Androgen Regulation of Prostate Morphoregulatory Gene Expression: Fgf10-Dependent and -Independent Pathways

Abstract

Androgens are essential and sufficient for prostate gland morphogenesis; however, the downstream gene targets that mediate this action are unclear. To identify androgen-regulated genes involved in prostate development, we used short-term organ culture and examined the effect of testosterone on the expression of several critical prostate morphoregulatory genes. Rat ventral prostates (VP) and lateral prostates (LP) were collected at birth, and contralateral lobes were cultured for 18 h in the presence or absence of 10 nM testosterone with or without OH-flutamide to block residual androgens. Gene expression was quantitated using real-time RT-PCR. Although expression of Fgf10, Nkx3.1, and Ptc was increased in both prostate lobes, other genes were regulated by testosterone in a lobe-specific manner. This included up-regulation of epithelial genes FgfR2iiib, Shh, Hoxb13, and Bmp7 in the VP specifically and down-regulation of mesenchymal genes Wnt5a (VP) and Bmp4 (LP). Thus, in addition to stimulation of homeobox genes and paracrine-acting growth factors, androgens may positively regulate prostatic development through suppression of growth inhibitory genes. Because previous studies revealed a similar gene regulation pattern in response to exogenous Fgf10, experiments were performed to identify androgen-regulated genes mediated through Fgf10 signaling. Short-term VP and LP cultures with FgfR antagonist PD173074 and Mek inhibitor U0126 identified epithelial Shh and Hoxb13 up-regulation by androgens to be Fgf10-dependent. We propose that androgen regulation of prostate development is mediated through positive and negative regulation of multiple morphoregulatory genes acting in combination through complex gene networks. Lobe-specific responses may provide a developmental basis for prostate gland heterogeneity.

Androgens are both essential and sufficient for prostate development, where they stimulate ductal outgrowth, branching morphogenesis, cellular differentiation, and secretory function (1, 2). The evidence for the absolute necessity of androgen comes from the observation of prostatic absence in mice or humans with complete dysfunctional androgen receptors (AR) (3). It has also been shown that male offspring of pregnant rats treated with the AR antagonist flutamide from gestation d 12–21 possess malformed ventral prostates (VP) and agenesis of the dorsolateral lobe (4). Furthermore, prostatic budding can be induced in the female urogenital sinus exposed to androgens in organ culture (5), indicating that androgens are sufficient for initiation of prostate morphogenesis.

Despite this clear role for androgens, the direct androgen targets during prostate gland development are not well understood. It has been shown by classic tissue recombinant experiments that AR in the mesenchyme, and not epithelial AR, is responsible for prostatic morphogenesis (6). When wild-type murine urogenital sinus mesenchyme was recombined with AR-deficient murine urogenital sinus epithelium and grafted under the renal capsule, the AR-deficient epithelium underwent androgen-dependent ductal morphogenesis, epithelial proliferation, and columnar cytodifferentiation forming glandular epithelium that resembled normal prostate. On the contrary, when AR-deficient urogenital sinus mesenchyme was recombined with wild-type urogenital sinus epithelium, vaginal-like differentiation occurred. Although further analysis revealed that epithelial AR are required for expression of secretory proteins in mouse (7) and rat prostates (8), epithelial proliferation and cytodifferentiation appear to be driven by paracrine factors under mesenchymal AR control.

Searching for the androgen-regulated, mesenchymal-derived, paracrine-acting factors that mediate prostate morphogenesis has remained a challenge, although a few candidate genes have been proposed. Among these, members of the fibroblast growth factor (Fgf) family, particularly Fgf7 (also known as keratinocyte growth factor or Kgf) and Fgf10, have been proposed to be candidate stromal-to-epithelial cell “andromedins” based upon their synthesis and secretion by mesenchymal cells, receptors on epithelial cells, and stimulation by androgens in vitro (9, 10). However, subsequent studies indicated that these Fgf family members were not directly regulated by androgen in vivo (11, 12). Consequently, the role of Fgf7 and Fgf10 as potential stromal-to-epithelial andromedins in the prostate gland remains unresolved.

As has been shown for the murine prostate (13–15), recent work from our laboratory has demonstrated important morphoregulatory roles for several paracrine-acting, secreted molecules during rat prostate development including mesenchymal Fgf10 via its cognate epithelial receptor FgfR2iiib, mesenchymal Bmp4, epithelial Bmp7, and epithelial-produced Shh via its cognate mesenchymal receptor ptc and downstream gli1–3 transcription factors (16–18). In addition, epithelial homeobox genes, including Hoxb13 and Nkx3.1, have also been shown to play important developmental roles in the rat prostate (19, 20). Furthermore, tightly controlled feedback loops between morphogens were shown to coordinate prostate branching morphogenesis (17). In this regard, addition of exogenous Fgf10 to neonatal VP organ cultures up-regulated expression of Shh/Ptc, Bmp7, Nkx3.1, and Hoxb13 and down-regulated Bmp4, establishing Fgf10 as a proximate regulator of prostatic gene networks (17).

