WO2001034190A2

WO2001034190A2 - Live bacterial vaccines against escherichia coli o157:h7

Info

Publication number: WO2001034190A2
Application number: PCT/CA2000/001321
Authority: WO
Inventors: Malcolm B. Perry; J. Wayne Conlan
Original assignee: National Research Council Of Canada
Priority date: 1999-11-10
Filing date: 2000-11-10
Publication date: 2001-05-17
Also published as: CA2390304A1; AU1373901A; WO2001034190A3; EP1227839A2

Abstract

Disclosed are novel live bacterial vaccines against E. coli O157:H7, to treat or prevent colonization of the gastrointestinal tract of a vertebrate by the pathogen. The vaccines comprise an effective amount of non-pathogenic bacteria naturally expressing the O157 antigen or a structural mimic thereof as a part of their lipopolysaccharide. In a preferred embodiment, the non-pathogenic bacteria are selected from bacterial strains of the genus Salmonella or Citrobacter. The vaccines of the invention are particularly useful in maintaining cattle herds free of E. coli O157:H7 and in reducing carriage and fecal shedding of E. coli O157:H7 prior to slaughter, thus potentially reducing the clin ical incidence of E. coli O157:H7infections in humans.

Description

LIVE BACTERIAL VACCINES AGAINST ESCHERICHIA COLI 0157:H7

FIELD OF THE INVENTION

The invention relates to live bacterial vaccines against pathogenic strains of enteric bacteria, in particular to novel Salmonella- and C/Yrobac-er-based live vaccines against Escherichia coli strain 0157:H7, for preventing, reducing or eliminating the colonization of the gastrointestinal tract of cattle by E. coli 0157:H7.

BACKGROUND OF THE INVENTION

Escherichia coli strain 0157:H7 is an enteric bacterial pathogen that can cause severe local and systemic disease in susceptible humans, especially young children and the elderly (reviewed in: Su et al, Ann. Intern. Med., 123: 698 - 714 (1995)). Unusually for an enteric pathogen, the organism can cause severe infection even when ingested in very small numbers (Neill, M.A., Curr. Opin. Infect. Dis., 7: 295 - 303 (1994)).

Cattle are considered to be a major reservoir for this organism, which appears to be widely disseminated amongst North American cattle herds. Cattle can harbour E. coli 0157:H7 in their intestinal tracts asymptomatically, and they can shed it in their feces for prolonged periods (Hancock et al., in: Escherichia coli 0157:H7 and other shiga toxin-producing E. coli, pp. 85 - 91 , Kaper, J.B. and O'Brien, A.D. (editors), American Society for Microbiology, Washington D.C. (1998)). Fecal contamination of meat during slaughter, and the use of contaminated feces as fertilizer are two ways by which this organism can enter the human food supply. Contamination of the potable water supply by contaminated farm yard run-off is another major way by which this pathogen disseminates from cattle to humans. In humans, ingested E. coli 0157:H7 attaches to the intestinal mucosa wherein it triggers an inflammatory response that can develop into a severe, sometimes fatal hemorrhagic colitis (Su et al, supra). In addition, this pathogen elaborates toxins, including shiga-like toxins and endotoxin (Wadolkowski et al, Infect. Immun., 58: 3959 - 3965 (1990); Fujii et al, Infect. Immun., 62: 3447 - 3453 (1994); Karpman et al, J. Infect. Dis., 175: 611 - 620 (1997)), which can disseminate from the gut to cause systemic pathology including hemolytic uremic syndrome (HUS) which may progress to kidney failure and death.

Some epidemiological studies indicate that antibiotic therapy might promote E. coli 0157: H7-associated disease (reviewed in: Neill, M.A., in: Escherichia coli 0157:H7 and other shiga toxin-producing E. coli, pp 357 - 363, Kaper, J.B. and O'Brien, A.D. (editors), American Society for Microbiology, Washington D.C. (1998)). Because of this possibility, other measures to counter this pathogen are being examined. In this regard, preventing or reducing the level of carriage of this organism in cattle could help reduce the clinical incidence of E. co//^' 0157:H7 infection.

Presently explored potential methods to reduce bovine carriage of this organism include feeding calves with various normal flora organisms that inhibit the growth of E. coli 0157:H7 (Zhao et al, J. Clin, Microbiol., 36: 641 - 647 (1998); see also US 5,965,128), changing cattle diet from grain to hay to prevent the development of gastrointestinal acidity which favours the growth of acid resistant E. coli 0157:H7 (Diez-Gonzalez et al, Science, 281 : 1666 - 1668 (1998)), and vaccination directed to 0157 antigen (Konadu et al, Infect. Immun. 62: 5048 - 5054 (1994); Konadu et al, J. Infect. Dis., 177: 383 - 387 (1998)).

Vaccination of cattle to prevent or abort colonization with E. coli 0157:H7 is another possible approach to achieving this aim. E. coli 0157:H7 O-antigen is considered a good vaccination target because it is abundantly expressed and exposed at the bacterial surface. Furthermore, others (Konadu et al, supra (1994); Konadu et al, supra (1998)) have demonstrated that antibodies directed against this antigen can kill E. coli 0157:H7 in vitro in a complement- dependent manner. Additionally, there are several precedents showing that antibodies directed against lipopolysaccharides of other enteric bacterial pathogens afford protection against the respective organisms in vivo (Michetti et al, Infect. Immun., 60: 1786 - 1792 (1992); Winner et al, Infect. Immun., 59: 977 - 982 (1991 )). Finally, because the 0157 antigen is essentially specific for the pathogen, the possibility of immune elimination of potentially beneficial commensal strains of E. coli is curtailed.

The 0157 antigen was chemically identified as a polysaccharide composed of repeating tetrasaccharide units (Perry et al, Biochem. Cell Biol., 64: 21 - 28 (1985) having the structure:

4) - β- D - Glcp - (1 - 3) -α- D - GalpNAc - (1 - 2) - α -D -Rhap4Nac - (1 - 3) - α- L - Fucp - (1

and monoclonal antibodies directed to the antigen were made and used for the identification of the E. coli serotype 0157 strains (Perry et al, J. Clin. Microbiol., 26: 2391 - 2394 (1988); Todd et al, Appl. Environ. Microbiol, 54: 2536 - 2540 (1988)).

Following reports in the literature of the serological cross-reactions between E. coli 0157 polyclonal antisera and unrelated Gram-negative bacteria (Bettelheim et al, J. Clin. Microbiol, 31 : 760 - 761 (1993); Borczyk et al, Int. J. Food Microbiol., 4: 247 - 249 (1987); Perry et al, in: Advances in Brucellosis Research (Adams L. G., Ed.), pp. 76 - 88, Texas A&M University Press (1990)), structural analyses of the lipopolysaccharide (LPS) O-antigens from the cross-reactive bacteria were undertaken. The LPS O-antigens produced by Brucella abortus (Perry et al, supra; Bundle et al, Biochem. Soc. Trans., 13: 890 - 892 (1985); Caroff et al, Infect. Immun., 46: 384 - 388 (1984)), Brucella melitensis (Bundle et al, Biochemistry, 26: 8717 - 8726 (1987); Bundle et al, FEMS Microbiol. Lett., 126: 261 - 264 (1987)), Salmonella O:30 (group N) (Bundle et al, Can J. Chem., 64: 255 - 264 (1986); Perry et al, Carbohydr. Res., 156: 107 - 122 (1986)), Escherichia hermanii (Perry et al, Infect. Immun., 58: 1391 - 1395 (1990)), Pseudomonas maltophilia 555 (DiFabio et al, Biochem. Cell Biol., 65: 968 - 977 (1987)), Yersinia entercolitica 09 (Bundle et al, Infect. Immun., 46: 389 - 393 (1984); Caroff et al, Eur. J. Biochem., 139: 195 - 200 (1984)), and Vibrio cholerae 01 (Keene et al, Carbohydr. Res., 100: 341 - 349 (1982) revealed that a structural common epitope consisting of 2-substituted N-acyl derivatives of 4-amino-4,6- di-deoxy-K-D-mannopyranosyl residues were present in each respective LPS O-PS antigen, and was the factor responsible for the serological cross- reactions. Only Salmonella serogroup O:30 O-antigens have a structure identical with that of E. coli 0157 O-PS.

