INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

INFLAMMATION UNDERLIES ALL major bladder pathologies. Indeed, areas of bladder inflammation are found during chronic implantation of catheters (22, 88), as an integrative part of bladder responses to intravesical bacillus Calmette-Guerin (BCG) (10, 12) and capsaicin therapy (34), during radiation cystitis, and in cyclophosphamide-induced cystitis (86, 87), interstitial cystitis (28, 68), and bladder cancer (3-13).

Chronic inflammation in the bladder mucosa is typical in spinal cord-injured patients with a chronic indwelling bladder catheter (22, 88). Inflammation is also observed secondarily to intravesical treatments. Indeed, inflammation is part of the response to intravesical BCG instillation. However, some authors have reported that BCG inflammation obviously differs from nonspecific inflammation by its quality and subclinical duration (12), whereas others have qualified BCG-induced bladder inflammation as granulomatous (10). In addition, inflammation does not explain BCG immunotherapy for superficial bladder carcinoma; this complex response requires an immune response as well as a tumor response for its effects (43). Inflammation occurs also secondarily to intravesical capsaicin therapy for detrusor hyperreflexia (34), and part of this response is modulated by ethanol, used as a vehicle for capsaicin (23).

Inflammation seems to underlie bladder cancer. In this context, epidemiological studies indicate that the development of squamous cell carcinoma of the urinary bladder is closely associated with chronic inflammation of the urinary tract. Chronic inflammation was present in all cases; radiation-induced pseudocarcinomatous proliferations of the urinary bladder (3) and chronic subepithelial inflammation are also commonly present in intraepithelial neoplasms (13).

It seems that the very same cells and molecules that mediate bladder inflammation and the response to pathogens and trauma are integral parts of bladder cancer processes. Indeed, adhesion molecules known to have a role in inflammation are now being postulated to play a central role in the development and progression of bladder cancer (79). Another molecule that participates in bladder inflammation and cancer is IL-6. IL-6 is the primary cytokine elevated in the urine of interstitial cystitis patients (26). IL-6 functions as an autocrine growth factor for bladder carcinoma cells but not for normal urothelial cells, and it may be a factor accounting for the marked enhancement of inflammation-associated bladder carcinogenesis and tumor growth (60). A common pathway leading to a bladder response to inflammatory stimuli is NF-B (89). A central role for NF-B in bladder inflammation was proposed on the basis of the finding that this transcription factor stimulates the expression of enzymes whose products contribute to the pathogenesis of the inflammatory process, including the inducible form of nitric oxide synthase and inducible cyclooxygenase-2 (61). Abnormalities in the regulation of the NF-B pathway are frequently seen in a variety of human malignancies including leukemias, lymphomas, and solid tumors (66). These abnormalities result in constitutively high levels of NF-B in the nucleus of a variety of tumors including breast, ovarian, prostate, and colon cancers.

Neuroimmune mechanisms have been suggested to play an important role in the development of inflammatory reactions in the urinary bladder (40). One potential pathway that may mediate such neurogenic inflammation is the interaction between mast cells and the neurotransmitter substance P. The responses to substance P and other tachykinins such as neurokinin A, neurokinin B, and hemokinin B are mediated by three different G protein-coupled neurokinin receptors via changes in cytosolic calcium concentration (90). Neurokinin-1 (NK₁) receptors (NK1Rs) are the predominant receptor subtype involved in inflammation and are expressed by endothelial cells, submucosal glands, and circulating leukocytes (85). In the urinary bladder, plasma extravasation induced by substance P seems to be modulated by NK1Rs, because it was inhibited by a specific receptor antagonist (33). More recently, we presented evidence of increased NK1R density during bladder inflammation (39). In addition, we have shown that in the urinary bladder, substance P receptors (NK1Rs) are downstream of NF-B activation and that NK1R knockout mice failed to mount a bladder inflammatory response (89). Others have shown an increase in NK1R numbers, and distribution may underlie persistent pain such as that observed during chronic inflammation (1, 9, 55). In contrast, NK2Rs mediate detrusor muscle contraction in humans (63, 95), dogs (56, 71), hamsters (83), and mice (57), whereas NK3Rs have been found in the ganglia (35) and vascular smooth muscle (45), but they have yet to be identified in the urinary system (4).

