A Signaling Network in Phenylephrine-Induced Benign Prostatic Hyperplasia

Abstract

Benign prostatic hyperplasia (BPH) is an age-related disease of unknown etiology characterized by prostatic enlargement and coinciding with distinctive alterations in tissue histomorphology. To identify the molecular mechanisms underlying the development of BPH, we conducted a DNA microarray study using a previously described animal model in which chronic α(1)-adrenergic stimulation by repeated administration of phenylephrine evokes histomorphological changes in the rat prostate that resemble human BPH. Bioinformatic tools were applied to microarray data obtained from prostate tissue to construct a network model of potentially relevant signal transduction pathways. Significant involvement of inflammatory pathways was demonstrable, including evidence for activation of a TGF-β signaling cascade. The heterodimeric protein clusterin (apolipoprotein J) was also identified as a prominent node in the network. Responsiveness of TGF-β signaling and clusterin gene and protein expression were confirmed independently of the microarray data, verifying some components of the model. This is the first attempt to develop a comprehensive molecular network for histological BPH induced by adrenergic activation. The study also implicated clusterin as a novel biochemical target for therapy.

A novel signal transduction network, revealed from studies of α-adrenergic receptor-mediated benign prostatic hyperplasia (BPH) in the rat, identifies the apoptosis-related protein, clusterin, and the TGF-β pathway as likely signaling mediators of histomorphologic changes in the prostatic epithelium.

Benign prostate hyperplasia (BPH, also known as nodular hyperplasia, benign prostatic hypertrophy, or benign enlargement of the prostate) is a hormone- and age-related disease, characterized by histological changes in the prostate gland and variable increases in prostatic size. BPH can lead to lower urinary tract symptoms and, in some cases, bladder obstruction. Histological changes in BPH tissues are characterized by stromal and/or epithelial hyperplasia within discrete nodules that are generally located in the periurethral region of the prostate (1,2,3).

Stromal-epithelial interactions play a critical role in normal prostatic development and growth (4). Prostatic stromal cells mediate phenotypic events in epithelial cells, including proliferation, ductal branching morphogenesis, and basal and luminal cytodifferentiation. It appears that, in many cases, pathogenic conditions in the prostate can be triggered by imbalances in the interactions between the stroma and the epithelium (5,6). During BPH pathogenesis, altered expression of growth-promoting factors may elicit changes in downstream signaling pathways and in the tissue microenvironment. ELISA analysis revealed that human BPH prostate tissue contains increased levels of fibroblast growth factor (FGF)-7 compared with normal prostate (7). Enhanced mRNA levels of IL-2, -4, and -17 in human prostate specimens from BPH patients were found using an RT-PCR approach (8,9). However, despite these and other studies (2,9,10), the molecular mechanisms underlying BPH are poorly understood, and a direct functional link between BPH pathogenesis and specific signaling pathways has not yet been demonstrated.

A rodent model of BPH has been described in which repeated sc injection of Wistar rats with phenylephrine (PE), an α(1)-adrenoceptor agonist, induces atypical prostatic hyperplasia (11,12). Histomorphological changes seen in this model include widespread atrophy of acini, increased accumulation of concretions in glandular lumen, infiltration of inflammatory cells, and focal epithelial atypia. Although the initial stimulus in the PE model is likely α(1)-adrenoceptor activation of the prostatic smooth muscle, the mechanisms underlying the subsequent changes in the prostatic epithelium are unknown. The purpose of the present study was to use the Wistar rat PE model of BPH, in combination with DNA microarray analysis and bioinformatics approaches, to identify one or more signaling networks involved in the histological changes seen in this model.

Materials and Methods

Reagents

PE was purchased from Sigma-Aldrich (St. Louis, MO). Antibodies against clusterin, phospho-Smad2, and β-actin were obtained from Santa Cruz Biotechnology (Santa Cruz, CA; anti-clusterin for Western blot analysis); Upstate Biotechnology [Lake Placid, NY; anti-clusterin for immunohistochemistry (IHC)]; and Cell Signaling Technology Inc. (Danvers, MA). Antibody against TGF-β ligands (all three isoforms) was a generous gift from Dr. Kathleen C. Flanders (National Cancer Institute, Bethesda, MD). Rat AceGene Chips 30K were obtained from Hitachi Software Engineering Co., Ltd. (Tokyo, Japan). All IHC reagents were purchased from Sigma-Aldrich and Dakocytomation (Carpinteria, CA). The clusterin expression construct, pIRES-clusterin, and vector were kindly provided by Dr. Saverio Bettuzzi (University of Parma, Parma, Italy).

