Expression of Pleiotrophin in the Prostate is Androgen Regulated and it Functions as an Autocrine Regulator of Mesenchyme and Cancer Associated Fibroblasts and as a Paracrine Regulator of Epithelia

BACKGROUND Androgens and paracrine signaling from mesenchyme/stroma regulate development and disease of the prostate, and gene profiling studies of inductive prostate mesenchyme have identified candidate molecules such as pleiotrophin (Ptn). METHODS Ptn transcripts and protein were localized by in situ and immunohistochemistry and Ptn mRNA was quantitated by Northern blot and qRT-PCR. Ptn function was examined by addition of hPTN protein to rat ventral prostate organ cultures, primary human fetal prostate fibroblasts, prostate cancer associated fibroblasts, and BPH1 epithelia. RESULTS During development, Ptn transcripts and protein were expressed in ventral mesenchymal pad (VMP) and prostatic mesenchyme. Ptn was localized to mesenchyme surrounding ductal epithelial tips undergoing branching morphogenesis, and was located on the surface of epithelia. hPTN protein stimulated branching morphogenesis and stromal and epithelial proliferation, when added to rat VP cultures, and also stimulated growth of fetal human prostate fibroblasts, prostate cancer associated fibroblasts, and BPH1 epithelia. PTN mRNA was enriched in patient-matched normal prostate fibroblasts versus prostate cancer associated fibroblasts. PTN also showed male enriched expression in fetal human male urethra versus female, and between wt male and ARKO male mice. Transcripts for PTN were upregulated by testosterone in fetal human prostate fibroblasts and organ cultures of female rat VMP. Ptn protein was increased by testosterone in organ cultures of female rat VMP and in rat male urethra compared to female. CONCLUSIONS Our data suggest that in the prostate Ptn functions as a regulator of both mesenchymal and epithelial proliferation, and that androgens regulate Ptn levels. Prostate 71:305–317, 2011. © 2010 Wiley-Liss, Inc.


INTRODUCTION
Stromal-epithelial interactions are involved in the regulation of prostate organogenesis and tumorigenesis. During prostate development, androgen action in mesenchyme regulates proliferation and differentiation of the prostatic epithelium, and in turn the epithelium regulates differentiation of the mesenchyme. Importantly, paracrine signaling from the mesenchyme to the epithelium acts to (1) specify prostatic epithelium identity, (2) induce bud formation, (3) elicit prostatic bud growth and ductal branching, (4) promote differentiation of secretory epithelia, and (5) specify the types of secretory proteins that are expressed, reviewed in Ref. [1]. This suggests that mesenchyme is the source of potent paracrine factors, yet few studies have endeavored to directly identify them [2,3]. The stroma is also involved in neoplastic prostate growth (reviewed in Ref. [4]) and cancer associated fibroblasts (CAFs) exhibit a variety of functional differences relative to normal stroma (reviewed in [5]). Tumor stroma is no longer able to restrain prostatic epithelial proliferation, but instead CAFs stimulate tumor growth [6,7] and angiogenesis [8].
To determine which molecules and pathways are active within developmental mesenchyme, we performed serial analysis of gene expression (SAGE) gene profiling of inductive mesenchyme of the ventral mesenchymal pad (VMP) [2]. This identified mesenchymal expression of pleiotrophin (Ptn) and suggested a localization to the VMP. Fgf10 [9], BMP 4 [10], Scube1 [2], Notch2, Dlk1, [11] and Wnt 5a [3] have previously been shown to be expressed in prostate mesenchyme and most of these have a functional role in prostate organogenesis. Since androgens stimulate prostate organogenesis via the mesenchymal compartment, this has led to the hypothesis that there may be ''andromedins'' made by the mesenchyme that act upon epithelia and which are produced in response to androgens. However, there is no direct evidence that ''andromedins'' identified in developing prostate are essential for androgen action, and there is controversy regarding direct or indirect regulation [12,13].
Ptn is a secreted, highly conserved cytokine, and belongs to a novel two member family of heparin binding molecules, sharing structural and functional similarities with midkine [14]. Ptn is an 18 kDa protein, also known as heparin affin regulatory peptide (HARP) and heparin-binding growth-associated molecule (HB-GAM). Ptn protein interacts with heparin sulfate proteoglycans (HSPGs) at the extracellular matrix [15]. Biological actions of Ptn are reported to be mediated by at least three different receptors; receptor protein tyrosine phosphatase, RPTB/x [16], syndecan-3 (N-syndecan) [17], and anaplastic lymphoma kinase (ALK) [18]. Ptn is responsive to regulation by steroid hormones, and Ptn transcript levels are upregulated in the presence of dihydrotestosterone or estradiol, in vitro and in vivo, respectively [15,19].
Ptn plays a key role in cellular growth and differentiation and has been implicated as playing an important role during development. Ptn expression is present in tissues during late embryogenesis and perinatal growth, and expression usually decreases around the time of birth [20,21]. Ptn is most frequently reported in tissues derived from mesoderm and in organs where mesenchymal-epithelial interactions play an important role, such as the salivary gland, lung, pancreas, kidney, and mammary gland [22][23][24]. In branching organs Ptn may also be important for branching morphogensis [25,26]. Ptn has not previously been studied in the developing prostate, although it is reported to play a role in prostate cancer [15]. Ptn-deficient mice have no gross anatomical abnormalities [27] and it has been proposed that Ptn and midkine, which are often co-expressed, might compensate for each other [23]. Mice lacking both Ptn and midkine show female infertility and reproductive abnormalities though no effects upon male fertility or reproductive anatomy were reported [28].
Ptn has diverse activity in vitro, stimulating the proliferation of a wide range of cells including epithelial, endothelial, and fibroblastic cell lines [29,30], and also stimulates progenitor cells in primary culture to enter lineage-specific differentiation pathways [20,31]. Ptn expression is associated with inflammatory diseases and tumor growth progression, and Ptn may recruit stromal tissue and vasculature to tumors [32][33][34][35]. Ptn expression has been observed in prostate, ovarian, testicular, pancreatic and breast cancer, solid gliomas, neuroblastomas, melanomas, and in several malignant cell lines of different origin, reviewed in Ref. [36].
We have examined the expression, distribution and function of Ptn in the developing prostate, as it was previously suggested to show mesenchymal expression [2], and we hypothesized that it may function as a paracrine regulator of prostate development. Ptn expression was restricted to VMP and mesenchyme of the VP as well as the smooth muscle of the urethra. Ptn protein was located in the mesenchyme and at the surface of epithelial cells, and displayed a gradient showing the highest levels at the tips of epithelial ducts. Recombinant hPTN increased proliferation and branching morphogenesis in cultures of VP organs grown in vitro, and stimulated the growth of fibroblasts derived from developing prostate and prostate cancer and prostate epithelial cells. Expression of Ptn mRNA and protein was increased by androgens in developmental mesenchyme, and showed sexually dimorphic expression.

