A possible mechanism underlying mood disorders associated with LUTS: Chronic bladder outlet obstruction causes NLRP3-dependent inflammation in the hippocampus and depressive behavior in rats

Abstract

Aims:

Reports link urinary dysfunction and mood disorders, such as depression, but a causative mechanism has never been postulated. Contemporary discoveries demonstrate a local inflammatory response in peripheral organs can trigger inflammation in the brain, particularly the hippocampus, mediated through the NLRP3 inflammasome. Critically, central inflammation causes depressive behavior. Since bladder outlet obstruction (BOO) evokes a local inflammatory response in the bladder, we hypothesize it will induce NLRP3-dependent inflammation in the hippocampus and depressive behavior.

Methods:

There were four groups of rats: control, sham, BOO, or BOO + glyburide (an NLRP3 inhibitor). BOO was created by urethral ligation over a 1 mm catheter. Sham was tied loosely. Glyburide was provided by slow-release pellet (subcutaneous 50 mg, 21 day, replaced as needed). Rats were analyzed 12 weeks post-op for: hippocampal inflammation, microglial density, neurogenesis, and depression symptoms (open field and sucrose preference).

Results:

BOO elicited hippocampal inflammation, accompanied by an increase in activated microglia (22%) and a decrease in neurogenesis (35%), which was blocked by glyburide. In addition, BOO rats displayed anxiety (57% decrease in exploratory behavior in the open field assay) and anhedonia (21% decrease in sucrose preference), two symptoms of depression. Like inflammation, these symptoms were diminished by glyburide to levels not statistically significantly different from controls.

Conclusions:

BOO, a bladder-localized event, stimulates NLRP3-dependent inflammation in the rat hippocampus after 12 weeks and this inflammation causes depressive behavior. This is the first mechanistic explanation of the link between BOO and depression and provides evidence for a distinct bladder-brain axis.

1 |. INTRODUCTION

The long suspected association between lower urinary tract symptoms (LUTS) and mood disorders, such as depression and anxiety, was recently confirmed by population-based studies. The large epiLUTS study¹ found 30% of men and 38% of women with LUTS self-report sufficient criteria for clinical depression. The association is more striking for clinical anxiety, with 35.9% of men and 53.3% of women meeting criteria. Similarly, the National Health and Nutrition Examination Survey showed that patients with more LUTS had greater odds of having depression.² In addition, several papers have drawn correlations between the specific diseases that underlie LUTS and mood disorders, including interstitial cystitis/bladder pain syndrome,³ recurrent urinary tract infections,⁴ overactive bladder,⁵ and incontinence.⁶

This study focuses on bladder outlet obstruction (BOO), most commonly found secondary to benign prostatic hyperplasia (BPH). BOO has been associated with mood disorders⁷ although a causative mechanism has never been proposed. Intuitively, BOO could exacerbate underlying mood disorders given its negative impact on quality of life, or patients may be more likely to report minor urinary disorders because depressed patients are more likely to somatize and engage in catastrophic thinking. However, work in other fields have shown an immunological underpinning for some psychiatric disease associated with peripheral dysfunction. In particular, localized inflammatory insults in peripheral organs may induce inflammation in the central nervous system (CNS), which triggers clinical depression.⁸ Since BOO triggers a chronic inflammatory response in the bladder,^9–11 we propose the exploratory hypothesis that it will also elicit inflammation in the CNS that will precipitate mood disorders. Interestingly, Johnson et al.¹² first proposed inflammation as the central commonality between LUTS and mood disorders in 2010. Moreover, it was recently shown that cyclophosphamide-induced hemorrhagic cystitis leads to inflammation in the hippocampus of rats and induces depressive behavior.¹³ Although that model has substantial limitations, it suggests similar results may be found with a more bladder-centric and clinically important disorder such as BOO.

