Horizontally transferred genetic elements and their role in pathogenesis of bacterial disease.

This article reviews the roles that laterally transferred genes (LTG) play in the virulence of bacterial pathogens. The features of LTG that allow them to be recognized in bacterial genomes are described, and the mechanisms by which LTG are transferred between and within bacteria are reviewed. Genes on plasmids, integrative and conjugative elements, prophages, and pathogenicity islands are highlighted. Virulence genes that are frequently laterally transferred include genes for bacterial adherence to host cells, type 3 secretion systems, toxins, iron acquisition, and antimicrobial resistance. The specific roles of LTG in pathogenesis are illustrated by specific reference to Escherichia coli, Salmonella, pyogenic streptococci, and Clostridium perfringens.

Alteration of the glycolipid binding specificity of the pig edema toxin from globotetraosyl to globotriaosyl ceramide alters in vivo tissue targetting and results in a verotoxin 1-like disease in pigs.

All members of the verotoxin (VT) family specifically recognize globo-series glycolipids on the surface of susceptible cells. Those toxins that are associated with human disease, VT1, VT2, and VT2c, bind to globotriaosyl ceramide (Gb3) while VT2e, which is associated with edema disease of swine, binds preferentially to globotetraosyl ceramide (Gb4). We were recently able to identify, using site-directed mutagenesis, amino acids in the binding subunit of these toxins that are important in defining their glycosphingolipid (GSL) binding specificity (Tyrrell, G. J., K. Ramotar, B. Boyd, B. W. Toye, C. A. Lingwood, and J. L. Brunton. 1992. Proc. Natl. Acad. Sci. USA. 89:524). The concomitant mutation of Gln64 and Lys66 in the VT2e binding subunit to the corresponding residues (Glu and Gln, respectively) found in VT2 effectively converted the GSL binding specificity of the mutant toxin from Gb4 to Gb3 in vitro. We now report that the altered carbohydrate recognition of the mutant toxin (termed GT3) has biological significance, resulting in a unique disease after intravascular injection into pigs as compared with classical VT2e-induced edema disease. The tissue localization of radiolabeled GT3 after intravascular injection was elevated in neural tissues compared with VT2e accumulation, while localization of GT3 to the gastrointestinal tract was relatively reduced. Accordingly, the pathological lesions after challenge with GT3 involved gross edema of the cerebrum, cerebellum, and brain stem, while purified VT2e caused hemorrhage and edema of the cerebellum, and submucosa of the stomach and large intestine. In addition, both radiolabeled toxins bound extensively to tissues not directly involved in the pathology of disease. VT2e, unlike GT3 or VT1, bound extensively to red cells, which have high levels of Gb4. The overall tissue distribution of VT2e was thus found to be influenced by regional blood flow to each organ and not solely by the Gb4 levels of these tissues. Conversely, the distribution of GT3 (and VT1), which cleared more rapidly from the circulation, correlated with respective tissue Gb3 levels rather than blood flow. These studies indicate the primary role of carbohydrate binding specificity in determining systemic pathology, suggest that the red cells act as a toxin carrier in edema disease, and indicate that red cell binding does not protect against the pathology of systemic verotoxemia.

Naturally occurring plasmid carrying genes for enterotoxin production and drug resistance.

Escherichia coli strain 86, isolated from a piglet with diarrhea, carries plasmid-linked genes for resistance to tetracycline, streptomycin, and sulfonamides and for production of heat-labile and heat-stable enterotoxin. Results of (i) genetic experiments involving conjugal transfer and phage P1-mediated transduction and (ii) physical experiments involving electron microscopic examination of plasmid DNA and heteroduplex analysis show that a single conjugative plasmid carries the genes for drug resistance and production of enterotoxin.

Bacteriophages for prophylaxis and therapy in cattle, poultry and pigs.

The successful use of virulent (lytic) bacteriophages (phages) in preventing and treating neonatal enterotoxigenic Escherichia coli infections in calves, lambs and pigs has prompted investigation of other applications of phage therapy in food animals. While results have been very variable, some indicate that phage therapy is potentially useful in virulent Salmonella and E. coli infections in chickens, calves and pigs, and in control of the food-borne pathogens Salmonella and Campylobacter jejuni in chickens and E. coli O157:H7 in cattle. However, more rigorous and comprehensive research is required to determine the true potential of phage therapy. Particular challenges include the selection and characterization of phages, practical modes of administration, and development of formulations that maintain the viability of phages for administration. Also, meaningful evaluation of phage therapy will require animal studies that closely represent the intended use, and will include thorough investigation of the emergence and characteristics of phage resistant bacteria. As well, effective use will require understanding the ecology and dynamics of the endemic and therapeutic phages and their interactions with target bacteria in the farm environment. In the event that the potential of phage therapy is realized, adoption will depend on its efficacy and complementarity relative to other interventions. Another potential challenge will be regulatory approval.

