Search

Enterotoxigenic Escherichia coli isolates obtained in Mexico from adult subjects with diarrhea and from healthy controls were examined for the production of heat-stable enterotoxin (ST) and heat-labile enterotoxin (LT) after serial passage in the laboratory. Isolates were found to be either stable for the production of ST and LT or unstable with respect to ST, LT, or both. Unilateral loss of either ST or LT production allowed classification of E. coli isolates into four groups according to stability/instability of enterotoxin production. Fewer serotypes, with more representative isolates, were in group I (stable) than in group IV (completely unstable). Isolates from Dacca, Bangladesh, could be similarly classified into stability groups. There is an apparent relationship between serotype, stability of enterotoxin production, particularly LT, and isolation from diarrhea cases as opposed to isolation from healthy controls.

The nature of the adhesive capacity of three hemagglutinating Escherichia coli strains that had earlier been described as nonfimbriated was studied. The strains that were isolated from human disease adhered to human buccal and urinary tract epithelial cells, an adhesion that was not inhibited by D-mannose. By crossed immunoelectrophoresis it was shown that the three strains produced a common antigen, Z1, developed after growth at 37 degrees C but not 18 degrees C. One of the strains produced an additional antigen, Z2, of almost the same electrophoretic mobility in crossed immunoelectrophoresis. A mutant of this strain deficient of its polysaccharide K antigen had maintained the adhesive capacity, indicating that the K antigen was not responsible for adhesion. A further mutant of the acapsular mutant produced a strongly reduced amount of the Z antigens and had lost the ability to adhere. The Z1 (and Z2?) antigens were therefore deemed to be responsible for adhesion. In sodium dodecyl sulfate-polyacrylamide gel electrophoresis of extracts of cells of the three strains, a heavy Coomassie-blue stained line was seen, indicating the presence of a protein subunit of molecular weight slightly above 14,400. By immunoblotting with absorbed antiserum, it was shown that this protein was the same as that detected by crossed immunoelectrophoresis. Protease from Streptomyces griseus, but not trypsin, digested the protein. Heating to 100 degrees C did not affect it. By immunoelectron microscopy of embedded and sectioned bacteria that had first been treated with specific antisera and ferritin-labeled antirabbit immunoglobulin, the protein adhesin-antibody complex was found to surround the bacteria as a heavy capsule. After negative staining with uranylacetate (pH approximately 4), the capsule appeared as a mesh of very fine filaments. The possible role of this capsule in the pathogenesis of disease is discussed.

The serotypes of 386 enterotoxigenic Escherichia coli (ETEC) isolated from 82 individuals with and without diarrhea in Thailand and the Philippines were determined. The 136 strains producing both heat-labile toxin (LT) and heat-stable toxin (ST) belonged to 12 different O serogroups; however, 83% (113/136) were of one of four serogroups (O6, O8, O25, and O78), and 76% of (104/136) belonged to one of seven O:K:H serotypes. Only 14% (28/196) of LT-only-producing ETEC belonged to serogroups most common among LT and ST strains, and these 196 strains belonged to 35 different O:K:H serotypes. Three O serogroups (O20, O27, and O78) accounted for 94% (52/54) of strains producing only ST. Although only 4% (2/54) of ST-only ETEC belonged to the seven serotypes most commonly found among strains which produced LT and ST, 85% of ETEC belonged to three other serotypes, O20:K?:H21, O27:K?:H7, and O78:H-. A total of 46% (37/80) of ETEC of serotypes O6:H16, O8:H9, O25:H42, and O78:H12 were resistant to two or more antibiotics in comparison to 68% (208/306) of ETEC of other serotypes (P less than 0.001). In Thailand and the Philippines, E. coli which produced LT and ST or ST alone, but not those which produced LT alone, were restricted in their O:K:H serotypes.

Fimbriae from three Escherichia coli strains, C1212, C1214, and C1023, isolated from urinary tract infections, have been purified and characterized by determination of the N-terminal sequences, amino acid composition, and molecular weights of their respective subunits. Furthermore, their immunological interrelationships have been investigated. The three strains all harbored more than one fimbrial species each. Immunologically different type 1-like fimbriae, termed 1A, 1B, and 1C, with highly homologous N-terminal sequences were isolated, of which strain C1212 possessed 1A and 1C, strain C1214 possessed 1A and 1B, and strain C1023 possessed 1A and 1C. Type 1A is known to cause a mannose-sensitive hemagglutination similar to that described for type 1 fimbriae, whereas the functions of types 1B and 1C are not yet known. Strain C1212, in addition, harbored the F7 fimbrial antigen which causes mannose-resistant hemagglutination and adherence to urinary epithelial cells. The N-terminal structure of this antigen seems to indicate a possible evolutionary kinship to the type 1-like fimbriae, although they are immunologically unrelated. Our results indicate that fimbriation of pathogenic wild-type strains can be of an intricate variety.

