Wnt Inhibitory Factor 1 (Wif1) Is Regulated by Androgens and Enhances Androgen-Dependent Prostate Development

Abstract

Fetal prostate development from urogenital sinus (UGS) epithelium requires androgen receptor (AR) activation in UGS mesenchyme (UGM). Despite growing awareness of sexually dimorphic gene expression in the UGS, we are still limited in our knowledge of androgen-responsive genes in UGM that initiate prostate ductal development. We found that WNT inhibitory factor 1 (Wif1) mRNA is more abundant in male vs. female mouse UGM in which its expression temporally and spatially overlaps androgen-responsive steroid 5α-reductase 2 (Srd5a2). Wif1 mRNA is also present in prostatic buds during their elongation and branching morphogenesis. Androgens are necessary and sufficient for Wif1 expression in mouse UGS explant mesenchyme, and testicular androgens remain necessary for normal Wif1 expression in adult mouse prostate stroma. WIF1 contributes functionally to prostatic bud formation. In the presence of androgens, exogenous WIF1 protein increases prostatic bud number and UGS basal epithelial cell proliferation without noticeably altering the pattern of WNT/β-catenin-responsive Axin2 or lymphoid enhancer binding factor 1 (Lef1) mRNA. Wif1 mutant male UGSs exhibit increased (Sfrp)2 and (Sfrp)3 expression and form the same number of prostatic buds as the wild-type control males. Collectively our results reveal Wif1 as one of the few known androgen-responsive genes in the fetal mouse UGM and support the hypothesis that androgen-dependent Wif1 expression is linked to the mechanism of androgen-induced prostatic bud formation.

Androgens initiate prostatic bud formation from the urogenital sinus (UGS) and stimulate prostatic bud elongation, ductal branching morphogenesis, and differentiation of mature prostatic ductal epithelium (1, 2). During fetal prostate development, androgens activate androgen receptors (ARs) in UGS mesenchyme (UGM) to induce prostatic bud formation in UGS epithelium (UGE) (1, 3–5). The developing prostate has therefore been used as a model to assess the role of androgens in mesenchymal/epithelial interactions. A longstanding question in the prostate development field is how ARs in UGM communicate with UGE to establish the pattern and quantity of prostatic buds that will form. Continued investigation is expected to shed light on how the developing prostate microenvironment influences prostate epithelial cell fate. It may also elucidate how adult prostate stromal ARs reactivate developmental signaling pathways to cause inappropriate proliferative growth during prostate disease (6–13).

Numerous gene expression profiling studies have been conducted to identify androgen-responsive mRNAs in fetal prostate (14–18). The WNT pathway has been identified as a major androgen-responsive pathway in developing UGS and many Wnts are more abundant in male compared with female mouse UGS (14, 15, 19). Several WNT antagonists are also present during prostate development, including but not limited to WNT inhibitory factor 1 (Wif1) and secreted frizzled related proteins (Sfrps) 1 and -2 (14, 15, 20). Tight regulation of WNT signaling is required for the normal program of prostatic branching morphogenesis (20, 21).

Despite growing awareness of sexually dimorphic gene expression in the UGS, we are still limited in our knowledge of androgen responsive genes in UGM that initiate prostate ductal development. We conducted a high-throughput in situ hybridization (ISH) screen in conjunction with the GenitoUrinary Development Molecular Anatomy Project (GUDMAP) to resolve developing prostate molecular anatomy and identify candidate androgen-responsive mRNAs in fetal mouse UGM (www.gudmap.org). Because mesenchymal AR signaling is necessary for prostate development, we screened for mRNAs present in fetal male UGM during prostatic bud formation at embryonic day (E) 17.5 and absent, less abundant, or differently patterned in E17.5 female UGM. Our screen revealed a sexually dimorphic expression pattern for the secreted WNT-binding protein Wif1, which was more abundant in male compared with female UGM. In this manuscript, we tested hypotheses that androgens are necessary and sufficient for Wif1 mRNA expression in developing and sexually mature mouse prostate and that exogenous WIF1 protein increases androgen-dependent prostatic bud formation in UGS explant cultures. Our results reveal Wif1 as one of the first androgen induced enhancers of prostatic bud formation and thereby provide new insight into the mechanism of androgen action during mouse prostate development.

Materials and Methods

Animals

Wild-type C57BL/6J mice were purchased from Jackson Laboratory (Bar Harbor, ME). Wif1^tm1Dmth targeted mutant mice (Wif1^lacZ) (22) were from Dr. Igor Dawid (Eunice Kennedy Shriver National Institute of Child Health and Human Development, Bethesda, MD) and were maintained on a C57BL/6J × 129/SV background. Mice were housed in polysulfone cages containing corn cob bedding and maintained on a 12-h light, 12-h dark cycle at 25 ± 5 C and 20–50% relative humidity. Feed (Diet 2019 for males and Diet 7002 for pregnant females; Harlan Teklad, Madison, WI) and water were available ad libitum. All procedures were approved by our institutional Animal Care and Use Committee and conducted in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals. To obtain timed-pregnant dams, females were paired overnight with males. The next morning was considered E0.5. Dams were euthanized by CO₂ asphyxiation. Wif1 embryos were genotyped with PCR primers for wild-type allele (5′-CGAGAACTTCACAAGCAGCACAGG-3′, and 5′-CCTGTTACAAATCTGCAGTCAGG-3′, 500 bp) and mutant allele (5′-CTGTGGCCGGCTGGGTGTGGCGG-3′ and 5′-AGCACTCTAGCCTGATGGGCTC-3′, 500 bp). For in vivo experiments, wild-type timed pregnant dams were injected on E13.5 and E14.5 or E17.5 and E18.5 with sterile corn oil (5 ml/kg sc maternal dose) containing 10% ethanol alone or containing flutamide (200 mg/kg · d maternal dose; no. F9397-1G; Sigma, St. Louis, MO). UGSs were collected at E15.5 and postnatal day (P) 0.

