Uropathogenic Escherichia coli Outer Membrane Antigens Expressed during Urinary Tract Infection

Abstract

Uncomplicated urinary tract infection (UTI) caused by uropathogenic Escherichia coli (UPEC) represents a prevalent and potentially severe infectious disease. In this study, we describe the application of an immunoproteomics approach to vaccine development that has been used successfully to identify vaccine targets in other pathogenic bacteria. Outer membranes were isolated from pyelonephritis strain E. coli CFT073 cultured under conditions that mimic the urinary tract environment, including iron limitation, osmotic stress, human urine, and exposure to uroepithelial cells. To identify antigens that elicit a humoral response during experimental UTI, outer membrane proteins were separated by two-dimensional gel electrophoresis and probed using pooled antisera from 20 CBA/J mice chronically infected with E. coli CFT073. In total, 23 outer membrane antigens, including a novel iron compound receptor, reacted with the antisera and were identified by mass spectrometry. These antigens also included proteins with known roles in UPEC pathogenesis, such as ChuA, IroN, IreA, Iha, IutA, and FliC. These data demonstrate that an antibody response is directed against these virulence-associated factors during UTI. We also show that the genes encoding ChuA, IroN, hypothetical protein c2482, and IutA are significantly more prevalent (P < 0.01) among UPEC strains than among fecal-commensal E. coli isolates. Thus, we suggest that the conserved outer membrane antigens identified in this study could be rational candidates for a UTI vaccine designed to elicit protective immunity against UPEC infection.

Urinary tract infection (UTI) is a prevalent infectious disease with potentially severe complications. Each year, approximately 11.3 million community-acquired UTIs occur in the United States, with an annual cost of $1.6 billion (8). If left untreated, these infections can lead to more serious conditions including acute pyelonephritis, bacteremia, and renal scarring. Furthermore, increasing rates of antimicrobial resistance among uropathogens will likely complicate future treatment of these infections (13, 21). Consequently, there is an urgent public health need to develop an efficacious vaccine to prevent UTI.

Uropathogenic Escherichia coli (UPEC), the most common etiological agent of community-acquired UTIs, accounts for >80% of these infections (31). A number of virulence determinants facilitate the ability of UPEC to colonize the urinary tract and exert cytopathic effects, including type 1 fimbriae (6), P fimbriae (39), Dr adhesins (12), hemolysin (52, 53), cytotoxic necrotizing factor 1 (37), flagella (25), capsule polysaccharide (2), lipopolysaccharide O antigen (44), and TonB-dependent iron transport systems (50). Recently, the determination of the in vivo transcriptome of UPEC further emphasized the importance of adhesion and iron acquisition during UTI, because genes involved in these processes were highly upregulated during experimental infection (46).

Due to the medical and economic impact of UPEC and UTI, several of these virulence-associated factors have been tested as vaccine targets. For example, immunization with FimH, the type 1 fimbrial adhesin, significantly reduced bladder colonization in C3H/J mice (27) and demonstrated protection in a primate model of UTI (26). Additionally, a subunit vaccine using PapG, the P fimbrial adhesin, complexed with its periplasmic chaperone, PapD, significantly protected primates from histological indications of pyelonephritis (38). Hemolysin (33), Dr fimbriae (11), and the siderophore receptor IroN (42) have also been used in attempts to generate protective immunity against UPEC, with limited success. Recently, mucosal immunization with a mixture of heat-killed uropathogens significantly decreased recurrent UTI incidence among women in a phase II clinical trial (18). However, long-term protection has not been demonstrated for any of these vaccine preparations. Therefore, there is a need to identify additional antigens that may be exploited for the development of a vaccine against UPEC.

While previous efforts to develop a UPEC vaccine were based primarily on specific virulence factors or whole cells, genomic and proteomic methods offer a broader approach to vaccine design. Recently, a technique termed reverse vaccinology was used to screen the genome of serogroup B Neisseria meningitidis and identified a number of novel surface-exposed antigens that are conserved among N. meningitidis strains (35). The antigens that induced the strongest antibody response in immunized animals were then used successfully to develop a universal multivalent vaccine against this pathogen (10). Additionally, immunoproteomic methods, which involve the screening of bacterial proteomes using sera from infected individuals, have been used to identify antigens in pathogens including Campylobacter jejuni (36), Anaplasma marginale (28), Bartonella henselae (4), and Klebsiella pneumoniae (24). An advantage of these genomics and proteomics techniques is the inclusion of novel proteins and nonvirulence factors as candidates for immunization, proteins that are normally excluded from conventional vaccine design strategies.

The immune response to UTI includes both innate and adaptive mechanisms. In addition to Toll-like receptor-mediated acute inflammatory responses (43, 55) and antimicrobial peptides (5), neutrophil infiltration is thought to be the primary mechanism of the innate immune response to control UTI (17). Indeed, neutrophil-depleted mice have an impaired ability to clear UPEC infection compared to neutrophil-replete animals (17). However, adaptive immune responses also contribute to immunity against UPEC. Severe combined immunodeficient mice display increased susceptibility to infection with UPEC, indicating that T- and B-cell-mediated immunity contributes significantly to the clearance of infecting bacteria (19). Consequently, a multifaceted immune response is elicited by natural infections with UPEC, and this evidence suggests that a vaccine that generates a humoral response could prevent uncomplicated UTI.