Bacterial agglutination and crossed immunoelectrophoresis were compared as techniques for the subdivision of mannose-resistant hemagglutinating Escherichia coli fimbrial antigens. A total of 22 fimbrial strains, 15 of which had earlier been grouped into two fimbrial agglutination groups, were examined for the presence of fimbrial antigens F7 go F12 by crossed immunoelectrophoresis. Most of the strains were isolated from infants with diarrhea; they were neither enteropathogenic nor enterotoxigenic serotypes. Of the 10 strains in agglutination group 1, 4 had fimbrial antigens F7 and 1C, 4 had antigens F8 and 1C, and in 1 strain, only antigen 1C was demonstrated. The tenth strain did not, by the crossed immunoelectrophoresis test, fit into agglutination group 1. Agglutination group 2 comprised five strains. Four of these had antigens in common with the still unnumbered fimbrial antigens of an O4:K12:H5 strain. In the fifth strain, no known F antigens were demonstrated. The previously found correlation between some O:K:H serotypes and fimbrial antigens in strains from urinary tract infections was confirmed in this study in strains from diarrhea cases. We concluded that, although the agglutination test can indicate the presence of certain fimbriae, it cannot be used presently for an exact demonstration of these antigens because each strain often produces several different fimbriae.

Sixty-four verotoxin-producing (VT+) Escherichia coli strains were analyzed for VT1- and VT2-specific DNA sequences and for production of hemolysin. Strains of human origin were of the following serotypes: O157:H7 or H-, O111:H8 or H-, O26:H11, O114:H4, and rough:H7. Strains of serotypes O157:H7, O113:H21, O116:H21, and rough:H- were from cattle, while those of serotype O139:K12:H1 were from pigs. All 64 isolates carried either VT1 or VT2 or both genes. Sixty of the strains (93.8%) were hemolytic (Hly+). The three O139:K12:H1 strains examined produced alpha-hemolysin, as shown by their reaction with the alpha-hemolysin-specific monoclonal antibody h2A and by DNA hybridization with an alpha-hly gene probe. The remaining 57 Hly+ strains (95%) produced a different type of hemolysin (enterohemolysin), which is genetically and serologically unrelated to alpha-hemolysin. The two types of hemolysin are further distinguished by the appearance of the lysis zone on blood agar and by the time interval for the detection of hemolysis. In contrast to alpha-hemolysin, enterohemolysin can be detected only on blood plates containing washed erythrocytes. The frequent association of enterohemolysin with verotoxin production (89%) makes it useful as an epidemiological marker for rapid and simple detection of potential VT+ E. coli.

The extent of chromosomal-gene diversity among 261 isolates of Escherichia coli sharing single O, K, or H antigens and various combinations thereof was estimated by multilocus enzyme electrophoresis, which detects allelic variation in structural genes. The results of this study indicate that the genetic diversity among isolates sharing single antigenic determinants can approach or equal that observed among randomly chosen strains; that the magnitude of the diversity varies among antigens; and that the genetic diversity is reduced, but not eliminated, among isolates sharing two antigenic determinants. With one exception, isolates of the same O:K:H serotype were of identical or closely related electrophoretic types (ETs). Isolates of the same ET generally shared the same combination of antigenic determinants, but some ETs included isolates of different serotypes. The implications of these findings for epidemiological research and the clone hypothesis of population structure are discussed, and possible evolutionary mechanisms causing antigenic divergence and convergence are considered.

