Role of Motility in the Colonization of Uropathogenic Escherichia coli in the Urinary Tract

http://www.100md.com 感染与免疫杂志 2005年第11期

     Department of Microbiology and Immunology

    Division of Infectious Diseases

    Department of Epidemiology, University of Maryland School of Medicine

    Research Service, Department of Veterans Affairs, Baltimore, Maryland 21201

    Department of Microbiology and Immunology, University of Michigan Medical School, Ann Arbor, Michigan 48109

    ABSTRACT

    Uropathogenic Escherichia coli (UPEC) causes most uncomplicated urinary tract infections (UTIs) in humans. Flagellum-mediated motility and chemotaxis have been suggested to contribute to virulence by enabling UPEC to escape host immune responses and disperse to new sites within the urinary tract. To evaluate their contribution to virulence, six separate flagellar mutations were constructed in UPEC strain CFT073. The mutants constructed were shown to have four different flagellar phenotypes: fliA and fliC mutants do not produce flagella; the flgM mutant has similar levels of extracellular flagellin as the wild type but exhibits less motility than the wild type; the motAB mutant is nonmotile; and the cheW and cheY mutants are motile but nonchemotactic. Virulence was assessed by transurethral independent challenges and cochallenges of CBA mice with the wild type and each mutant. CFU/ml of urine or CFU/g bladder or kidney was determined 3 days postinoculation for the independent challenges and at 6, 16, 48, 60, and 72 h postinoculation for the cochallenges. While these mutants colonized the urinary tract during independent challenge, each of the mutants was outcompeted by the wild-type strain to various degrees at specific time points during cochallenge. Altogether, these results suggest that flagella and flagellum-mediated motility/chemotaxis may not be absolutely required for virulence but that these traits contribute to the fitness of UPEC and therefore significantly enhance the pathogenesis of UTIs caused by UPEC.

    INTRODUCTION

    Uropathogenic Escherichia coli (UPEC) infections account for more than 80% of uncomplicated urinary tract infections (UTIs) (40). UPEC infections occur in otherwise healthy individuals, with the majority of infections affecting women (40). In the year 2000, urinary tract infections resulted in an estimated 6.8 million physician visits, 1.3 million emergency room visits, and 245,000 hospitalizations of women, with an annual cost of more than $2.4 billion (23). Even for men, urinary tract infections resulted in 1.4 million physician visits, 424,000 emergency room visits, and 121,000 hospitalizations in the year 2000, with an economic burden of more than $1 billion (23).

    UPEC strain CFT073 (O6:K2:H1) was isolated from the blood and urine of a hospitalized patient with acute pyelonephritis (26). The O6 clonal group is documented as one of the most frequent causes of UPEC infection (17, 39). Some factors that have been shown to enhance UPEC infection include type 1 fimbriae, Dr fimbriae, TonB, and DegS (8, 30, 33, 37). Although never proven, flagellum-mediated motility has been hypothesized to play a role in the pathogenesis of UTIs caused by UPEC (11, 27). Flagella have also been implicated in the virulence of other E. coli pathotypes, by inducing interleukin-8 expression and Toll-like receptor 5 activation and by mediating the adhesion of enteropathogenic E. coli to epithelial cells in vitro (10, 12). Moreover, flagella have been shown to contribute to the virulence of other uropathogens, such as Proteus mirabilis (25).

    Flagella, complex organelles that protrude from the exterior of the bacterial outer membrane, mediate directed motility and chemotaxis. The bacterial flagellum consists of a basal body, hook, and filament. These flagellar components are essentially constructed in an "inside-out" fashion, with the basal body embedded in the membrane prior to the addition of the hook and filament (13). Genes for flagellum synthesis form an ordered and highly regulated cascade of three classes (7, 18-20, 22, 24). Class 1 genes include flhDC, which encode the transcription factor necessary for transcription of the class 2 genes. Class 2 genes encode the basal body and hook of the flagellum, in addition to FliA (28) and FlgM (anti-28 factor) (18). FliA is the sigma factor that has been shown to be necessary and specific for transcription of the class 3 flagellar genes (21, 28). Intracellular FlgM inhibits FliA activity to ensure that the basal body and hook of the flagella are assembled before the class 3 genes are transcribed (15). The class 3 genes encode hook-associated proteins and the filament of the flagellum (FliC), as well as proteins necessary for motility and chemotaxis (such as MotA, MotB, CheW, and CheY) (18). We postulate that some advantages of utilizing flagellum-mediated motility during urinary tract colonization include the ability to disseminate to new sites of the urinary tract in order to obtain nutrients as well as to escape host immune responses.

    To assess the contribution of flagellum-mediated motility to the pathogenesis of UTI, defined mutations of UPEC strain CFT073 were made in genes that affect distinct stages of the flagellar synthesis cascade. The mutants constructed in this study were shown to have four different flagellar phenotypes: fliA and fliC mutants do not produce flagella; the flgM mutant has similar levels of extracellular flagellin as the wild type but exhibits less motility than the wild type, as seen in such mutants of Salmonella enterica serovar Typhimurium (32); the motAB mutant is nonmotile; and the cheW and cheY mutants are motile but nonchemotactic. By using the well-established CBA mouse model of ascending UTI (16), each of these mutants was examined using the experimental model of infection to determine whether flagella in general, in addition to other flagellar components, are required for colonization of the urinary tract. While these mutants were not attenuated in independent challenges as assessed 3 days postinoculation, they were significantly attenuated to various degrees at specific time points as assessed by coinfections where the wild type and the mutant were mixed 1:1 and transurethrally inoculated into the mouse. The results of another study, conducted concurrently with this study, are consistent with our findings in that a fliC mutant of UPEC strain UTI89 was at a competitive disadvantage during mixed infection with the parent strain 2 weeks after infection, but not during independent challenge (42). Altogether, these results demonstrate that flagella and flagellum-mediated motility/chemotaxis may not be absolutely required for virulence but suggest that they are important for colonization, contribute to the fitness of the bacterium, and therefore significantly enhance the pathogenesis of UTI caused byUPEC.

    MATERIALS AND METHODS

    Bacterial strains and growth conditions. E. coli strain CFT073 was isolated from the blood and urine of a patient diagnosed with acute pyelonephritis (26). CFT073 possesses the genes necessary for flagellum-mediated motility and chemotaxis and has been shown to be motile in 0.3% Luria-Bertani (LB) agar (30, 31, 41).

