RET-mediated glial cell line-derived neurotrophic factor signaling inhibits mouse prostate development

The prostate gland is formed from the embryonic urogenital sinus (UGS), where the pelvic urethra joins the bladder. Morphogenesis of the mouse prostate begins in the pelvic urethra of the UGS around embryonic day (E) 16.5 and is dependent on the inductive properties of the urethral mesenchyme (UrM) (Georgas et al., 2015). In rodents, the UrM is composed of discrete spatial regions with different properties. The peri-urethral mesenchyme (PUM) surrounds the urethral epithelium (UrE) and newly formed prostatic buds. The inductive UrM, which includes the ventral mesenchymal pad (VMP), is peripheral to the PUM and is crucial for prostatic induction (Thomson et al., 2002; Timms et al., 1995, 1994). During induction, mesenchymal cells at the interface of the PUM and VMP differentiate into smooth muscle cells, which surround the prostate epithelial (PrE) ducts as they elongate in the multilobed rodent prostate (Hayward et al., 1998; Thomson et al., 2002; Timms et al., 1995, 1994).

Androgen receptor (AR) signaling is required for prostate development (Cunha and Chung, 1981; Cunha and Lung, 1978; Donjacour and Cunha, 1988; Lasnitzki and Mizuno, 1980), and disruption of AR-mediated prostate development predisposes humans and rodents to prostate neoplasia by altering the phenotype of the gland early in life (Anway and Skinner, 2008; Gupta, 2000; Ho et al., 2006; Hu et al., 2012; Prins et al., 2014, 2008; Ramos et al., 2001; Timms et al., 2005). Testosterone and its potent metabolite 5α-dihydrotestosterone (DHT) activate the AR during prostate development. AR action in the inductive UrM is thought to induce formation of PrE progenitors/buds in the UrE through paracrine signaling (Cunha and Chung, 1981; Cunha and Lung, 1978; Lasnitzki and Mizuno, 1980). Androgen-dependent signaling also stimulates outgrowth, cytodifferentiation and branching morphogenesis of the PrE buds as they give rise to a network of prostatic ducts (Cunha and Chung, 1981; Cunha and Lung, 1978; Donjacour and Cunha, 1988; Sugimura et al., 1986).

AR signaling acts in concert with other signaling pathways to regulate prostate development. Outgrowth of the PrE is dictated by time-specific and cell-specific expression of morphogenetic genes (Prins and Putz, 2008; Thomson and Marker, 2006), including secreted signaling ligands of the hedgehog (Berman et al., 2004; Doles et al., 2006; Freestone et al., 2003; Lamm et al., 2002; Podlasek et al., 1999; Pu et al., 2004; Wang et al., 2003), Wnt (Allgeier et al., 2008; Huang et al., 2009; Joesting et al., 2008; Keil et al., 2012), FGF (Alarid et al., 1994; Donjacour et al., 2003; Sugimura et al., 1996; Thomson and Cunha, 1999) and TGFβ (Cancilla et al., 2001; Cook et al., 2007; Grishina et al., 2005; Itoh et al., 1998; Lamm et al., 2001; Tanji et al., 1994; Tomlinson et al., 2004) gene super-families. These morphogens are thought to communicate autocrine and paracrine signals through their cognate receptors and downstream signaling pathways. However, the nature of the AR-regulated morphogenetic genes is not understood.

Glial cell line-derived neurotrophic factor (GDNF) is a member of the TGFβ protein super-family and regulates spermatogenesis, neuronal survival and differentiation, and kidney development (Arenas et al., 1995; Davies et al., 1995; Henderson et al., 1994; Lin et al., 1993; Meng et al., 2000; Pepicelli et al., 1997; Pichel et al., 1996; Sainio et al., 1997; Vega et al., 1996). Cellular responses to GDNF are mediated by a multicomponent receptor complex consisting of GDNF family receptor α1 (GFRα1) and RET, a receptor tyrosine kinase (Jing et al., 1996; Trupp et al., 1996; Vega et al., 1996). GDNF signaling plays crucial roles in development; however, its role in prostate development has not been explored previously.

We recently reported the localization of GDNF, GFRα1 and RET in the murine UGS. Prior to prostatic induction, GDNF and GFRα1 are expressed in the urethral mesenchyme (UrM) and epithelium (UrE), whereas RET is restricted to the UrM, including the VMP (Park and Bolton, 2015). In addition, we demonstrated that GDNF increases proliferation of UrM cells and UrE cells in naïve UGSs, which were not androgen induced (Park and Bolton, 2015). Our previous findings suggest that GDNF signaling may influence additional aspects of UGS development. Based on our published data and on the requirement of AR signaling in prostatic induction and development, we explore the intriguing hypothesis that the AR and GDNF signaling pathways may interact and influence androgen-induced prostate development.