Assay for β-galactosidase activity

LacZ-dependent β-galactosidase activity was assessed as described previously (19).

Castration and testosterone treatment

Forty-eight- to fifty-d-old adult male wild-type mice were given buprenorphine (0.1 mg/kg) for analgesia and isoflurane (1–2% inhalant) for anesthesia. A scrotal incision was made, the testes and associated fat pads were externalized, ligated, and excised, and the incision was closed with a suture. Sham castrations were performed as described above except testes and fat pads were not ligated or removed. Some mice received 1 cm SILASTIC brand (Dow Corning, Midland MI) capsule implants (empty or testosterone filled), which were engineered in our laboratory from SILASTIC brand tubing (1.57 mm inside diameter, 3.18 mm outside diameter,) and inserted into the sc space above the scapular fat pad.

In situ hybridization

To allow for ISH comparisons between males and females and among treatment or genotypic groups, tissues from each comparison group were dissected, fixed, dehydrated, and stored under identical conditions. Tissues were processed for ISH at the same time and in the same vessel so qualitative comparisons could be made among or between groups. UGSs were fixed overnight in 4% paraformaldehyde and stained in whole mount, cut with a vibrating microtome into 50 μm sections, or frozen in optimal cutting temperature (O.C.T.) medium and cut with a cryotome into 10-μm sections. Fresh adult prostate was frozen in O.C.T. medium and cut with a cryotome into 10-μm sections. PCR-based riboprobe synthesis (www.gudmap.org) and ISH on vibrating microtome-cut sections and UGS whole mounts are described elsewhere (23, 24) (www.gudmap.org). Primer sequences for generating PCR-based riboprobes are listed in Supplemental Table 1, published on The Endocrine Society's Journals Online web site at http://endo.endojournals.org. For ISH of frozen tissue sections, tissues were embedded in O.C.T. medium, frozen, sectioned, air dried for 10 min, and fixed for 1 h in PBS containing 4% paraformaldehyde. ISH was conducted as described previously (23) with the following modifications. Washes were conducted in slide mailer tubes with constant agitation. Tissues were permeabilized for 8 min at room temperature with 5 μg/ml Proteinase K. Slides were coverslipped and placed in a humidified hybridization chamber for prehybridization, probe, and antibody solution incubations. The hybridization solution was supplemented with 5% dextran sulfate. The duration of all previously described washes (23) was reduced by half. BM Purple alkaline phosphatase precipitating chromagen (Roche Diagnostics, Indianapolis, IN) was used for detection of digoxigenin-labeled riboprobes. Incubation times were identical for all samples within a comparison group and staining times for each riboprobe are indicated in figure legends. For some UGS explants, light micrographs of BM Purple-stained tissue sections were captured, inverted, pseudocolored green, and merged with immunofluorescent images. The staining pattern for all the riboprobes was assessed in at least two tissue sections from at least two litter-independent mice. Gene expression patterns were annotated using the GUDMAP anatomical ontology.

Organ culture

E14.5 wild-type male and female UGSs were placed on 0.4-μm Millicell-CM filters (Millipore, Billerica, MA) and cultured for 1 or 4 d as described previously (25). Media were supplemented with 5α-dihydrotestosterone (DHT; 10 nm), hydroxyflutamide (OHF; 10 nm) (Sigma-Aldrich, St. Louis, MO) and/or recombinant human WIF1 protein (0–2000 ng/ml; R&D Systems, Minneapolis, MN). Media and supplements were changed every 2 d.

Immunohistochemistry (IHC)

Immunofluorescent staining of ISH-stained, frozen, paraffin, and vibrating microtome sections was performed as described previously (19, 26). Primary antibodies were diluted as follows: 1:200 rabbit anticadherin 1 (CDH1; 3195, Cell Signaling Technology, Beverly, MA), 1:250 mouse anti-smooth muscle actin-α2 (ACTA2; Leica Microsystems, Buffalo Grove, IL), 1:250 rabbit antiphospho-histone H3 (PH3; 9701s; Cell Signaling Technology) and 1:50 mouse anti-keratin 14 (KRT14; ms-115-p0; Thermo Scientific, Waltham, MA). Secondary antibodies were diluted as follows: 1:250 Dylight 549-conjugated antirabbit IgG (111-507-003; Jackson ImmunoResearch Laboratories, West Grove, PA) and Dylight 488-conjugated goat antimouse IgG (115-487-003; Jackson ImmunoResearch Laboratories). Immunofluorescently labeled tissues were counterstained with 4′,6-diamidino-2-phenylindole, dilactate (DAPI) and mounted in antifade media (PBS containing 80% glycerol and 0.2% n-propyl gallate).

