Association of Escherichia coli with the Small Intestinal Epithelium I. Comparison of Enteropathogenic and Nonenteropathogenic Porcine Strains in Pigs

Two enteropathogenic strains of Escherichia coli (EEC) differed from a nonenteropathogenic strain of E. coli (NEEC) in their association with porcine small intestinal epithelium. The EEC characteristically were found along villi from tip to base and contiguous to the brush border. They were not in crypts. In contrast, the NEEC characteristically remained in the central lumen near the tips of villi and was only occasionally contiguous to the brush border. No organisms were detected within epithelial cells. The difference in distribution between EEC and NEEC was apparent in ligated jejunal loops 45 min postexposure. The association between host and bacterial cells was most consistently demonstrated on frozen sections of intestine, as other histological techniques removed many bacteria. However, cellular details of the association were best demonstrated in chemically fixed tissues.     

Host specificity of enteropathogenic Escherichia coli from rabbits: lack of correlation between adherence in vitro and pathogenicity for laboratory animals.

The pathogenicity of four attaching and effacing strains of enteropathogenic Escherichia coli originally isolated from diarrheic rabbits was investigated by inoculating them perorally into rabbits, guinea pigs, and mice. The ability of the four strains to adhere to cultured epithelial cells, erythrocytes, and intestinal brush borders from various animal species, including rabbits, guinea pigs, and mice, varied considerably. Only one strain carried AF/R1 fimbriae, which are believed to determine the host specificity of these bacteria. Despite these differences, the pattern of behavior of the four strains in experimentally infected animals was similar. Each strain caused fatal diarrhea in rabbits (although the virulence of individual strains for rabbits differed significantly), and none was virulent for guinea pigs or mice. None of the strains colonized the intestinal tract of guinea pigs, but all were able to cause attaching-effacing lesions in ligated loops of guinea pig small intestine. By contrast, all four strains colonized mice, in particular the distal intestine, but none induced attaching-effacing lesions in mouse intestinal loops. These findings suggest that there may be previously unrecognized host-restricted adhesins in enteropathogenic E. coli and indicate that adherence to erythrocytes or intestinal brush borders in vitro does not necessarily reflect colonizing ability or pathogenicity in vivo.     

Attaching and effacing activities of rabbit and human enteropathogenic Escherichia coli in pig and rabbit intestines.

Three strains of enteropathogenic Escherichia coli (EPEC), originally isolated from humans and previously shown to cause diarrhea in human volunteers by unknown mechanisms, and one rabbit EPEC strain were shown to attach intimately to and efface microvilli and cytoplasm from intestinal epithelial cells in both the pig and rabbit intestine. The attaching and effacing activities of these EPEC were demonstrable by light microscopic examination of routine histological sections and by transmission electron microscopy. It was suggested that intact colostrum-deprived newborn pigs and ligated intestinal loops in pigs and rabbits may be useful systems to detect EPEC that have attaching and effacing activities and for studying the pathogenesis of such infections. The lesions (attachment and effacement) produced by EPEC in these systems were multifocal, with considerable animal-to-animal variation in response to the same strain of EPEC. The EPEC strains also varied in the frequency and extent of lesion production. For example, three human EPEC strains usually caused extensive lesions in rabbit intestinal loops, whereas two other human EPEC strains usually did not produce lesions in this system.     

Colonization of porcine small intestine by Escherichia coli: ileal colonization and adhesion by pig enteropathogens that lack K88 antigen and by some acapsular mutants.

Seven K88-negative porcine enteropathogenic Escherichia coli, representing three different serogroups, caused severe diarrhea and characteristically colonized the ileum, but not the jejunum, of intragastrically exposed newborn pigs. Bacterial counts of intestinal contents and wall, fluorescence, and scanning electron microscopy all suggested that these strains colonized the ileum by adhesion to the villous epithelium. However, in ligated intestinal loops, these enteropathogenic E. coli strains adhered to jejunal epithelium as well as to ileal epithelium. Acapsular (K-) mutants, derived from one of the principal strains, retained their colonizing and adhesive abilities, whereas K- mutants from three other enteropathogenic E. coli strains did not. It is suggested that: (i) these K88-negative enteropathogenic E. coli colonize the ileum by adhesion, and (ii) the adhesion of some K-88-negative strains is mediated by surface factors other than, or in addition to, the polysaccharide K antigen.     

