Canonical Wnt signaling regulates Nkx3.1 expression and luminal epithelial differentiation during prostate organogenesis

Authors

  • Marianna Kruithof-de Julio,

    1. Departments of Medicine and Genetics and Development, Columbia University Medical Center, New York, New York
    2. Herbert Irving Comprehensive Cancer Center, Columbia University Medical Center, New York, New York
    Search for more papers by this author
    • Drs. Kruithof-de Julio and Shibata contributed equally to this work.

  • Maho Shibata,

    1. Departments of Medicine and Genetics and Development, Columbia University Medical Center, New York, New York
    2. Herbert Irving Comprehensive Cancer Center, Columbia University Medical Center, New York, New York
    Search for more papers by this author
    • Drs. Kruithof-de Julio and Shibata contributed equally to this work.

  • Nishita Desai,

    1. Center for Advanced Biotechnology and Medicine, University of Medicine and Dentistry of New Jersey - Robert Wood Johnson Medical School, Piscataway, New Jersey
    Search for more papers by this author
  • Melissa Reynon,

    1. Center for Advanced Biotechnology and Medicine, University of Medicine and Dentistry of New Jersey - Robert Wood Johnson Medical School, Piscataway, New Jersey
    Search for more papers by this author
  • M. Vivienne Halili,

    1. Center for Advanced Biotechnology and Medicine, University of Medicine and Dentistry of New Jersey - Robert Wood Johnson Medical School, Piscataway, New Jersey
    Search for more papers by this author
  • Ya-Ping Hu,

    1. Center for Advanced Biotechnology and Medicine, University of Medicine and Dentistry of New Jersey - Robert Wood Johnson Medical School, Piscataway, New Jersey
    Search for more papers by this author
  • Sandy M. Price,

    1. Center for Advanced Biotechnology and Medicine, University of Medicine and Dentistry of New Jersey - Robert Wood Johnson Medical School, Piscataway, New Jersey
    Search for more papers by this author
  • Cory Abate-Shen,

    1. Departments of Medicine and Genetics and Development, Columbia University Medical Center, New York, New York
    2. Herbert Irving Comprehensive Cancer Center, Columbia University Medical Center, New York, New York
    Search for more papers by this author
  • Michael M. Shen

    Corresponding author
    1. Departments of Medicine and Genetics and Development, Columbia University Medical Center, New York, New York
    2. Herbert Irving Comprehensive Cancer Center, Columbia University Medical Center, New York, New York
    • Correspondence to: Michael M. Shen, Departments of Medicine and Genetics and Development, Herbert Irving Comprehensive Cancer Center, Columbia University Medical Center, New York, NY 10032 E-mail: mshen@columbia.edu

    Search for more papers by this author

Abstract

Background: The formation of the prostate gland requires reciprocal interactions between the epithelial and mesenchymal components of the embryonic urogenital sinus. However, the identity of the signaling factors that mediate these interactions is largely unknown. Results: Our studies show that expression of the prostate-specific transcription factor Nkx3.1 is regulated by the canonical Wnt signaling pathway. Using mice carrying a targeted lacZ knock-in allele of Nkx3.1, we find that Nkx3.1 is expressed in all epithelial cells of ductal buds during prostate organogenesis. Addition of Wnt inhibitors to urogenital sinus explant culture greatly reduces prostate budding and inhibits Nkx3.1 expression as well as differentiation of luminal epithelial cells. Analyses of a TCF/Lef:H2B-GFP transgene reporter show that canonical Wnt signaling activity is found in urogenital mesenchyme but not urogenital sinus epithelium before prostate formation, and is later observed in the mesenchyme and epithelium of prostate ductal tips. Furthermore, TCF/Lef:H2B-GFP reporter activity is reduced in epithelial cells of Nkx3.1 null neonatal prostates, suggesting that Nkx3.1 functions to maintain canonical Wnt signaling activity in developing prostate bud tips. Conclusions: We propose that activated canonical Wnt signals and Nkx3.1 function in a positive feedback loop to regulate prostate bud growth and luminal epithelial differentiation. Developmental Dynamics, 242:1160–1171, 2013. © 2013 Wiley Periodicals, Inc.

INTRODUCTION

During normal development, the induction and morphogenesis of tissues requires precise regulation of growth factor signaling pathways, whereas many of these same pathways become de-regulated during cancer progression. In the case of the prostate gland, recent gene expression profiling analyses have shown that embryonic programs of gene expression are reactivated in prostate cancer (Schaeffer et al., 2008; Pritchard et al., 2009). However, relatively little is known about critical signaling pathways that regulate either normal development or carcinogenesis of the prostate gland.

Organogenesis of the prostate gland is androgen-dependent and involves reciprocal signaling events between the urogenital sinus epithelium and surrounding mesenchyme. These tissue interactions result in the initial emergence of prostate buds from the rostral end of the urogenital sinus, which occurs at 17.5 days postcoitum (dpc) in the mouse (Cunha et al., 1987). Subsequently, the prostatic epithelial buds undergo extensive ductal outgrowth and branching into the surrounding mesenchyme during the first 3 weeks of postnatal development, giving rise to the distinct anterior, dorsolateral, and ventral lobes of the prostate (Lung and Cunha, 1981; Sugimura et al., 1986; Cunha et al., 1987; Prins and Putz, 2008). During this process of organogenesis, the primary epithelial cell types of the prostate epithelium are specified, corresponding to the luminal secretory cells and the underlying basal epithelium, as well as a rare population of neuroendocrine cells (Shen and Abate-Shen, 2010).