Recently there has been a fundamental shift in understanding how inflammation is triggered. In sterile inflammatory conditions such as BOO, there is a release of intracellular molecules, referred to as damage associated molecular patterns (DAMPs). DAMPs promote the formation of a multimeric structure known as the inflammasome in many cells, including urothelia^9–11 and microglia.¹⁴ The most common inflammasome involved in sterile inflammation is the NACHT, LRR, and PYD domains-containing protein 3 (NLRP3) inflammasome. The result of any inflammasome activation is the activation of the protease caspase-1. Caspase-1 cleaves pro-IL-1β and pro-IL-18 into their active forms, which then precipitate a wider inflammatory reaction. Recent work has demonstrated clear roles for NLRP3 in bladder dysfunction and deterioration during BOO,^9–11 and others have implicated it in central inflammation and depression.¹⁵ Therefore, we hypothesize that this structure plays a central role in BOO-induced mood disorders.

2 |. MATERIALS AND METHODS

2.1 |. Animals

Animal protocols were approved by the Institutional Animal Care and Use Committee at Duke University and performed within the guidelines set forth in the Guide for the Care and Use of Laboratory Animals published by the National Institutes of Health (USA). Sprague Dawley Rats (female, ~ 50 days, ~ 200 g) were purchased from Envigo (Indianapolis, IA). Although in humans BOO predominates in males, the standard for rodent BOO studies is the female rat due to the tortuosity of the male urethra which often leads to physical damage and inflammation during catheterization. Other concerns are complications from ducts associated with the prostatic gland and seminal vesicles. Thus, male animals are contraindicated in rodent BOO studies, particularly when examining inflammation. For most studies there were four groups; (a) control, (b) Sham, (c) BOO, or (d) BOO + glyburide (Gly). An additional group (BOO + fluoxetine) was used for behavior assays. For Sham and all BOO groups, animals were anesthetized (ketamine hydrochloride [90 mg/kg], xylazine [10 mg/kg]; Intraperitoneal [IP]) and a 1 mm OD catheter (P50 tubing) inserted transurethrally. A 5-0 silk suture was passed around the urethra and tied securely for BOO and loosely for Sham. The catheter was removed and the abdominal wall closed. For BOO + Gly, a single 50 mg, 21-day slow-release glyburide pellet (Innovative Research of America, Sarasota, FL) was placed in a subcutaneous pocket on the side of the neck. A new pellet was placed (contralateral side) after 21 days. Sides were alternated thereafter. No signs of urinary tract infection were seen in any animal.

For behavior assays, an additional group of BOO rats were provided with fluoxetine (Sigma, St Louis, MO) in drinking water (0.50 mg/mL) for the last 4 weeks. The dose was adjusted twice weekly to insure ~ 20 mg/kg/day.

2.2 |. Evans blue assay

Rats were injected (intravenous 3 mL/kg) with 2% Evans blue dye in sterile saline by an investigator blinded to the groups.¹⁶ After 1 hour, rats were anesthetized (ketamine hydrochloride [90 mg/kg], xylazine [10 mg/kg]; IP) and then transcardially perfused with cold phosphate-buffered solution to remove intravascular dye before being killed by thoracotomy. The brain was isolated and the hippocampus dissected, weighed, and placed into formamide (0.25 mL) overnight (56°C). Absorbance (620 nm) was measured and compared with a standard curve to calculate pg Evans blue/μg tissue.

2.3 |. Immunocytochemistry and quantitation of microglia and neurogenesis

Brains were fixed (10% neutral buffered formalin; 48 hours, room temperature), grossly cut (coronally) to the hippocampus and embedded in paraffin (plane of the hippocampus on the block face). Coronal sections (10 μm) were stained with anti-IbA1/AIF1 (1:500) (NBP2-19019; Novus Biologicals, Centennial, CO) or anti-Ki67 (1:500) (ab15580; Abcam, Cambridge, MA) using standard methodology and citrate antigen retrieval. IbA1 was visualized via horseradish peroxidase development (Vectastain; Vector, Burlingame, CA), biotinylated secondary antibody provided. Ki-67 was visualized using a goat antirabbit secondary antibody conjugated to Alexa Fluor 488 (111-545-144; Jackson Labs, West Grove, PA). Slides were coverslipped using Vectashield Antifade mounting medium with 4′-6-diamidino-2-phenylindole (DAPI; Vector, Burlingame, CA).