    For growth on solid medium, strains were streaked onto LB agar plates (per liter, 10 g tryptone, 5 g yeast extract, 0.5 g NaCl, 15 g agar) and incubated at 37°C for 18 h. For growth in liquid culture, strains were inoculated into LB broth (per liter, 10 g tryptone, 5 g yeast extract, 0.5 g NaCl) and incubated at 37°C for 18 h with aeration (200 rpm). Kanamycin (25 μg/ml) and chloramphenicol (20 μg/ml) were added as needed.

    Construction of UPEC flagellar mutants. Deletion mutants of flagellar genes were generated using the lambda red recombinase system designed by Datsenko and Wanner (9). Primers homologous to sequences within the 5' and 3' ends of the target genes were designed and were used to replace these genes with a nonpolar kanamycin resistance cassette derived from the template plasmid pKD4 (Table 1) (9). Less than 10% of the targeted gene sequence (with the exception of cheY, at 30%) remained after homologous recombination. Kanamycin (25 μg/ml) was used for selection of all mutant strains.

    Genotypic analysis of mutants. To determine whether the kanamycin resistance cassette recombined within the target gene site, primers that flank the target flagellar gene sequence were designed (Table 1). Both wild-type and mutant gene sequences were amplified with each set of primers using Taq DNA polymerase (New England Biolabs). After amplification, each PCR product was digested with the restriction enzyme EagI (New England Biolabs) overnight at room temperature. The kanamycin resistance gene carries a single EagI site, whereas the wild-type flagellar genes do not possess any EagI restriction sites. Both the PCR products and restriction digests were electrophoresed on a 0.8% agarose gel for visualization of the amplified and digested DNA. A 1 kb + DNA ladder (Invitrogen) was used for determining the sizes of DNA fragments.

    Motility assays. Motility was first evaluated using soft-agar plates (1% tryptone, 0.5% NaCl, 0.25% agar), which were prepared the day prior to use and left at room temperature overnight. A sample (400 μl) of overnight culture of each strain was used to inoculate 20 ml of sterile LB broth in a 125-ml Erlenmeyer flask and incubated at 37°C with aeration (200 rpm) to an optical density at 600 nm (OD600) of 0.90 to 1.10. The cultures were stabbed into the middle of the soft-agar plates using a sterile inoculating needle (care was taken not to touch the bottom of the plate so as to avoid possible twitching motility). Plates were incubated at 30°C for approximately 11, 17, 22, 25, and 39 h depending on the amount of motility of each strain.

    Motility was also determined by viewing wet mounts of bacterial cultures (prepared exactly as for the soft-agar assay) by phase-contrast microscopy at x400 magnification. Microscopy was performed by using a Zeiss Axioplan microscope with a Zeiss AxioCam MRm camera attached. Wild-type E. coli CFT073 served as the positive control, and the nonmotile strain MC4100 served as the negative control.

    Detection of flagellar expression. Bacterial concentrations from overnight cultures were adjusted to an OD600 of 0.80. To isolate flagella from each strain, 200 μl of standardized culture was vortexed at speed 6 for 3 min at room temperature. Cultures were then centrifuged (8,000 x g, 5 min, room temperature), and the supernatant (containing sheared flagellar filaments) was retained. Proteins were electrophoresed using discontinuous 1-dimensional sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) as described by Ausubel et al. (3) in a MiniPROTEAN II cell (Bio-Rad). After electrophoresis, gels were transferred to a polyvinylidene difluoride membrane (Immobilon-P; Millipore Corp.). The blot was incubated with a rabbit polyclonal antiserum to H1 flagellin (Statens Serum Institute, Copenhagen, Denmark) followed by alkaline phosphatase-conjugated goat anti-rabbit immunoglobulin G (Sigma). Blots were developed using the chromogenic BCIP/NBT (5-bromo-4-chloro-3-indolylphosphate/nitroblue tetrazolium) phosphate substrate system (KPL). The primary antibody was used at a final dilution of 1:40,000, and the secondary antibody was used at a final dilution of 1:2,000 in blocking buffer (5% phosphate-buffered saline [pH 7.5]-1% Tween 20 with 5% milk).

    Transmission electron microscopy (TEM). E. coli MC4100 (aflagellated control), CFT073, its various motility mutants, and the complemented mutants were prepared as described for the motility assays, with the exception that they were grown to an OD600 of 0.300. On the day of sample preparation, carbon-only-coated copper grids (Electron Microscopy Sciences, Fort Washington, PA) were coated with poly-L-lysine (Sigma) to enhance the attachment of bacteria to the grid. A drop (8 μl) of bacterial culture was placed on the grid for 5 min. Bacteria were then fixed with 2.5% glutaraldehyde (Electron Microscopy Sciences, Fort Washington, PA) and stained with 1% sodium phosphotungstic acid (pH 5.8) for 30 s. The grids were examined with a Philips CM-100 transmission electron microscope at an operating voltage of 60 kV. Bacteria were examined between the magnifications of x10,500 and x19,000. Digital images of bacteria were captured with an automated compustage and a Kodak 1.6 Megaplus high-resolution digital camera.

    In vitro complementation of mutants. The fliC, motABcheAW, cheW, cheY, fliA, and flgM genes were amplified using the primers listed in Table 1. The promoters for fliC, motAB, and fliA were also included in amplifications. For cheY, the upstream-most promoter was amplified separately; then both the 5' region of the cheY gene and the 3' region of the cheY promoter DNA were digested with SpeI (New England Biolabs) and ligated together (T4 DNA ligase; Promega). The original primer set that flanked both the cheY promoter and the gene was used for amplification of the cheY gene with its respective promoter attached.

    Each PCR fragment was subcloned into pCR2.1-TOPO, pCR-Blunt II-TOPO, or pCR-XL-TOPO (all from Invitrogen), depending on whether Taq (New England Biolabs) or Pfu (Stratagene) polymerase was used, and then transformed into electrocompetent TOP10 E. coli cells (Invitrogen). The cheW, fliA, and flgM genes were isolated from pCR2.1-TOPO by digestion with SphI and BamHI (New England Biolabs). The cheY and fliC genes were isolated from pCR-BluntII-TOPO by digestion with EcoRV (Invitrogen) and BamHI (New England Biolabs). The motAB genes were isolated from pCR-XL-TOPO (the vector used to clone motABcheAW) by digestion with EcoRV (Invitrogen) and SalI (New England Biolabs). Each digest was electrophoresed in a 0.8% agarose gel and purified using the QIAQuick gel extraction system (QIAGEN).

