Abstract Pathogens often inhabit the body asymptomatically, emerging to cause disease in response to unknown triggers. In the bladder, latent intracellular Escherichia coli reservoirs are regarded as likely origins of recurrent urinary tract infection (rUTI), a problem affecting millions of women worldwide. However, clinically plausible triggers that activate these reservoirs are unknown. Clinical studies suggest that the composition of a woman’s vaginal microbiota influences her susceptibility to rUTI, but the mechanisms behind these associations are unclear. Several lines of evidence suggest that the urinary tract is routinely exposed to vaginal bacteria, including Gardnerella vaginalis, a dominant member of the vaginal microbiota in some women. Using a mouse model, we show that bladder exposure to G. vaginalis triggers E. coli egress from latent bladder reservoirs and enhances the potential for life-threatening outcomes of the resulting E. coli rUTI. Transient G. vaginalis exposures were sufficient to cause bladder epithelial apoptosis and exfoliation and interleukin-1-receptor-mediated kidney injury, which persisted after G. vaginalis clearance from the urinary tract. These results support a broader view of UTI pathogenesis in which disease can be driven by short-lived but powerful urinary tract exposures to vaginal bacteria that are themselves not “uropathogenic” in the classic sense. This “covert pathogenesis” paradigm may apply to other latent infections, (e.g., tuberculosis), or for diseases currently defined as noninfectious because routine culture fails to detect microbes of recognized significance.

Author summary Millions of women suffer from recurrent urinary tract infections (rUTI) and the only treatment option is prophylactic antibiotics, which contributes to antibiotic resistance. In experimental models, Escherichia coli, the dominant UTI pathogen, establishes reservoirs inside the bladder lining; it is believed that some cases of rUTI in women may be due to these reservoirs awakening in response to triggers that are still unknown. Here we present a new mouse model that demonstrates the first clinically plausible trigger of rUTI arising from these reservoirs. Specifically, we show that bladder exposure to Gardnerella vaginalis, a common member of the vaginal microbial community, can drive the emergence of E. coli from bladder reservoirs. Furthermore, upon its exposure to the urinary tract, this vaginal organism caused severe kidney damage and other complications, suggesting that carriage of particular vaginal bacteria could also impact a woman’s risk for kidney infection. Bladder exposure to G. vaginalis is likely to occur during sexual activity in many women. Taken together, these data provide the first explanation for why certain characteristics of the vaginal microbiota have been linked with rUTI. Finally, our findings suggest that targeting specific members of the vaginal community may be an effective strategy for treating rUTI.

Citation: Gilbert NM, O’Brien VP, Lewis AL (2017) Transient microbiota exposures activate dormant Escherichia coli infection in the bladder and drive severe outcomes of recurrent disease. PLoS Pathog 13(3): e1006238. https://doi.org/10.1371/journal.ppat.1006238 Editor: Harry L.T. Mobley, University of Michigan Medical School, UNITED STATES Received: December 7, 2016; Accepted: February 14, 2017; Published: March 30, 2017 Copyright: © 2017 Gilbert et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability: All relevant data are within the paper and its Supporting Information files. Funding: This work was supported by the Center for Women’s Infectious Disease Research at Washington University School of Medicine (Pilot Research Award to NMG), by the American Heart Association: #12POST12050583 (NMG) and #14POST20020011 (NMG), by the National Science Foundation (Graduate Research Fellowship to VPO#DGE - 1143954), and by the National Institutes of Health, NIAID: R01 AI114635 (ALL) and NIDDK: R21 DK092586 (ALL) and P50 DK064540-11 (SJH, project II PI:ALL). Some of the animal studies were performed in a facility supported by NCRR grant C06 RR015502. Initial SEM studies were performed by the Research Center for Auditory and Vestibular Studies, supported by the NIH NIDCD Grant. Additional SEM was performed at the Washington University Center for Cellular Imaging (WUCCI) supported by Washington University School of Medicine, The Children’s Discovery Institute of Washington University and St. Louis Children’s Hospital, the Foundation for Barnes-Jewish Hospital and the National Institute for Neurological Disorders and Stroke (NS086741). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing interests: The authors have declared that no competing interests exist