The tissue interactions necessary for prostate development have been studied extensively using tissue recombination assays, in which dissociated epithelial and mesenchymal cells are recombined and grafted under the renal capsule of immunodeficient mice (Cunha et al., 1987; Cunha, 2008). Importantly, tissue recombination studies have shown that prostate differentiation requires both epithelial and mesenchymal components, as mature cell types fail to differentiate in the absence of either component. Although specificity for the urogenital mesenchyme is relatively stringent in these assays of prostate formation, there is relatively less specificity for the epithelial component because a wide range of epithelial cell types can form prostate when combined with urogenital mesenchyme (Cunha et al., 1987; Bhatia-Gaur et al., 1999; Cunha, 2008; Taylor et al., 2009).

Several recent studies have provided evidence for a central role of canonical Wnt signals in prostate formation (Zhang et al., 2006; Schaeffer et al., 2008; Mehta et al., 2011; Simons et al., 2012). In the canonical Wnt pathway, the binding of secreted Wnt ligands to Frizzled/Lrp receptor complexes activates a signal transduction pathway that results in accumulation of nuclear β-catenin and transcription of downstream target genes by β-catenin together with transcription factors of the TCF/LEF family (Logan and Nusse, 2004). A role for canonical Wnt signals in early prostate development has been suggested by the expression of numerous Wnt ligands in the urogenital mesenchyme as well epithelium before and during prostate formation (Zhang et al., 2006; Prins and Putz, 2008; Schaeffer et al., 2008; Yu et al., 2009). Indeed, several Wnt ligands as well as co-activators of the canonical Wnt pathway display a sexually dimorphic expression pattern and are detected specifically in the male urogenital sinus as well as in prostate buds (Mehta et al., 2011). Furthermore, recent studies have shown that deletion of β-catenin results in loss or reduction of prostate bud formation, suggesting an essential role for canonical Wnt signaling in prostate formation (Simons et al., 2012; Francis et al., 2013; Mehta et al., 2013). However, it remains unclear whether canonical Wnt signals are required for prostate induction per se, or instead are required for early events in prostate ductal outgrowth that are distinct from tissue induction.

Previous studies from our laboratory have shown that the homeodomain transcription factor Nkx3.1 is the earliest known molecular marker of the prostate epithelium (Bhatia-Gaur et al., 1999). Expression of Nkx3.1 is restricted to the epithelial cells within the nascent prostate buds of the ventral, dorsolateral, and anterior lobes shortly after their emergence at 17.5 dpc (Bhatia-Gaur et al., 1999; Berman et al., 2004). During postnatal organogenesis, Nkx3.1 continues to be expressed in the epithelium of all three prostate lobes, and becomes largely restricted to luminal cells by adulthood (Bhatia-Gaur et al., 1999; Chen et al., 2005; Wang et al., 2009). Although Nkx3.1 is not essential for prostate formation, Nkx3.1−/− null mutant mice display abnormal prostate ductal morphogenesis, with a 25–40% decrease in the number of ductal tips (Bhatia-Gaur et al., 1999; Schneider et al., 2000; Tanaka et al., 2000). Nkx3.1−/− adult prostates also display reduced production of secretory proteins, as well as epithelial hyperplasia and dysplasia that leads to formation of prostatic intraepithelial neoplasia (PIN) with increased age (Bhatia-Gaur et al., 1999; Tanaka et al., 2000; Kim et al., 2002a). Finally, Nkx3.1 expression marks a luminal stem cell population in the adult prostate and is also required for stem/progenitor maintenance during prostate regeneration (Wang et al., 2009). Thus, Nkx3.1 is likely to represent a key regulator of prostate organogenesis and luminal differentiation, as well as epithelial progenitor activity.

Based on these observations, we have investigated the hypothesis that canonical Wnt signals produced by the urogenital sinus mesenchyme regulate epithelial expression of Nkx3.1 during prostate development. In our studies, we have used a targeted lacZ knock-in allele of Nkx3.1 as well as highly sensitive transgenic reporter for canonical Wnt signaling activity to examine their tissue distribution during prostate formation. Using these mouse reagents, we find that canonical Wnt signaling activity is not detected in the urogenital sinus epithelium before prostate formation, but is found in both epithelium and mesenchyme after the onset of prostate budding. In urogenital sinus organ culture assays, we find that inhibition of canonical Wnt signaling disrupts prostate bud growth and luminal differentiation and down-regulates Nkx3.1 expression. Conversely, we also find that Nkx3.1 regulates canonical Wnt activity at prostate bud tips during neonatal prostate development. We integrate these findings to propose a model for prostate formation in which canonical Wnt signals act together with Nkx3.1 during organogenesis to regulate ductal outgrowth and epithelial differentiation.