Although there is considerable evidence that NK1Rs seem to play a critical role in the transmission of noxious stimuli (9, 47), NK1R antagonists were described to be of low efficacy in the clinical treatment of pain (41). The apparently low efficacy of NK1R antagonists in the treatment of pain in humans has been attributed to inadequate clinical trials (84). Alternatively, several dual antagonists of NK1Rs and NK2Rs have been developed (67, 77). With the exception of ZD-6021, which has been shown to dose dependently attenuate plasma extravasation in guinea pigs (6), little if any investigation of the therapeutic potential of dual antagonists has been done. Alternatively, several authors have used a combination of NK₁, NK₂, and NK₃ antagonists to reduce visceral hyperalgesia (44).

Sensory nerves within the bladder contain high concentrations of substance P, which is a potent inducer of mast cell degranulation (93). In addition, mast cells are found in increased numbers in bladder biopsies obtained from patients with interstitial cystitis (42, 62), suggesting that mast cells potentially influence this condition. The release of mast cell mediators such as histamine, proteases, arachidonic acid metabolites, and TNF- and other cytokines can potentially initiate and/or amplify inflammatory responses in the bladder (32) by stimulating sensory nerves to release substance P (11). Indeed, the inflammatory responses to mast cell tryptase are mediated through substance P release (21). This indirect evidence implies a link between substance P and mast cells in the pathogenesis of bladder inflammation, but a definitive demonstration of such a pathophysiological relation is lacking.

One area of controversy that continues to limit our understanding of substance P-mast cell interactions is the mechanism through which substance P induces mast cell degranulation. In other systems, tachykinins such as substance P interact with a family of receptors termed neurokinin receptors. The high-affinity receptor for substance P is NK1R. A straightforward speculation has been that substance P interacts with NK1Rs on mast cells. However, there is conflicting evidence in the literature as to whether mast cells express NK1Rs or whether substance P induces mast cell activation through an alternative, receptor-independent mechanism.

For example, evidence supporting mast cell expression of NK1Rs includes radiolabel binding assays suggestive of a single high-affinity binding site for substance P on freshly isolated rat peritoneal mast cells (PMCs) (59), the ability to inhibit substance P-induced degranulation of rat PMCs (58, 59) and a murine mast cell line (46) with specific NK1R antagonists, and the identification of NK1R mRNA by RT-PCR in freshly isolated PMCs (59). However, many of these findings using rat PMCs appear to depend on the strain of rat studied (58, 59). Similarly, the presence of NK1R mRNA was found in cultured RBL-2H3 cells, a rat mast cell line (18). By contrast, substantial evidence supports the concept that substance P induces mast cell degranulation by direct activation of certain G proteins (2, 53). Experiments using labeled substance P and patch-clamp techniques indicate that substance P translocates into mast cells and activates degranulation by a pertussis toxin- and GDPS-sensitive mechanism (53). Our group and others have been investigating the role of substance P and NK1Rs in experimental cystitis. Our laboratory previously reported a mouse model of antigen-induced bladder inflammation characterized by histological evidence of mast cell degranulation, edema, and infiltration of the bladder by neutrophils (73). Using genetically engineered mice deficient in the expression of NK1Rs, we were able to directly evaluate the importance of NK₁-substance P interactions in the development of cystitis. We reported that NK1R^/ mice exhibited no inflammation in response to antigen challenge, indicating that NK1Rs play a critical role in antigen-induced inflammation and suggesting that substance P is involved in the pathogenesis of this response (73).

In a separate study, we investigated the role of mast cells in this antigen-induced model of bladder inflammation (72). For this purpose, we examined the responses in mast cell-deficient Kit^W/Kit^W-v, congenic normal (+/+), and Kit^W/Kit^W-v mice that were reconstituted with +/+ bone marrow cells (BMR-Kit^W/Kit^W-v). This procedure repopulates mast cells and other bone marrow-derived elements in Kit^W/Kit^W-v mice. Antigen-induced bladder inflammation occurred in +/+ and BMR-Kit^W/Kit^W-v mice but not in mast cell-deficient Kit^W/Kit^W-v mice. Gene expression was determined using mouse cDNA expression arrays and indicated that gene expression associated with the inflammatory response was dependent on the presence of mast cells, i.e., certain genes were upregulated in bladders isolated from antigen-challenged +/+ and BMR-Kit^W/Kit^W-v mice but were not altered in Kit^W/Kit^W-v mice. Taken together, these studies indicate an important role for mast cells and NK1Rs in this antigen-induced model of cystitis (72-73) but do not address the nature of the interaction of substance P and mast cells.