Isolation and Culture of Primary Human Cells
Human fetal prostate tissue was obtained following medical termination of pregnancy. Consent was obtained in accordance with UK guidelines and the study was approved by the Lothian Research Ethics Committee. The bladder and urethra were excised, and the prostate microdissected prior to either RNA isolation or primary culture of embryonic prostate fibroblasts (hEPFs). Human adult prostate tissue was obtained from patients undergoing surgery at the Western General Hospital, Edinburgh. Primary CAFs were derived from tissue obtained from patients at the time of transurethral resection of the prostate (TURP). Patient consent was obtained prior to surgery and the study was approved by ethical review (MREC 02/5/63). To culture primary stromal cells from embryonic or cancer samples, tissue specimens were cut into small pieces (1 mm Â 1 mm Â 1 mm), and plated on a flask coated with fetal calf serum (FCS). Primary cells were grown in DMEM supplemented with 10% FCS and 1Â penicillin/streptomycin and CAFs were tested for tumorigenic activity in vivo [7].

Tissue Isolation From Rodents
Prostate and VMP tissues were microdissected from the urogenital tract of Wistar rats and ARKO mice. The day of copulatory plug observation was taken as e0.5, and day of birth was designated P0. Animal procedures were conducted in accordance with UK law regulated by the Home Office. The P0 female bladder and urethra were removed and dissected to isolate the urethra adjacent to the bladder termed the VSU, which comprised the VMP, smooth muscle, and urethral epithelia. Additionally, P0 VMP and ventral prostate (VP, from males) were microdissected. ARKO mice were produced by mating female mice heterozygous for the X-linked hypoxanthine phosphoribosyltransferase-Cre transgene with male ARflox mice [37]. This resulted in females carrying one deleted allele and one wild-type (WT) allele of the X-linked AR gene, which were subsequently mated to produce ARKO males and control littermates. ARKO males were identified by examination upon dissection and lacked prostate and Wolffian derived organs.

