Bladder infection with uropathogenic Escherichia coli increases the excitability of afferent neurons

Abstract

Urinary tract infections (UTIs) cause bladder hyperactivity and pelvic pain, but the underlying causes of these symptoms remain unknown. We investigated whether afferent sensitization contributes to the bladder overactivity and pain observed in mice suffering from experimentally induced bacterial cystitis. Inoculation of mouse bladders with the uropathogenic Escherichia coli strain UTI89 caused pelvic allodynia, increased voiding frequency, and prompted an acute inflammatory process marked by leukocytic infiltration and edema of the mucosa. Compared with controls, isolated bladder sensory neurons from UTI-treated mice exhibited a depolarized resting membrane potential, lower action potential threshold and rheobase, and increased firing in response to suprathreshold stimulation. To determine whether bacterial virulence factors can contribute to the sensitization of bladder afferents, neurons isolated from naïve mice were incubated with supernatants collected from bacterial cultures with or depleted of lipopolysaccharide (LPS). Supernatants containing LPS prompted the sensitization of bladder sensory neurons with both tetrodotoxin (TTX)-resistant and TTX-sensitive action potentials. However, bladder sensory neurons with TTX-sensitive action potentials were not affected by bacterial supernatants depleted of LPS. Unexpectedly, ultrapure LPS increased the excitability only of bladder sensory neurons with TTX-resistant action potentials, but the supplementation of supernatants depleted of LPS with ultrapure LPS resulted in the sensitization of both population of bladder sensory neurons. In summary, the results of our study indicate that multiple virulence factors released from UTI89 act on bladder sensory neurons to prompt their sensitization. These sensitized bladder sensory neurons mediate, at least in part, the bladder hyperactivity and pelvic pain seen in mice inoculated with UTI89.

NEW & NOTEWORTHY Urinary tract infection (UTI) produced by uropathogenic Escherichia coli (UPEC) promotes sensitization of bladder afferent sensory neurons with tetrodotoxin-resistant and tetrodotoxin-sensitive action potentials. Lipopolysaccharide and other virulence factors produced by UPEC contribute to the sensitization of bladder afferents in UTI. In conclusion, sensitized afferents contribute to the voiding symptoms and pelvic pain present in mice bladder inoculated with UPEC.

INTRODUCTION

Urinary tract infections (UTIs) are among the most common bacterial infections and constitute the second cause of antibiotic prescription worldwide (1–3). UTIs have a detrimental influence on the quality of life and are more prevalent in women, affecting ∼13% of women aged 18 yr or older in the United States (4–6). Almost 90% of UTIs are caused by uropathogenic Escherichia coli (UPEC) (2, 5, 7). These infections are often recurrent and are becoming increasingly difficult to treat due to the widespread emergence of antibiotic-resistant uropathogens.

The internal surface of the urinary bladder is lined by the urothelium, a barrier epithelium that prevents the entry of microorganisms and restricts the passage of ions and metabolic products present in the urine into the bladder interstitium (8). Although the urothelium is stratified, its outermost umbrella cell layer forms the primary barrier to urine components and microorganisms. The umbrella cell barrier includes a mucin layer with antiadherent properties, a high-resistance apical membrane with a unique lipid and protein composition, and impermeable tight junctions (8). To successfully establish an infection, UPEC must enter into the bladder lumen and adhere to the host urothelium. UPEC expresses several bacterial virulence factors such as lipopolysaccharide (LPS), flagella, and α-hemolysin, which enable them to overcome the host immune response and successfully initiate and maintain urothelial infection (7, 9–11). A critical first step in the infection is the attachment of adhesive type 1 pili, expressed at the surface of most UPEC strains, to the uroplakins found at the apical membrane in umbrella cells (7, 9, 12, 13). In turn, this process initiates the internalization of the bacteria, formation of intracellular bacterial colonies, and subsequent colonization of the urothelium (7). The infection of the bladder by UPEC triggers a number of host responses that promote clearance of the bacterial infection, including local production of cytokines/chemokines, leukocytic infiltration, and exfoliation of infected bladder umbrella cells (for reviews, see Refs. 1, 3, 7, 14). These processes concur with the appearance of painful sensations as well as urinary urgency and frequency (15–17).

Bladder sensory information is carried by afferent fibers from the pelvic and hypogastric nerves (18–20). In mice, the cell bodies of afferent neurons from the pelvic nerve are located in the dorsal root ganglion (DRG) at the lumbosacral (L6−S2) level, whereas those from the hypogastric nerve reside in DRGs at the thoracolumbar (T13−L2) level. The afferent fibers that innervate the urinary bladder have conduction velocities in the C and Aδ range (18, 20). The sensory endings of unmyelinated C-fibers are more prominent in the lamina propria and underneath the urothelium, and those of Aδ-fibers are enriched in the bladder musculature (21, 22). Early reports have indicated that mechanosensitive Aδ-fibers respond to bladder distention in the physiological range, whereas C-fibers are insensitive to bladder filling under physiological conditions and respond to bladder distention only at elevated pressure (18, 23–26). This gave rise to the concept that the normal sensation of bladder filling and generation of storage and voiding reflexes depend on Aδ-fibers. However, other studies have described a mechanosensitive subpopulation of bladder unmyelinated C-fiber that responded to bladder distention in the physiological range of pressures, indicating that these fibers may also play a role in normal micturition (27–31). There is a consensus on the existence of a subpopulation of “silent” C-fiber afferents that do not respond to normal distention but become mechanosensitive after exposure to chemical mediators and irritants (18–20, 32, 33). Sensitization is a key feature of bladder afferent neurons (34–37). This process occurs when noxious or harmful stimuli affect the function and/or expression of ion channels that regulate the excitability of sensory neurons (38–44). As a result, sensory neurons exhibit decreased thresholds for action potential firing and an increase in the magnitude of the response to stimulation (i.e., aberrant firing). Sensitization of both Aδ and C bladder primary sensory neurons by inflammatory mediators, noxious stimuli, and increased urothelial permeability results in bladder overactivity and pain (18, 40, 42–47).