RNA isolation and real-time RT-PCR

E17.5 UGE was enzymatically and mechanically separated from UGM and homogenized as described previously (25). RNA was purified with the Illustra RNAspin minikit (GE Healthcare, Pittsburgh, PA) and reverse transcribed with the SuperScript III first-strand synthesis system (Invitrogen, Carlsbad, CA). Real-time PCR was performed in 10.5-μl reactions containing 1× SsoFast EvaGreen Supermix (Bio-Rad Laboratories, Hercules, CA), 0.48 μm PCR primers, and 3.75 μl cDNA and amplified using the CFX96 PCR machine (Bio-Rad Laboratories). PCR primers were as follows: Wif1, 5′-GAG AAA GCC CTG TGC ATA CC-3′ and 5′-ACT GCT CTC TCC CTC GAG TCC-3′; Krt14, 5′-AAT TCT CCT CTG GCT CTC AGT CAT CC-3′ and 5′-AGC TTT AGT TCT TGG TGC GCA GGA C-3′; Acta2, 5′-CTG CCG AGC GTG AGA TTG-3′ and 5′-AAT GAA AGA TGG CTG GAA GAG AG-3′; and peptidyl prolyl isomerase a (Ppia), 5′-TCT CTC CGT AGA TGG ACC TG-3′ and 5′-ATC ACG GCC GAT GAC GAG CC-3′. Relative mRNA abundance was determined by the ΔC_t method as described previously (27) and normalized to Ppia abundance.

Statistical analyses

For RT-PCR, three tissue pools consisting of two to three UGSs per pool were used to measure mRNA abundance. For prostatic bud counting, UGSs were stained by ISH for NK3 homeobox 1 (Nkx3-1) mRNA and three individuals, blinded to treatment conditions, counted the total number of Nkx3-1 positive prostatic buds in each of at least four UGSs per experimental group. Nkx3-1 stained urethral gland buds were excluded from analysis (26). For immunolabeled cell counting, PH3-positive, KRT14-positive and double-positive cells were counted in at least two sections per UGS in at least three UGSs per treatment group. For the assessment of castration efficiency, lower urinary tract (LUT; including prostate, urethra, bladder, and seminal vesicle) wet weight was measured relative to body weight at time of euthanasia. Statistical analysis was performed using R version 2.13.1. Homogeneity of variance was determined using a Bartlett's test. A Student's t test or ANOVA, followed by Fisher's least significant difference test were used to identify significant differences (P ≤ 0.05) between or among treatment groups.

Results

Demonstration of Wif1 riboprobe specificity

We created a Wif1 riboprobe and demonstrated its ability to recapitulate the Wif1 expression pattern reported previously for a different Wif1 riboprobe (28) in E13.5 mouse handplate mesenchyme (Supplemental Fig. 1A). To further establish selectivity of our riboprobe, we showed that the Wif1 expression pattern in mouse UGS as revealed by ISH (described in the section below) was similar to the lacZ expression pattern in Wif1^lacz/+ UGS (Supplemental Fig. 1B).

Wif1 mRNA is more abundant in male compared with female UGM and overlaps androgen-responsive Srd5a2 mRNA from the earliest stages of mouse prostate development

Our next objective was to analyze the Wif1 expression pattern over a broad developmental period to include stages E14.5 to P5. This period spans the four earliest stages of prostate ductal development: prostatic bud specification, when molecular signals specify where prostatic buds will form from UGE (E14.5–16.5); prostatic bud initiation, when buds first project from UGE (E16.5–18.5); prostatic bud elongation, when prostatic buds extend into UGS stroma and prostatic ductal branching morphogenesis (E16.5 to P5) (1, 29, 30).

Fetal and neonatal mouse UGS sections were stained by ISH to visualize Wif1 mRNA. To determine whether Wif1 expression in mouse UGS is associated with AR activity, we compared its temporal and spatial expression to a known androgen-responsive gene, Srd5a2 (steroid 5α-reductase 2) (26, 31). To determine whether Wif1 is present in UGS before prostatic specification, we compared its temporal expression pattern with the earliest known mRNA marker of the mouse prostate field, Nkx3-1 (24, 32). To determine the onset and duration of sexually dimorphic Wif1 mRNA expression, we compared its expression in male and female UGS at all examined developmental stages (Fig. 1). Tissues were also stained by IHC to visualize epithelium with CDH1 (red) and ACTA2 (green) so that Wif1 expression could be assessed and described in the context of other tissue compartments (26).

Fig. 1.

Wif1, Srd5a2, and Nkx3-1 mRNA expression in developing male and female mouse UGS. A–Y, Near midsagittal sections (50 μm) of E14.5 to P5 male and female UGSs were stained by ISH to visualize mRNA expression (purple) of Wif1, Srd5a2 (marker of prostatic stromal AR activity), and Nkx3-1 (marker of prostatic buds). Sections were then immunofluorescently stained with antibodies against CDH1 (red) to label all epithelium, and ACTA2 (green) present in mesenchyme. Results are representative of three litter-independent samples for each gender and stage. bl, Bladder; ed, ejaculatory duct; lv, lower vagina; md, Müllerian duct; sv, seminal vesicle; uv, upper vagina; v, vagina; wd, Wolffian duct. Arrowheads indicate prostatic buds. A black line marks the interface between UGS epithelium and mesenchyme. BM Purple detection time was the same for males and females and was as follows: Wif1 (62 h), Srd5a2 (31 h), Nkx3-1 (47 h).