RNA Extraction and Quantitative RT-PCR
Total RNA was extracted using the RNeasy TM Mini kit (Qiagen, Crawley, UK) and RNA quality was measured on the Agilent 2100 Bioanlayser (Agilent Biotechnologies, Santa Clara, CA). Quantitative PCR's were performed on ABI 7500 machine, using the PowerSybr PCR mastermix (ABI, Warrington, UK). Transcript abundances were normalized to tbp/TBP expression. The PCR primer sequences are provided below:

Northern Blotting Analysis
DNA templates for Ptn (rat and human) and Gapdh were synthesized by RT-PCR from Rat P0 UGT cDNA or human embryonic prostate cDNA. The DNA template was subcloned into pGEM 1 -Teasy Vector (Promega, Southampton, UK). The primer sequences were; Rn Ptn forward-GCAGTCTGAAGAGAGCTCTG, Rn Ptn reverse-CCACTGGCAGAGACAATG and Rn Gapdh forward-TTAGCACCCCTGGCCAAGG, Rn Gapdh reverse-CTTACTCCTTGGAGGCCATG, Hs PTN forward-CAGCTGTGGATACTGCTG, Hs PTN reverse-GACAGTCTTCTGGCATTCG. Northern blot hybridization used 5 mg total RNAs as described in Ref. [2]. We observed one band ($1.4 kb) on the Northern blot, corresponding to the predicted size of the Rn Ptn transcript. The intensity of the bands was quantified using a phosphoimager, and Ptn expression was normalized to Gapdh.

Whole-Mount In Situ Hybridization
The rat Ptn template was prepared by TA cloning an 681 bp amplified PCR fragment corresponding to nucleotides 426-1,107 into pGEM 1 -Teasy. RNA probes were prepared by in vitro transcription using T7 and T3 RNA polymerases (DIG RNA labeling kit, Roche, Burgess Hill, UK). Dissected rat UGT tissue was fixed in 4% paraformaldehyde overnight, dehydrated through graded methanol and stored in 100% methanol at À208C. RNA in situ hybridization on embryonic and P0 urogenital tracts and cultured ventral prostates were performed using the InsituPro VS robot (Intavis, Bioanalytical Instruments, AG, Cologne, Germany) as

Western Blot Analysis
Protein was isolated from frozen tissue (10-30 organs) using radioimmunoprecipitation assay (RIPA) lysis buffer, and concentration was determined using a Bio-Rad (Hemel Hempstead, UK) protein assay kit. Twenty micrograms of protein extract was run on a Bis-Tris 4-12% gel (Novex pre-cast gels, Invitrogen), and electrotransferred overnight onto nitrocellulose membranes (Immobilon-FL; Millipore, Bedford, MA). The membranes were blocked for 1 hr in 5% milk/TBS, 50 mM Tris-HCl, and 150 mM NaCl and incubated overnight at 48C with anti-PTN antibody (R&D Systems, Abingdon, UK) diluted 1:100 and anti-btubulin antibody (Santa Cruz Biotechnologies, Inc., Santa Cruz, CA) diluted 1:1,000 in 5% milk/TBST (TBS containing 0.05% Tween-20). The membranes were washed in TBST before addition of the appropriate secondary antibody, goat anti-rabbit IRDye 800 (LI-COR Biosciences, Cambridge, UK) and donkey anti-goat Alexa fluor 680 (Invitrogen) diluted 1:10,000 in TBST. Membranes were washed in TBST prior to scanning on the LI-COR (LI-COR Biosciences). Antibody specificity was confirmed by the detection of only one band at the expected size. The intensity of the bands was then quantified, and corrected for loading using b-tubulin.

Organ Culture
Serum-free organ culture was performed as previously described [38]. P0 VSUs were placed in culture for 24 hr followed by treatment for 6 and 24 hr AE10 À8 M testosterone. P0 VPs were cultured for 6 days AE10 À8 M testosterone and AE3 mg/ml recombinant human Ptn protein (Peprotech, London, UK). Images of organ cultures were captured on a Leica MZ6 dissecting microscope (Leica, Deerfield, IL) with a Leica ICA camera. Organ perimeter and size was measured using NIH image software. Organs were incubated with 100 mg/ml BrdU (Sigma) for 2 hr prior to harvesting.

Cell Culture
Cells were seeded at 2 Â 10 3 cells/well in 96-well plates (for growth assays) or 6-well plates (for RNA preparation) and grown in DMEM supplemented with 2% or 10% charcoal-stripped fetal calf serum (CSFCS) and 1Â penicillin/streptomycin for 24 hr. For RNA, cells were treated AEtestosterone (10 À8 M) for 6 or 24 hr in duplicate. For growth assays, cells were treated with fresh medium AEtestosterone AErhPTN 50 ng/ml-1 mg/ ml for 4 days. To measure cell proliferation the Cell Titer 96 AQueous One Solution Cell Proliferation assay was used (Promega UK Ltd., Southampton, UK). The Cell Titer Solution was added directly to the cells in culture medium in a ratio of 1:6 and incubated at 378C for 100 min followed by absorbance reading at 490 nm.

