The Expression of Inflammatory Mediators in Bladder Pain Syndrome

Background Bladder pain syndrome (BPS) pathology is poorly understood. Treatment strategies are empirical, with limited efficacy, and affected patients have diminished quality of life. Objective We examined the hypothesis that inflammatory mediators within the bladder contribute to BPS pathology. Design, setting, and participants Fifteen women with BPS and 15 women with stress urinary incontinence without bladder pain were recruited from Cork University Maternity Hospital from October 2011 to October 2012. During cystoscopy, 5-mm bladder biopsies were taken and processed for gene expression analysis. The effect of the identified genes was tested in laboratory animals. Outcome measures and statistical analysis We studied the expression of 96 inflammation-related genes in diseased and healthy bladders. We measured the correlation between genes and patient clinical profiles using the Pearson correlation coefficient. Results and limitations Analysis revealed 15 differentially expressed genes, confirmed in a replication study. FGF7 and CCL21 correlated significantly with clinical outcomes. Intravesical CCL21 instillation in rats caused increased bladder excitability and increased c-fos activity in spinal cord neurons. CCL21 atypical receptor knockout mice showed significantly more c-fos upon bladder stimulation with CCL21 than wild-type littermates. There was no change in FGF7-treated animals. The variability in patient samples presented as the main limitation. We used principal component analysis to identify similarities within the patient group. Conclusions Our study identified two biologically relevant inflammatory mediators in BPS and demonstrated an increase in nociceptive signalling with CCL21. Manipulation of this ligand is a potential new therapeutic strategy for BPS. Patient summary We compared gene expression in bladder biopsies of patients with bladder pain syndrome (BPS) and controls without pain and identified two genes that were increased in BPS patients and correlated with clinical profiles. We tested the effect of these genes in laboratory animals, confirming their role in bladder pain. Manipulating these genes in BPS is a potential treatment strategy.


Introduction
Bladder pain syndrome (BPS) is pain related to the urinary bladder accompanied by frequency, urgency, or nocturia, with the exclusion of any other diseases of the lower urinary tract [1,2]. Although the disease affects both sexes, women are more commonly affected than men by 5:1 [3]. Some patients display a mild form of the disease, with treatment generally orchestrated in the outpatient setting; in other cases, the disease is debilitating and requires prolonged hospitalisation and, often, repeated surgical intervention.
The aetiology of BPS is unknown. Multiple theories exist, including epithelial disruption and mast cell and vascular abnormalities [4,5]. The contributions of peripheral neuronal mechanisms remain unclear. We explore the hypothesis that many patients have a peripheral inflammatory disorder and that the expression of inflammatory mediators in the bladder wall activates and sensitises the bladder sensory afferents, driving BPS symptoms. Because there are currently no disease-modifying treatments for BPS, we postulated that the identification of novel inflammatory mediators associated with the disease might be manipulated to alter the disease's course.
We used quantitative gene expression analysis of 96 inflammatory mediators to measure gene expression levels in BPS and control samples. We then tested the activity of the identified mediators in an animal model of BPS using the enzymes chondroitinase and heparanase to digest the proteoglycan barrier [6,7].

Materials and methods
We performed a prospective observational study of 15

Statistical analysis
Data analysis was performed using the DCT method [8]. We applied the

Behavioural assessment
Pelvic pain response was assessed using calibrated Von Frey monofilaments. Von Frey withdrawal is typically performed on the hind paw of laboratory animals. We performed mechanical withdrawal assessment on the suprapubic region, which is reported as a valid method for the assessment of referred hyperalgesia and mechanical allodynia in animal models of bladder hypersensitivity [11]. Tactile sensitivity of the suprapubic region was assessed using the Chaplan method [12].

CCL21 atypical receptor knockouts
We received 17 C57/BL6/J mice from The Beatson Institute for Cancer Research, Glasgow: 9 ACKR4 À/À receptor knockout and 8 wild-type littermates [13]. The mice were anaesthetised with urethane, catheterised, and bladder permeabilised using 150 ml of 0.25 IU chondroitinase ABC and heparanase III. Then, 10 ml of 250 ng/ml CCL21 was instilled for 2 h, following which all animals were sacrificed, perfused, fixed, and spinal c-fos staining performed.