Age-specific colonization of porcine intestinal epithelium by 987P-piliated enterotoxigenic Escherichia coli.

Neonatal (less than 1-day-old), 3- and 7-day old, and older (3-week-old postweaning) pigs were challenged by intragastric inoculation with 987P-piliated (987P+) enterotoxigenic Escherichia coli (ETEC) 987. Neonatal pigs were colonized (i.e., there were greater than or equal to 10(8) CFU of test strain per 10-cm ileal segment) and developed diarrhea. Intestinal colonization and the incidence and severity of diarrhea were lower in 3- and 7-day old pigs than in neonates. Older pigs were not colonized and did not develop diarrhea following oral inoculation with five strains of 987P+ ETEC. Strain 987 (987P+) adhered in vitro to intestinal epithelial cell brush borders isolated from both neonatal (sensitive) and older (resistant) pigs. The in vivo growth and expression of 987P pilus by strain 987 in ligated ileal loops created in neonatal and older pigs were similar. The in vivo adherence of 987P+ ETEC to intestinal epithelium in ligated ileal loops in neonatal and older pigs was compared. In neonatal pigs, most of the bacteria were in layers associated with the villous epithelium. In older pigs, most of the bacteria were associated with mucus-like material in the intestinal lumen. We concluded that swine develop an innate resistance to 987P+ ETEC by 3 weeks of age. This resistance does not appear to be due to an absence of 987P-specific receptors in the intestines of the older pig or to an inability of 987P+ bacteria to grow and express pili in the older pig. We hypothesized that the resistance of older pigs to 987P-mediated disease is due to release of 987P-specific receptors into the intestinal lumen, where these receptors facilitate bacterial clearance rather than bacterial adherence to intestinal epithelium and colonization.     

Differential Regulation of Enteric and Systemic Salmonellosis by slyA

Mutation of slyA, which reduces Salmonella typhimurium virulence in mice, caused only minor attenuation of S. typhimurium virulence in orally inoculated calves. This correlated with modest reductions in intestinal invasion and enteropathogenic responses in bovine ligated ileal loops. slyA appears to regulate virulence genes involved in systemic, but not enteric, salmonellosis.     

Use of the K88 antigen for in vivo bacterial competition with porcine strains of enteropathogenic Escherichia coli.

Infant mice were used to measure the amount of fluid accumulation (enterosorption) in the intestinal tract after oral inoculation of a porcine strain of enteropathogenic Escherichia coli (K88-+, Ent-+). Significant reduction in the amount of fluid found in the intestinal tract was observed if the mice were first inoculated with a K88-possessing, non-enterotoxigenic strain of E. coli. The protection provided is thought to be due to specific competition for attachment sites on cells of the small intestine.     

Susceptibility of porcine intestine to pilus-mediated adhesion by some isolates of piliated enterotoxigenic Escherichia coli increases with age.

