:2
B1	+	−	+	1:256	1:1,024
B2	−	+	−	<1:2	<1:2
C1	+	−	+	1:128	1:256
C2	−	+	−	<1:2	<1:2
B2(Φ258320)-1	+	−	+	1:64	1:64
B2(Φ258320)-2	+	−	+	1:64	1:128
B2(Φ258320)-3	+	−	+	1:128	1:256
B2(Φ258320)-4	+	−	+	1:128	1:128
C2(Φ258320)-1	+	−	+	1:128	1:256
C2(Φ258320)-2	+	−	+	1:64	1:128
C600	−	+	−	<1:2	<1:2
C600(Φ258320)	+	−	+	1:16	1:32

^a+, the locus is intact; −, the locus is occupied by foreign DNA.

^b+, a phage with the int gene homologous to that of phage Φ258₃₂₀ is integrated in yecE; −, the phage is absent.

^cVerotox-F (Denka Seikesn, Tokyo, Japan).

FIG. 1.

PCR analysis of occupation of yecE in SF EHEC O157:NM clinical isolates, their stx-negative derivatives, and lysogens. The strains tested, PCR targets, and lengths of PCR amplicons are listed across the top and to the left and right of the rows of amplicons, respectively. Strain B2 (258/98), harboring phage Φ258₃₂₀ integrated in yecE (11), and E. coli K-12 C600, which has yecE intact (13), were used as controls. In PCR targeting yecE, the presence of an amplicon indicates that the locus is intact, whereas the absence of an amplicon (or a very weak amplicon) indicates that the locus is occupied by foreign DNA. In PCR connecting yecE with the integrase gene (int) of phage Φ258₃₂₀, the presence of an amplicon indicates that a phage with a homologous int gene is integrated in yecE; the absence of an amplicon indicates the absence of such phage.

Transduction of stx-negative SF E. coli O157:NM with an stx₂-harboring phage from SF EHEC O157:NM.

A stable (for at least 12 months) infection with phage Φ258₃₂₀ could be achieved in two of the three stx-negative SF E. coli O157:NM strains. The remaining strain (A2) could be transiently infected but lost the phage after the second passage on LB agar. Four stable lysogens (among 246 transduced colonies) originated from the cognate stx-negative strain B2, an stx-negative derivative of strain B1 (258/98) (transduction frequency of 1.6%) and were designated B2(Φ258₃₂₀)-1 to B2(Φ258₃₂₀)-4. Two stable lysogens (among 221 transduced colonies) originated from a noncognate stx-negative strain (C2) derived from EHEC strain C1 (transduction frequency of 0.9%) and were designated C2(Φ258₃₂₀)-1 and C2(Φ258₃₂₀)-2. The lysogenization of the stx-negative strains with stx₂-harboring phage Φ258₃₂₀ resulted in the acquisition of stx₂ by each of the lysogens, as demonstrated by PCR with primers LP43-LP44 (Table 2; Fig. 1, lanes 7 to 12). Moreover, phage Φ258₃₂₀ also transduced the stx₂ gene into the control strain E. coli C600 (Table 2; Fig. 1, lane 14).

Stx production.

The latex agglutination assay confirmed that all three parental stx₂-containing SF EHEC O157:NM strains produced Stx2 (titers of 1:64 to 1:256) and that their stx-negative derivatives did not (titers of <1:2) (Table 2). Each of the six lysogens produced Stx2 in titers (1:64 to 1:128) comparable to that of the donor strain B1 (1:256) (Table 2). Stx2 produced by the lysogens was cytotoxic toward Vero cells (Table 2), indicating that the biological activity is similar to that of Stx2 produced by the wild-type SF EHEC O157:NM strains.

Phage integration sites.

In each lysogen, the acquisition of the stx₂-harboring phage Φ258₃₂₀ resulted in the occupation of yecE (Table 2; Fig. 1, lanes 7 to 12), which had been intact in the stx-negative recipient strains B2 and C2 (Table 2; Fig. 1, lanes 5 and 6, respectively). We confirmed that yecE in the lysogens was occupied by phage Φ258₃₂₀, using PCR to link yecE with the integrase gene of phage Φ258₃₂₀. The PCR produced an amplicon of 425 bp in strain B1 (258/98) (and also A1 and C1) (Fig. 1, lanes 1 to 3) and each of the lysogens (Fig. 1, lanes 7 to 12), demonstrating that all these strains contained a phage homologous to phage Φ258₃₂₀ integrated in yecE. As expected, phage Φ258₃₂₀ also integrated into yecE in the control E. coli strain C600 (Fig. 1, lane 14). Taken together, these data demonstrate that stx-negative SF E. coli O157:NM strains can be converted to EHEC O157:NM via transduction with an stx₂-harboring phage originating from an EHEC O157:NM strain and that yecE plays an essential role in this process.

DISCUSSION

One noteworthy aspect of EHEC infections is that the infecting organism can undergo major genotype and phenotype changes during the course of infection in humans (10, 21, 34). We have demonstrated interconversion between various pathotypes of SF E. coli O157:NM, in which EHEC O157:NM strains harboring stx₂ lose stx₂ and stx-negative E. coli O157:NM can be transduced by an stx₂-harboring phage from an EHEC O157 patient's isolate. In these events, yecE is a hot spot for integration and excision of stx₂ phages. The ability to undergo bidirectional conversion from one pathotype to another enables recycling of an important pathogenic trait in SF E. coli O157:NM and indicates that these organisms can respond to environmental influences with a high degree of flexibility and that this property increases their capacity to survive under a variety of conditions in the human host and reservoir(s).

