Article | Published:

A new biological agent for treatment of Shiga toxigenic Escherichia coli infections and dysentery in humans

Nature Medicine volume 6, pages 265270 (2000) | Download Citation



Gastrointestinal disease caused by Shiga toxin-producing bacteria (such as Escherichia coli O157:H7 and Shigella dysenteriae) is often complicated by life-threatening toxin-induced systemic sequelae, including hemolytic–uremic syndrome. Such infections can now be diagnosed very early in the course of the disease, but at present no effective therapeutic intervention is possible. Here, we constructed a recombinant bacterium that displayed a Shiga toxin receptor mimic on its surface, and it adsorbed and neutralized Shiga toxins with very high efficiency. Moreover, oral administration of the recombinant bacterium completely protected mice from challenge with an otherwise 100%-fatal dose of Shiga toxigenic E. coli. Thus, the bacterium shows great promise as a ‘probiotic’ treatment for Shiga toxigenic E. coli infections and dysentery.


Shiga toxin-producing Escherichia coli (STEC) strains, including serotype O157:H7, are important causes of diarrhea and hemorrhagic colitis in humans. This can lead to potentially fatal systemic sequelae, such as hemolytic–uremic syndrome (HUS), the leading cause of acute renal failure in children1,2,3,4. The mortality rate for HUS is 5–10%; other acute complications include stroke, diabetes mellitus and necrotizing colitis2. In recent years, there have been large outbreaks of STEC disease in North America, the UK, Europe, Japan and Australia. Certain other Enterobacteriaceae also produce Shiga toxin, the most notable being Shigella dysenteriae type 1, the causative agent of bacillary dysentery, which is often associated with Shiga toxin-induced neurological sequelae and HUS (ref. 1).

Shiga toxin (Stx) is a compound toxin, comprising a catalytic A subunit, which inhibits eukaryotic protein synthesis, and a pentameric B subunit responsible for binding to glycolipid receptors in target cell membranes2. There are two main classes of Stx (Stx1 and Stx2), and Stx produced by S. dysenteriae is essentially identical to E. coli Stx1. Within the Stx2 class, additional subtypes have been distinguished on the basis of receptor binding efficiency (Stx2c and Stx2e) or activation by intestinal mucus (Stx2d). All Stx types associated with human disease (Stx1, Stx2, Stx2c and Stx2d) recognize the same glycolipid receptor, globotriaosyl ceramide (Gb3), which has the structure Galα[1→4]Galβ[1→4]Glc-ceramide5. The Stx2e subtype is produced only by STEC strains associated with piglet edema disease and has a different receptor specificity, recognizing globotetraosyl ceramide (Gb4; GalNAcβ[1→3]Galα[1→4]Galβ[1→4]-Glc-ce-ramide) preferentially over Gb3 (ref. 6).

The pathological features of STEC disease are directly attributable to Stx, which is essential for virulence. Pathogenesis initially involves colonization of the gut by STEC, which does not invade the epithelium. Locally produced Stx is absorbed into the circulation and targets specific tissues in accordance with their Gb3 content. In humans, Gb3 is found in its highest concentrations in renal tissue and in microvascular endothelial cells, thereby accounting for the characteristic features of HUS (microangiopathic hemolytic anemia, thrombocytopenia and renal failure).

Development of rapid and sensitive methods for early diagnosis of STEC infection has created a ‘window of opportunity’ for therapeutic intervention. Indeed, STEC infection may be detected almost a week before symptoms of HUS become apparent7,8. Furthermore, increased awareness during major outbreaks will result in more patients presenting during the prodromal stage. Contacts of persons with proven or suspected STEC infection could also be treated. Unfortunately, antibiotic therapy is contraindicated for STEC infection, because it increases free Stx in the gut lumen by releasing cell-associated toxin and inducing toxin gene expression2,4. Thus, adsorption or neutralization of Stx in the gut is a potentially important alternative therapeutic strategy. Accordingly, we inserted genes encoding synthesis of the oligosaccharide Galα[1→4]Galβ[1→4]Glc- into E. coli such that the outer core region of the E. coli lipopolysaccharide (LPS) mimics the Stx receptor. We then assessed the capacity of this bacterium to adsorb and neutralize Stx in vitro and in vivo .