RESULTS

Nkx3.1lacZ Reporter Recapitulates Nkx3.1 Expression During Prostate Organogenesis

In the adult mouse prostate, Nkx3.1 protein is expressed in the luminal epithelium of all three prostate lobes (Bhatia-Gaur et al., 1999; Kim et al., 2002a; Wang et al., 2009). To investigate the molecular basis for prostate induction and differentiation, we have generated a Nkx3.1lacZ reporter line of mice, in which the development of prostate epithelium can be readily monitored. For this purpose, we have used gene targeting to generate knock-in mice where lacZ expression is placed under the control of the Nkx3.1 promoter by insertion of a promoter-less lacZ expression cassette into the 5′ untranslated region of Nkx3.1 (Fig. 1A–C). The remaining lacZ reporter cassette creates a null allele for Nkx3.1, and consequently Nkx3.1lacZ/+ mice are heterozygous. This strategy includes the use of a self-excising tACE-Cre/PolII-neo cassette (Bunting et al., 1999), thereby eliminating potential nonspecific effects of the PolII promoter on expression of the lacZ reporter gene.

Figure 1.

Generation and analysis of β-galactosidase expression in Nkx3.1lacZ/+ knock-in mice. A: Targeting strategy. B: Southern blot confirmation of ACE-Cre and PolII-neo excision using probe B on HindIII digested DNA. Results from a male germline chimera and three F1 progeny. C: Southern blot confirmation of germline transmission in F2 mice, using probe A for hybridization to HindIII-digested tail DNA. Band sizes are 7.5 kb wild-type Nkx3.1, 4.5 kb targeted, 5.5 kb chimeric (Chi) allele. D–L: Expression of β-galactosidase in Nkx3.1lacZ/+ prostate epithelium. D-K: β-galactosidase expression in whole-mount UGS. D: Frontal view at 17.5 days postcoitum (dpc). E,F: Frontal view at postnatal day (P) 1 (E); also shown in high-power (F); no staining is observed in other urogenital tissues, including bladder, testis, and seminal vesicle. G–I: Lateral (G), ventral (H), dorsal (I) views at P4. J: Ventral view at P8. K: Dorsal view at P12. L: Section of a P12 dorsolateral prostate stained for β-galactosidase activity and counter-stained with hematoxylin; X-gal staining is only detected in epithelial cells. AP, anterior prostate; DLP, dorsolateral prostate; DP, dorsal prostate; E, EcoRI; H, HindIII; LP, lateral prostate; SV, seminal vesicle; UGS, urogenital sinus; VP, ventral prostate; X, XbaI. Scale bars = 500 μm.

Using embryonic and neonatal Nkx3.1lacZ/+ mice, we have investigated the distribution of Nkx3.1 expression during prostate organogenesis starting from the formation of ductal buds. For this purpose, we can readily detect β-galactosidase expression in the developing prostate epithelium of Nkx3.1lacZ/+ mice by histochemical staining for X-Gal (5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside) (Fig. 1D–L). In whole-mounts of embryonic urogenital sinus, β-galactosidase was detected in newly formed prostatic buds at 17.5 days postcoitum (dpc) (Fig. 1D). In whole-mounts of intact urogenital systems from neonates, the individual ducts of all three prostate lobes were stained throughout the length of the ducts, and were readily distinguished from neighboring unstained ducts of the seminal vesicle (Fig. 1E–K). Notably, during ductal outgrowth at postnatal day 4 (P4), β-galactosidase expression appeared highest at the distal ends of the prostate ducts (Fig. 1H,I). In addition, as shown in sections from P12 mice, β-galactosidase expression was specific for epithelial cells, and was not found in mesenchymal cells surrounding the developing prostate ducts (Fig. 1L). These findings indicate that the pattern of β-galactosidase expression in Nkx3.1lacZ/+ mice is concordant with the pattern of endogenous Nkx3.1 expression as previously described by in situ hybridization (Bhatia-Gaur et al., 1999; Keil et al., 2012), and with the expression of β-galactosidase in a transgenic line containing a lacZ reporter inserted into a 17 kb genomic region flanking the Nkx3.1 locus (Chen et al., 2005).

Canonical Wnt Signaling Activity Occurs in the Urogenital Sinus Mesenchyme and Prostate Epithelial Buds

Previous studies have suggested that canonical Wnt signals may mediate androgen-dependent responses in the embryonic urogenital sinus (Schaeffer et al., 2008; Simons et al., 2012). Consistent with this idea, numerous Wnt ligands are expressed in the urogenital sinus epithelium as well as mesenchyme (Zhang et al., 2006; Mehta et al., 2011). However, despite the urogenital expression of several Wnt ligands, it has been difficult to detect canonical Wnt signaling activity during the early stages of prostate formation. In particular, the existing BAT-gal, BATLacZ, and TOPGAL transgenic mouse lines do not appear to be sensitive reporters of canonical Wnt activity in the embryonic urogenital sinus (Mehta et al., 2011).

Consequently, we have used a highly sensitive green fluorescent protein (GFP) -expressing reporter line to visualize the dynamic spatial and temporal changes in the localization of canonical Wnt signaling activity in the embryonic urogenital sinus and postnatal urinary tract. For this purpose, we have used a mouse line carrying a TCF/Lef:H2B-GFP transgene that contains six TCF/Lef DNA binding sites and a hsp68 minimal promoter driving expression of H2B-GFP (Ferrer-Vaquer et al., 2010). Thus, the expression pattern of GFP in these mice corresponds to cells that receive canonical Wnt signals, which can be detected at single-cell resolution.