Here, we report that GDNF signaling antagonizes androgen-induced prostate morphogenesis. In support of this, we show that androgen induces PrE budding, proliferation and cytodifferentiation in the developing prostate, whereas GDNF signaling inhibits these androgen-dependent effects. Using Ret-disrupted mice, we further demonstrate that RET-mediated signaling in the UrM is responsible for the inhibitory effects of GDNF on prostate development, including suppression of androgen-induced proliferation and differentiation of PrE cells. In addition to Ar being GDNF-repressed by RET-mediated signaling, Gdnf and Gfra1 are androgen repressed during prostate development, thus establishing reciprocal regulatory crosstalk between AR and GDNF signaling in androgen-induced prostate development.

We investigated the functional roles of AR and GDNF signaling in androgen-induced prostate development using UGS organ culture, which recapitulates the earliest stages of prostate development in vivo (Doles et al., 2005; Ghosh et al., 2011; Kuslak and Marker, 2007). As AR activation is necessary for prostate development in vivo and ex vivo (Allgeier et al., 2010; Cunha and Chung, 1981; Cunha and Lung, 1978; Doles et al., 2005; Marker et al., 2003), we cultured UGSs from E15.5 male embryos in defined organ culture medium supplemented with 10 nM DHT for 7 days. These UGSs were also co-treated with 10-100 ng/ml GDNF or vehicle as a control. As expected, DHT induced PrE budding in UGSs cultured in the absence of GDNF (Fig. 1A), which is consistent with previous reports of prostate development ex vivo (Allgeier et al., 2010; Doles et al., 2005; Marker et al., 2003). UGSs co-treated with GDNF displayed thicker UrM (Fig. 1A), which is consistent with the effect of GDNF in naïve UGSs (Park and Bolton, 2015). In addition, the co-treated UGSs showed striking changes in morphology, including significant decreases in the number and length of PrE buds that emerged from the UrE (Fig. 1B,C).

To assess further the PrE-budding defect in the developing prostate rudiments caused by GDNF, we used immunohistochemistry to visualize the outgrowth of the ventral and lateral PrE buds. Immunodetection of epithelial keratins has been shown to identify the epithelial cells of the UrE and PrE buds and ducts, whereas vimentin localization identifies mesenchymal cells. Consistent with our PrE-budding data (Fig. 1B,C), the number and length of PrE buds that elongated from the UrE were dramatically decreased in DHT-induced UGSs co-treated with GDNF compared with vehicle (Fig. 1D). Thicker UrM were also evident in UGSs co-treated with GDNF (Fig. 1D). Taken together, these results demonstrate that GDNF signaling inhibits androgen-induced prostate development.

In a previous study, we demonstrated that UrM thickness in the naïve UGS is selectively regulated by GFRα1 and RET kinase activity in response to GDNF (Park and Bolton, 2015). Here, we investigated the effects of androgen and GDNF on proliferation of mesenchymal cells and epithelial cells in the developing prostate by quantifying bromodeoxyuridine (BrdU) incorporation in the DNA of dividing cells. Androgen-supplemented E15.5 UGSs were cultured with 100 ng/ml GDNF or vehicle for 2-7 days, and BrdU was added before tissue processing for immunohistochemistry.

Immunodetection of BrdU revealed an increase in BrdU-positive UrM cells and a decrease in BrdU-positive PrE cells in androgen-induced UGSs co-treated with GDNF for 2 days (Fig. 2A). Quantification of the percentage of BrdU-positive cells in the UrM and the PrE outgrowths from the UrE indeed demonstrates a robust increase in proliferation of UrM cells and a substantial decrease in proliferation of PrE cells in DHT-induced UGSs co-treated with GDNF compared with vehicle (Fig. 2B). Notably, GDNF induced proliferation to a greater extent in the peripheral, inductive UrM compared with the PUM (Fig. 2A).

DHT-induced UGSs co-treated with GDNF for 7 days displayed thicker UrM (Fig. 2C). However, the percentage of BrdU-positive UrM cells was similar in UGSs co-treated with GDNF or vehicle (Fig. 2D). In comparison with the robust increase of BrdU-positive UrM cells in UGSs co-treated with DHT and GDNF on day 2 (Fig. 2B), the results on day 7 indicate that GDNF-induced proliferation of UrM cells is time-sensitive during prostate development (compare Fig. 2B,D). Quantification of BrdU-positive cells in the PrE outgrowths from the UrE showed prolonged suppression of DHT-induced proliferation in UGSs co-treated with GDNF for 7 days (Fig. 2D). Although PrE growth suppression was less pronounced on day 7 compared with day 2 (compare Fig. 2B,D). These findings indicate that elevated levels of GDNF induce morphological changes in the developing prostate that are driven by enhanced proliferation of UrM cells and inhibited proliferation of PrE cells.