Wif1 mRNA was detected in male and female UGS at E14.5, before prostate field specification as noted by the absence of Nkx3-1 expression (Fig. 1, A–E). Wif1 was more abundant in male compared with female UGS stroma beginning at E15.5 (Fig. 1, F and G) and continuing until at least P5 (Fig. 1, U and V). Intriguingly, the onset and duration of sexually dimorphic Wif1 expression in UGS stroma coincided temporally with onset and duration of sexually dimorphic Srd5a2 mRNA expression in UGS stroma (Fig. 1, H and I). Beginning at E15.5 and continuing until at least P5, Wif1 and Srd5a2 mRNAs were particularly concentrated in male urethral lamina propria mesenchyme (Fig. 1). This stromal subcompartment, positioned between UGE (labeled by CDH1, Fig. 1, red) and urethral muscularis propria (labeled by ACTA2, Fig. 1, green) (Fig. 1) (26), was previously identified as a key site of AR action during prostate development (33). Wif1 also exhibited sexually dimorphic expression in UGE. Beginning at E16.5 and continuing until at least P5, Wif1 was coexpressed with Nkx3-1 in male UGS and prostatic bud epithelium (Fig. 1). Wif1 was also detected in the male Wolffian duct, seminal vesicle, and ejaculatory duct mesenchyme between E14.5 to P5 and in urethral smooth muscle between stages E16 and P5. In female UGS, Wif1 localized to lower vagina (sinus vagina) mesenchyme and to a discrete population of ventral urethral basal epithelium (Fig. 1, F, K, and P). The latter is an anatomical region in which the small Nkx3-1-positive epithelial buds have been documented in female C57BL/6J mouse fetuses (34). Collectively these results reveal localized Wif1 expression in androgen-responsive male UGS stroma and in epithelial bud structures of male and female UGS.

To confirm ISH results showing Wif1 is more abundant in male vs. female UGS, cDNA from microdissected E17.5 male and female UGE and UGM was analyzed by real-time PCR. We confirmed the purity of each cDNA preparation by demonstrating selective expression of Acta2 mRNA in the UGS mesenchymal cDNA preparation and Krt14 in the UGS epithelial cDNA preparation (Supplemental Fig. 1C). We then demonstrated that Wif1 mRNA was more abundant in male compared with female UGM and UGE (Supplemental Fig. 1C). Taken together, these results reveal that Wif1 is present in the fetal mouse UGS in which it is more abundant in male compared with female fetuses.

Androgens are necessary and sufficient for Wif1 mRNA expression in fetal mouse prostate mesenchyme

Prostate development requires AR activation in UGM (4, 5). The observation that Wif1 is more abundant in male compared with female UGM, coupled with the fact that Wif1 expression overlaps spatially and temporally with androgen-responsive Srd5a2, raised the hypothesis that Wif1 mRNA expression in UGM is regulated by androgens. We assessed whether androgens are necessary for Wif1 mRNA expression in UGM before prostatic bud formation (E14.5), remain necessary after prostatic bud initiation (E17.5), and are sufficient for de novo Wif1 expression in E17.5 female UGM. The E14.5 male and E17.5 male and female UGSs were incubated for 24 h in organ culture media containing DHT and either vehicle alone or vehicle containing the AR antagonist OHF. After the culture period, UGSs were stained by ISH to visualize Wif1 mRNA (pseudocolored green) and by IHC with anti-CDH1 (red, all epithelium) and DAPI (blue) to localize Wif1 mRNA expression.

Androgen-responsive Srd5a2 mRNA (26) was also visualized to confirm pharmacological sufficiency of AR agonist and antagonist treatments. In male UGSs, OHF treatment noticeably reduced Srd5a2 and Wif1 abundance in UGM but not in prostatic bud epithelium (Fig. 2, A–J, insets). In female UGSs, DHT treatment activated Srd5a2 and Wif1 expression specifically in UGM, and this action was blocked by OHF (Fig. 2, E, F, K, and L, insets). To independently confirm the requirement of androgens for Wif1 expression in UGM, timed pregnant dams were dosed with AR antagonist flutamide before prostatic bud formation (E13.5 and E14.5) or during prostatic bud formation (E17.5 and E18.5). Male UGSs were collected on E15.5 and P0 and stained by ISH to visualize Wif1 and Srd5a2 mRNA (purple) and by IHC with anti-CDH1 (red, all epithelium) and anti-ACTA2 (green, smooth muscle) (Fig. 2, M–T). Flutamide treatment noticeably reduced Srd5a2 and Wif1 abundance in male UGM (Fig. 2, M–T, insets) but not Wif1 abundance in male prostatic bud epithelium (Fig. 2, S and T, arrowheads). These results support the hypothesis that androgens and active AR signaling are necessary and sufficient for Wif1 mRNA expression in fetal UGM but not UGE.

Fig. 2.

AR signaling is necessary and sufficient for Wif1 mRNA expression in fetal mouse prostate stroma. A–L, E14.5 male or E17.5 male and female UGS explants were cultured for 24 h in media containing DHT (10 nm) or DHT and the AR antagonist OHF (10 μm). Near midsagittal frozen sections (10 μm) were stained by ISH to visualize Srd5a2 (A–F) and Wif1 (G–L) mRNAs (green). Sections were then immunofluorescently stained with anti-CDH1 (red) antibody to label all epithelium and DAPI (blue). M–T, Timed pregnant dams were dosed with corn oil alone or corn oil containing flutamide on E13.4 and E14.5 and male UGSs collected on E15.5 or dosed on E17.5 and E18.5 and male UGSs collected on P0. Near midsagittal sections (50 μm) were stained by ISH to visualize Srd5a2 (M–P) and Wif1 (Q–T) mRNAs (purple). Sections were then immunofluorescently stained with anti-CDH1 (red) antibody to label all epithelium and anti-ACTA2 (green) to label smooth muscle. Insets represent magnified images. Results are representative of three litter-independent samples per treatment group and were processed as a single experimental unit. bl, Bladder; sv, seminal vesicle; u, urethra; wd, Wolffian duct. Arrowheads indicate presence of epithelial Wif1 in prostatic buds. A white or black line marks the interface between UGS epithelium and mesenchyme. BM Purple detection time was the same for all samples across experimental groups and was as follows: Wif1 (97 h, 10 μm frozen sections; 22 h, 50 μm sections); Srd5a2 (97 h, 10 μm frozen sections; 22 h, 50 μm sections).