Irritative voiding symptoms, including urinary urgency and frequency, combined with the presence of suprapubic pain are defining characteristics of UTIs (48). In the present study, we assessed whether UPEC-induced bladder infection promotes the sensitization of bladder afferent neurons. The results of our study show that upon intravesical inoculation with UPEC strain UTI89, mice presented the hallmarks of UTIs including lymphocytic and neutrophilic infiltration, edema in the bladder wall, increased voiding frequency, and pelvic allodynia. These histological and functional changes were correlated with an increase in the excitability of bladder afferent neurons of C and Aδ origin. In addition, we show that LPS and UPEC-derived virulence factors can directly promote the sensitization of bladder afferents. Taken as a whole, the results of our study indicate that sensitized afferents contribute to the voiding symptoms and pelvic allodynia seen in mice with UTIs.

MATERIALS AND METHODS

Reagents

All chemicals were purchased from Sigma-Aldrich (St. Louis, MO) unless otherwise specified. Ultrapure LPS from E. coli 055:B5 was acquired from InvivoGen (San Diego, CA). Tetrodotoxin (TTX), a Na⁺ channel blocker, was obtained from Abcam (Cambridge, MA). Rabbit polyclonal antibody to cytokeratin-20 (KRT20) was purchased from Abcam. Minimal cross-reactivity goat secondary antibody conjugated to DyLight 649 was purchased from Jackson ImmunoResearch Laboratories (West Grove, PA).

Experimental Animals

Care and handling of the animals were carried out in accordance with the University of Pittsburgh Institutional Animal Care and Use Committee. Female C57BL/6 mice (Jackson Labs, Bar Harbor, ME) between 8 and 10 wk of age were used in this study. Mice were housed under 12:12-h light-dark cycles with free access to food and water. Mice were randomly assigned to control and treated groups. Histological and physiological analyses were performed with the researcher blinded to mouse treatment. Animals were euthanized by CO₂ inhalation, followed by a thoracotomy.

Bacterial Strain and Growth Conditions

To prepare the inoculum, UTI89 bacteria (kindly provided by Dr. Hultgren, Washington University in St. Louis) were grown under type 1 pilus-inducing conditions as previously described in the literature (49–51). Briefly, bacteria were streaked onto a Luria-Bertani (LB) agar plate from frozen stock and incubated overnight at 37°C. A single colony was then picked from the plate, inoculated into 10 mL of LB media, and statically incubated at 37°C overnight. The following day, bacteria were subcultured into fresh LB media (1:1,000 dilution) and statically grown at 37°C overnight. On the day of infection, 25 mL of bacterial culture were centrifuged at 5,000 g for 5 min at 4°C in a Sorvall Lynx 4000 centrifuge. The supernatant was decanted into a clean 50-mL Falcon tube and retained for further use, whereas the bacterial pellet was resuspended in sterile Dulbecco’s PBS (DPBS; Invitrogen, Waltham, MA) to a concentration of ∼2 × 10⁹ colony-forming units/mL.

Preparation of Bacterial Supernatants

UTI89 culture supernatants (from the stationary culture described in Bacterial Strain and Growth Conditions) were centrifuged at 5,000 g for 30 min at 4°C. The supernatant was filtered through a 0.45-µm filter (Sigma-Aldrich), collected in a clean Falcon tube, and stored at −20°C. The concentration of LPS in the supernatants was reduced using Pierce high-capacity endotoxin removal resin (Thermo Fisher Scientific) following the manufacturer’s instructions. Briefly, the supernatant was incubated with high-capacity endotoxin removal resin in batch mode (resin-to-sample ratio of 1:4) overnight with gentle shaking at 4°C. The supernatant was collected and incubated with new high-capacity endotoxin removal resin in batch mode for another night with gentle shaking at 4°C. To remove the components of the LB media and concentrate the samples, the bacterial supernatants were loaded into the receptacle of an Amicon ultra 3K concentrating device (Millipore, Burlington, MA) and centrifuged at 4,000 g for 30 min at 4°C. The concentrated samples were then diluted with sterile DPBS and centrifuged again to a final volume of 400 µL. This procedure was repeated three times. An aliquot of the concentrated supernatant was plated on LB agar to ensure lack of viable bacteria. A separate aliquot was used to determine the concentration of LPS in the sample using a chromogenic endotoxin quantification kit (A39552) as specified by the manufacturer (Thermo Fisher Scientific). Supernatants were stored at −20°C until further use.

Instillation of E. coli Into Mouse Bladders

UPEC was delivered to the bladder via a transurethral catheter as previously described (49–51). Briefly, mice were anesthetized with isoflurane, and a 24-gauge Teflon catheter (Smith Medical ASD, Minneapolis, MN) was inserted in the urethra to drain the urine. The urinary bladder was infused thereafter with 50 µL of UPEC inoculum. Controls were mock infected with 50 µL DPBS. In all cases, a slow instillation rate was used to avoid vesicoureteral reflux. After a 30-min incubation, the bladder was emptied, and mice were allowed to recover. Unless otherwise specified, experiments were performed 24 h after the inoculation of mouse bladders.

Immunofluorescent Labeling of Urinary Bladders

Bladder cross sections or whole mount tissue were processed as previously described (52–55). Briefly, animals were euthanized with CO₂, and the urinary bladder was removed, cut open, and pinned mucosal side up onto a rubber sheet submerged in Krebs buffer solution [containing (in mM) 110 NaCl, 25 NaHCO₃, 5.8 KCl, 2 CaCl₂, 1.2 MgSO₄, 1.2 KH₂PO₄, and 11 glucose, buffered at pH 7.4 by gassing with a mixture of 95% O₂-5% CO₂ (vol/vol)]. After a 30-min incubation, the tissue was fixed with 4% (vol/vol) paraformaldehyde in 100 mM sodium phosphate buffer [containing (in mM) 19 NaH₂PO₄ and 81 Na₂HPO₄; pH 7.4] for 30 min at 37°C. Fixed tissues were rinsed with PBS and then incubated in 30% (wt/vol) sucrose dissolved in PBS for ∼2 h at 4°C until the tissue lost its buoyancy and sank to the bottom of the 15-mL tube. To obtain bladder cryosections, the tissue was embedded in Optimal Cutting Temperature compound before storage at −70°C. Frozen tissue blocks were sectioned at 8 µm using a CM1950 cryostat (Leica, Buffalo Grove, IL). Sections and whole mounted bladders were labeled with primary antibodies or fluorophore-labeled probes using our previously described techniques (52–55). Umbrella cells were labeled with an antibody against KRT20, the DNA of urothelial cells and bacteria was labeled with YO-PRO-1 (Thermo Fisher Scientific), and cortical actin was labeled with Alexa 546 conjugated-phalloidin (Life Technologies). The antibody against KRT20 was validated in our previous reports (42, 52, 54–57). Images were captured using a Leica TCS SP5 CW-STED confocal microscope equipped with a ×63 glycerol objective (numerical aperture: 1.3) and low-noise hybrid detectors. The captured images were contrast corrected using Volocity (version 6.3, Perkin-Elmer, Waltham, MA).