Experimental animals

The Animal Care Committee of Ehime University Graduate School of Medicine approved all procedures, and the animals received humane care during the experimental period. Male Wistar rats of age 7 wk were used for these experiments. The animals were housed in rooms with controlled temperature and a 12-h light, 12-h dark cycle and had free access to standard rat chow pellets and water. The animals were randomly divided into eight groups, each consisting of six to seven subjects. PE dissolved in saline was injected sc daily. The control group received injections of saline only. For a positive control for clusterin expression, three rats were castrated for 4 d. The three prostatic lobes were harvested separately while the animals were under ether anesthesia at 4, 7, 14, and 28 d. Half of the tissues were stored at −80 C for later use, and the remaining tissues were fixed immediately in 0.1 m phosphate-buffered 10% formalin (pH 7.4) for 48 h and then embedded in paraffin. Serial 4-μm-thick sections from each tissue specimen were prepared and mounted on poly-l-lysine-coated glass slides.

IHC analysis of rat prostate tissues

Hematoxylin and eosin staining was performed to observe histomorphology. For analysis of clusterin expression, sections from the paraffin-embedded tissue blocks were mounted on charged glass slides and baked at 60 C for 1 h, dewaxed, and rehydrated. The tissue sections were treated using 2% H₂O₂ in 50% methanol for 15 min to eliminate endogenous peroxidase. The sections were heated in 10 mm citrate buffer (pH 6.0) for 30 min in a microwave oven (Sharp, Tokyo, Japan) to facilitate antigen retrieval. This was followed by incubation for 30 min at room temperature with normal rabbit serum. Subsequently, the sections were incubated with specific antibodies (clusterin, 1/200; TGF-β, 1/100; p-Smad2, 1/200 dilution) for 16 h at 4 C in a humidified chamber. After washing with PBS, the sections were incubated for 30 min at room temperature with N-Histofine Simple Stain Rat MAX PO (M) (Nichirei Biosciences Inc., Tokyo, Japan). The immunoreaction was visualized using 3,3′-diaminobenzidine (Sigma-Aldrich). All sections were lightly counterstained with hematoxylin. The negative controls comprised serial sections that were stained using equivalent concentrations of nonimmune mouse IgG in place of the primary antibodies. The level of staining was evaluated independently by three observers (J.K., T.K., and N.J.) blinded to experimental conditions. Based on predetermined criteria, 10 fields from at least three rats were scored as negative or positive. Positive cells were then transformed into percentage of total cells within a field. Mean and sd were graphed with P values.

Microarray gene expression profiling

Total RNA was isolated from rat ventral prostate harvested 4 d after PE treatment. After cDNA was synthesized and purified, amino allyl-labeled RNA was synthesized using an amino allyl MessageAmp aRNA kit (Ambion, Austin, TX). The rat AceGene oligo chip was used for microarray analysis. Initial data analysis for each chip was performed using DNASIS-Array software (Hitachi Software Engineering). Bioinformatics resources, Ingenuity Pathway Analysis (Ingenuity Systems, Redwood City, CA), and MetaCore (GeneGo, Encinitas, CA) pathway analysis, were used for functional pathway and network analysis.

Cell culture and transfection

NbE-1 rat prostate epithelial cells (13,14) were maintained in T-medium (Life Technologies, Inc., Grand Island, NY) supplemented with 10% heat-inactivated fetal bovine serum (Life Technologies) containing 100,000 U/liter penicillin and 100 μg/liter streptomycin at 37 C in a humidified atmosphere (95% air/5% CO₂). NbE-1 cells in 100-mm dishes at about 80% confluence were transfected with small interfering RNAs (siRNAs) or plasmid vectors using Lipofectamine 2000 according to the manufacturer’s instructions (Invitrogen, Carlsbad, CA). To knock down clusterin protein levels cells were transiently transfected with siRNA targeting clusterin. Clusterin targeting and scrambled siRNAs (control) were from Dharmacon, Inc. (Chicago, IL). For enforced expression of clusterin, cells were transfected with clusterin expression constructs (15,16).