Enzymatic activity of Campylobacter jejuni hippurate hydrolase.

The hippurate hydrolase enzyme of Campylobacter jejuni was expressed in Escherichia coli as a six-histidine-tagged fusion protein. The purified recombinant enzyme was characterized to gain an understanding of the structure and activity of the hippurate hydrolase. The recombinant enzyme had a native molecular mass of 193+/- 11 kDa a reduced molecular mass of 42.4+/- 0.8 kDa, and possessed 1.98+/- 0.68 molecules of zinc per enzyme subunit molecule, suggesting that it was a homotetramer with two associated zinc ions. The enzyme was a metallocarboxypeptidase that was sensitive to silver, copper and ferrous ions, and displayed optimal activity at pH 7.5 and 50 degrees C. It hydrolyzed carboxypeptidase substrates in vitro, displaying its highest activity against N-benzoyl-linked small aliphatic amino acids. A high proportion of the enzyme structure consisted of highly ordered alpha-helix and beta-sheet sequences. An alignment of the amino acid sequence of the hippurate hydrolase enzyme with those of related enzymes with similar activities revealed several conserved amino acids, which might be involved in enzyme catalysis or metal ion binding for the enzyme. Site-directed mutagenesis of the recombinant enzyme demonstrated that the Asp(76), Aps(104), Glu(134), Glu(135), His(161) and His(356) positions were important for the catalytic activity of the enzyme.

Comparative cytotoxicity of purified Shiga-like toxin-IIe on porcine and bovine aortic endothelial and human colonic adenocarcinoma cells.

Porcine and bovine aortic endothelial cells and human colonic adenocarcinoma cells were compared for their susceptibility to the toxic effect of purified Shiga-like toxin IIe (SLT-IIe), measured by the neutral red cytotoxicity assay. Cytotoxicity correlated with toxin binding as indicated by fluorescence activated cell sorter analysis and with the globotriosylceramide (Gb3) and globotetraosylceramide (Gb4) content of cells determined by high pressure liquid chromatography. One line of porcine aortic endothelial cells was 1400-fold more susceptible than the line of bovine aortic endothelial cells that was tested, but a second line of porcine aortic endothelial cells was highly refractory to SLT-IIe. Human colonic adenocarcinoma cells lacked detectable levels of Gb4 and were least susceptible to SLT-IIe.

A comparison of procedures involving single step Salmonella, 1-2 Test, and modified semisolid Rappaport-Vassiliadis medium for detection of Salmonella in ground beef.

Procedures involving Single Step Salmonella (SSS), 1-2 Test, and Modified Semisolid Rappaport-Vassiliadis Medium (MSRV) were compared for their speed and sensitivity in detection of Salmonella in ground beef contaminated with one isolate of each of five Salmonella serotypes. Inocula of 10, 10(2) and 10(3) CFU/g of ground beef were used. When pre-enrichment in buffered peptone water and selective enrichment in tetrathionate broth were used, SSS and MSRV detected all five species of Salmonella at all levels of contamination, whereas the 1-2 Test was positive in 0, 12, and 15 of 15 tests at 10, 10(2) and 10(3) CFU/g, respectively. When only pre-enrichment was used, the results with MSRV were unchanged but the SSS test failed to detect S. typhimurium. Thus, with pre-enrichment the MSRV and SSS procedures were equally sensitive and both produced a result on the third day. The 1-2 Test was less sensitive and slower, with results available on day-4. The MSRV protocol was best overall.

Immunization of pigs with a purified Shiga-like toxin II variant toxoid.

