The IrgA Homologue Adhesin Iha Is an Escherichia coli Virulence Factor in Murine Urinary Tract Infection

ABSTRACT The role of the Escherichia coli iron-regulated gene homologue adhesin (Iha) in the pathogenesis of urinary tract infections (UTIs) is unknown. We performed a series of complementary analyses to confirm or refute the hypothesis that Iha is a virulence factor in uropathogenic E. coli. Fecal E. coli isolates exhibited significantly lower prevalences of iha (range, 14 to 22%) than did clinical isolates from cases of pediatric cystitis or pyelonephritis, adult pyelonephritis or urosepsis, or bacteremia (range, 38 to 74%). Recombinant Iha from E. coli pyelonephritis isolate CFT073 conferred upon nonadherent E. coli ORN172 the ability to adhere to cultured T-24 human uroepithelial cells. In a well-established mouse model of ascending UTI, CFT073 and its derivative UPEC76 (a pap [P fimbriae] mutant version of strain CFT073) each significantly outcompeted their respective iha deletion mutants in CBA/J mice 48 h after bladder challenge (P < 0.03 for urine, both kidneys, and bladders of both constructs, except for bladders of mice challenged with UPEC76 and its deletion mutant, where P = 0.11). These data suggest that IhaCFT073 is a virulence factor and might be a target for anti-UTI interventions.

Extraintestinal pathogenic Escherichia coli cells produce diverse factors (6,13,16) that allow them to overcome or subvert host defenses and to colonize, injure, and invade host cells or tissues. Delineation of the roles of these factors could lead to measures to prevent or attenuate infections caused by the organisms that express them (28,29).
iha O157:H7 and open reading frame R4 in pathogenicityassociated island 1 (PAI-1) of the well-studied E. coli pyelonephritogenic strain CFT073 (serotype O6:K2:H1) (12) differ by five synonymous nucleotide changes and one nonsynonymous nucleotide change. Strain UPEC76, a derivative of CFT073 in which both pyelonephritis-associated pap (pilus associated with pyelonephritis; P fimbriae) operons, in separate PAIs, have been inactivated by deletions, colonizes the urinary tract in a mouse model of ascending urinary tract infection (UTI) as well as does its parent (32). This finding suggests that CFT073 requires non-pap loci for virulence, with iha being one such candidate critical locus. Also, the virulence of a Proteus mirabilis strain, in which irgA was identified by signature-tagged mutagenesis, was attenuated in a mouse model of ascending UTI (4).
Because Iha's status as a virulence factor has been incompletely elucidated, we performed complementary studies to test the hypotheses that iha ϩ extraintestinal pathogenic E. coli iha is epidemiologically associated with human disease, that cloned iha confers the ability to adhere to uroepithelial cells, and that Iha CFT073 is needed for complete urovirulence in mice challenged with strain CFT073 or UPEC76.
Deletion of iha CFT073 from CFT073 and UPEC76. To create an in-frame, unmarked, isogenic deletion of iha in strains CFT073 and UPEC76, we cloned an in-frame deletion construct of the target gene into suicide plasmid pCVD442 (5). We used inverse PCR (42) to obtain a candidate DNA sequence (not shown) 5Ј to iha CFT073. We then used primer C (5ЈGCAGAGCTCCCTTGCAAGAGGG CGTCGAGC3Ј) and 5ЈGCTATGGATCCGGCTGAAAATCCGAGACAGGG3Ј to produce a 704-bp amplicon that includes the 5Ј terminus of iha CFT073 and 630 bp of the 5Ј noncoding region. This amplicon was sequentially cloned into, and excised from, the pGem-TEasy vector and pSKϩ with SacI and BamHI. Next, the cloned 558-bp amplicon that spans the iha O157:H7 3Ј terminus (41), which differs from the corresponding region in CFT073 by only 3 nucleotides in the downstream noncoding region of the gene, was excised from pSKϩ with BamHI and XhoI and ligated to the SacI-BamHI fragment spanning the 5Ј iha CFT073 terminus. This fusion, representing a central truncation of iha CFT073 , was then cloned into and excised from the pSK ϩ SacI and XhoI sites and finally inserted into the SacI and XhoI sites of pCVD442 (5), producing pCVD442⌬iha CFT073 . Though the 5Ј end of primer C imperfectly represents the iha upstream region, the deletion mutant is, as intended, identical to its parent at that site (sequence data not shown).