Histopathology and Image Analysis

Mouse urinary bladders were rinsed with Krebs buffer, cut open, and incubated in gassed Krebs buffer for 30 min at 37°C. The tissue was then fixed with 4% (vol/vol) paraformaldehyde in PBS (Thermo Fisher Scientific) for 30 min at 37°C and then cut into three to four ∼2-mm-wide strips. To ensure vertical sections through the full thickness of the bladder wall, the strips were oriented along their long axis, cut surface up, and placed side by side in slotted histology cassettes. Tissue was maintained in neutral buffered formalin for at least 24 h before being embedded in paraffin. Paraffin sections (∼5 µm) were mounted on slides and stained with hematoxylin and eosin. Sections were histologically evaluated for inflammation (edema and presence of inflammatory cells such as lymphocytes and neutrophils) and disruptions of the mucosa, submucosa, and muscular tissue by a board-certified pathologist who was blinded to the treatment. Three random images were collected from each of the bladder strips using a Leica DM6000B upright microscope fitted with a ×20 HC PL-APO objective (numerical aperture: 0.8). Images were captured using a Gryphax Prokyon (Jenoptik, Jupiter, FL) color digital camera interfaced with an Apple iMac computer running Gryphax software (Jenoptik). Images were analyzed with Fiji ImageJ (National Institutes of Health). The “Cell Counter” plugin was used to assist in counting the number of inflammatory cells. In each image, the urothelium and lamina propria were selected, and the total number of lymphocytes and neutrophils was counted. The data are reported as average numbers of inflammatory cells (lymphocytes or polymorphonuclear neutrophils) per image for each individual bladder.

Tissue Edema Quantification

To assess bladder edema, urinary bladders were harvested, carefully dried with filter paper to eliminate any fluids, and weighed to obtain the wet mass. To determine the dry mass, samples were placed in an incubator at 55°C until constant weight was reached (typically after 5 days). The percentage of water tissue content was calculated using the following equation:

$$\text{Percent}\text{water}\text{content}=\frac{\text{WM}-\text{DM}}{\text{WM}}\times 100$$

where WM is wet mass and DM is dry mass.

Bacterial Load Assessment

Mouse urinary bladders, harvested from euthanized animals, were aseptically removed and placed into a tube containing 1 mL of sterile PBS. Bladders were homogenized using a 7-mL Potter-Elvehjem glass tissue grinder (Wheaton). Samples were serially diluted and plated on LB agar to determine the number of colony-forming units as previously described in the literature (49).

Assessment of Pelvic Allodynia

Fifty percent withdrawal thresholds to mechanical stimuli applied to the pelvic area were estimated with von Frey filaments (Touch Test Sensory Evaluators, North Coast Medical, Gilroy, CA) using the up-down method (58, 59). Mice were individually placed in elevated Plexiglas modular cages (Bioseb, Pinellas Park, FL) with a wire mesh platform that allow access to the lower abdominal area. Mice were acclimatized for at least 1 h before the behavioral test. Testing was initiated with a von Frey filament with a calibrated force of 0.16 g. Stimuli were applied to the lower abdominal area close to the urinary bladder for 1–3 s. Abdominal withdrawal (either contraction of the abdominal musculature or postural retraction of the abdomen) and licking or scratching in the pelvic area in response to von Frey filament application were considered a positive response. When a negative response was observed, the next stronger filament was applied, and when a positive response was observed, the next weaker stimulus was applied. After the response threshold was first crossed, four additional filaments were applied that varied sequentially up or down based on the previous response. The 50% response threshold was calculated as previously described by Chaplan et al. (59).

Video-Monitored Void Spot Assay

To evaluate micturition in awake, freely moving mice, we used a modified version of the standard void spot assay that incorporates video monitoring as we have recently described (60, 61). The system consists of an upper compartment, which housed two mouse cages side by side, each made of acrylic sheets with a ultraviolet-transmitting acrylic bottom (dimensions of 37 × 25 × 20 cm), and a lower compartment, which houses ultraviolet tube lights (model T8-F20BLB24, 24 in., ADJ). Mouse cages were furnished with the following: an igloo-shaped sleeping chamber, an Eppendorf tube as a “play toy,” and a dish with standard mouse chow and water in the form of Hydrogel (ClearH2O, Westbrook, ME). The bottom of the cages was covered with chromatography paper (Cat. No. 057144, Thermo Fisher). The lower compartments have reflective mirrored walls, which evenly illuminate the chromatography paper from below. Each mouse was monitored by wide-angle cameras (C930e, Logitech), one positioned above the cage and another mounted at the base of the lower compartment. Mice were routinely housed in a facility with 12:12-h light-dark cycles, with 7:00 AM being zeitgeber time (ZT) = 0 (start of the light cycle). Mice were introduced into the cages, and micturition behavior was recorded during the light cycle of the day (ZT = 2–3 to ZT = 9–10) for 6 h following an acclimatization period of 1 h. The streamed video was captured with an Apple iMac computer running SecuritySpy (Ben Software) at 1 frame/s with a 1,920 × 1,080 pixel resolution. Voiding events were identified by visual inspection of the movies. The void spot area was computed from video images using ImageJ Fiji, as we have previously described (60, 62). The parameters evaluated in this study were the number of voids per hour and urine volume per void (in μL). Mice that did not void during the 6-h period or those animals that chewed the paper before or during the time window of analysis were not included in the analysis. Void spot volume was estimated from a calibration curve generated with known amounts of urine (2–750 µL) spotted on the chromatography paper.

Retrograde Labeling of Bladder Sensory Neurons

Bladder afferent neurons were labeled with the fluorescent dye DiI (Invitrogen), as previously described (63). Briefly, mice were anesthetized with isoflurane, and the bladder was exposed through an abdominal incision (∼1 cm in length). Dil [5% (wt/vol) in DMSO] was injected at three to four sites (total volume: 15–20 μL) in the bladder wall with a 28-gauge syringe. After each inoculation, the needle was kept in place at the injection site for 20–30 s. The muscle layers and skin incision were individually closed with 5.0 PDO absorbable monofilament surgical suture (AD Surgical, Sunnyvale, CA). Mice received a subcutaneous injection of ketoprofen (5 mg/kg, Zoetis, Parsippany-Troy Hills, NJ) as postoperative analgesia for 3 days. Ampicillin (10 mg/kg, Boehringer Ingelheim Vetmedica, Saint Joseph, MO) was administrated for 3 days to prevent infections. Mice were housed under the conditions described earlier for at least 7 days before any further procedure was performed.