A paper by Bettelheim et al (supra) reported that a biochemically typical strain of C. freundii carried the E. coli 0157 antigen as indicated by serological reaction with specific E. coli 0157 antisera. Park et al (J. Clin. Microbiol, 30:1408 - 1409) reported that a strain of C. sedlakii also showed strong serological reaction with E. coli 0157 antisera. In the belief that the basis for the serological cross reactivity resides in the LPS O-PSs of the two Citrobacter strains, a structural characterization of the O-antigens produced by the bacteria was undertaken by the inventors, also in view of using such strains as live vaccines against E. coli 0157:H7.

It is widely believed that to elicit local immunity in the gut, vaccines must be delivered orally to the Peyer's patches (PP) and other intestinal lymphoid sites where inductive events that give rise to specific immunity can occur (reviewed in: McGhee et al, Int. J. Technological Assessment in Health Care, 10: 93 - 106 (1994)). However, it has proven notoriously difficult to elicit prolonged protective immunity against enteric pathogens using traditional nonviable vaccine preparations delivered per os. It is believed that this is in part due to the fact that the hydrolytic environment of the gut ensures that too little ingested nonviable antigen survives the journey to sites of immune induction. Many experimental solutions to this problem have been developed including encapsulation of antigen in liposomes or other coatings (see, for example, US 5,730,989) that can resist the corrosive environment of the gut, admixing of antigen with mucosal adjuvants such as cholera toxin, and expression of antigen in heterologous, viable, attenuated, enteric bacterial and viral carriers that naturally target the PP (reviewed in: Shalaby, W.S.W., Clin. Immunol. Immunopathol., 74: 127 - 134 (1995)). However, all such approaches still have many hurdles to overcome before becoming clinically feasible.

Overall, live vaccines suffer few of the immunological shortcomings associated with nonviable oral vaccines, but are currently out of favour with the regulatory authorities. In this regard, various Salmonella species, usually attenuated by selective mutagenesis or otherwise genetically engineered (see, for example, US 6,040,421 ; US 5,792,452; US 5,730,989; US 4,472,378; US 4,350,684; WO 00/04919; WO 98/02552; WO 96/38177; WO 90/08184; EP 564,689) have been used as carriers of a multitude of experimental vaccines against a wide range of pathogens (reviewed in: Curtiss, R., in: New Generation Vaccines, pp 161 - 168, Woodrow, G.C. and Levine, M.M. (editors), Marcel Dekker, New York (1990); Hackett, J., Vaccine, 8: 5 - 11 (1990)). Salmonellae are well suited for this purpose because being enteric pathogens themselves they naturally elicit a robust mucosal immunity.

The present invention provides a novel Salmonella-based live vaccine against Escherichia coli strain 0157:H7, for preventing, reducing or eliminating the colonization of the gastrointestinal tract of vertebrates, in particular cattle, by the pathogen. SUMMARY OF THE INVENTION

According to one aspect, the present invention provides a live vaccine for the immunization of a vertebrate against a first Gram-negative enteric bacteria, said vaccine comprising a live non-pathogenic second Gram-negative enteric bacteria naturally expressing an antigen of the first gram-negative enteric bacteria or a structural mimic thereof.

According to another aspect, the present invention provides a method for inducing in a vertebrate an immune response to a first Gram-negative enteric bacteria, said method comprising administering to said vertebrate an effective amount of a live, non-pathogenic second Gram-negative bacteria naturally expressing an antigen of the first Gram-negative enteric bacteria or a structural mimic thereof.

According to yet another aspect, the present invention provides a method for preventing or treating carriage of a first Gram-negative enteric bacteria by a vertebrate, said method comprising administering to said vertebrate an effective amount of a live, non-pathogenic second Gram-negative enteric bacteria naturally expressing an antigen of the first Gram-negative enteric bacteria or a structural mimic thereof.

In a preferred embodiment, the first Gram-negative enteric bacteria is a strain of Escherichia coli, in particular the strain E. coli 0157:H7, and the second Gram-negative enteric bacteria is a bacterial strain of the genus Salmonella or Citrobacter, in particular a wild strain of Salmonella landau, Citrobacter sedlakii or Citrobacter freundii naturally expressing the 0157:H7 antigen or a structural mimic thereof as part of its lipopolysaccharide. The vaccine and the methods of the present invention are particularly useful in maintaining cattle herds free of E. coli 0157:H7 and in reducing carriage and fecal shedding of E. coli 0157:H7 prior to slaughter, thus potentially reducing the clinical incidence of E. coli 0157:H7 infections in humans.

BRIEF DESCRIPTION OF THE DRAWINGS

Figure 1 is a graph showing fecal shedding period of S. landau by mice. Mice were gavaged with 2.84 x 10¹⁰ CFU (group A), 2.84 x 10⁹ CFU (group B), 2.84 x 10⁸ CFU (group C), or 2.84 x 10⁷ CFU (group D) of S. landau, and their feces were monitored for the presence of the organism.

Figure 2 is a graph showing levels of 0157-antigen-specific antibodies in serum and feces of mice following primary exposure to S. landau. Sera and feces were collected from a group of unexposed control mice (group E) and from the mice shown in Fig. 1 (groups A - D) on day 15 post gavage with S. landau, and serum IgG (O), serum IgA (□), and fecal IgA (Δ) antibody titres against 0157 antigen were determined by ELISA. Data points below the horizontal broken line depict negative samples (serum titre < 25; fecal titre < 4).

Figure 3 is a graph showing fecal shedding of S. landau by mice following secondary challenge. Mice shown in Fig. 1 were all regavaged with 8 x 10⁹ CFU of S. landau, and their feces were monitored for the presence of the organism.

Figure 4 is a graph showing levels of 0157-antigen-specific antibodies in the sera and feces of mice following secondary exposure to S. landau. Sera and feces were collected from mice shown in Fig. 3 on day 11 of secondary exposure to S. landau, and serum IgG (O), serum IgA (□), and fecal IgA (Δ) antibody titres against 0157 antigen were determined by ELISA. Data points below the horizontal broken line depict negative samples (serum titre < 25; fecal titre < 4).

Figure 5 is a graph showing fecal burdens of E. coli 0157:H7 in control mice and mice previously exposed to S. landau. Mice exposed twice by gavage to S. landau (groups A - C), and control mice (Group E) never exposed to this organism were all gavaged with 2.9 x 10¹⁰ CFU of a test isolate of E. coli 0157:H7. Fecal burdens of E. coli 0157:H7 were measured on days 2, 4, 7, 10, 16 post gavage. ^* indicates significantly smaller (P < 0.05) mean fecal load than group E controls by Mann Whitney Rank Sum Test; n = 5 or 6 mice per group.