    The resulting flagellar genes were cloned into the tetracycline resistance gene of pACYC184 and were used to transform their respective flagellar mutants. Specifically, the cheW, fliA, and flgM genes were cloned into SphI-BamHI-cut pACYC184. The cheY and fliC genes were cloned into EcoRV-BamHI-cut pACYC184. Finally, the motAB genes were cloned into EcoRV-SalI-cut pACYC184. Each flagellar gene was cloned into the 5' region of the tetracycline resistance gene and was in the same orientation so that the flagellar genes would be transcribed by the upstream tetracyline resistance promoter. Mutants carrying the desired flagellar-gene-pACYC184 constructs were selected for kanamycin and chloramphenicol resistance, as well as for tetracycline sensitivity. The constructs were also confirmed by isolation (QIAprep MiniPrep; QIAGEN) and digestion with restriction enzymes known to cut each of the flagellar genes and the pACYC184 vector. In vitro complementation was assessed by restoration of the wild-type phenotype in each of the mutants with their respective flagellar-gene-pACYC184 constructs by using the phenotypic assays described above.

    In vitro growth curves. To verify that the independent growth of each flagellar mutant was comparable to that of wild-type CFT073, an overnight culture of each strain was standardized to an OD600 of 0.8 and diluted separately 1:100 in 100 ml LB in a 250-ml flask. The cultures were incubated at 37°C and aerated (200 rpm). The growth of the wild type and of each mutant was monitored approximately every 20 min for 8 h (until growth reached stationary phase). Growth was assessed by measuring OD600.

    In vitro coculture. Overnight cultures were diluted 1:100 in 5 ml of fresh LB and incubated at 37°C with aeration (200 rpm) until the OD600 reached 1.0. At this point, the cultures were individually standardized to an OD600 of 0.8 and the wild-type culture was mixed 1:1 with each mutant culture. From these mixtures, 50-μl aliquots of 10–4 and 10–5 dilutions were spiral plated using an Autoplate 4000 (Spiral Biotech) onto plain LB plates and LB plates containing 25 μg/ml of kanamycin to determine the wild-type and mutant input CFU/ml. Wild-type counts were obtained by subtracting the CFU/ml of the LB plates with kanamycin from the CFU/ml of the plain LB plates.

    The mixed cultures were repassaged every 8 and 16 h for as long as 72 h. Specifically, 10 μl of the standardized mixed culture was inoculated into 5 ml of fresh LB and incubated for 16 h at 37°C with aeration (200 rpm). A sample of this culture (100 μl) was subsequently passaged into 5 ml of fresh LB and incubated for 8 h at 37°C with aeration (200 rpm). These series of passages were repeated for as long as 72 h; after 24, 48, and 72 h, 50 μl of 10–4- and 10–5-diluted culture was spiral plated using an Autoplate 4000 (Spiral Biotech) onto plain LB plates and LB plates containing 25 μg/ml of kanamycin to determine the wild-type and mutant output CFU/ml. All spiral plates were read using a Q-Count automated colony counting system and accompanying software (Spiral Biotech) to determine CFU/ml.

    CBA mouse model of ascending UTI. Mouse studies were performed as described previously (16). The urine of 6-to 8-week-old female CBA/J mice (20 to 22 g; Harlan Sprague-Dawley) was screened 24 h prior to challenge. Mice with >102 CFU/ml (limit of detection) of bacteria in their urine were removed from the study. The remaining mice were anesthetized with pentobarbital and inoculated transurethrally over a 30-s period with a 50-μl bacterial suspension delivering 107 CFU per mouse by using a sterile polyethylene catheter (inner diameter, 0.28 mm; outer diameter, 0.61 mm) connected to an infusion pump (Harvard Apparatus).

    For the independent challenges, overnight cultures for each strain were adjusted to deliver 107 CFU per mouse. Dilutions of these inocula were spiral plated using an Autoplate 4000 (Spiral Biotech) to determine the input CFU/ml for each strain. After 72 h postinfection (hpi), urine was collected, weighed, and spiral plated onto plain LB plates with or without antibiotic. Mice were sacrificed by overdose with isoflurane after 72 hpi, and the bladder and kidneys were aseptically removed, weighed, and homogenized in sterile glass grinders (Kontes) with 5 ml of phosphate-buffered saline. The homogenized tissue was then spiral plated onto plain LB plates with or without antibiotic to determine the output CFU per ml or per g of tissue for each strain.

    For the cochallenge studies, mice were transurethrally inoculated as described above. To prepare the inocula for the different cochallenges, wild-type CFT073 was mixed 1:1 with each of the mutant strains and the resulting mixture was adjusted to deliver 107 CFU per mouse. Dilutions of these inocula were spiral plated to determine the input CFU/ml for each strain. At 6, 16, 24, 48, 60, and 72 hpi, urine was collected, the mice were sacrificed, and the bladder and kidneys were aseptically removed and homogenized. The urine and homogenized tissue were then spiral plated onto plain LB plates with or without antibiotic to determine the output CFU per ml or per g of tissue for each strain. Wild-type counts were obtained by subtracting the CFU/ml of the LB plates with kanamycin from the CFU/ml of the plain LB plates. All spiral plates were read using a Q-Count machine and software (Spiral Biotech) to determine CFU/ml.

    Statistical analysis. For coculture studies, the mean CFU/ml of each strain at each time point was calculated. A paired one-tailed Student t test (InStat; GraphPad Software) was used to determine significant differences between the numbers of wild-type and mutant CFU in vitro.

    Data generated from the independent challenges and cochallenges were generally more skewed (that is, not symmetrical about the mean) than the in vitro coculture data. For this reason, the median CFU per ml of urine or per g of tissue is reported so as to more accurately represent the data. For the independent challenge experiments, where the mutant and the wild type were inoculated into separate mice, the Mann-Whitney U test (a nonpaired test from InStat [GraphPad Software]) was used to determine significant differences between the numbers of wild-type and mutant CFU. For the cochallenge experiments, significant differences between the numbers of wild-type and mutant CFU recovered throughout infection were initially determined using a repeated-measure analysis of variance with rank order data (a ranked-sum test, STATA software; Stata Co.). This test resembles the more familiar Wilcoxon matched-pairs ranked-sum test, a nonparametric test that prevents extreme outliers from skewing the analysis, but in addition analyzes the overall differences in the number of CFU recovered at all sites of infection as a whole (4).