Construction of recombinant bacteria expressing Galα[1→4]Galβ[1→4]Glc- on their surface

Surface display of host epitopes is a strategy adopted by several mucosal pathogens for evasion of immune defenses. Indeed, Galα[1→4]Galβ[1→4]Glc- is one of the alternative structures for the outer core of the lipooligosaccharides (LOS) of Neisseria meningitidis9, N. gonorrhoeae10 and Haemophilus influenzae11,12. The N. gonorrhoeae and N. meningitidis loci encoding outer core LOS biosynthesis are conserved between the two species and contain five glycosyl transferase genes, lgtA, lgtB, lgtC, lgtD and lgtE, arranged as an operon (refs. 13,​14,​15 and GenBank accession number U65788). lgtA, lgtC and lgtD contain poly-G tracts, which renders them very susceptible to ‘slipped-strand’ mispairing during replication, causing frame-shift mutations14. Thus, the actual outer core oligosaccharide displayed on the bacterial surface will depend on which of the lgt genes are encoding functional enzymes at a given point in time. The specificity of the five gonococcal transferases has been determined by mutational analysis14. Expression of the so-called L1 immunotype LOS, which contains the Galα[1→4]Galβ[1→4]Glc-epitope, requires functional lgtE and lgtC genes, which encode the transferases responsible for linking the β-galactosyl and α-galactosyl residues to glucose at the distal end of the inner core oligosaccharide.

Accordingly, we amplified the lgtC and lgtE genes by PCR from N. meningitidis and N. gonorrhoeae chromosomal DNA, respectively, mutated the poly-G tract in lgtC to stabilize expression, and cloned both genes in tandem into the plasmid vector pK184 (Fig. 1 ). We then transformed the recombinant plasmid, pJCP-Gb3, into a derivative of E. coli R1 (CWG308) that has a mutation in the waaO gene. CWG308 produces an LPS core consisting of the inner core plus glucose linked to the terminal heptose residue16 ( Fig. 2), which closely resembles the natural acceptor for LgtE. Thus, we predicted that E. coli CWG308:pJCP-Gb3 would express a chimeric LPS core terminating in Galα[1→4]Galβ[1→4]Glc-, a mimic of the Stx receptor (Fig. 2). Purified LPS from E. coli CWG308:pJCP-Gb3 migrated slower through SDS–polyacrylamide gels than LPS from E. coli CWG308 (data not shown), which is consistent with its predicted higher molecular size. Glycosyl composition analysis of E. coli CWG308:pJCP-Gb3 LPS indicated that galactose (which is absent in the LPS of CWG308) was present at a molar ratio with respect to heptose of 1.86:3, which is close to the 2:3 ratio expected for the structure proposed (Fig. 2). Production of LPS with a terminal Galα[1→4]Galβ[1→4]Glc- epitope by CWG308:pJCP-Gb 3 was also consistent with weak immunoblot reactivity of crude cell lysates with a monoclonal antibody specific for the N. meningitidis L1 immunotype; in contrast, there was no reactivity with CWG308 lysates (data not shown).

Figure 1: Construction of pJCP-Gb3.
Figure 1

The lgtC (C) and lgtE (E) genes were amplified by PCR from N. meningitidis and N. gonorrhoeae chromosomal DNA, respectively, incorporating either EcoRI (Ec), HindIII (Hi), or BamHI (Ba) restriction sites at the termini. Stabilization of the poly-G tract in lgtC by overlap–extension PCR used PCR product cloned in pK184. The stabilized lgtC gene (*C) and lgtEwere then inserted in tandem, in the same orientation as the vector promoter (Plac) into pK184, which includes a kanamycin resistance marker (kanR).

Figure 2: Proposed structure of the LPS of E. coli CWG308:pJCP-Gb3.
Figure 2

Bottom row: right, the portion present in the LPS of E. coli CWG308 (based on refs. 16,37); left, the epitope that functions as the Stx receptor mimic. Wide downward arrows, linkages formed by the galactosyl transferases LgtC and LgtE. Gal, galactose; Glc, glucose; Hep, L-glycero-D-manno-heptose; Kdo, 3-deoxy-D-manno-oct-2-ulosonic acid.

Adsorption/neutralization of Stx by CWG308:pJCP-Gb3

We next assessed the capacity of CWG308 and CWG308:pJCP-Gb3 to adsorb and neutralize Stx in lysates of wild-type STEC strains or E. coli JM109 derivatives expressing cloned stx genes. We incubated suspensions of CWG308 or CWG308:pJCP-Gb3 cells with each of the Stx extracts and determined Stx activity in filter-sterilized supernatant fractions by Vero cytotoxicity assay, comparing each with that of the respective Stx extract incubated without added bacteria (Table 1). Treatment with CWG308:pJCP-Gb3 neutralized more than 98% of the cytotoxicity for each of the Stx types associated with human disease (Stx1, Stx2, Stx2c and Stx2d). However, neutralization of the variant toxin (Stx2e) produced by STEC strains associated with piglet edema disease was less efficient (87.5% neutralization). This is consistent with the fact that Stx2e has a different glycolipid receptor specificity from the other members of the Stx family. In contrast, treatment with CWG308 did not result in any detectable neutralization of any of the Stx extracts (Table 1). Expression of both lgtC and lgtE was essential for toxin adsorption, as incubation of Stx extracts with CWG308 derivatives carrying either gene (cloned individually in pK184) did not result in detectable neutralization of cytotoxicity (data not shown).