Passive transfer of neutralizing antibodies and active immunization with a toxoid of purified Shiga-like toxin II variant (SLT-IIv, edema disease toxin) were used to protect pigs against challenge with SLT-IIv. Six 6-week-old pigs were passively immunized by intraperitoneal administration of an immunoglobulin preparation from porcine antiserum against purified SLT-IIv. Six 6-week-old pigs and twelve 2-week-old pigs were actively immunized by two intramuscular injections of 25 micrograms of SLT-IIv toxoid given 2 weeks apart. The 24 immunized pigs and an equal number of age-matched unimmunized control pigs were all challenged by intravenous injection of purified SLT-IIv (6 ng/kg body weight). The six passively immunized pigs acquired neutralizing SLT-IIv antibody titers of 1280 or 2560 and the 18 actively immunized pigs mounted a serum immune response with toxin neutralizing titers ranging from 128 to 10,240. All 24 immunized pigs survived the challenge but nine pigs with the lowest titers showed mild clinical signs of edema disease. All 24 control pigs died within 30 hours of challenge.

Shiga toxin genes in avian Escherichia coli.

The purpose of this study was to determine the presence of stx genes in avian pathogenic Escherichia coli (APEC). We examined 97 APEC isolates: 34 from lesions of avian cellulitis, 31 from avian septicemia, 13 from swollen head syndrome (SHS) in chickens, and 19 from diseased turkeys. We also examined five isolates from the feces of healthy chickens. All 102 E. coli isolates were tested for the presence of stx genes by PCR amplification and by colony blots using probes specific for stx1 and stx2. Fifty-three percent (52) of the 97 APEC carried stx gene sequences: one isolate carried stx2 sequences, two carried both stx1 and stx2 sequences, and the remaining 49 isolates carried only stx1 sequences. Twenty-six isolates were positive by both hybridization and PCR amplification, 10 were positive by PCR only, and 16 were positive by hybridization only. All the stx-positive isolates were negative by PCR for the eae and E-hlyA genes. The five isolates from healthy chickens were all negative for stx. All 13 SHS isolates were positive for the stx1 gene and had low titres for cytotoxicity in the Vero cell assay (VCA). Other stx-positive isolates were negative in the VCA. The stx1 gene from one SHS E. coli isolate was cloned and sequenced and shown to be identical to that of the stx gene of Shigella dysenteriae. The observations indicate that stx1 gene sequences are widespread among APEC but that cytotoxicity on Vero cells is uncommon.

A comparison of extracted proteins of isolates of Dermatophilus congolensis by sodium dodecyl sulphate-polyacrylamide gel electrophoresis and Western blotting.

Antigenic diversity within a collection of 18 isolates of Dermatophilus congolensis from different Continents was examined by sodium dodecyl sulphate-polyacrylamide gel electrophoresis (SDS-PAGE) and by Western blotting with sera from cattle with clinical dermatophilosis using whole cell extracts obtained by three methods and one extract of extracellular products of D. congolensis. One of the methods involving the release of a lysostaphin-solubilized protein (LSP) of whole cells of D. congolensis revealed a number of discrete and easily-identifiable bands in SDS-PAGE which were found suitable for characterizing protein patterns and was, therefore, subsequently used for a comparative analysis of the proteins of all the D. congolensis isolates. Six electropherotypes (ET) of D. congolensis were identified among the 18 isolates using the protein profiles based on the presence of four protein bands at Molecular weights (MW) 62, 28, 17.4 and 16.4 kDa. The ETs were found among isolates from different animal species and from different sources with ET1 consisting of three bovine and two equine isolates; ET2, two bovine and three ovine isolates; ET3, two bovine isolates; ET4, two bovine isolates; ET5, one bovine and one ovine isolates and ET6, two bovine isolates. Immunoblotting of the extracts of D. congolensis isolates with sera from cattle with clinical dermatophilosis infection demonstrated protein bands of MW ranging from 9 kDa to 188 kDa. Sera from chronic dermatophilosis infection demonstrated a 28 kDa protein which was immunodominant in the LSP extracts of all the 18 isolates of D. congolensis tested while sera from mild infections demonstrated mainly the 62 kDa protein in the same extracts. However, many protein bands were demonstrated in surface membrane (TSMP) and extracellular protein extracts with sera from only mildly infected animals. The protein patterns observed in all isolates of D. congolensis revealed global antigenic similarities and distinct differences among isolates which could not be associated with either geographic, climatic or host factors. Also sera from infected animals from endemic regions of dermatophilosis could not differentiate isolates of D. congolensis. This suggests the possibility that such sera must have come from animals that had been infected by a multitude of D. congolensis strains present in the herd environment and strains an animal could have come across during the 'ritual' annual cross-country migration of the cattle herds.

Tagging and elimination of plasmids in Salmonella of avian origin.