PtnTranscript Expression in the Developing Prostate
Ptn was identified in the rat female VMP by LongSAGE transcriptional profiling with a ratio of 74:63 tags between the VMP/VSU [2]. The VSU consists of VMP, smooth muscle and urethra and transcripts enriched or specific to the VMP are diluted in the VSU relative to pure VMP. Using Northern blot and qRT-PCR, Ptn transcript levels were $2.6-fold (Northern) (P < 0.05) and 1.5-fold (qRT-PCR, data not shown) higher in the VMP than VSU. The temporal expression of Ptn mRNA in the prostate was studied by qRT-PCR (Fig. 1B). Ptn was most abundant during the neonatal (P0) and perinatal period (P4, P10) and expression subsequently decreased until adulthood. There was also evidence of higher Ptn levels in males versus females at e17.5 (P < 0.001). To examine the spatial distribution of Ptn in the male and female UGT, wholemount in situ hybridization was used to localize Ptn transcripts. Ptn mRNA was observed in the female in VMP and urethral stroma (Fig. 1C), and in the male in the prostatic mesenchyme (all lobes) and the urethral stroma (Fig. 1D). Additionally, to determine whether Ptn mRNA distribution was affected by testosterone, Ptn mRNA was localized in male VPs grown in culture AET (Fig. 1E,F). Organ growth was stimulated by testosterone, and Ptn was present in mesenchymal cells throughout the VP, and the distribution of Ptn was similar between organs grown with or without testosterone. There was no evidence of epithelial Ptn expression (insets, Fig. 1E,F) though our studies do not rule out low level epithelial expression or expression within epithelial subsets.

Ptn Protein Localization in the Male and Female
Urethra and Prostate Ptn protein was localized in the male and female urethra and prostate by immunohistochemistry, and showed a similar pattern to the in situ hybridization data. Ptn was observed in the VMP, periurethral mesenchyme and smooth muscle layer of the female ( Fig. 2A, arrows), and the mesenchyme, periurethral mesenchyme, and smooth muscle of the prostate (Fig. 2B). The strongest mesenchymal Ptn staining in the VP was at the periphery. Ptn was also localized to the ductal epithelia of the VP, where a gradient of Ptn expression was observed; Ptn was present at the tips of the ducts and was absent proximal to the urethra. Additionally, Ptn expression was examined in VP organs cultured AET (Fig. 2C,D) and Ptn distribution was similar to the VP in vivo (Fig. 2B); androgens did not alter Ptn distribution. To examine the precise localization of Ptn in VP epithelia, we co-localized Ptn with b-catenin, a cell surface marker (Fig. 2E,F). Ptn was expressed in the mesenchyme of the VP, and Ptn and b-catenin were co-expressed at the cell surface of ductal epithelial cells. The gradient of Ptn expression was confirmed with strong Ptn staining at the ductal tips and less staining proximal to the urethra. Additionally, co-localization of Ptn with p63 showed that Ptn was associated with both p63 positive and p63 negative cells ( Supplementary Fig. 1). Ptn distribution was partially inverse to the pattern associated with smooth muscle differentiation around epithelial ducts. We propose that, since Ptn mRNA was not observed in epithelia, Ptn at the epithelial surface was derived from the mesenchyme.

Ptn Is Involved in Prostatic Growth and Branching
The role of Ptn in prostate growth and morphogenesis was investigated using cultures of VP organs grown in vitro and treated with testosterone and recombinant hPTN protein (Fig. 3A). Addition of testosterone led to an increase in organ size and branching; however, addition of hPTN had relatively modest effects either in the presence or absence of testosterone (n ¼ 5 experiments, 45 organs). The two dimensional area of VP organs grown AET, AE3 mg/ml hPTN was measured using NIH imaging software The Prostate   Table A). Addition of hPTN to VP organs had no statistically significant effect upon organ area, in the presence (P ¼ 0.11) or absence (P ¼ 0.4) of testosterone. hPTN (Peprotech) was added to organ cultures at physiological levels ($0.94 mg/ml tissue volume) as used previously, to determine a functional response in vitro [39]. To investigate a role for hPTN in branching morphogenesis, the number of epithelial bud tips around the periphery of VPs cultured AET and AEhPTN were counted. The number of tips was expressed as a ratio to organ perimeter (tips per 1,000 pixels perimeter) to control for changes in organ size (Fig. 3C, and    an increase in the number of tips/1,000 pixels perimeter in the presence (P < 0.01) or absence (P < 0.001) of testosterone, and we suggest that Ptn has a role in increasing branching morphogenesis in the prostate.