Table 1: Neutralization of Stx in crude E. coli lysates

We also assessed the capacity of killed CWG308:pJCP-Gb3 cells to bind and neutralize cytotoxicity in extracts, prepared in a French pressure cell, containing Stx1 or Stx2c. Cell suspensions were killed by being heated at 65 °C for 3 hours, or by being treated with 1% formaldehyde for 16 hours at 4 °C. Heat-killed cells neutralized 93.7% and 96.8% of the Stx1 and Stx2c, respectively, compared with 99.2% and 99.6%, respectively, for live cells. However, formaldehyde-killed CWG308:pJCP-Gb3 cells were as effective as live cells, neutralizing 99.6% of the Stx1 and 99.2% of the Stx2c.

Adsorption studies using purified Stx1 and Stx2

To determine the total adsorption/neutralization capacity of CWG308:pJCP-Gb 3 cells, we incubated suspensions containing 5 × 108 colony-forming units (CFU) (equivalent to 1 mg dry weight) with aliquots (ranging from 1 ng to 640 μg) of purified Stx1 or Stx2, in a final volume of 0.5 ml. We compared the residual cytotoxicity with that for similar aliquots of toxin incubated with or without CWG308 (Fig. 3). Each 1 mg dry weight of CWG308:pJCP-Gb3 was able to neutralize 200 ng of either Stx1 or Stx2 with more than 99.9% efficiency; 40-μg doses of Stx1 or Stx2 were neutralized with more than 96% efficiency. Even at doses of 160 μg/mg cells, CWG308:pJCP-Gb3 could still neutralize 87.5% of either Stx1 or Stx2.

Figure 3: Total Stx adsorption/neutralization capacity of E. coli CWG308:pJCP-Gb3.
Figure 3

Suspensions of 5×108 cells (equivalent to 1 mg dry weight) were incubated with purified Stx 1 (filled bars) or Stx2 (shaded bars) (amounts, horizontal axis). Residual cytotoxicity was assayed using Vero cells and % neutralization was then calculated.

To confirm that expression of the Gb3 receptor mimic results in binding of toxin to the bacterial surface, we incubated CWG308:pJCP-Gb 3 and CWG308 suspensions with purified Stx1 or Stx2 (10 μg toxin per mg dry weight of cells), then analyzed serial 1:2 dilutions of cell suspensions by immunoblotting using monoclonal antibodies specific for the respective Stx type. We were able to detect both Stx1 and Stx2 on the surface of CWG308:pJCP-Gb 3 even in suspensions diluted 1:64, but could not detect immunoreactivity in CWG308 suspensions diluted 1:2 (data not shown).

We also showed adsorption of Stx1 to the surface of CWG308:pJCP-Gb 3 by immunofluorescence microscopy using monoclonal antibody against Stx1 (Fig. 4). We saw no fluorescence, however, for immunostained Stx1-treated suspensions of CWG308, or in CWG308:pJCP-Gb 3 suspensions stained without prior exposure to Stx1 ( Fig. 4).

Figure 4: Immunofluorescent staining.
Figure 4

Suspensions of E. coli CWG308 incubated with purified Stx1 (a) or CWG308:pJCP-Gb3 incubated with Stx1 (b) or without Stx1 (c), were reacted with monoclonal antibody against Stx1, and antibody against mouse IgG conjugated to FITC. Left, epi-fluorescent microscopy; right, phase contrast microscopy of the same fields.

In vivo protection studies

We assessed the in vivo protection afforded by oral administration of CWG308:pJCP-Gb3 using a streptomycin-treated mouse model of lethal renal damage induced by STEC. We used two wild-type STEC strains: B2F1 (an O91:H21 STEC that produces Stx2d) and 97MW1 (an O113:H21 STEC that produces Stx2). B2F1 is a well-characterized strain with very high virulence in this model17; 97MW1 is from our own collection and is equally virulent. We challenged two groups of eight streptomycin-treated BALB/c mice orally with approximately 1 × 108 CFU of B2F1, and challenged another two groups of eight mice with 97MW1. We then administered oral doses of approximately 4 × 109 CFU (equivalent to 8 mg dry weight) of either CWG308 or CWG308:pJCP-Gb3 twice daily, and monitored survival time (Fig. 5). Both STEC strains used for challenge rapidly colonized the gut, and fecal pellets collected 24 hours after challenge contained more than 1 × 109 CFU of the respective STEC per gram. Fecal pellets from groups that received CWG308:pJCP-Gb3 also contained approximately 1 × 103 CFU/g of this strain; similar numbers of CWG308 were present in fecal samples collected from the remaining groups. For both the B2F1 and 97MW1 groups, all of the mice that received oral CWG308 died, with a median survival time of approximately 4 days (Fig. 5). This is similar to that seen for unprotected mice challenged with the same strains (data not shown). However, the mice challenged with STEC that received CWG308:pJCP-Gb3 remained well, even though fecal pellets collected on day 8 still contained more than 1 × 109 CFU/g of the respective STEC. The numbers of CWG308:pJCP-Gb 3 in these fecal samples were approximately 1 × 105 CFU/g. We discontinued administration of CWG308:pJCP-Gb3 on day 12, and ended the experiment on day 15, at which time, all of the mice in the groups that received CWG308:pJCP-Gb3 were alive and well (Fig. 5). The difference in survival rate relative to mice given CWG308 (eight of eight compared with none of eight) was highly significant (P 0.005, Fisher exact test) and demonstrates unequivocally that oral administration of CWG308:pJCP-Gb3 is capable of preventing the fatal systemic complications of STEC disease. None of the fecal samples collected from the survivors on day 15 contained detectable levels of CWG308:pJCP-Gb3, indicating that the recombinant bacterium had been spontaneously cleared from the gut.