Figure 6 is a graph showing fecal shedding of E. coli 0157:H7 by mice previously exposed to S. landau, and by unexposed control mice. Mice exposed twice by gavage to S. landau (groups A - C), and control mice (Group E) never exposed to this organism were all gavaged with 2.9 x 10¹⁰ CFU of a test isolate of E. coli 0157:H7, and the presence of the latter organism in the feces was monitored. The fecal shedding period was calculated as the last day on which E. coli 0157:H7 was recovered by culture from the feces. Fecal shedding periods were analysed statistically by log- rank test of survival curves.

Figure 7 is a graph showing levels of 0157-antigen-specific serum and fecal antibodies in control and vaccinated mice following challenge with E. coli 0157:H7. Sera and feces were collected from mice shown in Fig. 5 on day 8 post-exposure to a challenge gavage of 2.9 x 10¹⁰ CFU of an isolate of E. coli 0157:H7, and serum IgG (O), serum IgA (□), and fecal IgA (Δ) antibody titres against 0157 antigen were determined by ELISA. Data points below the horizontal broken line depict negative samples (serum titre < 25; fecal titre < 4). Figure 8 is a graph showing fecal loads of E. coli 0157:H7 in control mice and mice repeatedly exposed to S. landau. Two weeks following a fourth weekly exposure to S. landau, the mice depicted in Table 1 (□), and a group of control mice (O) were challenged by gavage with 3.94 x 10¹⁰ CFU of a test isolate of E. coli 0157:H7. Feces were collected from both groups on days 3, 7, 9, 14, 18 post challenge with E. coli 0157:H7, and were examined for the presence of this organism by bacteriological culture.

Figure 9 is a graph showing levels of anti-0157-antigen serum and coproantibodies in mice exposed to multiple inocula of S. landau. Sera and feces were collected from both the mice vaccinated with S. landau (group A), and the control mice (group B) depicted in Fig. 8, on day 8 and 18 post challenge with E. coli 0157:H7, and serum IgG (O), serum IgA (□), and fecal IgA (Δ) antibody titres against 0157 antigen were determined by ELISA. Data points below the horizontal broken line depict negative samples (serum titre < 25; fecal titre < 4).

Figure 10 is a graph showing the ¹H -¹³C-NMR correlation spectrum and 1 D ¹H-NMR spectrum of the LPS O-polysaccharide of Citrobacter sedlakii (NRCC 6070).

Figure 1 1 is a graph showing the ¹H -¹³C-NMR correlation spectrum and 1 D ¹H-NMR spectrum of the LPS O-polysaccharide of Citrobacter freundii (NRCC 6052).

Figure 12 is a graph showing the serological response of cattle to S. landau immunization. Groups of eight calves were vaccinated per os with either placebo or 10⁶ cfu (low), 10⁸ cfu (medium), and 10¹⁰ cfu (high) doses of S. landau on days 0 (Sep 01 ), 21 (Sep 22), and 42 (Oct 06), then challenged 14 days later (Oct 20) with 10⁸ cfu of a test isolate of E. coli 0157:H7. Animals were bled on each of these days to generate sera in which levels of antibody specific to S. landau LPS were determined by ELISA or titration.

Figure 13 is a graph showing the total number of days in which animals of each group of Fig. 12 shed detectable levels of E. coli 0157:H7, after being challenged with a 10⁸ cfu dose of E. coli 0157:H7. Shedding was monitored on a daily basis, between days 4 and 14 post challenge, using enrichment culture and direct planting of fecal samples on Rainbow Agar. The total number of eligible days was 88 per group.

Figure 14 is a graph showing the number of animals of each group of Fig. 12 shedding greater than 100,000 cfu of E. coli 0157:H7 per gram of feces between days 4 and 14 post challenge.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

As used herein, the term "non-pathogenic enteric bacteria" refers to bacteria which do not cause pathological conditions in an animal when administered to the animal in the amount necessary to elicit the desired immunological reaction. The term "structural mimic of an antigen" refers to an antigen which is not chemically identical with the antigen it mimics but which shows cross- reaction with a monoclonal antibody specific to the mimicked antigen.

Konadu et al, supra (1994) have developed a conjugate vaccine consisting of the O- specific carbohydrate moiety of E. coli 0157 lipopolysaccharide (LPS) coupled to a protein from Pseudomonas aeruginosa. This vaccine elicits systemic IgG and IgA antibodies to the 0157 antigen when administered parenterally to mice and humans (Konadu et al, supra (1994); Konadu et al, supra (1998)). In the presence of xenogeneic complement, these antibodies proved to be bactericidal for E. coli 0157:H7 in vitro. However, to be effective in vivo, these antibodies must be available at foci of E. coli 0157:H7 colonization in the intestinal mucosa, and it is believed that systemically- elicited antibodies will naturally transude to these sites. However, inventors' findings failed to support this hypothesis. Thus, inventors' studies showed that vaccinating mice parenterally with an 0157 antigen glycoconjugate, whilst capable of eliciting a systemic humoral immune response, did not result in the appearance of 0157-antigen-specific copro-antibodies (Conlan et al, Can. J. Microbiol, 45: 279-286). Nor did this vaccination strategy ameliorate colonization of the murine intestinal tract by a test isolate of E. coli 0157:H7. In the wake of this failure, inventors embarked on studies to determine whether a mucosal immune response to the 0157 antigen might afford protection from colonization and infection by E. coli 0157:H7.

The aforementioned result was not entirely unexpected since it is generally believed that to elicit local immunity in the gut, vaccines must be delivered orally to the Peyer's patches and other intestinal lymphoid sites where inductive events that give rise to specific immunity can occur. Subsequently, specific T cells and IgA secreting B cells generated in these sites migrate throughout the lamina propria and mucosa of the gastrointestinal tract to exert their antibacterial effects (reviewed in: James et al, Alimentary Pharmacology and Therapeutics, 10 (supplement 2): 1 - 8 (1996)). It is not straightforward to elicit this gastrointestinal immunity since antigens fed orally generally fail to stimulate an appropriate immune response. The immunogenicity of many orally fed antigens can be improved by various measures including admixing with cholera toxin which acts as a potent mucosal adjuvant (Jackson et al, Infect. Immun., 62: 3594 - 3597 (1993)). However, in the inventors' studies, oral vaccination of mice with 0157:H7-antigen glycoconjugate admixed with cholera toxin failed to elicit a local immune response to 0157 antigen, despite eliciting a robust systemic and local immune response to the toxin adjuvant (Conlan et al, Can. J. Microbiol, 46: 283-290.).

Live Salmonella-based vaccines have proven to be highly suitable for eliciting mucosal immune responses in the gut (Curtiss, supra (1990); Hackett, supra (1990)). Live Salmonella-based vaccines are now being used in the poultry industry, and it is likely that they will ultimately gain acceptance elsewhere in the veterinary and medical arena. Usually the Salmonella strains employed for this purpose are deliberately attenuated by selective mutagenesis of bacterial genes crucial for the expression of virulence. According to the present invention, inventors used a wild type Salmonella landau strain that naturally expresses the 0157 antigen as part of its lipopolysaccharide, because its rarity as a clinical entity suggests that innately it might not be particularly virulent. Moreover, the organism appeared to be avirulent for mice even when administered in high doses. However, other Salmonella species, most notably S. typhi, which are naturally avirulent for mice are extremely virulent for other mammalian hosts. Some evidence of enteroinvasion and systemic infection by S. landau was obtained in the present study, suggesting that the wild type strain might need attenuation before being used as a clinical vaccine. With these considerations in mind, the virulence of wild-type S. landau in cattle is presently being examined.