Because an antibody response is likely a significant component of the adaptive immune response to UPEC (48, 51), ideal vaccine targets should be surface exposed and accessible to circulating immunoglobulins. In gram-negative bacteria such as E. coli, surface-exposed proteins are anchored in the outer membrane. Thus, the outer membrane proteins (OMPs) of UPEC represent a group of prospective vaccine candidates. Furthermore, ideal vaccine candidates should be specific to pathogenic E. coli to avoid cross-reactivity with commensal strains. In this study, we utilized an immunoproteomics approach to identify potential vaccine targets in UPEC. By screening OMPs purified from E. coli CFT073 grown under conditions that mimic the urinary tract with antisera from chronically infected mice, we identified 23 antigenic OMPs that elicited an immune response during infection. Several UPEC-specific OMPs as well as a novel iron-induced protein were identified. We suggest that these antigenic OMPs represent newly identified targets for the development of a multivalent vaccine against UPEC.

MATERIALS AND METHODS

Bacterial strains and culture conditions.

UPEC strain CFT073 was isolated from the urine and blood of a patient with acute pyelonephritis (30). For hybridization studies, UPEC strains (n = 55) were randomly sampled from a collection of 67 isolates cultured from blood and urine samples of patients with acute pyelonephritis (30) and a collection of 38 isolates cultured from urine samples of women with cystitis (9, 47). Fecal-commensal E. coli strains (n = 30) included the well-characterized strains MG1655 and HS as well as 28 isolates from fecal samples from healthy women (30). Ten diarrheagenic E. coli strains were randomly sampled from a collection provided by J. Nataro (University of Maryland, Baltimore, MD) (14).

All cultures were incubated at 37°C with aeration for ∼16 h unless otherwise noted. For iron limitation conditions, bacteria were cultured in Luria broth (LB) containing 10 mM deferoxamine mesylate (Sigma) or 400 μM 2′,2′-dipyridyl (Sigma). Osmotic stress conditions were achieved by culturing bacteria in W salts medium (45) supplemented with 2% NaCl, 0.4% glucose, 0.005% thiamine, 10 mM NH₄Cl, and 15 μM FeCl₂. For nitrogen-limiting conditions, W salts medium was supplemented with 0.4% glucose, 0.005% thiamine, and 1 mM NH₄Cl. Human urine was pooled from 10 male and female donors and sterilized by vacuum filtration through a 0.22-μm-pore-size filter. Urine cultures were incubated at 37°C without aeration until the optical density at 600 nm (OD₆₀₀) reached ∼0.4. All LB or supplemented LB cultures were inoculated with a single colony, whereas minimal medium and urine cultures were inoculated 1:100 from an LB culture grown overnight.

Mice and sera.

The CBA/J mouse model of ascending UTI was used as previously described (15, 22). To achieve chronic infections, mice were transurethrally inoculated with 1 × 10⁹ CFU of E. coli CFT073 three times over a 12-week period, on days 0, 6, and 65. After 12 weeks, the infection was assessed histologically, with all mice displaying signs of chronic pyelonephritis. Median bacterial loads at this time were 3.9 × 10⁷ CFU/g bladder and 5.1 × 10⁵ CFU/g kidney. Serum was collected at 12 weeks from each animal, and equal volumes from 20 individual mice were pooled. These anti-CFT073 sera were used as primary antibodies in subsequent experiments. Sera were also collected and pooled from three uninfected mice for use as nonimmune controls.

Outer membrane isolation.

Outer membranes were isolated as described previously by Molloy et al. (32), with the following modifications. Briefly, bacterial cells were collected by centrifugation (8,000 × g for 10 min at 4°C), and the pellet was resuspended and washed in 10 mM HEPES (pH 7.0). After the addition of 100 U Benzonase Ultrapure nuclease (Sigma), bacterial cells were lysed by two passages through a French pressure cell at 20,000 lb/in². Unbroken cells and cell debris were removed by centrifugation of the lysate (8,000 × g for 10 min at 4°C). Supernatants were diluted in 0.1 M sodium carbonate (pH 11) to a final volume of 60 ml and stirred on ice for 1 h. Carbonate-insoluble membranes were collected by ultracentrifugation (112,000 × g for 1 h at 4°C). Membrane pellets were washed with 10 mM HEPES (pH 7.0) and collected by ultracentrifugation (112,000 × g for 30 min at 4°C). To remove inner membrane contaminants, pellets were resuspended in 2% sodium lauryl sarcosine in 10 mM HEPES, incubated for 30 min at room temperature, and collected by ultracentrifugation (112,000 × g for 30 min at 4°C). The resulting outer membrane pellet was solubilized in 300 to 800 μl isoelectric focusing (IEF) solution (7 M urea, 2 M thiourea, 40 mM Tris-HCl [pH 7.5], 1% amidosulfobetaine-14, 2 mM tributylphosphine, 0.5% BioLyte 3-10 [Bio-Rad], 0.001% bromophenol blue). Soluble OMPs contained within the fraction were quantified using the 2-D Quant kit (Amersham).