We have examined GFP expression in whole-mounts of TCF/Lef:H2B-GFP mice from embryonic 15.5 dpc, before initial prostate bud formation, through postnatal day 8 (P8), during ductal elongation and branching morphogenesis. At these stages, we observed high levels of GFP expression in the seminal vesicle and ductus deferens (Fig. 2A,B,F,G), which are derivatives of the Wolffian duct. We also observed lower levels of GFP expression in the urogenital sinus region at 17.5 dpc as well as in prostate ductal tips of all three lobes at postnatal stages (Fig. 2A–J). Analysis of sections showed that the GFP expression in the urogenital sinus at 15.5 and 17.5 dpc occurred in urogenital sinus mesenchyme (UGM), predominantly on the dorsal side, with no expression detected in the urogenital sinus epithelium (UGE) (Fig 2K,L). This urogenital mesenchyme expression of GFP persisted at P1, when prostate ducts are elongating, but was much sparser at P8 (Fig. 2M–P). In the epithelium, we detected GFP expression at P1 and P8 in epithelial cells located in the distal tip regions of the elongating prostate buds (Fig. 2M–P); of interest, GFP was not detected in all distal epithelial cells, suggesting heterogeneity among this epithelial population (Fig. 2O,P). Finally, to determine whether cells that receive canonical Wnt signals also express Nkx3.1, we examined P4 prostates from TCF/Lef:H2B-GFP neonates and observed co-expression of GFP and Nkx3.1, with all GFP-expressing prostate epithelial cells also expressing Nkx3.1 (Fig. 2Q,R). These findings indicate that canonical Wnt signaling is active in both mesenchyme and epithelium in prostate organogenesis, and suggest distinct cell-type specific roles for Wnt signals at early stages during bud initiation and at later stages during ductal elongation.

Figure 2.

TCF/Lef:H2B-GFP Wnt reporter expression during prostate development. A–J: Whole-mount images of bright field (A–E) and direct visualization of TCF/Lef:H2B-GFP expression (F–J) at 17.5 days postcoitum (dpc) (A,F), postnatal day (P) 1 (B,G), and P8 (C–E, H–J). Brackets (F) indicate mesenchymal green fluorescent fluid (GFP) expression, and arrowheads (G–J) indicate GFP expression in prostate buds. K–P: Sections of TCF/Lef:H2B-GFP-expressing prostates at 15.5 dpc (K), 17.5 dpc (L), P1 (M–O), and P8 (P) immunostained for GFP (green) and E-cadherin (Cdh1; red). Arrows (M,N) indicate GFP expression at prostate bud tips. Q,R: Expression of Nkx3.1 and TCF/Lef:H2B-GFP in neonatal prostates. Q,R: Section from TCF/Lef:H2B-GFP prostate at P4 immunostained for GFP (green) and Nkx3.1 (red); panel R shows red channel only. Arrowheads indicate GFP-expressing epithelial cells that also express Nkx3.1. All sections were counterstained with DAPI (4′,6-diamidine-2-phenylidole-dihydrochloride). AP, anterior prostate; DD, ductus deferens; DLP, dorsal lateral prostate; DP, dorsal prostate; LP, lateral prostate; SV, seminal vesicle; UGE, urogenital sinus epithelium; UGM, urogenital sinus mesenchyme; VP, ventral prostate. Scale bars = 500 μm in A–J, 50 μm in K–R.

Canonical Wnt Signaling Is Required for Proper Nkx3.1 Expression and Prostate Bud Outgrowth

To address the function of canonical Wnt signaling during prostate development, we used an explant culture methodology to assess the effects of soluble factors on prostate formation and ductal outgrowth (Lopes et al., 1996; Berman et al., 2004; Doles et al., 2005). In this assay, we cultured urogenital sinus (UGS) explants from Nkx3.1lacZ/+ embryos at 15.5 dpc in defined medium for 7 days in the presence or absence of canonical Wnt pathway inhibitors, followed by X-gal staining to examine β-galactosidase expression. In control UGS explants, we found that β-galactosidase expression was undetectable at day 0, but was observed in prostate buds that had formed in explants cultured for 7 days (Fig. 3A). In contrast, treatment with the soluble canonical Wnt inhibitors Dkk1, Dkk2, or Dkk3 reduced the expression of β-galactosidase and the extent of prostate budding (Fig. 3A). Quantitation of the reduction in bud number as detected by β-galactosidase staining showed significant reduction of prostate budding in response to Dkk1, Dkk2, or Dkk3 relative to control explants (Fig. 3B). We confirmed the reduction in Nkx3.1 expression after inhibition of canonical Wnt signaling by Nkx3.1 immunostaining of sections from control and treated explants (Fig. 3C–E). Consistent with the reduced growth of UGS explants after treatment with Wnt inhibitors, we found that Ki67 immunostaining was reduced in Dkk1-treated explants, indicating defects in cell proliferation in response to canonical Wnt inhibition (Fig. 3F,G).

Figure 3.