    To rule out the possibility that any mutant could be outcompeted by the wild type in vitro, further analysis of the coculture data was done to help gauge whether mutant attenuations observed in vivo were genuine. Specifically, the competitive indices for both the coculture and cochallenge data at each time point after inoculation were calculated and compared to each other. The competitive index was calculated as (wild-type CFU recovered/mutant CFU recovered)/(wild-type CFU inoculated/mutant CFU inoculated). With this calculation, values of >1 indicate that the wild type outcompeted the mutant, while values of <1 indicate that the mutant outcompeted the wild type. The competitive indices of the cocultures were then used to create a threshold. If any competitive index in vivo fell below that in vitro threshold, whether or not it was deemed significant by the ranked-sum test, the mutant was determined not to be significantly attenuated in vivo. Using both the ranked-sum test and the competitive index comparison, significant differences in the number of CFU recovered during in vivo competition between the wild type and each mutant were confirmed and reported.

    RESULTS

    Construction of flagellar mutants. To determine the roles of flagella and flagellum-mediated motility and chemotaxis in the pathogenesis of UTI by UPEC, defined mutations affecting various stages of the flagellar synthesis cascade were made in genes of UPEC strain CFT073. The lambda red recombinase one-step PCR inactivation method (9) developed by Datsenko and Wanner was used to create six mutants of strain CFT073 with disruptions in flagellar synthesis, motility, and chemotaxis. The genes targeted for mutation included fliC (flagellin), motAB (motility), cheW (chemotaxis), cheY (chemotaxis), fliA (28), and flgM (anti-28). Primers for mutagenesis are listed in Table 1.

    To verify that these genes were specifically targeted for mutation by disruption with a kanamycin resistance cassette, a genetic PCR digest assay was designed. Primers that flank the target flagellar gene sequences were designed (Table 1). Both wild-type and mutant gene sequences were amplified by PCR and digested with restriction enzyme EagI. The kanamycin resistance cassette carries a single EagI site, whereas none of the flagellar gene sequences carry an EagI restriction site. After amplification of wild-type genomic DNA with the different primer sets, the expected sizes for the fliC, motAB, cheW, cheY, fliA, and flgM PCR fragments were observed (data not shown). PCR amplification of the mutant genomic DNA with the same primer sets yielded fragments of different sizes corresponding to the size of the kanamycin resistance gene (1,496 bp) with some remaining flagellar gene sequence. Moreover, the wild-type PCR fragments were not susceptible to digestion with EagI, whereas each of the mutant PCR fragments was digested with EagI, resulting in two fragments (data not shown). In all cases, the expected bands were observed. Thus, all mutations were verified as designed.

    Motility phenotypes of the mutants. After completion of the genetic assay, we assessed the motility and flagellation of each mutant to determine whether the mutation in each flagellar gene resulted in the expected phenotype. The six flagellar mutants were expected to have four different phenotypes: the fliA and fliC mutants should be aflagellate and nonmotile (24); the flgM mutant should have a level of flagellar expression similar to that of the wild type but should exhibit less motility than the wild type, as observed in such mutants of Salmonella serovar Typhimurium (32); the motAB mutant should have paralyzed flagella and therefore be nonmotile (24); and the cheW and cheY mutants should produce motile flagella but be nonchemotactic (24).

    Motility was first assessed by stabbing late-exponential-phase cultures of the wild type and each mutant into 0.25% tryptone broth agar. The soft-agar plates were incubated at 30°C for 11 to 40 h depending on the amount of motility of the individual strain. Motile strains would be expected to disperse through the soft agar, causing the agar to appear turbid. Nonmotile strains should be able to grow only within the confines of the stab and thus not change the turbidity of the agar. It is noteworthy that this test is not only a test of motility but also a test of chemotaxis in that nonchemotactic strains appear only slightly motile in soft agar (2). As expected, wild-type CFT073 was motile in soft agar (Fig. 1A). In contrast, the negative control, MC4100 (carrying a mutation in flhD [6]), appeared nonmotile (Fig. 1B). The flgM mutant displayed motility in soft agar, albeit to a lesser extent than the wild type; this was the predicted result (Fig. 1M). As expected, the fliC, motAB, and fliA mutants were all nonmotile in soft agar (Fig. 1C, E, and K). The chemotaxis mutant cheW appeared nonmotile in this assay, while the other chemotaxis mutant, cheY, appeared only slightly motile, as predicted (Fig. 1G and I). Motility was further assessed by phase-contrast microscopy of exponential-phase cultures. While the cheY mutant was clearly motile, the motility of the cheW mutant was not as obvious, in that only a small percentage of the bacterial population was motile (Fig. 1). Phase-contrast microscopy was also utilized to confirm the motility phenotypes of the wild type, MC4100, and the rest of the mutants (Fig. 1). As observed previously in the soft-agar assays, the wild type and the flgM mutant were motile, while the negative control MC4100 and the fliC, motAB, and fliA mutants were nonmotile (Fig. 1).

    Western blot detection of flagellin. To determine whether mutant strains produced flagella, expression of flagella was assessed by Western blot analysis using H1 flagellin antiserum (Statens Serum Institute, Copenhagen, Denmark). Flagella were isolated, denatured, electrophoresed by SDS-PAGE, and subjected to Western blot analysis. As expected, wild-type CFT073 was positive for flagellin expression (Fig. 2, lane 1). The flgM mutant was also positive for flagellin expression (Fig. 2, lane 12). Moreover, as predicted, the amount of flagellin expression in this mutant was comparable to that of the wild type (Fig. 2, lanes 1 and 12). The fliC and fliA mutants did not produce detectable levels of flagellin expression (Fig. 2, lanes 2 and 10), whereas the cheY mutant was shown to have reduced but moderate flagellin expression (Fig. 2, lane 8). Interestingly, flagellin expression was not observed for the motAB and cheW mutants (Fig. 2, lanes 4 and 6). However, we were able to detect flagellar expression of these two mutants by TEM (data not shown). Only about 10% of both mutant populations were observed to express flagella, explaining why the expression of flagellin was undetectable by Western blot analysis. Therefore, these findings confirm our prediction that the motAB and cheW mutants express flagella; however, the level of flagellar expression was affected by mutation.