Thus to more precisely define the role of mast cells and NK1Rs in experimental bladder inflammation, we examined the bladder inflammatory reactions in +/+, mast cell-deficient Kit^W/Kit^W-v, and BMR-Kit^W/Kit^W-v mice. The controversy concerning the expression of NK1Rs on mast cells aside, this model would permit us to examine experimental cystitis in the absence of mast cells in a host that expresses NK1Rs and then compare the reactions in the same host that now has mast cells that do not, and cannot, express NK1Rs.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Reagents. Dinitrophenyl (DNP) was conjugated to ovalbumin (OVA) or human serum albumin (HSA; Sigma, St. Louis, MO) as previously described (52). Alum was purchased from Intergen (Purchase, NY) and used as an adjuvant. Substance P was purchased from Sigma. All reagents were prepared in pyrogen-free saline immediately before use.

Animals. All animal experimentation described here was performed in conformity with the "Guiding Principles for Research Involving Animals and Human Beings" (Oklahoma University Health Sciences Center Animal Care and Use Committee, protocol no. 00-109).

/

Bone marrow reconstitution of mast cell-deficient Kit^W/Kit^W-v mice. Mice with double mutations at the W loci have a variety of phenotypic abnormalites, including a profound deficiency in the numbers of tissue mast cells, macrocytic anemia, age-dependent changes in intraepithelial lymphocyte populations in the gastrointestinal tract, and other nonmyeloid abnormalites (92). Bone marrow transplantation repairs both the mast cell deficiency and the anemia in Kit^W/Kit^W-v mice (92). We utilized this approach to reconstitute Kit^W/Kit^W-v mice with bone marrow from NK1R^/ mice, and thus the mast cells that develop in the bladder would not be capable of expressing the NK1R.

/

Antigen sensitization protocol. All mice in this study were sensitized intraperitoneally with 1 µg DNP₄-HSA in 1 mg alum on days 0, 7, 14, and 21. In normal mice, this protocol induces sustained levels of IgE antibodies up to 56 days postsensitization (38), and mast cell-deficient Kit^W/Kit^W-v mice have been shown to develop normal IgE responses in similar active immunization protocols (30). One week after the last sensitization, an inflammatory reaction was elicited in the bladder (see Induction of experimental cystitis).

Induction of experimental cystitis. Cystitis was induced as described previously (69-70, 72-73). Briefly, sensitized Kit^W/Kit^W-v, +/+, and BMR-NK1R^/ mice were anesthetized (40 mg/kg ketamine and 2.5 mg/kg xylazine ip), transurethrally catheterized (Angiocath, Becton Dickinson, Sandy, UT), and drained of any urine present by the application of slight digital pressure to the lower abdomen. The urinary bladders were instilled with 150 µl of either pyrogen-free saline, substance P (10 µM), or DNP₄-OVA (antigen; 1 µg/ml), infused at a slow rate to avoid trauma and vesicoureteral reflux. To ensure consistent contact of substances with the bladder, infusion was repeated twice within a 30-min interval, and a 1-ml Tb syringe maintained on the catheter retained the intravesical solution for at least 1 h. The catheter was then removed, and the mice were allowed to void normally. Twenty-four hours after challenge, the mice were killed with pentobarbital sodium (20 mg/kg ip) and bladders were removed.

/

Histological assessment of bladder inflammation. A cross section of bladder wall was fixed in formalin, dehydrated in graded alcohol and xylene, embedded in paraffin, and cut serially into four 5-µm sections (8 µm apart) stained with either hematoxylin and eosin or Giemsa. The urinary bladders were evaluated for the presence of edema, extent of inflammatory cell accumulation, and number of mast cells present. A semiquantitative score using defined criteria of inflammation severity was used to evaluate cystitis (69, 70, 72, 73). Slides were scanned using a Nikon digital camera (DXM1200) mounted on a Nikon microscope (Eclipse E600). Image analysis was performed using the MetaMorph Imaging System (Universal Imaging, West Chester, PA). The severity of lesions in the urinary bladder was graded as follows: 1+, mild (minimal infiltrate of neutrophils in the lamina propria and little or no interstitial edema); 2+, moderate (infiltration of neutrophils into the lamina propria and moderate interstitial edema); and 3+, severe (diffuse infiltration of a large number of neutrophils in the lamina propria and severe interstitial edema) (73). Identification of mast cells and quantification of their degree of degranulation was performed in Giemsa-stained sections. The extent of degranulation is presented as the percentage of mast cells per cross section that exhibited morphological evidence of degranulation.