To begin to investigate the cellular and molecular mechanism responsible for the effects of GDNF on prostate development, we examined the expression and localization of RET, which forms a co-receptor complex with GFRα1 and mediates GDNF signaling (Jing et al., 1996; Trupp et al., 1996; Vega et al., 1996). Using real-time quantitative polymerase chain reaction (QPCR) analysis, we determined the extent of RET regulation by GDNF signaling in androgen-induced prostate development. Ret mRNA increased substantially in DHT-induced UGSs co-treated with GDNF compared with vehicle (Fig. S1A). Transcripts coding for sprouty 1 (Spry1), an antagonistic regulator of RET-mediated signaling, also increased in co-treated UGSs (Fig. S1B). Consistent with the increase in Ret mRNA, GDNF exposure increased RET protein expression (Fig. S1C). Immunohistochemistry analysis revealed a robust increase in the number of UrM cells expressing RET and a significant increase in RET expression per cell in the VMP of DHT-induced UGSs co-treated with GDNF compared with vehicle (Fig. S1D,E). In contrast, RET localization was absent from the UrE and PrE (Fig. S1D). The increases in RET expression and localization suggest that downstream targets of GDNF signaling in the UrM may regulate transcription of the Ret gene.

In addition, we examined RET localization in the prostate of postnatal day (P) 0 neonates. Immunodetection of fibroblast growth factor 10 (FGF10), which is expressed and secreted from the inductive UrM cells of the VMP (Thomson and Cunha, 1999), and RET revealed robust expression and co-localization of RET and FGF10 in inductive UrM cells of the VMP (Fig. S1F). RET localization was absent from UrE and PrE regions in P0 prostate (Fig. S1F). The colocalization of RET and FGF10 in the VMP demonstrates that RET is indeed expressed in inductive UrM cells during prostate development.

During prostatic induction, mesenchymal cells at the interface of the PUM and VMP undergo smooth muscle differentiation, which has been suggested to influence PrE budding and outgrowth from the UrE (Hayward et al., 1998; Thomson et al., 2002; Timms et al., 1995, 1994). To determine whether GDNF signaling affects smooth muscle differentiation in androgen-supplemented UGSs, we examined the localization and expression of smooth muscle actin (SMA), a marker of smooth muscle cells. Immunohistochemistry revealed layers of smooth muscle cells encircle the ventral PrE in DHT-induced E15.5 UGSs regardless of whether they were co-treated with GDNF or vehicle for 7 days (Fig. S2A). Immunoblot analysis of DHT-induced UGSs also demonstrates that smooth muscle actin expression was unaffected by GDNF exposure for 2-7 days (Fig. S2B). Additional immunohistochemistry analysis revealed similar numbers of UrM cells expressing SMA and no significant difference in SMA expression per cell at sites of prostatic budding in DHT-induced UGSs co-treated with GDNF compared with vehicle (Fig. S2C,D). These results indicate that smooth muscle differentiation during prostate development is unaffected by GDNF signaling.

Androgen-dependent signaling during prostate development stimulates outgrowth and cytodifferentiation of PrE buds as they give rise to a network of prostatic ducts (Cunha and Chung, 1981; Cunha and Lung, 1978; Donjacour and Cunha, 1988; Sugimura et al., 1986). Since GDNF antagonizes androgen-induced PrE budding, we sought to determine whether GDNF influences cytodifferentiation of PrE buds formed in the presence of GDNF. We examined PrE cytodifferentiation based on expression and localization of Keratins 8 and 18 (KRT8 and KRT18), which are abundantly expressed in luminal epithelial cells of the prostate (Achtstätter et al., 1985; Hsieh et al., 1992; Soeffing and Timms, 1995; Verhagen et al., 1988). In baseline experiments, we induced prostate development in E15.5 UGSs with DHT for 0-7 days and assessed expression and localization of KRT5 (a basal epithelial keratin) and KRT8 (a luminal epithelial keratin) using immunohistochemistry (Fig. 3A). As expected, colocalization of KRT5 and KRT8 (orange) was observed in the UrE throughout the time course, and KRT5 localization was abundant in PrE bud cells that emerged from the UrE (Fig. 3A). By day 7 in DHT-induced UGSs, abundant KRT8 localization was evident in the central epithelial cells of elongated prostatic buds, and KRT5 localization persisted (Fig. 3A).