Sixty-three Escherichia coli strains isolated from neonatal sepsis or meningitis were studied and compared with previous data on fecal or urinary pyelonephritis-associated isolates from children. Characteristics significantly associated with neonatal infection were capsular type K1 (54%), O group 18 (27%), rough-type lipopolysaccharide together with K1 capsule (19%), and S fimbriae (29%). Within the neonatal infection group, the K1 capsule and rough lipopolysaccharide were most common among the youngest infants (0 to 21 days old) and in meningitis. Hemolysin production, P fimbriae, and X adhesions (adhesions not identifiable as type 1, P, or S) were significantly more common in the two materials from infections as compared with the fecal isolates. One large clone of 11 strains (O18:K1:H7, with both type 1 and S fimbriae) and three smaller ones (O7:K1:H1 and O6:K2:H1, both with type 1 and P fimbriae and X adhesions; and R:K1:H33 with no adhesions) were identified among the strains from neonatal infections. Only O6:K2:H1 strains were also common among the strains from pyelonephritis.

K88 antigen-negative enterotoxigenic Escherichia coli and non-enterotoxigenic strains isolated from piglets with diarrhea were examined for K99 antigen by agglutination tests after growth on Minca-IsoVitaleX (BBL Microbiology Systems, Cockeysville, Md.) agar medium. Of 64 K88-negative enterotoxigenic strains from as many piglets, 17 were found to be K99 positive. Of these, 10 were Swedish and 7 were of Norwegian origin. All 17 produced heat-stable enterotoxin detectable in the infant mouse assay, but only 2 gave positive ligated loop tests in 3- to 7-week-old piglets. Ligated loop tests in 5- to 12-day-old piglets were positive for each of the 15 K99-positive strains tested. Each of the Swedish K99-positive isolates was from a piglet of lesser than or equal to 7 days of age. One piglet harboring a K99-positive strain also harbored an O141:K88 enterotoxigenic strain producing only heat-stable enterotoxin. Five of the Swedish piglets yielding K99-positive isolates were from dams vaccinated wih a multicomponent bacterial vaccine containing K88 antigen. The K99 strains were O:K:H serotyped. The O serogroups represented were O8, O9, O64, O101, and O140. None of the 101 non-enterotoxigenic porcine isolates, representing 42 serogroups and non-O-groupable and rough strains, was found to be K99 positive. The findings indicate that so-called class 2 or atypical porcine enterotoxigenic E. coli should be routinely examined for the presence of K99 antigen.

Variation in chromosomal DNA in Escherichia coli was studied with probes specific for the P-associated-pilus (pap) region. The presence of DNA homologous to pap was determined by dot blots. Variation in the number of copies of pap and in the organization of internal and flanking sequences was determined by Southern blot hybridization. The 229 strains studied were also classified by O:K:H serotyping and multilocus enzyme electrophoresis. There was considerable heterogeneity in the presence of pap and distribution of pap-homologous DNA in these E. coli strains from natural sources. In general, there was less variation in pap among strains of the same specific O:K:H serotype and enzyme electrophoretic type than among random isolates. There were, however, E. coli strains identified as members of the same clone by O:K:H serotyping and enzyme electrophoresis that were pap positive and pap negative or had different Southern blot patterns for the pap probes (pap type). There were also isolates of the same pap type that differed in two of three O:K:H serotype antigens and the majority of enzymes that determined their enzyme electrophoretic type. These latter two observations were interpreted as evidence for the horizontal (infectious) transfer of the pap-homologous sequences among clones of E. coli.

A total of 568 strains of Escherichia coli isolated from healthy and diarrheic rabbits were separated into 11 different biotypes according to the fermentation patterns of four carbohydrates. Strains belonging to biotypes 1 to 3, 6, and 8 induced lesions characteristic for attaching and effacing E. coli (AEEC). They attached to the intestinal epithelium of the terminal small intestine and the large intestine of 5-week-old rabbits after experimental infection and caused effacement of the microvillous brush border. However, pathogenicity for weaned rabbits, as judged by diarrhea score, anorexia, and reduced weight gain, varied according to the biotypes of the strains. Strains belonging to biotypes 1 and 6 produced only discrete clinical signs, strains belonging to biotypes 2 and 3+ (motile) induced diarrhea and growth depression, whereas strains belonging to biotypes 3- (immotile) and 8 caused severe clinical signs and high mortality. This confirms evidence from the field. Biotypes 3- and 8, accounting for 35.5 and 7.1% of AEEC strains in weaned diarrheic rabbits, respectively, were not detected in weaned healthy rabbits, while biotype 2 was the predominant strain in weaned healthy rabbits (62.3%). Finally, serotyping showed a close relationship between biotype and serotype of the AEEC examined. Most strains of biotypes 1+ and 2+ tested were O109:K-:H2 and O132:K-:H2, respectively, whereas all strains tested of biotype 3- were O15:K-:H- and those of biotype 8 were O103:K-:H2. These data indicate that specific clones of AEEC might be involved in juvenile rabbit enteritis. It was concluded that determination of biotypes allows the screening of highly pathogenic AEEC in weaned rabbits (biotypes 3- and 8).

