- Volume 39, Issue 2, 1993

The expression of Escherichia coli lipopolysaccharide (LPS) and the binding capacity of anti-LPS monoclonal antibodies (MAbs) to E. coli grown in the presence or absence of subinhibitory concentrations of various antibiotics was studied. Four E. coli strains (three clinical blood-culture isolates and an isogenic, non-capsulate mutant of the O18:K1 parent) were grown in the presence of the β-lactam antibiotic, ampicillin, the aminoglycoside gentamicin, the fluoroquinolone ciprofloxacin and chloramphenicol. The techniques of silver staining, immunoblotting, whole-cell ELISA and flow cytometry were all used to monitor the expression of LPS on the bacteria and the binding of the anti-LPS MAbs. Treatment with ampicillin, chloramphenicol and ciprofloxacin resulted in enhanced binding of anti-core reactive MAbs to most E. coli strains. Overall, treatment with gentamicin produced the least effect on MAb binding. The presence of chloramphenicol decreased the expression of high molecular mass O-antigen or increased the expression of low molecular mass substituted E. coli LPS or both. These results further illustrate that LPS core, especially the inner-core region, becomes more accessible to antibodies when bacteria are grown in the presence of certain antibiotics. Possible synergy between antibodies and antibiotics for treatment of septicaemia and septic shock remains an intriguing possibility.

Sixty-eight isolates of Aeromonas spp. were examined biochemically and their cell proteins were analysed by silver-stained SDS-PAGE. Protein fingerprints did not correlate with phenotype. However, consideration of both phenotype and fingerprint showed clustering of epidemiologically related isolates. There was also evidence that similar strains could be found in infected people and water or other environmental samples.

Faecal samples from 123 infants who died with sudden infant death syndrome (SIDS) and from a comparative group of 52 age-matched babies were analysed for toxigenic bacteria and their toxins. Serum samples from the SIDS infants were also analysed for these toxins. A significantly higher proportion of toxigenic bacteria and their toxins were found in faecal samples of SIDS babies than in samples from the comparative group. These toxins were also found in serum from the SIDS babies. Clostridium perfringens was found in 54 (45.4 %) of 119 SIDS cases compared with 10 (19-6%) of 51 healthy babies (χ 2 = 101, p < 0.01); C. difficile in 33 (27.7 %) of 119 SIDS cases compared with 8 (14.8 %) of 54 healthy babies (χ 2 = 3.43ns, p < 0.1); Staphylococcus aureus in 12 (27.3%; 66.7% enterotoxigenic) of 44 SIDS cases compared with 12 (85.7%; non-enterotoxigenic) of 14 healthy babies = 14.9, p < 0.001); C. botulinum in 6 (5.0 %) of 120 SIDS cases compared with 0 of 53 healthy babies (%2 = 2.74, p < 0.1). Campylobacter jejuni, Yersinia enterocolitica, Vibrio parahaemolyticus, salmonellae and Bacillus cereus were not detected. Heat-labile toxin, lethal to mice (HLML) was found in 32 (27.1 %) of 118 SIDS faecal samples compared with 5 (10.6 %) of 47 healthy babies (j 2 = 5.24, p < 0.05); cytotoxins in 38 (30.9%) of 123 SIDS faecal samples compared with 0 of 21 of healthy babies (χ 2 = 8.8, p < 0.01) and 24 (27.6%) of 87 SIDS serum samples. C. perfringens enterotoxin was detected in 33 (34.4%) of 96 SIDS faecal extracts compared with 0 of 23 of healthy babies (χ 2 = 10.94, p < 0.001), and in 27 (24.5 %) of 110 SIDS serum samples. C. perfringens a-toxin (presumptive) was detected in 14 (17.5%) of 80 SIDS faecal extracts compared with 0 of 17 from healthy babies (χ 2 = 3.5ns, p — 0.05) and in 2 (2.3 %) of 87 SIDS serum samples. C. difficile toxin was detected in four SIDS faecal samples and two serum samples. C. botulinum toxin was detected in only one of 120 SIDS faecal samples compared with none of 49 from healthy babies. Staphylococcal enterotoxins were detected in 8 (19.5 %) of 41 SIDS faecal samples compared with 0 of 19 from healthy babies (χ 2 = 4.278, p < 0.05), and in 4(10.8 %) of 37 SIDS serum samples. Toxigenic and non-toxigenic strains of C. perfringens and C. difficile occurred in faecal samples of both SIDS and healthy babies. Formula-fed SIDS babies had a significantly higher incidence of C. difficile (j 2 = 6.654, p < 0.01), C. perfringens (χ 2 = 6.422, p < 0.05), and its enterotoxin (j 2 = 7.787, p < 0.01) in faeces, and a higher incidence (non-significant) of C. perfringens enterotoxin in their serum, faecal HLML toxin, and S. aureus and its enterotoxin, than breast-fed babies. Male SIDS babies had a significantly higher incidence of C. perfringens (χ 2 = 7.687, p < 0.01) and higher incidences (non-significant) of C. perfringens enterotoxin, HLML toxin, C. difficile, and S. aureus and its enterotoxin than female babies. SIDS babies dying in winter had a significantly higher incidence of C. difficile than those dying in summer (χ 2 = 5.328, p < 0.05) and spring (χ 2 = 4.444, p < 0.05). C. perfringens, S. aureus and their enterotoxins occurred in more babies dying in autumn and winter than in spring and summer. The incidence of these bacteria and their toxins did not differ for position of death. These results provide some support for the idea that intestinal toxins have a pathogenic role in SIDS.