Figure 5: Survival of mice challenged with virulent STEC.
Figure 5

Groups of streptomycin-treated mice were challenged with either B2F1 or 97MW1, and then treated with twice daily oral doses of E. coli CWG308 or CWG308:pJCP-Gb3. Data represent the survival time of each mouse.


This study is the first report to our knowledge of the deliberate engineering of expression of an oligosaccharide receptor mimic on the surface of a recombinant bacterium. Moreover, it is the prototype for a new ‘probiotic’ strategy for the prevention and treatment of a wide range of enteric diseases. Many bacterial and viral pathogens exploit oligosaccharide moeities of glycoproteins or glycolipids on the surface of eukaryotic cells as receptors for toxins, adhesins or other ligands. Construction of a given receptor mimic requires the identification of the specific glycosyl transferases required for its synthesis, and insertion of genes encoding these into a heterologous host producing an appropriate surface-expressed acceptor molecule. In the prototypic example, expression of two Neisseria galactosyl transferase genes in a waaO mutant of E. coli R1 resulted in production of a modified LPS that terminates in the trisaccharide Galα[1→4]Galβ[1→4]Glc-. This mimics the natural receptor for all Stx types associated with human disease, and the recombinant bacterium (E. coli CWG308:pJCP-Gb3) showed a very high capacity to adsorb and neutralize these toxins. Experiments using purified toxins indicated that at saturation, the equivalent of 1 mg dry weight of CWG308:pJCP-Gb3 cells could adsorb 50–100 μg of Stx1 or Stx2. This binding capacity is approximately 10,000 times greater than that reported in other studies for a chemically synthesized Stx adsorbent (Synsorb-Pk), which consists of Galα[1→4]Galβ[1→4]Glc- covalently linked to silica particles18,19. The massive difference in toxin binding between the natural and synthetic adsorbents may be due to the density of receptor mimics displayed on the particle or bacterial surface. Alternatively, presentation of the receptor mimic on LPS embedded in the fluid bacterial outer membrane may facilitate molecular rearrangements that optimize interaction between toxin and receptor.

We have also shown here that twice-daily oral administration of CWG308:pJCP-Gb 3, but not CWG308, completely protected streptomycin-treated mice from an otherwise 100% fatal challenge with either of two highly virulent STEC strains (B2F1 and 97MW1), each producing different Stx2 subtypes. This degree of protection was noteworthy, because very large amounts of STEC were maintained in the mouse gut throughout the experiment, outnumbering CWG308:pJCP-Gb 3 approximately 1 × 109 to 1 × 10 5. In a previous report, oral administration of Synsorb-Pk did not protect mice from oral challenge with B2F1, although it did delay death by 1 day (ref. 20). The much-better in vitro Stx-binding capacity of CWG308:pJCP-Gb3 and the absolute protection afforded by oral administration of this bacterium in the mouse model indicate that it may also be very protective in humans.