In the present study, expression changes of several known critical morphoregulatory genes were characterized in organ cultures of the developing rat prostate lobes as a function of testosterone treatment to directly assess potential downstream androgen targets and regulatory gene networks that direct prostate morphogenesis. The potential carryover of endogenous testosterone in control prostate lobes was taken into consideration and blocked with OH-flutamide. To further dissect the function of the Fgf10 signaling pathway in androgen regulation of prostatic morphoregulatory genes, a Mek inhibitor UO126 or an Fgf10 receptor antagonist PD173074 were added to the culture medium. By analysis and comparison of the testosterone effects in the presence and absence of Fgf10 signaling, lobe specificity of testosterone action in the separate prostate lobes and the role of Fgf10 as a candidate stromal-to-epithelial andromedin are proposed.

Materials and Methods

Animals

All rats were handled in accordance with the principles and procedures of the Guiding Principles for the Care and Use of Animal Research and the experiments were approved by the Institutional Animal Care Utilization Committee. Timed pregnant female Sprague Dawley rats were purchased from Zivic-Miller (Pittsburgh, PA), housed individually in a temperature (21 C)- and light (14 h light/10 h dark)-controlled room, and given standard Purina rat chow (Ralston-Purina, St. Louis, MO) and water ad libitum. They were monitored daily for delivery of pups, and the day of birth was designated as d 0.

Prostate organ culture

VP and lateral prostate (LP) were isolated on the day of birth by manual dissection under a microscope and immediately cultured on Millicell-CM filters (Millipore Corp., Bedford, MA) floating in 2 ml nutrient medium in BD Falcon six-well plates (BD Biosciences, San Jose, CA) as previously described (16, 17). The basic organ culture medium (BOCM) consisted of DMEM/F-12 (Invitrogen, Carlsbad, CA), 50 μg/ml gentamicin, and 1× insulin-transferrin-selenium (Invitrogen). The VPs and LPs were cultured in a humidified 5% CO₂ incubator for 18 h, which allowed sufficient time for steroid action on gene expression while avoiding shifts in the mesenchymal to epithelial ratio observed with growth during extended culture (Fig. 1). Culture was terminated at 18 h, and RNA was immediately isolated from the individual prostate lobes for subsequent use in real-time RT-PCR. Four separate sets of experiments were performed.

Fig. 1.

Newborn rat prostate lobes cultured with 10 μM OH-flutamide or 10 nM testosterone for 18 h. A, Images of contralateral VP lobes in culture at 0 and 18 h in the presence of OH-flutamide or testosterone with elongating and branching ducts outlined in red. Bar, 50 μM. B, Ductal area and total prostate area were calculated for six separate sets of VPs as described in A, and the ratios of ductal area to total area were compared between 0 and 18 h with OH-flutamide or testosterone. No differences were observed between time points and treatment groups. C, VPs immunohistochemically stained for p63 in basal-type epithelial cells at 0 and 18 h of culture with OH-flutamide or testosterone. At birth (0 h), several main prostatic ducts are visualized. After 18 h of culture in either OH-flutamide or testosterone, ductal elongation and branching was observed with no difference between the treatment groups. Bar, 200 μM. D, Ratio of CK19 to vimentin mRNA levels in contralateral VP or LP lobes cultured for 18 h in BOCM alone or with OH-flutamide (n = 5–6). E, Ratio of CK19 to vimentin mRNA levels in contralateral VP or LP lobes cultured for 18 h with OH-flutamide or testosterone (n = 10). The CK19 to vimentin ratio serves as a surrogate marker of the E:M cell ratio in the cultured tissues. No statistical differences were observed between the treatment groups for either prostate lobe.

Experiment 1

One VP and LP lobe from each animal (n = 8) were individually cultured for 18 h in basal organ culture medium (BOCM), whereas the contralateral lobe from each animal was individually cultured in BOCM with 10 nM testosterone (Sigma-Aldrich, St. Louis, MO).

Experiment 2

Circulating androgen levels are high in the neonatal rat during the first 12 h after birth (21), which may lead to a high background androgen level within the prostatic tissue. To block endogenous androgens retained in the prostatic cells and the viscous mesenchyme surrounding the rudimentary gland, experiments were next performed with OH-flutamide, a potent nonsteroidal AR antagonist (Schering Corp., Bloomfield, NJ). In initial studies, individual VPs and LPs were cultured in BOCM alone, whereas contralateral lobes from each animal were cultured in BOCM plus 10 μM OH-flutamide. Results revealed that OH-flutamide alone had limited effects on prostatic morphoregulatory gene expression. Next, individual VP and LP lobes from 10 neonatal male rats were separately cultured for 18 h in the complete absence of androgen action (BOCM plus 10 μM OH-flutamide), whereas the contralateral lobe from each animal was cultured in BOCM plus 10 nM testosterone.

Experiment 3

To determine whether prostatic Fgf10 mediates testosterone action on prostatic morphoregulatory gene expression, testosterone regulation of gene expression was next assessed in the absence of Fgf10 action. Because the Mek inhibitor UO126 was shown in our previous study to completely block Fgf10-induced budding and elongation in mesenchyme-free VP ductal cultures (17), contralateral VP and LP lobes (n = 10) were cultured in the absence (BOCM plus 10 μM OH-flutamide) and presence of androgens (BOCM plus 10 nM testosterone) as described for experiment 2, with the addition of 20 μM UO126 (Calbiochem, San Diego, CA) to both treatment groups to block the downstream Fgf10 signaling pathway.

