INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Pathogenic Escherichia coli cells, which cause intestinal and extraintestinal infections in humans, generally adhere to mucosal epithelia early in the colonization of host tissues (14). These bacteria produce a wide variety of adhesive proteins and organelles. Adhesins are often assembled into hairlike fibers called fimbriae (or pili) and are classified based on their adhesive properties. Type 1 adhesins that bind to mannose-containing host cell receptors (adhesins mediating mannose-sensitive hemagglutination [MSHA]) are produced by a wide variety of pathogenic and nonpathogenic E. coli strains yet have been implicated only in the pathogenicity of uropathogenic E. coli (41). There are many adhesins that mediate mannose-resistant hemagglutination (MRHA). They are produced by a large number of pathogenic E. coli isolates associated with different intestinal and extraintestinal infections. Some MRHA adhesins do not form fimbriae: among these are the AFA afimbrial adhesive sheaths (AFAs) that are encoded by the afa gene clusters. Several studies have strongly suggested that afa-positive strains play an important role in urinary tract infection (UTI) pathogenesis (1, 2, 6, 9, 32). Such strains are especially common in pregnant woman (44), children (2), and patients with recurring UTIs (10). Furthermore, using an experimental model of mouse chronic pyelonephritis, Goluszko et al. (20) demonstrated that an isogenic mutant that did not produce the Dr adhesin (encoded by the afa-related dra operon) was less virulent in terms of causing persistent UTI than the parental wild-type strain (20). An unusual feature of the afa-positive strains is their additional association with intestinal infections in children (17, 19, 37, 38). These diarrhea-associated isolates are E. coli organisms of the diffusely adherent pathotype (DAEC) (26, 35).

The first set of afa gene clusters to be described originated from human uropathogenic and diarrhea-associated strains. It contained very similar operons that could be detected by a PCR assay based on the sequence of the afaB and afaC genes from the afa-3 operon (36). This assay also detected the dra (45) and daa (5) operons from the same family of gene clusters. Unlike the other afa genes, afaE, the structural-adhesin-encoding gene, was found to be highly heterogeneous, producing antigenically different adhesins (30). Of the various AfaE subtypes, the AfaE-I, AfaE-III, Dr, and F1845 adhesins, encoded by the afa-1, afa-3, dra, and daa operons, respectively, have been extensively studied (3, 4, 8, 12, 13, 21, 22, 28, 31, 32, 37, 39). They mediate MRHA of human erythrocytes expressing the Dr blood group antigen on the decay-accelerating factor (DAF, or CD55) (43). These so-called Afa/Dr⁺ adhesins also mediate diffuse adhesion of the bacteria to human epithelial cells by recognizing the short consensus repeat-3 (SCR-3) domain of the DAF molecule as a receptor (42). The relative distribution of each of these Afa/Dr⁺ adhesin subtypes in a large collection of strains from patients with UTI showed that afaE3 and afaE1 are frequently expressed (47). Their distribution among afa-positive diarrheagenic strains is unknown. Interestingly, the same afaE1-positive strain has been implicated as the causative agent of consecutive diarrhea and cystitis in an individual child (16).

We recently reported the cloning and characterization of the afa-7 and afa-8 gene clusters from bovine isolates (33). Although these two operons have a genetic organization very similar to that of the afa gene clusters from human isolates, strains carrying them test negative for afa sequences by PCR. The AfaE-VII and AfaE-VIII adhesins do not bind to human DAF (Afa/Dr adhesins) (33). Preliminary epidemiological results showed a high prevalence of afa-8 genes in E. coli isolates from animals with extraintestinal infections and indicated that afa-8 sequences were present in human extraintestinal clinical isolates (15, 33). From these data, it appears that the afa operons are widely distributed among pathogenic E. coli. However, they encode a large variety of AfaE adhesins, the distributions and the receptors of which have not always been identified.