Androgens are necessary and sufficient for Wif1 expression in the adult mouse prostate stroma

We next tested whether androgens are necessary for Wif1 expression in adult prostate stroma. Male mice were either sham castrated (control) or castrated to reduce circulating androgens and induce prostate involution. There was a significant reduction in relative LUT wet weight among 3-d castrated males (3dC) compared with 3-d sham-castrated controls (3dS; Supplemental Fig. 2). Tissues were stained by ISH to visualize Srd5a2 and Wif1 mRNA (purple) and by IHC with anti-CDH1 (red), anti-ACTA2 (green) and DAPI (blue) to localize mRNA expression. Srd5a2 and Wif1 mRNAs were detected in ventral prostate stroma of both experimental groups, but both mRNAs were visibly less abundant in 3dC vs. 3dS (Fig. 3, A, B, E, and F). We observed a similar trend in other prostate lobes and in seminal vesicle (Supplemental Fig. 3). These results suggest a testicular androgen requirement for normal Wif1 mRNA expression in adult mouse prostate stroma.

Fig. 3.

AR signaling is necessary for Wif1 mRNA expression in mature prostate stroma. Adult male mice (48–50 d) were castrated or sham castrated. Prostate tissue was obtained from some mice 3 d after sham castration (3dS) or 3 d after castration (3dC), whereas other mice received empty (14dC+3dE) or testosterone-filled (14dC+3dT) SILASTIC brand capsule implants (Dow Corning) 14d after castration and prostate tissue was obtained 3 d later. ISH was used to visualize Srd5a2 (A–D) and Wif1 (E–H) mRNA expression in frozen ventral prostate sections (10 μm). ISH sections were then immunofluorescently labeled with antibodies against CDH1 (red) to label all epithelium, ACTA2 (green) to label smooth muscle and DAPI (blue). Results are representative of three litter-independent samples per treatment group. Arrows represent Wif1 mRNA-positive muscle cells. A black line marks the interface between UGS epithelium and stroma. BM Purple detection time was the same for all samples across experimental groups and was as follows: Wif1 (75 h), Srd5a2 (75 h).

We next tested whether exogenous androgen would restore Wif1 expression in castrated mouse prostate stroma. Fourteen-day castrated mice were implanted with empty or testosterone-filled SILASTIC brand capsules (Dow Corning) and 3 d later were euthanized. Relative LUT wet weight was significantly greater among 14-d castrated males implanted with testosterone capsules (14dC+3dT) than among males implanted with empty capsules (14dC+3dE; Supplemental Fig. 2). Wif1 and Srd5a2 mRNAs were weakly expressed in 14dC+3dE ventral prostate stroma, and both were noticeably more abundant in 14dC+3dT ventral prostate stroma (Fig. 3, C, D, G, and H). The same trend existed in the other prostate lobes and seminal vesicle (Supplemental Fig. 3). Collectively these results indicate that androgens are required for Wif1 expression in the intact adult mouse prostate and are sufficient for reactivating Wif1 expression during regeneration of the castrated adult mouse prostate.

Exogenous WIF1 protein enhances androgen-dependent prostatic bud formation and basal epithelial cell proliferation

Formation and continued development of Nkx3-1-positive prostatic buds requires ARs in UGM (4, 5), but AR-responsive genes mediating this action are largely unknown. We hypothesized that WIF1 enhances AR action by stimulating prostatic bud formation. To test this, E14.5 male UGSs were grown in UGS organ culture media containing DHT and graded concentrations of exogenous recombinant human WIF1 protein (Fig. 4, A–F). Recombinant human WIF1 protein is 93% homologous to the mouse ortholog, and its ability to impair WNT/β-catenin signaling was demonstrated previously in an array of species and tissues including Xenopus embryo, chicken heart, and mouse thymus (35–37). Whole-mount ISH was performed as described previously to visualize and quantify the Nkx3-1-positive prostatic buds formed per the UGS (24). Exogenous recombinant WIF1 protein elicited a graded increase in the Nkx3-1-positive buds (Fig. 4G), which supports the hypothesis that WIF1 enhances androgen-dependent prostatic bud formation.

Fig. 4.

Exogenous recombinant WIF1 enhances androgen-dependent prostatic bud formation in vitro. A–F, E14.5 wild-type male UGSs were cultured for 4 d in the presence of 10 nm DHT and graded concentrations (0–1000 ng/ml) of exogenous recombinant WIF1 protein. Whole-mount tissues were then stained by ISH to visualize and quantify Nkx3-1 mRNA-marked prostatic buds (purple). G, Results are mean ± sem (n = 4 litter-independent samples per group). Asterisks indicate a significant difference compared with 0 ng/ml WIF1 (P < 0.05). bl, Bladder. BM Purple detection time was the same for all samples across experimental groups and was as follows: Nkx3-1 (21 h).

Because WIF1 enhances prostatic bud formation and developing prostatic buds are comprised of basal epithelium, we next tested the hypothesis that WIF1 enhances androgen-induced UGS basal epithelial cell proliferation. UGS explants were incubated in organ culture media as described above, cut into sagittal sections, and immunofluorescently stained to reveal mitotically active cells (phosphohistone H3 positive, PH3+, red) and basal epithelial cells (KRT14+, green) (Fig. 5, A–D). Exogenous WIF1 concentrations that significantly increased Nkx3-1-positive prostatic bud number (500 ng/ml, 1000 ng/ml; Fig. 4) also significantly increased the mean fraction of PH3⁺KRT14⁺/total KRT14⁺ cells (Fig. 5E). These results show a pro-proliferative action of WIF1 on UGS basal epithelium.