Experiment 4

To address the possibility that other paracrine signaling factors may also function through the MAPK/Erk signaling pathway in the developing prostates, we performed studies with Fgf receptor (R1 and R2) antagonist PD173074 (gift from Pfizer, New York, NY) to specifically block Fgf10 action at the receptor level. Contralateral VP and LP lobes were cultured as described for experiment 3 except that 100 nM PD173074 was added to cultures in the absence and presence of androgen stimulation for 18 h.

Real-time RT-PCR

Real-time RT-PCR was performed using published sequences for primers and probes for Fgf10, FgfR2iiib, Shh, Ptc, Bmp7, Bmp4, Hoxb13, Nkx3.1, and Rpl19 (16, 17). Wnt5a expression was measured using forward primer: 5′-gagaaagggaacgaatccac-3′, reverse primer: 5′-catggcacttacaggctacatc-3′, and dual-labeled probe 5′-FAM-acaatgaagcaggtcgcaggacag-3′. Cytokeratin19 (CK19) expression was measured using forward primer: 5′-gctggcctacctgaagaaga-3′, and reverse primer: 5′-aatccacctccacactgacc-3′. Vimentin expression was measured using forward primer: 5′-aaattgcaggaggagatgct-3′and reverse primer: 5′-aggtcaagacgtgccagag-3′. The amplicon size of Wnt5a, CK19, and vimentin was 126, 86, and 98 bp. For all genes, primers were designed across exon boundaries to minimize the effect of potential genomic DNA contamination. The primers and dual-labeled probes were designed on the Primer3 website (http://www-genome.wi.mit.edu/cgi-bin/primer/primer3_www.cgi) and secondary structure was avoided when possible. The melting temperature (Tm) of primers was 58–60 C and the Tm of dual-labeled probes was 68–70 C. For Nkx3.1, Bmp7, CK19, and vimentin, a SYBR green assay with melting curve was used to ensure no nonspecific amplification. For all genes except CK19 and vimentin, plasmids containing each DNA sequence were cloned with TOPO TA cloning kit (Invitrogen) and used for standard curves, which were run in parallel for each reaction. The actual amount of target DNA in each experimental sample was directly calculated from each plasmid DNA standard curve. For CK19 and vimentin, cDNA from a pool of cultured d 6 rat VP total RNA was serial diluted and used for standard curve. Ribosomal protein L19 (Rpl19) was quantitated and served as an internal reference gene for normalization. Under these conditions, it was determined that as low as 100 copies in a single neonatal prostate lobe could be reliably measured with enough mRNA for assay of multiple genes from the same tissue.

Total RNA was extracted with RNeasy (Qiagen, Valencia, CA) and On-Column DNase I Digestion was performed. Total RNA was reverse transcribed with iScript cDNA Synthesis Kit (Bio-Rad, Hercules, CA) and the cDNA was 5× diluted and stored at −20 C. Quantitative real-time PCR was performed as previously described (16, 17) using an iCycler (Bio-Rad) and duplex wells with iQ Supermixture (Bio-Rad). The reactions contained 0.2 mM of each dNTP, 3 mM MgCl₂, 200 nM each primer, 200 nM dual-labeled probe, and 2 μl 5× diluted cDNA in a 25 μl PCR mixture. Cycle conditions were 95 C for 3 min, followed by 42 cycles of 95 C denaturation for 15 sec and 60 C annealing/extension for 30 sec. Optical data obtained by real-time PCR was analyzed by the manufacturer’s software (iCycle Optical System Interface Version 3.1). Each data point was repeated eight to 10 times with prostate tissue from different animals. The ratio of Amplicon cDNA copy number of a specific gene to Rpl19 was calculated, and statistical analysis was performed with two-tailed Student’s t test (Sigma Plot, version 8.02; SPSS Inc., Chicago, IL).

Immunohistochemistry

To histologically examine prostate lobe growth and branching after 18 h of organ culture, contralateral lobes were removed at birth and one lobe was frozen in Tissue-Tek O.C.T. compound (Sakura Finetek, USA, Inc., Torrance, CA), whereas the other was cultured in the presence of 10 nM testosterone or the presence of 10 μM OH-flutamide to block androgen action as described above. After 18 h culture, the lobes were mounted in OCT, frozen in liquefied propane, and stored in liquid nitrogen. Frozen sections of the contralateral 0- and 18-h lobes were mounted on a single glass slide and immunostained for p63, a basal cell marker, as previously described (22). In brief, sections were fixed in 2% paraformaldehyde, blocked with 2% serum and incubated overnight at 4 C with rabbit anti-p63 antibody (1:500, sc-8343; Santa Cruz Biotechnology, Inc., Santa Cruz, CA). The sections were reacted with biotinylated anti-IgG (Vector Laboratories, Inc., Burlingame, CA) and detected with avidin-biotin peroxidase (ABC-Elite; Vector Laboratories, Inc.) using diaminobenzidine tetrachloride as chromagen. For controls, normal rabbit IgG was substituted for primary antibody. The sections were counterstained with Gill’s #3 hematoxylin (1:4).