Introduction Many disease-causing microorganisms can inhabit the human body asymptomatically, but factors that activate potentially pathogenic reservoirs to cause disease are not well understood. After an initial infection, some pathogens establish states of latency that can reactivate to cause recurrent symptomatic infections. In such cases, the problems lie not in diagnosing the recurrent infection, but rather in identifying and preventing potential triggers of reactivation and predicting which patients are at risk for developing severe manifestations of disease. An important example of bacterial latency and reactivation is bladder infection (cystitis), which has the potential to develop into life-threatening kidney and systemic disease [1–4]. The bladder is one of the most common sites of infection in humans, and urinary tract infections (UTIs) are one of the primary reasons for clinical use of antibiotics, accounting for 9% of all antibiotic use in an ambulatory care setting [5,6]. UTIs also often recur; 25% of women afflicted with acute UTI experience a recurrence within six months, and an estimated 1% of women (70 million worldwide) suffer more than six recurrent UTIs (rUTIs) each year [6]. Infection is most often caused by uropathogenic E. coli, which can establish intracellular, antibiotic-resistant bladder reservoirs in mouse models [1–3,6,7]. In humans, episodes of rUTI are often caused by the same strain of E. coli as the initial infection (~50% of cases according to whole genome sequencing and up to 82% of cases by less stringent criteria) [8]. While rUTI sometimes occurs due to reinfection of the bladder by E. coli residing in external sites (gastrointestinal tract, vagina), bladder intracellular E. coli reservoirs are regarded as another likely source of rUTI in humans [2–4,8–10]. However, clinically plausible triggers of the transition from latent bladder reservoirs to rUTI have not been identified. Several lines of evidence implicate the vaginal microbiota in host susceptibility to bladder and kidney infections. First, women with bacterial vaginosis (BV), a dysbiosis characterized by overgrowth of fastidious anaerobes including Gardnerella vaginalis, are at higher risk of experiencing UTI than women with normal vaginal communities (composed mainly of Lactobacillus spp.) [11–14]. Second, women with rUTI experience fewer recurrences after vaginal interventions that influence the microbiota, such as administration of Lactobacillus crispatus, than do those receiving placebo [15,16]. Third, sexual activity is one of the strongest risk factors for both BV and UTI and is an independent risk factor for the development of ascending kidney infection and pyelonephritis, a more serious form of UTI [6,16–22]. These clinical findings support the longstanding idea that classic uropathogens gain access to the urinary tract by mechanical transfer from nearby mucosal sites, such as the vagina. However, uropathogenic bacteria exist in mucosal sites within the context of the host microbiota, not in isolation. Thus, bacteria other than classic uropathogens like E. coli are likely to be frequently transferred from the vagina to the urinary tract. Consistent with this idea, multiple studies have isolated bacterial species from urine (Gardnerella and lactobacilli are among the most prominent) that are common members of the vaginal microbiota, supporting the interpretation that the bladder is likely to be regularly exposed to vaginal bacteria [23–26]. Therefore, we hypothesized that certain vaginal bacteria play an active role in the bladder, potentially triggering E. coli emergence from bladder reservoirs to cause rUTI and influencing the propensity for more serious forms of disease.

