Adherence of an enterotoxigenic Escherichia coli strain, serotype O78:H11, to purified human intestinal brush borders.

The human enterotoxigenic Escherichia coli pathogen designated H10407 expresses two different types of surface pili, one designated type 1 pili and the other designated colonization factor antigen I (CFA/I), CFA/I pili are thought to promote the adherence of H10407 to the mucosa of the human small bowel. H10407 was grown under conditions which promoted the expression of either type 1 pili or CFA/I pili, and in each case, the adherence of H10407 to purified human intestinal brush borders was quantitated. The adherence assays revealed that H10407 adhered to human brush borders only when it expressed CFA/I pili. It appears that in vitro adherence of H10407 to human intestinal epithelial cells is dependent on the expression of CFA/I.     

Adhesion of enterotoxigenic Escherichia coli to human small intestinal enterocytes.

An improved enterocyte adhesion assay has been used to examine a collection of 44 strains of enterotoxigenic Escherichia coli (ETEC) for their ability to adhere to the brush border of isolated human duodenal enterocytes. Fourteen strains showed good adhesion; in each case the ability to adhere correlated with the production of colonization factor antigen I or II (CFA/I or CFA/II) fimbriae. CFA/II-positive producing coli surface antigens 1 and 3 (CS1 and CS3), coli surface antigens 2 and 3 (CS2 and CS3), and only coli surface antigen 3 (CS3) each showed good adhesion. CS3-mediated brush border attachment of CFA/II-positive ETEC was demonstrated by electron microscopy with monospecific antibody and an immunogold labeling technique. One CFA/I-positive ETEC strain was nonadherent in the assay, as were ETEC producing type 1 somatic fimbriae. Five animal ETEC strains producing K88, K99, F41, and 987P fimbriae were slightly more adhesive than control strains, but adhesion was significantly less than that of CFA-positive ETEC. Twenty five human ETEC strains that lacked CFA/I and CFA/II were nonadherent, suggesting either that the surface antigens responsible for adhesion to human intestinal mucosa in these strains were not being produced or that mucosal receptors for these strains are present in regions of the small intestine other than the duodenum.     

Characteristics of adherence of enteroaggregative Escherichia coli to human and animal mucosa.

An Escherichia coli strain (serotype O127a:H2) that had been isolated from a child with diarrhea in Thailand and that was negative for the virulence factors of the four categories of diarrheagenic E. coli (enterotoxigenic, enteropathogenic, enteroinvasive, and enterohemorrhagic) and that showed an aggregative pattern of adherence to HeLa cells was investigated for adherence to native or Formalin-fixed human and animal mucosa. The hemagglutinating activity and adherence ability of the bacteria were resistant to D-mannose and were strictly regulated by environmental conditions. Genetic data supported the close relation between the hemagglutinating activity and adherence ability. In accordance with the adherence pattern on tissue-cultured cells, the bacteria adhered to human and animal mucosa, as evidenced by a direct gold-labeling analysis. In human intestines, Formalin-fixed mucous coatings, epithelial cells of colonic mucosa, epithelial cells of ileal single lymphoid follicles and Peyer's patches, and the absorptive cells of jejunal or ileal villi provided adherence targets. Adherence to M cells in the Peyer's patch-associated epithelium was also confirmed. The adherence levels to native jejunal or ileal human villi were low, as was the case with the corresponding Formalin-fixed villi. In human urinary tract, the superficial epithelial cells of both native and Formalin-fixed ureter provided striking adherence targets. In animal (porcine and rabbit) small intestines, the bacteria adhered to the native villi to a lesser extent than to the Formalin-fixed villi. The adherence levels were compared with those of enterotoxigenic E. coli with colonization factor antigen (CFA)/I pili or CFA/II pili. The data suggested unique mucosa adherence characteristics of the enteroaggregative E. coli strain. The possibility of the adherence ability as a virulence factor was discussed.     

Detection of enterotoxigenic Escherichia coli colonization factor antigen I in stool specimens by an enzyme-linked immunosorbent assay.