Deletion mutant characterizations. Genomic DNA from the presumptive mutants and their parents was digested with BstXI, electrophoresed, transferred to a MagnaCharge nylon membrane (GE Osmonics, Minnetonka, Minn.), and probed with iha O157:H7 . A 371-bp amplicon produced by primers A and B from each candidate deletion mutant was cloned into pSKϩ and sequenced. The deletion mutants were serotyped by Flemming Scheutz, digested with XbaI, and analyzed by pulsed-field gel electrophoresis, and tested by PCR for 11 (iha, papA, papC, sfa/focDE, focG, fimH, iutA, fyuA, iroN, hlyD, and malX) (11,26) putative or demonstrated urovirulence loci that are known to be present in the parent plus, specifically, the F7-1 and F7-2 papA alleles (21,22). The deletion mutants and their parents were also evaluated for mannose-sensitive agglutination of guinea pig erythrocytes and baker's yeast and for mannose-resistant agglutination of human erythrocytes after overnight growth in static broth or on agar plates to indicate expression of type 1 or P fimbriae, respectively (24), and for differences in growth rates (compared to parents) over 18 h in LB broth shaken at 37°C. Finally, the mutants were hybridized with digoxigenin-labeled pCVD442 to confirm suicide plasmid loss.
Anti-Iha antibodies. The PolyQuik protocol (Zymed Laboratories, South San Francisco, Calif.) and affinity purification produced lapine polyclonal antibodies against an Iha epitope. The resulting affinity-purified antibodies (designated anti-Iha antibodies) do not detect an externally directed Iha epitope (41).
OMP analyses. OMPs were prepared (1) from bacteria grown overnight at 37°C in 100 ml of LB broth or Dulbecco's minimal essential medium (D-5030; Sigma, St. Louis, Mo.) containing 0.45% glycerol, 25 mM HEPES, 10 mM sodium pyruvate, 4.5 mN sodium hydroxide, 4 mM L-glutamine, and 44 mM sodium bicarbonate (DMEM). Protein concentrations were determined (protein assay kit; Bio-Rad, Hercules, Calif.). OMPs (2 g) in loading buffer were electrophoretically separated (sodium dodecyl sulfate-10% polyacrylamide) and transferred to membranes (Immobilon-P; Millipore, Billerica, Maine) or Coomassie stained to confirm equal loading of analyte. The membranes to which the proteins were transferred were blocked overnight in antibody buffer (phosphatebuffered saline with 0.05% [vol/vol] Tween 20) containing 5% nonfat dried milk and 0.02% sodium azide) and washed once with antibody buffer. The membranes were then incubated overnight with anti-Iha antibodies diluted 1:2,000 in antibody buffer, washed three times with antibody buffer, incubated for 30 min with affinity-purified goat anti-rabbit immunoglobulin G (heavy plus light chains) peroxidase conjugate (Boehringer Mannheim, Indianapolis, Ind.) diluted 1:2,000 in antibody buffer, and washed again three times with antibody buffer. Membranes were incubated at 4°C or room temperature (blocking or all subsequent steps, respectively). Bound antibodies were detected with the use of the Super-Signal Western blotting (Pierce, Rockford, Ill.).
Adherence assay. T-24 human bladder epithelial cells (30), after growth to confluence at 37°C in 5% CO 2 in plastic flasks in culture medium (McCoy's 5a medium; Sigma; M-4892) supplemented with 10% (vol/vol) heat-inactivated fetal bovine serum, were trypsinized, pelleted, resuspended in the above medium, and then added to wells (in 2-ml volumes) of a six-well plastic flat-bottom cell culture plate (Costar 3516; Corning, Corning, N.Y.) containing sterile plastic 25-mmdiameter coverslips. When cells were ca. 75% confluent, the medium was replaced with fresh medium containing mannose (0.5%) and ampicillin, and adherence assays were performed.