Sample preparation for cDNA expression arrays. cDNA expression arrays were performed as previously described (69, 72). Briefly, three bladders from each group were pooled and homogenized in Ultraspec RNA solution (Biotecx Laboratories, Houston, TX) for isolation and purification of total RNA (we had previously determined that the pooling of 3 bladders is necessary to provide enough RNA for gene array analysis). RNA was DNAse treated according to the manufacturer's instructions (Clontech Laboratories, Palo Alto, CA), and the quality of 10 µg RNA was evaluated by denaturing formaldehyde-agarose gel electrophoresis. This procedure was replicated using an additional three bladders per each experimental group. Therefore, two pools of RNA were generated per experimental group, and two separate hybridizations were performed per group.

Background subtraction. Background represented by empty rows and columns was averaged. This background intensity was deducted from the intensity of each spot. In addition, the intensity of negative controls M13 mp18 (+) strand DNA, -DNA, and pUC18 DNA were taken as unspecific hybridization and averaged. This unspecific hybridization was deducted from the intensity of each spot. Only signals with 3 SD above background were used. This was done to avoid ratios either mathematically impossible or astronomically high. The values for all genes with background-subtracted intensities >1 were then exported to an Excel spreadsheet to generate a sorted data matrix for all arrays within each experiment. At this point, data were filtered to eliminate genes detected fewer than two times in each condition and replicate array values normalized by the sum-of-intensities method. Total intensities from the two conditions were then scaled, and mean intensities were calculated for scatterplot visualization of the fold-changes. The signal intensities were normalized to the mean signal intensity of all the spots on the array. Duplicate spots of each cDNA were averaged. To compare two arrays (2 conditions), expression was calculated as the percentage of -actin, as previously described (72).

Experimental criteria. To determine which genes were dependent on the presence of tissue mast cells but independent of NK1Rs, we set the following arbitrary criteria: 1) the gene should be upregulated at least threefold in response to antigen or substance P challenge in +/+ tissues; 2) the same gene should not be upregulated in the absence of mast cells (Kit^W/Kit^W-v mice); and 3) upregulation should be demonstrated in BMR-NK1R^/ mice.

Cluster analysis using self-organizing maps. Genes fulfilling the above criteria were identified on the basis of a Euclidean distance metric, after the elimination of genes that failed to vary in expression level within an experiment by a factor of 3 SE and an absolute value of 100% of -actin and normalization within experiments to a mean of 0 and an SD of 1.

Data preprocessing. Using GeneCluster, SOMs were constructed by choosing a 6 × 4 grid that generated 24 clusters. It was also our concern to present as few clusters as possible that would still give a clear picture of substance P- and antigen-induced gene expression. Increasing the number of clusters by increasing the grid did not gave us any additional correlation between genes. A variation filter was used to eliminate genes that did not change significantly across samples.

Array reproducibility. To determine the reproducibility of the arrays, we performed a regression analysis using an Excel spreadsheet. This test compared the results obtained with two pools of RNA isolated from sensitized wild-type mice that were challenged with saline and hybridized separately. The same analysis was performed using RNA isolated from sensitized wild-type mice that were challenged with antigen and substance P.

Venn diagrams. For representations of the numbers of genes up- or downregulated in each group as well as genes common to two or more groups, we used Venn diagram analysis performed by GeneSpring software [Silicon Genetics, Redwood City, CA (http://www.silicongenetics.com/cgi/SiG.cgi/index.smf.)] using raw data and filtering genes that were upregulated at least threefold in a comparison of substance P-, antigen-, and saline-challenged bladders isolated from +/+ , Kit^W/Kit^W-v, and BMR-NK1R^/ mice. Another set of analyses was performed by comparing genes upregulated or downregulated in antigen- and saline-challenged groups. Finally, we determined genes that satisfied all three conditions in response to either antigen or substance P.