Western blot analysis

Prostate tissues were quickly harvested, flash frozen in liquid nitrogen, and stored at −80 C. After homogenization, total protein was extracted in lysis buffer [1% Nonidet P-40, 50 mm Tris (pH 7.4), 10 mm NaCl, 1 mm NaF, 5 mm MgCl₂, 0.1 mm EDTA, 1 mm phenylmethylsulfonyl fluoride, and Complete protease inhibitor cocktail tablet (Roche Diagnostics GmbH, Mannheim, Germany)] at the indicated conditions and centrifuged at 12,000 × g for 15 min. Supernatants were subjected to SDS-PAGE and transferred to nitrocellulose membranes. After blocking with 10% skim milk/PBS, membranes were incubated with specific antibodies. The blots were visualized by enhanced chemiluminescence (ECL; Amersham Bioscience, Little Chalfont, Buckinghamshire, UK).

Real-time RT-PCR

Total RNA was extracted from the snap-frozen tissue samples using TRI reagent (Sigma-Aldrich) according to the manufacturer’s instructions. RT was performed in a 30-μl reaction mixture. The RT reaction contained 2 μg RNA, 10 mm Tris-HCl (pH 8.3), 50 mm KCl, 1.5 mm MgCl₂, 2.5 mm dithiothreitol, 500 μmol/liter each of dATP, dCTP, dGTP, and dTTP (Amersham Biosciences), 40 U RNasin (Roche), 25 μg/ml oligo dT [pd(T)_12–18; Amersham Biosciences] and 200 U Moloney murine leukemia virus reverse transcriptase (Invitrogen). The reaction mixture was incubated at 37 C for 90 min and then heated to 100 C for 5 min. The resultant cDNA was used for PCR. For quantitative real-time RT-PCR, we prepared appropriate dilutions of each single-strand cDNA followed by normalizing of the cDNA content using β-actin as a quantitative control. Quantitative PCR amplification was performed with a 20-μl final volume consisting of 0.5 μl RT reaction mixture, 3 mm MgCl₂, 0.3 μm of each sense and antisense primer, and 2 μl LightCycler Fast Start DNA Master SYBR Green I (Roche Diagnostics). We used the following primers: rat clusterin forward, 5′-CTGACCCAGCAGTACAACGA-3′, and reverse, 5′-TGACACGAGAGGGGACTTCT-3′ (196 bp), and rat β-actin forward, 5′-GTCACCCACACTGTGCCCATCT-3′, and reverse, 5′-ACAGAGTACTTGCGCTCAGGAG-3′ (542 bp). PCR conditions were as follows: initial denaturation at 95 C for 10 min and 40 cycles of denaturation at 95 C for 10 sec, annealing at 62 C for 10 sec, and elongation at 72 C for 8 sec. We carried out real-time quantitative PCR using a LightCycler Quick System 350 (Roche Diagnostics) coupled to LightCycler software version 3.5 (Roche Diagnostics). Clusterin expression in the test samples was normalized to the corresponding β-actin level and was reported as the fold difference relative to the β-actin gene expression.

Statistical analysis

All experiments were repeated at least three times, and results are shown as mean and sd from independent experiments. Data were compared using a paired Student’s t test. P values < 0.05 were considered to be statistically significant.

Results

Stimulation of α(1)-adrenoceptor induces histological BPH

To understand the molecular basis underlying the reported effects of sympathomimetic stimulation on prostatic glandular histomorphology (11,17), we performed DNA microarray analysis using a previously described animal model in which PE is applied sc once a day. Wistar rats were randomly selected and divided into PE and vehicle-only (control) groups as described in Materials and Methods. Animals in the PE group were injected sc with PE at varying doses (0, 1, 5, or 10 mg/kg · d) for 4, 7, 14, or 28 d. At the time of killing, the entire prostate complex was removed and the ventral prostatic lobe (VP) and dorsolateral and anterior prostatic lobes were separated for evaluation.