This study has been initiated to increase our knowledge of the various AfaE adhesins and to better understand their role in the pathogenicity of extraintestinal and intestinal E. coli isolates. The first goal was to develop a new PCR assay (using the afa-f and afa-r primers) for the detection of all the members of the afa family of gene clusters, including the afa-7 and afa-8 operons. We then used this assay to collect a large number of afa-positive strains associated with different pathologies. The main objective was to compare the adhesins produced by intestinal and extraintestinal isolates. We identified the subtypes of the AfaE adhesins by PCR and showed that afaE8 is the most prevalent adhesin subtype in human pyelonephritis and blood isolates. We then studied the receptor specificities of some as-yet-uncharacterized AfaE adhesins. These studies confirmed that Afa/Dr⁺ adhesins are produced by both extraintestinal and intestinal pathogenic human isolates, whereas Afa/Dr adhesins are produced only by extraintestinal pathogenic strains. Based on our results, we have a PCR assay for the detection of both Afa/Dr⁺ and Afa/Dr adhesins and a PCR assay for the detection of Afa/Dr⁺ adhesins only.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains, plasmids, and culture conditions. Three collections of human pathogenic E. coli, previously partially characterized, were used in this study. Ninety-seven E. coli strains were isolated from urine specimens from patients (children and adults) clinically diagnosed with pyelonephritis (1). Eighty-eight E. coli strains were isolated from stool specimens from children with diarrhea (18). These strains did not produce heat-stable or heat-labile enterotoxins or Shiga-like toxins and were noninvasive. All adhered to HEp-2 and HeLa cell monolayers. One hundred fifteen strains of E. coli were isolated from blood cultures from cancer patients (25). Additional afa-positive strains were also used: seven human E. coli isolates, including strains KS52 and A30, from which the afa-1 and afa-3 operons have been cloned, respectively (30, 37), and the diarrhea-associated isolate C1845 (5), kindly provided by S. Moseley (University of Washington, Seattle). Twenty-two isolates from calves (19 strains) and piglets (3 strains) with intestinal and extraintestinal disorders were also studied. These isolates were previously reported to carry either the afa-7 (bovine strain 262 KH 89) or the afa-8 (21 strains, including the bovine strain 239 KH 89) gene clusters (33).

Molecular biology techniques. Restriction endonuclease digestion and other common DNA manipulations were performed according to standard procedures (40). PCR assays were performed as previously described (36) using the sets of primers listed in Table 1. Amplification was carried out over 25 cycles of 94°C for 1 min, 65°C for 1 min, and 72°C for 2 min in a thermal cycler (Perkin-Elmer Cetus).

Hemagglutination and adhesion assays. Adhesion to HeLa cells and the hemagglutination of washed human erythrocytes in the presence of 2% (wt/vol) -methyl mannoside were assessed as described elsewhere (1, 28). CHO cell-binding assays were performed as previously described (42). For hemagglutination and adhesion assays, human erythrocytes and CHO cells were preincubated with monoclonal antibodies (MAbs) directed against various domains of the human DAF.

Antibodies. The DAF-specific MAbs 1H4 and 8D11 (directed against the SCR-3 and SCR-4 domains, respectively), were kindly provided by D. M. Lublin (Washington University, St. Louis, Mo.), and BRIC110 (directed against SCR-2) and BRIC230 (directed against SCR-1) were purchased from the International Blood Group Reference Laboratory (Bristol, United Kingdom).

Immunofluorescence. HeLa cells on glass coverslips were infected by incubation with afa-expressing strains for 3 h. Briefly, the cells were fixed in 4% (wt/vol) paraformaldehyde (Merck, Darmstadt, Germany) in 0.1 M phosphate buffer (pH 7.4) for 15 min, incubated in 50 mM NH₄Cl in phosphate-buffered saline (PBS) for 30 min, and then permeabilized by adding 0.1% saponin (Sigma-Aldrich Chimie, St. Quentin-Fallavier, France) in PBS containing 0.2% bovine serum albumin (BSA; Sigma-Aldrich Chimie) and incubating them for 15 min. DAF molecules were detected by incubating the cells for 30 min at room temperature with BRIC230 and then with a fluorescein-conjugated secondary antibody. Coverslips were mounted in mowiol and examined under an Olympus microscope (model BH-2).