This study compared the effectiveness of a number of procedures designed to label and eliminate plasmids that may play a role in virulence in Salmonella. Twenty strains of Salmonella of 9 serovars were subjected to 3 methods for labelling plasmids with transposons. Strains containing labelled and unlabelled plasmids were exposed to physical and chemical curing agents. Plasmids in 9 of 20 strains of Salmonella were tagged by conjugation with a donor Escherichia coli containing a temperature-sensitive RP4 plasmid that carried the Tn1 transposon. Plasmids in 2 of 5 strains of Salmonella were labelled by conjugation with a donor E. coli that contained a F' tslac::Tn5 plasmid. Transduction of Salmonella with a P22 bacteriophage that carried a temperature-sensitive Tn10 transposon resulted in chromosomal insertion of Tn10 in 2 of 10 strains. Use of chemical curing agents resulted in curing of plasmids in only 6 of 17 strains. Two strains were cured by ethidium bromide, two by a combination of ethidium bromide and novobiocin, two by a combination of imipramine and methylene blue, and none by acridine orange, novobiocin, sodium dodecyl sulfate or rifampicin. In contrast, plasmids in 14 of 17 Salmonella strains were eliminated by incubation at 45.5 degrees C.

Characteristics of the Shiga-like toxin produced by Escherichia coli associated with porcine edema disease.

Shiga-like toxin (SLT-IIv) from Escherichia coli strains associated with edema disease of pigs was characterized and compared with SLT-I, SLT-II, and the SLT of E. coli strain HI8 (SLT-HI8). SLT-IIv from an E. coli K12 in which the genes for SLT-IIv had been cloned was indistinguishable from SLT-IIv of wild strains of E. coli from edema disease. There was cross-neutralization among all SLTs except SLT-I. The different SLTs could be distinguished by heat lability, with the descending order of heat lability being SLT-IIv, SLT-II, SLT-I, and SLT-HI8. SLT-IIv and SLT-HI8 had lower cytotoxic titers on HeLa cells compared with Vero cells and were more active on MDBK cells than were the other SLTs. All SLTs were enterotoxic in rabbit but not in pig intestine and SLT-IIv was less enterotoxic than SLT-I. SLT-IIv had a lower LD50 in mice than did the other SLTs.

Resistance of broiler chickens to Escherichia coli respiratory tract infection induced by passively transferred egg-yolk antibodies.

Egg-yolk antibodies induced by immunizing hens with selected Escherichia coli antigens were evaluated for their ability to protect broiler chickens against respiratory/septicemic disease caused by avian pathogenic E. coli (APEC). Seven groups of broiler breeder hens were vaccinated three times, 1 week apart with live E. coli, killed E. coli, E. coli antigens [lipopolysaccharide (LPS), type 1 pilus adhesin (FimH), P pilus adhesin (PapG), aerobactin outer membrane receptor (IutA)] or phosphate buffered saline (PBS). An O78 APEC strain was used for preparation of all the antigens. Egg yolk immunoglobulins (IgY) were purified from eggs of each group and antibody activity in serum and purified IgY was determined by enzyme-linked immunosorbent assay (ELISA). IgY (100mg) was injected intramuscularly into 11-day-old broiler chickens, which were challenged 3 days later with homologous (O78) or heterologous (O1 or O2) E. coli by the intra-air sac route. Mortality was recorded and surviving chickens were euthanized 1 week after the challenge and examined for macroscopic lesions. Passive antibodies against all antigens except FimH were protective (90-100%) against the homologous challenge, but only anti-PapG and anti-IutA were effective against heterologous challenge. Anti-PapG IgY provided the greatest protection against the three serogroups of E. coli used for challenge. Hence vaccination of broiler breeders to induce anti-PapG and anti-IutA antibodies may provide passive protection of progeny chicks against respiratory/septicemic disease caused by APEC.

In vitro comparison of bovine mastitis and fecal Escherichia coli isolates.