Inhibition of canonical Wnt signaling affects prostate bud outgrowth and luminal epithelial differentiation in explant culture. A: X-gal staining of 15.5 days postcoitum (dpc) urogenital sinus explants from Nkx3.1lacZ/+ mice treated with Dkk1, Dkk2, or Dkk3 containing media compared with controls at 0 days and 7 days of culture. B: Quantitation of the number of epithelial buds after 7 days of culture and comparison to controls (n = 20 for each treatment); error bars represent one standard deviation. C–O: Immunostaining of sections from 15.5 dpc urogenital sinus explants cultured for 7 days in control media (C,F,H,J,L,N), Dkk1-containing media (D,G,I,K,M,O), or Dkk2-containing media (E) for Nkx3.1 and p63 (C–E), Ki67 (F,G), CK5 and Foxa1 (H,I), p63 and CK18 (J,K), p63 and CK8 (L,M), and AR and Cdh1 (N,O). Arrows show reduction of CK18 (J,K) and CK8 (L,M) expression following Dkk1 treatment. P: Percentage of epithelial cells in sections of prostate buds that express p63 (n = 9 explants), Nkx3.1 (n = 9), Ki67 (n = 9), or CK18 (n = 2); error bars represent one standard deviation. d, day. Scale bars = 500 μm in A, 50 μm in C–O.

In parallel experiments, we treated UGS explants with IWR-1, a small molecule inhibitor of canonical Wnt signaling (Karner et al., 2010). First, we performed a time course analysis using Nkx3.1lacZ/+ UGS explants cultured in the presence of IWR-1, and observed similar defects in prostate budding and β-galactosidase expression (Fig. 4A–K). Second, we confirmed that treatment of UGS explants with IWR-1 inhibited expression of the TCF/Lef:H2B-GFP reporter in whole-mounts, both in the seminal vesicles and prostate bud tips (Fig. 4L–O,Q–T), which was confirmed by GFP immunostaining in sections (Fig. 4P,U). Taken together, these findings indicate that canonical Wnt signals are required for proper Nkx3.1 expression and prostate bud growth in UGS explant cultures.

Figure 4.

Inhibition of canonical Wnt signaling with IWR-1 also affects prostate bud outgrowth. A–E,G–K: Urogenital sinus explants from Nkx3.1lacZ/+ mice cultured in control (A–E) or IWR-1 containing media (G–K) for 7 days. X-gal stained images of explants in panels D, J, are shown in panels E, K. F: Quantitation of the number of epithelial buds after 7 days of culture and comparison to controls (n = 6 for each treatment); error bars represent one standard deviation. L–U: TCF/Lef:H2B-GFP expression in cultured 15.5 days postcoitum (dpc) urogenital sinus explants. Time course with bright-field images (L,N,Q,S), direct imaging of GFP expression (M,O,R,T), and green fluorescent protein (GFP) immunostaining of sections (P,U) of explants cultured in control media (L–P) or media containing 100 μM IWR-1 (Q–U). Arrowheads (O) indicate GFP-expressing cells at prostate bud tips. d, day; SV, seminal vesicle. Scale bars = 500 μm in A–E,G–K,L–O,Q–T, 50 μm in P,U.

To determine whether inhibition of canonical Wnt signaling affected the differentiation of prostate epithelial cells, we examined the expression of basal and luminal epithelial markers in Dkk1-treated explants. We found that Foxa1, which is expressed in all urogenital epithelial cells (Gao et al., 2005), as well as the basal markers cytokeratin 5 (CK5) and p63 were expressed in wild-type patterns in explants treated with Dkk1 (Fig. 3H–M). In contrast, the expression of the luminal markers cytokeratin 18 (CK18), cytokeratin 8 (CK8), and androgen receptor (AR) were reduced in Dkk1-treated explants (Fig. 3J–O). Quantitation of these immunostaining data showed significant differences in the percentages of epithelial cells that expressed Nkx3.1 or CK18, or were positive for Ki67 immunoreactivity (Fig. 3P). Furthermore, we found that the percentage of epithelial cells co-stained for Ki67 and p63 was similar between control and Dkk1-treated explants (2.58 ± 0.20% vs. 2.13 ± 0.29%), indicating that the observed decrease in cell proliferation occurs primarily in luminal cells. Taken together, these findings suggest that inhibition of canonical Wnt signals results in decreased proliferation and impaired differentiation of luminal cells.

Nkx3.1 Regulates Canonical Wnt Signaling in Prostate Epithelial Bud Tips

Finally, because Nkx3.1 expression and cells receiving canonical Wnt signals are colocalized at prostate ductal tips, we investigated whether Nkx3.1 might regulate canonical Wnt signaling in the developing prostate. We compared the expression of the TCF/Lef:H2B-GFP transgene in control wild-type prostates with its expression in Nkx3.1−/− mutant prostates. In Nkx3.1−/− neonates at P1, we found that GFP expression in the seminal vesicles and ductus deferens was normal (Fig. 5A–D), consistent with the lack of Nkx3.1 expression in these tissues (Bhatia-Gaur et al., 1999). However, we observed a variable reduction of GFP expression in prostate bud tips in Nkx3.1−/− neonates that is not statistically significant (Fig. 5E–H,O). In contrast, at the later P4 stage, we found a significant reduction in GFP expression in Nkx3.1−/− prostates, evident in both whole-mounts and sections (Fig. 5I–O). These findings indicate that Nkx3.1 is required after the initial stages of prostate organogenesis for proper levels of canonical Wnt signaling activity in prostate epithelial cells.