The use of live bacteria for the treatment and prevention of en-teric disease and its complications has considerable cost advantages over the use of synthetic toxin adsorbents. Expression of LPS O-antigen by enteric bacteria increases their capacity to survive in and colonize the gut. However, CWG308:pJCP-Gb 3 lacks the portion of the LPS outer core to which O-antigen is attached in wild-type strains16, and consequently is attenuated. Thus, the numbers of CWG308:pJCP-Gb3 in the mouse gut were low during the treatment period, and stable colonization was not established, as judged by its spontaneous clearance soon after withdrawal. Although these are desirable properties from the human safety and regulatory viewpoint, improving the survival of the bacterium in the gut may enable reduction in the dose and frequency of administration. Another safety issue is that administration of a human receptor mimic might ‘break’ tolerance and induce autoimmunity. However, this is unlikely because just the presence of an antigen in the gut is not sufficient to invoke autoimmune responses. Indeed, phase I clinical trials have shown that Synsorb-Pk, which displays the same epitope, is safe for administration to children21. Induction of systemic immune responses to heterologous antigens through the enteric route typically requires expression by a live bacterium with the capacity to penetrate the gut mucosa, or co-administration of a strong mucosal adjuvant22 (Stx has no such activity). However, for self antigens, even co-administration with strong adjuvants is incapable of ‘breaking’ an established state of immune tolerance23. Furthermore, Galα[1→4]Galβ[1→4]Glc- is one of the alternative LOS epitopes displayed by N. meningitidis and H. influenzae, presumably as a strategy to evade host immune defences24. Both of these organisms asymptomatically colonize the nasopharynx of a substantial proportion of the human population, but so far, no autoimmune disorder has been associated with such carriage, nor indeed with invasive diseases caused by either pathogen. Nevertheless, caution should be exercised, and additional animal studies are required to ensure that oral administration of CWG308:pJCP-Gb3 does not elicit inappropriate immune responses. Formaldehyde-killed E. coli CWG308:pJCP-Gb3 could be used if administration of live cells is considered imprudent, as such treatment does not diminish Stx binding capacity.

This study unequivocally demonstrates the therapeutic potential of recombinant bacteria expressing an Stx receptor mimic. Oral administration of this new agent to individuals diagnosed with, or at risk of, STEC or S. dysenteriae infection has the potential to produce adsorption and neutralization of free Stx in the gut lumen, thereby preventing absorption of toxin into the bloodstream and the concomitant life-threatening systemic renal and neurological sequelae associated with STEC disease and dysentery in humans.


Bacterial strains and plasmids.

The bacterial strains and plasmids used or generated in this study are described in Table 2. CWG308 was provided by C. Whitfield (University of Guelph, Ontario, Canada); 128/12 was provided by C. Stevens (Queensland Department of Primary Industry, Toowoomba, Australia.). All E. coli strains were routinely grown in Luria-Bertani (LB) medium25 with or without 1.5% bacto-agar. Where appropriate, 50 μg/ml ampicillin or 50 μg/ml kanamycin was added to growth media.

Table 2: Bacterial strains and plasmids

Cloning and mutagenesis of galactosyl transferase genes.

Oligo-nucleotide primers for PCR amplification of lgtC and lgtE genes from either N. gonorrhoeae or N. meningitidis were designed with reference to sequence data in GenBank (accession numbers U14554 and U65788). For lgtC, the primers used were 5′–GAACAG GAATTCGGCAAGATTATTGTGCC–3′, which incorporates an Eco RI site (underlined), and 5′–TACGTCGGATCCCGTCT-GAAGGCTTCAGAC–3′, which incorporates a BamHI site (underlined). For lgtE, the primers used were 5′–GCCCTTGGATCCACCGCAGC-TATTGAAACC-3′, incorporating a BamHI site (underlined) and 5′–CCATT-T AAGCTTTTAATCCCCTATATTTTACAC–3′, incorporating a HindIII site (underlined). These were used to amplify the complete open reading frame and the ribosome-binding site for lgtC and lgtE, using N. meningitidis and N. gonorrhoeae chromosomal DNA as template, respectively. The PCR products were then cloned into the vector pK184 after digestion of both vector and PCR product with EcoRI and BamHI (for lgtC ) or BamHI and HindIII (for lgtE), and were used to transform E. coli JM109. Because the lgtC gene has a poly-G tract, it was necessary to mutate this region to stabilize expression of the encoded transferase. The DNA sequence of N. meningitidis lgtC nucleotides 157–171 of the open reading frame is 5′–CGGG-GGGGGGGGGGT–3′, which encodes the amino-acid sequence Arg-Gly-Gly-Gly-Gly. This was mutated to 5′–CGTGGCGGTGGCGGT–3′ by overlap extension PCR, which involved separate amplification of overlapping 5′ and 3′ portions of the lgtC gene cloned in pK184. The 5′ portion was amplified using the universal M13 reverse sequencing primer and a primer with the sequence 5′–ATATTACCGCCACCGCCACGCAAATTGGCGGC–3′, whereas the 3′ portion was amplified using the universal M13 forward sequencing primer and another primer with the sequence 5′–AATTTGCGTGGCGGTG-GCGGTAATATCCGCTT–3′. The two PCR products were then purified, aliquots were mixed, and full-length lgtC with the desired modifications was amplified by PCR using the M13 forward and reverse primers. The PCR product was digested with EcoRI and BamHI and once again cloned into similarly digested pK184, and subjected to sequence analysis to confirm mutagenesis of the poly-G tract. This eliminated the possibility of ‘slipped-strand’ mispairing without affecting the amino acid sequence of the encoded protein. The mutated lgtC gene was then excised from the pK184 construct by digestion with EcoRI and BamHI and was cloned into the compatible restriction sites in the pK184 derivative containing lgtE. This construct, pJCP-Gb 3, places the lgtC and lgtE genes in tandem in pK184, in the same orientation as the vector lac promoter ( Fig. 1).