Results

Validation of short-term culture method for assessment of prostatic gene expression

To determine whether the short-term culture system permits accurate measurement of epithelial- and mesenchymal-specific gene expression during development without the influence of changing epithelial and mesenchymal cell fractions, a number of endpoints were assessed. First, photographs were taken of each VP culture at 0 h and again at culture termination after 18 h. The visualized ducts were outlined and ductal area was calculated using Zeiss Image (Fig. 1A). A similar outline of the entire organ was used to calculated organ area and the ratio of ductal to total prostate area was derived. Due to growth of both cell compartments during the 18 h culture period, there was no change in the total ductal area relative to total organ area over time (Fig. 1B). Furthermore, there was no change in the area ratios when VPs were cultured with or without testosterone or in the presence of OH-flutamide (Fig. 1B) due to similar growth and branching in the absence vs. presence of androgen action (Fig. 1A). Similar findings were observed for the cultured LPs. Next, a set of contralateral VP lobes were either frozen at 0 h or cultured for 18 h in the absence of androgen action (10 μM OH-flutamide) or presence of 10 nM testosterone followed by freezing. Immunohistochemistry staining for p63 of tissue sections from both times and treatment groups was performed to visualize the epithelial cell fraction and ductal branching patterns (Fig. 1C). At 0 h, we consistently observed four to five main ducts with sections through a few secondary branches, whereas after 18 h of culture, the contralateral lobes showed further elongation and tertiary branching. Importantly, there were no differences in this pattern or the epithelial cell fraction when comparing cultures without androgens (OH-flutamide) to culture with testosterone.

A final assessment was made by measuring the mRNA levels for CK19 and vimentin as surrogate markers of the epithelial and mesenchymal cell fractions. CK19 is expressed by all undifferentiated epithelial cells during early prostate development (23), whereas vimentin is expressed exclusively by cells of mesenchymal origin. To examine the action of OH-flutamide alone on the epithelial to mesenchymal (E:M) ratio, CK19 and vimentin expression was measured for VPs and LPs (n = 5–6) cultured in either BOCM alone (no testosterone) vs. with OH-flutamide (Fig. 1D). The CK19 to vimentin ratio was not significantly different in either lobe as a function of OH-flutamide treatment, indicating that no marked shifts in the E:M ratio occur during this short time frame. The ratios of CK19 to vimentin were next measured for the VPs and LPs (n = 10) cultured for 18 h in the absence of androgen action (OH-flutamide) vs. the presence of testosterone (Fig. 1E). In the VPs, the epithelial cell fraction relative to mesenchymal cells slightly decreased with testosterone, whereas this ratio modestly increased in the LP. Because neither was significantly different between treatments, we conclude that alterations in the E:M cell ratio due to the presence or absence of androgen action do not occur during the initial first day of prostate organ culture. Thus valid comparisons of gene expression changes between treatment groups can be made.

Testosterone regulation of morphoregulatory genes in separate rat prostate lobes

Short-term culture of contralateral neonatal VP and LP lobes with or without exogenous testosterone (experiment 1) resulted in minimal stimulation of morphoregulatory gene expression by testosterone. As shown in Fig. 2, there was no change in mRNA levels for Fgf10 or its cognate receptor FgfR2iiib, Shh or its cognate receptor Ptc, Bmp7 or Hoxb13 after 18 h of exposure to 10 nM testosterone. The only gene examined with significant stimulation in response to exogenous testosterone was Nkx3.1, which increased 59% and 83% in the VPs and LPs, respectively. Because previous studies have shown that prostatic Nkx3.1 is a direct androgen target gene (24), the present results confirm this regulation in the developing rat prostate and help to validate the experimental system. However, because Nkx3.1 is a nuclear transcription factor localized to prostate epithelium, it cannot be considered as a candidate mesenchymal mediator of androgen action in the developing prostate. A notable observation in experiment 1 was a significant decrease in the expression of Bmp4 (both lobes) and Wnt5a (P < 0.05 for VP only) in response to testosterone exposure. Because Bmp4 (14) and Wnt5a (25) are known secreted inhibitory factors produced by the developing prostate mesenchyme, these findings suggest that testosterone may drive prostate growth through the suppression of growth inhibitory genes.

Fig. 2.

Effect of exogenous testosterone on expression of prostatic morphoregulatory genes. Contralateral VP and LP lobes from postnatal d 0 rats were cultured in BOCM in the absence (open bar) or presence (solid bar) of 10 nM testosterone for 18 h, and gene expression was quantitated by real-time RT-PCR. Epithelial gene Nkx3.1 was up-regulated by testosterone in both lobes, whereas mesenchymal inhibitory genes Wnt5a (significant in VP only) and Bmp4 (both lobes) were down-regulated by testosterone. Other genes including Fgf10/FgfR2iiib, Shh/Ptc, Bmp7, and Hoxb13 showed no alteration by testosterone. Bars represent the mean ± SEM for eight samples per treatment. *, P < 0.05; **, P < 0.01; ***, P < 0.001; BOCM vs. BOCM plus testosterone.