E. coli strain B171 and E. coli ORN172 containing pSKϩ Iha-CFT073 or pSKϩ were grown in LB broth with ampicillin at 37°C without shaking, before assessment in the adherence assay. Cells and bacteria were incubated for 1 (wild type and wild-type derivatives) or 2 h (ORN172 transformants) at 37°C in 5% CO 2 , after which they were washed eight times with sterile phosphate-buffered saline, fixed (100% methanol, 1 min), stained (0.5% crystal violet, 15 min), and thoroughly washed with water to remove background stain. Coverslips were then air dried and mounted on glass slides, and adherent organisms were enumerated by a microscopist unaware of the identity of the organisms in each well. The test organisms were concurrently assessed in duplicate on each of multiple days. Five fields were randomly selected for analysis from each of two duplicate wells per test organism. The adherent organisms were counted separately for each cell within each selected field. Molecular epidemiological analyses. iha was sought in E. coli from the feces of healthy volunteers and noninfected hospitalized subjects and from E. coli recovered from the urine, blood, or cerebrospinal fluid of infected patients by PCR with internal primers D (5ЈCTGGCGGAGGCTCTGAGATCA3Ј) and E (5ЈTC CTTAAGCTCCCGCGGCTGA3Ј) and/or by dot blot hybridization with an internal digoxigenin-labeled iha probe (synthesized with primers D and E) (22). Isolates were tested in duplicate, with independently prepared boiled lysates, for the presence of iha. Strains CFT073 and K-12 were the positive and negative controls, respectively (22,25).
Mouse models of UTI. In an extensively evaluated ascending atraumatic model of UTI, 6-to 10-week-old CBA/J mice were anesthetized and inoculated via the urethra with 1.0 l of the test bacterial suspension/g of body weight Ϫ1 , under conditions that avoided vesicoureteral reflux (14,15,35). Challenge bacteria were grown in shaking, overnight LB broth cultures at 37°C, pelleted, and resuspended in LB broth before challenge. After a standard 2-day period, mice were euthanized, and aseptically harvested urine and bladder and kidney homogenates were cultured quantitatively on agar. Dual-strain challenges were used to compare the colonizing abilities of CFT073 and of UPEC76 to those of their respective iha deletion mutants. Intra-animal competition assessments were used in these studies because they minimize the impact of mouse-to-mouse variation and maximize the ability to identify differences among test strains. The challenge inocula contained ca. 2 ϫ 10 9 total CFU, which is standard in our use of this model and which is within the range used by others (7,32,43) in various mouse models of UTI.
Because we used unmarked mutations in competition experiments, neither CFT073 nor UPEC76 could be easily phenotypically differentiated from their respective iha CFT073 mutants in the postmortem cultures in mouse UTI models. Therefore, we tested colonies from each mouse culture, randomly selected to preclude systematic bias favoring one test strain over the other, by using dot blot hybridization with an internal iha probe or PCR with internal iha primers D and E (25) to determine the relative proportions of recovered organisms that contained an intact copy of iha. Each isolate was analyzed in duplicate, with the corresponding challenge strains used as controls. Colonies from the inoculum suspension cultures were similarly analyzed to define the relative abundances of the two test strains as administered to the mice, and this proportion (the input ratio) was used to adjust the postmortem quantitative culture results (the output ratio) from the mouse infection experiments to obtain the competitive index (CI). Between 10 and 64 colonies per culture (i.e., from inoculum suspensions and postmortem samples) were so assessed. In addition, to exclude contamination, one putative representative of each test strain from each positive culture from the mouse experiments was compared with the actual test strain by random amplified polymorphic DNA analysis (22,44).
