Author's, address (for correspondence): Natali Krekeler, Faculty of Veterinary Science, The University of Melbourne, 250 Princes Highway, Werribee 3030, Victoria, Australia. E-mail: firstname.lastname@example.org
Pyometra, a prevalent infectious uterine disease that affects intact middle-aged bitches, is typically associated with Escherichia coli. Our hypotheses were (i) that bacterial adhesion to canine endometrium differs between different stages of the oestrous cycle and (ii) that the adhesin FimH facilitates this adhesion. Twelve post-pubertal, ovariectomized greyhound bitches were treated with exogenous hormones to simulate different stages of the oestrous cycle. Tissue samples from each uterus were incubated with a pathogenic E. coli strain carrying the fimH gene, but no other adhesin genes (P4-wt)–or an E. coli strain in which fimH was insertionally inactivated (P4-∆fimH::kan)–or with phosphate-buffered saline as a negative control. After washing, tissue samples were homogenized for quantification of adherent bacteria. The differences in binding to canine endometrium at different stages of the oestrous cycle were not significant. However, the mean difference in binding of the P4-wt and the P4-∆fimH::kan across all stages of the simulated oestrous cycle was significant (p < 0.001 by paired t-test on geometric means). Individual differences in numbers of P4-wt bacteria bound between dogs might suggest genetic variations or epigenetic differences in FimH receptor expression by the endometrium, unrelated to the stage of the oestrous cycle.
Uropathogenic Escherichia coli is the pathogen most commonly isolated from canine uteri with pyometra (Chen et al. 2003). Bacterial attachment to mucosal membranes, the first step in infection, is commonly facilitated by adhesins, which bind to glycoconjugate receptors on the mucosal surface (de Man et al. 1988). The fimH adhesin gene can be detected in 100% of E. coli pyometra isolates from bitches by PCR (Chen et al. 2003). FimH has been shown to mediate E. coli binding to canine endometrial epithelial cells (Krekeler et al. 2012). The majority of dogs with pyometra present clinically in diestrus (Blendinger and Bostedt 1991).
The number of E. coli binding to canine uterine epithelium has been found to vary with different stages of the oestrous cycle, peaking during diestrus (Sandholm et al. 1975; Ishiguro et al. 2007) or during simulated diestrus (Arora et al. 2008), rather than in oestrus/simulated oestrus or anestrus. The objective of this study was to examine the role of different stages of the canine oestrous cycle on bacterial binding and to determine the role of FimH on this binding.
Materials and Methods
Escherichia coli strain
An E. coli strain (P4), described previously (Chen et al. 2003), derived from a bitch with pyometra was used in this study.
Detection of fimH by PCR
Strain P4 carried fimH but not the other known E. coli adhesin genes, papGIII and sfa. The chromosomal gene fimH was insertionally inactivated with a kanamycin resistance gene cassette to generate a knockout mutant (P4-∆fimH::kan), as described previously (Krekeler et al. 2012).
Preparation of bacteria for in vitro inoculation with canine endometrial biopsies
One bacterial colony each of P4-wt and P4-∆fimH::kan were inoculated into separate 5-ml volumes of LB broth supplemented with ampicillin 100 μg/ml and incubated for a total of 48 h at 37°C, being subcultured after 24 h. The bacteria were then washed and collected by centrifugation. The optical density was adjusted to approximately 0.4 at 600 nm.
Tissue incubation with bacteria
Four uterine punch biopsies were used for each treatment. The biopsies were incubated with 3 ml of either P4-wt or P4-∆fimH::kan in PBS (107 cells/ml) or with PBS as a negative control, for one hour at room temperature on a rocking platform. After incubation, the biopsies were washed twice with PBS for 5 min on a rocking platform.
Enumeration of bacteria
Four biopsies from each treatment group were homogenized in a microcentrifuge tube with 1 ml of PBS using a pestle and then plated onto nutrient agar in serial dilutions (100, 10−1, 10−2, 10−3, 10−4 and 10−5) to determine the bacterial concentration per gram of tissue. Agar plates were incubated for 16 h at 37°C prior to counting the number of colonies.
This experiment was conducted with the approval (# 0811068.1) of the Animal Experimentation Ethics Committee, Faculty of Veterinary Science, the University of Melbourne. Twelve post-pubertal, ovary-intact greyhound bitches, 1.5 to 2 years of age, with no history of prior treatment with exogenous hormones or pregnancy were used in this study. At the commencement of the study, the bitches were all confirmed to be in anestrus by vaginal cytology, and a serum progesterone concentration of less than 0.2 ng/ml. All bitches were bilaterally ovariectomized using standard aseptic surgical procedures. The twelve bitches were randomly assigned to one of the three treatment groups (oestrus, diestrus and anestrus), with four animals allocated to each group. One to three weeks after ovariectomy, simulated stages of the oestrous cycle were induced using exogenous hormones, as described previously (Arora et al. 2006). Briefly, estradiol benzoate (Intervet, Bendigo East, Vic, Australia) at a daily dose of 0.6 to 4.8 μg/kg, im, for 13 days was followed by 2 mg/kg megestrol acetate (Jurox Pty Ltd., Rutherford NSW, Australia) orally, once a day for 3 or 16 days to simulate oestrus or diestrus, respectively. Untreated animals served as anestrus controls.