To address the possibility that physiologically active levels of androgens were retained within the tissues cultured for 18 h in the absence of testosterone, which in turn, moderated the effects of exogenous androgens, studies were performed with OH-flutamide to block AR action. To first establish the basal effect of OH-flutamide, prostate lobes were cultured in BOCM without testosterone in the presence and absence of OH-flutamide for 18 h. No change in gene expression was observed in the VPs after short-term OH-flutamide exposure (data not shown), whereas in the LPs, androgen blockade significantly increased Wnt5a by 38% (P < 0.01), Hoxb13 by 40% (P < 0.05), and FgfR2iiib by 31% (P < 0.05). Next, comparisons were made between VP and LP lobes cultured in the absence of androgen action (BOCM plus OH-flutamide) and the presence of exogenous androgen (BOCM plus 10 nM testosterone). In addition to a significant increase in Nkx3.1 expression in both lobes, short-term androgen exposure increased Fgf10 expression in VP and LP lobes by 46 and 79%, respectively, suggesting that it is a mesenchymal target of androgen action in the developing prostate (Fig. 3). Furthermore, several other genes were moderately yet significantly stimulated by testosterone action specifically in the VP, including the epithelial Fgf10 receptor FgfR2iiib (42%), Shh (32%), Bmp7 (40%), Hoxb13 (32%) and the mesenchymal Ptc gene (30%). Although suppression of Bmp4 by testosterone was further enhanced in the LP, the suppression of Bmp4 and Wnt5a expression by testosterone in the VP observed in experiment 1 was lost when androgen blockade was used as a control in this set of experiments.

Fig. 3.

Effect of exogenous testosterone on prostatic morphoregulatory genes compared with absence of androgen action using AR antagonist, OH-flutamide. Con-tralateral VP and LP lobes from postnatal d 0 rats were cultured with 10 μM OH-flutamide (open bar) or 10 nM testosterone (solid bar) for 18 h. Mesenchymal Fgf10 and epithelial Nkx3.1 were up-regulated by testosterone in both lobes, whereas epithelial FgfR2iiib, Shh, Bmp7, Hoxb13, and mesenchymal ptc were up-regulated in the VP only. Mesenchymal Bmp4 was down-regulated in the VP specifically by testosterone. Bars represent the mean ± SEM for 10 samples per treatment. *, P < 0.05; **, P < 0.01; ***, P < 0.001, testosterone vs. OH-flutamide treatments.

To determine whether the aforementioned gene expression changes were a function of an E:M cell ratio shift after culture in the presence of testosterone or the absence of androgen action (i.e. with flutamide), epithelial-specific CK19 and mesenchymal-specific vimentin mRNA levels were measured by real-time RT-PCR in all samples. The calculated CK19 to vimentin ratio served as a surrogate E:M ratio for the different treatments. After 18 h of culture, there was no change in the E:M ratio in the VP lobes cultured in the presence of flutamide (0.53 ± 0.11; mean ± SEM) or testosterone (0.46 ± 0.11; P = 0.66), indicating that the observed increases in epithelial FgfR2iiib, Shh, Bmp7, Hoxb13, and Nkx3.1 gene expression with testosterone treatment are not a function of a relative increase in epithelial cell number. On the other hand, the modest increase in the relative mesenchymal fraction in the testosterone-treated VPs compared with the flutamide-treated VPs may account for the loss of the modest suppression of Wnt5a and Bmp4 previously observed in experiment 1. In the LPs, there was no statistical difference between the E:M ratio in organs cultured in the presence of flutamide (1.13 ± 0.12) or testosterone (1.49 ± 0.14; P = 0.065). Although there was a trend for relatively higher epithelial levels in the testosterone-treated group, it is noteworthy that LP gene expression changes were limited to increased mesenchymal Fgf10, increased epithelial Nkx3.1, and decreased mesenchymal Bmp4. Thus it is unlikely that these disparate responses are related to an increased E:M ratio.

Testosterone regulation under Fgf10 signaling blockage

Our previous studies implicated Fgf10 as an important regulator of growth factor networks and homeobox genes critical for prostate branching morphogenesis and differentiation (17). Exogenous Fgf10 protein added to VP cultures for 24 h stimulated expression of multiple genes including Bmp7, Nkx3.1, Hoxb13, Shh, and Ptc in a manner similar to the effects of exogenous testosterone observed in the present studies. To examine whether the testosterone effects on gene expression may be mediated through induction of Fgf10 expression, experiments 3 and 4 were performed, which blocked Fgf10 action at different sites in the presence and absence of testosterone. In the prostate as in other tissues, Fgf10 signaling is mediated through the ras/raf/Mek pathway (17). Addition of the Mek1/2 inhibitor UO126 to prostates cultured without (BOCM plus OH-flutamide) or with androgens (BOCM plus 10 nM testosterone) resulted in a loss of testosterone-induced increases in epithelial Shh, Bmp7, Nkx3.1, and Hoxb13 mRNA in the VP (Fig. 4), suggesting that testosterone stimulation of these genes may be mediated through paracrine Fgf10 action. In the absence of Mek activity, Shh expression was actually reduced (P < 0.05) by exogenous testosterone in the VP. In contrast, in the absence of Mek1/2 action, up-regulation of mesenchymal Ptc expression by testosterone continued in the VP and LP, suggesting Fgf10-independent regulation by androgens. Furthermore, some lobe specificity in androgen action was observed because Nkx3.1 remained stimulated and Bmp4 remained suppressed by androgens in the LP alone, whereas Wnt5a expression remained suppressed in the VP alone in the absence of Mek1/2 action.

Fig. 4.

