TEXT

Top Abstract Text References

The virulent strains of Escherichia coli that cause urinary tract infections (UTIs) and other extraintestinal infections in humans (i.e., extraintestinal pathogenic E. coli [ExPEC]) owe their pathogenic potential largely to the presence of specialized virulence factors (VFs) which are absent from commensal members of the species and which allow ExPEC strains to colonize host mucosal surfaces, injure and invade host tissues, foil host defense mechanisms, and incite an injurious host inflammatory response (4, 5, 9, 20). Currently recognized putative VFs (PVFs) of ExPEC include adhesins, siderophores, toxins, protectins, and invasins, some of which are encoded on pathogenicity-associated islands (PAIs) (2, 7, 8, 17, 24, 26). PAIs are large blocks of established or suspected virulence genes that are inserted into the E. coli genome (often at tRNA loci) and which may provide a mechanisms for coordinate horizontal transfer of virulence genes between lineages within E. coli and even between species (3, 7, 21, 24, 25).

Two recently described PAI-linked PVF genes of ExPEC, iha and iroN from E. coli (iroN_{E. coli}), are of interest both in their own right and because of their potential utility as markers for their respective PAIs of origin (23, 27). iha, an novel nonhemagglutinating adhesin which in vitro confers HeLa cell adherence capability to (nonadhering) E. coli K-12, was first identified as part of a tellurite resistance-associated PAI (termed TAI, the tellurite resistance-adherence-conferring island) from an E. coli O157:H7 isolate from a patient with hemorrhagic colitis (27). iha exhibits nearly perfect sequence identity with the sequenced portion of an open reading frame (ORF) of unknown function (ORF "R4") from a PAI in archetypal ExPEC strain CFT073 (8, 17). In addition to the iha homologue and multiple other ORFs of unknown significance, this PAI from strain CFT073 also contains an hly (hemolysin) operon and one of the strain's two pap operons (named for pilus associated with pyelonephritis; P fimbriae), which includes the F7-2 allele of papA (the P fimbrial structural subunit gene) and allele II of papG (the P fimbrial digalactoside-specific adhesin gene) (8, 16, 17). Consistent with the hypothesis that iha is a PVF in ExPEC, probes derived from PAI regions immediately adjacent to iha as it occurs in strain CFT073 hybridized significantly more frequent with UTI or bacteremia isolates of E. coli than with commensal E. coli (8).

iroN_{E. coli}, a novel catechole siderophore receptor which exhibits increased expression in urine, was recently identified in archetypal ExPEC strain CP9 as part of a PAI which also includes one of this strain's two pap operons (i.e., the pap operon containing the F14 papA allele and papG allele III), cnf1 (named for cytotoxic necrotizing factor 1), hly, and foc (F1C fimbriae) (16, 23). iroN_{E. coli}, like iha, was also found by probe hybridization to be more prevalent among E. coli isolates from patients with UTI or bacteremia than among commensal strains, consistent with its being a VF in extraintestinal infections (23).

In the present study, we used PCR- and probe hybridization-based gene detection to determine the prevalence of iha and iroN_{E. coli} within a well-characterized collection of E. coli blood isolates from patients with urosepsis. By comparing newly determined results for iha and iroN_{E. coli} with previously determined extended PVF genotypes (15), O:K:H serotypes (13), carboxylesterase B electrophoretic types (which reflect membership in virulence-associated phylogenetic group B2 versus in other phylogenetic groups) (11), antimicrobial resistance profiles (12), and host compromise data (12), we were able to define and compare the associations of iha and iroN_{E. coli} with these other factors. We present data demonstrating that iha and iroN_{E. coli} exhibit widely divergent associations with all of the above factors and are not consistently linked even with PVFs from their own presumed PAIs of origin.

Strains. The clinical collection studied comprised 67 well-characterized blood culture isolates of E. coli collected from adults with urosepsis in Seattle, Wash., in the mid-1980s (12). The status of these strains with respect to multiple characteristics has been reported previously (10-13, 15). (Eight strains from the 75-member collection which exhibited amplification fingerprints inconsistent with their putative O:K:H serotype and carboxylesterase B type in experiments not yet published were excluded from the present study, leaving 67 evaluable strains as studied herein.)

Molecular detection of iha and iroN_{E. coli}. iha and iroN_{E. coli} were each detected in duplicate both by PCR and by dot blot hybridization. Primers for iha, which were based on GenBank accession no. AF126104 (27), were 5'-CTGGCGGAGGCTCTGAGATCA-3' (IHA-F; forward) and 5'-TCCTTAAGCTCCCGCGGCTGA-3' (IHA-R; reverse) (827-bp product). Primers for iroN_{E. coli}, which were based on GenBank accession no. AF135597 (23), were 5'-AAGTCAAAGCAGGGGTTGCCCG-3' (IRONEC-F; forward) and 5'-GACGCCGACATTAAGACGCAG-3' (IRONEC-R; reverse) (665-bp product). Probes were generated and PCR labeled as previously described (14) using the same primers as for PCR-based gene detection. Amplification and blotting procedures were as previously described (14, 15). Appropriate positive and negative controls were included in each assay. Discrepancies between duplicate blot or PCR determinations were investigated with additional replicates as needed.