Purification and analysis of LPS.

LPS was purified from E. coli CWG308 and E. coli CWG308:pJCP-Gb3 according to a published method26. LPS extracts were separated by SDS–PAGE and silver-stained as described27. Glycosyl compositions was analyzed by the Complex Carbohydrate Research Center (University of Georgia, Athens, Georgia).

Purified toxins and monoclonal antibodies.

Purified Stx1 and Stx2 and monoclonal antibodies 13C4 (specific for the Stx1 B subunit) and BB12 (specific for he Stx2 B subunit) were obtained from Toxin Technologies (Sarasota, Florida). When tested using the Vero cytotoxicity assay described below, the purified Stx1 had a specific activity of 2.5 × 10 8 cytotoxic doses (CD)/mg, whereas the purified Stx2 had a specific activity of 1.3 × 108 CD/mg. Monoclonal antibody 17-1-L1, which is specific for N. meningitidis L1 immunotype LOS, was provided by P. van der Ley (National Institute for Public Health and the Environment (RIVM), Bilthoven, the Netherlands).

Crude Stx extracts.

Crude Stx extracts were prepared by growing STEC strains or E. coli JM109 derivatives expressing various cloned stx genes in 10 ml LB broth overnight at 37 °C. Cells were collected by centrifugation and were resuspended in 10 ml phosphate-buffered saline (PBS), pH 7.2, and lysed in a French pressure cell operated at 12,000 p.s.i. Lysates were then sterilized by passage through a 0.45-μm filter.

Stx adsorption/neutralization assay.

E. coli CWG308 or E. coli CWG308: pJCP-Gb3 cells were grown overnight in LB broth supplemented with 20 μg/ml IPTG, and 50 μg/ml kanamycin for CWG308:pJCP-Gb3. Cells were collected by centrifugation, washed and resuspended in PBS at a density of 1 × 109 CFU/ml. Aliquots (250 μl) of each Stx extract (or purified Stx) were incubated with 500 μl of either CWG308 or CWG308:pJCP-Gb3 suspension, or PBS, for 1 h at 37 °C with gentle agitation. The mixtures were then centrifuged and filter-sterilized. Cytotoxicity was then assayed using Vero (African green monkey kidney) cells, which are very susceptible to all Stx-related toxins2. Twelve serial dilutions of 1:2 were prepared in tissue culture medium (Dulbecco's modified Eagle medium buffered with 20 mM HEPES, and supplemented with 2 mM L-glutamine, 50 IU/ml penicillin and 50 μg/ml streptomycin), starting at a dilution of 1:20, or greater when high concentrations of purified Stx1 or Stx2 were tested. For each dilution, 50 μl was transferred onto washed Vero cell monolayers in 96-well tissue culture trays, and after 30 min of incubation at 37 °C, a further 150 μl of culture medium was added to each well. Cells were examined microscopically after 72 h of incubation at 37 °C, and were assessed for cytotoxicity. The end-point Stx titer (CD/ml) was defined as the reciprocal of the highest dilution resulting in cytotoxicity in at least 10% of the cells in a given monolayer. As a permanent record, cell monolayers were then fixed in 3.8% formaldehyde in PBS and stained with crystal violet. The percent of Stx adsorbed/neutralized was calculated using the formula 100 – (100×CDCELLS ÷ CD PBS), where CDCELLS is the Stx titer in the extracts incubated with either CWG308 or CWG308:pJCP-Gb3, as appropriate, and CD PBS is the Stx titer in the respective Stx extract treated only with PBS.

Immunoblot and immunofluorescent analysis.

For immunoblot analysis, cell suspensions were incubated 1 h at 37 °C with purified Stx1 or Stx2 at a dose of 10 μg toxin per mg dry weight of cells. Then, cells were centrifuged and washed four times with PBS. Serial 1:2 dilutions of each cell suspension, in 5-μl aliquots, were spotted onto nitrocellulose filters, which were then fixed with 10% methanol. Filters were then blocked using 5% skim milk and reacted with the appropriate monoclonal antibody. Monoclonal antibodies against Stx1 and Stx2 were used at a concentration of 2 μg/ml; for monoclonal antibody 17-1-L1, hybridoma culture supernatant was used at a dilution of 1:10. Filters were developed using goat antibody against mouse IgG conjugated to alkaline phosphatase (BioRad Laboratories, Richmond, California), and immunoreactive spots were visualized using chromogenic substrate (4-nitro blue tetrazolium and X-phosphate).