Effect of exogenous testosterone on prostatic morphoregulatory genes with concomitant blockade of Fgf10 signaling with Mek inhibitor UO126. Contralateral VP and LP lobes from postnatal d 0 rat were cultured with 10 μM OH-flutamide + 20 μM UO126 (open bar) or 10 nM testosterone + 20 μM UO126 (solid bar) for 18 h. See Results for description. Bars represent the mean ± SEM for eight samples per treatment. *, P < 0.05; **, P < 0.01; OH-flutamide + UO126 vs. testosterone + UO126.

Because other signaling pathways may also converge on Mek1/2 within the developing prostate, Fgf10 action was alternatively blocked at the receptor level using the specific FgfR1/R2 antagonist PD173074 in the presence and absence of testosterone stimulation. The responses to androgens were similar to those observed after Mek inhibition for several genes (Fig. 5). Epithelial Shh and Hoxb13 stimulation by testosterone in the developing VP was blocked by the FgfR antagonist indicating that androgen regulation of these genes is indirectly mediated through paracrine Fgf10 action. Similarly, Ptc up-regulation and Wnt5a down-regulation in the VP by testosterone continued in the absence of Fgf10 action, further indicating that androgen regulation of these mesenchymal genes is independent of Fgf10. In contrast to the findings observed with the Mek inhibitor, testosterone up-regulation of epithelial genes Bmp7 (in the VP) and Nkx3.1 in both lobes continued in the absence of FgfR action. The differential results using an FgfR inhibitor and a Mek inhibitor indicate that testosterone regulation of Bmp7 and Nkx3.1 in the VP may be indirectly mediated through an alternate growth factor that acts through Mek activation.

Fig. 5.

Effect of exogenous testosterone on prostatic morphoregulatory genes with concomitant blockade of Fgf10 receptor with antagonist PD173074. Contralateral VP and LP lobes from postnatal d 0 rat were cultured with 10 μM OH-flutamide + 100 nM PD173074 (open bar) or 10 nM testosterone + 100 nM PD173074 (solid bar) for 18 h. Gene expression was quantitated with real-time RT-PCR. See Results for description. Bars represent the mean ± SEM for nine samples per treatment. *, P < 0.05; **, P < 0.01; ***, P < 0.001; OH-flutamide + PD173074 vs. testosterone + PD173074.

Discussion

It is accepted that androgens are essential for prostate development and that morphogenesis is driven by androgen-dependent, mesenchymal-derived, paracrine-actingfactors. Despitethis, identification of these androgen targets has been elusive. Through a series of carefully controlled organ culture experiments, the present study provides clear evidence that androgens regulate the expression of several prostatic morphoregulatory genes including up-regulation of Fgf10 expression and signaling in the developing rat VP and LP lobes. The present findings also show that epithelial expression of Shh and Hoxb13 is indirectly up-regulated by androgens through Fgf10, thus establishing a stromal-to-epithelial paracrine role for Fgf10 in testosterone action. Because other epithelial genes (Nkx3.1 and Bmp7) were identified as androgen targets through Fgf10-independent pathways, additional mesenchymal-derived paracrine factors must be involved in mediating androgen action during prostate development. The present study further identified additional mesenchymal genes as androgen targets. Interestingly, testosterone down-regulated expression of two genes encoding secreted mesenchymal factors involved in growth suppression during prostate development, namely Bmp4 and Wnt5a. Thus in addition to stimulation of paracrine-acting growth factors, androgens may positively regulate prostatic development through suppression of growth inhibitory genes. Based on the present findings, a model for the lobe-specific androgenic regulation of prostatic genes is presented in Fig. 6.

Fig. 6.

A schematic representation of differential testosterone action on morphoregulatory gene expression in the rat VP and LP lobes. In this model, testosterone might be considered a proximate regulator of a cascade of both growth stimulatory and inhibitory genes. Genes in rectangles represent indirect testosterone targets through the Fgf10 signaling pathway. Genes in circles represent potential direct targets of testosterone in stroma or epithelial testosterone target genes that are independent of Fgf10 signaling. Pointers indicate similar responses in the VP and LP lobes. Genes shown in red are growth inhibitory genes. Exogenous testosterone up-regulates Fgf10/FgfR in both prostate lobes. In the VP, elevated Fgf10 signaling drives increased expression of epithelial Shh and Hoxb13. Testosterone also increases Nkx3.1 and Bmp7 in the VP through a Mek pathway that is independent of Fgf10. In contrast, testosterone increases Nkx3.1 expression in the LP through an Fgf10 and Mek-independent pathway and has no effect on epithelial Hoxb13, Bmp7, or Shh in that lobe. Mesenchymal expression of Ptc is directly increased by androgens in both lobes. Of the two mesenchymal inhibitory factors examined, Wnt5a expression is down-regulated by testosterone in the VP, whereas Bmp4 is down-regulated by androgen in the LP. We propose that androgen regulation of prostate development is mediated through positive and negative regulation of multiple genes acting in combination and through complex gene networks that include cross-regulation of the genes themselves in a time- and location-specific manner.