Discussion Latent E. coli reservoirs are regarded as a source of rUTI, but until now, clinically relevant triggers of reactivation have been unknown. Here, we used a new mouse model to show that transient exposures to G. vaginalis, a prevalent member of the vaginal microbiota and a dominant species in women with BV, is sufficient to drive E. coli emergence from reservoirs and the development of more serious manifestations of UTI, such as severe kidney infection (see model in Fig 5). Our study both shows a direct role for the vaginal microbiota in UTI pathogenesis and provides a mechanistic understanding of the cellular events leading to rUTI. We identified a single bacterial species that may explain the long-standing but poorly understood clinical association between BV and UTI. We note two possible translational outcomes of this finding. First, our data suggest that therapies aimed at reducing G. vaginalis colonization of the vagina could protect against E. coli rUTI. Second, given our finding that G. vaginalis exposures cause kidney damage and increased the likelihood of kidney and systemic infections by E. coli in mice, G. vaginalis colonization could be a risk factor for pyelonephritis in women. The possibility of preventing rUTI and subsequent pyelonephritis and systemic infection by targeting G. vaginalis is an exciting concept given the alarming global rise in multi-drug resistant E. coli [43]. PPT PowerPoint slide

PowerPoint slide PNG larger image

larger image TIFF original image Download: Fig 5. Model of Gardnerella vaginalis “covert pathogenesis.” In women, approximately half of recurrent urinary tract infection (rUTI) episodes are caused by an E. coli strain identical to the strain that caused the initial infection. Here we present a model of one mechanism of rUTI: activation of latent intracellular E. coli reservoirs in the bladder. G. vaginalis urinary tract exposure, likely a consequence of sexual activity, results in exfoliation of the bladder epithelium and damage to the kidney. Subsequent to exfoliation, E. coli emerges from intracellular reservoirs into the bladder lumen, where it can ascend into the kidney, sometimes causing severe inflammation and systemic infection. Often E. coli rUTI occurs after G. vaginalis clearance from the urinary tract. These findings have important implications for our understanding of rUTI etiology and point to G. vaginalis colonization as a potentially important marker of rUTI risk. https://doi.org/10.1371/journal.ppat.1006238.g005 Our model provides a new lens through which to view clinical findings relevant to the urinary tract. For instance, it is widely thought that the strong relationship between sexual activity and the risk of UTI [19–21] exists because E. coli or other classical uropathogens are transferred to the bladder from surrounding niches, such as the vagina. Our data suggest an alternative, albeit not mutually exclusive, model in which urinary tract exposure to certain members of the vaginal microbiota, such as G. vaginalis, can elicit E. coli emergence from reservoirs within the bladder. On this note, we emphasize that vaginal colonization with G. vaginalis can occur in women who do not meet the diagnostic criteria for BV (for example, see the table of strains and BV status in [44]). However, the finding that women with BV generally have high vaginal burdens of G. vaginalis may make urinary tract exposure upon sexual activity (and the potential consequences thereof) all the more likely. Previous reports that vaginal administration of L. crispatus reduces the incidence of subsequent rUTI have suggested that L. crispatus may act by reducing the vaginal reservoir of E. coli [15]. In light of our data, we propose an additional potential explanation for the protective effect of probiotic L. crispatus: the displacement of G. vaginalis from the vagina, thereby reducing the likelihood of G. vaginalis-induced recurrent E. coli UTI. Taken together, these observations suggest that repeated transient exposures to G. vaginalis may be broadly important in determining susceptibility to E. coli rUTI and may act through previously unrecognized routes. Our results shed light on established, albeit largely ignored, clinical findings of G. vaginalis in both the bladder and the kidney [45–51]. With respect to the bladder, our discovery that G. vaginalis causes damage to the urothelium are relevant to a study in which 25% of women with urinary frequency or dysuria had urine samples containing