For immunofluorescent staining, suspensions of E. coli CWG308 or CWG308:pJCP-Gb3 were incubated with or without Stx1, reacted with monoclonal antibody against Stx1, and stained with goat antibody against mouse IgG conjugated to fluorescein isothiocyanate (FITC) (Sigma), as described28. Phase contrast and epi-fluorescence microscopy were done using an Olympus BH2 microscope with FITC filters and a 100× oil immersion objective.

In vivo protection studies.

The streptomycin-treated mouse model of renal injury induced by STEC has been described17,29,30. Mice were given oral streptomycin (5 mg/ml in drinking water) for 24 h before oral challenge with streptomycin-resistant derivatives of the STEC strain. Two groups of eight streptomycin-treated BALB/c mice were challenged with 1×108 CFU STEC B2F1; another two groups of eight mice were challenged with 1×108 CFU STEC 97MW1. Mice were then given oral doses of approximately 4×109 CFU of either CWG308 or CWG308:pJCP-Gb3, suspended in 60 μl of 20% sucrose and 10% NaHCO3, twice daily for up to 12 days. CWG308 is streptomycin-resistant, and treatment with streptomycin (given orally) was continued throughout the experiment. The amount of STEC, as well as of either CWG308 or CWG308:pJCP-Gb 3, as appropriate, was monitored in fecal samples from each group. The survival times of mice in each of the groups were also recorded. The differences in survival rate between mice challenged with STEC and treated with CWG308 or CWG308:pJCP-Gb3 were analyzed using the Fisher exact test.


  1. 1.

    & Shiga and shiga-like toxins. Microbiol. Rev. 51, 206–220 (1987).

  2. 2.

    Infection by verocytotoxin-producing E. coli. Clin. Microbiol. Rev. 2, 15–38 ( 1989).

  3. 3.

    & Diarrheagenic Escherichia coli. Clin. Microbiol. Rev. 11, 142– 201 (1998).

  4. 4.

    & Pathogenesis and diagnosis of Shiga toxin-producing Escherichia coli infections. Clin. Microbiol. Rev. 11, 450–479 ( 1998).

  5. 5.

    Role of verotoxin receptors in pathogenesis. Trends Microbiol. 4, 147–153 ( 1996).

  6. 6.

    , , , & Globotetraosyl ceramide is recognized by the pig edema disease toxin. J. Biol. Chem. 264, 12520 –12525 (1989).

  7. 7.

    et al. Molecular microbiological investigation of an outbreak of hemolytic uremic syndrome caused by dry fermented sausage contaminated with Shiga-like toxin-producing Escherichia coli. J. Clin. Microbiol. 34, 1622–1627 ( 1996).

  8. 8.

    , , , & Molecular characterization of a Shiga-toxigenic Escherichia coli O113:H21 strain lacking eae responsible for a cluster of cases of hemolytic-uremic syndrome. J. Clin. Microbiol. 37, 3357–3361 (1999).

  9. 9.

    et al. Structure of an alpha-2,6-sialylated lipooligosaccharide from Neisseria meningitidis immunotype L1. Eur. J. Biochem. 254, 626–633 (1998).

  10. 10.

    & Lipooligosaccharides (LOS) of mucosal pathogens: molecular mimicry and host modification of LOS . Immunobiology 187, 382– 402 (1993).

  11. 11.

    et al. Structural analysis of the lipopolysaccharide oligosaccharide epitopes expressed by a capsule deficient strain of Haemophilus influenzae Rd. Eur. J. Biochem. 261, 171– 180 (1999).

  12. 12.

    , , , & Structure of the variable and conserved lipopolysaccharide oligosaccharide epitopes expressed by Haemophilus influenzae serotype b strain Eagan. Biochemistry 36, 2091– 2103 (1997).

  13. 13.

    Genetic locus for the biosynthesis of the variable portion of Neisseria gonorrhoeae lipopolysaccharide. J. Exp. Med. 180 , 2181–2190 (1994).

  14. 14.

    & Variation of gonococcal lipooligosaccharide structure is due to alterations in poly-G tracts in lgt genes encoding glycosyl transferases. J. Exp. Med. 183, 323–327 (1996).

  15. 15.

    , , , & Molecular analysis of a locus for the biosynthesis and phase-variable expression of the lacto-N-neotetraose terminal lipopolysaccharide structure in Neisseria meningitidis. Mol. Microbiol. 18, 729–740 (1995).

  16. 16.

    , , & The assembly system for the outer core portion of R1- and R4-type lipopolysaccharides of Escherichia coli. J. Biol. Chem. 273, 29497–29505 (1998).

  17. 17.

    , & Virulence of enterohemorrhagic Escherichia coli O91:H21 clinical isolates in an orally infected mouse model. Infect. Immun. 61, 3832–3842 (1993).

  18. 18.

    , & and Investigation of Shiga-like toxin binding to chemically synthesized oligosaccharide sequences. J. Infect. Dis. 164, 1160–1167 (1991).