Mice gavaged with S. landau developed high titres of systemic and local antibodies to the 0157 antigen. These antibodies persisted for several months post-vaccination. Moreover, mice exposed to S. landau displayed some evidence of protection from colonization, measured as decreased fecal shedding, against a subsequent challenge with an isolate of E. coli 0157:H7. This protection could be due to immunity elicited against shared antigens other than the 0157 antigen. However, this seems unlikely given that the normal flora of mice includes benign strains of E. coli (Miller et al, J. Infect. Dis., 113: 59 - 60 (1963)) which might be expected to share greater non-O- antigen-based antigenicity with E. coli 0157:H7 than S. landau. In the present study, any such naturally-elicited background immunity to the former organism would be equally well expressed in control and test mice. This argument extends to cattle and humans too. Thus, any protective immunity against E. coli 0157:H7 elicited by S. landau most probably resides in the immune response to the 0157 antigen common to both organisms.

The protection results of this study might be a reflection of the inadequacies of the mouse model rather than of the vaccine used. For example, the E. coli 0157:H7 isolate used only colonizes the intestinal tracts of nonimmune mice for up to two weeks, making it difficult to assess the meaningfulness of any vaccine induced immunity that might reduce this period by only a day or so. By contrast, cattle can harbour and excrete the pathogen for several months (Hancock et al., supra). If vaccination could reduce this colonization period to only a week or so it would be considered a significant achievement. An additional caveat of the mouse model is that the inoculum size needed to ensure that mice become colonized is of the order of 10⁸ fold higher than that required to initiate severe infection in susceptible humans. Clearly, pre- existing mucosal immunity to the 0157 antigen is likely to prove to be more protective in the latter than the former situation. Thus, Salmonella landau based vaccination against E. coli 0157:H7 has a potential to be more successful in the field than in the laboratory. In that case, it would be also possible to engineer S. landau to express antigens of other exogenous mucosal pathogens, thereby creating a pluripotent vaccine.

The structural basis for the literature reported serological cross-reactions of E. coli 0157 specific O-antigen antibody with strains of C. freundii (Bettelheim et al, supra) and C. sedlakii (Park et al, supra) were found to reside in the O- polysaccharides of their respective LPS components.

In the case of C. sedlakii, its O-PS was revealed to be identical in structure with that of the O-PS of E. coli 0157:H7 and, as expected, its serological reactivity with homologous and heterologous antisera were essentially identical. It is interesting to note that the only previous report of structural identity of an E. coli 0157 O-antigen in another bacterial species was that in strains of Salmonella O:30 (Group N) (Bundle et al, supra; Perry et al, supra).

The LPS O-PS of C. freundii strain which showed strong cross-reactivity with E. coli 0157 O-antigen had its own unique structure differing from that of the E. coli 0157 antigen and surprisingly, unlike other investigated crossreacting LPS O-PS antigens, did not contain a 2-substituted 4-amino-4,6-dideoxy-D- mannopyranosyl residue which has been implicated as the epitope involved in previously examined Gram-negative bacteria. It is possible that the 2- substituted K-D-Rhap residue in the C. freundii O-PS structurally mimics the 2-substituted 4-acetamido-4,6-dideoxy-K-D-mannopyranosyl residue (K-D- Rhap4Nac residue) in the E. coli 0157 O-PS; however, modeling and oligosaccharide inhibition experiments will be required to establish this point.

Similarly as S. landau (0:30), which naturally expresses the E. coli 0157 O- PS structure, the strains of C. sedlakii and C. freundii naturally expressing the 0157 antigen or its structural mimic can be used as a live vaccine against E. coli 0157:H7 gut colonization.

In the following, the invention will be described in still greater detail, by way of examples and with respect to the preferred embodiments.

SALMONELLA LANDAU VACCINATION OF MICE

Fate of Salmonella landau inoculated per os.

In an initial experiment, groups of mice were inoculated with 2.8 x 10¹⁰ CFU of S. landau (group A) or serial ten-fold dilutions thereof (groups B - D). Monitoring showed that mice rapidly ceased shedding this organism in their feces at all test doses (Fig. 1 ). Serological testing of mice 11 days following initial gavage with S. landau showed that the majority of mice receiving the highest doses of the organism had developed detectable levels of 0157 antigen-specific copro-antibodies, but not circulating antibodies by this time (Fig. 2). By contrast, the group of mice (group D) that received the smallest inoculum of S. landau and a group of control mice (group E) that were not exposed to S. landau were uniformly negative for 0157 antigen-specific fecal and circulating antibodies at this time. None of the mice became ill following primary exposure to S. landau at any dose tested.

Next, groups A - D were rechallenged with 0.8 x 10¹⁰ CFU of S. landau. All mice had stopped shedding S. landau by 11 days of rechallenge (Fig. 3). Moreover, all but one of the mice that had been exposed to high inocula of S. landau initially (Group A and B in Fig. 1 ) had already stopped shedding the secondary inoculum of the organism by day 4 of rechallenge. This was a shorter fecal shedding period than observed for mice initially gavaged with a comparable inoculum (groups A and B in Fig. 1 ), suggesting that these mice acquired immunity against recolonization by S. landau as a result of initial exposure to a large bolus of the organism. One group of mice (group D) was killed on day 11 following secondary challenge with S. landau and their Peyer's patches, livers, and spleens were removed, homogenized, and plated on SBHI agar to check for evidence of enteroinvasion (Carter et al, J. Exp. /Wed., 139: 1189 - 1203 (1974)). Trace quantities of S. landau were found in the Peyer's patches (150 CFU/tissue) of one mouse and in the spleen of another (200 CFU/tissue). All other tissues examined were sterile. Again, no mice died as a result of re-exposure to S. landau. At 10 days post- rechallenge, 0157-specific IgA antibodies were found in the feces of all remaining mice (Fig. 4). Circulating IgA was detected in the sera of all but one mouse initially exposed to any test dose of S. landau. Circulating IgG was present in all group A - C mice, but absent from all but one mouse in group D that received the smallest initial inoculum of S. landau. The latter result was the only statistically significant difference (P<0.05) found among antibody titre comparisons from groups A to D following re-exposure to S. landau. Resistance of mice exposed to S. landau to colonization by E. coli O157.H7

On day 15 post re-exposure to S. landau, the aforementioned mice (groups A - C) and a group of control mice (group E) that had never been exposed to this organism were all challenged by gavage with an inoculum (2.9x10¹⁰ CFU/mouse) of a test isolate of E. coli 0157:H7. Subsequently, the feces of all mice were monitored for the presence of E. coli 0157:H7 by bacteriology coupled with confirmatory slide agglutination test of random colonies. The results of this examination are shown by Fig 5. Compared to control mice (group E), all groups (group A - C) of mice previously exposed to S. landau exhibited lower mean fecal loads of E. coli 0157:H7 on days 7 and 10 of challenge that in most instances were statistically significant (P<0.05). Similarly, compared to control (group E) mice, the mean period over which E. coli 0157:H7 was detectable in the feces was reduced in all groups (groups A - C) of mice previously exposed to S. landau (Fig. 6). However, this decrease was only statistically significant in the case of group C versus group E, implying that none of the observed reductions in fecal shedding period among the test groups were biologically meaningful.