Microscopy. Interacting bacteria and erythrocytes were fixed by incubation in 1.6% buffered (0.1 M phosphate buffer, pH 7.4) glutaraldehyde for 1 h, postfixed for 2 h with 2% buffered osmium tetroxide, and embedded in 4% agarose type VII at 37°C. The agar was solidified on ice, and the embedded specimens were cut into 1-mm³ blocks, which were dehydrated in a graded series of ethanol solutions, treated with epoxy-1,2-propane, and then embedded in epoxy resin. Semithin sections (0.5 mm thick) were cut with a diamond knife, placed on glass slides, and examined in phase contrast with a Leica microscope. Bacterial suspensions were examined by electron microscopy for fimbrialike structures as previously described (37). For immunocytochemistry, the infected monolayers were treated as previously described (23). Briefly, cells were fixed in situ with 4% formaldehyde (freshly made from paraformaldehyde) and 0.2% glutaraldehyde in 0.1 M phosphate buffer (pH 7.4), scraped off the slides with a rubber policeman, and embedded in 10% gelatin. Small blocks were infused with 1.7 M sucrose-15% (wt/vol) polyvinylpyrrolidone (average molecular weight, 10,000) for at least 2 h, frozen in liquid nitrogen, and cut at 120°C with a cryostat; the sections were transferred to Formvar-coated nickel grids. The grids were floated on droplets of the following solutions in succession: 50 mM NH₄Cl in PBS; 1% BSA-1% normal goat serum in PBS; anti-DAF antibodies (IA10) in 1% BSA-1% normal goat serum in PBS; three times on 0.1% BSA in PBS; IgG (H+L) anti-mouse immunoglobulin-gold conjugate; and 0.01% gelatin in PBS. The grids were washed with PBS, fixed with 1% glutaraldehyde in PBS, washed again with water, and then incubated with 1% methyl cellulose-0.3% uranyl acetate, air dried, and examined with a Jeol 1010 electron microscope at 80 kV.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Selection of oligonucleotides for the detection of afa-related sequences in pathogenic E. coli. The first set of primers, afa1 and afa2, used to detect afa sequences was based on the partial sequence of the afa-1 gene cluster (36). These oligonucleotides flanked a 750-bp DNA segment overlapping the afaB and afaC genes (Table 1). Comparison of the nucleotide sequences of the afa-3, afa-7, and afa-8 operons showed that these primers did not detect all the afa-related gene clusters. Thus, based on the alignment of the sequences of the afaC3, afaC7, and afaC8 genes, we selected a new set of primers (afa-f and afa-r) specific for the afa family. These primers flanked a 672-bp DNA segment internal to the afaC genes (Table 1). The specificity of the new afa oligonucleotide pair for afa gene clusters was evaluated by testing representative strains. Strains KS52, A22, A30, AL851, 262 KH 89, and 239 KH 89, from which the afa-1, afa-2, afa-3, afa-5, afa-7, and afa-8 operons, respectively, have been cloned, showed positive amplification (Table 2). An amplification product was also obtained from the diarrhea-associated C1845 strain, which carried the daa operon (Table 2). No amplification product was obtained for the E. coli K-12 strain, HB101, used as a negative control. As for the afa1-afa2 set of primers, the afa-f and afa-r oligonucleotides could be used in the multiplex PCR assay developed to detect in a single step the three common adhesin-encoding operons (pap, sfa, and afa) in uropathogenic E. coli (36). The distribution and sizes of the amplification products were as predicted (data not shown).