A Zeiss Axio Imager 2 microscope (Zeiss, Oberkochen, Germany) running Zen software (Zeiss) using the tiling and stitching feature was used to scan one entire hemisphere. Images were imported into NIS-Elements software (Nikon Co, Tokyo, Japan), calibrated and quantitated by a researcher blinded to the groups. For microglia, we demarcated 600 000 to 700 000 μm² of the fascia dentata, counted the number of black/brown spots with two or greater associated tendrils and calculated microglial density. For Ki-67⁺ cells we demarcated the entire dentate gyrus, CA1, CA2, and CA3 regions, quantitated nuclei (DAPI⁺) that were Ki-67⁺ (green florescence) and calculated the density.

2.4 |. Behavioral assays

2.4.1 |. Open field

Rats were placed in an acrylic box (45 cm × 45 cm × 30 cm, white bottom divided into 16 equal squares, clear sides) and video recorded (10 minutes). The video was scored by blinded individuals and the amount of time in which two or more of the rats paws are inside the central four squares (time in middle) recorded.

2.4.2 |. Sucrose preference

Rats were simultaneously provided with bottles containing 2% sucrose or drinking water. The location of the bottles are alternated after 24 hours. After 48 hours the remaining liquid was measured and sucrose preference calculated as a percentage of the total liquid intake.

2.5 |. Statistical analysis

Using Stata software (College Station, TX) analysis of variance examined differences across the four groups, followed by a Dunn’s post hoc analysis to enable pairwise comparisons. To examine the effect of BOO on bladder weight, inflammation, neurogenesis, and behavioral dysfunction, we compared control (and sham) group to the untreated obstructed group. To examine whether treatment returned outcomes to baseline with bladder weight, inflammation, neurogenesis, and behavioral dysfunction, we compared control (and sham) group with the treated obstructed group. Lastly, to examine whether treatment ameliorated changes in bladder weight, inflammation, neurogenesis, and behavioral dysfunction, we compared untreated and treated obstructed groups. Comparisons were specified before beginning work. Results were considered statistically significant if P < .05 and exact P values can be found in Table S1, although because this is an exploratory study the P-values should be interpreted as descriptive only. Due to the novelty of these studies, size effects were difficult to estimate making a power analysis ineffective. However, the experimental methods were familiar to the investigators and variability known from published studies. Therefore, group sizes were estimated from previous experience.

3 |. RESULTS

3.1 |. BOO increased bladder weight and this was partially blocked by inhibiting NLRP3

There was a small increase in bladder weight in sham-operated rats compared with control (Figure 1) that is attributed to the general pelvic inflammation expected from sham surgery. However, after 12 weeks of BOO, bladder weights increased well over 10-fold. Glyburide suppresses this increase, although the values were considerably larger than controls.

FIGURE 1.

Bladder weights are greatly increased 12 weeks after BOO and this is partially inhibited by the NLRP3 inhibitor glyburide. Bladders from the various groups indicated were weighed when they were removed for the various endpoints in this experiment. Results are the mean ± 95% confidence intervals; *P < 0.05, ***P < .005 by ANOVA and Dunn’s test. (n = 45, 42, 38, and 35). ANOVA, analysis of variance; BOO, bladder outlet obstruction; Gly, glyburide-treated; NLRP3, NACHT, LRR, and PYD domains-containing protein 3; Veh, vehicle-treated

3.2 |. BOO triggers NLRP3-dependent inflammation in the hippocampus

BOO triggered a statistically significant increase in the leakage of Evans blue into the hippocampus (Figure 2), indicating inflammation and disruption of the blood brain barrier (BBB). Leakage was maintained at control levels by glyburide treatment. There was no difference between sham and control. To confirm this we quantitated the density of microglia in the fascia dentate. Figure 3 shows a statistically significant increase in the density of activated microglia in the hippocampus (no difference in sham) which was prevented by glyburide.