Two-dimensional (2D) gel electrophoresis and Western blot analysis.

Duplicate 17-cm, pH 4 to 7 immobilized pH gradient strips (Bio-Rad) were passively rehydrated overnight with 500 μg outer membrane sample in 325 μl IEF solution. IEF was performed with a 250-V linear ramp for 20 min, a 10,000-V linear ramp for 2.5 h, and a 10,000-V rapid ramp for 40,000 V·h in a protean IEF cell (Bio-Rad). Prior to second-dimension separation, immobilized pH gradient strips were equilibrated for 20 min with gentle shaking in ∼5 ml buffer containing 6 M urea, 2% sodium dodecyl sulfate (SDS), 20% glycerol, 5 mM tributylphosphine, and 2.5% acrylamide in 0.15 M bis-Tris-0.1 M HCl (32). Second-dimension SDS-polyacrylamide gel electrophoresis (PAGE) was completed on 10% polyacrylamide gels as previously described (32). One duplicate gel was stained overnight with colloidal Coomassie G-250, while the other gel was transferred onto a polyvinylidene difluoride membrane for Western blotting. Membranes were probed with pooled antisera at a dilution of 1:5,000. Horseradish peroxidase-conjugated goat anti-mouse secondary antibody was used at 1:100,000 and detected using the ECL Plus Western blotting detection system (Amersham).

Mass spectrometry.

Immunoreactive 2D gel spots were excised from a colloidal Coomassie-stained gel and submitted for peptide mass fingerprinting or tandem mass spectrometry (MS/MS) at the University of Michigan Protein Structure Facility or the Michigan Proteome Consortium (Ann Arbor, MI), respectively. Samples were subjected to in-gel trypsin digestion, and mass spectra for peptide mass fingerprinting were acquired on a Micromass TofSpec2E MALDI mass spectrometer. For MS/MS analysis, spectra were acquired on an Applied Biosystems 4700 Proteomics Analyzer (tandem time of flight) from 800 to 3,500 Da, and the eight most intense peaks in each spectrum were selected for MS/MS. All peptide identifications were made using the MASCOT search engine, and spectra were searched against the NCBInr database.

Enzymatic assays.

The activities of cytoplasmic, inner membrane, and outer membrane marker enzymes (glucose-6-phosphate dehydrogenase, NADH oxidase, and esterase, respectively) in the outer membrane preparations were determined. Glucose-6-phosphate dehydrogenase activity was assayed as described previously (54). Stock solutions of 45 mM NADP and 110 mM glucose-6-phosphate were diluted 1:100 into buffer containing 55 mM Tris-HCl (pH 7.5) and 11 mM MgCl₂. This substrate solution (950 μl) was added to 50 μl of sample (25 to 50 μg protein), and the increase in the OD₃₄₀ per min was measured at 25°C for 5 min. As described previously for NADH oxidase activity (49), 900 μl of substrate (50 mM Tris-HCl [pH 7.5], 0.2 mM dithiothreitol, 0.12 mM NADH) was mixed with 100 μl of sample (50 to 100 μg), and the decrease in the OD₃₄₀ per min was measured at 25°C for 5 min. Esterase activity was determined as described previously (49). In a microtiter plate, 20 μl of sample (25 to 50 μg) was mixed with 180 μl of substrate solution containing 10 mM MgSO₄, 1 mM n-hexanoic acid 4-nitrophenyl ether (p-nitrophenyl caproate) (Tokyo Kasei Kogyo Co., Tokyo, Japan), and 5% ethanol in 100 mM potassium phosphate buffer (pH 7). Reaction mixtures were incubated for 10 min at 25°C, and the OD₄₁₀ was measured. All enzymatic assays were standardized using commercially purified enzymes: glucose-6-phosphate dehydrogenase (Fluka), NADH oxidase (Calbiochem), and esterase (Sigma).

Cell culture.

UM-UC-3 human urinary bladder epithelial cells (ATCC CRL-1749) were maintained at 37°C in a humidified 5% CO₂ environment. Cells were cultured in Dulbecco's modified Eagle medium (Gibco) supplemented with 10% fetal bovine serum, 2 mM l-glutamine, 100 U/ml penicillin, and 100 μg/ml streptomycin. For bladder cell infections, monolayers of ∼80 to 90% confluence were washed in Dulbecco's phosphate-buffered saline (pH 7.4) and supplied with fresh medium lacking antibiotics. Cells were overlaid with E. coli CFT073 at a multiplicity of infection of 100 for 2 h. To harvest cell-associated bacteria, monolayers were washed with phosphate-buffered saline. To detach and lyse the uroepithelial cells, 20 ml of sterile water was added to each 150-cm² flask and incubated at 37°C for 10 to 15 min. Eukaryotic cells were lysed by vortexing vigorously for 2 min, and bacteria were collected by centrifugation (8,000 × g for 10 min at 4°C). Outer membranes were isolated from the bacterial pellet as described above.

DNA dot blot hybridizations.