    Complementation. To determine whether the mutations made were nonpolar and disrupted only the flagellar gene of interest, each wild-type flagellar gene was cloned into plasmid pACYC184 and transformed into the respective mutant strain. The presence of the correct flagellar gene cloned into pACYC184 in each mutant was confirmed by restriction digestion (data not shown). We assessed whether complementation of each mutant with the wild-type flagellar gene restored the motility and flagellation phenotypes. The motility of the fliC and fliA mutants was restored by the addition of complementation plasmids pfliC and pfliA, respectively, as assessed by motility in soft agar and phase-contrast microscopy (Fig. 1D and L). This also corresponded to restoration of wild-type levels of flagellin expression (Fig. 2, lanes 3 and 11). Flagellation of the wild type, the fliC mutant, and its complemented mutant was also assessed by TEM. For both the wild type and the fliC complemented mutant, flagellated cells were observed by TEM (Fig. 3). In contrast, the fliC mutant and MC4100 (aflagellated control) were never observed to be flagellated (Fig. 3), in agreement with the data obtained from the motility and Western blot analyses.

    The flgM mutant carrying the flgM-pACYC184 construct exhibited less motility in soft agar and under phase-contrast microscopy (Fig. 1N), as well as lower levels of flagellin expression (Fig. 2, lane 13), than the wild type and the flgM mutant. These data are consistent with a study by Hughes et al. (15), where the introduction of a plasmid expressing the flgM gene resulted in loss of flagellin expression in a wild-type strain of Salmonella serovar Typhimurium. For the cheY mutant, motility was restored in soft agar with the addition of the complementing plasmid pcheY (Fig. 1J). As expected, the level of flagellin expression did not change in this mutant with the addition of the cheY-pACYC184 construct (Fig. 2, lane 9). Oddly, the addition of complementing plasmids pmotAB and pcheW to the motAB and cheW mutants, respectively, was not able to restore flagellin expression to a level detectable on the Western blot (Fig. 2, lanes 5 and 7); however, we were able to confirm by TEM that small percentages of these complemented mutant populations were flagellated (data not shown). Moreover, addition of complementing plasmids to the motAB and cheW mutants was able to restore the motility of these two mutants when stabbed into soft agar and observed under phase-contrast microscopy (Fig. 1F and H). Overall, the ability to restore or partially restore wild-type motility and flagellar expression in each of the mutants decreases the likelihood that the mutagenesis caused polar or pleiotropic effects.

    In vitro growth of mutants versus wild-type CFT073. To determine whether there was a difference in growth rates between the wild type and each mutant, two different in vitro growth tests were performed. Growth was first assessed by calculation of the growth rates of the wild type and the mutants when cultured independently in LB at 37°C with aeration. The growth rates of the wild type and the flagellar mutants were found not to be significantly different (data not shown).

    Next, the growth of the wild type and the mutants was assessed by viable counts following in vitro coculture. We consider this assay more sensitive for measuring subtle differences in growth between the wild type and mutants. The wild type and each mutant were mixed 1:1, passaged every 8 and 16 h for as long as 72 h in fresh LB, and incubated at 37°C with aeration. The concentrations (CFU/ml) of the wild type and each mutant were reported after 24, 48, and 72 h postinoculation. The competitive index was calculated as described in Materials and Methods and was reported for each time point postinoculation (Table 2). For all mutants except fliA, the number of CFU throughout coculture showed a modest but nevertheless statistically significant decline over time compared to the number of wild-type CFU (P < 0.05) (Table 2). The number of CFU of the fliA mutant was significantly different only after 72 h of in vitro coculture (P < 0.05) (Table 2). The coculture data allowed the establishment of a baseline that was used to evaluate the in vivo cochallenge data reported below.

    Individual mouse infections with the wild type and flagellar mutants. To assess whether the flagellar mutants could establish a urinary tract infection in the well-established mouse model, approximately 107 CFU of the wild type or each mutant strain was independently transurethrally inoculated into the bladders of CBA/J mice. After 3 days, the CFU/ml of urine and CFU/g of bladder and kidney tissue were determined. Each of the flagellar mutants remained capable of establishing an infection in mice at levels similar to wild-type levels (Fig. 4A to F). In some instances, there appeared to be major differences between the numbers of wild-type and mutant CFU recovered from the bladder after infection; however, these differences were not found to be statistically significant by the Mann-Whitney U test. These results indicated that flagellar motility and chemotaxis are not absolutely required for colonization of the urinary tracts of mice in the absence of competing strains.

    Coinfections with the wild type and flagellar mutants. The coinfection or cochallenge assay is a more sensitive method for determining subtle differences in attenuation during colonization between a wild type and a mutant (5). To determine if the flagellar mutants were attenuated during cochallenge with the wild type, each mutant was mixed 1:1 with the wild type and approximately 107 CFU of this mixture was used for transurethral inoculation into the bladders of CBA/J mice. After 6, 16, 24, 48, 60, and 72 hpi, the CFU/ml of urine and CFU/g of bladder and kidney tissue were determined. Each of these mutants was significantly attenuated at various time points throughout infection (Fig. 5A to F). The significance of the attenuation was initially determined using a repeated-measure analysis of variance with rank order data (a ranked-sum test, STATA software; Stata Co.) (Fig. 5). The competitive index was calculated as described in Materials and Methods and was reported for each time point postinoculation (Table 2).

    Due to the observation that the number of CFU for each mutant declined modestly over time during in vitro coculture, an additional criterion was developed to analyze the significance of mutant attenuation observed in vivo. Specifically, the in vitro competitive indices at the 24-, 48-, and 72-h time points were utilized to create a threshold of significance for mutant attenuation observed in vivo. For example, the number of fliC mutant CFU was observed to decline significantly over time in coculture (Fig. 6A). Both the in vitro and in vivo log10 competitive indices were plotted over time (Fig. 6B). The in vitro competitive index line was extrapolated to 1, since the input CFU are equivalent for the wild type and the mutant; this helped to better visualize the threshold that was created (Fig. 6B). As seen in Fig. 5A, the fliC mutant was initially determined to be significantly attenuated during cochallenge in the urine at 6, 16, 48, and 72 hpi, in the bladder at 6, 16, 48, and 72 hpi, and in the kidneys at 6, 16, 48, and 60 hpi. However, at 60 and 72 hpi, the competitive indices in the kidneys and bladder, respectively, were observed to be at or below the in vitro competitive index threshold (Fig. 6B). Due to the fact that the number of fliC mutant CFU declines over time during in vitro coculture with the wild type, and since these two in vivo competitive indices fell at or below the in vitro competitive index threshold, the mutant attenuations observed at these sites of infection and designated time points were no longer considered statistically significant (marked by section signs [] in Fig. 5 and Table 2). Therefore, upon the addition of this new criterion for significance, the fliC mutant was determined to be significantly outcompeted only at 6, 16, 48, and 60 hpi in the urine, at 6, 16, and 60 hpi in the bladder, and at 6, 16, and 48 hpi in the kidneys (marked by asterisks in Fig. 5 and Table 2).