Fig. 5.

Exogenous recombinant WIF1 enhances UGS basal epithelial cell proliferation in vitro. A–D, E14.5 wild-type male UGSs were cultured for 4 d in the presence of 10 nm DHT and graded concentrations (0–1000 ng/ml) of exogenous recombinant WIF1 protein. Near midsagittal sections (5 μm) were immunofluorescently stained to visualize and count the percentage of KRT14 (green)-positive basal epithelial cells with detectable nuclear PH3 (red). Cell nuclei were visualized with DAPI (blue). E, Results are mean ± sem (n = 3 litter-independent samples per group). Asterisks indicate a significant difference compared with 0 ng/ml WIF1 (P < 0.05). Square insets represent magnified images. bl, Bladder; bud, prostatic bud. A white line marks the interface between UGS epithelium and mesenchyme.

Exogenous WIF1 does not change the pattern of WNT/β-catenin target genes in prostatic buds

The most widely accepted Wif1 mechanism of action is functional antagonism of WNT/β-catenin signaling (35, 36, 38). Yet in developing mouse prostate, Wif1 is expressed in UGM and UGE and overlaps WNT/β-catenin-responsive Axin2 and Lef1 (lymphoid enhancer binding factor 1) expression (14, 15, 19). Coexpression of Wif1 with abundant WNT/β-catenin-responsive mRNA expression would not be predicted if WIF1 functioned as a strong WNT/β-catenin inhibitor in control UGS. We tested whether prostatic bud-enhancing concentrations of exogenous WIF1 protein visibly altered the pattern of WNT/β-catenin-responsive Axin2 and Lef1 mRNA within and immediately surrounding prostatic bud tips in which both mRNAs were detected previously (19). E14.5 male UGSs were grown for 4 d in UGS organ culture media containing DHT and graded recombinant WIF1 protein concentrations. ISH was used to visualize Axin2 and Lef1 mRNAs. Axin2 and Lef1 were detected in prostatic bud epithelium in all experimental groups, and exogenous recombinant WIF1 did not visibly alter their patterning (Fig. 6). Presence of WNT/β-catenin target genes and Wif1 at the same time and location during embryonic prostatic development suggests at least two possible mechanisms for prostatic bud enhancement by WIF1: a subtle β-catenin-dependent action we are unable to detect or a β-catenin-independent action.

Fig. 6.

Exogenous recombinant WIF1 does not visibly alter WNT/β-catenin-responsive Axin2 and Lef1 mRNA expression during androgen-dependent prostatic bud formation in vitro. E14.5 wild-type male UGSs were cultured for 4 d in the presence of 10 nm DHT and graded concentrations (0–1000 ng/ml) of exogenous recombinant WIF1 protein. Near midsagittal UGS sections (10 μm) were stained by ISH to visualize Axin2 (A–D) and Lef1 (E–H) mRNAs (green). Sections were then immunofluorescently stained with an antibody against CDH1 (red) to label all epithelium and with DAPI (blue). Results are representative of three litter-independent samples per treatment group. A white line marks the interface between the UGS epithelium and mesenchyme. BM Purple detection time was the same for all samples across the experimental groups and was as follows: Axin2 (72 h), Lef1 (72 h).

Exogenous WIF1 is not sufficient to induce prostatic budding in the absence of androgens

To determine whether WIF1 is the only androgen-induced signal needed for prostatic bud formation, E14.5 female UGSs, naïve to high levels of androgens, were cultured for 4 d in androgen-free media containing recombinant WIF1. As a positive control, female UGSs were cultured in media containing 10 nm DHT and no recombinant WIF1. After the culture period, ISH was used to visualize three different prostatic bud-marking mRNAs: Nkx3-1, Wnt10b, and Edar (19, 24, 26). Although Nkx3-1 is considered the best marker of prostatic buds, Wnt10b and Edar are additional mRNAs restricted to distal tips of prostatic buds and are not detected in urethral gland buds (19, 24, 26, 32). As expected, DHT induced the formation of Nkx3-1-, Wnt10b-, and Edar-positive prostatic buds in female UGS explants (Fig. 7, D, H, and L, solid arrowheads). The only Nkx3-1 positive buds formed in androgen-free medium were in the ventral UGS region in which small buds were previously shown to form by an androgen-independent mechanism (Fig. 7, A–C, open arrowheads) (34). Exogenous WIF1 protein concentrations greater or equal to those that increased prostatic bud number in the presence of androgens (Fig. 4; ≥ 500 ng/ml) did not increase Nkx3-1- (Fig. 7, A–C), Wnt10b- (Fig. 7, E–G), or Edar (Fig. 7, I–K)-positive prostatic bud number in the absence of androgens. Therefore, the formation of large prostatic buds that normally occur in male UGSs likely requires the actions of other androgen-induced factors in collaboration with WIF1.

Fig. 7.

Exogenous recombinant WIF1 does not stimulate androgen-independent prostatic bud formation in vitro. E14.5 wild-type female UGSs were cultured for 4 d in media containing 10 nm DHT (positive control) or in androgen-free media containing exogenous recombinant WIF1 protein. Whole-mount tissues were then stained by ISH to visualize Nkx3-1 (A–D), Wnt10b (E–H), and Edar (I–L) mRNA-marked prostatic buds. Results are representative of five litter-independent samples per treatment group. bl, Bladder. A–C, Open arrowheads indicate Nkx3-1-positive epithelial buds observed in females. D, H, and J, Solid arrowheads indicate prostatic and urethral gland buds. BM Purple detection time was the same for all samples across experimental groups and was as follows: Nkx3-1 (21 h); Wnt10b (50 h); Edar (36 h).