One bitch of each treatment group (simulated oestrus day 5, simulated diestrus day 10 and simulated anestrus) was euthanized on each of the four experimental days with intravenous pentobarbitone (Lethobarb; Virbac, Peakhurst, NSW, Australia). The bitches were euthanized rather than hysterectomized because these were former racing greyhounds deemed unfit for adoption. Each uterus was weighed and opened longitudinally. A total of 12 full-thickness punch biopsies were obtained using a 6-mm-diameter biopsy punch (Kai Medical, Oyana, Japan).
Incubation of uterine biopsy specimens with bacteria
After collection, the uterine specimens were washed twice for 5 min on a rocking platform. After washing, the 12 uterine biopsies from each bitch were randomly allocated into three treatment groups and incubated with strain P4-wt, strain P4-∆fimH::kan or in PBS (as negative controls) for 60 min on a rolling platform. Then, the tissue samples from each group were washed separately in PBS to remove any unattached bacteria.
Enumeration of bacteria
From each bitch, four biopsies from each of the three treatment groups were homogenized using a pestle in a microcentrifuge tube with 1 ml of PBS. Serial tenfold dilutions (100, 10−1, 10−2, 10−3, 10−4 and 10−5) were then inoculated onto nutrient agar plates. The plates were incubated for 16 h at 37°C, and the number of colonies was then counted. The bacterial concentration on the homogenized uterine tissues was calculated per cm2 of endometrium.
Overall differences in bacterial adherence to uterine epithelium between stages of the simulated oestrous cycle were evaluated on log10-transformed data using a two-way analysis of variance (anova), taking into consideration the stage of the cycle and the experimental treatment days. The interaction between experimental days and stage of cycle was evaluated using Tukey's test for non-additivity. Tukey's multiple comparison test was used to compare bacterial adherence at different stages of the simulated oestrous cycle. Differences in adherence to canine endometrium between P4-wt and P4-∆fimH::kan were evaluated using a paired t-test on the log10-transformed data. For all analyses, a p value of < 0.05 was considered significant. Minitab v15.1.30 (Minitab Inc., State College, PA, USA) and Stata (Stata Corporation, College Station, TX, USA) software were used for the statistical calculations.
Confirmation of insertional inactivation of fimH
Successful deletion of fimH was confirmed by PCR with primers up- and downstream of the inserted construct and subsequent DNA sequencing (Applied genetics Diagnostics, the University of Melbourne, Australia).
Quantification of P4-wildtype binding in relation to the stage of the simulated oestrous cycle
To investigate the influence of the stage of the oestrous cycle on E. coli binding to canine endometrial cells, the binding of P4-wt was assessed for each of the three simulated oestrous cycle groups. The mean number of viable P4-wt bound per cm2 of uterine tissue on the log10 scale was 5.71 (SD 0.95), 5.05 (SD 0.76) and 5.09 (SD 0.54) in simulated diestrus, oestrus and anestrus, respectively (Fig. 1). Differences in bacterial binding to uterine epithelium between experimental days within the diestrus, oestrus and anestrus groups were significant (p < 0.01; as determined by two-way anova). However, the differences in bacterial binding between stages of the simulated oestrous cycle were not significant between experimental days (p = 0.39; as determined by Tukey's test for non-additivity). The p value for the overall effect of the simulated stage of the oestrous cycle on binding of the P4-wt strain in this experiment was 0.088 when a two-way anova that included both the effect of day and stage was performed.
Quantification of P4-∆fimH::kan binding in relation to the stage of the simulated oestrous cycle
To assess the influence of the stage of the simulated oestrous cycle on FimH-mediated binding, the binding of the P4-∆fimH::kan strain to canine endometrium was assessed for each of the three simulated oestrous cycle groups. The mean number of P4-∆fimH::kan bound (in log10 CFU/cm2 of uterine tissue) was 3.54 (SD 0.76), 2.85 (SD 0.88) and 3.20 (SD 1.23) in simulated diestrus, oestrus and anestrus, respectively (Fig. 2). Again the P4-∆fimH::kan binding varied significantly between experimental days (p < 0.05 as determined by two-way anova). However, the differences between simulated stages of the oestrous cycle were also consistent between experimental days (as determined by Tukey's test for non-additivity; p = 0.66). The overall effect of the simulated stage of the oestrous cycle in this experiment on the binding of the P4-∆fimH::kan strain was not significant (p = 0.35) when a two-way anova that included both the effect of day and stage was performed.