The histological analyses revealed that administration of PE resulted in dramatically altered glandular morphology. In contrast to the dilated acini surrounded by a single layer of columnar cells in the control group, the PE group showed decreased number and size of acini, with a response seen 7 d after the beginning of the experiment. Columnar epithelium in the PE group was more elongated and stratified with intraluminal projections compared with control, resulting in histomorphological changes with a more distorted, folded shape (Fig. 1, A and B). At higher magnification (×40), nuclei were enlarged and arranged in an irregular manner in the PE group compared with control tissues. Changes in cell morphology were more robust with increasing time of treatment (Fig. 1B). These histological responses are consistent with the morphological features that have been reported for this model (11,17) and are suggestive of prostatic hyperplasia as previously described. The vehicle group had no discernible prostatic morphological changes.

Figure 1.

Chronic treatment with PE induces histomorphological changes in the rat prostate. A, Adolescent male Wistar rats were administrated daily with saline solution (control) or PE for 28 d with increasing doses (1, 5, or 10 mg/kg). Prostates were separated into constituent lobes [anterior (AP), dorsolateral (DLP), and VP] and analyzed histologically (hematoxylin and eosin). B, Rats were administrated with 10 mg/kg PE daily. VPs were harvested at 4, 7, 14, or 28 d after PE administration. Histomorphology of the PE and the control group is shown at ×10, ×20, and ×40 magnification (hematoxylin and eosin).

Construction of a network model from DNA microarray data

cDNA microarray analysis was performed with RNAs extracted from VPs obtained from either the 10 mg/kg/d PE or saline groups 4 d after the start of the experiment. VP tissue was used as representative because histomorphological changes were similar in all prostatic lobes. We queried 30,000 transcripts using Rat AceGene 30K Chips (Hitachi Software Engineering). After filtering out data with low quality scores due to low intensity values, and normalizing gene expression profiles to fold change compared with control, we generated a primary gene list with statistically significant differences in expression between the two conditions (using a statistical cutoff at P < 0.001).

To identify important functional modules perturbed in response to PE, and to select a subset of genes potentially related to the hyperplastic alterations, we interrogated the primary gene list with the Ingenuity Pathways Analysis bioinformatics tool (http://www.ingenuity.com), which is useful for constructing interaction models that objectively arise from microarray datasets. In modeling the PE data, we uncovered a series of independent interactions that involved clusterin (apolipoprotein J), which is expressed in many tissues and is implicated in various biological processes such as aging, cancer, vascular damage, and diabetes (18,19,20,21). Clusterin was the most prominent node identified in the pathway analysis (Fig. 2A). Solid lines in the figure indicate direct interactions (physical binding events between molecules), arrows designate a regulatory relationship (positive or negative), and the shapes are indicative of the molecule class/protein family.

Figure 2.

Clusterin as a critical signaling node. A, Candidate genes identified by DNA microarray were analyzed with the Ingenuity software tool to construct an objective signaling network. Up-regulated genes in the PE-treated rats are shown in red; down-regulated genes are depicted in green. B and C, Analysis of the network demonstrates a link between clusterin and the TGF-β signaling pathway. Components of the network are listed in B with fold change compared with control. Regulatory network showing potential interactions between molecules is shown in C.

We also identified 18 genes within the TGF-β pathway that were altered in the PE group in comparison with the control group (Fig. 2B). TGF-β is a multifunctional cytokine involved in the regulation of cell proliferation, differentiation, survival/apoptosis, and inflammation (22). Notably, clusterin is networked with TGF-β receptor 1 (TGFBR1), TGFBR2, and the Smad family. Collectively, the model shown in Fig. 2 suggested that PE induces changes in the TGF-β signaling pathway in the prostate in a manner that involves clusterin as a central component of the response.

Pathways activated by chronic treatment with PE

To determine whether the tissue changes seen with PE administration could be objectively associated with one or more biological processes, we employed functional enrichment analysis based on models of pathway maps and biological processes using GeneGo’s MetaCore (http://www.genego.com) software tool. This bioinformatics program compares groups of genes derived from each expression profile to an annotated database generated from reported protein or genetic networks. It can extract pathways from input data that reflect the given experimental perturbation. Pathway and biological process associations identified by MetaCore from genes whose expression was greatly increased or decreased (>2.5-fold) in the PE condition compared with the vehicle condition are shown in Table 1. These independent analyses using different bioinformatics programs suggest that immune responses and processes related to inflammation or cell cycle regulation are markedly altered in the prostate upon chronic PE treatment. A heat map showing the inflammation-related gene expression changes in order of intensity is shown in supplemental Fig. S1 (published as supplemental data on The Endocrine Society’s Journals Online web site at http://endo. endojournals.org).