Prostatic bud number is unaltered in Wif1 mutant mice

We next assessed whether endogenous Wif1 is required for androgen-dependent prostatic bud formation in vivo. To test this hypothesis, we used a Wif1 targeted mutant mouse containing a knock-in lacZ transgene that disrupts the Wif1 coding sequence (22). E18.5 male Wif1 wild-type (Wif1^+/+) and mutant (Wif1^lacZ/lacZ) whole-mount UGSs were stained by ISH to visualize and quantify Nkx3-1-positive prostatic buds. There were no significant genotypic differences in pattern or number of ventral, anterior, dorsolateral, or total prostatic buds formed (Fig. 8, A–C). Therefore, Wif1 is not required for normal prostatic bud formation. We investigated the possibility that WIF1 acts by a functionally redundant mechanism to promote prostatic bud formation. Specifically, we tested whether Wif1 mutant mice exhibited a visible compensatory increase in Sfrp mRNAs. SFRPs have been identified in the developing prostate and are thought to function in a manner similar to Wif1 (14, 15, 20). E18.5 Wif1 mutant and wild-type UGSs were stained by ISH to visualize mRNA expression patterns of Sfrp1, Sfrp2, Sfrp3 (also known as Frzb), and Sfrp4 mRNAs. There was no noticeable genotypic difference in Sfrp1 or Sfrp4 (results not shown), but there was a noticeably wider band of Sfrp2 and Sfrp3 mRNAs in Wif1 mutant UGSs compared with wild-type (Fig. 8, D–G). Increased expression of Sfrp2 and Sfrp3 mRNAs in Wif1 mutant UGSs suggests balanced WNT signaling is likely to be important in prostatic bud formation.

Fig. 8.

Wif1 is not required for normal prostatic bud formation in vivo. E18.5 wild-type control (Wif1^+/+) (A) and Wif1 mutant (Wif1^lacZ/lacZ) male mouse UGSs (B) were stained by ISH to visualize and quantify the number of Nkx3-1 mRNA-marked prostatic buds per UGS. C, Results are mean ± sem (n = 5 litter-independent samples per group). D–G, Near midsagittal sections (50 μm) of E18.5 Wif1^+/+ and Wif1^lacZ/lacZ male UGSs were stained by ISH to visualize mRNA expression (purple) of Sfrps2 and Sfrps3. Sections were then stained by IHC with antibodies against CDH1 (red) to label all epithelium and ACTA2 (green). Results are representative of two litter-independent samples per genotype. Asterisks indicate regions of differential mRNA expression between Wif1^+/+ and Wif1^lacZ/lacZ male UGSs. bl, Bladder. A black line marks the interface between UGS epithelium and mesenchyme. BM Purple detection time was the same for wild-type and mutant UGSs and was as follows: Nkx3-1 (38 h); Sfrp2 (5.5 h); Sfrp3 (63 h).

Discussion

Fetal mouse urethral lamina propria mesenchyme is revealed as a key site of Wif1 expression and AR action

It has long been known that AR action in UGS stroma is necessary for prostate development (1, 3–5). Yet neither the cellular composition of UGS stroma nor AR activity within UGS stroma is homogeneous (26). This study, in conjunction with other recent studies, refines the subanatomical location of AR action in the developing mouse prostate. Fetal urethral lamina propria mesenchyme, positioned between urethral urothelium and muscularis propria was hypothesized long ago to be an important site of AR action (33). More recent studies identified this subanatomical region as the predominant site of AR-responsive Srd5a2 expression in mouse (26), rat (39), and male-enriched Sfrp2 expression in mouse (15). We have now identified Wif1 as an androgen-responsive mRNA in fetal mouse urethral lamina propria mesenchyme. As the prostate development field moves forward, a focus on this particular fetal prostate subcompartment, which accounts for only about 10% of total UGS stroma, will likely yield information about AR-mediated mechanisms of prostate morphogenesis.

We know very little about prostate stromal cell fate, but identification of fetal mouse urethral lamina propria mesenchymal cells as a key site of AR action may help to close this knowledge gap. For example, epithelial-mesenchymal interactions are crucial for fetal prostate mesenchyme differentiation (40). Whether urethral lamina propria mesenchyme gives rise to smooth muscle or other UGS stromal cell populations or participates in the instructive patterning of other UGS stromal constituents (blood vessels, nerves, urethral sphincter skeletal muscle) is unknown. Furthermore, whether urethral lamina propria cells expand clonally during prostatic bud elongation and branching to produce AR-positive periductal mesenchyme cells or whether AR activity is stimulated de novo in these cells is unknown. A transgenic mouse that expresses an inducible tag would allow for lineage tracing in urethral lamina propria mesenchyme and would resolve some of these questions.

Wif1 appears to be differently regulated in UGS mesenchyme compared with UGS epithelium

A better understanding of Wif1 transcriptional regulation during prostate development will shed light on the mechanism of AR-induced prostatic bud formation. We identified Wif1 in UGM and UGE. Differences in onset, duration, pattern, and androgen-dependence of Wif1 expression in UGM and UGE suggest distinct Wif1 transcriptional regulatory mechanisms in these fetal prostate tissue compartments.