Next, we cultured androgen-supplemented E15.5 UGSs with GDNF or vehicle for 7 days, and assessed expression and localization of KRT5 and KRT8 (Fig. 3B). In DHT-induced UGSs co-treated with vehicle, abundant localization of KRT5 and KRT8 was evident in the central epithelial cells of the prostatic buds. In contrast, the small prostatic buds formed in UGSs co-treated with GDNF displayed abundant KRT5 localization, but they lacked KRT8 (Fig. 3B). Consistent with our immunohistochemistry data, QPCR analysis revealed significant downregulation of luminal Krt8 and Krt18 mRNAs, and upregulation of basal Krt14 and Krt5 mRNAs in DHT-induced UGSs co-treated with GDNF compared with vehicle (Fig. 3C). Taken together, these results suggest that GDNF signaling inhibits AR-mediated cytodifferentiation of luminal PrE cells.

In addition to RET-mediated signaling by GDNF, RET-independent mechanisms of GDNF action have also been reported (Cao et al., 2008; Panicker et al., 2003; Paratcha et al., 2003; Poteryaev et al., 1999; Trupp et al., 1999). To determine the functional role of RET-mediated signaling in androgen-induced prostate development, we assessed RET protein expression and examined PrE budding using UGSs from wild-type (Ret WT) and Ret-null (Ret KO) embryos. Notably, Ret KO mice die at birth due to complications associated with renal agenesis (Enomoto et al., 2001; Jain et al., 2004; Schuchardt et al., 1994). The Ret WT and mutant genotypes of embryos were distinguished using PCR (Fig. 4A). Immunoblot analysis of DHT-induced E15.5 UGSs confirmed the presence of RET in Ret WT UGSs and an absence of RET in Ret KO UGSs (Fig. 4B).

To determine the role of RET in PrE budding, androgen-supplemented E15.5 UGSs from Ret WT males and Ret KO males were co-treated with GDNF or vehicle for 7 days (Fig. 4C). DHT induced PrE budding in Ret WT UGSs and Ret KO UGSs co-treated with vehicle, indicating that RET-mediated signaling is not required for prostatic induction and PrE budding (Fig. 4C). PrE-budding deficiency and thicker UrM were observed in DHT-induced Ret WT UGSs co-treated with GDNF (Fig. 4C). Further, quantification of PrE budding showed significant decreases in PrE bud number and length in Ret WT UGSs co-treated with GDNF compared with vehicle (Fig. 4D,E). However, these effects of GDNF were absent from DHT-induced Ret KO UGSs exposed to GDNF (Fig. 4C-E).

Using immunohistochemistry, we also compared the outgrowth of PrE buds from Ret WT UGSs and Ret KO UGSs (Fig. 4F). Of the androgen-induced UGSs from Ret WT males and Ret KO males, only Ret WT UGSs co-treated with GDNF displayed thicker UrM (Fig. 4F). Consistent with our PrE-budding data (Fig. 4D,E), the number and length of PrE buds that elongated from the UrE decreased substantially in DHT-induced Ret WT UGSs co-treated with GDNF compared with vehicle. In contrast, PrE budding and UrM thickness of Ret KO UGSs were unaffected by GDNF exposure (Fig. 4F). Taken together, these results demonstrate that RET-mediated GDNF signaling inhibits androgen-induced PrE budding.

The role of RET signaling in proliferation of UrM and PrE cells was examined by immunohistochemistry analysis of androgen-supplemented E15.5 UGSs from Ret WT males and Ret KO males co-treated with GDNF or vehicle for 7 days (Fig. 5). Of the DHT-induced UGSs from Ret WT males and Ret KO males, only Ret WT UGSs co-treated with GDNF displayed thicker UrM (Fig. 5A), which is consistent with results in Fig. 4C,F. Quantification of BrdU-positive UrM cells revealed similar proportions of proliferating cells in Ret WT UGSs and Ret KO UGSs regardless of GDNF exposure (Fig. 5B), which is consistent with GDNF-induced proliferation of UrM cells being time-sensitive during prostate development (Fig. 2C,D). The percentage of BrdU-positive cells in the PrE outgrowths from the UrE showed prolonged suppression of DHT-induced proliferation in Ret WT UGSs co-treated with GDNF for 7 days (Fig. 5C). However, the inhibitory effect of GDNF on DHT-induced PrE proliferation was abolished in Ret KO UGSs (Fig. 5C). These data confirm that RET-mediated GDNF signaling is a positive regulator of UrM cell proliferation and a negative regulator of DHT-induced PrE cell proliferation in the developing prostate.

Using immunohistochemistry, we also determined the role of RET signaling in PrE cytodifferentiation during the outgrowth of PrE buds from DHT-supplemented Ret WT UGSs and Ret KO UGSs. KRT8-positive luminal epithelial cells were absent from the PrE buds of DHT-induced Ret WT UGSs co-treated with GDNF (Fig. 5C), which is consistent with GDNF inhibiting DHT-induced PrE cytodifferentiation (Fig. 3B). In contrast, KRT8-positive luminal epithelial cells were present at the midline of the PrE buds of Ret KO prostate rudiments regardless of GDNF exposure (Fig. 5C). These results demonstrate that RET-mediated GDNF signaling is a negative regulator of androgen-induced PrE cytodifferentiation.