Detection of afa operons in clinical isolates. Operons of the afa family have been detected in E. coli isolates associated with human extraintestinal and intestinal infections. To validate the new PCR approach for detecting afa-related sequences in clinical isolates, we investigated three collections of E. coli isolates (Table 3). This PCR assay was demonstrated to be sensitive because all the strains that previously gave positive results with the afa1-afa2 pair and all but one of the afa-8-carrying strains tested gave positive results with this test. PCR investigation with the afa-f-afa-r set identified a total of 27 positive pyelonephritis-associated E. coli strains. Only 12 of these strains were positive with the afa1-afa2 pair of primers. Eleven of 115 blood isolates were positive with the afa-f-afa-r set, even though for 8 of these isolates no amplification product was obtained with the afa1-afa2 pair. Twenty-eight of the 88 diarrheal isolates, positive with afa1-afa2 primers, tested positive with the afa-f-afa-r pair. These results confirmed that afa operons are present in E. coli strains associated with various diseases and indicated that the new PCR assay is more sensitive than the original assay for the detection of afa-expressing strains. This assay shows that the frequency of afa-positive strains among pyelonephritis and blood isolates is much higher than predicted with the afa1-afa2 set: 28 instead of 12.4 and 9.6 instead of 2.6%, respectively. The frequencies of afa-positive strains among diarrheal strains are similar (32%) whatever the set of primers.

Distribution of afaE subtypes in pathogenic E. coli isolates. The various afa operons carried by human and animal pathogenic E. coli were characterized by subtyping the afaE gene, using a PCR approach, as previously reported (47). Seven pairs of primers, each specific for an afaE subtype, were defined (Table 1). The specificity of each set of primers was evaluated by testing representative strains (Table 2): (i) the strains from which the afa-1, afa-2, afa-3, afa-5, afa-7, afa-8, and daa operons were cloned and (ii) strains reported to produce adhesins previously designated AFA-I and AFA-II on the basis of their biochemical and antigenic properties (30).

Receptor specificities of AfaE adhesins. Four afaE subtypes (afaE1, afaE3, draE, and daaE) expressed by strains positive in afa1-afa2 PCR encoded adhesins binding the SCR-3 domain of human DAF (42). In contrast, adhesins produced by afa1-afa2 PCR-negative strains, such as AfaE-VII and AfaE-VIII, did not recognize human DAF molecules as receptors (33). We investigated whether other AfaE subtypes (AfaE-II, AfaE-V, and AfaE-X) produced by strains positive in the afa1-afa2 PCR recognized DAF as a receptor.

(i) Binding specificity of AfaE-V. Strains producing AfaE-V have an MRHA-negative phenotype. As an initial approach for evaluating the receptor specificity of AfaE-V, we compared the interactions with human erythrocytes of MRHA-positive strains producing AfaE-I and AfaE-III and that of an MRHA-negative AfaE-V-producing strain by light microscopy (Fig. 1). AfaE-I- and AfaE-III-producing strains adhered to erythrocytes, causing extensive agglutination (Fig. 1A and B). The AfaE-V-producing strain also bound to erythrocytes, suggesting that the cell receptor for the AfaE-V adhesin was present on erythrocyte membranes (Fig. 1C). However, this strain caused a lower level of erythrocyte aggregation.

View larger version (73K): [in this window] [in a new window]   FIG. 1.   Light micrographs of semithin sections of human erythrocytes interacting with strains KS52 (A), A30 (B), and AL851 (C). Note that the aggregates resulting from the cross-linking of erythrocytes (arrowheads) with bacteria (arrows) producing AfaE-I (A) and AfaE-III (B) are larger than those with AfaE-V-producing bacteria (C).

(ii) Binding specificity of AfaE-II and AfaE-X adhesins. Inhibition of hemagglutination was used to test the specificity of the receptor for the AfaE-II and AfaE-X adhesins produced by HB101(pILL1019), three afaE2-expressing isolates, and five isolates carrying an afaEX gene. In all cases, MRHA was affected by the prior treatment of erythrocytes with SCR-3 MAb (Fig. 2). The other afaEX-expressing strains could not be tested due to the lack of MRHA phenotype or P adhesin production. Thus, the SCR-3 domain appears to be essential for AfaE-II and some AfaE-X attachment.

View larger version (59K): [in this window] [in a new window]   FIG. 2.   Hemagglutination of erythrocytes by the AL657 strain producing an AfaE-X adhesin without (A) and with (B) prior treatment of erythrocytes with SCR-3 MAb. Bar, 0.5 µm.