Eight days after gavage with E. coli 0157:H7, serum and copro-antibody levels against 0157 antigen were determined, and the results are shown in Fig. 7. All mice previously exposed to S. landau (groups A - C) had circulating anti-0157 antigen IgA and IgG antibodies, whereas the control group (group E) had no detectable systemic antibodies to this antigen. In contrast, most group E mice had detectable levels of fecal IgA against the 0157 antigen, however the mean level of these was significantly lower (P<0.05) than those found in the feces of mice in groups A - C. No 0157-antigen specific IgG was detected in the feces of any mouse (data not shown). Immune status of mice following multiple exposures to S. landau.

In a complementary experiment, a group of mice was gavaged a total of four times with approximately 2.5 x 10¹⁰ CFU of S. landau at weekly intervals. The fecal burdens of S. landau were assessed three days following each gavage and the results are shown in Table 1.

Table 1. Fecal shedding of S. landau by mice following each of four successive challenges by gavage with the pathogen.

No. of mice Log₁₀ mean ± SD

Exposure to Inoculum size shedding S. CFU S. landau I S. landau / mouse landau in feces 100 mg feces for on day 3 post shedding mice exposure

1st 2.45 x 10¹⁰ 5/5 3.35 ± 0.29 2nd 2.72 x10¹⁰ 3/5 2.40 ± 0.16 3rd 2.11 x 10¹⁰ 1/5 2.31 4th 3.89 x 10¹⁰ 0/5

It shows that the bacterial burden as measured by the proportion of mice shedding the organism in feces, fell following each exposure implying that the initial exposures to S. landau generated protective immunity against colonization by subsequent inocula of the organism. Mice developed 0157- specific IgA copro-antibodies following primary exposure to S. landau, and this titre rose significantly (p<0.05) following secondary exposure, but not in response to subsequent exposures (Table 2).

Table 2. Anti-0157 antigen IgA levels in feces following sequential exposure by gavage to inocula of S. landau

Fecal anti-0157 antigen IgA titre ± SD following exposure to S. landau^

1st exposure 2nd exposure 3rd exposure 4th exposure

7.7 ± 6.6^* 218 ± 115 241 ± 119 188 ± 50

^f feces were collected 3 days following each gavage from 5 mice. ^* Titre significantly lower than those obtained from other examination times.

Additionally, mice had developed substantial systemic IgA (430 ± 109, n=5) and IgG (1140 ± 585, n=5) titres against the 0157 antigen by 9 days following the final exposure to S. landau. Control mice exhibited no 0157-specific antibodies at any period of examination.

Two weeks following the final exposure to S. landau the aforementioned mice, and a group of five control mice that had never been exposed to this organism, were challenged by gavage with an inoculum of 3.94 x 10¹⁰ CFU/mouse of E. coli 0157:H7. Subsequently, the fecal burden of E. coli 0157:H7 in both groups of mice was monitored (Fig. 8). At each period examined the mean fecal burden was slightly lower in group A versus group B mice, however all these differences were statistically insignificant. There was no significant difference either in the mean fecal shedding period between the two groups.

Sera and feces from both groups of mice were collected on days 8 and 18 following exposure to E. coli 0157:H7, and were examined for the presence of 0157-specific antibodies (Fig. 9). As expected, serum and fecal antibody levels remained high in mice previously exposed to S. landau. No fecal 0157-antigen specific IgG was detected in these mice. Serum and fecal antibody titres were similar whether whole 0157 LPS or delipidified 0157 antigen was used as ELISA antigen (not shown). By contrast, none of the sera or fecal samples reacted when E. coli O70 LPS was used as ELISA antigen (not shown). Control mice had not developed a detectable humoral immune response to 0157 antigen by day 8 post gavage with E. coli 0157:H7. The latter mice had developed a weak fecal IgA anti-0157 antigen titre by day 18 post gavage. However, these control mice essentially failed to mount a systemic antibody response to this antigen.

Long-term immune status of mice following multiple exposures to S. landau

In the experiment detailed in the previous section, a group of ten mice exposed four times to S. landau were left unchallenged with E. coli 0157:H7. At 5 week intervals following the last S. landau exposure, sera and feces from these mice, and a group of five control mice were examined for the presence of 0157-antigen specific antibodies. Additional feces collected at these times were cultured for the presence of S. landau. The results summarized in Table 3 show that 0157 antigen specific antibodies persisted for at least 15 weeks following the final exposure to S. landau. No 0157-specific antibodies were detected in the feces or sera of control mice. No S. landau -like colonies were recovered from the feces of any mice at any of these extended time points.

Table 3. Persistence of anti-0157 antibodies in mice following exposure to S. landau.

week post final S. Anti-0157- •antigen antibody titre ± SD^* landau exposure

Serum IgG Serum IgA Fecal IgA

5 3789 ± 2647 357 ± 126 169 ± 82

10 2626 ± 2368 301 ± 110 125 ± 62

15 3751 ± 4082 254 ± 77 189 ± 129 ^*n= 5 mice per group. ^f mice received 4 inocula of S. landau as detailed in Table 1.

SEROLOGICAL CROSS-REACTIONS OF C. SEDLAKII AND C. FREUNDII

Slide agglutination tests using 0157-specific monoclonal antibodies confirmed the previously reported (Bettelheim et al, supra; Park et al, supra) serological cross reactivities between E. coli 0157:H7 and C. freundii (NRCC 6052) and C. sedlakii whole cells. Antisera specific for the 0157 antigen obtained from mice orally challenged with viable E. coli 0157:H7 (Conlan et al, supra) reacted with high titre against C. freundii LPS as coating antigen in ELISAs. The same antisera uniformly failed to react with E. coli O70 LPS. However, when these LPS preparations or their respective O-PS derivatives were used as coating antigens in an ELISA assay, one 0157-specifc Mab reacted with substantially higher titre against the homologous versus the heterologous polysaccharide (data not shown). Moreover, in immunodiffusion assays, O-PS prepared from E. coli 0157:H7 and C. freundii run in adjacent wells gave precipitin lines of only partial identity against an 0157-specific Mab placed in the central well of the assay plate. Together, these results imply that the 0157 antigen and the O-PS of the C. freundii are not identical.

Mild acid hydrolysis of the LPSs produced by E. coli 0157:H7, C. freundii (NRCC 6052) and C. sedlakii (NRCC 6070) gave an insoluble lipid A (12%) and water-soluble products which on Sephadex G-50 column chromatography yielded an O-PS fractions (K_av 0.03 - 0.15, 80 - 83%), core oligosaccharide fraction (K_av 0.56, 4 - 6%), and a fraction (K_av 0.98, 12%) containing KDO and phosphate.

The LPS prepared from E. coli 0157:H7 was chemically and serologically identical to the preparation described previously (Perry et al, supra). It afforded an O-PS having [α]_D +41.0° (c 0.2, water) and gave H- and ¹³C-NMR spectra identical and in agreement with published data (Perry et al, supra).

The O-PS from the LPS of the C. sedlakii strain had [α]_D +35.2° (c 1.3, water), gave chemical analysis composition and structural data identical with the data from E. coli 0157 O-PS. Furthermore, ¹H- and ¹³C-NMR analysis of the O-PS gave spectra (Fig. 10) consistent with published data and indistinguishable from those given by the E. coli 0157 O-PS.