As GDNF signaling inhibits androgen-induced prostate development, we hypothesized that GDNF signaling may influence Ar gene expression. We first assessed AR and RET protein expression in E15.5 UGSs, which were not cultured. AR protein is expressed in the UrM and to a lesser extent in the UrE of E15.5 UGSs, whereas RET expression is limited to the UrM (Fig. 6A). We then cultured androgen-supplemented E15.5 UGSs with GDNF or vehicle for 2-7 days, and assessed AR expression. In comparison to vehicle as a control, GDNF exposure decreased the expression of AR mRNA (Fig. 6B) and AR protein (Fig. 6C). Immunohistochemistry analysis revealed robust AR expression and nuclear localization in nearly all UrM cells of UGSs treated with DHT alone and the Ret-negative UrM cells of UGSs co-treated with DHT and GDNF (Fig. 6D). In contrast, decreases in AR expression and nuclear localization were evident in RET-positive VMP cells of UGSs co-treated with DHT and GDNF (Fig. 6D). These results indicate that GDNF signaling downregulates AR expression in androgen-induced UGSs, establishing a novel regulatory interaction between GDNF signaling and the AR.

To determine whether RET mediates downregulation of AR, we quantified AR expression in Ret WT UGSs and Ret KO UGSs cultured with DHT for 7 days to induce prostate development (Figs 3A, 4C). AR mRNA and protein expression were increased significantly in Ret KO prostate rudiments compared with Ret WT rudiments (Fig. 6E,F), thus implicating RET-dependent signaling in the downregulation of AR in the inductive UrM.

As GDNF signaling regulates Ar gene expression, we examined the responsiveness of Gdnf, Gfra1 and Ret to androgen in the embryonic UGS. Using QPCR analysis, we found that Gdnf and Gfra1 mRNAs were androgen-repressed in UGS culture, whereas Ret mRNA levels were unresponsive to androgen (Fig. 6G). These results suggest that expression of GDNF and its receptor, GFRα1, are regulated by AR signaling in DHT-induced UGSs. Taken together, our gene expression results establish reciprocal regulatory crosstalk between AR and GDNF signaling in androgen-induced prostate development (Fig. 7).

GDNF signaling plays crucial roles in development; however, its role in prostate development has not been explored previously. Here, we show for the first time that elevated levels of GDNF in the UGS inhibit androgen-induced prostate development. In addition to promoting the survival and differentiation of neurons in the nervous system (Arenas et al., 1995; Henderson et al., 1994; Lin et al., 1993), GDNF dose in the urogenital system regulates cell fate decisions of germline progenitors in testis and ureteric branching morphogenesis in kidney development. In mammalian seminiferous tubules, Sertoli cells dictate the microenvironment of the spermatogonial stem cell niche through release of GDNF. At low GDNF levels, the spermatogonial stem cells differentiate, whereas at high levels, the stem cells self-renew and are unable to differentiate (Meng et al., 2000). During formation of the permanent kidney, GDNF, which originates from the metanephric mesenchyme adjacent to the nephric duct epithelium, acts as a mesenchyme-derived paracrine signal to promote RET-mediated ureteric bud outgrowth and branching (Hellmich et al., 1996; Pachnis et al., 1993; Pepicelli et al., 1997; Pichel et al., 1996; Sainio et al., 1997; Vega et al., 1996). Somewhat analogous to the role of GDNF in spermatogonial stem cell self-renewal, our results demonstrate that RET-mediated GDNF signaling functions as a positive regulator of inductive UrM cell self-renewal and a negative regulator of PrE cytodifferentiation in androgen-induced prostate rudiments. The inhibitory effects of GDNF signaling on PrE proliferation and cytodifferentiation are also intriguing because they differ dramatically from the stimulatory effect that GDNF signaling has on ureteric bud outgrowth and branching during kidney morphogenesis.

GDNF-induced proliferation of UrM cells in androgen-induced UGSs (Fig. 2) is consistent with our previous report of cell-specific regulation of UrM proliferation by RET-mediated activation of the MEK-ERK pathway (Park and Bolton, 2015). Curiously, we found that GDNF-induced proliferation of UrM cells is time sensitive during androgen-induced prostate development (Fig. 2), despite persistent upregulation of RET (Fig. S1). Hence, we propose that chronic GDNF-induced proliferation of inductive UrM cells in the VMP is dependent on persistent activation of RET and the MEK-ERK pathway (Fig. 7). By day 7, an inhibitory feedback signal is likely activated, as Spry1 mRNA increased substantially in DHT-induced UGSs co-treated with GDNF (Fig. S1). Studies by Basson et al. have demonstrated that Spry1 is an antagonistic regulator of RET-mediated GDNF signaling during kidney morphogenesis (Basson et al., 2005, 2006). Although the mechanism whereby sprouty proteins exert their function is unclear, most studies support a role for them as antagonists of MEK-ERK pathway activation.