Results
BPS participants had urodynamic and cystoscopic evidence of disease. Control participants had no urologic disease or pain symptoms. All participants were free from urinary tract infections. See supplementary material for further phenotypic details. Gene expression analysis of 96 inflammationrelated genes revealed 15 differentially regulated genes (fold change >2; p < 0.05), highlighting a clear inflammatory process in the BPS pathology (Fig. 1A). The volcano plot shows the five most dysregulated genes (CCL21, IL12A, CXCL1, TNF, and FGF7 [shown in red]), which were all significantly dysregulated after correcting p values using the FDR method of Benjamini-Hochberg, with a 5% FDR.

3.1.
Gene correlation with clinical phenotypes Spearman rank correlation was used to compute correlation between the dysregulated genes and patient clinical phenotypes; it revealed that CCL21 and FGF7 were positively correlated with patient clinical phenotypes for ICS/PI symptom and problem indices, respectively (Fig. 2). In the case of CCL21, the correlation was statistically significant.

Principal component analysis
We applied principal component analysis to the entire gene expression data set (15 BPS and 15 controls) to visualise variability in the data. Figure 3A shows the first two dimensions of this analysis, from which we identify a cluster of seven BPS and seven controls that were tightly grouped. Further analysis of these principal component analysisselected cohorts by hierarchical clustering revealed, as expected, a striking reproducibility among patients for all differentially regulated genes (Fig. 3B). A DCT analysis of this refined group revealed 35 significantly dysregulated inflammatory genes, including CCL21 and FGF7.

Replication study
We performed a transcriptional analysis on an independent cohort of patients. In this cohort, hierarchical clustering revealed large biologic variability (Fig. 4A), but principal component analysis revealed a cluster of 5 pain and 6 controls, DCT analysis of which showed significant dysregulation of 15 inflammatory genes, including again CCL21 and FGF7 (Fig. 4B). This gene dysregulation pattern for both the original cohort and the principal component analysis-selected cluster correlated with that of the primary study (Fig. 4C).
[ ( F i g . _ 1 ) T D $ F I G ] To assess the effect of CCL21 and FGF7 in laboratory animals, we permeabilised the bladders of female Wistar rats using chondroitinase ABC and heparanase III to digest proteoglycans of the mucosal barrier, then applied the ligands intravesically. Cystometric analysis revealed an increase in bladder contraction number following enzymatic deglycosylation, which was sustained but not additive after CCL21 treatment (p = 0.0067). There was no change from baseline following FGF7 treatment (p = 0.320) (Fig. 5A and 5B). In addition, behavioural pain assessment showed a significant decrease in the mechanical withdrawal threshold following deglycosylation, with withdrawal thresholds [ ( F i g . _ 2 ) T D $ F I G ] further decreasing following CCL21 treatment. There was no difference in withdrawal threshold in the FGF7 or control rats (Fig. 5C).

CCL21 leads to upregulation of spinal c-fos
The effect of the ligands on spinal neurone activation was assessed by immunolocalisation of the immediate-early gene, c-fos, in anaesthetised rats following bladder permeabilisation. Analysis revealed a significant increase in the c-fos-positive cell number following urothelial permeabilisation compared to the saline-treated controls (p = 0.036). There was a further significant increase following CCL21 treatment (p = 0.042). There was no significant difference between the animals after permeabilisation alone compared to those after FGF7 treatment (p = 0.27) (Fig. 6B).
To assess the role of the CCL21 atypical chemokine receptor 4 (ACKR4) on bladder pain, we performed immunolocalisation of c-fos on ACKR4 knockout mice and wildtype littermates. Assessment revealed a significant increase in the number of c-fos-positive cells in the ACKR4 knockout mice compared to the wild-type controls (Fig. 6A and 6C).