An enzyme-linked immunosorbent assay (ELISA) was employed to detect and quantitate the fimbrial colonization factor antigen (CFA/I) of enterotoxigenic Escherichia coli in stool specimens obtained from adult cases of diarrhea in which CFA/I-positive E. coli was the known causative agent. The inhibition method, or blocking technique, was used. In this method, a standardized dilution of human anti-CFA/I serum was preincubated with dilutions of stool extract before transfer to CFA/I-coated microtiter plate wells, and then ELISA was performed with alkaline phosphatase-conjugated anti-human immunoglobulin. CFA/I purified from E. coli strain H-10407 (O78:H11) was used. Acute-phase diarrheal stool specimens were found to contain approximately 3.0 mg of antigen (mean value) per g stool, whereas control (CFA/I-negative) specimens contained insignificant amounts (less than 0.03 mg/g) of antigen. Also, CFA/I was detected in culture fluids of CFA/I positive enterotoxigenic E. coli belonging to a variety of serotypes and was undetectable in similar preparations from P-strains (spontaneous CFA/I-negative derivatives) of the same test cultures. Equivalent results were obtained in ELISA tests by using bacterial cells taken from isolated colonies grown on CFA agar. These results indicate that the ELISA technique will be useful for the diagnosis of diarrhea caused by CFA/I-positive enterotoxigenic E. coli.     

Hemagglutination of Human Group A Erythrocytes by Enterotoxigenic Escherichia coli Isolated from Adults with Diarrhea: Correlation with Colonization Factor

Enterotoxigenic Escherichia coli (ETEC) of several different serotypes isolated from adults with diarrhea and known to possess the colonization factor antigen (CFA) were found to cause mannose-resistant hemagglutination (HA) of human group A erythrocytes. CFA-negative E. coli isolated during the same study did not possess the mannose-resistant hemagglutinin, although some non-ETEC, CFA-negative isolates did exhibit mannose-sensitive HA activity. The mannoseresistant hemagglutinin of ETEC was found to possess many characteristics previously associated with CFA, which is a surface-associated fimbriate heatlabile antigen, and the functionally and morphologically similar K88 and K99 antigens of animal-specific ETEC. Mannose-resistant HA and CFA titers were maximal when ETEC cells were grown on an agar medium (CFA agar) composed primarily of 1% Casamino Acids and 0.15% yeast extract, pH 7.4. Neither CFA nor HA were produced at a growth temperature of 18 degrees C; HA was completely inhibited by pretreatment of CFA-positive cells with the anti-CFA serum. The mannose-resistant hemagglutinin was lost spontaneously and simultaneously with CFA when clinical ETEC isolates were passaged on artificial medium in the laboratory, indicating plasmid control of both entities. The mannose-resistant hemagglutinin of ETEC was shown to be thermolabile, i.e., sensitive to heating at 65 degrees C, as was the CFA. Also, there was correlation between possession of CFA, as detected serologically and by demonstration of biological activity (adherence in the infant rabbit small intestine), presence of CFA-type fimbriae, and the ability of various E. coli isolates to cause mannose-resistant HA of human group A erythrocytes. These results indicate that the mannose-resistant HA of ETEC is another manifestation of CFA.     

Glucose up-regulates expression of the differentiation-associated brush border binding site for enterotoxigenic Escherichia coli colonization factor antigen I in cultured human enterocyte-like cells.

The association of enterotoxigenic Escherichia coli expressing colonization factor antigen I (CFA/I) with the cultured human colon adenocarcinoma cell, a model of the mature enterocyte of the small intestine, is dependent on the binding of CFA/I to a brush border-associated component. Binding of the purified radiolabeled [125I]CFA/I- and 14C-labeled CFA/I-positive bacteria could be displaced by an increasing concentration of unlabeled CFA/I. Moreover, we showed that expression of the specific CFA/I binding developed as a function of cell differentiation in Caco-2 cells, whereas expression of the nonspecific binding did not. Expression of the brush border differentiation-associated component acting as a binding site for CFA/I was up-regulated by glucose. Indeed, the enterocyte-like HT-29 glc- cell subpopulation not expressing the CFA/I binding site when cultured in dialyzed serum and hexose-free medium regained the ability to bind CFA/I when the cells were returned to culture medium containing glucose. Furthermore, expression of the brush border-associated CFA/I binding site in the enterocyte-like Caco-2 cells was repressed when the cells were cultured in hexose-free conditions.     

Differences in hydrophobic surface characteristics of porcine enteropathogenic Escherichia coli with or without K88 antigen as revealed by hydrophobic interaction chromatography.