Dot blot hybridizations were performed as previously described (14). Cultures of all UPEC, fecal-commensal, and diarrheagenic E. coli strains grown overnight were standardized to an OD₆₀₀ of ∼4.0 to 5.0. Equal volumes of lysis buffer (0.2 M NaOH, 0.6 M NaCl, 0.8% SDS) and culture were added to 96-well plates. After 10 min of lysis at room temperature, 90 μl of each sample was applied onto ZetaProbe (Bio-Rad) nylon membranes using a BioDot vacuum apparatus (Bio-Rad). Membranes were rinsed in 2× SSC (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate) and air dried. Probes were constructed by PCR amplifying 550-to 650-bp fragments (near the 3′ end of each gene) from CFT073 using the primers listed in Table 1. Probes do not share significant homology with any other genes in currently sequenced E. coli strains. Probe labeling, membrane hybridization, and signal detection were performed using the ECL Direct Nucleic Acid Labeling and Detection system (Amersham) according to the manufacturer's instructions. All blots were performed in duplicate; ambiguities were validated by PCR. A two-tailed Fisher's exact test was used to compare the prevalence of each gene among groups.

TABLE 1.

Primer sequences for dot blot probe construction

Probe	Sequence (5′→3′)
	Forward	Reverse
chuA	GGAGCAAGGCTGGAGAAC	CGTCATCACATCCCAGCC
c2482	CAAGCAAATCTCTGACCG	AGTCCCCATGTTTTTGCC
iha	CATCTGGTTACGGTGGGG	CGCACCTTTGTTATCACC
ireA	GTGATGACTCAGCCACGG	CCCTTTCAGCGACTTGCC
iroN	CCTGACGGTGAATGACAG	GCTTCGATACCGTCCACC
iutA	CGGCTGGCAAATCACCTG	GTTGTCACCGGTAAAGCG

RESULTS

Specificity of antisera from chronically infected mice.

To identify E. coli antigens to which a humoral response was generated, antisera were pooled from 20 CBA/J mice chronically infected with the pyelonephritis strain E. coli CFT073. To determine the reactivity of these sera, whole-cell CFT073 lysates separated on a 12% SDS-polyacrylamide gel were probed with antisera from infected mice or nonimmune sera from uninfected mice (Fig. 1). Antisera from infected mice reacted strongly with CFT073 lysate, while nonimmune sera yielded a minimal reaction. Thus, these antisera contain antibodies specific to protein and nonprotein components of E. coli strain CFT073 that were elicited during chronic infection.

FIG. 1.

Specificity of antisera generated against E. coli CFT073. Western blots of a CFT073 whole-cell lysate electrophoresed on a 12% polyacrylamide gel and probed with nonimmune sera from naïve CBA/J mice (left) or antisera from chronically infected mice (right) are shown. Molecular mass standards are shown in kDa.

OMPs from UPEC are antigenic.

Outer membrane fractions were isolated from E. coli CFT073 cells cultured in LB. To confirm the successful fractionation of outer membranes from other cellular components, cell fractions were assayed for compartment-specific enzymes (Table 2). As expected, esterase, a marker enzyme for the outer membrane, showed significantly higher activity in the outer membrane fraction (P < 0.05) than in other fractions. Esterase activity in the whole membrane fraction is due to the presence of outer membranes in this preparation. NADH oxidase and glucose-6-phosphate dehydrogenase activities, markers for the inner membrane and the cytoplasm, respectively, were significantly lower (P < 0.05) in the outer membrane fraction than in the native fractions of the enzymes. Together, these data indicate that outer membranes were successfully separated from cytoplasmic and inner membrane compartments and suggest that antigens subsequently identified from outer membrane fractions are indeed localized to the bacterial outer membrane.

TABLE 2.

Activities of compartment-specific enzymes in E. coli CFT073 cell fractions

Enzyme	Mean sp act (U/mg) ± SDa
	Cytoplasm	Whole membraneb	Outer membrane
Esterase	0.4 ± 0.2	5.5 ± 0.1	16.5 ± 6.6*
NADH oxidase	0.1 ± 0.2	23.4 ± 3.9	2.0 ± 3.5*
Glucose-6-phosphate dehydrogenase	1.7 ± 0.8	ND	ND

^aND, no detectable activity; *, P < 0.05 by a Student's t test (compared to activity in other fractions).

^bContains inner and outer membrane fractions.

To identify antigens in the outer membrane of E. coli CFT073, proteins in outer membrane fractions were separated by 2D gel electrophoresis, transferred onto polyvinylidene difluoride membranes, and probed with pooled antisera from infected mice. The majority of OMPs detectable by colloidal Coomassie staining (Fig. 2A) reacted with the antisera by Western blot (Fig. 2B). However, a minority of stained OMPs were not seroreactive, further demonstrating the specificity of this approach to identify antigens. Of the >60 open reading frames predicted to encode OMPs in an E. coli genome (32), we identified 23 seroreactive proteins by mass spectrometry (Table 3). Seroreactivity did not always correlate with protein abundance (as determined by colloidal Coomassie stain intensity). For example, the intensely stained OmpC spot (Fig. 2A) was weakly seroreactive (Fig. 2B), while OmpX displayed the opposite effect and was the most seroreactive antigen identified.