The adhesive fimbrial antigen F12 from a strain of uropathogenic Escherichia coli has been isolated and characterized. The antigen was purified by ammonium sulfate precipitation and gel chromatography. The protein subunit of the F12 fimbria has a molecular weight of 18,200; the N-terminal amino acid sequence of the subunit shows close resemblance to that of the subunits of other F fimbriae and the type 1 fimbriae. We identified in these proteins a pattern of alternating conserved and variable amino acid residues which could indicate a special structural and functional feature.

Enterotoxigenic Escherichia coli (ETEC) isolates from 36 persons with acute traveler's diarrhea from whom no other pathogens were recovered were tested (after no more than three subcultures) for the presence of colonization factor antigens I and II (CFA/I and CFA/II) and type 1 somatic pili. CFA/I or CFA/II was identified in 7 of 10 strains with heat-labile and heat-stable enterotoxins (LT+/ST+), but in only 2 of 12 LT-/ST+ (P less than 0.05) and 0 of 14 LT+/ST- (P less than 0.02) strains. CFA pili were not found among 74 non-enterotoxigenic E. coli strains. Type 1 somatic pili were demonstrable in 42% of the 36 ETEC and in 49% of the 74 non-enterotoxigenic E. coli isolates. The nine ETEC isolates bearing a CFA were serially subcultured on 10 consecutive days and retested for CFA and toxin. After five subcultures only one strain had lost a CFA, but after 10 passages three strains were negative: two lost CFA/I and one lost CFA/II. The strain that lost CFA/II became negative for both LT and ST as well and was found to lack a 48- and a 60-megadalton plasmid. The two strains that lost CFA/I also became negative for ST, but plasmid analysis revealed no plasmid loss. Disappearance of the CFA/I phenotype without loss of a plasmid can be explained by phase variation, as exhibited by type 1 somatic pili, or by rearrangement of base sequences in the CFA/I plasmid genome. If purified pili vaccines are to provide broad-spectrum protection against ETEC diarrhea, the search must be intensified to identify the antigens responsible for adhesion to intestinal mucosa in the many ETEC strains that lack CFA/I and CFA/II.

The Escherichia coli K13, K20, and K23 capsular polysaccharide antigens are serologically related. All of these polysaccharides contain ribose and 2-keto-3-deoxyoctonate in equimolar quantities. The K13 and K20 polysaccharides are partially O-acetylated. A comparison of these polysaccharides after O-deacetylation, by nuclear magnetic resonance and permethylation analysis, showed that these polysaccharides contained the disaccharide repeat unit leads to)-beta-ribofuranosyl-(1 leads to 7)-beta-2-keto-3-deoxyoctonate. They differed in the presence and location of an acetyl moiety. The K13 polysaccharide was O-acetylated at C-4 of the 2-keto-3-deoxyoctonate. The K20 antigen was O-acetylated at C-5 of the ribose moiety. The K23 polymer was nonacetylated. The cross-reactivity of these antigens was demonstrated by tandem-crossed immunoelectrophoresis. Antibodies to K23 could be completely absorbed from OK K23 serum by K13, K20, and K23 antigenic extracts. The K13 and K20 antibodies could be completely absorbed from their respective antisera only by homologous antigenic extracts. Monoclonal antibodies were prepared against a protein conjugate of the K13 polysaccharide. Analyses of the reactions of these antibodies with the three polysaccharides suggest that the K13 polysaccharide has at least three antigenic sites, one of which is common to the K13, K20, and K23 polysaccharides.

Two hundred and ninety-four isolates of Escherichia coli, including 105 from blood cultures, 94 from stools of hospital inpatients, and 96 from rectal cultures of healthy young adults, were biotyped by using the API-20E system and tested for the presence of K1 antigen. The overall frequency of K1 strains was 14.2% and was similar among the three sources. Forty-eight biotypes were observed, but two-thirds of all isolates, including two-thirds of the K1 strains, belonged to only five biotypes. Among the five commonest biotypes, the distribution of K1 strains was nonrandom, since 23 of the 27 K1 strains belonged to only two biotypes. Analysis of the O and H antigens of K1 strains indicated that this correlation of biotype with K1 antigen was due to a restricted number of serovars ("clones") that were repeatedly isolated from the population studied. These serovas included O18:K1:H7, O1:121:H6 and O16:K1:H6. Although a statistically significant correlation between biotype and K1 antigen was observed, the correlation was not sufficiently great to alow biotyping to be of significant predictive value as a marker for the K1 antigen.