Ptn Effects Upon Cellular Proliferation and Differentiation
To examine the effect of hPTN upon the cellular differentiation and proliferation of VPs grown in vitro (n ¼ 3, 27 organs) we examined the organs by histology and calculated proliferative rates using immunohistochemistry for BrdU incorporation and stromal and epithelial markers (Fig. 4). Histology of the VP organs showed no substantial changes in organ morphology or stromal and epithelial distribution (Fig. 4A). Analysis of stromal and epithelial differentiation using SM aactin and p63 showed no difference in cellular differentiation after treatment with hPTN, but confirmed increased branching of ductal tips at the periphery of the VP organs ( Supplementary Fig. 2). The proliferative index of the epithelial and mesenchymal compartments in the peripheral region (distal to the urethra) was calculated for VPs cultured AET, AEhPTN (Fig. 4B,C and values are listed in Supplementary Tables C and  D). hPTN significantly increased epithelial (P < 0.05) and stromal (P < 0.01) proliferation in the absence of testosterone.

PTNIncreases Proliferation of Embryonic and Cancer-Associated Fibroblasts, and BPH1Cells
The effect of hPTN upon embryonic human prostate fibroblasts and adult CAFs was measured by proliferation assay (MTT) (Fig. 5A and Supplementary Table E). Addition of hPTN (1 mg/ml) in the absence of testosterone and hPTN (200 ng/ml-1 mg/ml) in the presence of testosterone, led to a statistically significant (P < 0.01, one-way ANOVA) increase in the number of embryonic human prostate fibroblasts. We next examined the effect of Ptn upon the proliferation of adult prostate CAFs. Average values for cancer associated fibroblasts, derived from eight patients are shown in Figure 5B (and Supplementary Table F). Addition of recombinant hPTN protein (200 ng/ml-1 mg/ml) in the absence of testosterone or 1 mg/ml hPTN in the presence of testosterone led to a statistically significant increase in CAF cell number (P < 0.01) (one-way ANOVA). To investigate the effect of hPTN on epithelial cells we used BPH1 cells (ÀT, AEhPTN The Prostate Fig. 4. The effectofhPTNoncellularproliferationin theproximal and distal regions of the prostate. Rat VPs (P0) were grown AETand AEhPTN for 6 days and BrdU was added to the culture medium for 2 hr prior to harvesting.Panel A:Histology (hematoxylin andeosin staining) of cultured VPs after 6 days, asterisks (*) mark epithelial buds at the periphery of the organ. Scale bar represents 100 mm. Panel B: Immunohistochemistry of VP epithelial ducts distal to the urethra under different treatment conditions showing BrdU incorporation (green) in the epithelial cells (red); nuclei were stained withTopro (blue). Scale bar represents100 mm. Panel C: Graphs of BrdU incorporation AESEM illustrating the rates of proliferation of stroma and epithelia. Addition of hPTN led to anincreasein stromal and epithelial cell proliferation only in the absence of testosterone (Student's t-test, **P < 0.01). 50 ng/ml-1 mg/ml) (Fig. 5C and Supplementary  Table G). BPH1 cells were used as a model of nontumorigenic human prostate epithelia which do not express a functional androgen receptor, and thus testosterone was not added to the treatment groups. Addition of hPTN (1 mg/ml) to BPH1 cells led to a statistically significant increase in cell number (P < 0.01) (one-way ANOVA). Taken together, Ptn protein stimulated the growth of human embryonic fibroblasts, CAFs and BPH1 cells in the presence or absence of testosterone.

Differential Expression of Ptn mRNA Between Normal and Cancer-Associated Fibroblasts, and Males and Females
To examine if Ptn mRNA was differentially expressed between CAFs and NPFs, we compared Ptn transcript levels in eight pairs of functionally tested patient-matched NPF and CAF samples [7]. qRT-PCR analysis (Fig. 6A) determined that there was between 1.6-and 2.5-fold (P < 0.05-0.001) less Ptn expression in CAFs compared to NPFs in 5 of 8 samples. In samples 3 and 6, Ptn increased in CAFs versus NPFs and this indicates the heterogeneity inherent in patient CAF samples. CXCL12 was found to be increased in six of eight of the CAF samples, similar to findings in breast cancer stroma (data not shown) [40,41]. Next, we investigated whether Ptn exhibited sexually dimorphic expression in the developing human urethra and prostate. Ptn transcript expression was compared between the male and female human urethra (15-18 weeks) by qRT-PCR and Northern blot. qRT-PCR analysis of PTN mRNA expression demonstrated PTN transcript levels were 1.5-fold higher in the male compared to female (P < 0.001) (Fig. 6B). Northern blot analysis determined PTN transcript expression was $2-fold higher in the male compared to female (data not shown). Ptn mRNA (P < 0.001, Fig. 1B), and protein (1.3-fold, Fig. 6H) expression was also higher in e17.5 male versus female rat UGT, where there is little difference in tissue morphology. We have shown sexually dimorphic expression of Ptn in the developing UGT and as Ptn is expressed in the mesenchyme only of the rat and human prostate [15], it is not due to the presence of prostatic ducts in the male and their absence in female.