  19. 19.

    , , , & In vitro assessment of a chemically synthesized Shiga toxin receptor analog attached to chromosorb P (Synsorb Pk) as a specific absorbing agent of Shiga toxin 1 and 2. Microbiol. Immunol. 43, 331–337 (1999).

  20. 20.

    , & Abstract V149/VII, 114 (3rd International Symposium and Workshop on Shiga Toxin (Verotoxin)-producing Escherichia coli Infections, Baltimore, Maryland, June, 1997). (Lois Joy Galler Foundation for Hemolytic Uremic Syndrome, Melville, New York, 1997).

  21. 21.

    et al. A phase I study of chemically synthesized verotoxin (Shiga-like toxin) Pk-trisaccharide receptors attached to chromosorb for preventing hemolytic-uremic syndrome. J. Infect. Dis. 171, 1042– 1045 (1995).

  22. 22.

    & Production of recombinant subunit vaccines: protein immunogens, live delivery systems and nucleic acid vaccines. J. Biotech. 73, 1– 33 (1999).

  23. 23.

    , & Adjuvant activity of Escherichia coli heat-labile enterotoxin and effect on the induction of oral tolerance in mice to unrelated protein antigens. Vaccine 6, 269– 277 (1988).

  24. 24.

    , & Molecular mimicry of host structures by bacterial lipopolysaccharides and its contribution to disease. FEMS Immunol. Med. Microbiol. 16, 105–115 (1996).

  25. 25.

    , & in Molecular cloning: A Laboratory Manual p. 68 (Cold Spring Harbor Laboratory, Cold Spring Harbor, New York, 1982).

  26. 26.

    & Procedure for isolation of bacterial lipopolysaccharides from both smooth and rough Pseudomonas aeruginosa and Salmonella typhimurium strains. J. Bacteriol. 155, 831–838 (1983).

  27. 27.

    , , , & Effect of lipopolysaccharide core synthesis mutations on the production of Vibrio cholerae O-antigen in Escherichia coli K-12. FEMS Microbiol. Lett. 663, 279 –285 (1991).

  28. 28.

    , & Extracellular transport of cholera toxin B subunit using Neisseria IgA protease beta-domain: conformation-dependent outer membrane translocation. EMBO J. 9, 1991– 1999 (1990).

  29. 29.

    , & Mouse model for colonization and disease caused by enterohemorrhagic Escherichia coli O157:H7. Infect. Immun. 58, 2438–2445 ( 1990).

  30. 30.

    , , , & Acute renal tubular necrosis and death of mice orally infected with Escherichia coli strains that produce Shiga-like toxin type II. Infect. Immun. 58, 3959– 3965 (1990).

  31. 31.

    et al. Shiga-like toxin-converting phages from Escherichia coli strains that cause hemorrhagic colitis or infantile diarrhea. Science 226, 694–696 ( 1984).

  32. 32.

    , & Improved M13 phage cloning vectors and host strains: nucleotide sequences of the M13 mp18 and pUC19 vectors. Gene 33, 103–119 (1985).

  33. 33.

    , & Polymerase chain reaction amplification, cloning and sequencing of variant Escherichia coli Shiga-like toxin type II operons . Microb. Pathog. 15, 77– 82 (1993).

  34. 34.

    , & Heterogeneity of the amino-acid sequences of Escherichia coli Shiga-like toxin type-I operons. Gene 153, 71–74 (1995).

  35. 35.

    , , , & Cloning and nucleotide sequencing of Vero toxin 2 variant genes from E. coli O91:H21 isolated from a patient with HUS. Microb. Pathogen. 8, 47–60 (1990).

  36. 36.

    & Construction of vectors with the p15a replicon, kanamycin resistance, inducible lacZαand pUC18 or pUC19 multiple cloning sites. Nucleic Acids Res. 18, 5315–5316 (1990).

  37. 37.

    et al. The structures of the carbohydrate backbones of the lipopolysaccharides from Escherichia coli rough mutants F470 (R1 core type) and F576 (R2 coretype). Eur. J. Biochem. 261, 629– 639 (1999).

Download references


The assistance of L. van den Bosch and E. Parker is acknowledged. This work was supported by grants from the National Health and Medical Research Council of Australia, and the Channel Seven Children's Research Foundation.

Author information


  1. Molecular Microbiology Unit, Women's and Children's Hospital, North Adelaide, S.A., 5006, Australia

    • Adrienne W. Paton
    •  & James C. Paton
  2. Department of Microbiology and Immunology, University of Adelaide, Adelaide, S.A., 5005, Australia

    • Renato Morona


  1. Search for Adrienne W. Paton in:

  2. Search for Renato Morona in:

  3. Search for James C. Paton in:

Corresponding author

Correspondence to James C. Paton.

About this article

Publication history





Further reading