OUP user menu

The role of cytotoxic necrotizing factor-1 in colonization and tissue injury in a murine model of urinary tract infection

David E. Johnson, Cinthia Drachenberg, C. Virginia Lockatell, Michael D. Island, John W. Warren, Michael S. Donnenberg
DOI: http://dx.doi.org/10.1111/j.1574-695X.2000.tb01454.x 37-41 First published online: 1 May 2000


Cytotoxic necrotizing factor-1 (CNF1) is commonly found in Escherichia coli isolates from patients with urinary tract infection (UTI). To determine whether CNF1 is an important UTI virulence factor we compared the ability of a clinical E. coli UTI isolate and a CNF1-negative mutant of that isolate to colonize and induce histological changes in the urinary tract in a murine model of ascending UTI. We found no evidence that the mutant strain was attenuated.

  • Urinary tract infection
  • Pathogenesis
  • Cytotoxic necrotizing factor
  • Escherichia coli

1 Introduction

Escherichia coli is the most common cause of urinary tract infections (UTIs), responsible for 75–80% of all community-acquired and a substantial minority of complicated UTIs [1]. It is estimated that 40–50% of women will have a UTI during their lifetime [2]. It has been appreciated for many years that E. coli strains isolated from clinical cases of UTI in normal hosts (uropathogenic E. coli or UPEC) differ from strains isolated from the feces of healthy individuals [3]. Putative virulence factors of UPEC have generally been identified through epidemiological studies that have revealed a higher prevalence of certain phenotypic traits in these strains than in strains of E. coli from other sources. Thus, P fimbriae, members of the Dr family of adhesins, members of the Sfa family of fimbriae, type 1 fimbriae, hemolysin, cytotoxic necrotizing factor 1 (CNF1), aerobactin, serum resistance, the ability to grow in urine, capsule and certain O antigen groups have all been implicated in the pathogenesis of UTI on epidemiological grounds [3,4]. However, associations between phenotypic traits and strain origin are not sufficient to prove that these traits contribute to virulence. The appreciation that the genes encoding many of these putative virulence factors are linked within pathogenicity islands underscores the fallacy of such assumptions [57]. In contrast, careful studies using precise mutations of the genes encoding putative virulence factors in UPEC strains followed by comparative studies in relevant animal models can be invaluable in confirming or refuting hypotheses regarding the roles of these factors in pathogenesis. Ideally, these studies include a group of animals challenged with the mutant strain complemented with the wild-type allele of the gene under investigation before the trait is considered to be required for virulence [8]. Thus far only type 1 fimbriae have been confirmed by studies of this type to be important for acute infections of the urinary tract [9].

CNF1 is a fascinating protein that induces the formation of stress fibers in cells by deamidating and constitutively activating members of the Rho family of small actin regulatory GTPases [10,11]. CNF1 is found with higher frequency in UPEC than fecal E. coli, which is not surprising given that CNF1-positive strains are almost always positive for hemolysin, with which CNF1 is genetically linked on the same pathogenicity island [1215]. Treatment of HEp-2 cells with CNF1 causes them to be able to internalize latex beads and non-invasive bacteria [16]. CNF1 also induces effacement of microvilli [17] and, according to some reports, increases permeability in polarized intestinal cell monolayers [18]. These effects suggest that CNF1 may play a role in UTI. However, no direct experimental evidence that CNF1 contributes to the pathogenesis of UTI has been published and CNF1 was not found to play a role in enteropathogenicity in rabbits [19]. We found no evidence that CNF1 and hemolysin act in concert to damage human bladder cells in vitro [20]. However, we recently reported that CNF1 impedes wound healing in an in vitro model using human bladder epithelial cells [21]. This observation led to the hypothesis that CNF1 and hemolysin work in concert to damage the bladder epithelium, expose the underlying extracellular matrix, and facilitate attachment of bacteria to underlying receptors. The purpose of this study was to test the hypothesis that CNF1 contributes to colonization, persistence and tissue injury by E. coli in the urinary tract in a murine model of ascending UTI.

2 Materials and methods

2.1 Strains

E. coli strain F11 was recovered from the urine of a patient with symptoms of cystitis [22]. DNA probe testing revealed that strain E. coli F11 is positive for hemolysin, P fimbriae, a member of the Dr family of adhesins, and aerobactin [20,22]. This strain causes cystitis in a murine model of ascending UTI and, in fact, colonizes the bladder in higher numbers than does a prototypic pyelonephritis strain [23]. The construction of E. coli strain F11.297, a CNF1-negative derivative of F11 that contains an insertion mutation in the cnf1 gene, and verification that it lacks CNF1 activity have been described previously [20].