Comparison of binding of P4-wildtype and P4-∆fimH::kan strains to canine endometrium
The mean difference in binding of the P4-wt and the P4-∆fimH::kan across all stages of the simulated oestrous cycle on the log10 scale was 2.08 (SD 0.55) and was significant (p < 0.001; as determined by paired t-test). The binding of the P4-∆fimH::kan strain was 0.8% of that of the P4-wt strain (Fig. 3).
The binding of pathogenic E. coli to canine endometrial epithelium was investigated during different stages of the simulated oestrous cycle. Bacterial binding did not vary significantly with the cycle stage. These findings contrast with those of other studies, in which significantly greater bacterial binding was seen in uteri from bitches in diestrus (Sandholm et al. 1975; Ishiguro et al. 2007) or simulated diestrus (Arora et al. 2008). Arora et al. (2008), using an E. coli strain that demonstrated mannose-resistant binding, demonstrated that more bacteria bound to endometrial epithelial cells during simulated diestrus than in simulated oestrus or anestrus. However, the overall proportion of bacteria binding was low. Moreover, accurate enumeration of bound bacteria was hindered by the insensitivity of light microscopy and the inability to ensure that the bacteria bound were indeed E. coli. Furthermore, it was not unequivocally determined in the study which adhesin was responsible for bacterial binding, as the E. coli strain utilized possessed three different adhesin genes (fimH, papGIII and sfa). However, it is unlikely that the binding was facilitated by FimH, as in this case mannose-sensitive binding would have been expected. In contrast, in the current study, the bacterial binding was proven to be mediated by FimH. It is possible that canine endometrial expression of receptors for different glycoconjugates is differentially affected by the hormonal status and thus that binding might be influenced in an adhesin-specific manner. This could be investigated in future studies.
Only one earlier study has investigated the influence of the oestrous cycle on bacterial binding to canine endometrium using similar means of bacterial quantification and a comparable number of dogs with these described in this chapter (Ishiguro et al. 2007). These authors showed significantly greater bacterial binding in diestrus. The differences in binding between the cycle stages were comparable to those observed in the current study, but the standard deviation in their data was significantly smaller. The significance of the influence of the stage of the oestrous cycle on bacterial binding in their study was assessed using a one-way anova (Ishiguro et al. 2007). It is possible that either there was no significant difference between experimental days or that all experiments were performed on the same day.
One explanation for the failure of the current study to demonstrate increased bacterial binding in diestrus is that there was significant variation in binding between experimental days. The potential for such variation had been anticipated and was addressed by processing one uterus from each oestrous cycle stage group per experimental day so that variation in preparation of inocula did not confound the results. The fact that the effect of the stage of the oestrous cycle did not vary between experimental days validates this approach. In a previous model, different bacterial strains were used to assess binding in the presence or absence of certain adhesins (Arora et al. 2008). In the current model, the insertional inactivation of virulence factors allowed the use of only one bacterial strain, thereby eliminating the possibility of confounding effects of other differences between strains.
In the current study, it is likely that differences between individual dogs underlie the large variance within each oestrous cycle group. On experimental day 2, the diestrus uterus had the least amount of P4-wt binding, whereas on all other experimental days the diestrus uterus bound the greatest number of bacteria. Quite possibly, genetic or epigenetic differences in FimH receptor expression by the endometrium that are unrelated to the stage of the oestrous cycle were responsible for this observation. It is also possible that differences in uterine innate immune defences in the dogs resulted in variation in bacterial binding. This suggests that individual differences between dogs might be the cause for differences in bacterial binding during different stages of the cycle. Recently, it has been confirmed that type 1 fimbriae facilitate adhesion of E. coli to the canine endometrium (Krekeler et al. 2012), suggesting that FimH is an important adhesin for pathogenic E. coli in the canine uterus. The present data show that the binding of a pathogenic E. coli strain, isolated from a canine uterus affected by pyometra, is facilitated by type-1 fimbriae. Insertional disruption of the fimH gene results in significantly reduced binding to canine endometrium. The formation of CFU in the P4-∆fimH::kan strain was only 0.8% that of the P4-wt strain.
In summary, the pathogenic E. coli strain P4, carrying fimH, isolated from a canine uterus affected by pyometra did not bind in significantly higher numbers to canine endometrium in any specific stage of the simulated oestrous cycle. The bacterial binding in the current study was mediated by the E. coli adhesin FimH as the P4-∆fimH::kan strain's binding ability was significantly reduced compared with that of the P4-wt strain.
Conflicts of interest
None of the authors have any conflicts of interest to declare.
N Krekeler performed all the experiments and prepared the manuscript, KM Lodge helped in design and execution of the experiments, GA Anderson undertook the statistical analysis, GF Browning and PJ Wright were instrumental in experimental design and manuscript revision, JA Charles critically revised the manuscript and assisted in the experiments.