Table 1.

Pathways significantly associated with PE-treated prostate tissue compared with control

	Pathway or biological process models	P value (−log)
Up-regulated genes (>2.5-fold) in PE model
Pathway models
1	Immune response: alternative complement pathway	7.2
2	Immune response: classicaI complement pathway	4.88
3	Development: leptin signaling via PI3K-dependent pathway	4.88
4	Regulation of lipid metabolism: insulin regulation of glycogen metabolism	4.73
5	Immune response: role of the C5b-9 complement complex in survival	3.88
Biological processes models
1	Inflammation: complement system	4.83
2	Signal transduction: leptin signaling	3.14
3	Proteolysis: ECM remodeling	2.45
4	Neurophysiological process: transmission of nerve impulse	2.39
5	Inflammation: IFN-γ signaling	2.07
Down-regulated genes (>2.5-fold) in PE model
Pathway models
1	Retinol metabolism	3.9
2	Cell cycle: role of APC in cell cycel regulation	3
3	Cell cycle: cell cycle (generic scheme)	1.51
4	Cell cycle: chromosome condensation in prometaphase	1.51
5	Cell cycle: sister chromatid cohesion	1.49
Biological processes models
1	Cytoskeleton: spindle microtubules	3.03
2	Cell cycle: mitosis	1.91
3	Development: neurigenesis	1.91
4	Cytoskeleton: intermediate filaments	1.68
5	Cell cycle: core	1.41

GeneGo’s MetaCore functional enrichment analysis was performed using genes whose expression was increased or decreased over 2.5-fold compared with control. 

Clusterin expression is increased in response to PE

To test the validity of our in silico analyses, we initially assessed the level of clusterin mRNA expression by real-time RT-PCR. Consistent with the network model, clusterin mRNA levels were significantly higher 4 d after initiating PE treatment and recovered to basal levels at 14 d (Fig. 3A). The clusterin mRNA level seen with control VP tissues was not significantly altered at all experimental time points. Western blot analysis of VP tissues demonstrated that clusterin protein was robustly higher in the PE group (Fig. 3B), in agreement with the mRNA data. Examination of the IHC data showed that clusterin was increased in the PE group (Fig. 3C). As a positive control for clusterin immunostaining, we used VP tissue from castrated rats (data not shown) (23).

Figure 3.

Increased prostatic clusterin in response to PE. Rat VPs from PE-treated or control groups were harvested at the indicated time points (4, 7, 14, or 28 d after treatment). A, Clusterin mRNA levels were measured in PE-treated and control rats and quantified by normalization with mRNA level of β-actin intensity. B, Western blot analysis using anti-clusterin antibody reveals an increase of clusterin protein in the PE-treated group compared with control. C, Clusterin IHC was performed for VP tissues harvested from PE rat (left) and control (right). Data shown are representative of at least three independent experiments.

TGF-β and active Smad2 are increased in response to PE

As another test of the network model, we compared TGF-β immunostaining intensity in the PE vs. control VP groups. Our IHC experiments demonstrated that TGF-β expression was still low 14 d after saline treatment (control) but was significantly increased 14 d after PE treatment (Fig. 4A). IHC analysis at 28 d demonstrated that TGF-β protein expression was higher in the PE group (Fig. 4B). In the canonical TGF-β signaling pathway, Smad2 is one of the key intracellular signal transducers for responses induced by TGF-β. Upon binding of TGF-β to its receptors (TGFBR1 and TGFBR2), Smad2 becomes activated after phosphorylation at serine 465/467, transits to the nucleus, and participates in transcriptional activation (24,25). Analysis of the VP tissues demonstrated that Smad2 phosphorylation level was also enhanced in the PE group compared with control. Over 40% of prostate epithelial cells were positively stained with phosphospecific Smad2 antibody 14 d after PE treatment, and over 60% of cells were positive at the 28-d time point (Fig. 5B, black arrows). Significantly, active Smad2 was predominantly located in cell nuclei, suggesting the protein is active as a transcription factor.