For rapid identification of Staphylococcus aureus, a monoclonal antibody (MAb)-biotin-avidin-peroxidase complex, directed against the S. aureus thermostable nuclease (TNase), was formed and used in a rapid three-step sandwich enzyme-linked immunofiltration assay (sELIFA) and a three-step sandwich enzyme-linked immunosorbent assay (sELISA). The MAb-peroxidase complex was formed by incubating the biotinylated MAbs with a streptavidin-peroxidase conjugate and the complex was purified by gel permeation chromatography. When compared with a four-step MAb-based sELISA described previously, this complex permitted one reagent step to be omitted in a three-step sELISA, and the test time was significantly reduced. The test sensitivity was slightly reduced in the three-step ELISA (detection limit 1.0—2.0 ng of TNase/ml) when compared to the four-step sELISA (detection limit 0.5—1.0 ng of TNase/ml). The sELIFA method was based on the filtration of bacterial culture supernates through nitrocellulose membrane disks pre-spotted with a MAb directed against the S. aureus TNase, followed by detection with the MAb-peroxidase complex (three-step sELIFA). A detection limit of 0.5—2.0 ng of TNase/ml was achieved with the three-step sELIFA, depending on the filtrate volume of culture supernates. The total test time was 10—15 min when pre-spotted and blocked membranes were used. A total of 85 bacterial strains was tested in the sELIFA. All the 28 S. aureus strains showed positive results, but none of the 57 non-S. aureus strains did so, although some of these produced thermostable nuclease activity. When 75 blood cultures were tested directly in the sELIFA, 87% of the cultures with growth of S. aureus gave a positive result whereas all of the cultures with non-S. aureus gave negative results, a diagnostic sensitivity similar to that of the routine TNase enzyme test. Thus, the three-step sELIFA has potential for the rapid confirmation of S. aureus bacteraemia and, possibly, also for detecting S. aureus by direct testing of other clinical specimens.

The serum IgG response of human volunteers challenged with Vibrio cholerae O1 was analysed for reactivity to V. cholerae O1 outer-membrane antigens by enzyme-linked immunosorbent assay (ELISA) and the immunoblot technique. Purified outer-membrane antigen preparations from vibrios grown in low-iron conditions were separated by SDS-PAGE. Specific immunoblot reactions of human sera showed that an 18-kDa antigen, cholera protective antigen, was the major antigen with which sera reacted. ELISA revealed an increase in antibody to the 18-kDa antigen in nine of 10 challenged volunteers. This response was independent of the biotype and serotype of the V. cholerae O1 challenge strain. Cholera protective antigen appears to be one of the major outer-membrane antigens involved in the human immune response to infection with V. cholerae.

Mice that had been treated with cyclophosphamide and ampicillin were fed with Pseudomonas aeruginosa. These procedures induced an endogenous septicaemia under conditions mimicking the pathophysiology of the disease in man. This model was used to compare the mortality rates in mice infected with P. aeruginosa isolates from various clinical sources. Mortality rates in mice given isolates from blood cultures had a broad range (0—100%), but the mean rate was significantly higher than with isolates from other infection sites. Moreover, blood isolates persisted in the intestines of mice after oral inoculation, whereas most isolates from other sources were gradually eliminated. Most P. aeruginosa isolates from blood culture produced significantly higher levels of exotoxin A and total proteases than isolates from other infection sites. Amongst the blood isolates, all but one of the lethal strains produced large quantities of exotoxin A or total proteases or both. Taken together, the results suggest that the ability of P. aeruginosa to adhere to the intestinal tract and to produce high levels of exo-enzymes may contribute to the development of fatal septicaemia.

Experiments were performed to determine whether a modern flow cytometer could be used to study bacterial populations in suspension, with particular reference to their morphological characteristics and their responses to antibiotics. The FACScan, a commercial benchtop flow cytometer fitted with an air-cooled laser, designed primarily for the study of eukaryotic peripheral blood mononuclear cells, yielded reproducible data relating to bacterial shape and internal architecture. It was sensitive enough to detect changes in bacterial morphology on entry into the growth cycle and after exposure to antibiotics. Antibiotic-induced morphological changes affecting subpopulations of bacteria were sufficiently specific to allow differentiation between antibiotics with different cell-wall enzyme targets. Simultaneously, the effect of such antibiotics on the integrity of the outer cell membrane of Escherichia coli was assessed by measurement of the association of the nucleic acid-binding dye propidium iodide with the bacteria. These experiments demonstrated complex patterns of probable cell-wall leakage, related to the modes of action of the antibiotics. The FACScan is a useful and sensitive tool for the study of the morphology and physiology of bacterial populations in suspension, and is especially applicable to the study of antibiotic action.

Potassium tellurite was assessed for the selection of verocytotoxigenic (VT+) Escherichia coli O157. MICs were higher for VT+ E. coli O157 than for other strains of E. coli and for Aeromonas spp. MacConkey medium containing sorbitol, tellurite and cefixime (TC-SMAC) permitted the growth of VT+ E. coli O157 and Shigella sonnei but partially or completely inhibited the growth of 67% of other strains of E. coli and all or most strains of other sorbitol-non-fermenting species tested. Of 391 rectal swabs from cattle screened on TC-SMAC medium, 26 yielded isolates of VT+ E. coli O157 whereas sorbitol-MacConkey medium with cefixime and rhamnose yielded only nine isolates. Inclusion of potassium tellurite in sorbitol-MacConkey agar markedly increased the rate of isolation of VT+ E. coli O157 from cattle rectal swabs and may do so for other types of specimen.