Although yecE is only rarely occupied by phages in EHEC O157:H7 (16, 46), our analysis of consecutive stx₂-positive and stx-negative SF E. coli O157:NM strains confirms our previous observation that yecE is an important integration site for Stx2-encoding phages in SF EHEC O157:NM strains (11). It is of interest that up to now no other function is known for the yecE gene or its product. yecE in all stx₂-positive SF EHEC O157:NM strains was occupied by an stx₂-converting phage, whereas in all stx-negative SF E. coli O157:NM strains the yecE genes were intact (Fig. 1). Lysogenization of stx-negative strains with stx₂-containing phage Φ258₃₂₀ originating from a SF EHEC O157:NM clinical isolate resulted in the occupation of yecE in the lysogens (Fig. 1). The presence of phage Φ258₃₂₀ in yecE in the genomes of the lysogens could be confirmed by the signal elicited from all lysogens in the PCR connecting yecE with the integrase gene of phage Φ258₃₂₀ (Fig. 1). Together with acquiring the stx₂ gene, all lysogens also acquired the ability to produce Stx2 cytotoxic to Vero cells. This demonstrates that via transduction with the stx₂ phage, the stx-negative, eae-positive SF E. coli O157:NM strains were converted to biologically potent EHEC. In contrast to the case for SF EHEC O157:NM, stx₂ phages in EHEC O157:H7, including both sequenced strains EDL933 and Sakai (24, 32, 38, 40) and other clinical isolates (6, 47), use wrbA as a common genomic integration site, whereas yehV is frequently used for integration of stx₂ phages in EHEC O157:H7 isolated from cattle (46). In addition, other, mostly yet-unidentified chromosomal loci have been proposed to serve as integration sites for stx₂-harboring phages in the genome of EHEC O157:H7 (6, 36, 47). Although it is presently unknown why yecE, which is intact in E. coli O157:H7 strains EDL933 and Sakai (24, 38) and other isolates (unpublished data), does not regularly serve as an stx₂ phage integration site in these organisms (46), this fact is in accordance with a recent observation that the selection of an stx phage integration site preferentially depends on the host strain rather than on the phage (46). The greater stability of stx₂ in EHEC O157:H7 than in SF EHEC O157:NM (21, 34) might be attributable to the difference in the integration sites for stx₂-harboring phages in the two organisms.

The observation that stx-negative SF E. coli O157:NM strains occur more frequently than stx-positive SF E. coli O157:NM strains in the gastrointestinal tracts of animals (31, 43, 59) suggests that an stx-negative phase might be a normal occurrence in the life cycle of SF EHEC O157:NM in the environment. The stx-negative condition might enable the pathogens to avoid lysis triggered by various phage-inducing stimuli, such as H₂O₂ (52), thereby conferring a selective advantage to survive under different conditions. If analogous to E. coli O157:H7 (15), the absence of an stx phage would not influence the ability of stx-negative SF E. coli O157 organisms to colonize the intestines of animals and thus to be spread in the environment via fecal contamination. We propose that such stx-negative, eae-positive SF E. coli O157:NM strains, which can be converted to EHEC by transduction with stx₂ phages from EHEC O157 in vitro, represent “EHEC wannabes.” The acquisition of an stx phage by such organisms not only makes them highly pathogenic for the human host but also could play a role in their ecology. It has been recently shown that the carriage of an Stx-encoding prophage increases the rate of survival of E. coli O157:H7 within grazing protozoa (Tetrahymena pyriformis) (50) and thus in the environment. Apparently, both gain and loss of stx phages can contribute to the evolution of SF E. coli O157:NM. Because of their ability to recycle stx₂, which is the major virulence trait, SF E. coli O157:NM can be considered ephemeral EHEC strains that can exist as stx-negative organisms during certain phases of their life cycle.

A change in the pathotype of the infecting strain during infection has consequences regarding the laboratory diagnosis and epidemiological investigations. Loss of the stx gene makes it impossible to correctly identify stx-negative variants using methods based on the detection of only stx and/or Stx. To overcome this limitation, we recommend the parallel use of procedures independent of the presence of stx, in particular the detection of eae using PCR (34). In the case of SF E. coli O157:NM, detection of O157 lipopolysaccharide antigen by slide agglutination and/or its encoding gene using PCR and, especially, detection of the sfpA gene, which is a unique marker for SF E. coli O157:NM (19), allow identification of both stx-positive and stx-negative strains (19). Furthermore, loss of stx genes in strains that are related epidemiologically can change the pulsed-field gel electrophoresis pattern (11), thereby complicating epidemiological investigations of SF E. coli O157:NM disease (11). Therefore, the possibility of an altered pulsed-field gel electrophoresis pattern arising from the loss of stx should be always considered when interpreting the DNA fingerprints of potentially epidemiologically related SF E. coli O157:NM strains.

Ongoing studies in our laboratories aim to investigate the conditions favoring the loss of stx and the possibility of promoting the excision of stx and conversion to less pathogenic stx-negative EHEC. In the absence of specific anti-EHEC treatment, such an intervention would offer a new and powerful preventive and therapeutic approach.