Forty-six Escherichia coli isolates of serotype O2:K1 from human urinary tract infections, chicken sepsis, and bovine mastitis were obtained from laboratories in England, Denmark, Sweden, and Finland. The bacteria were compared for outer membrane protein (OMP) pattern, lipopolysaccharide pattern, electrophoretic mobilities of enzymes, and flagellar serotype and were tested for fimbriation, biotype, hydroxamate production, hemolysin production, antibiotic resistance, plasmid content, colicin production, and virulence in neonatal rats. Isolates from humans were assigned to two clonal groups; poultry isolates belonged to one of these clonal groups, whereas bovine isolates belonged to the other. Poultry and human isolates of the same clonal group could be distinguished only by their plasmid content. Strains within this group were heterogeneous with respect to biotype, fimbriation, virulence, and flagellar serotype. Human and bovine isolates of the second clonal group were distinguished by a minor change in OMP pattern and by their plasmid content. It is concluded that meaningful clonal groupings are best recognized by the combination of OMP and electrophoretic enzyme patterns. The O:K serotype can aid in the recognition of important subclones, whereas the other microbiological properties tested can vary widely within clonal groupings. Furthermore, we conclude that certain O:K serotypes can contain very different clonal groupings having little genetic relatedness.

Hybridoma-derived monoclonal antibodies were produced with fimbrial preparations from Escherichia coli 20025 (04:K12:H-) with fimbrial (F) antigens 1C, 13, one related to 12, and one preliminarily termed y and from E. coli 2980 (018ac:K5:H-) with F antigens 1C and 8. Two clones of subclonal hybrid cells were studied which produced monoclonal antibodies (mc-20025-F2b, immunoglobulin G2b [IgG2b]; mc-2980-F2, IgG1) that were reactive with E. coli 20025, 2980, and a number of additional strains which exhibited the F1C antigen. Results of enzyme-linked immunosorbent assay and Western blot analysis indicated that the antibodies had F1C specificity, and competitive enzyme-linked immunosorbent assay with 125I-labeled antibodies showed that they recognized different epitopes on the fimbrial subunit. Neither of the antibodies agglutinated F1C-fimbriated E. coli but bound to the bacteria. There was no binding to E. coli without F1C fimbriae.

A total of 237 Escherichia coli strains isolated from the urine of patients with various forms of urinary tract infection or from feces of healthy children were analyzed for O group, possession of K1 capsule, type 1 fimbriae, P fimbriae, X adhesin, and production of hemolysin. Some of the strains were also analyzed for K and H antigens, outer membrane protein pattern, and plasmid content. P fimbriation, hemolysin production, and certain O groups were found to be significantly correlated with pyelonephritogenicity. Possession of type 1 fimbriae or of K1 capsule or plasmid content did not significantly correlate with virulence. Outer membrane protein patterns in 139 strains of the more common O groups were analyzed. Only one to three patterns, which varied between serotypes, were usually found within any one O group. Distinctive groups (clones) were found when the strains were grouped according to complete serotype, fimbriation, hemolysin production, and outer membrane protein pattern; also, the mean number of plasmids was typical of the strains in a given clone. Seven clones associated with pyelonephritis were found; together they accounted for 57% of the O serotypable strains from the pyelonephritis patients. The seven clones were P fimbriated but differed in their serotypes as follows: O1:K1:H7, O4:K12:H1, O4:K12:H5, O6:K2:H1, O16:K1:H6, or O18ac:K5:H7. All O1:K1:H7 strains observed fell into two clones according to the presence or absence of type 1 fimbriae and hemolysin production. One clone associated with cystitis was also found; this consisted of O6:K13:H1 strains lacking P fimbriae. Not a single representative of these eight clones was found among the fecal strains from the healthy children. They are proposed to represent virulent clones with special ability to cause human urinary tract infection.