Figure 5.

TCF/Lef:H2B-GFP expression is reduced in neonatal Nkx3.1−/− prostate epithelial cells. A–N: Images of whole-mounts viewed under bright-field (A–B,I–J), or by epifluorescence for GFP expression (C–F,K,L), or sections (G,H,M,N) from TCF/Lef:H2B-GFP-expressing wild-type (A,C,E,I,K) and Nkx3.1−/− (B,D,F,J,L) neonates. A–F,I–N: Urogenital regions at postnatal day (P) 1 (A–F, dorsal view) and P4 anterior prostate (I–N) are shown. Boxed regions in panels C,D are shown at higher magnification in panels E,F; arrowheads indicate clusters of cells expressing TCF/Lef:H2B-GFP at prostate ductal tips. G–H,M–N: Sections were stained using anti-GFP (green) and anti-Cdh1 (red) antibodies and counterstained with DAPI (4′,6-diamidine-2-phenylidole-dihydrochloride). O: Percentage of epithelial cells in sections of prostate buds that express GFP (n = 3 animals); error bars represent one standard deviation. DD, ductus deferens; SV, seminal vesicle; WT, wild-type. Scale bars = 500 μm in A–D,I–L, 100 μm in E,F, 50 μm in G,H,M,N.

DISCUSSION

In this study, we have shown that canonical Wnt signaling activity is active in both mesenchymal and epithelial compartments during prostate bud formation and early ductal outgrowth and branching. Using an explant culture assay together with highly sensitive mouse reporters to detect Nkx3.1 expression and canonical Wnt signaling responses, we have found that inhibition of canonical Wnt signaling disrupts prostate budding. These defects in bud formation and outgrowth are associated with decreased cellular proliferation, reduced expression of Nkx3.1, and impaired luminal differentiation in the prostate epithelium. Furthermore, we have found that canonical Wnt signaling is active at ductal tips during prostate ductal outgrowth, where it is also colocalized with Nkx3.1 expression, while loss of Nkx3.1 function leads to reduction or absence of canonical Wnt signaling activity at ductal tips.

Based on our findings, we propose the following schematic interpretation of our findings on canonical Wnt and Nkx3.1 function in prostate development (Fig. 6). Before initial prostate budding at 17.5 dpc, canonical Wnt activity is present in the urogenital sinus mesenchyme, but these Wnt signals are not received by the urogenital sinus epithelium. After the prostate buds have begun to form, canonical Wnt signals from the mesenchyme directly or indirectly up-regulate Nkx3.1 expression as well as epithelial Wnt activity at the nascent bud tips. At neonatal stages, Nkx3.1 and canonical Wnt signals in the epithelium create a positive feedback loop to promote branching morphogenesis and luminal differentiation within the elongating prostate buds. Notably, such a model is consistent with the known functions of Nkx3.1 in ductal branching and luminal differentiation (Bhatia-Gaur et al., 1999). Moreover, canonical Wnt signaling at prostate bud tips is likely to regulate ductal outgrowth and branching, consistent with its similar roles in other organ systems including developing kidney (Bridgewater et al., 2008), lung (Cardoso and Lu, 2006), salivary gland (Haara et al., 2011; Patel et al., 2011), and mammary gland (Sternlicht et al., 2006).

Figure 6.

Proposed roles of canonical Wnt signaling during prostate development. Before prostate bud formation, canonical Wnt signaling activity (green) is detected only in the urogenital mesenchyme and not the urogenital epithelium. After bud formation, canonical Wnt signals from the mesenchyme activate epithelial Nkx3.1 expression, and also up-regulate canonical Wnt signaling activity in prostate bud epithelial cells (blue arrows). At neonatal bud tips, Nkx3.1 functions together with canonical Wnt signals in a positive feedback loop to regulate ductal branching and luminal differentiation. dpc, days postcoitum; P, postnatal day; UGE, urogenital epithelium; UGM, urogenital mesenchyme.

In support of a role for canonical Wnt signaling in branching morphogenesis, we have observed high levels of Wnt reporter activity in prostate bud tips during the first week of neonatal prostate development, when the numbers of distal tips and branch points increase at the highest rate (Sugimura et al., 1986). Also consistent with this model, deletion of β-catenin in epithelial cells in organ culture or treatment of P2 rat prostates with Dkk1 in organ culture results in reduced ductal branching, although with a decreased number of basal epithelial cells (Wang et al., 2008; Francis et al., 2013). Furthermore, we have found that Wnt reporter activity was reduced in prostate bud tips of Nkx3.1−/− prostates, in which the number of prostate ducts is decreased compared with wild-type (Bhatia-Gaur et al., 1999), indicating that Nkx3.1 can regulate canonical Wnt activity at bud tips during prostate organogenesis. Notably, because not all Nkx3.1-expressing cells also express the TCF/Lef:H2B-GFP reporter, it is likely that canonical Wnt signals are not required cell autonomously for maintenance of Nkx3.1 expression.