Isolation of Bladder Sensory Neurons

L6−S2 DRGs were harvested from two to three mice, as we have previously described (42, 61–64). Isolated DRGs were transferred to a 3.5-cm cell culture dish containing Neurobasal-A medium (Invitrogen). DRGs were minced and transferred to a cell culture dish containing 3 mL of Neurobasal-A medium supplemented with 10 mg of collagenase type IV (Worthington Biochemical, Lakewood, NJ) and 5 mg of trypsin (Worthington Biochemical). The cell culture dish containing the tissue was incubated for 30 min at 37°C with agitation. The tissue fragments were then gently triturated with a fire-polished glass pipette, and the cell suspension was centrifuged at 460 g for 5 min at room temperature in a Sorvall ST8 centrifuge. The pellet, containing DRG somas, was resuspended in 5 mL of complete neuro media (Neurobasal-A medium supplemented with 5% of B27 supplements, 0.5 mM l-glutamine, 100 U/mL penicillin, and 100 μg/mL streptomycin). The centrifugation and resuspension steps were repeated three times. The pellet from the last centrifugation was resuspended in 1.5 mL of complete neuro media, and the suspension was plated on coverslips coated with ornithine and laminin (Invitrogen) placed in a six-well cell culture plate. After 2 h of incubation at 37°C inside a CO₂ incubator, 3 mL of warm complete neuro media were added to each well, and the tissue culture plate was returned to the incubator. Electrophysiological experiments were performed within 2 and 10 h of plating unless otherwise stated.

Electrophysiological Experiments

Whole cell patch-clamp recordings of DRG neurons were performed using the amphotericin-perforated patch technique as we have previously described (42, 62, 63). Briefly, glass coverslips containing DRG neurons were transferred to a chamber mounted on the stage of a Nikon Ti inverted microscope equipped with a Sedat Quad set (Chroma Technology, Brattleboro, VT), a Lambda XL light source (Sutter Instruments, Novato, CA), and an ORCA-Flash 2.8 camera (Hamamatsu, Bridgewater Township, NJ). Micropipettes were pulled from borosilicate glass capillary tubes with a PP-830 puller (Narishige, Amityville, NY). Fire-polished micropipettes with a tip resistance of ∼1.5–3 mΩ were used for current-clamp recordings. The pipette filling solution contained (in mM) 145 KCl, 1 MgCl₂, 0.1 CaCl₂, 1 EGTA, and 10 HEPES (pH 7.2). The extracellular bath solution contained (in mM) 135 NaCl, 5 KCl, 1 MgCl₂, 2.5 CaCl₂, 10 glucose, and 10 HEPES (pH 7.4). Experiments were performed at room temperature with a PC-505B patch-clamp amplifier (Warner Instruments, Hamden, CT). Signals were low-pass filtered at 1 kHz (four-pole Bessel filter) and digitized with a Digidata 1440A (Molecular Devices, San Jose, CA) at 5 kHz. Cell capacitance was obtained by reading the value for whole cell input capacitance neutralization directly from the amplifier. Command protocols as well as data acquisition and analysis were controlled with pClamp 10 (Molecular Devices).

Data Analysis

The passive and active electrical properties of DRG neurons were determined in the current-clamp mode, as we have previously reported (42, 62). Bladder sensory neurons were classified based on the sensitivity of their action potential to 1 µM TTX as TTX sensitive (TTX-S) or TTX resistant (TTX-R). Based on the presence or absence of spontaneous action potentials during a 5-min period immediately after whole cell configuration was achieved, sensory neurons were classified as silent or spontaneously active. For silent neurons, only those that had a resting membrane potential more negative than −40 mV and generated action potentials with a distinct overshoot higher than 0 mV in response to depolarizing current injections were studied. The following electrical properties were measured for silent bladder sensory neurons: resting membrane potential (RMP), action potential overshoot above 0 mV, action potential duration at 0 mV, magnitude of afterhyperpolarization below RMP, action potential threshold, and rheobase. Action potential rheobase and threshold were defined as the minimum depolarizing current injection necessary to evoke an action potential and the maximum membrane potential depolarization obtained in the absence of an action potential, respectively. To examine the excitability of bladder sensory neurons, a series of 500-ms depolarizing rectangular current pulses equivalent to 1, 1.5, 2, 2.5, and 3 times the rheobase were injected every 4 s, and the voltage responses were recorded. Stimulus response relationships were generated by plotting the number of spikes evoked by the electrical stimulation as a function of the injected current.

Statistical Analysis

Data are expressed as means ± SE; n is the number of independent experiments. For statistical analyses, the normality of the data was tested. Based on this result, a parametric or nonparametric test was used to make statistical comparisons. Comparison between two groups was performed with a t test (parametric) or Mann–Whitney test (nonparametric). For parametric multiple comparisons, we used ANOVA followed by a Tukey’s multiple comparisons test; for nonparametric multiple comparisons, we used a Kruskal–Wallis test followed by a Dunn’s multiple comparisons test. P values of <0.05 were considered statistically significant. Fitting and statistical comparisons were performed with Clampfit (Molecular Device) or GraphPad Prism 8 (GraphPad Software, San Diego, CA).

RESULTS

UPEC Increases Voiding Frequency and Causes Pelvic Pain

To establish a model of UTI, we instilled the well-characterized UPEC clinical cystitis isolate UTI89 (49, 51, 65) into the bladders of 8- to 10-wk-old female mice. Significant bacterial loads were observed 24 h after bladder inoculation with UTI89, including large collections of bacteria attached to the surface of the bladder epithelium (Fig. 1, A and C). Histological analysis of paraffin-embedded bladder sections from mice inoculated with UTI89 and stained with hematoxylin and eosin showed edema in the lamina propria along with a lymphocytic and neutrophilic infiltrate (Fig. 1, B, D, and E). The integrity of the urothelium was preserved, and no desquamation of urothelial cells was observed 24 h after the inoculation with UTI89. However, we observed some areas with urothelial hypertrophy (Fig. 1B). There was no evidence of inflammation in the mucosa or lamina propria in the bladders harvested from mock-infected mice (Fig. 1B). Consistent with the histological findings, the tissue water content, a measurement of edema, was significantly elevated in bladders from mice inoculated with UTI89 compared with controls (Fig. 1F).