The chemical basis for the alternating antigenic change called form variation noted for the Escherichia coli K1-capsular polysaccharide has been shown by 13C nuclear magnetic resonance to be a result of random O-acetylation of C7 and C9 carbons of the alpha-2-8-linked sialic acid homopolymer. A serologic method (antiserum agar) was developed to identify and isolate the form variants. The O-acetyl positive and O-acetyl negative K1 polysaccharides had unique biochemical and immunologic properties. The O-acetyl-positive variants resisted neuraminidase hydrolysis in contrast to the susceptibility of the O-acetyl negative variant to this enzyme. In addition, O-acetylation altered the antigenicity of the O-acetyl polysaccharides. When injected as whole organisms, O-acetyl positive organisms produced anti-K1 -antibodies in rabbits specific for this polysaccharide variant. O-acetyl negative organisms were comparatively less immunogenic; however, antibodies induced by these organisms reacted with both K1 polysaccharide variants. Burros, injected with either variant, produced antibodies reactive with both K1 polysaccharides.

Eleven Escherichia coli strains, crossreactive with the capsular polysaccharide (CPS) of Neisseria meningitidis group A (GrA), were detected among 645 stool isolates from healthy families in Cairo, Egypt. 10 of these strains were of the O107:K93:H27 or O107:K93:SP serotypes and may be considered descendents of a single bacterium or as a clone. The remaining crossreactive strain was of the O7:K51:H18 serotype. None of the 11 strains produced enterotoxins and none were enteroinvasive. The purified CPS of these E. coli strains, as well as a polysaccharide (PS) from B. pumilis, strain Sh17, precipitated with equine GrA (H49) antiserum. A partial identity between the E. coli K93, K51 and Sh17 PS on the one hand and the GrA CPS on the other was observed by double immunodiffusion when reacted against the H49 antiserum. Four K93 strains and one K51 strain were found among 320 E. coli strains from patients at the Clinical Center, National Institutes of Health, and three K93 strains were found in 105 stool samples from children in Copenhagen. The data from these three surveys suggest that these crossreactive E. coli are common organisms and could serve as a stimulus for "natural" GrA CPS antibodies. Quantitative precipitation analysis showed that K51, K93, and Sh17 PS precipitated 25, 46.8, and 50% of H49 antibodies, respectively. Absorption of H49 antiserum with the GrA CPS removed its precipitating activity with the E. coli K93, K51, and Sh17 PS. Absorption of H49 antiserum with either K51 CPS or Sh17 PS removed the homologous crossreactivity only, whereas K93 CPS absorbed both K93 and K51 reactivities. Antibodies, raised by intravenous injection of formalinized E. coli K93 or K51 cells into rabbits, precipitated with GrA CPS and were bactericidal against GrA meningococci. The crossreaction between the E. coli K93 and the GrA CPS was unexpected since these two CPS are compositionally so dissimilar.

The incidence of enterotoxigenic Escherichia coli (ETEC) was investigated in 95 E. coli strains isolated from 48 infants with diarrhea in Santiago, Chile. By using standard biological assays and DNA-DNA hybridization procedures, ETEC was found in 31.2% of the cases: 14 strains produced heat-stable enterotoxin (ST) only, three strains produced heat-labile enterotoxin (LT) and ST, and two strains produced LT only. DNA probes detected all enterotoxin producers except one ST-producing strain. The ST strains hybridized with one or both of the human ST probes (ST Ib and ST A2). Two of the LT-ST strains hybridized with the ST Ia and ST Ib probes, and the third strain did not hybridize with any of the ST probes. Only the ST group expressed multiple resistance (85.7%) and colonization factor antigen I (CFA I) (92.8%); CFA II was found in two of three LT-ST strains. The O153:H45 serotype was found in 10 of 14 ST strains, and O6:K15:H16 was found in one LT strain and in two LT-ST strains. These findings suggest that ETEC, especially strains that produce ST, may be an important cause of diarrhea among Chilean infants.