View larger version (93K): [in this window] [in a new window]   FIG. 3.   Electron micrographs of immunogold staining of uninfected HeLa cells (A) and HeLa cells incubated with HB101(pILL1147) bacteria producing AfaE-V (B). DAF molecules were detected by incubation with anti-DAF antibodies followed by anti-mouse immunoglobulin G antibodies conjugated to 10-nm-diameter gold particles. Bar, 0.5 µm.

View larger version (38K): [in this window] [in a new window]   FIG. 4.   Immunofluorescence micrographs of uninfected HeLa cells (A) and HeLa cells incubated with HB101(pILL1101) (B), HB101(pILL1019) (C), or the clinical isolate 1548 (D) producing the AfaE-III, AfaE-II, and AfaEX adhesins, respectively. DAF molecules were detected by incubation with anti-DAF antibodies followed by fluorescein-conjugated secondary antibodies. Bar, 1 µm.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

The AFA adhesive sheath, encoded by the afa operons, is reported to be a virulence factor in E. coli strains that cause intestinal infections and UTIs in humans. Our recent description of new members of the afa family of gene clusters, afa-7 and afa-8, two operons from E. coli strains pathogenic in calves (33), strongly suggested that the prevalence and significance of these virulence factors has been underestimated. In this study, we developed a PCR approach, using a single set of primers designed such that the sequence of each of the members of the afa family of operons would be amplified, and validated it by testing clinical isolates from patients with various diseases (intestinal and extraintestinal infections). This new PCR assay appears to detect all afa-related sequences in human and animal strains. We observed that the frequencies of afa-positive strains in pyelonephritis isolates and in blood isolates were higher (twice and three times, respectively) than those obtained with the previously described PCR. Characterization of the AfaE adhesin subtype by specific PCR assay made it possible to determine in both cases the increase in detection of the afa-8 operon.

Investigation of the AfaE subtypes showed that afaE1, afaE2, afaE3 (and draE, which is 99.4% identical to afaE3), afaE5, and daaE were found in both diarrhea and uropathogenic human isolates, suggesting that, regardless of the afaE subtype, strains expressing these operons may cause both intestinal infections and UTIs. An AfaE-I-producing clone lacking other virulence factors has been reported to be the causative agent of both diarrhea and cystitis (16). We found that the afaE1 subtype was one of the most frequent in isolates from patients with pyelonephritis, sepsis, and diarrhea. In addition, afaE3 and afaE5, two subtypes that have been reported to predominate in cystitis isolates (47), together with afaE1, were also observed in this study to predominate in strains associated with diarrhea. In contrast, we did not detect the afaE7 subtype in human pathogenic strains. These data are consistent with the lack of significance of this subtype suggested by studies with animal pathogenic strains (15, 33).

Previous studies have detected the afa-8 operon in E. coli strains causing neonatal septicemia and diarrhea in farm animals, especially calves (15, 33), and have reported the presence of afa-8-positive strains in human E. coli isolates, especially in human extraintestinal isolates producing cytotoxic necrotizing factor 1 (CNF1) (15). Our results confirm the association of afa-8 with human extraintestinal isolates. Our approach, using two different PCR assays to detect the conserved AFA biogenesis region and the adhesin-encoding gene, generated results suggesting that the entire afa-8 operon is present in all of the positive strains. We demonstrated the presence of the afa-8 operon in blood isolates from cancer patients with various underlying diseases and immune statuses and with and without a possible urinary source for bacteremia. We found that afa-8 was carried by 75% of the afa-positive bacteremia isolates. In addition, we report for the first time that afa-8 is also the most prevalent afa operon in a well-documented collection of pyelonephritis isolates. All these data strongly suggested that afa-8-positive bacteria are associated with severe human extraintestinal infections. afa-positive strains are also associated with infections of the lower urinary tract (10). The possible presence of afa-8 in cystitis isolates should be evaluated.