Figure 1.

Bladder infection with UTI89 induces cystitis. A: immunofluorescence analysis of UTI89-infected urothelium. Top: en face view of the umbrella cell layer stained with antibody against cytokeratin-20 (KRT20, blue, marker of umbrella cells), Alexa Fluor 546-conjugated phalloidin (red, marker of cortical actin cytoskeleton), and YO-PRO-1 (green, which recognizes bacterial and host DNA). Images are three-dimensional reconstructions of confocal Z-series. Bottom: cross section of a UTI89-infected mouse bladder stained with the same antibodies. B: hematoxylin and eosin-stained bladder sections from control (Ctrl) mice or animals with a urinary tract infection (UTI). Left: cross section of the bladder wall from control mouse. The urothelium was intact, and there was no evidence of inflammation. The region in the red dashed box is the magnified area shown in the inset. Fibroblasts (F) are indicated. Right: the bladder wall from a mouse with a UTI. The urothelium (UT) was intact but hypertrophied; note the leukocytic infiltrate and edema, particularly in the lamina propria (LP). The region in the red dashed box is the magnified area shown in the inset. Representative polymorphonuclear neutrophils (PMNs) and lymphocytes (L) are indicated. C: bacterial colonization of the bladder. Total uropathogenic Escherichia coli colony-forming units (CFU) were counted 24 h after infection for Ctrl (black circles, n = 8) or mice with UTIs (red circles, n = 8). D: quantitation of the average number of PMNs present in the urothelium and lamina propia per field for controls (black circles, n = 4) and mice with UTIs (red circles, n = 6). E: quantitation of the average number of lymphocytes (LYMs) present in the urothelium and lamina propia per field for controls (black circles, n = 4) or mice with UTIs (red circles, n = 6). F: water content of bladder for controls (black circles, n = 7) and mice with UTIs (red circles, n = 8). Statistically significant differences between sample means are indicated as **P < 0.01 and ***P < 0.001 (Mann–Whitney nonparametric test).

Urinary urgency and frequency as well as pain in the lower pelvic area are cardinal symptoms of UTI (3, 15, 48). Voiding function in awake control mice or those inoculated with UTI89 was evaluated using a recently described video-monitored void spot assay (60, 61). Control mice exhibited a characteristic voiding pattern with large voids on the edge of the cage (Fig. 2A, top). In contrast, mice inoculated with UTI89 exhibited an overactive bladder phenotype with high numbers of voiding events (Fig. 2, A, bottom and B). Although most of the urine voids appeared on the edges of the box in mice inoculated with UTI89, numerous void spots were present in the central area of the cage. Mice inoculated with UTI89 exhibited higher voiding activity (Fig. 2C) and a smaller mean volume per void than controls (Fig. 2D). To assess whether bladder infection with UTI89 produced changes in somatic sensitivity in the pelvic area, we measured the 50% withdrawal threshold (g) to von Frey filaments (a measure of referred pelvic allodynia) applied to this region. Consistent with the presence of bladder-derived pain, mice infected with UTI89 exhibited a lower mechanical threshold to von Frey filaments applied to the pelvic area than controls 24 h after inoculation (Fig. 2E). In summary, mice inoculated with the UPEC strain UTI89 present with the hallmark characteristics of UTIs including inflammation in the bladder mucosa and lamina propria, increased voiding frequency, and pelvic allodynia.

Figure 2.

Bladder infection with UTI89 increases voiding frequency and induces pelvic allodynia. A: voiding behavior in awake mice was evaluated with a video-monitored void spot assay. Representative images captured using the camera located under the chamber that housed a control (Ctrl) mouse (top) or a mouse with urinary tract infection (UTI; bottom) are shown. Urinary spots are marked with a red arrow. Note that the area of the urinary spot for the control mouse was significantly larger than that of the infected mouse. B: representative tracings of voiding activity during a 6-h period for a control mouse (black line) or a mouse with UTI (red line). Each step represents a single voiding event. C: number of voids for Ctrl mice (black circles, n = 12) or mice with UTIs (red circles, n = 12). D: urine volume per void for Ctrl mice (black circles, n = 12) or mice with UTIs (red circles, n = 12). E: UTI reduced the withdrawal threshold to mechanical stimuli applied to the pelvic area. The 50% mechanical withdrawal threshold (g) was estimated with von Frey filaments applied in an up-down testing paradigm on the lower pelvic area for Ctrl mice (black circles, n = 10) or mice with UTIs (red circles, n = 10). Statistically significant differences between sample means are indicated as **P < 0.01 and ***P < 0.001 (Mann–Whitney nonparametric test).

UTI Sensitizes Bladder Sensory Neurons

Nociceptive information from visceral structures is transmitted to the central nervous system by DRG neurons. An important property of these neurons is their ability to sensitize, a process that can affect organ function and results in the development of pain (34, 66). To assess whether bacterial cystitis alters the excitability of bladder afferents, we evaluated the active and passive electrical properties of bladder sensory neurons harvested from mice inoculated with UTI89 or DPBS. Four distinct TTX-S subunits (1.1, 1.2, 1.6, and 1.7) and two TTX-R subunits (1.8 and 1.9) of voltage-gated Na⁺ (Na_v) channels are known to be differentially expressed in subpopulations of DRG neurons (67–69). Although the cell bodies of C-fibers express both TTX-S (1.7) and TTX-R (1.8, and 1.9) Na_v subunits, myelinated sensory neurons (i.e., Aδ-fibers) predominantly express TTX-S (1.1, 1.6, and 1.7) Na_v subunits (67–73). Functional experiments showed that most (∼70%) of the neurons innervating the urinary bladder are neurofilament 200-negative C-fibers that are sensitive to capsaicin and exhibit TTX-R action potentials, whereas the other 30% of the neuronal population is comprised of myelinated Aδ afferent neurons that are insensitive to capsaicin and have action potentials sensitive to TTX (TTX-S) (40, 74, 75). Our experiments indicated that a large proportion (∼70%) of bladder sensory neurons harvested from mice exhibited TTX-R action potentials (Table 1). Sensory neurons with TTX-R action potentials from control mice presented action potentials with longer duration and a more depolarized activation threshold than neurons with TTX-S action potentials (Table 1). Whereas bladder sensory neurons harvested from control mice were electrically silent (0 of 45 neurons), a small fraction (2 of 44 neurons) of bladder sensory neurons from mice inoculated with UTI89 exhibited spontaneous action potential firing (data not shown). Consistent with sensitization, electrically silent bladder sensory neurons from mice inoculated with UTI89 exhibited a depolarized RMP, lower action potential threshold, and lower action potential rheobase than those from control mice (Table 1).