Ptn Expression Is Increased by Androgens and Androgen Receptor
To determine whether androgens were involved in the regulation of Ptn mRNA expression we examined its expression in feminized ARKO male mice that lack AR and which cannot respond to androgens. Ptn transcripts were measured by qRT-PCR in the UGT of ARKO males, wt males, and wt females (Fig. 6C). As controls, we included Nkx3.1 a gene expressed in epithelia and known to be androgen regulated, as well as Fgf10, a mesenchymally expressed gene unlikely to be regulated by androgens. Nkx3.1 mRNA levels, were substantially higher in wt males compared to ARKO males and females (P < 0.001, one-way ANOVA) The Prostate Fig. 5. The effect of hPTN on cultures of human prostate cancer fibroblasts and BPH1 cells. Panel A: Human embryonic prostate fibroblasts (EPFs) isolated from 15 to 18 weeks male UGT were grown AE testosterone and with increasing amounts of hPTN (501 ,000 ng/ml). Panel B: Human prostate CAFs isolated from adult maleTURP. Panel C: BPH1 cells.The proliferation of the cells was measured by MTTassay and the values at day 4 are presented and the measurement at 490 nm is relative to the cell number. Addition of hPTN (50 ng/ml^1 mg/ml) led to a significant increase in the cell numberofEPFs,CAFsandBPH1cellsafter4daysbothinthepresence or absence of testosterone.One-way ANOVAwithTUKEYmultiple comparison; *P < 0.05, **P < 0.01, ***P < 0.001. Fig. 6. Expression of PTN mRNAin CAFs versus NPFs, maleversus femaleurethra, andregulationby androgens.Panel A:The expression of PTN mRNA showed a decrease in CAFs compared to NPFs in five of eight CAF/NPF pairs, when quantified by qRT-PCR. Panel B: qRT-PCR analysis of PTN mRNA expression in human embryonic male and female urethra/UGT (15^18 weeks).The embryonic male UGT is exposed to significantlyhigherlevels ofcirculating testosterone, producedby the fetal testes, than the embryonic female UGT [48].PTN mRNAlevels were $1.5-foldhigherin the embryonicmaleversus femaleUGT.Panel C: QuantitativePCRforPtnmRNAin theUGTofmalemice,ARKOmalemice and female mice. ARKO mice lack AR and are a model for androgen action; they develop testes but lack secondary sex accessory tissues. Ptn mRNAwasless abundantinARKOmalesandwtfemalescomparedtowtmales.NKx3.1,anepithelialmarkerofprostateidentity whichisknown to be regulated by androgens, was included as a control and showed a marked reduction in level between wt males and ARKOs males or wt females.Fgf10, a ligand expressedin the stroma, showedno change betweenwtmales, ARKO males, or females, suggesting thatitis not androgen-regulated and that the mesenchymal composition between these samples is similar. Panel D: qRT-PCR analysis of rat VSUs grown AE testosterone for 6 and 24 hr; testosterone increased Ptn mRNA levels by twofold at 24 hr but not at 6 hr.Panel E: qRT-PCR for PTN mRNA in human embryonic prostate fibroblasts culturedin the absence of testosterone followedby 6 and 24 hr treatment with testosterone. PTN transcriptexpressionincreasedin thepresence of testosteroneat6 hr andwaspartiallyelevatedat24 hr.Panels F,G:Westernblotanalysis of Ptn (19 kDa) and b-tubulin (50 kDa); male and female rat UGTat e17.5, e19.5 and P0 (F), and VSU organs grown in the presence or absence of testosterone for 24 and 48 hr (G). b-tubulin was used as the loading control and for normalization. Panel H: Quantification of Western blot analysis in panels F and G, showing fold difference in Ptn expression inVSU organs grown ÀTversus þT 24 and 48 hr; testosteroneincreased Ptn levels at 24 and 48 hr anda differencein Ptnexpressionbetweenfemaleversusmalerat UGT (e17.5,19.05, and P0).Panels A,B,D,E, Student'st-test; *P < 0.05, **P < 0.01, ***P < 0.001.Panels C,H: One-way ANOVAwithTUKEYmultiple comparison; (C) ***,þþþ P < 0.001, (H) b^d P < 0.001. consistent with its epithelial expression, while Fgf10 mRNA levels were similar in all three groups. Ptn mRNA was 1.6-fold lower in ARKO males and wt females (P < 0.001, one-way ANOVA), compared to wt males, suggesting that androgens and AR increase Ptn transcript levels (Fig. 6C). Next, we examined whether Ptn mRNA levels were affected by testosterone using female VMP organ rudiments grown in vitro (AET). Female VMP was cultured for 24 hr followed by 6 or 24 hr of treatment with testosterone (Fig. 6D). VMPs (n ¼ 4, 40 organs) and not VPs were used for this experiment to minimize the potential carry over of residual testosterone in male VPs. No changes in gross tissue morphology of the stroma or epithelia were observed during treatment of VMPs for 6-24 hr in vitro. The results, expressed as fold difference in Ptn mRNA expression relative to ÀT were as follows:

The Prostate
Ptn transcript levels were upregulated by testosterone in cultures of rat female VMP. We next examined the effects of testosterone upon PTN mRNA levels in primary human embryonic prostate fibroblast cells (n ¼ 3), treated AET for 6 and 24 hr. The fold difference in PTN mRNA levels were: ÀT ¼ 1 versus þT 6 hr ¼ 1.52 AE 0.09 (Student's t-test, P ¼ 0.0082) and ÀT versus þT 24hr¼ 1.20 AE 0.08 (Student's t-test, P ¼ 0.046) (Fig. 6E). To determine if PTN mRNA expression in CAFs was androgen-regulated, PTN mRNA was measured in CAFs grown in the presence or absence of testosterone for 6 or 24 hr (n ¼ 3) by qRT-PCR and Northern blotting ( Supplementary Fig. 3). PTN transcript levels in CAFs showed very small changes in response to testosterone at any of the time points. We determined AR mRNA and protein expression in CAFs by RT-PCR and immunohistochemistry (data not shown), prior to performing the experiments. We note that throughout our studies Ptn transcript levels were detectable in females or the absence of testosterone, and that androgens and AR increased these levels. This suggests that there is a component of Ptn expression which is androgen independent.
To determine whether Ptn protein expression was regulated by androgens, we examined sexually dimorphic expression of Ptn and its response to androgens in vitro. We examined Ptn protein expression in developing male and female rat prostate/urethra (e17.5, e19.5, and P0) using Western blot analysis, and observed sexually dimorphic expression (Fig. 6F). Also, we measured Ptn in VSU organs grown AET for 24 or 48 hr using Western blot analysis (Fig. 6G) (n ¼ 3, 38 organs). Quantification of Ptn protein in VSU organs grown in the presence of testosterone was expressed as fold difference relative the absence of testosterone at each time point. The data were as follows: ÀT versus þT 24 hr ¼ 1.6 AE 0.1 (Student's t-test, P ¼ 0.002), ÀT versus þT 48 hr 2.1 AE 0.1 (Student's t-test, P ¼ 0.0003) (Fig. 6H). The data demonstrated that Ptn protein was upregulated by testosterone in cultures of rat female VMP. Quantification of Ptn protein in the prostate was expressed as fold difference relative to the female urethra at each time point. The results were as follows: e17.5 female versus e17.5 male ¼ 1.3 AE 0.1 (Student's t-test, P ¼ 0.057), e19.5 female versus e19.5 male ¼ 2.2 AE 0.3 (Student's t-test, P ¼ 0.014), P0 female versus P0 male ¼ 2.5 AE 0.2 (Student's t-test, P ¼ 0.007) (Fig. 6H). Ptn protein levels showed a sexually dimorphic expression, with higher Ptn expression in the male versus female.