The relationship between bacterial characteristics and the severity of urinary tract infection in adults has not been clarified. In this study, Escherichia coli strains (n = 178) were prospectively collected from women with community-acquired urinary tract infection. The isolates were identified by O:K:H serotype and characterized for adherence, hemolysin production, and serum bactericidal resistance. The patients had acute pyelonephritis with or without complicating factors and acute cystitis. Nine serotypes (O1:K1:H7, O1:K1:H-, O2:K1:H-, O4:K12:H1, O7:K1:H-, O9:K34:H-, O16:K1:H6, O16:K1:H-, and O75:K5:H-) comprised 65% of the strains in uncomplicated pyelonephritis, but were significantly less often encountered in complicated pyelonephritis or cystitis. Adherence was the single property most characteristic of the pyelonephritogenic clones. Adhesins specifically recognizing Gal alpha 1----4Gal beta-containing receptors occurred in 80% of strains in uncomplicated pyelonephritis, in 50% of strains in complicated infections, and in 37% of cystitis strains. Hemolysin production and serum resistance did not correlate with any disease pattern. Advanced age did not seem to reduce the selection of virulent E. coli to cause pyelonephritis. These results demonstrate in women a relationship between E. coli virulence and the severity of urinary tract infection analogous to that previously observed in pediatric populations and also illustrate the balance between host resistance and bacterial virulence in the urinary tract.

The repeating pentasaccharide of O-antigen from Escherichia coli O111 contains galactose, glucose, N-acetylglucosamine, and colitose, the latter representing the major antigenic determinant. Phenol extraction of this strain was previously shown to release two fractions (I and II) containing O-antigen carbohydrate, and both fractions were believed to be lipopolysaccharide. We have now characterized fractions I and II and conclude that only fraction II represents lipopolysaccharide. Fraction II contains phosphate, 2-keto-3-deoxyoctonate, beta-hydroxymyristic acid, and potent endotoxin activity, whereas fraction I was deficient in all of these properties of the lipid A and core oligosaccharide regions of lipopolysaccharide. Fractions I and II each represented 50% of the total cellular O-antigen, and both were present on the cell surface. Both fractions were metabolically stable, and no precursor-product relationship existed between them. Fraction II had a number-average molecular weight of 15,800, corresponding to an average of 12 O-antigen repeats per molecule. In contrast, fraction I had a number-average molecular weight of 354,000, corresponding to an average of 404 O-antigen repeats per molecule. Before heat treatment, cells of E. coli O111 are poorly agglutinated by O-serum; although this indicates the presence of a capsule, the corresponding K-antigen was never detected. We conclude that fraction I, when present on the cell surface, inhibits agglutination of unheated cultures of E. coli O111 by O-serum because: (i) a variant strain which lacks fraction I was agglutinated by O-serum without prior heating; (ii) erythrocytes coated with purified fraction I behaved like bacteria containing fraction I in showing inhibition of O-serum agglutination; and (iii) heat treatment released fraction I and rendered bacterial cells agglutinable in O-serum.

Two Escherichia coli O6:K2:H1 strains, C1212 and C1214, isolated from urinary tract infections, were compared for their capacity to adhere to various cells. After growth on solid medium, only C1212 bacteria agglutinate human erythrocytes and attach to urinary epithelial cells. Both of these reactions are mannose resistant. In contrast, C1214 bacteria cause a mannose-sensitive agglutination of guinea pig erythrocytes, show a mannose-sensitive attachment to buccal epithelial cells, and attach to urinary mucus. Immunoelectron microscopy revealed that C1214 bacteria possess type 1 fimbriae (mannose sensitive), which are not present in C1212 bacteria when this strain is grown on solid medium. The fimbriae of C1212 (mannose resistant) were also demonstrated by immunoelectron microscopy. We call these fimbriae demonstrated in C1212 the E. coli F7 antigen. Urinary mucus, and probably mucous material elsewhere, may function as a trap for Enterobacteriaceae with type 1 fimbriae by the specific adherence of such bacteria. We consider this a nonimmune resistance mechanism against disease caused by Enterobacteriaceae.

We examined the relationship between serotype and enterotoxin production in 109 enterotoxigenic Escherichia coli strains isolated from 109 patients with severe cholera-like diarrhea in Dacca, Bangladesh. Of 69 strains producing both heat-labile and heat-stable toxins, 59 (86%) belonged to the one of four O serogroups, and 56 (81%) of these strains belonged to one of six O:K:H serotypes. In contrast, 34 strains producing only heat-stable toxin were distributed among 15 O serogroups, and six strains producing only heat-labile toxin were distributed among six O serogroups. Twelve strains producing heat-labile and heat-stable toxins and five strains producing heat-stable toxin were found which had the same serotype (O78:K-:H12) and biotype. It appears that at least in one geographic setting E. coli strains producing both heat-labile and heat-stable toxins are more restricted in their O groups and O:K:H serotypes than E. coli that produce only heat-stable toxin and that certain serobiotypes may commonly include strains which produce both toxin types.