G. vaginalis in counts greater than 103 cfu/mL. Furthermore, G. vaginalis was isolated more frequently as a member of the “urinary microbiome” of women suffering from urgency urinary incontinence (a.k.a. overactive bladder syndrome) than healthy controls [23]. Although there is some debate regarding the impact of “urinary microbiome” bacteria and whether or not they represent an established community, our findings emphasize that long-standing colonization of the bladder may not be required for bacteria such as G. vaginalis to affect disease outcomes. In regards to the kidney, our finding that G. vaginalis was associated with markers of kidney injury parallel a study in which G. vaginalis was detected in urine collected directly from the bladder in 33% of patients with “sterile pyelonephritis” (i.e., patients with reflux scarring in the apparent absence of bacterial infection) [45]. Strikingly, G. vaginalis was localized to the upper urinary tract in 75% of these patients [45]. Despite these reports of G. vaginalis in the urinary tract, the classification of G. vaginalis as a “uropathogen” is currently controversial. We note that fastidious growth requirements make G. vaginalis unrecoverable, or at least unidentifiable, under conditions most often used by clinical microbiology labs for the culture and identification of potential uropathogens. Indeed, a recent investigation of 120 patient urines identified G. vaginalis by proteomic analysis in nearly 20% of samples, none of which were classified as containing G. vaginalis based on standard aerobic culture [52]. Thus, it is likely that the incidence of G. vaginalis in the urinary tract is underestimated. Taken together with this body of clinical findings, our data implicate G. vaginalis as a cause of urologic pathology, both together with and independent of E. coli, and strongly support further investigations into the possible role of this organism in urologic conditions of unknown or disputed etiology. Also of note, it is becoming increasingly evident that there is a broad range of genetic and phenotypic diversity among individual strains of G. vaginalis [53–55]. Future studies will reveal whether certain lineages of G. vaginalis are more frequently associated with negative effects in the urinary tract. Finally and more broadly, our work suggests a new paradigm for understanding and diagnosing recurrent infections. With few exceptions, the diagnosis and treatment of bacterial infections is founded on the assumption that the pathogen present at the time of clinical presentation is the main driver of disease. However, this classical view of pathogenesis may be overly simplistic. Our work provides a new paradigm we term “covert pathogenesis”, whereby transient exposure to organism A triggers disease caused by organism B, despite organism A being absent at the time of disease emergence. These short-lived bacterial exposures are likely to be missed in the clinical setting because they occur before disease symptom onset but nevertheless could drive both the recurrence and severity of disease caused by a recognized pathogen. “Covert pathogenesis” may be relevant in other contexts in which the natural triggers of disease have remained obscure. For instance, we see several striking parallels between our model and another extremely common latent bacterial infection: tuberculosis (TB). Approximately one-third of the world’s population harbors latent TB in the lungs, with 5–15% developing symptomatic disease [56]. The recent reports of the “lung microbiome” [57,58] echo the concept that the lung may be routinely transiently exposed to bacteria, and thus “covert pathogenesis” should be explored as a potential mechanism of reactivation of latent TB. Ultimately, our finding that short-lived microbial exposures influence disease pathogenesis may have far-reaching implications beyond the urinary tract.