Table 1.

Passive and active electrophysiological properties of bladder dorsal root ganglion neurons from control mice or mice with UTIs

	Control		UTI
	TTX-S	TTX-R	TTX-S	TTX-R
Number of cells, n (%)	14 (32)	31 (68)	12 (32)	31 (68)
Resting membrane potential, mV	−60 ± 1	−61 ± 1	−50 ± 2d	−50 ± 3d
Cm, pF	47 ± 4c	26 ± 2	48 ± 2c	24 ± 2
RIn, MΩ	283 ± 19c	582 ± 51	260 ± 31c	684 ± 70
Threshold, mV	−26 ± 1b	−22 ± 1	−30 ± 1b,d	−24 ± 1d
Duration, ms	2.2 ± 0.1c	3.4 ± 0.2	2.4 ± 0.1c	3.5 ± 0.2
Overshoot, mV	28 ± 1	25 ± 4	26 ± 2	29 ± 3
Rheobase, pA	829 ± 115	761 ± 107	292 ± 38d	320 ± 66f
Afterhyperpolarization magnitude, mV	−76 ± 1	−77 ± 1	−74 ± 1	−73 ± 2

Values are means ± SE. TTX-R, tetrodotoxin resistant; TTX-S, tetrodotoxin sensitive; UTI, urinary tract infection.

Statistically significant differences are indicated as follows: ^bP < 0.01 and ^cP < 0.001 vs. TTX-R counterparts and ^dP < 0.05 and ^fP < 0.001 vs. control counterparts (Kruskal–Wallis test followed by a Dunn’s multiple comparisons test).

To investigate whether UTIs alter the excitability of bladder sensory neurons, we injected current pulses equivalent to 1, 1.5, 2, 2.5, and 3 times the rheobase for 500 ms with 4-s intervals. Previous studies have revealed that sensory neurons exhibit a phasic pattern of action potential firing in response to suprathreshold stimulation (39, 40, 42, 74). In good agreement with previous studies, bladder sensory neurons with TTX-R (Fig. 3, A and B) and TTX-S (Fig. 3, C and D) action potentials from control mice discharged a single action potential in response to suprathreshold stimulation. In contrast, bladder sensory neurons with TTX-R (Fig. 3, A and B) and TTX-S (Fig. 3, C and D) action potentials from mice inoculated with UTI89 exhibited sustained firing in response to electrical stimulation. In summary, our results show that inoculation of UTI89 in mice promotes the sensitization of bladder sensory neurons with TTX-S and TTX-R action potentials. This finding supports the notion that the irritative voiding symptoms and pelvic allodynia seen in mice inoculated with UTI89 are mediated, at least in part, by sensitized bladder afferents.

Figure 3.

Urinary tract infection (UTI) sensitizes bladder sensory neurons. A: representative tracings of the action potential firing pattern in response to suprathreshold stimulation for bladder sensory neurons with tetrodotoxin-resistant (TTX-R) action potentials from control (Ctrl) mice or mice with UTIs. B: stimulus response relationships for bladder sensory neurons with TTX-R action potentials for control mice (black circles, n = 31 neurons from 3 independent preparations) or mice with UTIs (red circles, n = 31 neurons from 3 independent preparations). C: representative tracings of the action potential firing in response to suprathreshold stimulation for bladder sensory neurons with tetrodotoxin-sensitive (TTX-S) action potentials from control mice or mice with UTIs. The current pulse protocol is shown at the bottom. D: stimulus response relationships for bladder sensory neurons with TTX-S action potentials for control mice (black circles, n = 14 neurons from 3 independent preparations) or mice with UTI (red circles, n = 12 neurons from 3 independent preparations). Statistically significant differences with controls are indicated as †P < 0.05 and *P < 0.001 (Kruskal–Wallis test followed by a Dunn’s multiple comparisons test). Rh, rheobase.

Multiple Bacterial Factors Contribute to the Sensitization of Bladder Sensory Neurons

Bladder infections trigger a robust innate inflammatory response. It has been widely assumed that the pain related to bacterial infections is secondary to the activation of the immune system. However, recent studies have shown that several virulence factors derived from various bacterial strains can modulate the excitability of somatic and visceral nociceptive sensory neurons (76–83). To assess the contribution of the virulence factors released from UTI89 to afferent sensitization, we examined the excitability of bladder sensory neurons exposed in vitro to bacterial supernatants. Bladder sensory neurons isolated from naïve mice were incubated for 24 h in neurobasal media supplemented with either supernatant isolated from bacterial cultures [SUP(+); LPS concentration: ∼10 µg/mL] or supernatant filtered through high-capacity endotoxin removal resin [SUP(−); LPS final concentration: <0.05 µg/mL]. Consistent with sensitization, bladder sensory neurons with TTX-S and TTX-R action potentials incubated with SUP(+) exhibited a depolarized RMP and lower action potential threshold and rheobase than controls (Table 2). Moreover, bladder sensory neurons with TTX-R (Fig. 4, A and B) and TTX-S (Fig. 4, C and D) action potentials incubated with SUP(+) exhibited aberrant firing in response to suprathreshold stimulation. Of major significance, treatment with SUP(−) resulted in the sensitization of bladder sensory neurons with TTX-R action potentials (Fig. 4, A and B, and Table 2) but not those with TTX-S action potentials (Fig. 4, C and D, and Table 2). These results indicate that bacterial virulence factors derived from UPEC UTI89, including LPS, act directly on bladder DRG neurons to prompt their sensitization.

Table 2.