Two porcine isolates of enterotoxigenic Escherichia coli (ETEC) (serogroup O157 and O141) derived from fatal cases of postweaning diarrhea and lacking K88, K99, F41, and 987P pili (4P- ETEC) were tested for adhesiveness to small-intestinal epithelia of pigs of different ages. Neither strain adhered to isolated intestinal brush borders of newborn (1-day-old) pigs in the presence of mannose. However, mannose-resistant adhesion occurred when brush borders from 10-day- and 3- and 6-week-old pigs were used. Electron microscopy revealed that both strains produced fine (3.5-nm) and type 1 pili at 37 degrees C but only type 1 pili at 18 degrees C. Mannose-resistant in vitro adhesion to brush borders of older pigs correlated with the presence of fine pili. These strains produced predominantly fine pili in ligated intestinal loops of both older and newborn pigs, but adherence was greater in loops in older pigs. Immunoelectron microscopic studies, using antiserum raised against piliated bacteria and absorbed with nonpiliated bacteria, of samples from brush border adherence studies revealed labelled appendages between adherent bacteria and intestinal microvilli. Orogastric inoculation of pigs weaned at 10 and 21 days of age indicated significantly (P less than 0.001) higher levels of adhesion by the ETEC to the ileal epithelia of older pigs than to that of younger ones. We suggest that small-intestinal adhesion and colonization by these ETEC isolates is dependent on receptors that develop progressively with age during the first 3 weeks after birth. Furthermore, our data are consistent with the hypothesis that the fine pili described mediate intestinal adhesion by the 4P- ETEC strains studied.     

Failure to detect conventional enterotoxins in classical enteropathogenic (serotyped) Escherichia coli strains of proven pathogenicity.

Ammonium sulfate-precipitated supernatants of classical enteropathogenic Escherichia coli strains were negative when investigated for enterotoxin production in rabbit ligated ileal loops, rabbit skin vascular permeability factor tests, suckling mice, and Y-1 adrenal cells. They also failed to stimulate guanylate cyclase activity in homogenates of rabbit, rat, and infant mouse intestines. Furthermore, DNA from enteropathogenic E. coli lacked sequences that encode heat-labile and heat-stable enterotoxins. These studies fail to show conventional enterotoxin synthesis by classical enteropathogenic E. coli.     

Lesions produced by Clostridium butyricum strain CB 1002 in ligated intestinal loops in guinea pigs

Heated spores (80 degrees C, 10 min) of Clostridium butyricum strain CB 1002 isolated from a fatal case of necrotising enterocolitis in a human neonate were inoculated into ligated intestinal loops prepared in young conventional guinea pigs. Necropsy findings 18 h later included congestion, patchy haemorrhage of the intestinal mucosa and bacteraemia. No abnormalities were observed in control loops given inocula of inactivated spores (heated at 100 degrees C for 10 min) or TYG 6 medium. The results suggest that vascular lesions are produced by C. butyricum in the intestine of young conventional guinea pigs.     

Virulence inhibition by zinc in shiga-toxigenic Escherichia coli

Previously, our laboratories reported that zinc inhibited expression of several important virulence factors in enteropathogenic Escherichia coli (EPEC) and reduced EPEC-induced intestinal damage in vivo. Since EPEC is genetically related to Shiga-toxigenic E. coli (STEC), we wondered whether the beneficial effects of zinc extended to STEC as well. Treatment options for STEC infection are very limited, since antibiotics tend to exacerbate disease via enhanced toxin production, so a safe intervention for this infection would be welcome. In this study, we report that in STEC strains zinc inhibits adherence to cultured cells as well as expression of EHEC secreted protein A (EspA). In addition, zinc inhibits the expression of Shiga toxin (Stx) at both the protein and the RNA level. Zinc inhibits basal and antibiotic-induced Stx production and inhibits both Stx1 and Stx2 by ≥90% at a concentration of 0.4 mM zinc. Rabbit EPEC strains were selected for acquisition of Stx-encoding bacteriophages, and these rabbit STEC strains (designated RDEC-H19A and E22-stx2) were used to test the effects of zinc in vivo in ligated rabbit intestinal loops. In vivo, zinc reduced fluid secretion into loops, inhibited mucosal adherence, reduced the amount of toxin in the loops, and reduced STEC-induced histological damage (villus blunting). Zinc has beneficial inhibitory effects against STEC strains that parallel those observed in EPEC. In addition, zinc strongly inhibits Stx expression; since Stx is responsible for the extraintestinal effects of STEC infection, such as hemolytic-uremic syndrome (HUS), zinc might be capable of preventing severe sequelae of STEC infection.     