2.2 Murine model of ascending urinary tract infection

An inoculum of approximately 2×109 live E. coli F11 or E. coli F11.297 was introduced into the urinary tract of female CBA/J/Hsd mice via the transurethral route. The procedures for preparation of the inoculum, transurethral challenge, documentation that reflux of the inoculum into the kidneys did not occur and quantitative culture of kidneys, bladder and urine have been described in detail previously [23]. The geometric mean colony counts from both kidneys from each animal were analyzed as a single data point. Since the lower limit of detection in this model is greater than 100 colony forming units (cfu) per ml of urine or gram of tissue, samples from which no organisms were cultured were assigned this value.

2.3 Histological evaluation of kidneys and bladders

One half of each organ was preserved in 10% neutral buffered formalin for histological examination. The procedures for tissue preparation and blinded assessment of histological changes in the kidney have been previously reported [23]. In the bladder mucosa and submucosa the degree of acute inflammation was graded as follows. No inflammation equaled 0, few neutrophils equaled 1, scattered neutrophils not forming micro-abscesses equaled 2, and numerous neutrophils in clusters equaled 3. Chronic inflammation was graded based on the degree of lymphocytes and plasma cells in the submucosa as follows. None equaled 0, rare equaled 1, small aggregates measuring <100 µm equaled 2 and larger aggregates equaled 3. The thickness of the epithelium was evaluated and the degree of hyperplasia was graded from 0 to 4. Epithelial morphology identical to that of controls (2–3 cell layers) equaled 0, epithelium with 3–4 cell layers and normal cytoplasmic surface maturation equaled 1, epithelium with 3–4 cell layers and reduced cytoplasmic volume in surface cells equaled 2, irregular epithelial crowding with patchy areas showing >4 cell layers and nuclear crowding equaled 3, epithelium with diffuse thickening, >4 cell layers and nuclear crowding/palisading equaled 4. Increased epithelial cell turnover was graded as normal (0), mild (1), moderate (2) and severe (3) based on the presence of apoptotic bodies. None equaled 0, extremely rare equaled 1, occasional equaled 2, and numerous equaled 3. Mitoses were present in grade 3. Scores from both kidneys from each animal were averaged and analyzed as a single data point.

2.4 Statistical analyses

The Mann-Whitney test was used to compare the numbers of cfu recovered from urine and tissues. Student's t-test was used to compare the histological scores. The χ2 test was used to compare the proportions of urine and tissue samples that contained bacteria in numbers greater than the minimal detection level.

3 Results and discussion

3.1 Effect of CNF1 on colonization and persistence of E. coli in the urinary tract

We conducted three separate trials comparing the ability of strains F11 and F11.297 to colonize and persist in the urinary tracts of 10 mice each. The actual number of organisms inoculated, as assessed by viable counts, varied from 1×109 to 3.5×109 cfu and the ratio of the number of organisms inoculated in each trial did not exceed 1.2. In each trial, the CNF-negative mutant E. coli strain F11.297 colonized the mouse urinary tract in greater numbers than did wild-type E. coli strain F11 at 7 days after transurethral challenge. In the first trial (Fig. 1), differences in colonization (median log10 cfu, interquartile range F11 vs. F11.297) in urine (3.0, 2.0–4.9 vs. 5.0, 2.0–6.8), bladder (5.2, 5.1–5.2 vs. 4.9, 4.8–5.3), and kidneys (2.0, 2.0–3.9 vs. 3.8, 2.0–4.2) were not statistically significant (P values for all comparisons >0.2). In the second trial the CNF-negative mutant E. coli strain F11.297 colonized mouse urine and kidneys in significantly greater numbers than did wild-type E. coli strain F11 (P<0.001 for each). Upon closer inspection, it was apparent that the CNF-negative mutant E. coli strain F11.297 colonized the urine, bladder and kidneys in numbers that were similar to those of the first trial. However, wild-type E. coli strain F11 colonized the urine and kidney of only one of 10 mice, a result that was discordant from the first trial and from previously published experiments [23]. In contrast, colonization of the bladder by the mutant was similar in both trials, raising doubt as to the validity of these findings. These discordant results between trials 1 and 2 prompted a third trial. The results of this third trial were similar to those of the first trial. Thus, we decided to analyze the pooled results from trials 1 and 3, excluding the results from trial 2 in which the wild-type strain colonized at an unusually low level. These analyses revealed no significant differences between the wild-type E. coli strain F11 and the CNF1-negative mutant E. coli strain F11.297 in colonization of the urine (P=0.207), bladder (P=0.628) or kidneys (P=0.144) of mice at 7 days. We calculate that the power of our studies to detect a significant difference, had there been one, of 3.8 logs in the urine, 1.3 logs in the bladder and 1.6 logs in the kidneys was 80%. Similarly, an analysis of the proportion of specimens with cfu =102 per ml of urine or gram of tissue revealed no significant differences in the proportion of animals that were colonized at detectable levels at any site. The ratio (wild-type/mutant) of the proportion of tissues that were colonized at detectable levels was 1.40 (95% confidence interval 0.57–3.52) for urine, 1.00 (0.51–1.96) for bladder and 1.07 (0.49–2.32) for kidneys.