In the present study, testosterone was used in the organ cultures to mimic the physiologic condition where prostate cells take up circulating testosterone. It is well established that high intracellular levels of 5α-reductase within the prostate cells rapidly convert testosterone to dihydrotestosterone, the active androgen in the prostate (26). Because testosterone can also be aromatized to estradiol, it deserves mention that some of the downstream effects may also be mediated through estrogen action. However, it is noteworthy that more than 90% of testosterone is irreversibly metabolized to dihydrotestosterone in the prostate (27), and although aromatase has been reported in human prostates (28), this has not been confirmed in the rat prostate. Furthermore, at the time of birth, prostatic estrogen receptors (ER) are limited to ERα expressed in proximal duct mesenchymal cells (29), whereas epithelial ERβ protein is not detectable until postnatal d 5–10 (22, 30). Thus it is likely that most if not all of the actions of testosterone are androgenic in the present set of experiments and that contributions of estrogen action would be minor at best.

Androgens stimulate Fgf10/FgfR2iiib expression in the developing rat prostate

Fgf10 is expressed by mesenchymal cells and has been previously identified as a critical morphogen during prostate development (10, 12, 17, 31). Initial studies showed that Fgf10 expression was strongly up-regulated by testosterone in stromal cells in vitro (10), and because its cognate receptor, FgfR2iiib, is exclusively expressed by epithelial cells (17), Fgf10 was proposed as a key androgen-driven, paracrine-acting growth factor for prostate development. This concept was challenged by organ culture studies where Fgf10 transcript was increased by 1.5-fold in the rat VP after 4 d exposure to testosterone, but was not decreased after 24 h of cyproterone acetate treatment (12). Thus Thomson and Cuhna (12) concluded that Fgf10 was not androgen regulated in vivo and, although essential for morphogenesis, is not the elusive andromedin. In the present study, several potential confounding variables in the organ culture experiments were taken into account: 1) contralateral VPs and LPs were directly compared with decrease interanimal variability, 2) cultures of newborn prostates were limited to 18 h to avoid marked shifts in E:M cell ratios that may complicate data interpretation, and 3) residual androgens in the control cultures were blocked by OH-flutamide. Under these conditions, a consistent and significant up-regulation of Fgf10 expression by testosterone was observed in both the VP and LP lobes, indicating that Fgf10 is in fact an androgen-regulated gene. However, as previously noted (12), testosterone is not required for Fgf10 expression because it continues in the absence of androgen, albeit at lower levels, and after short-term exposure to the pure AR antagonist OH-flutamide. Thus rather than regulating an all-or-none synthesis of Fgf10, these findings indicate that androgens drive elevated expression levels of this critical growth factor during development. In addition, androgens increased FgfR2iiib expression in the VP, thereby increasing epithelial cell responsiveness to this secreted mesenchymal morphogen. Because our previous studies showed that FgfR2iiib is not autoregulated by Fgf10 (17), this androgenic regulation must be mediated through an Fgf10-indepenedent pathway. Together, we propose that androgen regulation of prostate development is mediated through positive and negative regulation of multiple genes acting in combination and through complex gene networks that include cross-regulation of the genes themselves in a time- and location-specific manner. In this model, testosterone might be considered a proximate regulator of a cascade of both growth stimulatory and inhibitory genes.

Testosterone stimulates epithelial Shh and Hoxb13 expression in the VP through an Fgf10-dependent pathway

Shh, an epithelial secreted glycoprotein, is an essential morphogen during prostatic development (16, 32–34). Our laboratory has recently demonstrated cross-regulation between prostatic Fgf10 and Shh where Fgf10/FgfR up-regulates Shh expression and Shh/Ptc down-regulates Fgf10 signaling (16, 17). Hoxb13, a homeobox gene expressed in prostate epithelium, is essential for epithelial cell differentiation (35, 36) and is up-regulated by Fgf10 in the rat VP (17). In the present study, testosterone stimulated a moderate increase in Shh and Hoxb13 expression in rat VP lobes. Additional studies with Mek inhibition and FgfR antagonism demonstrated that testosterone regulation of Shh and Hoxb13 were mediated through Fgf10 signaling, which identifies a specific molecular effect of elevated Fgf10 expression as a function of androgen exposure. In contrast to VP effects, Shh and Hoxb13 were not altered by testosterone in the LP lobe in either the presence or absence of Fgf10 signaling. However, previous studies showed that alterations in Fgf10 signaling are the proximate cause of Shh/ptc and Bmp7 down-regulation after estrogen exposure in the dorsolateral prostate (17). This suggests that, although Fgf10 is a proximate regulator of steroid action in the prostate, specific steroidal responses may differ between the prostatic lobes.

Fgf10-independent targets of testosterone in the prostatic mesenchyme and epithelium

Results from the present set of experiments identified additional prostatic genes as androgen targets that are independent of Fgf10 signaling. Nkx3.1, an NK family homeobox gene expressed in prostate epithelium (37), plays a role in epithelial cell proliferation and differentiation from early development through adulthood. Previous studies have shown direct transcriptional regulation of Nkx3.1 by androgens via the AR in vitro (24, 38). However, in the rodent, Nkx3.1 is strongly expressed in the fetal urogenital sinus and initial prostate buds before epithelial expression of AR, thus developmental regulation of this gene is not fully understood. Our prior studies have shown that Nkx3.1 is suppressed by estrogens (20) and up-regulated by Fgf10 during early postnatal life (17). The present studies demonstrate that Nkx3.1 is also strongly up-regulated by androgens during the first 24 h of life. Although the results using an FgfR antagonist showed that this was Fgf-independent, loss of androgen regulation with addition of a Mek inhibitor in the VP specifically suggests that a Mek-dependent pathway may mediate androgen up-regulation of Nkx3.1 in that lobe. Because prostatic epithelial cells begin to express low levels of AR shortly after birth (8), direct regulation of Nkx3.1 through epithelial AR is also possible at this time.