    The ranked-sum test in combination with the additional criterion for significance was then applied to the remaining flagellar mutant cochallenge data. The mutant attenuations that were determined to be significant by the ranked-sum test only are marked by section signs, whereas those found to be significant by both the ranked-sum test and the added criteria are marked by asterisks (Fig. 5 and Table 2). From these analyses, the motAB mutant was determined to be significantly attenuated in the urine at 48 and 60 hpi, in the bladder at 6 hpi, and in the kidneys at 60 hpi; the cheW mutant was determined to be significantly attenuated in the urine at 6, 16, 48, 60, and 72 hpi, in the bladder at 6, 16, 48, and 60 hpi, and in the kidneys at 16, 48, 60, and 72 hpi; the cheY mutant was determined to be significantly attenuated in the bladder at 60 hpi; the fliA mutant was determined to be significantly attenuated in the urine at 48 hpi; and the flgM mutant was determined to be significantly attenuated in the urine at 48 hpi and in the bladder at 48 hpi.

    DISCUSSION

    Neither synthesis of flagella nor motility nor chemotaxis is required for uropathogenic E. coli to colonize the bladder and kidney of the murine urinary tract. However, these traits clearly contribute to the fitness of the uropathogen and allow a strain that bears these traits to outcompete strains lacking these capabilities. It is likely that transient expression of flagellum-mediated motility and chemotaxis are necessary at specific stages of infection and at specific sites for most-efficient colonization of the urinary tract. It has been demonstrated that, overall, flagellar genes are down-regulated during infection (34). This strategy likely succeeds in avoiding the triggering of Toll-like receptor 5-mediated innate immunity.

    The key finding of this report is the differences in mutant growth observed during cochallenge experiments using the CBA mouse model of ascending UTI. Typically, independent challenges assess the contribution of a particular genetic factor to the virulence of the bacterium. If a mutation in the gene encoding a specific factor still allows for the bacterium to be as infectious as the nonmutated parent strain, then we would argue that this factor is not necessary for virulence. The cochallenge, or mixed infection, however, is a more sensitive method for assessing genetic factors and overall fitness that allow for the efficient colonization of a bacterium. Even though a mutant may not appear attenuated during independent challenge, it may still be significantly outcompeted by the wild type during cochallenge. These results would therefore suggest that certain factors may not be required for a bacterium to colonize a particular site; however, these traits may clearly contribute to the fitness of the bacterium and allow a strain that bears these traits to outcompete strains lacking these capabilities. These are the kinds of results that were observed using the flagellar mutants constructed in this study. From these results, we conclude that flagella and flagellum-mediated motility and chemotaxis contribute to the overall fitness of uropathogenic E. coli and allow this bacterium to outcompete strains lacking these capabilities. Interestingly, a similar study with a different uropathogenic E. coli strain presents data consistent with our findings in that a fliC mutant of UPEC was at a competitive disadvantage during mixed infection with the parent strain as assessed 2 weeks after infection; the mutant and the wild type, however, colonized to levels that were not significantly different during an independent challenge (42). If we look to examples with other genera, Ottemann and colleagues observed that competition of motility and chemotaxis mutants with wild-type Helicobacter pylori during mixed infections results in more attenuation than observed in independent challenges in FVB/N mice (29, 36).

    The well-established mouse model of ascending UTI that we utilized in this study began with infection of the bladder after transurethral inoculation of UPEC (16). To date, there have been no consistent models of ascending UTI developed that originate from the periurethral area. Therefore, in this particular study we are unable to make any assumptions about the role of flagella during periurethral colonization and subsequent ascension from the urethra into the bladder. However, the stages of coinfection during which the mutants were outcompeted in this study tell us something about the timing of flagellar gene expression after infection of the bladder. The attenuation of the fliC and motAB mutants early and midway through infection, but not later, suggests that flagellum-mediated motility aids in the early colonization of UPEC within the urinary tract but may not be required for maintenance of the infection. While the cheW mutant was significantly outcompeted by the wild type at practically all time points and sites during cochallenge, the cheY mutant was significantly outcompeted in the bladder only at 60 hpi. The difference in cochallenge results between the two chemotaxis mutants is most likely attributable to their different phenotypes (the cheY mutant is flagellated, slightly motile, and nonchemotactic, while the cheW mutant is nonmotile and nonchemotactic and has decreased expression of flagella as assessed by TEM). In any event, the attenuation of both of these mutants suggests that chemotaxis is important for the colonization of UPEC within the urinary tract. The attenuation of the fliA and flgM mutants observed 48 hpi during cochallenge may suggest a role for the proper regulation and balance of flagellar expression during infection. Finally, after 48 or 60 h, a general decline in the competitive index was observed for each of the mutants. This appears to indicate that the mutant could be considered, in a way, more fit than the wild type, allowing the mutant to "catch up" with the number of wild-type bacteria present. Therefore, these data may suggest (as well as confirm) that flagellum-mediated motility, chemotaxis, and regulation of flagellar expression are all more important during the early stages of infection but perhaps may not be as important for the maintenance of infection.

    Cochallenges, or mixed infections, have been routinely used to demonstrate subtle differences in colonization between two strains during infection, as thoroughly reviewed by Beuzon and Holden (5). With this in mind, to show that there are no underlying differences in growth in vitro, most researchers measure the growth rate of each strain cultured independently over time. In this study, we observed that two strains (the wild type and the mutant), when cultured independently, had the same growth rates. However, when these strains were mixed and cultured together, we were able to measure a subtle decline in the number of mutant CFU compared to wild-type CFU over time. The reasons for the differences observed during coculture versus independent culture are not well understood. It is our belief that the coculture is more sensitive in measuring subtle differences in growth between two strains in vitro and is also more relevant to in vivo cochallenge than monitoring of independent growth curves. The differences in the number of CFU of the wild type versus the mutant during in vitro coculture allowed the establishment of a baseline that was used to evaluate the in vivo cochallenge results.