Statistics. Differences in proportions were tested for significance by Fisher's exact test for unpaired comparisons or McNemar's test for paired comparisons. Differences involving continuous variables were assessed for significance by the Mann-Whitney U test and the Wilcoxon rank sum tests for unpaired and paired comparisons, respectively. The threshold for statistical significance was a P value Ͻ0.05, but selected values Ͻ0.10 are noted in Table 2 to demonstrate trends.
Nucleotide sequence accession number. The nucleotide sequence of the region upstream of iha CFT073 in CFT073 has been deposited in the GenBank database under accession number AF401752.

Molecular epidemiology of iha.
To determine the epidemiological association of iha with specific extraintestinal infection syndromes, iha was sought among 286 fecal E. coli isolates and 839 clinical isolates from diverse human populations (Table 1). Fecal isolates consistently exhibited a lower prevalence of iha (range, 14 to 22%) than did clinical isolates from all infection syndromes except cystitis in women and neonatal meningitis. The highest prevalences of iha (50 to 74%) occurred among cystitis and pyelonephritis isolates from children and pyelonephritis and urosepsis isolates from adults. Intermediate prevalences (38 to 39%) that still significantly exceeded control frequencies were observed among bacteremia isolates from diverse sources. These differences demonstrate that Iha fulfills the first of the molecular restatement of Koch's postulates, which requires that the property of interest be epidemiologically associated with disease (8).
Structure of iha CFT073 and its upstream region. Iha CFT073 has a deduced molecular size of 78 kDa and contains 696 amino acids. Iha CFT073 and Iha O157:H7 differ by only one amino acid (an Asp 469 3Asn 469 alteration), but the two iha genes differ by 6 nucleotides. Upstream of the iha ATG start codon, CFT073 and E. coli O157:H7 are 98.8 and 34.8% identical from position Ϫ1 to Ϫ489 and positions Ϫ490 through Ϫ630, respectively, consistent with CFT073's extensive chromosomal mosaicism (45).
Role of Iha CFT073 in adherence of E. coli to uroepithelial cells. pSKϩ Iha-CFT073 conferred on E. coli ORN172 the ability to adhere to cultured T-24 uroepithelial cells, whereas pSKϩ did not (Fig. 1). The adherence pattern observed for E. coli ORN172 transformed with pSKϩ Iha-CFT073 was diffuse, and there was interfield variability. This adherence pattern of Iha recombinants was also observed on HEp-2 cells (data not shown). Strain B171 displayed its characteristic localized adherence to cells (data not shown). Assessments of the cellular adherence of CFT073 or UPEC76, their iha deletion mutants, or their deletion mutants transformed with pSKϩ Iha-CFT073 and pSKϩ were precluded by the considerable cytotoxicity of these strains (31), even after only 1 h of incubation (data not shown).
In vivo challenge experiments. CBA/J mice were challenged via the urethra with mixtures of CFT073 and CFT073⌬iha CFT073 or mixtures of UPEC76 and UPEC76⌬iha CFT073 , in dual-strain competition model experiments. Urine, bladders, and kidneys harvested 48 h after inoculation uniformly showed greater overall intensity and prevalence of colonization with CFT073 or UPEC76 than with these strains' respective ⌬iha CFT073 mutants, as summarized for categorical outcomes in Table 2 and as illustrated for absolute log 10 CI values in Fig. 3. Of note, many cultures yielded only the parent strain, indicating that the mutants were present at levels below the detection threshold or were lost in vivo. These data establish the contribution of iha CFT073 to bladder and kidney colonization in mice challenged with a urovirulent E. coli strain, with or without an intact pap operon.