Figure 4.

TGF-β levels in PE-treated rats. To evaluate the level of TGF-β, IHC analysis was performed using VPs from PE and control groups. A, Positively stained cells and total cells were counted to calculate the extent of TGF-β positivity. At least 10 fields per time point and 100 cells per field were counted. The tabulated result is the mean of data obtained from three independent experiments. Error bars represent sd; *, P < 0.05. B, Representative IHC images at 14 and 28 d from the indicated condition are shown.

Figure 5.

Smad2 phosphorylation is enhanced in PE rats. A, VPs from BPH and control rats were used for IHC using phospho-specific anti-Smad2 antibody. Numbers of phospho (p)-Smad2-positive cells and total cells were counted to obtain the percentage of positive cells (shown in graph). Error bars represent sd; *, P < 0.05. B, Representative p-Smad2 images are shown. Arrows indicate examples of positively (black) or negatively (white) stained cells.

To test directly whether clusterin interacts with the TGF-β pathway in rat prostate epithelial cells, we altered clusterin expression levels in the immortalized rat prostate epithelial cell line NbE-1 (13,14). NbE-1 exhibits characteristics of normal prostatic epithelial cells and forms a functional epithelial monolayer in culture. Although enforced expression of clusterin sensitized and increased Smad2 activation upon TGF-β treatment (Fig. 6A), cells with reduced clusterin expression as a result of treatment with siRNA targeted to clusterin were less responsive to TGF-β stimulation compared with control (Fig. 6B).

Figure 6.

Expression level of clusterin affects TGF-β signal cascades in vitro. A, Normal rat prostate epithelial cells (NbE-1) were transiently transfected with a clusterin expression construct (C) or vector (V) only. At 24 h after transfection, cells were stimulated with TGF-β after serum starvation for 16 h. Phosphorylation level of Smad2 was determined by Western blot analysis using specific phospho (p)-Smad2 antibody. Expression levels of Smad2, clusterin, and β-actin were determined by Western blot. Forced expression of clusterin sensitizes Smad2 phosphorylation by TGF-β compared with control (vector). B, Clusterin protein levels in NbE-1 cells were reduced by siRNA. Smad2 phosphorylation by TGF-β treatment was abrogated by siRNA targeting clusterin (upper panel). Western blot using β-actin was used as loading control (lower panel).

Discussion

In this study, we used a combination of DNA microarray analyses and unbiased bioinformatics approaches to develop a new signal transduction network model for the molecular mechanism underlying histomorphological changes induced in the rat prostate by repeated α(1)-adrenoceptor agonist stimulation. We verified previous reports that the sympathetic agonist PE induces alterations in tissue architecture in all three prostatic lobes that resemble human BPH (11,12). Finally, we validated several features of this network model using conventional biochemical methods. Our results suggest a pivotal role for clusterin, the most prominent node in our network analysis, in the response of prostatic epithelia to repeated systemic PE treatment.

The principal rationale for conducting our study is that the molecular mechanism that results in BPH in humans is still obscure. To understand the etiology and pathogenesis of BPH, models have been developed using the chimpanzee, dog, rat, and mouse (26). The canine is generally accepted as a relevant animal model that recapitulates a condition resembling human BPH because urethral obstruction occurs spontaneously and also after application of steroids (e.g. 5α-dihydrotestosterone and 17β-estradiol). However, ethical concerns and high expense have limited the use of primates and the dog (27). The adult male C57/BL6 mouse undergoes a histomorphological response in the prostate similar to that seen in the Wistar rat after prolonged sympathomimetic stimulation, resulting in marked atypical prostatic epithelial hyperplasia characterized by multilayered epithelial cells and papillary unfolding into the lumen (17). Transgenic mouse lines that overexpress int-2 (28), IGF-I (29), or FGF-2 in prostatic stromal cells (30) also have been experimentally developed (31).