FIG. 2.

Identification of antigenic OMPs of E. coli CFT073. (A) Colloidal Coomassie-stained 2D-PAGE gel of outer membrane fractions isolated from CFT073 cultured in rich medium. (B) Western blot of 2D-PAGE gel probed with pooled antisera from chronically infected CBA/J mice. Proteins annotated in A are seroreactive and were identified by mass spectrometry. Molecular mass standards are shown in kDa.

TABLE 3.

Antigenic OMPs identified by 2D-PAGE and mass spectrometry

Locus tag	Protein	Molecular mass (kDa)a	pIa	Function	No. of peptides matched	MASCOT scoreb	In vivo gene expressionc	Culture conditionf
								LB	OS	NL	IL	U	UEC
c0185	FhuA	82.2	5.3	Ferrichrome siderophore receptor	8	34-100d	3.60	+	+	+	+	+	+
c0214	YaeT	90.5	4.9	Outer membrane assembly factor	8	20-125d	3.46	+	+	+	+	+	+
c0652	OmpT	35.6	5.6	Outer membrane protease	17	102	1.71	−	+	+	−	−	−
c0900	OmpX	18.8	6.6	Putative adhesin	8	64	5.99	+	+	+	+	+	+
c1071	OmpF	39.3	4.8	Porin	7	69	2.29	+	+	+	+	+	+
c1093	OmpA	41.2	6.2	Porin	7	35-98d	26.1	+	+	+	+	+	+
c1250	IroN	79.4	5.8	Salmochelin siderophore receptor	24	102	2.11	−	−	−	+	+	+
c3655	Ag43	107	6.0	Nonprotease autotransporter adhesin	2	58, 113e	2.16	+	+	+	+	+	+
c1560	NmpC	41.9	5.4	Porin	4	68-164e	0.27	+	+	+	+	+	+
c1722	OmpW	25.9	5.9	Outer membrane biogenesis factor	4	63-92e	4.78	+	+	+	+	+	+
c2187	YeaF	27.8	5.3	Mlt-interacting protein	26	134	0.53	+	+	+	+	+	+
c2338	FliC	60.8	4.7	Flagellin	8	64-187d	4.86	+	+	+	+	+	+
c2482		79.1	5.4	Putative iron/colicin receptor	8	40-170e	17.3	−	−	−	+	+	+
c2758	OmpC	41.2	4.6	Porin	8	45-106d	3.01	+	+	+	+	+	+
c3610	Iha	76.5	5.6	Iron-regulated adhesin	7	41-131d	7.84	−	−	−	+	+	+
c3623	IutA	84.1	5.2	Aerobactin siderophore receptor	7	59-108e	0.54	+	+	+	+	+	+
c3781	TolC	55	5.8	Multidrug efflux channel	8	52-163d	2.72	+	+	+	+	+	+
c4095	YheE	29.9	9.1	Putative secretion component PulC	13	60	0.19	+	+	−	+	+	+
c4308	ChuA	71.1	5.0	Heme/hemoglobin receptor	6	102-154e	5.25	−	−	−	+	+	+
c4894	Tsx	33	6.0	Nucleoside-binding OMP	2	92, 103e	0.47	+	+	+	+	+	+
c4929	BtuB	70.3	5.5	Vitamin B12 receptor	20	102	0.61	+	+	+	+	+	+
c5006	LamB	50	4.9	Maltoporin	5	87-110e	1.43	+	+	+	+	+	−
c5174	IreA	75.3	6.2	Iron-regulated OMP	6	80-132d	5.44	−	−	−	+	+	+

^aTheoretical molecular mass (kDa) and isoelectric point were determined using the ExPASy Proteomics Server UniProt Knowledgebase (http://us.expasy.org/).

^bRanges represent MS/MS ion scores, and single values indicate mass spectrometry scores determined by peptide mass fingerprinting (where peptide mass fingerprinting scores greater than 60 are considered to be significant [P < 0.05]).

^cRelative expression values represent average signal intensities from a CFT073-specific DNA microarray used to quantify bacterial transcripts isolated during murine experimental UTI (46).

^dIon scores greater than 32 are considered to be significant (P < 0.05).

^eIon scores greater than 51 are considered to be significant (P < 0.05).

^fPresence (+) or absence (−) of the antigen following culture under the condition listed. OS, osmotic stress; NL, nitrogen limitation; IL, iron limitation; U, urine; UEC, uroepithelial cell infection.

From CFT073 cultured in rich medium, porins were the most frequently recognized antigens, including OmpA, OmpC, OmpX, NmpC, and LamB. Outer membrane assembly factors, including YaeT and YeaF, as well as nucleoside and vitamin B₁₂ receptors Tsx and BtuB were also identified as being reactive antigens. Additionally, the secretion apparatus component TolC and a PulC homolog were also found to be antigenic. Therefore, the humoral response generated during UTI specifically targets these OMPs during chronic UTI in mice.

Antigens identified under conditions mimicking the urinary tract.