Phylogenetic analysis of PVF distribution. Strains were sorted according to carboxylesterase B electrophoretic type (i.e., B2 or B1) and then by O:K:H serotype within each carboxylesterase B type. Serotypes with two or more representatives in the study population were considered to exhibit homogeneity with respect to a particular PVF if the PVF was present in or absent from all representatives of the serotype or was present in all representatives but one (if three or more representatives total).

Statistical analyses. Comparisons of proportions were tested using Fisher's exact test. Comparisons of prevalence within the same population were tested using McNemar's test (6). Comparisons of antibiotic resistance scores and host compromise scores were tested using the Mann-Whitney U test. The criteria for the different categories of statistical significance follow: P > 0.10, indifferent association; P  0.10, borderline significant trend; P < 0.05, possible statistical significance, and P < 0.01, statistical significance. Associations of iha and iroN_{E. coli} with another PVF were considered to agree when they were in the same direction and exhibited P values of <0.10, to somewhat disagree (regardless of direction) when one exhibited a P value of <0.05 and the other exhibited a P value of >0.10, and to strongly disagree when the associations were in opposite directions and both exhibited P values of <0.10.

Prevalence of iha and iroN_{E. coli}. iha was detected in 37 (55%) and iroN_{E. coli} in 26 (39%) of the 67 urosepsis isolates (0.05 < P < 0.10, McNemar's test) (Table 1). Probe and PCR yielded identical results for each gene (not shown). Overall, iha and iroN_{E. coli} were indifferently associated with one another (P > 0.10).

Associations with other PVFs. iha and iroN_{E. coli} each exhibited positive, negative, and indifferent associations with diverse other PVFs (Table 2). Consistent with the occurrence of iha in archetypal uropathogenic strain CFT073 on a PAI that also contains papG allele II and hly (8, 15), iha was strongly associated with papG allele II (and all other pap elements except papG allele III) and with hlyA (Table 2). Similarly, consistent with the discovery of iroN_{E. coli} in archetypal uropathogenic strain CP9 on a PAI that also contains papG allele III, hly, cnf, and foc, iroN_{E. coli} was positively associated with each of these traits, despite not being significantly associated with other pap elements (Table 2).

Associations with papA alleles. iha and iroN_{E. coli} differed substantially with respect to their associations with individual papA alleles (Table 4). iha was significantly associated with the F7-2 allele, which is present in the PAI of strain CFT073 that contains iha, and with the F10 and F16 alleles, whereas it was significantly negatively associated with the F11 allele (Table 4). In contrast, iroN_{E. coli} was positively associated with the F14 allele (P = 0.029), which is present in the iroN_{E. coli}-containing PAI from strain CP9. iha and iroN_{E. coli} exhibited (nonsignificant) opposing associative trends with all the remaining papA alleles except F13 (Table 4).

Phylogenetic distribution of iha and iroN_{E. coli}. Although iha and iroN_{E. coli} were both somewhat more prevalent among carboxylesterase B2 strains, they exhibited divergent patterns of distribution with respect to individual O:K:H serotypes (Table 1). In contrast to the more homogeneously distributed iha, iroN_{E. coli} was sporadically distributed, occurring in many different serotypes but rarely in all members of any one serotype (Table 1). Although both PVFs occurred in a similar proportion of the 14 most prevalent O:K:H serotypes (i.e., 10 of 14 for iha versus 9 of 14 for iroN_{E. coli}), iha satisfied criteria for homogeneous distribution (as defined above [phylogenetic analysis]) in all 14 serotypes, as compared with only seven for iroN_{E. coli} (P < 0.02, McNemar's test) (Table 1).

Host compromise and antimicrobial resistance. iha and iroN_{E. coli} differed with respect to their associations (or lack thereof) with host compromise and antimicrobial resistance. iha was consistently more prevalent among strains from noncompromised hosts than those from compromised hosts, with the strength of the association depending on the type of host compromise (Table 5). In contrast, iroN_{E. coli} was more prevalent among strains from compromised hosts (Table 5). Host compromise scores were significantly lower among iha-positive strains (mean, 0.8 [range, 0-2]) than among iha-negative strains (mean, 1.4 [range, 0-3]; P = 0.002) but did not differ between iroN_{E. coli}-positive and -negative strains (means of 1.0 and 1.1, respectively [range for both groups, 0-3]; P > 0.10). Among strains from noncompromised hosts, iha was always somewhat or significantly more prevalent than was iroN_{E. coli}, whereas among strains from compromised hosts, iha and iroN_{E. coli} were similarly prevalent (Table 5).

Implications of findings. That iha and iroN_{E. coli} were both significantly associated with other PVF genes from their respective PAIs in archetypal ExPEC strains CFT073 (iha) and CP9 (iroN_{E. coli}) suggests, as a first approximation, that iha and iroN_{E. coli} occur consistently on the same (or similar) PAIs throughout the ExPEC population. Consequently, the striking differences documented in the present study between iha and iroN_{E. coli} with respect to their individual associations with other bacterial and host characteristics demonstrate that such associations are likely to be specific to individual PAIs and/or PVF genes rather than being common to all PAIs or to PAI-linked PVF genes in general. This implies that discovery that a particular PVF of ExPEC is linked to a PAI is insufficient to predict its associations with other bacterial and host characteristics, which instead can be discovered only through specific study of the particular PVF itself or the individual PAI.