Biological relationship between F18ab and F18ac fimbriae of enterotoxigenic and verotoxigenicEscherichia colifrom weaned pigs with oedema disease or diarrhoea

Comparative fimbrial expression and adhesion studies were made on enterotoxigenic and verotoxigenicE. coli(ETEC and VTEC) strains isolated from cases of porcine postweaning diarrhoea or oedema disease. F107(F18ab) fimbriae—monitored by polyclonal and monoclonal antibodies and by electron microscopy—were poorly expressed on most VTEC strains. In contrast, 2134P(F18ac) fimbriae were more readily detected on most ETEC strains. The F18ac strains adheredin vivoto ligated intestinal loops in weaned pigs while the F18ab strains did not adhere or adhered weakly. Similarly, the F18ac strains adhered to isolated intestinal brush borders in weaned pigs but the F18ab strains (except for the F107 referenceE. coli) did not adhere or adhered weaklyin vitro. Neither the F18ab nor F18ac strains adhered to brush borders from newborn pigs.In vitroadhesion of F18ab and F18ac strains was mannose resistant and receptors for F18 seemed to differ from receptors for K88(F4). It is concluded that the antigenic variants of F18 fimbriae (F18ab and F18ac) are biologically distinct. F18ab fimbriae are expressed poorly bothin vitroandin vivoand are frequently linked with the production of SLT-IIv and serogroup O139, while F18ac are more efficiently expressedin vitroandin vivoand most often are linked with enterotoxin (STa, STb) production, and serogroups O141, O157.     

Evaluation with an iuc::Tn10 mutant of the role of aerobactin production in the virulence of Shigella flexneri.

To evaluate the role of aerobactin production in the virulence of Shigella flexneri, a iuc::Tn10 insertion mutant was obtained from strain M90T, a serotype 5 isolate. This mutant was tested for its ability to invade and kill HeLa cells in monolayers, to elicit keratoconjunctivitis in guinea pigs and to infect ligated segments of rabbit ileal loops. Although this mutant did not grow in iron-depleted media, its ability to grow intracellulalry and eventually kill HeLa cells was unchanged from that of the wild-type strain. On the other hand, an inoculum-dependent effect was observed in the Sereny test, as well as in the rabbit ligated ileal loop model, which was monitored for fluid production and for both gross and microscopical alterations of the mucosa. Transduction of the mutation within a noninvasive plasmidless derivative of the parental strain did not alter growth within the intestinal lumen. We conclude that aerobactin production most probably provides invasive strains with a selective advantage for growth within tissues when located in extracellular compartments.     

Elicitation of enteroluminal neutrophils by enterotoxigenic and nonenterotoxigenic strains of Escherichia coli in swine.

In intact neonatal piglets, two strains of enterotoxigenic Escherichia coli (ETEC), which could adhere to epithelial cells and thus colonize the small intestine, attracted greater numbers of neutrophils into the lumen and wall of the intestine than did a nonenteropathogenic strain of E. coli. Ligated loops of small intestine in 8-week-old pigs were used in attempts to identify the attributes of ETEC involved in stimulating an increased enteroluminal migration of neutrophils. A nonenteropathogenic strain of E. coli did not attract neutrophils into the intestinal lumen in this model. However, three of the five ETEC strains tested did so. The three positive strains all produced heat-stable enterotoxin type b (STb). Neither of the negative ETEC strains produced STb. An STb-containing culture supernatant prepared from a strain of E. coli which contained an STb plasmid did not attract significantly more neutrophils than did a control supernatant prepared from the same strain of E. coli without the plasmid. The ETEC strains which attracted neutrophils in loops did not associate intimately with loop villi more consistently, nor did they grow to higher numbers in loops than strains which did not. It was concluded that there are increased numbers of neutrophils in the intestinal lumen during ETEC infection of newborn pigs. However, attempts to identify the attribute(s) of ETEC responsible for eliciting enteroluminal neutrophils were not successful.     