Bone morphogenetic protein 7 (Bmp7), a member of the Tgfβ superfamily, is expressed in the adult rodent prostate and is androgen dependent (39). Studies with Bmp7 null mice suggest an inhibitory role for this secreted protein during prostate branching morphogenesis (15). Although expression was low at birth, we noted marked epithelial induction by d 3 (17) with up-regulation by exogenous Fgf10 in short-term VP organ cultures (17). In the present study, Bmp7 transcripts were elevated after testosterone exposure in the VP, whereas expression in the LP was androgen-independent. However, the VP testosterone-driven expression was not mediated through elevated Fgf10 because the specific FgfR antagonist did not block the effect. However, the androgen-stimulated expression was blocked by a Mek inhibitor, which suggests that, similar to Nkx3.1, Bmp7 regulation by androgens may be mediated through alternate paracrine factors that converge on the Mek signaling pathway in the VP specifically. Known prostate factors that may signal through Mek include Egf (40), Tgfβ (41), and Sprouty (42) among others.

In addition to epithelial genes, expression of several mesenchymal cell genes in the neonatal prostate was influenced by short-term testosterone exposure. Because AR is expressed at high levels in most prostatic mesenchymal cells at the early stages of development (8), these genes may be direct androgen targets. Ptc, the transmembrane Shh receptor, localizes to the distal periductal mesenchymal cells in the developing rat prostate gland (16). In the present study, ptc was markedly up-regulated by androgens in the absence of Fgf10 signaling in both the VP and LP lobes. Although ptc expression is tightly regulated by Shh (32), it is noteworthy that the increased ptc expression in experiments 3 and 4 of the present study was not correlated with increased Shh expression, suggesting that androgens may directly increase ptc expression. This presents a novel pathway whereby androgens may contribute to prostate morphogenesis because elevated ptc levels will result in increased responsiveness to Shh known to be essential for prostate morphogenesis.

Another interesting observation in the present studies was down-regulation of Bmp4 and Wnt5a in the prostate lobes after androgen exposure in an Fgf10-independent manner. Bmp4 is broadly expressed by rat prostate mesenchymal cells during early development (18) and restricts prostate ductal budding and branching morphogenesis in the mouse (14). Furthermore, our previous studies demonstrated an inhibitory role for Bmp4 on gene networks during rat prostate development (16, 17). Wnt5a is a noncanonical Wnt gene expressed in mesenchyme of several branched structures during development and functions as a negative regulator of Shh and Fgf10 signaling during pulmonary branching (43, 44). We recently demonstrated an inhibitory role for mesenchymal Wnt5a during rat prostate morphogenesis (25). Thus direct down-regulation of these two inhibitory genes by androgens acting through mesenchymal AR will contribute to the global promotion of prostate morphogenesis by testosterone.

Lobe-specific response of morphoregulatory genes to testosterone

It has been previously demonstrated that the separate lobes of the rat prostate gland exhibit differential sensitivity to androgens in terms of growth, with the ventral lobe showing greater androgen responsiveness than the dorsal or lateral lobes (45–48). Additionally, AR expression is tightly regulated by androgens in the VP but is androgen-independent in the LP (49, 50). The present findings extend this lobe-specificity for androgen regulation to the expression of morphoregulatory genes (Fig. 6). Although Fgf10/FgfR2iiib, stromal Ptc, and epithelial Nkx3.1 were up-regulated by testosterone in both VP and LP lobes, several other epithelial genes were androgen up-regulated in the VP alone. Of these, Shh and Hoxb13 demonstrated dependency on the intact Fgf10 signaling pathway for androgen regulation, whereas Bmp7 and Nkx3.1 were dependent on Mek signaling through other up-stream factors. Notably, there was no evidence for Fgf10 or other Mek signaling factors acting as intermediaries for androgens in the LP. Of further note was the down-regulation of two inhibitory genes in mesenchyme by testosterone, with Wnt5a suppressed only in the VP and Bmp4 primarily in the LP. Because testosterone is essential and sufficient for prostate morphogenesis, the differential regulation of these developmental genes by testosterone may participate in the determination of tissue heterogeneity in the separate regions of the prostate gland.

In conclusion, testosterone regulates the expression of multiple morphoregulatory genes in the developing prostate gland in a lobe-specific manner as summarized schematically in Fig. 6. The present study suggests that Fgf10 is a critical and possibly direct target of testosterone action in the rat VP and LP lobes. Fgf10 up-regulates some VP epithelial genes in response to testosterone stimulation and as such, may be considered a stromal-to-epithelial andromedin in that region specifically. We postulate that testosterone regulates prostate development in the separate lobes though overlapping and distinct sets of regulatory morphogens, which forms a developmental basis for prostate gland heterogeneity.

Abbreviations

AR

Androgen receptor

BOCM

basic organ culture medium

CK19

cytokeratin19

E:M

epithelial to mesenchymal

ER

estrogen receptor

LP

lateral prostate

VP

ventral prostate