Porcine enteropathogenic Escherichia coli strains possessing or lacking K88 antigen were studied by using hydrophobic interaction chromatography on cross-linked agarose gels with alkyl or aryl substituents (amphiphilic gels) to determine whether or not they possessed surface-associated hydrophobic properties. Strains with K88ab or K88ac antigen adsorbed to phenyl and octyl Sepharose gels in the presence of 4 M sodium chloride. This property correlated with phenotypic expression of K88 antigen. Cells grown at 37 degrees C but not those grown at 18 degrees C possessed hydrophobic adsorptive characteristics in addition to the property of mannose-resistant hemagglutination of guinea pig erythrocytes. Adsorption of K88-positive strains to gels with hydrophobic ligands was independent of O group and enterotoxicity. Strains lacking K88 antigen did not adsorb to the hydrophobically substituted derivatives of Sepharose and lacked mannose-resistant hemagglutinating characteristics. Neither the presence of additional polysaccharide K antigens nor nonhemagglutinating pili conferred the property of hydrophobic interaction on the strains. K88-positive bacteria had a lower electrophoretic migration rate than did K88-negative bacteria of the same serotype in free-zone electrophoresis. K88-positive bacteria also adsorbed strongly to hydrophobic ligands in the presence of 1 M ammonium sulfate, whereas K88-negative strains did not. These observations provide evidence for the suspected role of hydrophobic interaction in the adhesive properties of certain enteropathogenic strains of E. coli. Moreover, hydrophobic interaction chromatography provides convenient and rapid alternative means of screening strains for a property potentially associated with adhesiveness.     

Immunity to Escherichia coli in Pigs: Adhesion of Enteropathogenic Escherichia coli to Isolated Intestinal Epithelial Cells

A method was developed to test for the ability of Escherichia coli to adhere to isolated intestinal epithelial cells. Of the E. coli tested, those having either K88ac or K88ab antigens adhered to the cells, and those which did not have these antigens did not. Since some enteropathogenic E. coli did not have the ability to adhere, it is assumed that adherence is not an essential factor of pathogenesis but rather should be considered an enhancement to the pathogenicity of some E. coli. None of the E. coli enteropathogens of cattle tested adhered to either pig or cattle cells. Similarly, human strains did not adhere to pig cells. Although the test system may not have been ideal for human or bovine E. coli, the results reported here suggest that adhesiveness is a property limited to porcine enteropathogenic E. coli carrying one of the K88 antigens. Adhesiveness is associated with the K88c or K88b antigens, and their adhesive ability is only neutralizable by the homologous antisera.     

Prediction of antigenic determinants and secondary structures of the K88 and CFA1 fimbrial proteins from enteropathogenic Escherichia coli.

Recently the amino acid sequences of the K88ab and the CFA1 fimbrial proteins from enteropathogenic Escherichia coli have become available. To elucidate the immunological interrelationships among the K88 family of antigenic variants, and to estimate the nature and position of the antigenic determinants in the K88 and CFA1 proteins, the positions of these features in the amino acid sequences have been predicted. This has been done by computerized algorithms, incorporating such factors as hydrophilicity and secondary structural potential. Some of the predicted determinants of K88 show good correlation with the known sequence differences between the antigenic variants of this protein. The presented results point to a possible use of synthetic vaccines against fimbriated pathogenic E. coli strains.     

Nonimmunoglobulin fraction of human milk inhibits bacterial adhesion (hemagglutination) and enterotoxin binding of Escherichia coli and Vibrio cholerae.

Human milk and colostrum samples were divided into an immunoglobulin and a nonimmunoglobulin fraction by immunosorbent chromatography. The ability of these fractions to inhibit bacterial cell adhesion and enterotoxin receptor binding of Vibrio cholerae and various Escherichia coli isolates was then tested by in vitro assays. The strongest effect was generally seen with the nonimmunoglobulin fractions, which were shown to significantly inhibit E. coli cell adhesion (hemagglutination) mediated by CFA/I, CFA/II, or K88 fimbriae (but not type 1 pili) and V. cholerae hemagglutination, as well as the binding of cholera toxin and E. coli heat-labile enterotoxin to GM1 ganglioside. Also, the immunoglobulin fractions had significant inhibitory activity in some of these systems. The results are interpreted to suggest that human milk and colostrum may contain secreted structure analogs of the cell receptors for some bacterial adhesions and enterotoxins; this might contribute to the protective effect of milk against enteric infections.     

Prevalence of the enterotoxigenic Escherichia coli colonization factors CFA/I, CFA/II and E8775 isolated in the South Pacific (New Caledonia, Republic of Vanuatu and Wallis and Futuna)

We tested the expression of adherence properties of enterotoxigenic Escherichia coli (ETEC) strains isolated in New-Caledonia, Vanuatu and Wallis and Futuna by examining for the presence of colonization factor E8775 using an agglutination test and an immuno-diffusion technique with specific antisera. Approximately 19% of ETEC strains possessed CFA/I and 21% a CFA/II. The E8775 antigen was found on 1.8% of the strains. This last factor was found on strains of the serogroup 025 from Vanuatu. Two strains 078 usually CFA/I+ possessed a CFA/II and three strains of the serogroup 0126 possessed a CFA/I. The results of this study emphasis the need to continue the search for other mechanisms of adhesion used by ETEC strains without any of the three factors of colonization.     