Characterization of plasmids that encode for the K88 colonization antigen.

K88 antigen, and important virulence factor in porcine enteropathogenic Escherichia coli (EEC), can be transferred along with the ability to ferment the trisaccharide raffinose (Raf). The plasmids from a number of EEC strains that encode these two properties were isolated and characterized. In most strains the K88 and Raf genes were found on a single nonconjugative plasmid approximately 50 x 10(6) daltons in size. This plasmid core was conserved with only slight variation among the strains tested. In some transconjugants, larger conjugative plasmids were observed that were apparently recombinants between the Raf/K88 plasmid and a transfer fa(tor. Occasionally plasmids carrying only the raffinose fermentation genes arose by deletion of a deoxyribonucleic acid segment of about 20 x 10(6) daltons that included the K88 antigen gene(s).     

Synergistic Effects of Clostridium perfringens Enterotoxin and Beta Toxin in Rabbit Small Intestinal Loops

The ability of Clostridium perfringens type C to cause human enteritis necroticans (EN) is attributed to beta toxin (CPB). However, many EN strains also express C. perfringens enterotoxin (CPE), suggesting that CPE could be another contributor to EN. Supporting this possibility, lysate supernatants from modified Duncan-Strong sporulation (MDS) medium cultures of three CPE-positive type C EN strains caused enteropathogenic effects in rabbit small intestinal loops, which is significant since CPE is produced only during sporulation and since C. perfringens can sporulate in the intestines. Consequently, CPE and CPB contributions to the enteropathogenic effects of MDS lysate supernatants of CPE-positive type C EN strain CN3758 were evaluated using isogenic cpb and cpe null mutants. While supernatants of wild-type CN3758 MDS lysates induced significant hemorrhagic lesions and luminal fluid accumulation, MDS lysate supernatants of the cpb and cpe mutants caused neither significant damage nor fluid accumulation. This attenuation was attributable to inactivating these toxin genes since complementing the cpe mutant or reversing the cpb mutation restored the enteropathogenic effects of MDS lysate supernatants. Confirming that both CPB and CPE are needed for the enteropathogenic effects of CN3758 MDS lysate supernatants, purified CPB and CPE at the same concentrations found in CN3758 MDS lysates also acted together synergistically in rabbit small intestinal loops; however, only higher doses of either purified toxin independently caused enteropathogenic effects. These findings provide the first evidence for potential synergistic toxin interactions during C. perfringens intestinal infections and support a possible role for CPE, as well as CPB, in some EN cases.     

Does enteropathogenic Escherichia coli produce heat-labile enterotoxin, heat-stable enterotoxins a or b, or cholera toxin A subunits?

Although most enteropathogenic Escherichia coli strains do not produce recognized enterotoxins, we wished to examine whether they produce any factors like heat-stable enterotoxin b or cholera toxin active subunits that might be missed by conventional assay methods. E. coli strains E851 (O142) and E2348 (O127) that had caused diarrhea in volunteers were negative for heat-labile enterotoxin and heat-stable enterotoxin a in Chinese hamster ovary cell and suckling mouse assays, failed to cause secretion in ligated small bowel loops from 6- to 8-week-old pigs after 4 to 5 h (used to show heat-stable enterotoxin b), and did not activate adenylate cyclase in pigeon erythrocyte lysates (used to demonstrate cholera toxin A subunit). We conclude that crude, unconcentrated culture filtrates and sonicates do not mimic heat-labile or heat-stable enterotoxins or cholera toxin or its A subunit and that enteropathogenic strains of E. coli probably have yet another mechanism or group of mechanisms by which they cause diarrhea.     