In vitro methods were used to test the hypothesis that Escherichia coli from bovine mastitis are essentially no different from isolates from bovine feces. Fifty E. coli isolates from bovine mastitic milk, 50 from feces of mastitic cows and 50 from feces of healthy cows were compared with respect to biochemical properties and certain potential virulence factors. There were no significant differences among the groups in tests for biotype; production of colicins, colicin V, or Vero cell cytotoxicity; and growth in 90% gnotobiotic calf serum or 90% normal milk whey. Resistance to killing in 90% gnotobiotic calf serum varied from 66 to 84%. Most isolates grew in normal whey: the percentage in a group varied from 86 to 96. Mastitic milk isolates were significantly different from the fecal isolates in adonitol fermentation (P < or = 0.006), production of aerobactin (P < or = 0.026), and ability to grow in 90% mastitic whey (P < or = 0.00004). However, only 40% of mastitis E. coli fermented adonitol and only 20% produced aerobactin. Ninety-six percent of mastitic milk E. coli grew in mastitic whey, whereas 64% and 60%, respectively, of mastitic fecal and normal fecal isolates grew in this medium. It is concluded that none of the properties that were investigated constitute potential virulence factors or markers for ability to induce mastitis; the data are consistent with the hypothesis that mastitic E. coli are simply opportunistic pathogens.

Comparison of an LPS-specific competitive ELISA with a motility enrichment culture method (MSRV) for detection of Salmonella typhimurium and S. enteritidis in chickens.

We report here evaluation of a competitive enzyme-linked immunosorbent assay (c-ELISA) for detection of Salmonella spp. in chicken organs and faeces. The c-ELISA used a monoclonal antibody (MAb), specific for a genus-specific epitope of the outer core oligosaccharide of salmonellae. Salmonella lipopolysaccharide (LPS) in samples competed with Salmonella LPS coated on microtitre plates, for binding to the MAb. Competition reduced binding of the MAb to the LPS on the plate and of the secondary antibody to the MAb hence reducing the chromogenic signal. Stable coating and minimal false positive were achieved by conjugating LPS to poly-L-lysine. The c-ELISA was compared with motility enrichment culture using modified semisolid Rappaport Vassiliadis (MSRV) medium, which detected less than 10(2) CFU/g, and did not allow migration of non-salmonella species. The c-ELISA detected 10(6) CFU of enriched culture or 10(2)-10(3) CFU of Salmonella/g of faeces. Its limit of detection was thus higher than that of MSRV culture and it had a sensitivity of 92.9% and a specificity of 96.7%.

Evaluation of an aroA mutant Salmonella typhimurium vaccine in chickens using modified semisolid Rappaport Vassiliadis medium to monitor faecal shedding.

In groups of chickens vaccinated orally or intramuscularly with a live aroA mutant Salmonella typhimurium vaccine strain and then experimentally inoculated with 10(8) CFU of wild type S. typhimurium or 10(9) CFU of S. enteritidis, faecal shedding of the vaccine and wild type strains was monitored by the buffered peptone water-modified semisolid Rappaport Vassiliadis medium method, which detected less than 10(2) CFU per gram of faeces. The vaccine strain was shed in the faeces for up to 26 days. Vaccination failed to reduce the faecal shedding of wild type S. typhimurium or S. enteritidis. The variation in the shedding patterns of chickens within each group was greater than between treatment groups.

Attaching and effacing Escherichia coli isolated from dogs in Brazil: characteristics and serotypic relationship to human enteropathogenic E. coli (EPEC).

Escherichia coli isolates recovered from 182 fecal specimens from dogs up to five months old from the cities of São Paulo and Campinas, SP, Brazil, were examined by polymerase chain reaction (PCR) for several virulence factors and properties. The eae gene was found in 23 isolates of E. coli from 22 dogs, 19 of 146 (13%) from dogs with diarrhea and 3 of 36 (8.3%) from dogs with no diarrhea. Two different eae+ isolates were recovered from one dog with diarrhea. Isolates from two dogs with diarrhea harbored the bfpA gene, and none of the isolates possessed genes for enterotoxins, the EAF plasmid or Shiga toxins. PCR showed that, among the 23 isolates, eight were positive for beta intimin, six for gamma, two for, one for alpha, one for kappa, and five showed no amplification with any of the nine pairs of specific intimin primers used. PCR also showed that the LEE (locus of enterocyte effacement) was inserted in selC in four isolates, likely in pheU in seven isolates, and in undetermined sites in twelve isolates. Fifteen isolates adhered to HEp-2 cells and were fluorescence actin staining (FAS) positive. The predominant adherence pattern was the localized adherence-like (LAL) pattern. The eae-positive isolates belonged to a wide diversity of serotypes, including O111:H25, O119:H2 and O142:H6, which are serotypes that are common among human EPEC. These results confirmed the presence of EPEC in dogs (DEPEC) with and without diarrhea. The virulence factors found in these strains were similar to those in human EPEC, leading to the possibility that EPEC may move back and forth among human and canine populations.