Several virulence factors associated with extraintestinal isolates of E. coli were detected with similar frequencies in afa-8-positive strains isolated from patients with sepsis and pyelonephritis. We found that iutC (aerobactin)- and papC (P fimbria)-positive strains occurred at high frequencies (70.8 and 58%, respectively) (data not shown). Some afa-8-positive strains were also found to carry sfa or foc (S and FIC fimbriae), hlyA (hemolysin), and cnf1 (CNF1 toxin) sequences. Moreover, 54% of the afa-8-positive strains carried two to five of these virulence factors (data not shown), indicating that these strains are true extraintestinal pathogens. However, three pyelonephritis isolates (12.5%) were positive for only afa-8. In two of these strains, afa-8 was reported to be carried by a genetic element similar to PAI I_AL862, the afa-8-carrying pathogenicity island of the human blood isolate AL862 (34). These data strongly suggested that these two strains are pathogenic and that afa-8 itself may be a key factor in pathogenesis. This idea is supported by a recent study that reported the presence of bmaE (encoding an M agglutinin very similar to AfaE-VIII [33, 46]) in urosepsis isolates of E. coli lacking other extraintestinal virulence factors (27). E. coli isolated from the feces of healthy volunteers is rarely positive for the presence of afa-8 sequences (data not shown). The normal niche for the extraintestinal pathogenic E. coli is the colonic microflora, from which it may spread and cause UTI or septicemia. We detected four afa-8-positive strains among 46 isolates from healthy people (data not shown). At least three of these strains could be considered extraintestinal pathogenic E. coli. They carried a genetic element similar to PAI I_AL862 and were positive for the presence of sequences encoding virulence factors (aerobactin and P fimbriae) associated with uropathogenic strains (data not shown). Interestingly, afa-8 was not detected in human isolates associated with diarrhea, consistent with the previously reported properties of AfaE-VIII, which binds to urothelial cell lines but does not bind to intestinal cell lines (33). It seems likely that afa-8 is involved exclusively in the development of extraintestinal infections in humans and animals. The afa-8 operon encodes AfaE-VIII adhesin and AfaD-VIII protein, which belongs to the AfaD family of invasins (11, 33). The characterization of adhesin and invasin receptors should provide important information about the role of this virulence factor in the pathogenic processes involved in the development of such infections.

Afa/Dr⁺ adhesins, encoded by the afaE1, afaE3, draE, and daaE genes, and Afa/Dr adhesins, encoded by the afaE7 and afaE8 genes, have been described (33, 43). The Afa/Dr⁺ adhesins bind to human DAF, and this attachment results in a dense local accumulation of DAF molecules readily detectable on HeLa and Caco-2 cells by immunofluorescence (22, 24; M. Jouve, M.-I. Garcia, P. Courcoux, S. Bouzari, A. Labigne, P. Gounon, and C. Le Bouguénec, Abstr. 98th Gen. Meet. Ave. Soc. Microbiol., abstr. B-304, 1998). Here, we demonstrate that the afaE2, afaE5, and afaEX subtypes encode adhesins that also recognize DAF, suggesting that all these adhesins, which do or do not mediate MRHA, belong to the Afa/Dr⁺ family. We screened all the afa-positive clones with the fluorescent DAF staining test and showed a correlation between the accumulation of DAF at the bacterium-HeLa cell interaction site and positive results in the PCR assay based on the afa1 and afa2 primers. Thus, the two afa PCR assays described here could be used for different purposes. The afa-f-afa-r primers detect all afa strains, irrespective of the afaE subtype and the binding properties of the adhesins (Afa/Dr⁺ and Afa/Dr). In contrast, the afa1-afa2 assay detects only strains encoding Afa/Dr⁺ adhesins.

The production of several AfaE adhesins by a single strain has been reported (30). Our results suggest that the various AfaE adhesins produced by a single strain have similar binding properties: we found strains that produced two different Afa/Dr⁺ adhesins among diarrhea and intestinal isolates, whereas no single strain simultaneously produced Afa/Dr⁺ and Afa/Dr adhesins. Similar mutual exclusion was observed between sfa or foc operons and afa operons encoding Afa/Dr⁺ adhesins. We have no explanation for this yet. Further investigation of these points might increase our understanding of how and why the genome of E. coli evolves to create new pathotypes and the limits of the evolutionary process.