Here, we determined that proliferation decreased in the PrE outgrowths from the UrE of DHT-induced UGSs co-treated with GDNF compared with vehicle (Fig. 2). As the negative effect of GDNF on androgen-induced proliferation of PrE cells is time sensitive and dependent on RET (Figs 2 and 5), we propose that an inhibitory feedback signal is likely activated. The upregulation of Spry1 by day 7 is consistent with inhibitory feedback on RET-mediated signaling in the UrM (Fig. 7). Previously, we have reported that a RET-independent mechanism is responsible for GDNF-induced proliferation of UrE cells in naïve UGSs (Park and Bolton, 2015) (Fig. 7). These differences in epithelial cell proliferation are likely due to intrinsic differences between the PrE cells and UrE cells that are bestowed during androgen-dependent prostatic induction and formation of PrE progenitors/buds (Cunha and Chung, 1981; Cunha and Lung, 1978; Lasnitzki and Mizuno, 1980).

GDNF and AR signaling components are expressed in the UGS; however, their interaction during prostate development has not been explored previously. GDNF, GFRα1, RET and AR are expressed in the UrM of the embryonic UGS and developing prostate (Park and Bolton, 2015) (Fig. 6), and AR signaling in the UrM is required for prostate development (Cunha and Chung, 1981; Cunha and Lung, 1978; Lasnitzki and Mizuno, 1980). Here, we have demonstrated that RET-mediated GDNF signaling downregulates AR expression and nuclear localization in the inductive UrM of androgen-induced UGSs (Fig. 6). These results suggest that AR downregulation by GDNF signaling may lead to partial androgen insensitivity in the developing prostate. This is an intriguing hypothesis because disruption of AR-mediated prostate development predisposes humans and rodents to prostate neoplasia by altering the phenotype of the gland early in life (Anway and Skinner, 2008; Gupta, 2000; Ho et al., 2006; Hu et al., 2012; Prins et al., 2014, 2008; Ramos et al., 2001; Timms et al., 2005). Consistent with Ar being a GDNF-repressed gene, and Gdnf and Gfra1 being androgen-repressed genes in the UGS and developing prostate (Fig. 6), we propose that these genes and their gene products mediate reciprocal regulatory crosstalk to tune prostate development (Fig. 7). The reciprocal regulatory crosstalk in such a circuit may serve to balance GDNF-induced proliferation of inductive UrM cells with androgen-induced PrE proliferation and cytodifferentiation, which are mediated by AR signaling in the inductive UrM cells. If GDNF and/or GFRα1 are not downregulated early in prostate development, elevated levels of GDNF may lead to partial androgen-insensitivity and disruption of AR-mediated prostate development. Additional studies are needed to elucidate the extent of crosstalk between the AR and GDNF signaling pathways.

We have determined that GDNF signaling through RET in UrM cells inhibits DHT-induced proliferation and differentiation of PrE cells in the developing prostate (Figs 4 and 5). We also interrogated smooth muscle differentiation, which may influence paracrine signaling between the inductive UrM and PrE during prostate development (Hayward et al., 1998; Thomson et al., 2002; Timms et al., 1995, 1994). We found that GDNF signaling in the developing prostate does not affect smooth muscle differentiation (Fig. S2), indicating that RET-mediated GDNF signaling in the UrM may selectively affect PrE proliferation and cytodifferentiation. These findings raise an interesting question: how does GDNF signaling in the UrM suppress androgen-induced proliferation and cytodifferentiation in the PrE?

We propose that these effects of GDNF signaling are mediated indirectly on PrE cells by a putative paracrine signal, originating from RET-expressing UrM cells in the VMP (Fig. 7). This mechanism is consistent with previous reports of paracrine signaling for secreted ligands of the WNT (Allgeier et al., 2008; Huang et al., 2009; Joesting et al., 2008; Keil et al., 2012), FGF (Alarid et al., 1994; Donjacour et al., 2003; Sugimura et al., 1996; Thomson and Cunha, 1999) and TGFβ (Cancilla et al., 2001; Cook et al., 2007; Grishina et al., 2005; Itoh et al., 1998; Lamm et al., 2001; Tanji et al., 1994; Tomlinson et al., 2004) protein super-families. For example, regulation of Wnt paracrine signaling from the inductive UrM to the adjacent epithelial cells has been proposed to mediate the effects of TGFβ signaling in the UrM (Li et al., 2009). Given that GDNF family ligands are members of the TGFβ protein super-family, RET-mediated GDNF signaling may regulate aspects of prostate development through communication between inductive UrM cells and PrE cells via a paracrine mechanism.