    Overall, this study provides evidence for the role of flagella and flagellum-mediated motility and chemotaxis in the colonization of uropathogenic E. coli in the urinary tract. The results are also consistent with a study conducted previously in our lab by Snyder et al. (34), comparing the transcriptome of UPEC when cultured in vitro and during infection in vivo. Snyder et al. (34) reported that genes required for flagellum-mediated motility and chemotaxis were down-regulated during infection in vivo compared to expression in vitro. The report by Snyder et al. (34) averaged expression levels over 10 days and would not have identified important factors such as flagella that are expressed only over short periods of time. Here, in this study, we show that flagellum-mediated motility and chemotaxis likely contribute to the early colonization of the urinary tract; however, it is possible that the expression of these class 2 and 3 flagellar genes is transient and is not easily detectable in vivo. This makes sense, since the expression of flagella during infection in vivo has been shown by many others to induce strong interleukin-8 production and inflammation that is most likely due to flagellar activation of Toll-like receptor 5 (1, 14, 35, 38, 43, 44). Therefore, we hypothesize that flagella and flagellum-mediated motility and chemotaxis are expressed transiently within the urinary tract to aid in colonization but then are quickly repressed to avoid activation of Toll-like receptor 5 and subsequent inflammatory responses.

    ACKNOWLEDGMENTS

    We thank Dotty Sorenson and Sasha Meshinchi of the University of Michigan Microscopy and Image Analysis Lab for help with the transmission electron microscopy.

    This work was supported by Public Health Service grant AI43363 from the National Institutes of Health and, in part, by the Department of Veterans Affairs.

    REFERENCES

    1. Adamo, R., S. Sokol, G. Soong, M. I. Gomez, and A. Prince. 2004. Pseudomonas aeruginosa flagella activate airway epithelial cells through asialoGM1 and toll-like receptor 2 as well as toll-like receptor 5. Am. J. Respir. Cell Mol. Biol. 30:627-634.

    2. Armstrong, J. B., J. Adler, and M. M. Dahl. 1967. Nonchemotactic mutants of Escherichia coli. J. Bacteriol. 93:390-398.

    3. Ausubel, F. M., R. Brent, R. E. Kingston, D. D. Moore, J. G. Seidman, J. A. Smith, and K. Struhl. 1995. Current protocols in molecular biology. John Wiley & Sons, Inc., New York, N.Y.

    4. Bahrani-Mougeot, F. K., E. L. Buckles, C. V. Lockatell, J. R. Hebel, D. E. Johnson, C. M. Tang, and M. S. Donnenberg. 2002. Type 1 fimbriae and extracellular polysaccharides are preeminent uropathogenic Escherichia coli virulence determinants in the murine urinary tract. Mol. Microbiol. 45: 1079-1093.

    5. Beuzon, C. R., and D. W. Holden. 2001. Use of mixed infections with Salmonella strains to study virulence genes and their interactions in vivo. Microbes Infect. 3:1345-1352.

    6. Casadaban, M. J. 1976. Transposition and fusion of the lac genes to selected promoters in Escherichia coli using bacteriophage lambda and Mu. J. Mol. Biol. 104:541-555.

    7. Chilcott, G. S., and K. T. Hughes. 2000. Coupling of flagellar gene expression to flagellar assembly in Salmonella enterica serovar Typhimurium and Escherichia coli. Microbiol. Mol. Biol. Rev. 64:694-708.

    8. Connell, I., W. Agace, P. Klemm, M. Schembri, S. Marild, and C. Svanborg. 1996. Type 1 fimbrial expression enhances Escherichia coli virulence for the urinary tract. Proc. Natl. Acad. Sci. USA 93:9827-9832.

    9. Datsenko, K. A., and B. L. Wanner. 2000. One-step inactivation of chromosomal genes in Escherichia coli K-12 using PCR products. Proc. Natl. Acad. Sci. USA 97:6640-6645.

    10. Donnelly, M. A., and T. S. Steiner. 2002. Two nonadjacent regions in enteroaggregative Escherichia coli flagellin are required for activation of toll-like receptor 5. J. Biol. Chem. 277:40456-40461.

    11. Emody, L., M. Kerenyi, and G. Nagy. 2003. Virulence factors of uropathogenic Escherichia coli. Int. J. Antimicrob. Agents 22(Suppl. 2):29-33.

    12. Giron, J. A., A. G. Torres, E. Freer, and J. B. Kaper. 2002. The flagella of enteropathogenic Escherichia coli mediate adherence to epithelial cells. Mol. Microbiol. 44:361-379.

    13. Harshey, R. M., and A. Toguchi. 1996. Spinning tails: homologies among bacterial flagellar systems. Trends Microbiol. 4:226-231.

    14. Hayashi, F., K. D. Smith, A. Ozinsky, T. R. Hawn, E. C. Yi, D. R. Goodlett, J. K. Eng, S. Akira, D. M. Underhill, and A. Aderem. 2001. The innate immune response to bacterial flagellin is mediated by Toll-like receptor 5. Nature 410:1099-1103.

    15. Hughes, K. T., K. L. Gillen, M. J. Semon, and J. E. Karlinsey. 1993. Sensing structural intermediates in bacterial flagellar assembly by export of a negative regulator. Science 262:1277-1280.

    16. Johnson, D. E., C. V. Lockatell, M. Hall-Craigs, H. L. Mobley, and J. W. Warren. 1987. Uropathogenicity in rats and mice of Providencia stuartii from long-term catheterized patients. J. Urol. 138:632-635.

    17. Johnson, J. R. 1991. Virulence factors in Escherichia coli urinary tract infection. Clin. Microbiol. Rev. 4:80-128.

    18. Kalir, S., J. McClure, K. Pabbaraju, C. Southward, M. Ronen, S. Leibler, M. G. Surette, and U. Alon. 2001. Ordering genes in a flagella pathway by analysis of expression kinetics from living bacteria. Science 292:2080-2083.

    19. Komeda, Y. 1982. Fusions of flagellar operons to lactose genes on a mu lac bacteriophage. J. Bacteriol. 150:16-26.

    20. Komeda, Y. 1986. Transcriptional control of flagellar genes in Escherichia coli K-12. J. Bacteriol. 168:1315-1318.

    21. Kutsukake, K., S. Iyoda, K. Ohnishi, and T. Iino. 1994. Genetic and molecular analyses of the interaction between the flagellum-specific sigma and anti-sigma factors in Salmonella typhimurium. EMBO J. 13:4568-4576.

    22. Kutsukake, K., Y. Ohya, and T. Iino. 1990. Transcriptional analysis of the flagellar regulon of Salmonella typhimurium. J. Bacteriol. 172:741-747.