Passive and active electrical properties of bladder dorsal root ganglion neurons treated with bacterial supernatants or controls

	Control		SUP(+)		SUP(−)		LPS		LPS + SUP(−)
	TTX-S	TTX-R	TTX-S	TTX-R	TTX-S	TTX-R	TTX-S	TTX-R	TTX-S	TTX-R
Number of cells, n (%)	29 (36)	51 (64)	22 (39)	34 (61)	14 (30)	33 (70)	6 (25)	18 (75)	9 (27)	24 (73)
Resting membrAne potential, mV	−61 ± 1	−60 ± 1	−53 ± 1a	−53 ± 1b	−61 ± 1	−50 ± 1b	−63 ± 1	−50 ± 1b	−51 ± 2b	−51 ± 1b
Cm, pF	51 ± 3	30 ± 1	43 ± 3	27 ± 1	52 ± 4	28 ± 1	43 ± 5	25 ± 1	48 ± 2	25 ± 1
RIn, GΩ	304 ± 19	730 ± 35	294 ± 24	719 ± 26	288 ± 16	739 ± 22	243 ± 56	647 ± 33	281 ± 37	701 ± 36
AP threshold, mV	−27 ± 1	−23 ± 1	−30 ± 1b	−26 ± 1b	−27 ± 1	−26 ± 1b	−28 ± 1	−28 ± 1b	−29 ± 1a	−28 ± 1b
AP duration, ms	2.1 ± 0.1	3.2 ± 0.1	2.1 ± 0.1	3.2 ± 0.1	2.1 ± 0.1	3.2 ± 0.1	2.1 ± 0.1	3.3 ± 0.2	2.2 ± 0.1	3.2 ± 0.1
AP overshoot, mV	33 ± 2	34 ± 1	31 ± 1	32 ± 1	29 ± 2	32 ± 1	34 ± 5	36 ± 3	36 ± 2	35 ± 1
AP rheobase, pA	1,227 ± 153	1,169 ± 114	111 ± 22b	114 ± 14b	914 ± 189	108 ± 15b	917 ± 202	106 ± 23b	171 ± 63b	96 ± 11b
Afterhyperpolarization magnitude, mV	−76 ± 1	−76 ± 1	−76 ± 1	−76 ± 1	−74 ± 1	−75 ± 1	−75 ± 1	−75 ± 1	−72 ± 1	−72 ± 1

Values are means ± SE. AP, action potential; SUP(+), supernatant isolated from bacterial cultures; SUP(−), supernatant filtered through high-capacity endotoxin removal resin; TTX-R, tetrodotoxin resistant; TTX-S, tetrodotoxin sensitive.

Statistically significant differences compared with control are indicated as follows: ^aP < 0.05 and ^bP < 0.001 (Kruskal–Wallis test followed by a Dunn’s multiple comparisons test).

Figure 4.

Bacterial cell products derived from UTI89 increase the excitability of bladder sensory neurons. A: representative tracings of the action potential firing in response to suprathreshold stimulation for bladder sensory neurons with tetrodotoxin-resistant (TTX-R) action potentials under control conditions, exposed to UTI89 supernatants [SUP(+)], exposed to lipopolysaccharide (LPS)-reduced supernatant [SUP(−)], exposed to 10 µg/mL ultrapure Escherichia coli-derived LPS, or exposed to LPS-reduced supernatant supplemented with 10 µg/mL ultrapure LPS [LPS + SUP(−)]. B: stimulus response relationships for bladder sensory neurons with TTX-R action potentials under control conditions (black circles, n = 51 neurons from 7 independent preparations), exposed to SUP(+) (blue circles, n = 34 neurons from 4 independent preparations), exposed to SUP(−) (red circles, n = 33 neurons from 3 independent preparations), exposed to LPS (open circles, n = 18 neurons from 2 independent preparations), or exposed to LPS + SUP(−) (green circles, n = 24 neurons from 2 independent preparations). C: representative tracings of the action potential firing in response to suprathreshold stimulation for bladder sensory neurons with tetrodotoxin-sensitive (TTX-S) action potentials under control conditions, exposed to SUP(+), exposed to SUP(−), exposed to LPS, or exposed to LPS + SUP(−). The current pulse protocol is shown at the bottom. D: stimulus response relationships for bladder sensory neurons with TTX-S action potentials under control conditions (black circles, n = 29 neurons from 7 independent preparations), exposed to SUP(+) (blue circles, n = 22 neurons from 4 independent preparations), exposed to SUP(−) (red circles, n =14 neurons from 3 independent preparations), exposed to LPS (open circles, n = 6 neurons from 2 independent preparations), or exposed to LPS + SUP(−) (green circles, n = 9 neurons from 2 independent preparations). Statistically significant differences compared with control are indicated as #P < 0.01 and *P < 0.001 (Kruskal–Wallis test followed by a Dunn’s multiple comparisons test).

To further assess whether E. coli-derived LPS per se can mediate bladder sensory neuron sensitization, we incubated these cells with 10 µg/mL of ultrapure LPS (from E. coli 055:B5) for 24 h. Our experiments revealed that bladder sensory neurons with TTX-R action potentials exposed to ultrapure LPS, but not those with TTX-S action potentials, exhibited a more depolarized RMP and lower action potential threshold and rheobase than controls (Table 2). The observed changes in the electrical properties of bladder sensory neurons with TTX-R action potentials correlated with increased excitability (Fig. 4, A and B). Although ultrapure LPS failed to sensitize bladder sensory neurons with TTX-S action potentials (Fig. 4, C and D), when it was combined with SUP(−) the electrical properties of both TTX-R and TTX-S bladder sensory neurons were altered (Table 2). After 24 h of incubation with LPS + SUP(−), both groups of bladder sensory neurons showed aberrant firing in response to suprathreshold stimulation (Fig. 4). These results show that multiple E. coli-derived virulence factors are capable of sensitizing different population of bladder sensory neurons, either acting on their own (e.g., LPS) or through synergistic mechanisms.

DISCUSSION

UTIs are among the most common infectious diseases, resulting in a major public health problem with a high economic burden (2). Although progress has been made in understanding the pathogenic mechanisms involved in the colonization of the urothelium by bacteria and the host immune response to UTI (for reviews, see Refs. 1, 2, 7, 14, 84–87), the molecular mechanisms that mediate UTI-associated sensory symptoms remain largely unknown. The goals of this study were to gain understanding of the mechanism involved in the generation of irritative voiding symptoms and pain in a murine model of UTI and to define the contribution that bacterial factors have to these processes.