In conclusion, we have determined that RET-mediated GDNF signaling in the UGS increases proliferation of UrM cells and suppresses androgen-induced PrE proliferation and cytodifferentiation, altering prostate morphogenesis. We also determined that Ar is GDNF repressed by RET-mediated signaling, and Gdnf and Gfra1 are androgen repressed in the developing prostate. These findings reveal that RET-mediated GDNF signaling serves as a negative regulator of prostate development via downregulation of AR expression, and they establish novel crosstalk between AR and GDNF signaling in the cellular and molecular mechanism of androgen-induced prostate development. Therefore, future studies should interrogate the crosstalk mechanism between these signaling pathways and explore whether AR-mediated prostate development and growth in vivo are regulated by other selective GDNF family receptors and activation of RET tyrosine kinase.

C57BL/6J mice were purchased from Jackson Laboratories. Heterozygous Ret mutant (Ret GFP/+) mice have a null Ret knock-in allele that has been described previously (a gift from Dr Sanjay Jain, Washington University, St Louis, MO, USA) (Hoshi et al., 2012). All animal experiments were approved by the University of Illinois at Urbana-Champaign Animal Care and Use Committee. Timed-pregnant females were generated by pairing males with females overnight. We define midday on the day of plug identification as E0.5. We mated Ret GFP/+ mice to produce Ret GFP/GFP (Ret-null, Ret KO), Ret GFP/+ and Ret +/+ control (Ret wild type, Ret WT) embryos.

Genotyping of mice was performed as described for the Ret locus (Hoshi et al., 2012; Jain et al., 2004) and fetal sex (Park and Bolton, 2015). PCR reactions were performed using wild-type Ret primers (5′-CAGCGCAGGTCTCTCATCAGTACCGCA-3′ and 5′-CAGCTACCCGCAGCGACCCGGTTC-3′ produce a 451 bp amplicon) and mutant Ret-GFP primers (5′-CAGCGCAGGTCTCTCATCAGTACCGCA-3′ and 5′-GCCGTTTACGTCGCCGTCCAGCTCGACCAG-3′ produce a 350 bp amplicon). The PCR was achieved in the presence of 1 M betaine using the following method: a denaturation and polymerase activation step at 94°C for 1 min and then 40 cycles consisting of 94°C for 10 s, 62°C for 10 s and 72°C for 20 s.

UGSs were dissected from E15.5 male embryos and prostates were collected from P0 neonates in DMEM/F12 (1:1) medium (21041-025, Invitrogen) and cultured on 0.4 µm Millicell CM filters (PICM01250, EMD Millipore) floating on organ culture medium [DMEM/F12 supplemented with 1% insulin, transferrin and selenium (51500-056, Invitrogen), and gentamicin and amphotericin B (R-015-10, Invitrogen)] in a 5% CO₂ atmosphere, as described previously (Park and Bolton, 2015). Additional supplements included 0.1% ethanol control, 10 nM DHT (A8380, Sigma Aldrich) in ethanol, 0.1% PBS control and 1-100 ng/ml recombinant GDNF (212-GD, R&D Systems) in PBS. Culture medium was changed every other day.

Morphological changes in UGS whole mounts were imaged using a Leica M205A stereoscope. The prostatic epithelial buds were counted from bright-field images and the length of each bud was measured by scaled reference measurement using ImageJ software. The mean bud number and length were quantified using n=4 or more UGSs for each condition. For cell proliferation experiments, UGSs were cultured in organ culture medium with the previously mentioned supplements for 2-7 days and 10 µM bromodeoxyuridine (BrdU, B5002, Sigma Aldrich) was added 2 h before fixation (Park and Bolton, 2015).

UGSs and developing prostates were processed for immunohistochemistry as described previously (Park and Bolton, 2015). Gill Method Hematoxylin Stain 3 (CS4021D, Fisher Scientific) and Eosin Y solution (HT110332, Sigma Aldrich) were used. Immunohistochemistry controls included pre-incubation of primary antibody with antigenic peptide, when available, prior to incubation with secondary antibodies and incubation in the absence of primary antibodies (Park and Bolton, 2015). Primary antibodies against the following proteins were used: vimentin (1:200, 5741S) and pan-keratin (1:200, 4545S) from Cell Signaling Technology; BrdU (1:200, 347580, BD Biosciences); AR (1:200, 06-680, EMD Millipore); RET (5 µg/ml, AF482, R&D Systems); alpha smooth muscle actin (1:1000, ab5694, Abcam); and CK5 (1:300, PRB-160P) and CK8 (1:500, MMS-162P) from Covance. The following secondary antibodies from Molecular Probes were used: AlexaFluor-594 anti-mouse (1:300, A-21203), AlexaFluor-594 anti-rabbit (1:300, A-21207), AlexFluor-594 anti-goat (1:300, A-11058), AlexaFluor-488 anti-mouse (1:300, A-21202) and AlexaFluor-488 anti-rabbit (1:300, A-21206). Tissue sections were then incubated with 0.6 µM 6-diamidino-2-pheylindole (DAPI), washed with PBS and overlaid with mounting medium (10981, Sigma Aldrich) and coverslips. Fluorescence microscopy images were collected and analyzed as described previously (Park and Bolton, 2015).