FIGURE 2.

After 12 weeks of BOO, inflammation is present in the hippocampus of rats. This inflammation is blocked by concomitant treatment with glyburide. Following the treatments indicated, inflammation was assessed by the Evans blue assay as described in the Materials and Methods section. Results are the mean ± 95% confidence intervals; *P < .05, **P < .01, ***P < .005 by ANOVA and Dunn’s test. (n = 8, 7, 12, and 8). ANOVA, analysis of variance; BOO, bladder outlet obstruction; Gly, glyburide-treated; Veh, vehicle-treated

FIGURE 3.

After 12 weeks of BOO the number of activated microglia in the hippocampus is increased and this increase was blocked by glyburide treatment. Activated microglia in 10 μm sections of brain were stained for IbA1/AIF1 and then visualized and quantitated as described in the Materials and Method section. The results are presented as the density of activated microglia per μm². Results are the mean ± 95% confidence intervals; *P < .05, **P < .01, ***P < .005 by ANOVA and Dunn’s test. (n = 7, 6, 8, and 7). ANOVA, analysis of variance; BOO, bladder outlet obstruction; Gly, glyburide-treated; Veh, vehicle-treated

3.3 |. BOO causes a decrease in the number of proliferating cells in the hippocampus

As shown in Figure 4, there was a statistically significant decrease in the concentration of proliferating cells (Ki-67⁺) in the hippocampus following BOO which was blocked with glyburide. There was no difference between sham and control.

FIGURE 4.

After 12 weeks of BOO, neurogenesis is statistically decreased in the hippocampus and this increase is blocked by glyburide treatment. Cells in 10 μm sections of brain were stained for Ki-67 and then visualized and quantitated as described in the Materials and Method section. The results are presented as the density of Ki-67⁺ cells per μm². Results are the mean ± 95% confidence intervals; *P < .05, ***P < .005 by ANOVA and Dunn’s test (n = 6, 6, 6, and 6). ANOVA, analysis of variance; BOO, bladder outlet obstruction; Gly, glyburide-treated; Veh, vehicle-treated

3.4 |. BOO rats show NLRP3-dependent signs of depression; anxiety and anhedonia

Two behavior assays that assess different signs of depression were performed. The open field test measures anxiety as a function of the rat’s propensity to explore the middle region of a square open field. Normal rats, being somewhat curious, will naturally explore this region while anxious rats refrain. As shown in Figure 5A, BOO rats spent less than half the time of the control and sham rats exploring the middle of the field. Interestingly, glyburide maintained this exploratory behavior demonstrating this anxiety is NLRP3-dependent. In addition, a set of BOO rats were given the antidepressant fluoxetine. Fluoxetine helps to differentiate true depression-related behavior from sick behavior, which is not affected by this antidepressant. As shown in Figure 5A, fluoxetine restored exploratory behavior back to levels not statistically different from control.

FIGURE 5.

After 12 weeks of BOO, rats show signs of depression. These behavior differences were not present when rats were given glyburide or fluoxetine (Flu), an antidepressant. A, The open field assay (a measure of anxiety). The assay was performed as described in the Materials and Methods section and scored by a blinded investigator. The results are presented as the time in which at least two paws were present in the middle section of the open field during the 10 minute test session. Veh = vehicle-treated, Gly = glyburide-treated. Results are the mean ± 95% confidence intervals; **P < .01, ***P < .005 by ANOVA and Dunn’s test (n = 26, 23, 15, 13, and 6). B, The sucrose preference assay (a measure of anhedonia). The assay was performed as described in the Materials and Methods section. The results are presented as the amount of sucrose laden water consumed as a percentage of the total volume imbibed. Results are the mean ± 95% confidence intervals; *P < .05, **P < .01, ***P < .005 by ANOVA and Dunn’s test (n = 24, 24, 14, 14, and 6). ANOVA, analysis of variance; BOO, bladder outlet obstruction; Gly, glyburide-treated; Veh, vehicle-treated