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.     

Virulence properties and attaching-effacing activity of Escherichia coli O45 from swine postweaning diarrhea.

Escherichia coli O45 isolates associated with swine postweaning diarrhea in Québec were characterized with respect to virulence determinants genetically and investigated for their attaching and effacing (A/E) activities by experimental inoculation of gnotobiotic piglets and by the HEp-2 cell adherence assay. All of 32 isolates tested were negative for enterotoxigenic and verotoxigenic E. coli virulence determinants, heat-labile enterotoxin (LT), heat-stable enterotoxins (STap, STb), verotoxins (VT1, VT2), and F4 (K88), F5 (K99), F6 (987P), and F41, except one STb-positive and two F4-positive isolates. A total of 25 isolates hybridized with an EaeA probe, and 11 hybridized with an enteropathogenic E. coli adherence factor (EAF) probe. None of 32 isolates hybridized with a bundle-forming pilus (BFP) probe. The EAF, EaeA, and BFP factors have been associated with human enteropathogenic E. coli strains. A total of 10 of 12 eaeA-positive porcine O45 isolates induced A/E lesions characterized by intimate adherence of bacteria to the intestinal epithelial cell membrane with effacement of the microvilli, similar to those of human attaching-effacing E. coli. However, A/E lesions were not observed in the piglets inoculated with any one of three eaeA-negative O45 isolates. All E. coli O45 isolates were non-adherent to HEp-2 cells. Thus, we have demonstrated the production of typical A/E lesions by nonenterotoxigenic E. coli O45 isolates from swine postweaning diarrhea. The results indicate the significance of the eaeA gene in A/E activities of these isolates and suggest that EAF and BFP are not involved in O45 E. coli infection of weaning piglets.     

Rapid isolation of K88+ Escherichia coli by using immunomagnetic particles.

Superparamagnetic polymer particles precoated with sheep anti-mouse immunoglobulin G were coated with immunoglobulin G2 monoclonal antibodies to the K88 antigen of Escherichia coli (MAb-K88). These immunomagnetic particles (IMP) were used for isolation and identification of K88 antigen-positive (K88+) E. coli. The bacteria presenting the K88 antigen were easily isolated in almost pure culture from a mixed culture of five different O serogroups of E. coli. Nonspecific binding of K88 antigen-negative (K88-) E. coli to the IMP was not observed. The sensitivity of the test to detect K88+ E. coli was found to be 4,000 CFU/ml with fluorescence microscopy. When bacteria attached to the MAb-K88 IMP were grown on blood agar, about 20% of the initial number of CFU was recovered. The test is promising as a rapid method for isolation and identification of K88+ E. coli from a mixed culture.     

Differences in susceptibility of inbred and outbred infant mice to enterotoxigenic Escherichia coli of bovine, porcine and human origin

Infant mice from outbred Swiss OF1 and from inbred DBA/2, C57BL/6, BALB/cBy and CBA strains were screened for usefulness in the diarrhoea model with enterotoxigenic Escherichia coli (ETEC) strains of bovine, porcine and human origin. Mouse strains were either weakly susceptible or not susceptible to ETEC strains of porcine or human origin bearing antigen K88, 987P, CFA/I or CFA/II. In contrast, some mouse strains were highly susceptible to bovine and porcine ETEC strains bearing K99 or F41 or both antigens. Swiss OF1 and CBA infant mice were highly susceptible to one bovine ETEC strain bearing antigen K99, whereas DBA/2, BALB/cBy and C57BL/6 mice exhibited nearly complete resistance to the same ETEC strain. Except DBA/2, all mouse strains were highly susceptible to bovine and porcine ETEC strains bearing antigen F41 alone or in combination with antigen K99. Challenge ETEC strains colonised intestines of all infant mice, but they reached very high levels soon after inoculation in the diarrhoeic ones only.     

Behavior of Escherichia coli K antigens K88ab, K88ac, and K88ad in immunoelectrophoresis, double diffusion, and hemagglutination.