Our analyses of canonical Wnt signaling and Nkx3.1 expression have been facilitated by two highly sensitive mouse reagents. First, to follow Nkx3.1 expression in urogenital sinus explants, we have generated a novel targeted Nkx3.1lacZ knock-in line in which β-galactosidase expression recapitulates endogenous Nkx3.1 expression. Interestingly, an independent Nkx3.1lacZ knock-in line has been reported to display β-galactosidase expression in several tissues that do not express Nkx3.1 (Schneider et al., 2000), which may be due to position effects from an unexcised PGK-neo selection cassette, similar to reported effects of unexcised selection cassettes at several distinct targeted loci (Olson et al., 1996; Pham et al., 1996; Moran et al., 1999).

Second, we have used a highly sensitive reporter for canonical Wnt signaling in the TCF/Lef:H2B-GFP transgenic line (Ferrer-Vaquer et al., 2010) to visualize the dynamic distribution of canonical Wnt signaling activity in vivo as well as in urogenital sinus explants. In particular, the sensitivity of this Wnt reporter mouse considerably extends previous findings in the literature. For example, we have been able to detect canonical Wnt signaling activity in the urogenital sinus mesenchyme before prostate formation, whereas previous studies using the BAT-gal reporter could only detect β-galactosidase activity in prostate epithelial cells at postnatal day 5 (Wang et al., 2008). Importantly, we believe that the TCF/Lef:H2B-GFP reporter accurately recapitulates the distribution of canonical Wnt activity because the pattern of GFP expression during prostate organogenesis matches the published description of Axin2 as well as Lef1 expression in ductal bud tips (Mehta et al., 2011; Wu et al., 2011; Francis et al., 2013). Despite the sensitivity of the TCF/Lef:H2B-GFP reporter for canonical Wnt activity, we have not detected GFP expression in urogenital sinus epithelium at 15.5 or 17.5 dpc, suggesting that Wnt signals may not be received by epithelial cells before prostate bud formation. In contrast, Axin2 expression in urogenital sinus epithelium at 16.5 and 17.5 dpc has been reported by quantitative reverse transcriptase polymerase chain reaction (RT-PCR; Simons et al., 2012) and by low-resolution whole-mount in situ hybridization (Mehta et al., 2011); however, Axin2 mRNA expression may not always correspond to canonical Wnt signaling activity (Jho et al., 2002).

Notably, recent work has used organ culture experiments to show that inducible deletion of β-catenin in both urogenital epithelial and mesenchyme cells results in failure of prostate budding and lack of Nkx3.1 expression, indicating a role in prostate specification (Simons et al., 2012). Two other recent studies have shown that inducible deletion of β-catenin specifically in the urogenital sinus epithelium leads to highly impaired budding and loss of Nkx3.1 expression in organ culture and in vivo (Francis et al., 2013; Mehta et al., 2013). These results are consistent with our finding that Nkx3.1 expression is greatly reduced following treatment of UGS explants with soluble Wnt inhibitors, which likely do not completely inhibit canonical Wnt signaling in our organ culture experiments. However, our observations on the absence of TCF/Lef:H2B-GFP reporter activity in the urogenital epithelium before prostate formation suggest that mesenchymal Wnt signaling is not required for prostate budding per se. Instead, we favor the hypothesis that mesenchymal and epithelial Wnt signals are likely to be necessary sequentially for prostate bud outgrowth and subsequent events in ductal branching and luminal epithelial differentiation (Fig. 6), such that the absence of canonical Wnt signaling in both compartments or in the epithelial compartment alone would result in defective bud growth and loss of Nkx3.1.

Our study provides new insights into the relationship of canonical Wnt signaling to androgen receptor (AR) function during prostate formation. In particular, AR is initially required in the urogenital mesenchyme to produce signals for prostate induction and growth, and is only necessary later in the epithelium for the secretory function of differentiated luminal epithelial cells (Cunha et al., 1987; Prins and Putz, 2008). Consequently, the current model for prostate induction involves the activation of AR by testicular androgens in the urogenital sinus mesenchyme, which in turn leads to production and/or activation of paracrine signals that act upon the urogenital epithelium to induce prostate budding (Cunha, 2008; Thomson, 2008). Such paracrine inductive signals have been termed andromedins, but the identities of candidate andromedins have remained unclear to date (Tenniswood, 1986; Thomson, 2008). Our results suggest that canonical Wnt signals from the mesenchyme might not function as true andromedins by themselves, but might act cooperatively with other mesenchymal signals during prostate induction and specification. This interpretation is also consistent with the observation that inducible overexpression of β-catenin in the urogenital sinus epithelium is not sufficient to induce Nkx3.1 expression or prostate bud formation (Mehta et al., 2013). Mesenchymal signals that might cooperate with canonical Wnt signals could include FGF10, which is expressed in the urogenital mesenchyme and is required for prostate formation, but does not itself appear to represent a sexually dimorphic signal (Lu et al., 1999; Thomson and Cunha, 1999; Donjacour et al., 2003; Pu et al., 2007; Thomson, 2008).

Finally, the epithelial–mesenchymal interactions that are critical for prostate budding and ductal morphogenesis during organogenesis are likely to be deranged during prostate tumor progression, as has been suggested by gene expression profiling studies (Schaeffer et al., 2008; Pritchard et al., 2009). For example, the ability of canonical Wnt signals to mediate communication between epithelial and mesenchymal cells and regulate Nkx3.1 expression may provide insights into mechanisms for how canonical Wnt signaling may be activated in castration-resistant prostate cancer (Kypta and Waxman, 2012). Notably, increased expression of Wnt16b in the tumor microenvironment enhances treatment resistance in human prostate tumors (Sun et al., 2012). Therefore, a greater understanding of canonical Wnt pathway function in normal prostate development may provide insights into the deregulation of this critical signaling pathway in prostate tumorigenesis.