We next performed the sucrose preference assay to assess anhedonia or the inability to feel pleasure. As shown in Figure 5B, there was a statistically significant decrease in the preference for sugar laden water in BOO rats which was maintained at control levels by glyburide. Fluoxetine also restored sucrose preference back to levels not statistically different from control. There was no difference between sham and control.

4 |. DISCUSSION

Initial studies of the innate immune system during BOO logically focused on the local response in the bladder^9–11 and found activation of NLRP3 in urothelia which initiated inflammation, fibrosis, and denervation. We have expanded that work to examine BOO-induced inflammation in the brain and changes in behavior. Initially, hippocampal inflammation was found after 6 weeks but we were unable to detect differences in sucrose preference at that time point (Figure S1), necessitating extending the time course to 12 weeks. This suggests a sequential series of events where neuroinflammation precedes behavioral changes.

At 12 weeks of BOO, bladder weight increase was attenuated, but not completely blocked, in the drug-treated animals. The reason for incomplete attenuation was not examined, but we speculate that weight gain is composed of inflammation/edema and muscle hypertrophy. Logically glyburide is expected to blunt inflammation/edema but is unlikely to do so for muscle hypertrophy caused by overuse.

Critical to our exploratory hypothesis was the detection of hippocampal inflammation using the Evans blue assay which measures extravasation potential of capillaries. When considering blood vessels traversing the brain, this equates to a measurement of the integrity of the BBB.¹⁶ Thus, after 12 weeks BOO has precipitated a notable degradation of the BBB. The presence of neuroinflammation was confirmed by the increase in activated microglia, the main drivers of neuroinflammation and the brain cells possessing NLRP3 (along with astrocytes). Importantly, both indicators of neuroinflammation were blocked by glyburide, indicating their dependence on NLRP3. It is unclear where, exactly, the glyburide is functioning; at the level of the bladder, the brain, or both. Undoubtedly, glyburide acts on the bladder urothelia, as there are several publications showing just that,^9–11 but glyburide activity in the brain is not so clear. In the serum, glyburide is mostly albumin-bound and does not cross the BBB.¹⁷ However, in the event of a breakdown in the barrier, even transiently, glyburide can cross into the brain,¹⁸ where it could function to block NLRP3, minimize neuroinflammation and preserve psychiatric health. Thus, we propose that the initial, major and perhaps only effect of glyburide is in the bladder. However, if its protective effect is overwhelmed and there is breakdown of the BBB, it may enter the brain and directly prevent a neuroinflammatory response.

In the hippocampus, BOO caused a statistically significant decrease in proliferating cells (Ki-67⁺). Decreases in neurogenesis, and more generally plasticity, in the hippocampus have been directly linked with neuroinflammation and depression.¹⁹ Therefore, we feel the difference in Ki-67⁺ cells likely reflect differences in neurogenesis, although we cannot rule out a contribution by proliferation of cells other than neurons. Such changes may lead to permanent, or long lasting, differences in cognitive function or mood that persists after the initiating stimulus is gone (in the case of BOO, after a TURP is performed). This decrease was blocked by glyburide, demonstrating the central role of NLRP3 and suggesting that NLRP3 inhibitors may further help prevent BOO-induced neurodeterioration by suppressing negative changes in neuroplasticity.

Our most important observation was likely the behavioral differences in BOO. Untreated, obstructed animals showed statistically significant signs of anxiety and anhedonia, two core symptoms of depression, and these behavioral differences were blocked by glyburide demonstrating the centrality of NLRP3. Importantly, behavioral differences could also be prevented by the antidepressant fluoxetine demonstrating they are not due simply to pain or “sick” behavior as fluoxetine would not help in that situation.