While 17 antigenic OMPs were identified from E. coli CFT073 cells cultured in LB medium, it is likely that every antigen expressed in vivo cannot be detected under these conditions. To identify additional antigens that may not be expressed in rich medium, CFT073 was cultured in vitro under conditions that more closely mimic the urinary tract. The recently described in vivo transcriptome of E. coli CFT073 demonstrated that genes involved in iron acquisition, osmoregulation, and nitrogen utilization are upregulated during infection of CBA/J mice, suggesting that the urinary tract is an iron- and nitrogen-limited environment for this pathogen as well as being hyperosmotic (46). Therefore, outer membrane antigens isolated from CFT073 cultured under iron-limiting, nitrogen-limiting, and high-osmolarity conditions were identified (Table 3). An additional antigen, OmpT, was identified under conditions of nitrogen limitation and osmotic stress. Iron-limiting conditions induced the expression of five additional antigens with known or putative roles in iron acquisition (Fig. 3). In addition to known UPEC iron compound receptors IroN, ChuA, Iha, and IreA, a novel iron-related OMP was also identified as being an antigen. Hypothetical protein c2482, annotated as a putative receptor for iron or colicin, was identified at 75 kDa and at an isoelectric point of 5.8. Therefore, the number of antigens identified using this immunoproteomics approach can be increased by culturing bacteria under conditions that more closely mimic the urinary tract environment.

FIG. 3.

Antigenic OMPs identified from E. coli CFT073 cultured under conditions of iron limitation. (A) Colloidal Coomassie-stained 2D-PAGE gel of outer membrane fractions isolated from CFT073 cultured under iron-limiting conditions. (B) Western blot of 2D-PAGE gel probed with antisera from chronically infected CBA/J mice. Annotated proteins were identified by mass spectrometry. Note that FhuA, shown in A, is visible on alternate exposures of the blot shown in B. Molecular mass standards are shown in kDa.

To further simulate the urinary tract and induce the expression of antigens found in vivo, human urine was examined as a growth medium. Human urine represents an ex vivo system for culturing UPEC, as it contains a complex array of nutrients, salts, minerals, and soluble cellular factors encountered by E. coli during infection. In view of this, CFT073 was cultured statically in pooled human urine, and seroreactive OMPs produced during this growth were identified by mass spectrometry (Table 3). These antigens include the previously identified porins, transporters, and receptors as well as all of the iron compound receptors detected under conditions of iron limitation. The use of this ex vivo system is important not only to confirm the data acquired under iron limitation conditions but also because antigens produced during growth in urine are likely expressed by UPEC in the bladder.

While the studies described above attempt to reproduce the chemical nature of the urinary tract, the contribution of host cells may be equally important. Because adherence to uroepithelial cells is thought to be critical for the colonization of the urinary tract by E. coli, we next sought to identify outer membrane antigens that are expressed by CFT073 during interactions with host cells. A confluent monolayer of UM-UC-3 cells, a human bladder epithelial cell line, was inoculated with E. coli CFT073 at a multiplicity of infection of 100. Two hours postinfection, cell-associated bacteria were harvested from the monolayer, and outer membrane antigens from these bacteria were identified (Table 3). Interestingly, the antigen profile of CFT073 in contact with cultured uroepithelial cells is nearly identical to the antigenic OMP profiles during growth either in human urine or under conditions of iron limitation. Specifically, all of the iron-related antigens identified in our screen, FhuA, IutA, IroN, ChuA, Iha, IreA, and c2482, were detected during the exposure of CFT073 to host cells. These antigens were also detected during growth in spent tissue culture medium from bladder epithelial cells but not fresh medium (data not shown), suggesting that uroepithelial cells sequester iron, making this environment iron limited for the bacteria. Although novel host cell association factors were not found in the outer membrane, the presence of numerous iron compound receptors provides further evidence that iron acquisition is critical for UPEC survival in vivo and confirms the relevance of these OMPs as antigens targeted by the immune system of the host.

Comparison of outer membrane antigen profiles under diverse culture conditions.

An additional application of this study was an analysis of outer membrane antigen profiles of E. coli CFT073 cultured under distinct conditions. Data in Table 3 demonstrate that the majority of antigens identified were detected in the CFT073 outer membrane in nearly all of the six culture conditions tested. Qualitative examination of Coomassie-stained 2D-PAGE gels indicated that while the level of each antigen varied tremendously with each environmental condition (data not shown), their presence remained constant among most conditions. Five outer membrane antigens with predicted iron-related functions are the exception to this observation. ChuA, IroN, Iha, IreA, and c2482 were detected only during growth under conditions of iron limitation, in human urine, and during contact with bladder cells. Thus, with the exception of several iron-regulated OMPs, the range of urinary tract-mimicking conditions tested did not affect the identified antigens' presence or absence in the outer membrane. These data refine our previous findings, showing that the use of multiple culture conditions can only modestly increase the number of outer membrane antigens identified using this immunoproteomics approach.

Prevalence of genes encoding outer membrane antigens among E. coli strains.

