A total of 43 strains of lactobacilli were isolated from the gastrointestinal tract of piglets at the time of weaning. Isolates, grown on solid media, were allocated to strongly adherent or non/weakly adherent groups on the basis of numbers attaching to isolated porcine enterocytes. Strains of Lactobacillus fermentum were disproportionally represented amongst the strongly-adherent strains and Lact. acidophilus and Lact. salivarius amongst the non/weakly-adherent group. Lactobacilli showed significantly better attachment ability when grown on agar than when grown in broth culture. Strongly adherent strains were not found to effect the attachment of enterotoxigenic Escherichia coli to porcine enterocytes, tested under the conditions of exclusion (lactobacilli added to the enterocytes before E. coli), competition (lactobacilli and E. coli added simultaneously) and displacement (E. coli added before lactobacilli). Tests were made with [14C]-labelled E. coli. Suspensions of bacteria and enterocytes were passed through a filter selected to retain enterocytes but pass free bacterial cells. Counts (dpm) obtained from filters after solubilization were taken as a measure of E. coli attachment. Some strains of lactobacilli coaggregated with enterotoxigenic E. coli with K88 fimbriae, but not with a K88-negative mutants strain. These were excluded from the competitive exclusion experiments. In the apparent absence of a direct effect on the association of E. coli with host tissue, removal of potential gut pathogens by aggregation could contribute to the probiotic properties ascribed to lactic acid bacteria.

The reversible and dose-dependent hyperplastic growth of the small intestine and accelerated epithelial cell turnover caused by feeding rats with diets containing kidney bean lectin (PHA) increased the proportion of immature cells on the villi whose membrane and/or cytoplasm contained mainly simple, polymannosylated glycans. These new alpha-linked mannosyl terminals, particularly of the damaged epithelium, facilitated the preferential adherence of opportunistic Escherichia coli with mannose-sensitive Type 1 fimbriae, and other coliforms, to the glycocalyx. Accordingly, the growth of the gut was accompanied by a reversible and PHA dose-dependent overgrowth with E. coli. As expected from their common carbohydrate specificity, the inclusion in the diet of the mannose-specific agglutinin from snowdrop (Galanthus nivalis) bulbs (GNA) significantly reduced the extent of E. coli overgrowth, but abolished neither the growth nor the damage caused by PHA to the small intestine. Thus, GNA and perhaps other mannose-specific lectins, especially when used in a preventive mode, can be used to specifically block the proliferation of Type 1 E. coli in the small intestine.

Salmonellosis in the rat has many similarities with the disease in humans, with the ileum thought to be the main site of colonization/invasion in both species. Thus, the rat may be a useful way to study the mechanism of infection by these pathogenic bacteria. A series of infection trials carried out with Hooded Lister rats showed that a salmonella infection persisted for an extended period of time and that salmonellae bind to the small intestinal epithelium as early as 4 h after intragastric intubation. Reinfection from the large intestine may not therefore initially play a significant role in the salmonella infection process. The rat model may therefore provide a means to test in vivo interventionist strategies, designed to block binding of the pathogens in the gastrointestinal tract. Road, Bucksburn, Aberdeen A B 2 9SB. UK (r-niail pjn'j"@rri.sari.ar..uk/. 0 1996 The Society for Applied Bacteriology 652 P.J. NAUGHTON E T A L . Correspondence to: Palrick J. Niiughlon, Division iflNutriiionu1 Sciences, Romeit Resrurch Institutr. Greenbiirn

An. in vitro fermentation system capable of maintaining at least part of the microbial population of the contents of the porcine ileum has been developed. The system was tested over the pH range 6.0-8.5, anaerobically and at dissolved oxygen concentrations within the ranges detected in the ileum of piglets at weaning (50 and 100 mumol l-1). The results demonstrated that changes in pH and dissolved oxygen within these ranges had relatively little effect on the total numbers of aerobic and anaerobic bacteria. Lactic acid bacteria, enumerated anaerobically, showed changes in viable counts in response to pH changes but were apparently unaffected by changes in dissolved oxygen, although the proportion of aerotolerant species within this group was increased at high concentrations of dissolved oxygen. The relative proportions of lactobacilli and coliforms were similar to those reported in vivo when dissolved oxygen was present at a concentration of 50 mumol l-1. Under these conditions the simulated population was consistently found to resist colonization by an enterotoxigenic Escherichia coli, of a serotype known to cause weaning scours in the piglet.

Mixtures of strains of Lactobacillus spp., two of strains inhibitory to growth of the porcine enteropathogen Escherichia coli O149: K88ac: K91 in a co-culture screen, and one of strains with no effect on this pathogen, were examined to determine their effects on the retention of this pathogen in an in-vitro continuous culture system designed to maintain a representation of the porcine ileal microflora. Strains effective in co-culture also increased the rate of decline of the pathogen in the continuous culture. The inhibitory effect exceeded the displacement effect observed with non-inhibitory strains.