Discussion
Our study shows upregulation of specific genes encoding inflammatory mediators in bladder biopsies of BPS patients.
We also showed a significant correlation between mRNA levels of the genes FGF7 and CCL21 with patient clinical profiles and disease severity scores. In addition, we highlighted the role of CCL21 as a pain mediator in an animal model of BPS.
In the past 20 yr, various animal models of cystitis have been created [14]. These models rely on inflammation of the bladder urothelium and so replicate only part of the pathology of the human disease. We have disrupted the bladder barrier by specific deglycosylation of urothelial proteoglycan molecules, thus replicating the disease, because 80% of BPS patient bladder biopsies show features of a defective barrier [15].
The key problem with human analysis is biologic variability, especially with complex diseases like BPS. In our study, principal component analysis revealed clear biologic variability in the participants analysed, but it also identified a cluster of patients who had distinct transcriptional profiles, suggesting that although a substantial number of patients share common inflammatory pathology, this does not apply to all patients. We therefore used this method to examine the biologic variability of the replication study, identifying a cluster with a strongly described inflammatory phenotype. The replication study was not designed to be a formal biologic replicate. The recruitment criteria for that study were slightly different and less  The second panel represents the seven bladder pain syndrome versus seven controls from the initial study, correlated with the five pain patients and six controls from the replication study (R 2 = 0.32; p = 0.01). Significance was calculated using Pearson R with a 2-tailed t test and a 95% confidence interval. FC = fold change.
exhaustive than the original one, but we took the opportunity to replicate the analysis in a similar cohort of patients with bladder pain.
The modest number of participants represents the main limitation of our study. This is a consequence of the difficulty of recruiting BPS patients resulting from the inherent diagnostic complexity of this syndrome. This sample size is, however, similar to previous studies in publication evaluating gene expression profiles of bladder biopsies from BPS patients as well as a quantitative analysis of gene expression, which, although small, were able to detect differences [16][17][18][19].
There is great heterogeneity of findings in the field of molecular characterisation of BPS. Results vary because of the complexity of the disease, the difficulty of diagnosis, the and phenotypic variability in BPS patients. We have confirmed an inflammatory component to the disease process. Several candidate genes that were found upregulated in our study have been reported as significant in BPS, including NOS, IL6, IL8, IL10, IL17A, NGF, and TNFA [16,20,21]. Of these, we identified two genes that correlated to patient phenotypes: FGF7 and CCL21. FGF7, also known as keratinocyte growth factor, expression is found to be upregulated in chronically injured tissue [22]. FGF7 expression is associated with healing and wound repair, supporting the integrity of the gastrointestinal tract mucosal barrier in chemotherapy patients who have oral mucositis [23,24]. It is possible that upregulation of FGF7 has a similar action in BPS patient bladders, strengthening urothelial barrier integrity. Our results corroborate these findings; FGF7 treatment following urothelial permeabilisation reduced bladder excitation on cystometry and behaviour analysis. In addition, FGF7 treatment did not lead to an increase in c-fos-positive cell numbers, suggesting that FGF7 may be involved in repair of the damaged urothelium.
CCL21, also known as secondary lymphoid tissue chemokine, is a potent inflammatory chemokine that regulates dendritic cell migration and has been implicated in various chronic inflammatory, fibrotic, and pain conditions, including rheumatoid arthritis, neuropathic pain, idiopathic pulmonary fibrosis, chronic hepatitis C, and primary biliary cirrhosis [25][26][27][28]. In the central nervous system, CCL21 expression is increased in damaged neurons, leading to increased microglial P2X4 expression [26]. It is possible that CCL21 may be involved in similar inflammatory cell activation in the periphery in response to injury. We have [ ( F i g . _ 5 ) T D $ F I G ] shown that CCL21 activity leads to an increase in bladder excitability and pain. The atypical CCL21 receptor ACKR4, found in the epithelial cells of the heart, thymus, skin, and urinary bladder, is a high-affinity receptor for CCL21 [29]. By internalising its ligands, this receptor diminishes the available circulating CCL21 levels, suppressing disease severity [13]. In our study, the increased number of spinal cells positive for c-fos in ACKR4 À/À suggests an increase in circulating CCL21 and a consequent increased disease severity. Thus, our study suggests that blockade of CCL21 in the bladder may have similar disease-modifying effects in BPS patients.

Conclusions
We have used medium-throughput quantitative polymerase chain reaction analysis to examine genes' differences in BPS and control biopsies and correlated gene expression to disease phenotype. In addition, we highlighted the role of CCL21 and FGF7 in BPS pain in rats and demonstrated an increase in pain processing in mouse knockouts for the CCL21 atypical receptor. Our results support the potential for the molecular evaluation and manipulation of these inflammatory mediators in the bladder as a possible new treatment strategy for BPS.  design and conduct of the study, the interpretation of the data, and approval of the manuscript.