Statistical analysis. For each group, 1,200 genes were analyzed by 2 different hybridizations using 2 different pools of RNA isolated from sensitized Kit^W/Kit^W-v, +/+, and BMR-NK1R^/ mice that were challenged with saline, antigen, or substance P. Therefore, a total of 21,600 data points were analyzed. SOMs were constructed by choosing a 6 × 4 grid that generated 24 clusters. As this type of analysis assumes that the data can be divided into a certain number of clusters and that they are well separated, it was also our concern to present the fewest clusters possible that would still give a clear picture of antigen- and substance P-induced gene expression. Increasing the number of clusters by increasing the grid did not produce any additional correlation between genes. GeneCluster software provides a list of genes present in each cluster and a centroid. However, to permit comparisons, in each cluster, gene expression was averaged and the SE was calculated. In each cluster, comparisons between gene regulation in response to antigen or substance P challenge were obtained by the ratio of gene expression in relation to the corresponding saline group. Significant differences were determined using an unpaired Student's t-test. The statistical analysis of histological data was performed using Wilcoxon's rank-sum test. Results are expressed as means ± SE. The n values reported refer to the number of animals used for each experiment. In all cases, a value of P < 0.05 was considered to represent a significant difference.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Morphological assessment of substance P-induced cystitis. We recently reported that mast cells were critical to the inflammatory response and gene expression associated with antigen-induced cystitis (72). In that study, we validated the approach of repopulating bladder mast cells in mast cell-deficient Kit^W/Kit^W-v mice by transplantation of bone marrow from normal +/+ mice (BMR-Kit^W/Kit^W-v).

/

/

/

/

/

Reproducibility of array hybridization. We previously presented evidence of the reproducibility of gene array methodology for the analysis of bladder inflammatory genes (69, 72) and verified the results using RNase protection assays (69). In the present work, we determined the reproducibility of our hybridization technique by performing regression analysis from values obtained with two different pools of RNA. Figure 1A represents regression analysis of RNA isolated from the bladder of sensitized +/+ mice that were challenged with saline (Fig. 1A). Similar analysis was performed using RNA isolated from the bladder of +/+ mice that were challenged with either substance P (Fig. 1B) or antigen (Fig. 1C). The calculated correlation coefficients for +/+ mice challenged with saline, antigen, or substance P were 0.9838, 0.9628, and 0.9718, respectively.

View larger version (15K): [in this window] [in a new window]   Fig. 1.   A-C: reproducibility of gene arrays. Regression analysis between 2 pools of RNA (1 and 2) isolated from sensitized normal (+/+) mice that were challenged with saline (A), substance P (SP; B), or antigen (C) was performed. cDNA probes were prepared from DNAse-treated RNAs, and 2 separate hybridizations to Atlas Mouse cDNA Expression Arrays were performed. Background counts were subtracted, and gene expression was calculated as percentage of -actin.

Role of mast cell NK1R expression on gene expression during experimental cystitis. Figure 2 is a Venn diagram showing the genes that were upregulated at least threefold in each animal group 24 h after substance P or antigen challenge (Fig. 2, A and B, respectively). The diagram indicates genes commonly upregulated between two or more groups. Interestingly, the genes expressed by mast cell-deficient Kit^W/Kit^W-v mice as a result of antigen challenge differed fundamentally from those observed with substance P challenge. In the mast cell-deficient Kit^W/Kit^W-v mice, only 13 genes were uniquely expressed after antigen challenge compared with 128 genes upregulated in response to substance P. In addition, the Venn diagram also highlights genes that were dependent on tissue mast cells for their upregulation but independent of the presence of NK1R on the mast cells. Thirty-two genes in the group of antigen-challenged mice and 21 genes in the group challenged with substance P fulfilled our criteria of being upregulated at least 3-fold in tissues isolated from +/+ and BMR-NK1R^/ but not in tissues isolated from mast cell-deficient Kit^W/Kit^W-v mice. In the presence of mast cells, only four genes were always upregulated in either inflammatory response (antigen or substance P). Those genes are represented by cluster 1 (Fig. 3A; Table 2) and include serine protease inhibitor (Spi) 2.2 (M64086), maspin (U54705), mitogen- and stress-activated protein kinase 2 (MSK2; AF074714), and macrophage colony-stimulating factor 1 (CSF-1 or M-CSF; X05010).