One intriguing question is the nature of the peripheral-to-central signal and currently three possible pathways are in vogue that are not mutually exclusive.¹⁵ The humoral pathway could involve cytokines produced in the bladder traveling systemically to the BBB where they trigger breakdown and enter into the CNS, initiating microglial activation and neuroinflammation. Given the breakdown of the BBB in this study we feel this pathway likely contributes to BOO-induced neuroinflammation. Another possibility is the neural pathway where sensory input along afferent nerves carries a retrograde signal for inflammation back to the hippocampus where it is translates into cytokine production. Finally, a cellular pathway may involve transmigration of immune cells activated in the bladder directly across the BBB. All three of these potential pathways warrant future testing.

The relationship between mood disorders and BOO (or more correctly LUTS) appears to be bidirectional. For example, two large prospective studies found that men with BPH had a two to three-fold greater risk of developing clinical depression,^20,21 whereas a 5-year prospective study showed that men with baseline depression were more likely to see a progression of LUTS.²² Working out the nature of each of the directional signals promises great fodder for future explorations, as this is hypothesis-generating work.

Our work shows that BOO causes central inflammation and mood disorders. It implies, therefore, that relieving the obstruction will relieve the mood disorder. Some clinical studies can attest to this relationship. For example, Cho et al²³ followed 120 BPH patients over 12 weeks of medical treatment and found the severity of depression decreased with the severity of LUTS. Likewise, O’Sullivan et al²⁴ found a reduction of depression associated with improvement in LUTS following a TURP. Probably the most telling was Queck et al²⁵ who compared medical therapy and TURP. Medical therapy does not completely normalize urodynamic function leaving low levels of bladder inflammation. TURP, however, removes the obstruction, greatly improves urodynamic parameters and drastically reduces bladder inflammation. Both cohorts showed improvement of depression although the TURP was substantially more effective. Besides supporting our exploratory hypothesis, this study also suggests that TURP may be the better choice for men with LUTS (secondary to BPH) suffering from depression or anxiety. Such correlations have prompted a call from others to screen BPH patients for depression⁷ and the current data support that call.

The primary limitation of this study is the use and dose of glyburide. While Lamkanfi et al²⁶ clearly showed this drug directly inhibits NLRP3, it is also a widely used sulfonylurea drug for the treatment of type 2 diabetes where it acts at ATP-sensitive K⁺ (KATP) channels and Kir6.2. Therefore, because of these pleiotropic effects we cannot state with 100% certainty the current outcomes are solely attributable to inhibition of NLRP3. However, given the correlation of multiple inflammatory parameters across multiple models, we feel it is very likely that inhibition of NLRP3 is the major, if not only, effect of this drug driving our results. In this study, and several others that use glyburide as an NLRP3 inhibitor,^9,10 we have used a dose common in the literature although considerably less than the amount (500 mg/kg/d) used in Lamkanfi’s original studies.²⁶ However, it does exceed the recommended doses for humans. The use of suprapharmacological doses of medications in rodent studies is common practice to ensure a robust response. This is often necessary due to (a) differences in metabolism between human and rodents and (b) the compressed time scales in rodent studies (eg, our studies lasted 12 weeks whereas humans may suffer with BOO for decades). Finally, an additional shortcoming is the measurement of neuroinflammation by only two methods. The work would have benefited from more extensive proof of neuroinflammation such as measurements of microglia function.

5 |. CONCLUSIONS

Using a rat surgical model of BOO, a clear bladder-localized model, we have shown an NLRP3-dependent increase in inflammation and a decrease in neurogenesis in the hippocampus along with the appearance of depressive behavior. Our data provided the first mechanistic description of an immunological bladder to brain axis and provide a framework for understanding the relationship between LUTS and mood disorders such as depression.