In fetal UGE, WNT/β-catenin signaling is more likely than AR signaling to activate Wif1 transcription. Although AR protein is present in UGE (26, 33), its presence there does not lead to appreciable binding of ligand (33), is expendable for prostatic bud formation (41), and is not needed for Wif1 expression (Fig. 2). In contrast, β-catenin activity associates with Wif1 expression in other tissues (42–44), is required for prostatic bud formation (45, 46), and positively regulates the target genes Axin2 and Lef1, which associate spatially and temporally with Wif1 in UGE (19) (www.gudmap.org).

In fetal UGM, AR signaling is more likely than WNT/β-catenin signaling to activate Wif1 transcription. AR protein is present in UGM, binds ligand, and is required for prostatic bud formation (1, 3–5). Although WNT/β-catenin-responsive Axin2 and Lef1 are present in UGM, both transcripts are detectable at an earlier developmental stage and in a different pattern than Wif1 (19) (www.gudmap.org), do not persist in UGM as long as Wif1, and are not reliant on a high level of androgen signaling in the UGM like Wif1. An important future direction of our research is to determine whether Wif1 transcription in UGM and UGE is regulated directly or indirectly by AR and by β-catenin activities.

Wif1 promotes prostatic bud formation and fetal prostate basal epithelial cell proliferation

A preponderance of evidence supports WIF1 as a suppressor of adult prostate epithelial cell proliferation (47–50). Although few studies have examined WIF1 expression and function during prostate development, Li et al. (51) overexpressed WIF1 in UGM, combined UGM with adult bladder epithelium, and grew the tissue recombinants under the kidney capsule of immunocompromised mice. WIF1 overexpression in UGM impaired its ability to reprogram adult bladder epithelium into prostate epithelium, which suggests that abnormally high Wif1 expression in UGM impairs the instructive signals needed for prostate development (51).

The results of the current study differ from those described above. We found exogenous recombinant WIF1 protein, at media concentrations of 500-1000 ng/ml, enhanced androgen-dependent mouse prostatic bud formation and basal epithelial cell proliferation (Figs. 4 and 5). Discrepancies in WIF1 action reported in our study vs. those described above are likely a function of prostate cell populations present, prostate developmental stages examined, and WIF1 concentrations used. Studies conducted in cell lines (47, 49) or human prostate carcinomas (50) vary in their AR responsiveness, exhibit a cancerous phenotype, and do not fully recapitulate the mesenchymal-epithelial interactions present during prostate development. Additionally, interactions between fetal UGM and adult bladder epithelium under the kidney capsule of an immunocompromised mouse (51) may not be the same as the interactions between intact fetal UGM and UGE in UGS culture medium.

Although we found that exogenous WIF1 protein stimulated prostatic bud formation and UGS basal epithelial cell proliferation, Wif1 was expendable for normal prostate development (Fig. 8). These results are consistent with observations made by others that Wif1 is not required for morphogenesis of most mouse tissues (22). A likely possibility is that other genes with a similar function to Wif1, such as Sfrps, are up-regulated to compensate for the deficient WIF1 activity in Wif1 mutant UGSs. In support of this possibility, we observed an expanded pattern of Sfrp2 and Sfrp3 in Wif1 mutant UGM compared with wild-type UGM. These results suggest a possible functional redundancy among SFRPs and WIF1 during prostate development.

Possible mechanisms by which WIF1 could enhance fetal prostatic bud formation

One of the most important aspects of the current study is that it reveals WIF1 as one of the first androgen-regulated genes that mediates androgen action during prostate development. An intriguing question raised as a result of our study is how WIF1 enhances prostatic morphogenesis. Although we did not directly answer this question, at least two possible mechanisms should be considered for future investigation: β-catenin-dependent or -independent mechanisms. Even though exogenous WIF1 concentrations that induced prostatic bud formation did not elicit noticeable changes in the pattern of WNT/β-catenin-responsive Axin2 and Lef1 mRNAs (Fig. 6), we cannot exclude the possibility that it causes small changes that we were unable to detect. We do not fully understand the concentration-response relationship between β-catenin activity and prostatic bud formation and minor changes in this signaling pathway could account for biologically significant changes in prostate development. In chicken proepicardial explant cultures, for example, subtle changes in β-catenin activity in either the positive or negative direction elicit the same response: cardiomyocyte proliferation (36). Therefore, although WNT/β-catenin-responsive Axin2 and Lef1 mRNA patterns were not noticeably changed in UGS explants exposed to WIF1, it is possible that we failed to detect subtle or finely localized WIF1-mediated changes in WNT signaling that were important in shaping the pattern and number of prostatic buds formed in the developing mouse UGS.

Finally, it is possible that WIF1 enhances prostatic bud formation by a β-catenin-independent mode of action. Addition of exogenous recombinant WIF1 did not appear to alter WNT/β-catenin target genes Axin2 and Lef1 (Fig. 6). Consistent with this finding, abundance of most WNT target genes was unaltered by inappropriately high levels of Wif1 reported previously in UGM in which Tgfrb2 was conditionally deleted (51). Vertebrate WIF1 typically functions as an inhibitor of WNT signaling, but Drosophila ortholog shifted modulates Hedgehog signaling (52). Although it is thought that vertebrate WIF1 does not bind hedgehog, recent evidence suggests that it may be capable of weakly modulating Hedgehog activity (53). WIF1 is also capable of binding connective tissue growth factor in vitro independently of its inhibitory action on WNT signaling (54). Whether WIF1 is capable of modulating Hedgehog signaling or whether WIF1 binds other signaling molecules during prostate development is unknown.