View larger version (15K): [in this window] [in a new window]   Fig. 2.   Cluster analysis of genes upregulated during SP- or antigen (Ag)-induced cystitis and the influence of the presence of mast cells and mast cells lacking neurokinin-1 receptors (NK1Rs). The ratio of gene expression obtained in bladders isolated from sensitized +/+, mast cell-deficient KitW/KitW-v (Kit), and Kit mice reconstituted with bone marrow mast cells from NK1R/ mice (BMR-NK1/) during SP (A)- or antigen (B)-induced cystitis was determined using GeneSpring software. Venn diagrams were generated using raw data and filtering genes that were upregulated at least 3-fold compared with respective saline-challenged mice. All mice were sensitized and killed 24 h after intravesical challenge. Yellow, congenic +/+ mice; blue, Kit mice; and red, BMR-NK1/ mice.

View larger version (41K): [in this window] [in a new window]   Fig. 3.   A-E: clusters 1-5. Kit, +/+, and BMR-NK1R/ mice were challenged with either Ag, SP, or saline. Twenty-four hours later, the urinary bladders of 3 mice were isolated, homogenized, and processed for isolation of RNA. The RNA was reverse transcribed to cDNA, labeled with [-32P]dATP, and hybridized to membranes containing Atlas Mouse cDNA Expression Arrays. Radioactivity was quantified using a PhosphorImager and interpreted by OptiQuant image-analysis software. Gene expression was normalized to -actin expression (72) and analyzed by GeneGluster software using self-organizing maps (SOMs) as described before (69, 72, 80). For each group, 1,200 genes were analyzed by 2 different hybridizations using 2 different pools of RNA isolated from sensitized Kit, +/+, and BMR-NK1R/ mice that were challenged with saline, Ag, or SP. Therefore, a total of 21,600 data points were analyzed. SOMs were constructed by choosing a 6 × 4 grid that generated 24 clusters, and 5 selected clusters are represented. The gene composition of cluster 1 is found in Table 2, and those for clusters 2-5 are presented in Table 3.

/

View larger version (52K): [in this window] [in a new window]   Fig. 4.   Sensory nerve-mast cell communication in cystitis. Activation of sensory nerves leads to release of SP, which stimulates mast cells to release their granules and to modulate gene upregulation independently of NK1Rs on mast cells. Mast cell mediators such as PGD2, histamine, and tryptase activate sensory nerves to further release SP. PGD2 activates prostaglandin receptors (PG), whereas tryptase responses are modulated by protease-activated receptors (PAR) present on bladder sensory nerves (21). The amplification of SP release by mast cells seems to be essential because mast cell-deficient mice presented a reduced inflammatory response compared with wild-type mice. Finally, NK1Rs present on blood vessels, endothelial cells, and smooth muscle cells play an absolutely mandatory role in inflammation because NK1R/ mice do not mount inflammation to SP and antigen stimulation despite an increased number of mast cells (73).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Several lines of indirect evidence suggest that substance P-mast cell interactions are involved in the development of interstitial cystitis. We now provide direct evidence that mast cells play a critical role in the pathogenesis of substance P-induced cystitis in mice. In the absence of mast cells, no demonstrable tissue edema or neutrophil infiltration was noted. However, these findings do not absolutely prove that mast cells are the cells responsible for this effect. For example, some other defect resulting from W mutations could theoretically account for the difference seen between +/+ and mast cell-deficient Kit^W/Kit^W-v mice.

To further investigate the role of mast cells in this reaction and whether mast cell-substance P interactions occurred via a mechanism utilizing the NK1R, we reconstituted mast cell-deficient Kit^W/Kit^W-v mice with bone marrow cells derived from NK1R-deficient mice. This unique animal model permitted us to examine the reaction in a tissue milieu where only the mast cell was deficient in NK1R expression. We found that inflammation induced by substance P was similar in BMR-NK1R^/ compared with normal +/+ mice. This finding provides definitive proof that mast cells are essential for substance P-induced bladder inflammation and that NK1Rs on mast cells are not required.