We hypothesized that several of the outer membrane antigens identified in this study might be more prevalent among pathogenic strains of E. coli compared to nonpathogenic commensal strains. Specifically, genes encoding the iron-related antigens ChuA, c2482, Iha, IreA, IroN, and IutA appeared to be absent from the prototypical commensal strain MG1655 (E. coli K-12) by sequence analysis. To determine the prevalence of each of these genes among a larger collection of strains, dot blot hybridizations were performed. Genomic DNA in cell lysate from 95 E. coli strains, including uropathogenic (n = 55), fecal-commensal (n = 30), and diarrheagenic (n = 10) isolates, was probed for each of the six genes encoding these antigens. chuA, c2482, iroN, and iutA were significantly more prevalent (P < 0.01) among UPEC strains than among fecal-commensal isolates (Table 4). Furthermore, these four genes were detected in 65 to 87% of the UPEC isolates tested. Interestingly, chuA and iroN were also found significantly more frequently (P < 0.01) in diarrheagenic isolates, suggesting that these genes might be more broadly conserved among pathogenic E. coli strains. These data clearly demonstrate the conservation of chuA, c2482, iroN, and iutA among UPEC strains and their absence from the majority of commensal E. coli strains, further supporting their potential for application as vaccine candidates.

TABLE 4.

Prevalence of genes encoding iron-related outer membrane antigens among pathogenic and nonpathogenic E. coli isolates determined by hybridization and PCR

Gene	No. (%) of isolates positive for genea
	Uropathogenic (n = 55)	Fecal-commensal (n = 30)	Diarrheagenic (n = 10)
chuA	48 (87)**	9 (30)	9 (90)*
c2482	38 (69)**	5 (17)	5 (50)
iha	25 (45)	11 (37)	4 (40)
ireA	11 (20)	5 (17)	2 (20)
iroN	39 (71)*	10 (33)	7 (70)
iutA	36 (65)**	5 (17)	8 (80)*

^a*, P < 0.01 by Fisher's exact test (compared to fecal-commensal prevalence); **, P < 0.0001 by Fisher's exact test (compared to fecal-commensal prevalence).

DISCUSSION

This study applies an immunoproteomics approach to identify OMP antigens produced during UPEC infection and represents the first broad screen for vaccine targets for this pathogen. Using immunoreactive antisera from chronically infected mice, we identified 23 outer membrane antigens from UPEC that are expressed in vivo and are capable of eliciting a humoral response. One protein identified in this screen, hypothetical protein c2482, is a novel antigen expressed under conditions of iron limitation. Furthermore, we demonstrated that the genes encoding at least four of these OMPs, chuA, c2482, iroN, and iutA, are significantly more prevalent among UPEC strains than among fecal-commensal E. coli strains. We suggest that these conserved antigenic OMPs may be useful targets for a vaccine against UPEC.

The evaluation of UPEC virulence factors as protective immunogens for the prevention of UTI has been extensively investigated (11, 27, 33, 38, 42). However, the heterogeneous nature of UPEC isolates (3) suggests that additional vaccine targets will be required to ensure protection against a broad array of strains. Thus, we were interested in identifying conserved outer membrane antigens of UPEC that may be exploited for use in a vaccine against UTI.

In addition to nutrient transporters, porins, and adhesins, our screen identified a novel UPEC outer membrane antigen. A putative iron receptor, c2482, was detected during conditions of iron limitation, growth in human urine, and contact with bladder epithelial cells (Fig. 3 and Table 3). These findings confirm the predicted outer membrane localization of this protein as well as provide indirect evidence of its involvement in iron acquisition. Recently, our laboratory also determined that the expression of c2482 is upregulated during murine UTI (46), and it is one of the most highly induced proteins during the growth of CFT073 in human urine (1), suggesting that this antigen is a potential vaccine target. The discovery of novel antigens is a notable advantage of immunoproteomics approaches to vaccine design, and the detection of this OMP is a major finding of this report.

This study also identifies proteins expressed by E. coli CFT073 during experimental UTI. While the in vivo transcriptome of murine UTI described transcript-level gene expression during infection (46), the current findings support and extend previous research by examining protein expression in vivo. Consequently, the 23 seroreactive OMPs identified are expressed in the UPEC outer membrane during infection. This agrees well with data from previous work, as 17 of these 23 seroreactive OMPs were among the top 30% of CFT073 transcripts detected in vivo (Table 3) (46). Indeed, 11 of these 17 OMPs were upregulated at least twofold during experimental UTI (46). In addition to potential applications for vaccine development, these findings also contribute to an understanding of the pathogenesis of UPEC.

A notable advantage of the current study over other immunoproteomics analyses is the inclusion of multiple culture conditions designed to mimic the in vivo environment of the pathogen. While these various conditions only modestly increased the number of antigens identified, the approach nevertheless bolstered our confidence that few major outer membrane antigens were omitted from the screen. This approach also revealed several iron-related antigens, ChuA, Iha, IroN, c2482, and IreA, which were detected during growth under only three culture conditions: iron limitation, human urine, and exposure to bladder epithelial cells (Table 3). As these environments likely induce iron deprivation, it is not surprising that they also induce the expression of additional proteins involved in iron acquisition. A recent study from our laboratory confirmed these results, showing the induction of these five OMPs during culture in human urine as well as their repression during growth in iron-replete medium (1). It is interesting that the genes encoding three of the iron-related antigens, chuA, iroN, and c2482, were also found to be both conserved and UPEC specific by dot blot hybridization (Table 4). Given the well-established role of iron acquisition in pathogenesis, we speculate that UPEC, compared to commensal E. coli, expresses a greater range of iron receptors in response to iron-limiting environments.