    23. Litwin, M. S., C. S. Saigal, E. M. Yano, C. Avila, S. A. Geschwind, J. M. Hanley, G. F. Joyce, R. Madison, J. Pace, S. M. Polich, and M. Wang. 2005. Urologic diseases in America Project: analytical methods and principal findings. J. Urol. 173:933-937.

    24. Macnab, R. M. 1996. Flagella and motility, p. 123-145. In F. C. Neidhardt, R. Curtiss III, J. L. Ingraham, E. C. C. Lin, K. B. Low, B. Magasanik, W. S. Reznikoff, M. Riley, M. Schaechter, and H. E. Umbarger (ed.), Escherichia coli and Salmonella: cellular and molecular biology, 2nd ed., vol. 1. American Society for Microbiology, Washington, D.C.

    25. Mobley, H. L., R. Belas, V. Lockatell, G. Chippendale, A. L. Trifillis, D. E. Johnson, and J. W. Warren. 1996. Construction of a flagellum-negative mutant of Proteus mirabilis: effect on internalization by human renal epithelial cells and virulence in a mouse model of ascending urinary tract infection. Infect. Immun. 64:5332-5340.

    26. Mobley, H. L., D. M. Green, A. L. Trifillis, D. E. Johnson, G. R. Chippendale, C. V. Lockatell, B. D. Jones, and J. W. Warren. 1990. Pyelonephritogenic Escherichia coli and killing of cultured human renal proximal tubular epithelial cells: role of hemolysin in some strains. Infect. Immun. 58:1281-1289.

    27. Mobley, H. L., M. D. Island, and G. Massad. 1994. Virulence determinants of uropathogenic Escherichia coli and Proteus mirabilis. Kidney Int. Suppl. 47:S129-S136.

    28. Ohnishi, K., K. Kutsukake, H. Suzuki, and T. Iino. 1990. Gene fliA encodes an alternative sigma factor specific for flagellar operons in Salmonella typhimurium. Mol. Gen. Genet. 221:139-147.

    29. Ottemann, K. M., and J. F. Miller. 1997. Roles for motility in bacterial-host interactions. Mol. Microbiol. 24:1109-1117.

    30. Redford, P., P. L. Roesch, and R. A. Welch. 2003. DegS is necessary for virulence and is among extraintestinal Escherichia coli genes induced in murine peritonitis. Infect. Immun. 71:3088-3096.

    31. Roesch, P. L., P. Redford, S. Batchelet, R. L. Moritz, S. Pellett, B. J. Haugen, F. R. Blattner, and R. A. Welch. 2003. Uropathogenic Escherichia coli use D-serine deaminase to modulate infection of the murine urinary tract. Mol. Microbiol. 49:55-67.

    32. Schmitt, C. K., S. C. Darnell, and A. D. O'Brien. 1996. The attenuated phenotype of a Salmonella typhimurium flgM mutant is related to expression of FliC flagellin. J. Bacteriol. 178:2911-2915.

    33. Selvarangan, R., P. Goluszko, J. Singhal, C. Carnoy, S. Moseley, B. Hudson, S. Nowicki, and B. Nowicki. 2004. Interaction of Dr adhesin with collagen type IV is a critical step in Escherichia coli renal persistence. Infect. Immun. 72:4827-4835.

    34. Snyder, J. A., B. J. Haugen, E. L. Buckles, C. V. Lockatell, D. E. Johnson, M. S. Donnenberg, R. A. Welch, and H. L. Mobley. 2004. Transcriptome of uropathogenic Escherichia coli during urinary tract infection. Infect. Immun. 72:6373-6381.

    35. Steiner, T. S., J. P. Nataro, C. E. Poteet-Smith, J. A. Smith, and R. L. Guerrant. 2000. Enteroaggregative Escherichia coli expresses a novel flagellin that causes IL-8 release from intestinal epithelial cells. J. Clin. Investig. 105:1769-1777.

    36. Terry, K., S. M. Williams, L. Connolly, and K. M. Ottemann. 2005. Chemotaxis plays multiple roles during Helicobacter pylori animal infection. Infect. Immun. 73:803-811.

    37. Torres, A. G., P. Redford, R. A. Welch, and S. M. Payne. 2001. TonB-dependent systems of uropathogenic Escherichia coli: aerobactin and heme transport and TonB are required for virulence in the mouse. Infect. Immun. 69:6179-6185.

    38. Urban, T. A., A. Griffith, A. M. Torok, M. E. Smolkin, J. L. Burns, and J. B. Goldberg. 2004. Contribution of Burkholderia cenocepacia flagella to infectivity and inflammation. Infect. Immun. 72:5126-5134.

    39. Vosti, K. L. 2002. Infections of the urinary tract in women: a prospective, longitudinal study of 235 women observed for 1-19 years. Medicine (Baltimore) 81:369-387.

    40. Warren, J. W. 1996. Urinary tract infections: molecular pathogenesis and clinical management, 1st ed. ASM Press, Washington, D.C.

    41. Welch, R. A., V. Burland, G. Plunkett III, P. Redford, P. Roesch, D. Rasko, E. L. Buckles, S. R. Liou, A. Boutin, J. Hackett, D. Stroud, G. F. Mayhew, D. J. Rose, S. Zhou, D. C. Schwartz, N. T. Perna, H. L. Mobley, M. S. Donnenberg, and F. R. Blattner. 2002. Extensive mosaic structure revealed by the complete genome sequence of uropathogenic Escherichia coli. Proc. Natl. Acad. Sci. USA 99:17020-17024.

    42. Wright, K. J., P. C. Seed, and S. J. Hultgren. 2005. Uropathogenic Escherichia coli flagella aid in efficient urinary tract colonization. Infect. Immun. 73:7657-7668.

    43. Zhang, J., K. Xu, B. Ambati, and F. S. Yu. 2003. Toll-like receptor 5-mediated corneal epithelial inflammatory responses to Pseudomonas aeruginosa flagellin. Investig. Ophthalmol. Vis. Sci. 44:4247-4254.

    44. Zhou, X., J. A. Giron, A. G. Torres, J. A. Crawford, E. Negrete, S. N. Vogel, and J. B. Kaper. 2003. Flagellin of enteropathogenic Escherichia coli stimulates interleukin-8 production in T84 cells. Infect. Immun. 71:2120-2129.(M. Chelsea Lane, Virginia)

    http://www.100md.com/html/DirDu/2006/10/18/25/43/36.htm