The C. freundii O-PS had [α]_D +15.5° (c 1.6, water) and preparative paper chromatographic separation of the O-PS hydrolysate (4 M TFA, 105°C, 4 h) gave only two glycoses which were identified as D-glucose and D-rhamnose (1 :2). Methylation analysis by GLC-MS showed the hydrolysis products of the methylated O-PS to be 3,4-di-O-methyl-D-rhamnose, 2,4-di-O-methyl-D- rhamnose and 2,3,6-tri-O-methyl-D-glucose (1 :1 :1 ) thus identifying the structural units -2)-D-Rhap-(1-, -3)-D-Rhap-(1-, and -4)-D-Glcp-(1- in the O- PS. The linkage sequence and anomeric configurations of the glycose units present in the C. freundii O-PS trisaccharide repeating unit were determined by NMR spectroscopy. The 1 D ¹H- and ¹³C-NMR spectra of the O-PS were consistent with a trisaccharide repeating unit. Two-dimensional NMR spectra 2D correlation spectroscopy (DQF COSY), total correlation spectroscopy (TOCSY), nuclear Overhauser effect spectroscopy (NOESY), and H-detected heteronuclear H, ¹³C single-quantum coherence (HSQC) (Fig. 11 ), were used for the assignment of proton and carbon signals as recorded in Table 4.

Table 4. ¹H and ¹³C-NMR assignments for the O-PS of the LPS of Citrobacter freundii

¹H Chemical shifts

Glycose residue H-1 H-2 H-3 H-4 H-5 H-6

(A) -2)-K-D-Rhap-(1- 5.14 4.22 3.87 3.44 3.86 1.27

(B) -3)-L-D-Rhap-(1- 4.70 4.10 3.68 3.46 3.45 1.30

(C) -c4)-L-D-Glcp-(1- 4.50 3.36 3.64 3.63 3.53 3.85

¹³C Chemical shifts

Glycose residue C-1 C-2 C-3 C-4 C-5 C-6

(A) -2)-K-D-Rhap-(1-c 101.6 79.0 70.7 73.6 70.7 18.0

(B) -3)-L-D-Rhap-(1- 01.1 72.2 81.8 72 5 73.6 18.0

(C) -c4)-L-D-Glcp-(1-c 102.9 73.8 75.4 80.2 75.9 61.6

The three O-PS component glycose units were given the designations A, B, and C in order of decreasing anomeric proton chemical shifts (Fig. 10). The two component D-Rhap residues (A and B) were identified on the basis of vicinal proton coupling constants and the D-Glcp residue (C) anomeric configuration was determined as β- from its J_1>2 coupling constant (8 Hz). The anomeric configuration of the D-Rhap residues followed from the position of their C-5 resonances: the high field shift for residue A (70.7 ppm) indicated an -configuration, whereas the C-5 of residueB gave a signal at 73.6 ppm characteristic for a β-D-configuration. The β-D-configu ration of residue B was confirmed from the observation of intraresidue NOEs between H-1 B and H-3B and H-5B. The sequence of the residues was determined from NOE data: correlations were observed between proton H-1 A and H-3B, H-1 B and H-4C, and H-1 C and H-1 A and H-2A, thus indicating a residue linkage order → A → B → C -» and the structure of the O-PS to be an unbranched linear polymer of the repeating unit:

- 2) - α- D - Rha p^A - (1 - 3) - β- D - Rha p^B - (1 - 4) - β- D - Glc p^c - (1 -

A 13 C-NMR spectrum of the Citrobacter O-PS was simulated on a computer (Lipkind et al, Carbohydr. Res., 175: 59 - 75 (1988)) assuming L- or D- configurations of both Rhap residues and the D-configuration for the Glcp residue. It was found that spectrum of the C. freundii O-PS, as expected, matches the calculated spectrum for D-Rhap constituent residues and not that of the one calculated for L-Rhap residues.

In keeping with the chemical analysis providing unequivocal evidence for the distinct O-PS structures for E. coli 0157:H7 and C. freundii, it was found that polyclonal antisera raised against the LPS of the latter continued to react strongly with C. freundii O-PS by ELISA even after cross-reacting antibodies to the O-PS of the former had been completely eliminated by extensive absorption against formalin-killed E. coli 0157:H7 bacilli. Moreover, by ELISA, this absorbed antiserum reacted less strongly against C. sedlakii O-PS than C. freundii as coating antigen, indicating that it contained predominantly C. freundii O-PS specific antibodies rather than common C/^'-robacter-specific anti-core oligosaccharide antibodies.

SALMONELLA LANDAU VACCINATION OF CATTLE

Seroconversion

Four groups of eight calves were vaccinated per os with either placebo or low (10⁶ cfu), medium (10⁸ cfu) or high (10¹⁰ cfu) doses of S. landau on days 0, 21 , and 42, then challenged 14 days later with approximately 10⁸ cfu of a test isolate of E. coli 0157:H7. Animals were bled on days 0, 21 , 42, and 56, and the serological response to immunization was determined in collected sera using an ELISA procedure with purified S. landau LPS as the coating antigen, or by titration. The numbers of animals which seroconverted in each group are shown in Table 5.

Table 5. Seroconversion following immunization of cattle

Vaccine Group Number Positive (n=8)

Placebo 1 Low Dose (10⁶ cfu) 0 Medium Dose (10⁸ cfu) 7 High Dose (10 ⁰ cfu) 7

Fig. 12 shows the serological response of animals to immunization, as a function of time and the vaccine dose. The levels of antibodies developed as a result of immunization were measured by titration and are shown as a mean optical density of a 100-fold serum dilution. As would normally be expected, the serological response is clearly stronger in animals of the medium and high dose vaccine groups. As no pathological conditions were observed in the immunized animals, these results indicate that cattle can tolerate even multiple high doses of S. landau applied for immunization purposes. It should be noted that the observed seroconversion was used only as a convenient marker for immunogenecity, as there is no known correlation between the level of the circulating antibody and the protective immunity against E. coli 0157:H7.

Shedding E. coli 0157:1-17 following challenge After all animals were challenged with doses of approximately 10⁸ cfu of E. coli 0157:H7, shedding of the latter was monitored on a daily basis using enrichment culture and direct plating of fecal samples on Rainbow agar. The shedding data during days 0 - 3 was typically similar in all groups, as during this time the challenge dose was cleared by animals. Consequently, data were collected only for days 4 - 14. All animals were culture-negative on day 0. Between 62.5% and 87.5% shed detectable levels of E. coli 0157:H7 during the trial, typical of what is normally observed.

The total number of days in which animals shed detectable levels of E. coli 0157:H7 between days 4 and 14 is shown for each vaccination group in Table 6 and illustrated in Fig. 13. The total number of eligible days was 88 days per group. Table 6. Total number of days per group in which animals shed E. coli 0157:H7 during days 4 - 14 post challenge

Vaccine Group Total Days Shedding Per Group

Placebo 24 Low Dose 18 Medium Dose 16 High Dose 14

This data shows a trend towards a decreased shedding with increasing S. landau vaccine dose, as reflected by a lower number of total days shedding.

The numbers of animals shedding greater than 100,000 cfu per gram of feces are shown for each group in Fig. 14. Also this data also shows that there is a trend towards decreased shedding in the medium and high dose vaccine groups, as reflected by fewer days shedding grater than 100,000 cfu per gram of feces.