At this time, we do not know the mechanism involved in mast cell activation by substance P in the bladder. Clearly, NK1Rs are not involved; however, NK2R may play a role. In guinea pigs, an NK2R-dependent mechanism has been suggested in the activation of lung mast cells by substance P (51). Alternatively, a receptor-independent mechanism may be involved (52). In either case, it is important to emphasize that our results pertain to murine mast cells in the bladder, and given the known heterogeneity that exists in mast cell populations, these results should not be generalized to other anatomic sites or species.

We observed that 50% of sensitized mice bearing NK1R^/ mast cells died as a consequence of intravesical antigen challenge. In more than six years of research challenging sensitized mice with intravesical antigen administration, this was the first time that we lost mice to anaphylaxis. Our previous work indicates that this outcome was not observed when mast cells from wild-type mice were used to reconstitute the mast cell-deficient mice (72). However, NK1R^/ mice have at least three times more mast cells in the urinary bladder (73), and the same is valid in the stomach and esophagus (Saban R, unpublished observations). It is not clear why the absence of substance P receptors would lead to an increase in tissue mast cell numbers (72). Together with the present findings, our data strongly indicate that mast cells from NK1R^/ mice present an overwhelming response to antigen stimulation. Future studies are necessary to fully explain the differences between wild-type and NK1R^/ mast cells.

We next focused our attention on the genes expressed during experimental bladder inflammation. For this purpose, we examined gene expression in antigen- and substance P-induced cystitis, both of which we have now characterized as mast cell-dependent forms of inflammation (72). We found that there were more mast cell-dependent genes expressed during substance P responses than in antigen-induced bladder inflammation; i.e., only 13 genes were upregulated in Kit^W/Kit^W-v mice during antigen-induced reactions compared with 128 genes upregulated during responses to substance P. This is not entirely unexpected because relatively few cell types express IgE receptors compared with multiple immune and nonimmune cells that express receptors capable of interacting with substance P.

Although our results indicate that an NK1R on the mast cell is not necessary for the communication between sensory nerves and this immune cell, others have presented indirect evidence that activation of PMNs by substance P requires NK1Rs. In this instance, substance P primes PMNs exposed to recombinant IL-8 and suggested that substance P-priming effects are receptor mediated (24). In addition, substance P activates PMNs to release cytokines such as TNF- and interferon-, and this effect is potentiated by bacterial LPS (70). However, further studies using adequate animal models such as the one described in the present work are necessary to definitively access the role of neurokinin receptors on inflammatory cells.

In normal +/+ or BMR-NK1R^/ mice, only four mast cell-dependent genes were upregulated by either substance P or antigen. Those included Spi 2.2, maspin, MSK2, and M-CSF (or CSF-1). Interestingly, all four genes modulate inflammatory responses.

M-CSF (or CSF-1) acts to regulate the development and function of cells of the macrophage lineage (19). Macrophages are involved in inflammatory responses and, on treatment with M-CSF, displayed greater secretion of IL-6 (5). It was suggested that IL-6 occupies a central role in the CSF-1-regulated macrophage response to infection (36). In addition, it was suggested that the interplay between IL-6 and M-CSF switches monocyte differentiation to macrophages rather than dendritic cells and that IL-6 is an essential factor in the molecular control of antigen-presenting cell development (17).

Spi 2.2 is a member of the serpin gene family (74). In terms of physiological regulation of rat Spi 2.1 and 2.2 mRNAs, Spi 2.1 is dependent on growth hormone (GH) for maximal mRNA content (48, 94), whereas Spi 2.2 mRNA is not GH dependent (74). Spi 2 not only acts as a basal promoter element but also mediates transcriptional activation by GH and IL-6 (50). Spi 2 is known to be associated with proinflammatory processes. In the liver, Spi 2 genes have been correlated with the acute-phase response (8). These effects are mediated by the products of stimulated monocytes and macrophages in combination with glucocorticoids (7). In rat hepatocytes, turpentine induced an acute-phase response and an increase in Spi 2.2 expression, which was mediated primarily by IL-6 (7). The role of IL-6 on Spi 2.2 upregulation in the bladder should be further pursued because this cytokine is consistently elevated in patients with cystitis (26).

Maspin, another serine protease inhibitor related to the serpin family, is a secreted protein encoded by a class II tumor suppressor gene, whose downregulation is associated with the development of breast and prostate cancers (54). Maspin is associated with tumorigenesis, inflammation, and protection from autolysis by granule proteinases (75). Maspin interacts with the p53 tumor-suppressor pathway and functions as an inhibitor of angiogenesis in vitro and in vivo (96).