Materials and methods Bacterial strains and growth conditions Uropathogenic Escherichia coli clinical cystitis isolate UTI89 containing a kanamycin resistance cassette (UTI89 kanR) [59] was grown static at 37°C under aerobic conditions in Lysogeny broth (LB) medium or on LB agar plates supplemented with 25 μg/mL kanamycin. A spontaneous streptomycin-resistant (SmR) strain of Gardnerella vaginalis derived from clinical isolate JCP8151B (from a BV-positive woman) was cultured in NYCIII media or on NYCIII agar plates supplemented with 1 mg/mL streptomycin at 37°C in an anaerobic chamber. Lactobacillus crispatus GED7756A was obtained from a vaginal swab of a BV-negative woman (Nugent score = 1). The vaginal swab was transported from the clinic to the lab using Port-A-Cu pre-reduced anaerobic transport media tubes (BD). Tubes were brought into a vinyl anaerobic airlock chamber (Coy Products) under an atmosphere maintained at approximately 1% hydrogen and 0 ppm oxygen. Within 24 hours the swab was streaked to isolation using de Man Rogosa and Sharpe (MRS) medium. All growth incubations were static at 37°C, under aerobic (E. coli) or anaerobic (G. vaginalis and L. crispatus) conditions, as we have previously described [30,44,60,61]. For infection experiments, E. coli was inoculated from an LB agar plate into 10 mL LB medium, grown ~24 h, then subcultured 1:10 in fresh LB for 18 h. E. coli inoculum was prepared in phosphate buffered saline (PBS) at OD 0.35 (~1–2 x 107 colony forming units (cfu) in 50 μL). G. vaginalis was inoculated from an NYCIII agar plate into 3 mL NYCIII media and grown overnight, ~18 h. G. vaginalis inoculum was prepared in PBS at OD 5 (~5 x 107–1 x 108 cfu in 50 μL). L. crispatus was inoculated from an MRS agar plate into 3 mL MRS media and grown overnight ~18 h. L. crispatus inoculum was prepared in PBS at OD 10 (~1 x 108 cfu in 50 μL). For experiments using heat-killed G. vaginalis, the inoculum was divided into two aliquots and one was incubated at 80°C for 10 min. Complete killing was confirmed by incubation on NYCII medium for 48 h anaerobically at 37°C. Ethics statement Mouse experiments were carried out in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals. The protocol was approved by the Animal Studies Committee of Washington University School of Medicine (Protocol Number: 20140114). Mice Five- to six-week old female C57BL/6 mice were obtained from Charles River Labs (NCI grantee order from Fredericks facility). Mice were given a regular chow diet in a specific pathogen-free facility at Washington University in St. Louis School of Medicine. Mouse infection/exposure model Mice (6–8 weeks of age) were anaesthetized with isofluorane and then inoculated transurethrally with 107 cfu E. coli (UTI89 kanR) in 50 μL sterile PBS as previously described [60,61]. Urine was collected and monitored for E. coli titers at 24 hours post infection (hpi) and weekly for 4 weeks. Mice with E. coli bacteriuria (any level of bacteria in urine) at 4 wpi, thus presumed to have E. coli kidney infection based on previous reports [62] and our own observations, were culled from the experiments. Mice that were negative for bacteriuria, thus presumed to contain intracellular E. coli reservoirs, were divided into groups for transurethral exposure to either PBS, G. vaginalis (7x107- 1x108 JCP-8151B SmR) or L. crispatus (1x108 GED7756A). For individual experiments, 20 mice per group were required, to be powered (0.8) to detect a statistically significant difference (p<0.05) between a 10% rate of spontaneous rUTI in the control group and 50% rate of rUTI in the G. vaginalis group. The groups were frequency matched based upon the time course of clearance of E. coli urine titers during the initial infection (also, E. coli titers at 24 hours post initial infection were not statistically different between groups). Mice were inoculated transurethrally with PBS, G. vaginalis or L. crispatus twice (12 h or 1 w apart). Urine was collected at 3, 6, 12, 24, 48 and 72 hpi (as indicated in text and figure legends) and bacterial titers enumerated by serial dilution and plating on selective media for either E. coli (LB+kanamycin) or G. vaginalis (NYCIII+streptomycin). Mice were humanely sacrificed (cervical dislocation under isofluorane anaesthesia) to aseptically harvest bladders, kidneys and spleens. Homogenates were prepared in 1 mL sterile PBS and plated on appropriate selective media. Bacterial burden in each sample was calculated as cfu/bladder or cfu/kidney pair. Samples with no colonies were plotted at one-half of the limit of detection. A 500 μL aliquot of each bladder and kidney homogenate was centrifuged at 13,000 rpm in a benchtop microcentrifuge for 5 minutes at 4°C and supernatants were removed and stored at -20°C for cytokine analysis, as described below. Urine sediment microscopy Eighty microliters of a 10-fold dilution of urine was cytospun onto coated Cytopro Dual microscope slides and stained with a Hema 3 staining kit (Fisher) to visualize epithelial cells and neutrophils (PMNs). Slides were observed on a Olympus Vanox AHBT3 