EXPERIMENTAL - MATERIALS AND METHODS

Bacteria

Salmonella landau (LCDC Strain # S-1358) originated as a clinical isolate. It was obtained in a frozen state from a culture collection maintained at the

Institute for Biological Sciences (Ottawa, Canada) by D. Griffith. It was revived by thawing and plating on brain heart infusion (BHl) agar. Large colonies grew following overnight incubation at 37°C. The identity of the revived organism was confirmed by Gram's stain, and by slide agglutination test using a monoclonal antibody specific for the 0157 antigen (Perry et al,

Biochem. Cell Biol., 64: 255 - 264 (1986)). A streptomycin-resistant variant was obtained by natural selection by re-plating heavy inocula of the revived organism onto BHl agar containing 50 g/ml streptomycin sulphate (Sigma

Chemical Co.). A single streptomycin-resistant colony, identified as S. landau as above, was used to seed a flask of BHl broth containing 50 g/ml streptomycin sulphate (SBHI broth). This culture was incubated at 37°C with shaking for 5 h, then sterile glycerol was added to a final concentration of 20% (v/v) and dispersed by shaking. This stock culture was then dispensed in 1.5 ml aliquots and frozen at -80°C until required. A stock of a streptomycin-resistant isolate of Escherichia coli 0157:H7 was similarly prepared, as described previously (Conlan et al, Can. J. Microbiol, 44: 800 - 805 (1998)). Upon ingestion, this isolate of E. coli is capable of colonizing the intestinal tract of Balb/c mice as evidenced by fecal shedding of the pathogen for periods of up to two weeks (Conlan et al, supra; Conlan et al, Can. J. Microbiol, in press (1999)).

For the study of serological cross-reactions and chemical structure of antigens, E. coli 0157:H7 (NRCC 4125) was obtained from the NRC culture collection, C. freundii (CDC 3488-90, NRCC 6052) kindly supplied by Dr. N.A. Strockbine and C. sedlakii (NRCC 6070) was kindly supplied by Dr. CH. Park. The bacteria were cultivated in 3.7% brain-heart infusion broth (Difco) for 18 h at 37°C in a 75-I fermenter with stirring and air aeration and killed cells (2% phenol) were harvested (yield ca. 450 g wet weight each).

Mice

Female Balb/c mice were obtained from Charles Rivers Laboratories (St.- Constant, Quebec) when they were 7 to 9 weeks old. Groups of 5 mice, including appropriate control groups were used throughout. Mice were maintained and used in accordance with the recommendations of the current edition of the Canadian Council on Animal Care Guide to the Care and Use of Experimental Animals (1993). Mice entered experiments within two weeks of receipt. Mice were inoculated by gavage with freshly grown cultures of S. landau or E. coli 0157:H7. Briefly, for each experiment, vials of frozen bacteria were thawed and the entire contents of each vial used to inoculate a flask containing 200 ml fresh SBHI broth. Flasks were incubated as above to an absorbance at 600 nm of approximately 1.1. Next, the cultures were harvested by centrifugation at 7200 g at ambient temperature and resuspended in phosphate buffered saline (PBS) to a concentration of approximately 10" CFU/ml for inoculation into mice. In each case, the viability of the inoculum was determined by plating serial dilutions of it on SBHI agar. Mice received approximately 2 x 10¹⁰ CFU of either organism in a volume of 0.2 ml delivered using a 1 ml tuberculin syringe fitted with a 20 g gavage needle. Mice were observed to ensure that they did not regurgitate or aspirate the inoculum.

At various intervals post-challenge, fresh fecal pellets expressed during a 15 min. collection time were obtained from individual animals and transferred into pre-weighed tubes of sterile PBS. Feces were thoroughly resuspended by vortex mixing and thturation using a 1 ml serological pipette, then, serially diluted in PBS and plated on SBHI agar. Plates were incubated overnight at 37°C and Salmonella- like or E. coli- like colonies were counted. In each case the identity of random colonies was confirmed by slide agglutination test. The SBHI medium used in the present study severely inhibited the growth of much of the autochthonous microflora. Moreover, the streptomycin-resistant normal flora that did grow during the incubation period employed, gave rise to small colonies that were readily distinguishable from the much larger colonies of S. landau or E. coli 0157:H7. In this regard, it should be noted that no streptomycin-resistant Salmonella- like or E. coli - like colonies were ever recovered from the feces of any of the mice used in this study prior to the deliberate inoculation of the test organisms.

Bacterial burdens per unit weight of feces were compared among various groups of mice using the Mann Whitney Rank Sum Test to assess statistically significant (P<0.05) differences. Fecal shedding periods between groups were also compared for statistically significant differences. For individual mice, the fecal shedding period was calculated as the last day of examination on which the pathogen was cultured from the feces. For each group, fecal shedding periods were plotted as survival curves which were then compared for statistical differences by log-rank test.

Vaccination

For vaccination studies, mice were inoculated with two doses of S. landau administered 14 days apart, or four doses administered 7 days apart. Two weeks after the final exposure to S. landau, mice were challenged with the test isolate of E. coli 0157:H7. Feces were monitored for the presence of either pathogen as detailed above. Additionally, sera and feces were collected at intervals for antibody determinations. Sera were prepared from venous blood obtained from a lateral tail vein. For copro-antibody determinations, freshly expressed fecal pellets were homogenized in PBS containing 0.05 % w/v sodium azide and 10 % v/v fetal calf serum as exogenous protein to quench proteolytic degradation of specific antibody (Elson et al, J. Immunol. Methods, 67: 101 - 108 (1984)). Each sample consisted of two fecal pellets homogenized in 1 ml of this diluent. Fecal homogenates were clarified by centrifugation at high speed in a microcentrifuge, and the supematants were stored frozen at -70°C until needed.

Antibody determinations

Sera and feces were screened for the presence of specific IgA and IgG isotype antibodies by ELISA using purified E. coli 0157 LPS, or delipidified 0157 LPS, or E. coli O70 LPS as antigen. Briefly, microtitre plates (Dynatech Immunlon II) were coated with antigen (100 μ/well), diluted to 10 μg/ml in carbonate buffer by incubating them overnight at 4°C Excess antigen was removed by washing the wells three times with PBS containing 0.05 % v/v Tween 20 using an automated plate washer. Next, sera at a starting dilution of 1/40, and feces at a starting dilution of 1/4 were titrated through a two-fold dilution series down a column of 8 wells. Plates were then incubated for 2 h at ambient temperature, then washed three more times as above. Next, goat- anti-mouse IgG or goat-anti-mouse IgA conjugated to alkaline phosphatase (Caltag Laboratories) at a dilution of 1 :2000 in antibody diluent was added at 100 μ/well and incubation continued for a further 2 h. After this time, plates were re-washed and 100 | of p-nitrophenylphosphate substrate in diethanolamine buffer (phosphatase substrate kit, Kirkegaard & Perry Laboratories) was added to each well. The yellow colour that developed was read at 410 nm using a microplate reader. Titres were determined from plots of absorbance at 410nm versus dilution and were defined as the reciprocal of the dilution giving an A_4W equivalent to 0.25. Because in previous studies (Conlan et al, supra (1998); Conlan et al, supra (1999)) 0157 antigen-specific IgG was never detected in the feces, it was not routinely assayed for in the feces in the present study. Instead, only selected fecal samples were assayed for the presence of anti-0157 antigen IgG. Standard negative and positive control sera or fecal supernatants identified by a preliminary ELISA of candidate samples were included on each plate. Titres were analysed statistically by Mann Whitney Rank Sum Test and were considered to be significantly different to comparative samples when P values <0.05 were obtained.