EXPERIMENTAL PROCEDURES

Gene Targeting

The targeting vector for the Nkx3.1lacZ allele was constructed using a 3′ arm that was a 4.1 kb PCR fragment from a Nkx3.1 genomic clone (Bhatia-Gaur et al., 1999), and a 5′ arm that was a 3.5 kb PCR fragment up to the translation initiation site of Nkx3.1 fused to the 5′ end of the promoter-less SDK-lacZ insert in the pSDKlacZpA vector (Shalaby et al., 1995). Positive selection was provided by the self-excising ACE-Cre/PolII-neo selection cassette from the pACN vector (Bunting et al., 1999), while negative selection was conferred by the PGK-tk cassette from the pPNT vector (Tybulewicz et al., 1991). Primers to amplify the 3′ arm were: 5′-CTA GTC TAG AGC GGC TCA CCT CCT TCC TCA-3′ and 5′-CTA GTC TAG AGG ATG GCA GGA GAG GTC ACT GC-3′; primers for generating the 5′ arm were: 5′-ACC GGA ATT CTC CGC TGC GCG CCG CTT TTG C-3′ and 5′-ACC CAA GCT TCA TGC CTG CAG GTC GGA GGC C-3′. Culture and transfection of mouse ES cells followed standard protocols (Nagy et al., 2003). Homologous recombinants in TC1 ES cells (Deng et al., 1996) were selected by positive-negative selection followed by Southern blot screening. Of 250 clones analyzed, we found proper targeting in a single clone, which was used for generation of germline chimeras by blastocyst injection.

Mouse Lines

Nkx3.1tm2(lacZ)Mms (Nkx3.1lacZ) mice were genotyped using lacZ primers 5′-CCG ACG GCA CGC TGA TTG AAG-3′ and 5′-TGC ACC GGG CGG GAA GGA T-3′. TCF/Lef:H2B-GFP mice (Ferrer-Vaquer et al., 2010) were obtained from Kat Hadjantonakis. Nkx3.1−/− mice have been previously described (Bhatia-Gaur et al., 1999). All mice were maintained on a C57/BL6 strain background.

UROGENITAL SINUS EXPLANT CULTURE

For explant culture experiments, 15.5 dpc urogenital sinus were obtained from timed matings and dissected and cultured for 7 days at 37°C with 5% CO2 on Millicell Culture Plate Inserts (Millipore) in Ham's F-12/DMEM (50:50) supplemented with Insulin-Transferrin-Sodium Selenite Supplement (Roche) and 1 × 10−8 M dihydrotestosterone (Sigma). Conditioned media containing Dkk1, Dkk2, or Dkk3 were generated by transfection of 293T cells with Lipofectamine (Invitrogen), using previously described expression plasmids (Mao et al., 2001). After growth for 4 days, conditioned media were harvested, centrifuged, and aliquots were stored at −80°C. For IWR-1 treatments, explants were cultured with 100 μM IWR-1 (Sigma) in 0.5% demethyl sulfoxide (DMSO); control explants were cultured in 0.5% DMSO. Media were changed every other day. Statistical analyses were conducted using the Student's t-test.

Immunofluorescence and β-Galactosidase Staining

Dissected prostates and tissue explants were fixed in 4% paraformaldehyde in phosphate buffered saline (PBS) at 4°C for 1 hr, placed on a 30% sucrose gradient overnight, and then a 1:1 30% sucrose:OCT mixture, followed by embedding in OCT. Antibody staining was conducted using conditions and primary antibodies as previously described (Wang et al., 2009). Primary antibodies used were anti-AR (Sigma 1:500), anti-Cdh1 (Cell Signaling; 1:200), anti-CK5 (Covance 1:500), anti-CK8 (Abcam 1:500), anti-CK18 (Abcam 1:100), anti-Foxa1 (Abcam; 1:500), anti-GFP (Abcam 1:1,000), anti-Ki67 (Dako 1:600), anti-Nkx3.1 (Kim et al., 2002b; 1:1,000) and anti-p63 (Santa Cruz 1:50). Fluorescent-conjugated secondary antibodies (Molecular Probes) were used at a dilution of 1:250 or 1:500. Slides were mounted using Vectashield Mounting Medium with DAPI (4′,6-diamidine-2-phenylidole-dihydrochloride; Vector Laboratories).

For analysis of lacZ expression, tissues were fixed in 4% paraformaldehyde in PBS at 4°C for 1 hr, and stained for β-galactosidase activity overnight at 4°C.

ACKNOWLEDGMENTS

We thank Dr. Kat Hadjantonakis for generously providing TCF/Lef:H2B-GFP mice, Dr. Christof Niehrs for Dkk1, Dkk2, and Dkk3 plasmids, Alessandro Pizzo for assistance with immunostaining, and members of the Shen laboratory for their comments and support. M.M.S. was supported by the NIH. In addition, M.K.-deJ. was supported by DOD Prostate Cancer Research Program postdoctoral award and M.S. received a NIH training grant.

Ancillary