Attaching and effacing Escherichia coli isolates from Danish children: clinical significance and microbiological characteristics

This study describes the prevalence, clinical manifestations and microbiological characteristics of attaching and effacing Escherichia coli isolates, i.e., enteropathogenic E. coli (EPEC) belonging to the classical EPEC serotypes, non-EPEC attaching and effacing E. coli (A/EEC) and verocytotoxin-producing E. coli (VTEC), isolated in a case-control study of Danish children aged <5 years. Among 424 children with diarrhoea and 866 healthy controls, EPEC and VTEC were more prevalent in cases (2.4% and 2.6%, respectively) than in controls (0.7% and 0.7%, respectively). There was a high frequency of A/EEC isolates (n = 121), but these were equally prevalent in cases (11.3%) and controls (12.5%), and comprised a heterogeneous distribution of O:H serotypes. The intimin (eae) subtypes in A/EEC isolates showed an even distribution; the eae-gamma subtype predominated in classical EPEC cases. The virulence genes encoding the bundle-forming pilus (bfpA) and enteroaggregative heat-stable enterotoxin (astA) were rare among all isolates, and seemed to be of limited pathogenic importance in this population. Virulence characterisation of A/EEC isolates did not reveal any significant differences between cases and controls. Colonisation of children with A/EEC was associated with contact with sheep or goats (OR 2.2). The role of A/EEC, not being VTEC or belonging to the classical EPEC serotypes, requires further clarification, but serotyping is useful in discriminating between EPEC and A/EEC strains.     

Prevalence of Pilus Antigens, Enterotoxin Types, and Enteropathogenicity Among K88-Negative Enterotoxigenic Escherichia coli from Neonatal Pigs

Enterotoxigenic Escherichia coli (ETEC) that were isolated from neonatal pigs and that did not react in preliminary tests for pilus antigen K88 were subjected to additional tests for K88 and for pilus antigens K99 and 987P. Four such isolates produced K88, 9 isolates produced K99, 55 isolates produced 987P, and the remaining 43 isolates produced none of the three pilus antigens (3P⁻). Immunofluorescence tests of ileal sections from pigs were more sensitive for 987P detection than was serum agglutination of bacteria grown from the ileum. Most ETEC that produced K88, K99, or 987P were enteropathogenic (adhered to ileal villi, colonized intensively, and caused profuse diarrhea) when given to neonatal pigs. In contrast, only 3 of the 43 ETEC that produced none of the pilus antigens were enteropathogenic. The isolates were also tested for the type of enterotoxin produced. The K88⁺ isolates all produced heat-labile enterotoxin (LT) detectable in cultured adrenal cells (i.e., were LT⁺). None of the 987P⁺, K99⁺, or enterpathogenic 3P⁻ isolates produced LT. However (except for a single K99⁺ isolate), they all produced heat-stable enterotoxin detectable in infant mice (STa⁺). Sixteen isolates produced neither LT nor STa but did produce enterotoxin detectable in ligated intestinal loops of pigs (STb). Most of these LT⁻ STa⁻ STb⁺ isolates were also K88⁻, K99⁻, and 987P⁻ and non-enteropathogenic. One of them was K99⁺ and enteropathogenic. Our conclusions are as follows. (i) Most enteropathogenic ETEC from neonatal pigs produce either K88, 987P, or K99; however, there are some that produce none of the three antigens. (ii) Immunofluorescence tests for pilus antigens produced in vivo are recommended for the diagnosis of ETEC infections. (iii) Reports of LT⁻ STa⁻ STb⁺ swine ETEC are confirmed; furthermore, such isolates can be enteropathogenic.     

Biological evaluation of a methanol-soluble, heat-stable Escherichia coli enterotoxin in infant mice, pigs, rabbits, and calves.

Escherichia coli P16 was shown to produce two heat-stable toxins (ST) with differing biological activity. The toxins were separated by methanol extraction, and the first, STa, was methanol soluble, partially heat stable, active in neonatal piglets (1 to 3 days old) and infant mice, but inactive in weaned pigs (7 to 9 weeks old); the second, STb, was methanol insoluble, active in weaned pigs and rabbit ligated loops, but inactive in infant mice. It is therefore suggested that use of suckling mice as indicators of ST production will fail to identify certain ST-producing strains.