DISCUSSION
The impaired abilities to colonize the mouse urinary tract displayed by isogenic, in-frame, iha CFT073 deletion mutants in two different E. coli backgrounds provide strong experimental support for the hypothesis that Iha is a urovirulence factor. That Iha CFT073 conferred the ability to adhere to T-24 cells is particularly interesting, because E. coli adherence to these cells of human bladder origin has been reported to be independent of known UTI-associated adhesins, such as type 1, P, or S fimbriae, or afimbrial adhesin I (30). These assay results also demonstrate that the single amino acid difference between Iha CFT073 and Iha O157:H7 does not negate the ability of the cloned protein to confer cellular adherence capability to nonadherent laboratory E. coli. This phenotype, though anticipated, warranted confirmation because of the known ability of single amino acid polymorphisms to alter the receptor binding characteristics of the pathogenetically important E. coli adhesin, FimH (39). Whether Iha augments adherence directly or indirectly via effects on other bacterial components and, if it acts directly, whether this phenotype involves receptor-specific or nonspecific binding to the host cell await further assessment.  Additionally, our work confirms the localization of Iha to the cell envelope of another iha ϩ strain and demonstrates that its presence in the cell envelope of CFT073, like the presence of the Iha homologue in the cell envelope of locus of enterocyte effacement-negative E. coli O91:H Ϫ strain 4797/97 (37), depends on the growth medium of the organism, with expression being suppressed in LB broth. Indeed, this diminished expression in LB broth might have actually led to an underattribution of the role of Iha in the mouse colonization experiments, because the challenge inocula were prepared in LB broth, which would tend to diminish the expression of Iha in the parent at the time of inoculation.
The mutations that produced the colonization deficiencies of CFT073 and UPEC76 were each in-frame deletions of iha CFT073 . Such deletions are quite unlikely to exert polar effects on downstream genes, although we cannot exclude completely the possibility that occult cis-acting effects of the deletion reduced colonization ability. Additionally, the deletion mutants were indistinguishable from the parent with respect to a broad panel of characteristics analyzed other than iha CFT073 and antigenic Iha CFT073 . Moreover, two independent deletions of iha CFT073 in two different organisms produced the same diminished-colonization phenotype, to similar extents, without perturbing other analyzed loci. Recent data demonstrate the fairly frequent occurrence of secondary mutations in extraintestinal pathogenic E. coli using pCVD442 (20); therefore, it is important to generate such confirmatory results to increase confidence that the altered phenotype actually was caused by the intended mutation. The extensive methods that we used to exclude unanticipated secondary mutations in the iha mutants exceed those customarily used for this purpose in studies of this sort and were the same as those that uncovered otherwiseoccult secondary mutations in the recent study; their use increases confidence that such alterations were absent from the present mutants.
Although inactivating papG in E. coli pyelonephritis strain DS17 (O6:K5) by introducing a premature stop codon reduced the ability of this strain to cause ascending pyelonephritis in monkeys (34), inactivation of both pap operons in CFT073 did not diminish CFT073's ability to colonize the mouse urinary tract (32). Likewise, inactivating the putative virulence factor sat (secreted autotransporter toxin) in CFT073 had no impact on in vivo colonization ability (32). Thus, iha joins tonB, iutA, chuA, and fimH as the only known putative urovirulence genes, the mutation of which in the CFT073 background significantly attenuates urovirulence (43). Moreover, iha CFT073 is the first such gene for which an unmarked, in-frame deletion was used to generate the evaluated CFT073 mutant.
Finally, our epidemiological data provide novel molecular evidence that iha is statistically significantly more frequent among diverse groups of extraintestinal E. coli pathogens than among fecal control isolates. The absence of a categorical association of iha with all infection syndromes and host groups suggests differing pathogenetic mechanisms across clinical and epidemiological settings, a worthy topic for future investigations.
In summary, Iha CFT073 's demonstrated pathogenetic importance in a mouse model of UTI complements (i) epidemiological data associating iha with recurrent or invasive UTI and diverse-source bacteremia and (ii) in vitro uroepithelial adherence data. However, even though iha CFT073 mutants exhibited impaired in vivo colonization in two different bacterial hosts and although this colonization deficit is plausibly attributable to iha's adherence-conferring function, Iha CFT073 has not yet been proven to serve as an adhesin in wild-type pathogens. Nonetheless, our data suggest that Iha deserves further scrutiny as a molecule to exploit for prevention or treatment of human UTIs. We thank Jennifer L. Bradley for assistance with the preparation of the manuscript, David Prentiss for preparing the mouse data figures, and Adam Stell for expert technical assistance.