Figure 1

The role of CNF1 in colonization of the mouse urinary tract. Results indicate paired comparisons from three separate trials in which mice were challenged by the transurethral route with wild-type E. coli F11 or CNF-negative mutant E. coli F11.297. Mice were cultured 7 days after challenge. The left ordinate indicates the median and interquartile range of the number of cfu recovered (in log10) per ml of urine or per gram of bladder or kidney homogenate. The right ordinate indicates the percentage of the 10 mice in each group (squares) that were colonized at levels exceeding 100 cfu per ml of urine or gram of bladder or kidney homogenate.

To determine whether CNF1 is required for colonization of the urinary tract at earlier time points, we conducted two trials in which we challenged 10 mice each with approximately 109 cfu of wild-type E. coli strain F11 and the CNF1-negative mutant E. coli strain F11.297 and determined the number of cfu per ml of urine or gram of tissue at 48 h. The results were similar to those obtained at 7 days: there was no significant difference in the proportion of animals colonized or the number of cfu recovered from the urine, bladder or kidneys between the groups (data not shown).

3.2 Effect of CNF1 on inflammation, hyperplasia and cell turnover in the urinary tract

We analyzed the bladder and kidney specimens from experiments 1 and 3. There were no consistent differences in any histological parameter between mice infected with the wild-type E. coli strain F11 and the CNF1-negative mutant E. coli strain F11.297 (Table 1). For unclear reasons, the CNF-1 mutant strain elicited more severe acute and chronic inflammation in the kidneys, but not the bladder in trial 1, but not trial 3. Conversely, the wild-type strain elicited more severe chronic inflammation than the CNF1-negative mutant E. coli strain in the bladder, but not the kidneys in trial 3, but not trial 1. Since the specimens were identified by code number, there was little chance of bias or labeling error to account for these findings.

View this table:
Table 1

Effect of CNF1 on inflammation, hyperplasia and cell turnover in the murine urinary tract

Histological parameterTrialWild-type strainCNF1-negative mutantP value
Number of observationsScore (mean±S.E.M.)Number of observationsScore (mean±S.E.M.)
Acute inflammation kidneys1101.1±0.2102.2±0.2<0.001
Chronic inflammation kidneys1100.9±0.1101.8±0.1<0.001
Acute inflammation bladder1100.4±0.390.7±0.30.54
Chronic inflammation bladder1100.5±0.290.6±0.20.82
Bladder hyperplasia1100.9±0.2101.2±0.30.43
Bladder cell turnover1100.5±0.2100.9±0.30.23
  • Bladder and kidney specimens from mice infected with wild-type E. coli strain F11 or CNF-negative mutant E. coli strain F11.297 were examined without knowledge of the identity of the infecting strain and scored according to the criteria described in the text.

3.3 Conclusions

CNF1 has been linked by epidemiological studies to strains that cause UTI. To test the hypothesis that CNF1 contributes to colonization of and tissue injury in the urinary tract, we compared two strains, a CNF1-positive isolate from the urine of a patient with acute cystitis and a genetically engineered CNF1-negative mutant of that strain, in a murine model of ascending UTI. Our results fail to support the hypothesis that CNF1 plays an important role in either urinary tract colonization or tissue injury in this model. That we did not detect a consistent difference in acute or chronic inflammation was somewhat surprising, since it has been reported that CNF1 can impede the influx of neutrophils by paralyzing the actin cytoskeleton [17].

Our results leave open the possibility that CNF1 is detrimental to the ability of E. coli to infect the urinary tract. Trends in several of the trials suggest that this may be the case. However, we did not design our experiments to test this hypothesis. Furthermore, our analyses indicate that any such effect, if present, is not dramatic and would require many mice as well as complementation experiments to verify. Our results do suggest that any advantage conferred to E. coli strains by producing CNF1 is probably enjoyed in a niche other than the urinary tract.


We are grateful to J. Richard Hebel for performing the statistical analyses. This work was supported by Public Health Service Award P01 DK49720 from the National Institutes of Health and by the Department of Veterans Affairs.


  1. [1].
  2. [2].
  3. [3].
  4. [4].
  5. [5].
  6. [6].
  7. [7].
  8. [8].
  9. [9].
  10. [10].
  11. [11].
  12. [12].
  13. [13].
  14. [14].
  15. [15].
  16. [16].
  17. [17].
  18. [18].
  19. [19].
  20. [20].
  21. [21].
  22. [22].
  23. [23].
View Abstract