In cell proliferation experiments, the BrdU-positive cells were detected using BrdU antibody (1:200, 347580, BD Biosciences) and counted in the UrM, UrE and regions where PrE outgrowth occurred. The UrM, and UrE and PrE cells were defined by immunostaining using antibodies for detection of mesenchymal vimentin (5741S, Cell Signaling Technology) and epithelial keratins (4545S, Cell Signaling Technology), respectively. Percentages of proliferating cells in each UGS tissue was calculated as described previously (Park and Bolton, 2015).

UGSs from E15.5 male embryos were separated into UrM and UrE components, and lysates were prepared and used for immunoblot analysis as described previously (Park and Bolton, 2015). PVDF membranes were blocked in TBST [20 mM Tris-HCl (pH 7.5), 150 mM NaCl and 0.1% Tween-20] containing 1% BSA. Membranes were incubated overnight with primary antibodies at the indicated dilutions in TBST+1% BSA. Primary antibodies included: RET (0.1 µg/ml, AF482, R&D Systems); α smooth muscle actin (1:2000, ab5694, Abcam); GAPDH (1:1000, 2118L), vimentin (1:1000, 5741S) and pan-keratin (1:1000, 4545S) from Cell Signaling Technology; and AR (1:800, 06-680, EMD Millipore). Membranes were washed in TBST and incubated for 1 h with 1:2000 dilutions of horseradish peroxidase-linked anti-rabbit or anti-goat IgG (Jackson ImmunoResearch Laboratories) in TBST+1% BSA. Immunoblots were visualized, and ImageJ software was used to quantify protein expression relative to GAPDH control.

UGSs from E15.5 male embryos were cultured in organ culture medium containing 0-10 nM DHT and 0-100 ng/ml GDNF for 2-7 days. RNA was extracted from homogenized UGSs and cDNA was analyzed by QPCR using a StepOnePlus Real-Time PCR System (Applied Biosystems) and validated primers as described previously (Park and Bolton, 2015). Primers included: Ret sense, 5′-TGACATCAGCAAGGATCTGG-3′; Ret antisense, 5′-AGAGCCCATCGTCATACAGC-3′; Pygo2 sense, 5′-CCCAGTCAACCCTTCAACC-3′; Pygo2 antisense, 5′-GAGATCATGGGACCAAATCC-3′; Krt18 sense, 5′-GACGCTGAGACCACACTCAC-3′; Krt18 antisense, 5′-TCCATCTGTGCCTTGTATCG-3′; Krt8 sense, 5′-GACACGTCTGTGGTGCTGTC-3′; Krt8 antisense, 5′-CAGGGTCTGCAATTCCTCAT-3′; Krt14 sense, 5′-GTATTGGTGATGGGCTCCTG-3′; Krt14, antisense, 5′-CACCTCCAGTTCAGTGTTGG-3′; Krt5 sense, 5′-TGCTGCCTACATGAACAAGG-3′; Krt5 antisense, 5′-GCTCCGCATCAAAGAACATC-3′; AR sense, 5′-GGCGGTCCTTCACTAATGTC-3′; AR antisense, 5′-GACAGGTGCCTCATCCTCAC-3′. Data were analyzed using the comparative threshold cycle (Ct) method (Schmittgen and Livak, 2008) and Pygo2 control gene, which is unresponsive to GDNF and DHT (Park and Bolton, 2015). Following normalization to Pygo2 cDNA levels, which is reflected in the ΔCt values, the relative quantification (RQ) of the fold change for each treatment compared with reference control was determined using the following equation: RQ=2^(–ΔCt)/2^{(–ΔCt reference)}. The mean of the log₂ RQ and standard error of the mean (s.e.m.) are plotted.

Significance differences between two treatment groups were determined using Student's t-test, whereas differences among multiple groups or time points were determined by analysis of variance (ANOVA), followed by Tukey's honest significant difference test. Significance levels of 0.05 and 0.01 were applied during data analysis using Student's t-test, and different significance levels are indicated (*P<0.05 and **P<0.01). A significance level of 0.05 was applied during data analysis using ANOVA, and different significance levels have been indicated using different lowercase letters.

We thank Hee Jung Chung, Lori Raetzman and Congcong Chen for insightful comments on the manuscript, and Congcong Chen for technical assistance.