MSK2 is a novel protein kinase that is activated in vitro and in vivo by either MAPK/ERKs or SAPK2/p38. MSK1 (and/or MSK2) mediates activation of the transcription factors calcium/cAMP response element binding protein and activating transcription factor 1 by either growth factors or stress signals (20). Others have shown that substance P activates MAPKs (15) and, at nanomolar concentrations, induces the human astrocytoma cell line U373 MG to produce IL-6 by a p38 MAPK-dependent pathway (29).

The result of upregulation of proteases such as Spi 2.2 and maspin, SAPK, and CSF-1 by either substance P- or antigen-induced responses indicates a common mast cell-dependent pathway involved in acute bladder inflammation. These results suggest that activation of mast cells initiates activation of monocytes and/or macrophages and their resulting products such as IL-6 may modulate bladder inflammation.

There are several available methods for the analysis of gene microarray data, based on different mathematical assumptions and algorithms. Each of the analysis methods has advantages and limitations (65, 76).

An increasingly common approach involves using gene expression behavior observed over multiple experiments to first cluster genes together into groups, either by manually examining the data (16) or by using statistical methods such as SOMs (80), K-means clustering, or hierarchical clustering (25, 78). K-means clustering is a completely unstructured approach, which proceeds in an entirely local fashion and produces an unorganized collection of clusters that is not conducive to interpretation (80). Several recent papers employed hierarchical clustering algorithms to organize genes into a phylogenetic tree, reflecting similarity in expression patterns. However, the interpretation of these clusters is left to the observer.

We preferred to use SOMs because they are potent tools for identifying clusters involved in biologically related pathways and mechanisms (25, 69, 80, 82). The basic assumption underlying this unsupervised analysis is that genes with similar expression behavior (for example, increasing and decreasing together under similar circumstances) are likely to be related functionally. Although not logically rigorous, the utility of the cluster approach has been demonstrated, as genes already known to be related do, in fact, tend to cluster together based on their experimentally determined expression patterns (69). The validity of this approach has been demonstrated for many genes in Saccharomyces cerevisiae, a simple organism for which the entire genomic sequence and the functional roles of ~60% of the genes are known (25, 91). One limitation of SOMs is that genes negatively associated are not clustered.

SOMs can be performed by using Cluster software (http://rana.lbl.gov/) developed by Michael Eisen's laboratory or GeneSpring (http://www.silicongenetics.com/cgi/SiG.cgi/index.smf.), which permit the calculation of SOMs, K-means, and principal component analysis on the same set of data. The reason for the use of SOMs with GeneCluster software developed by Tamayo et al. (80) is that it performs the analytical calculations and provides easy data visualization. The approach is made more systematic and statistically sound by calculating the probability that the observed functional distribution of differentially expressed genes could have happened by chance. The application of statistical rigor is essential to avoid overly subjective interpretations of the results based on the predispositions, prior knowledge, and interests of the individual researcher (49). The power of cluster analysis was enhanced in the present work by the use of an excellent biological paradigm and a stringent criterion for cluster selection.

To fairly interpret gene cluster analysis, we must be cognizant of a growing body of evidence of mechanisms that control the rate of synthesis and half-life of proteins (31) and that gene and protein changes can be dissociated (31, 37). Future proteomic correlation must determine how directly mRNA changes reflect translated protein levels and the physiological consequence of these proteins.

In conclusion, our results confirm a mandatory role of mast cells in substance P-induced bladder inflammation in mice and suggest that mast cells mediate inflammation and gene regulation in the absence of NK1Rs on its cell surface. The cDNA array experimental approach provides a global profile of gene expression changes in bladder tissue after stimulation with antigen or substance P. SOMs identified functionally significant gene clusters. These gene expression responses may represent a balance between the cytoprotective and inflammatory processes that accompany bladder response to injury. Gene cluster analysis techniques can be applied in the future to begin to understand clinically relevant issues, such as how and why the transition from acute to chronic inflammation occurs only under selected circumstances and which therapeutic strategies can be used to selectively target genes expressed in bladder inflammation. However, cluster analysis and gene array profiling will have a strong impact whenever unique animal models such as Kit^W/Kit^W-v, BMR, and BMR-NK1R^/ are included, which allow establishment of biological paradigms to be tested.