Use of a Shiga toxin (Stx)-enzyme-linked immunosorbent assay and immunoblot for detection and isolation of Stx-producing Escherichia coli from naturally contaminated beef.

The purpose of this study was to evaluate an enzyme-linked immunosorbent assay (ELISA) and an immunoblot procedure for detection and isolation of Shiga toxin-producing Escherichia coli (STEC) from beef, and to correlate the presence of STEC in beef with E. coli and total coliform counts. A total of 120 samples of boneless beef supplied to a meat processor in southern Ontario were tested for the presence of STEC, E. coli, and total coliforms. Following enrichment in modified tryptic soy broth, samples were screened for Shiga toxin (Stx) by a Stx-ELISA and a Vero cell assay (VCA). Samples that were positive in the Stx-ELISA were subjected to the Stx-immunoblot for STEC isolation. Overall, 33.3% of samples were positive in the VCA, and 34.2% were positive in the Stx-ELISA. There was almost complete agreement between the Stx-ELISA and the VCA results (kappa = 0.98). The sensitivity and specificity of the Stx-ELISA with respect to the VCA were 100% and 98.75%, respectively. STEC were isolated by the Stx-immunoblot from 87.8% of the samples that were positive in the Stx-ELISA. The STEC isolates belonged to 19 serotypes, with serotype O113:H21 accounting for 10 of 41 isolates. No STEC of serotype O157:H7 were isolated. There was a significant correlation between E. coli counts and total coliform counts (Spearman correlation coefficient = 0.68, P < 0.01). The E. coli count was positively correlated with detection of STEC by both the Stx-ELISA and the VCA (P < 0.01).

Infectious agents in acute respiratory disease in horses in Ontario.

A study of acute respiratory disease in horses in Ontario was undertaken to determine the identity of current causative infectious agents. A nasopharyngeal swab was designed and utilized to maximize isolation of viruses, mycoplasma, and pathogenic bacteria. Serum samples were collected for parallel determination of antibody titers to equine influenza virus type A subtype 1 (H7N7) and subtype 2 (H3N8), equine rhinovirus types 1 and 2, equine herpesvirus type 1, Mycoplasma equirhinius, and Mycoplasma felis. Equine rhinovirus type 2 was recovered from 28/92 horses tested, and equine influenza virus type A, subtype 2, was recovered from 5. The mycoplasma and bacteria isolated were consistent with those commonly associated with nonspecific respiratory diseases in horses, except that Streptococcus pneumoniae capsular type 3 was isolated from 10 horses.

Construction and characterization of avian Escherichia coli cya crp mutants.

We constructed delta cya delta crp mutants of two avian septicemic Escherichia coli strains and evaluated their attenuation in virulence. The P1 phage was used to transfer cya::Tn10 from an E. coli K-12 strain into virulent avian O78 and O2 E. coli isolates. Tetracycline-resistant transductants were plated on Bochner-Maloy Medium, and tetracycline-sensitive colonies were selected, then tested by polymerase chain reaction to confirm that they had deletions of the cya gene. Deletions of crp were created by the same technique in isolates with deletions in cya. The delta cya and delta cya delta crp derivatives had slower growth rates, smaller colonies, and impaired fermentation of carbohydrates compared with their wild parents, and they did not revert. Attenuation of the mutant strains was evaluated by subcutaneous (s.c.) inoculation of day-old chicks and by intratracheal (i.t.) inoculation of 9-day-old chicks previously inoculated intranasally with infectious bronchitis virus. For the wild O78 strain and its delta cya and delta cya delta crp derivatives, the percentages of chicks that died within 6 days of s.c. injection of approximately 5 x 10(7) organisms were 100, 60, and 0, respectively. The corresponding percentages for wild-type O2 and its delta cya and delta cya delta crp mutants were 100, 70, and 20 at a dose of approximately 2 x 10(5) organisms. Following i.t. inoculation, group scores based on pathologic and bacteriologic findings were 51%, 15%, and 9% for wild, delta cya, and delta crp O78 strains (inoculum approximately 2 x 10(7) organisms) and 98%, 31%, and 11%, respectively, for the corresponding O2 strains (inoculum approximately 4 x 10(6) organisms). This study demonstrated reduced virulence and stability of the double mutant, which may useful as a live attenuated vaccine against poultry colibacillosis.