LPS and LPS P-PS preparations

Bacterial cells ( V400 g wet weight) were extracted with 50% aqueous phenol as described by Johnson and Perry (Can. J. Microbiol, 22: 29 - 34 (1976)) and the water diluted (2 vols). Dialyzed solutions were subjected to ultracentrifugation (105 000 x g, 10 h, at 4°C) to yield precipitated LPS gels which were lyophilised from water solutions (yields 4.7 - 5.2 g). LPSs (1 g) dissolved in 2% (v/v) acetic acid (150 ml) were heated in a boiling water bath (2 h), precipitated lipid A (ca. 80 mg) was removed from the cooled solutions by centrifugation, and the water-soluble products were lyophilised, dissolved in pyridium acetate buffer (0.05 M, pH 4.8), and were fractionated by Sephadex G-50 gel filtration chromatography (column size, 2 x 100 cm). The eluates were monitored by refractive index, and the O-PS fractions eluting at the void volume of the column were collected and lyophilised (yield 370 - 420 mg).

Chemical analysis Gas-liquid chromatography (GLC), GLC-mass-spectrometry (GLC-MS), methylation analyses (Hakomori S., J. Biochem. (Tokyo), 55: 205 - 208 (1964)), hydrolyses, and paper and column chromatography were performed as previously described (Perry et al, Carbohydr. Res., 322: 57 - 66 (1999)).

NMR spectroscopy

¹H- and ³C-NMR spectra were recorded on a Varian Inova 500 MHz spectrometer in D₂0 at 25°C with acetone standard (2.225 ppm for ¹H and 31.5 ppm for ¹³C) using standard pulse sequences.

Serology

Whole cell slide agglutination tests were performed as described previously (Perry at al, J. Clin. Microbiol, 26: 2391 - 2394 (1988)) using viable E. coli 0157:H7, C. freundii, and Y. entercolitica 0:9 bacilli, and various 0157- antigen-specific Mabs (Perry et al, supra). Immunodiffusion assays (Bundle et al, Infect. Immun., 46: 389 - 393 (1984)) used the aforementioned Mabs and purified LPS and O-PS of E. coli 0157:H7 and C. freundii antigens. ELISAs were performed as described previously (Perry et al, supra) using E. coli 0157:H7 LPS or O-PS, C. freundii LPS or O-PS, C. sedlakii LPS or O-PS, or E. coli O70 LPS as coating antigens at a concentration of 5 g ml^"1.C. freundii- specific polyclonal antiserum was prepared by vaccinating 10-week-old female CD1 mice subcutaneously with purified C. freundii LPS admixed with incomplete Freund's adjuvant. Mice were inoculated subcutaneously with 12.5 g of LPS in a volume of 0.1 ml three times at 14-day intervals. Antisera were prepared from blood collected 1 week following the final immunization. Although various particular embodiments of the present invention have been described hereinbefore for the purpose of illustration, it would be apparent to those skilled in the art that numerous variations may be made thereto without departing from the spirit and scope of the invention, as defined in the appended claims.

Claims

WHAT IS CLAIMED IS:

1. A live vaccine for the immunization of a vertebrate against a first Gram- negative enteric bacteria, said vaccine comprising a live, non- pathogenic second Gram-negative enteric bacteria naturally expressing an antigen of the first Gram-negative enteric bacteria or a structural mimic thereof.

2. A live vaccine according to claim 1 , wherein the first Gram-negative enteric bacteria is a strain of Escherichia coli.

3. A live vaccine according to claim 2, wherein the strain of E. coli is the strain E. coli 0157: H7.

4. A live vaccine according to claim 3, wherein the second Gram-negative enteric bacteria is a bacterial strain of the genus Salmonella or Citrobacter.

5. A live vaccine according to claim 4, wherein the bacterial strain is a wild type strain.

6. A live vaccine according to claim 4, wherein the bacterial strain is a recombinant strain.

7. A live vaccine according to claim 5, wherein the bacterial strain naturally expresses the 0157 antigen or a structural mimic thereof.

8. A live vaccine according to claim 7, wherein the bacterial strain expresses the 0157 antigen or a structural mimic thereof as part of its lipopolysaccharide.

9. A live vaccine according to claim 8, wherein the bacterial strain is selected from the group consisting of strains of Salmonella landau, Citrobacter sedlakii, and Citrobacter freundii.

10. A live vaccine according to claim 9, wherein the bacterial strain is selected from the group consisting of S. landau LCDC Strain # S-1358, C. sedlakii NRCC 6070, and C. freundii CDC 3488-90, NRCC 6052.

11. A live vaccine according to claim 10, wherein the vertebrate is a mammal.

12. A live vaccine according to claim 11 , wherein the mammal is a bovine.

13. A method for inducing in a vertebrate an immune response to a first Gram-negative enteric bacteria, said method comprising administering to said vertebrate an effective amount of a live non-pathogenic second Gram-negative enteric bacteria naturally expressing an antigen of the first Gram-negative enteric bacteria or a structural mimic thereof.

14. A method according to claim 13, wherein the first Gram-negative enteric bacteria is a strain of Escherichia coli.

15. A method according to claim 14, wherein the strain of E. coli is the strain E. coli 0157:H7.

16. A method according to claim 15, wherein the second Gram-negative enteric bacteria is a bacterial strain of the genus Salmonella or Citrobacter.

17. A method according to claim 16, wherein the bacterial strain is a wild type strain.

18. A method according to claim 16, wherein the bacterial strain is a recombinant strain.

19. A method according to claim 17, wherein the strain of Salmonella naturally expresses the 0157 antigen or a structural mimic thereof.

20. A method according to claim 19, wherein the bacterial strain expresses the 0157 antigen or a structural mimic thereof as part of its lipopolysaccharide.

21. A method according to claim 20, wherein the bacterial strain is selected from the group consisting of strains of Salmonella landau, Citrobacter sedlakii, and Citrobacter freundii.

22. A method according to claim 21 , wherein the bacterial strain is selected from the group consisting of S. landau LCDC strain # S-1358, C. sedlakii NRCC 6070, and C. freundii CDC 3488-90, NRCC 6052.

23. A method according to claim 22, wherein the vertebrate is a mammal.

24. A method according to claim 23, wherein the mammal is a bovine.

25. A method for preventing or treating carriage of a first Gram-negative enteric bacteria by a vertebrate, said method comprising administering to said vertebrate an effective amount of a live non-pathogenic second Gram-negative enteric bacteria naturally expressing an antigen of the first Gram-negative enteric bacteria or a structural mimick thereof.

26. A method according to claim 25, wherein the first Gram-negative enteric bacteria is a strain of Escherichia coli.

27. A method according to claim 26, wherein the strain of E. coli is the strain E. coli 0157:H7.

28. A method according to claim 27, wherein the second Gram-negative enteric bacteria is a bacterial strain of the genus Salmonella or Citrobacter.

29. A method according to claim 28, wherein the bacterial strain is a wild type strain.

30. A method according to claim 28, wherein the bacterial strain is a recombinant strain.

31. A method according to claim 29, wherein the bacterial strain naturally expresses the 0157 antigen or a structural mimic thereof.

32. A method according to claim 31 , wherein the bacterial strain expresses the 0157 antigen or a structural mimic thereof as part of its lipopolysaccharide.

33. A method according to claim 32, wherein the bacterial strain is selected from the group consisting of strains of Salmonella landau, Citrobacter sedlakii, and Citrobacter freundii.

34. A method according to claim 33, wherein the bacterial strain is selected from the group consisting of S. landau LCDC strain # S-1358, C. sedlakii NRCC 6070, and C. freundii CDC 3488-90, NRCC 6052.

35. A method according to claim 34, wherein the vertebrate is a mammal.

36. A method according to claim 35, wherein the mammal is a bovine.