Porcine enteropathogenic Escherichia coli strains were found to possess a variant of the K88 antigen provisionally termed K88ad. We propose to include this antigen into the international E. coli typing scheme. Ultrasonic extracts of field strains of E. coli possessing the K88ab, K88ac, or K88ad antigen and their E. coli K-12 K88+ transconjugants showed a specific K88 precipitation line in immunoelectrophoresis and double diffusion only when grown at 37 degrees C, but not when grown at 18 degrees C. By using agarose gels, K88ab, K88ac, and K88ad antigens showed anodic mobility in immunoelectrophoresis. When using Difco Noble agar gels, K88ad was not mobile or anodic, K88ab was cathodic; K88ac of 17 strains was cathodic and of 24 strains was anodic. The immunoelectrophoretic behavior of a K88 antigen (K88ab, K88ac, or K88ad) did not alter after transfer of the corresponding plasmid to E. coli K-12. Anodic and cathodic K88ac antigens could not be distinguished serologically. The differences between the results obtained in Noble agar gels and agarose gels are due to electro-endosmotic flow. We describe a procedure which increases the detection level of K88+ transconjugants in a mating mixture. It is based on the specific mannose-resistant attachment of K88+ cells to guinea pig erythrocytes.     

A rns-like regulatory gene for colonization factor antigen I (CFA/I) that controls expression of CFA/I pilin.

Colonization factor antigens (CFA) are needed for adherence of human enterotoxigenic Escherichia coli (ETEC) strains to their hosts. The CFA/II antigens, CS1 and CS2, which are found in some ETEC strains, require the plasmid-encoded gene rns for expression (J. Caron, L. M. Coffield, and J. R. Scott, Proc. Natl. Acad. Sci. USA 86:963-967, 1989). Other ETEC strains express CFA/I, whose synthesis and assembly require genes on two unlinked regions (regions 1 and 2) of a plasmid (G. A. Willshaw, H. R. Smith, and B. Rowe, FEMS Microbiol. Lett. 16:101-106, 1983). We report that CFA/I region 2 DNA can substitute for rns to cause expression of CS1 and CS2. The cfaR gene in region 2 is defined by a mutation abolishing both expression of CFA/I and complementation of a rns mutant for expression of CS1 or CS2. In a strain containing only region 1, complementation for expression of CFA/I by a plasmid containing rns+ is inefficient but is adequate to cause hemagglutination by the CFA/I adhesin.     

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).     

Iron represses the expression of CFA/I fimbriae of enterotoxigenic E. coli.

Experiments were designed to study the effect of iron on the expression of CFA/I fimbriae by enterotoxigenic E. coli (ETEC). Addition of 0.05 mM ferrous sulfate to growth media decreased CFA/I antigen and fimbrial production by the CFA/I-positive ETEC strain H-10407 as measured by quantitative ELISA and hemagglutination assay. The repressive effect was reversed by the addition of the iron chelators, sodium citrate or dipyridyl. With a CFA/I subunit gene promoter-lacZ fusion, it was found that the activity of the subunit gene promoter was significantly higher in the presence of iron chelators than in medium containing iron in the fur+ strain DHB24. This difference was not observed in the fur mutant strain SBC24, suggesting that the global E. coli metalloregulatory protein Fur (ferric uptake regulation) is involved in the repression. The repressor may bind to the promoter of the CFA/I subunit gene since several potential Fur-binding sites were identified in the promoter area.     

Purification and characterization of the CFA/I antigen of enterotoxigenic Escherichia coli.

The fimbral colonization factor antigen CFA/I of enterotoxigenic Escherichia coli was purified and characterized. The initial purification step was release of these fimbriae from the bacterial cells by homogenization with a Waring blender. Common fimbriae and flagellar antigen were avoided by careful control of growth conditions and the use of a nonmotile (H-) mutant of the prototype strain H-10407 (O78:H11). The essential purification steps were membrane filtration (Millipore Corp.), ammonium sulfate fractionation, and negative diethylaminoethyl-Sephadex column chromatography. Yields were approximately 4.0 mg of CFA/I protein per g (wet weight) of bacteria. Purified CFA/I is a fimbrial molecule 7.0 nm in diameter and has an average molecular weight of 1.6 X 10(6), as determined by sedimentation equilibrium. CFA/I is a polymer of identical subunits of molecular weight 23,800 with an N-terminal valine, 37% hydrophobic amino acid residues, and 11 residues of proline per mol. The purified antigen retains its morphology, antigenicity, and biological activity. Purified antigen retains its morphology, antigenicity, and biological activity. Purified CFA/I exhibits mannose-resistant hemagglutination of human group A, bovine, and chicken erythrocytes, as do CFA/I-positive bacteria. This was demonstrated by sensitizing latex microbeads with the purified antigen since cell-free CFA/I fimbriae do not hemagglutinate erythrocytes. Thus, CFA/I detached from the bacteria are monovalent; however, purified CFA/I antigen retains an affinity for the epithelial cells of rabbit small intestine and blocks adhesion of CFA/I-positive bacteria. These results demonstrate that purified CFA/I is a good candidate for use in an oral vaccine for immunoprotection against diarrhea caused by CFA/I-positive enterotoxigenic E. coli.