Two other DNA microarray studies have been conducted previously on BPH tissues. Prakash et al. (32) identified over 500 human gene signatures that were differentially expressed in the symptomatic and asymptomatic BPH using the 42K Affymetrix HuGene FL array. Luo et al. (33) performed cDNA microarray analysis using nine human BPH tissues and 12 normal prostate tissues assayed using 6500 human gene probes. Consistent with our data showing a correlation between inflammation and BPH in the PE model, the clustering analysis performed by Prakash and co-workers (32) showed that symptomatic BPH correlated with proliferation and inflammation gene expression profiles, including involvement of chemokine receptors, Igs, and cytokines, that were distinct from the normal or asymptomatic BPH groups. Our unbiased network mapping approach identified a protein interaction module from rat VP (Fig. 2C), which was similar to the findings of the Luo et al. (33) report [IGF-I, IGF-II, TGF-β3, bone morphogenetic protein 5, latent TGF-β-binding protein 1 (LTBP1), and LTBP2] using human samples. Both studies suggest that a common network (including TGF-β and LTBP) is active in BPH. However, a role for clusterin was uniquely obtained in the present analysis but not these previous studies. A recent report identified up-regulation of phospho-Smad3, Snail, and Slug within the epithelial compartment of human BPH specimens, also consistent with a role for the TGF-β pathway in BPH (34).

Several lines of evidence point to the involvement of clusterin: 1) clusterin level is increased in response to PE, 2) clusterin is networked with a number of proteins differentially expressed in the experimental vs. the control group, and 3) the deduced network pointed to a link between clusterin and the TGF-β signaling pathway. Clusterin is expressed ubiquitously in human tissues; mediates tissue remodeling, lipid transport, and apoptosis; and is functionally involved in aging, cancer progression, and vascular damage (20). Clusterin was first identified as a gene that is overexpressed in the rodent prostate after androgen deprivation (23). Antisense oligonucleotide therapy directed against clusterin has been reported to improve success rates as an adjunct agent along with chemotherapy, radiation, and hormone depletion therapy in prostate cancer patients and in animal models of prostate cancer (21,35). However, the precise role of clusterin in tumor progression is in dispute. Increased (36,37) or decreased (15,38,39,40) clusterin expression has been reported in various tumor systems, suggesting that clusterin functions in either an antiapoptotic or proapoptotic role depending on the physiological context.

Another interesting finding in the present study is that immune response and inflammation pathways were markedly activated [with significance values (−log) of 7.2 and 4.83, respectively (Table 1)] in the prostates of PE-treated animals. Disruptions in immunological pathways have been reported in BPH (22), which has been linked to systemic immune responses like chronic inflammation (41). About 40% of BPH patients exhibit inflammatory features histologically (42), including infiltration by activated T cells, mast cells, and macrophages capable of secreting growth factors and proinflammatory cytokines such as IL-2, TGF-β, and IFN-γ (43). A heat map showing a number of inflammatory genes affected by PE treatment in the present study is shown in supplemental Fig. 1. This cohort includes clusterin, which has been reported to be involved in inflammation (44). Pathway analysis, similar to what we performed using the Wistar rat data, was performed on human microarray data deposited by Luo et al. (33). Supplemental Fig. 2 shows that immune responses/inflammation pathways are also affected in human BPH, consistent with our observations in the rat. Published data show that daily sc administration of PE to rats for 1 month induced fibrosis and inflammation in the prostate as evidenced by vascular dilatation, edema, and leukocytic infiltrate (e.g. ED1⁺ cells) (12). Expression of T cell-derived cytokines including IFN-γ, IL-2, IL-4, and IL-13 was reported to be significantly increased in human BPH tissues compared with normal prostate tissues (45,46). ILs (e.g. IL-8) induce prostatic growth factors, such as FGF-2 (an autocrine prostatic stromal and epithelial growth factor) and FGF-17 (a stromal mitogenic factor), which can promote tissue changes consistent with those seen in human BPH (7,47,48). Collectively, these data strongly argue for inflammatory responses to be an important cellular mechanism underlying histological tissue changes in BPH.

In conclusion, we describe a novel signaling network model as a potential mechanism for histomorphological changes in the prostate induced by persistent α(1)-adrenergic stimulation. The likely role of smooth muscle and other influences on BPH in humans implies that elements of this signaling network may be relevant to the human disease. A network model can be productively tested in numerous ways and can be used to frame new hypotheses about disease etiology. Ours is the first such large-scale signaling network proposed as a mechanism for epithelial dysplasia associated with an adrenergic cause of BPH.