Outer membrane-anchored surface structures, such as fimbriae, were largely absent from our Western blots. We predict, however, that the infected mouse antisera used in the screen contain fimbria-specific antibodies. Indeed, anti-P fimbria immunoglobulin G (IgG) was detected in the sera of primates infected with UPEC (38). However, because fimbrial proteins often require additional steps to solubilize (16) and are easily sheared from the surface during preparation, they were likely absent from our 2D gels. Furthermore, a comprehensive 2D-PAGE analysis of the E. coli outer membrane proteome also lacked the detection of fimbriae (32), indicating that this is an expected result.

While we are confident that we have identified the major outer membrane antigens expressed by E. coli CFT073 during UTI, there are several inherent limitations associated with our approach. Low-abundance OMPs, which may nevertheless be seroreactive, could have been below the limits of detection of our 2D-PAGE colloidal Coomassie staining and mass spectrometry analyses. Additionally, our screens were performed using anti-mouse IgG; thus, we detected primarily antigens that elicited an IgG response during infection. As UPEC is a mucosal pathogen, secretory IgA is thought to play a role in the clearance of UTI (20). Due to isotype switching, however, IgA-recognized antigens are likely also recognized by IgG. Therefore, while the immunoproteomics approach utilized in this study has limitations, we do not expect that these limits have adversely affected our findings.

Several OMPs identified as being potential protective antigens have roles in the virulence of UPEC and other pathogens. E. coli strains lacking the heme/hemoglobin receptor ChuA (50), the aerobactin receptor IutA (50), the salmochelin receptor IroN (41), the iron-responsive element IreA (40), or the iron-regulated adhesin Iha (23) were significantly outcompeted by the wild-type strain in a mouse model of UTI, demonstrating the importance of iron acquisition to the fitness of this pathogen in vivo. Additionally, isogenic mutants lacking the major flagellum subunit FliC were similarly outcompeted by wild-type E. coli CFT073 (25). Therefore, our data provide evidence that a humoral response is generated against these virulence-associated factors during murine infection. Furthermore, while a role in pathogenesis is likely not a requirement for a vaccine target, it may be beneficial, as neutralizing antibodies may block critical functions of such targets upon infection.

Further validating our findings, several of the outer membrane antigens identified here have been previously implicated as being immunogens. For example, OmpA, a strongly antigenic protein in our screen (Fig. 2), was found to be one of the most antigenic OMPs of Klebsiella pneumoniae by a similar immunoproteomics approach (24). In addition, the most strongly seroreactive antigen identified, OmpX, was previously demonstrated to have immunogenic properties, functioning as an immunological carrier to elicit a humoral response against a coadministered hapten molecule (29). Moreover, Russo et al. demonstrated that immunization with purified IroN affords significant protection in the kidneys of mice infected with UPEC (42). Similarly, a recent study of vaccine candidates against extraintestinal E. coli showed that vaccination with IroN protected mice from lethal septicemic challenge with a neonatal meningitis strain (7). While immunization with ChuA was not protective in that study, Durant et al. noted that ChuA does not contribute to the virulence of E. coli in their sepsis model (7), while it does contribute to the virulence of UPEC (50). Thus, this evidence suggests that our approach has successfully identified antigens that may provide protective immunity against UPEC.

The outer membrane antigens identified in the present study have additional characteristics that support their putative roles as vaccine candidates. Many of these outer membrane antigens, including the iron compound receptors, are predicted to form β-barrel structures in the outer membrane. Even while the majority of each β-barrel protein will be embedded in the membrane, extracellular loops provide surface-exposed regions (34). Because neutrophils are a significant component of the immune response to UTI (17), opsonizing antibodies against such surface-exposed proteins may be important to increase phagocytosis at the site of infection. In addition, many of the OMPs identified in this screen are involved in cellular processes that are critical for bacterial growth in vivo, such as iron acquisition. Blocking the function of these proteins via neutralizing antibodies may also facilitate bacterial clearance from the urinary tract. Finally, at least four of the antigens, ChuA, c2482, IutA, and IroN, are conserved among UPEC isolates and absent from most fecal-commensal E. coli strains (Table 4). Among sequenced strains of pathogenic E. coli, little genetic diversity is observed for each of these antigens; at the amino acid level, each antigen is 90 to 100% identical between strains. This suggests that, if exploited as vaccine targets, these proteins may generate protection against a broad array of pathogenic strains, with minimal cross-reactivity with the normal flora.

A vaccine that prevents uncomplicated UTI would have tremendous public health benefits. The data presented in this study represent a first step towards the development of such a broadly protective vaccine against UPEC. Most outer membrane antigens identified here have not yet been examined as vaccine targets for UPEC. Additionally, a novel antigen identified in this screen, c2482, represents not only a unique vaccine target but also a newly identified OMP that could function as an iron compound receptor. Clearly, much work is needed to extend these findings and investigate immunization with these antigens before an efficacious UTI vaccine can be developed.