microscope. Urine samples were scored qualitatively for the degree of epithelial exfoliation (ranging from 0 = none to 3 = robust) in a blinded fashion. Scanning electron microscopy Bladders were aseptically harvested, splayed, and fixed in EM fixative (2% paraformaldehyde, 2% glutaraldehyde in 0.1M sodium phosphate buffer, pH 7.4). Samples were prepared by critical point drying. Briefly, samples were post-fixed in 1.0% osmium tetroxide, dehydrated in increasing concentrations of ethanol, then dehydrated at 31.1°C and 1072 PSI for 16 minutes in a critical point dryer. Samples were mounted on carbon tape-coated stubs and sputter-coated with gold/palladium under argon. Bladders shown in Figs 2, 3 S5 and S6 were imaged on a Zeiss Crossbeam 540 FIB-SEM. For preliminary analysis samples were imaged on a Hitachi S-2600H SEM. ImageJ 1.49v (National Institutes of Health, USA) was used to add scale bars. Immunofluorescence and immunohistochemistry Bladders were fixed overnight in methacarn (60% methanol, 30% chloroform, 10% glacial acetic acid) at room temperature followed by paraffin embedding and histological slide preparation performed by the Washington University School of Medicine Histology Core. For immunofluorescence, slides were deparaffinized, hydrated, blocked with 1% BSA and 0.3% triton X-100 in PBS, incubated with primary antibody to uroplakin III (M-17 goat polyclonal IgG, Santa Cruz Biotechnology) in blocking buffer overnight at 4°C and secondary antibody in PBS for 30–60 minutes at room temperature. Antibodies are verified at 1DegreeBio (http://1degreebio.org/). Samples were stained with Hoechst for 5 min, mounted with ProLong Gold anti-fade mounting medium (ThermoFisher / Life Technologies) and then visualized on an Olympus BX61 fluorescent microscope using SlideBook 5.0 software. Terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) staining was performed using the In situ Cell Death Detection Kit (Roche) following manufacturer’s protocol for paraffin-embedded tissue. Immunohistochemistry analysis of cleaved caspase-3 was performed using SignalStain Apoptosis (Cleaved Caspase-3) IHC kit (Cell Signaling Technology) according to the manufacturer's protocol. Cytokine/chemokine analysis Tissue homogenate supernatants collected as described above were thawed on ice, centrifuged again at 4°C to remove any remaining particulates and then cytokine expression was measured using the Bio-Plex-Pro Mouse Cytokine 23-Plex Panel multiplex cytokine bead kit (Bio-Rad), which quantifies 23 different cytokines/chemokines. Assay was performed according to manufacturer instructions, except using 10-fold less standard and half the amount of coupled beads and detection antibodies indicated in the protocol. Serum analysis of kidney damage Blood was collected from mice at sacrifice by cardiac puncture and put into 400-μl Microtainer serum separation tubes (BD). After coagulation, Microtainer tubes were subjected to centrifugation at 15,000 × g for 5 min at room temperature and stored at −20°C. Serum was thawed on ice and creatinine levels were measured using the QuantiChrom Creatinine Assay Kit (BioAssay Systems) according to manufacturers’ instructions. Statistics The figures show individual data with each data point from a different animal with a line at the geometric mean or with box and whiskers (Min. to Max.) plots. The statistical tests used to analyze each set of data are indicated in the figure legends. For non-parametric analyses, differences between the experimental groups were analyzed with a two-tailed unpaired Mann-Whitney U test or Kruskal-Wallis test for comparisons of more than two groups using Prism GraphPad software. For analysis of cytokines, raw uncorrected P values are provided.

Acknowledgments We thank Francisca Chou, Nadum Member-Meneh and Lynne Foster for technical assistance, Deborah J. Frank, Karen Dodson, Stephen Beverley, David Sibley, and Scott Hultgren for critical reading of the manuscript, Jaclynn Lett, Matthew Joens and James Fitzpatrick for assistance with SEM analysis, Jennifer Lodge for use of her microscope and Scott Hultgren for the clinical E. coli isolate UTI89 and anti-uroplakin antibodies.

Author Contributions Conceptualization: NMG VPO ALL. Data curation: NMG. Formal analysis: NMG ALL. Funding acquisition: NMG ALL. Investigation: NMG VPO. Methodology: NMG VPO ALL. Project administration: ALL. Resources: ALL. Visualization: NMG VPO ALL. Writing – original draft: NMG. Writing – review & editing: NMG VPO ALL.