DISCUSSION
We propose that Ptn signaling is involved in regulation of growth and branching morphogenesis in the developing prostate, and that Ptn expression is increased by androgens. We identified Ptn in a gene profiling study of VMP mesenchyme [2] which is a specialized area of mesenchyme that contains prostate inductive activity [42]. Given the importance of VMP mesenchyme in regulating prostate organogenesis, the identification of molecules that can mediate its effects is important and we propose that Ptn is one of many paracrine regulators made in VMP mesenchyme that regulate prostate growth. Ptn expression is associated with the development of organs that are regulated by epithelial-mesenchymal interactions and has also been described as a proto-oncogene. Ptn expression in the mesenchyme plays a role in the developing kidney, lung, and mammary gland [25,26,43], and has been localized to the stroma of adult prostate, and prostate cancer [15]. The role of Ptn in organogenesis has been difficult to examine given co-expression and redundancy of midkine; however, mice doubly null for Ptn and midkine show female infertility and reproductive abnormalities but no effects in males have been reported [28]. Were Ptn an essential mediator of androgen action, loss of Ptn might be expected to show impairment of masculinization similar to that seen in Tfm or ARKO mice, and such molecules remain to be identified.
We suggest that Ptn is a factor that regulates the proliferation of human and rodent prostatic stromal and epithelial cells in developmental and disease models. Given that transcripts for Ptn are stromal, it suggests that Ptn functions as an autocrine regulator of stroma and a paracrine regulator of epithelia. The paracrine function of Ptn is supported by the localization of Ptn on the surface of epithelial cells in growing ductal tips, where Ptn protein was abundant, in contrast to its absence from ducts proximal to the The Prostate urethra. It will be important to establish the expression patterns of the three receptors that bind Ptn and transduce its signal, as this will define Ptn responsive cells within the stromal and epithelial compartments, and these may include key subsets of cells such as progenitors. Co-localization of Ptn with b-catenin on the surface of epithelial cells is notable since b-catenin is a substrate of receptor protein tyrosine phosphatase RPTB/x [16], a well-studied Ptn receptor. Ptn is known to interact with the extracellular matrix of undifferentiated cells [23] and Ptn distribution patterns similar to our observations have been reported in the mouse genital tubercle, where Ptn is expressed in the mesenchyme and on the surface of epithelial cells [23]. Ptn is also expressed in the metanephric mesenchyme and on the basement membrane of the uteric bud at the onset of kidney development [26]. A gradient of Ptn is present in the uteric bud, similar to our findings in VP epithelia. Sakurai et al suggested Ptn may be acting as ''a classic morphogen'' of the mesenchyme that induced phenotypic changes in epithelia according to its concentration. We observed that Ptn had a small effect on branching morphogenesis at the periphery of the developing prostate in vitro, and established that it did not lead to mis-differentiation of stroma or epithelia when examined by immunohistochemistry for p63 or SMA ( Supplementary Fig. 2). Furthermore, addition of hPTN to VPs in culture led to a significant increase in the number of bud tips and an increase in the stromal and epithelial cell proliferative rate at the periphery of the organs in the absence of testosterone. We propose that Ptn functions as a classic morphogen in the developing prostate, since it is made by inductive mesenchyme of the VMP and shows similar activities in other organs such as kidney and genital tubercle [23,26].
Ptn has been shown to be mitogenic for a variety of fibroblast, epithelial and endothelial cell lines, including normal human prostate (PNT1A) and prostate cancer epithelial lines (DU145, PC3, and LNCaP) [15,30]. We examined the effect of hPTN on primary human prostate embryonic fibroblasts, cancer associated fibroblasts and BPH1 cells in vitro, and found that addition of hPTN led to a significant increase in cell proliferation. Ptn mRNA and protein are expressed in the stromal cells of normal adult prostate, prostate cancer and benign prostatic hyperplasia (BPH), and Ptn protein is also expressed in prostate tumor epithelia [15]. Surprisingly, when we examined PTN mRNA expression levels in matched pairs of CAFs/NPFs, five out of eight samples showed lower levels of PTN transcripts in CAFs compared to NPFs, and a recent study has confirmed a reduction of PTN in reactive prostate cancer stroma [44]. Breast cancer studies have reported conflicting data on PTN mRNA regulation in normal and cancer cells, reviewed in Ref. [36]. Different approaches to attenuate Ptn function in tumor cells were found to decrease tumor growth in vivo, thus constitutive Ptn signaling was a rate-limiting factor in the pathogenesis of these tumors.
In the developing prostate and normal adult prostate the effects of androgens upon epithelial cells are mediated via AR that is expressed in the mesenchyme or stroma, respectively [45,46]. However, there is little knowledge of the genes and pathways that mediate the effects of androgens on normal prostate growth or exactly how the paracrine effects of androgens are mediated. PTN mRNA exhibited a sexually dimorphic pattern of expression in the human fetal prostate/ urethra; PTN transcript levels were higher in the male, with elevated androgen levels, compared to the female. Similarly there were lower levels of Ptn mRNA in the urethra of female mice and ARKO males (that lack AR) when compared with wt male mice. Ptn protein also showed a sexually dimorphic difference between male and female rat urethra/prostate. Both Ptn mRNA and protein were upregulated by testosterone in organ cultures of the female VMP. We have shown that androgens increase Ptn levels in several different systems; however, Ptn expression is evident in females or the absence of androgens, which indicates there is an androgen independent element of Ptn expression. Gene profiling studies have confirmed the correlation between androgen levels and Ptn transcript levels in the prostate [47]. Such correlative data do not provide direct evidence for Ptn function as an ''andromedin,'' which would require direct experimental proof. Since prostate cancer is sensitive to androgens, we examined whether Ptn mRNA was affected by exposure to testosterone in three CAF isolates. Surprisingly, we found that T did not significantly alter Ptn mRNA levels in CAFs ( Supplementary Fig. 3). There are several possible explanations for this, however, we suggest that it may reflect dysregulation of androgen signaling in CAFs and contributes to their protumorigenic signaling.

CONCLUSION
Our data suggest that in the prostate Ptn functions as a regulator of both mesenchymal and epithelial proliferation, and that androgens regulate Ptn levels.