The results of our study show that mice inoculated with the strain UTI89 present the hallmark features of UTI with inflammation in the bladder mucosa and lamina propria, increased voiding frequency, and pelvic allodynia. Pain is a complex sensory process mediated by peripheral and central neuronal pathways. Sensitization of primary afferent fibers is a fundamental process in pain perception, since these fibers transmit information from the peripheral organs to the central nervous system (34, 35, 66, 88). Sensitization of and persistent afferent drive contribute to acute and chronic bladder-derived pain, and blockage of nociceptive input into the central nervous can effectively attenuate ongoing bladder pain (40, 42, 43, 89). We showed that bladder sensory neurons with TTX-R or TTX-S action potentials from mice inoculated with UTI89 exhibited aberrant firing in response to suprathreshold stimulation as well as a reduction in the action potential threshold and rheobase. These findings indicate that afferent sensitization contributes, at least in part, to the pelvic allodynia seen in mice with UTI. Our data do not allow us to exclude a possible contribution of the detrusor to the bladder overactivity observed in mice with UTI. Although intravesical administration of UPEC has been shown to induce detrusor hyperreflexia in the short-term (1 h after infection), the contractile response of the bladder smooth muscle was normalized after 24 h of infection (90, 91). Because we conducted our experimental measurements 24 h after infection, it seems unlikely that the reported alterations in detrusor contractility contribute to the increased bladder activity observed in our study. Taking the aforementioned into consideration, our results support the notion that sensitization of bladder afferents plays a central role in the generation of voiding symptoms and pain observed in patients with UTIs.

UTIs induce a robust innate immune response with local production of cytokines and chemokines by urothelial and resident immune cells that rapidly initiate the migration of neutrophils to the bladder (1, 14). In addition to neutrophils, T cells (natural killer, CD4⁺, CD8⁺, and γδ) are also recruited to the infected mouse bladder (92–95). The adaptive immune response is thought to be ineffective in eradicating bacteria during bladder infections in humans, and it does not prevent recurrent infections (92, 96–99). A recent study has shown that a subtype of CD4⁺ T cells (type 2) that are activated and recruited into the bladder following infection are relevant for tissue repair but are markedly deficient in bacterial clearance properties, suggesting that lymphocytes may be involved in the repair of the urothelium during UTI (97). Inflammatory mediators are also known to promote or contribute to the sensitization of somatic and visceral sensory neurons (for reviews, see Refs. 19, 34, 66, 100–102). Recently, the effect of proinflammatory mediators released during UTI on bladder afferent nerves was investigated (103). An ex vivo study with a bladder-nerve preparation showed that the supernatants of bladder homogenates from mice with UTI, which contain elevated concentrations of inflammatory mediators, enhance the firing of high-threshold afferents in response to distention and recruit a population of silent nociceptors to became mechanosensitive (103). However, pain sensitivity during infections seems to correlate better with the bacterial load in the tissue than with the kinetics of edema or the accumulation of immune cells (78). Indeed, no correlation was found between the pain observed during UTI and the degree of inflammation (15, 17). In addition, several studies have shown that bacteria cell products can directly stimulate afferent fibers (76, 78, 80, 104–107). The results of our study support the notion that bacterial cell products also play an important role in the sensitization of afferent pathways in mice inoculated with UTI89.

Uropathogens express various virulence factors that enable them to successfully initiate and maintain urothelial colonization. Because terminal fibers of bladder sensory neurons reside in the urothelium and in a subepithelial plexus within the lamina propria (21), bacterial components and toxins released in the course of urothelial infections may affect their function. Previous studies have shown that exposure of mouse bladders to UPEC or purified LPS triggers pain responses by mechanisms that depend on Toll-like receptor 4 and are independent of the recruitment of immune cells (15, 16). To assess the contribution of bacterial virulence factors to bladder sensory neuron sensitization, we cultivated neurons isolated from naïve mice with supernatants collected from UTI89 cultures. In good agreement with our experiments in mice inoculated with UTI89, we found that the treatment of bladder sensory neurons in vitro with bacterial supernatants resulted in their sensitization. Of major significance, the results of our study demonstrate that bacterial supernatants with reduced LPS content or ultrapure E. coli-derived LPS selectively promote the sensitization of bladder sensory neurons with TTX-R action potentials. The effect of ultrapure E. coli-derived LPS on the excitability of TTX-S bladder sensory neurons was only evident when the endotoxin was combined with SUP(−). These findings indicate that 1) LPS sensitizes bladder sensory neurons with TTX-R action potentials; 2) UPEC release virulence factors, in addition to LPS, with potential to sensitize TTX-R bladder afferents; 3) the sensitization of TTX-S bladder sensory neurons is triggered by LPS only when the endotoxin is combined with other unknown virulence factors; and 4) neurons with TTX-S and TTX-R action potentials may express distinct receptors for UPEC virulence factors. Further studies are needed to define the identity of the virulent factors and receptors involved in the sensitization of bladder afferent neurons.

Perspectives and Significance

In summary, we found that UPEC infection in mice prompts an inflammatory response characterized by lymphocytic and neutrophilic infiltration as well as edema in the bladder wall, increased voiding frequency, and pelvic allodynia. These alterations were associated with an increase in the excitability of sensory neurons with TTX-S and TTX-R action potentials. In addition, the results of our study show that multiple virulence factors produced by UTI89 act directly on bladder afferents to promote their sensitization. The results described earlier support the notion that the combined effects of UTI89 virulence factors and the host immune response cause the sensitization of bladder afferents, which contributes to the bladder hyperactivity and pelvic pain seen in mice with UTI.

GRANTS

This work was supported by the Urology Care Foundation Research Scholar Award Program (to N.M.), National Institute of Diabetes and Digestive and Kidney Diseases Grants R01DK119183 (to M.D.C. and G.A.), and by the Physiology and Model Systems Core of the Pittsburgh Center for Kidney Research Grant P30DK079307.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

N.M., M.G.D., G.A., and M.D.C. conceived and designed research; N.M., M.G.D., S.I.B., D.R.C., W.G.R., and M.D.C. performed experiments; N.M., M.G.D., S.I.B., D.R.C., W.G.R., and M.D.C. analyzed data; N.M., M.G.D., S.I.B., G.A., and M.D.C. interpreted results of experiments; N.M. and M.G.D. prepared figures; N.M. drafted manuscript; N.M., M.G.D., G.A., and M.D.C. edited and revised manuscript; N.M., M.G.D., S.I.B., D.R.C., W.G.R., G.A., and M.D.C. approved final version of manuscript.