PI3K/mTOR Signaling Regulates Prostatic Branching Morphogenesis

Abstract

Prostatic branching morphogenesis is an intricate event requiring precise temporal and spatial integration of numerous hormonal and growth factor-regulated inputs, yet relatively little is known about the downstream signaling pathways that orchestrate this process. In this study, we use a novel mesenchyme-free embryonic prostate culture system, newly available mTOR inhibitors and a conditional PTEN loss-of-function model to investigate the role of the interconnected PI3K and mTOR signaling pathways in prostatic organogenesis. We demonstrate that PI3K levels and PI3K/mTOR activity are robustly induced by androgen during murine prostatic development and that PI3K/mTOR signaling is necessary for prostatic epithelial bud invasion of surrounding mesenchyme. To elucidate the cellular mechanism by which PI3K/mTOR signaling regulates prostatic branching, we show that PI3K/mTOR inhibition does not significantly alter epithelial proliferation or apoptosis, but rather decreases the efficiency and speed with which the developing prostatic epithelial cells migrate. Using mTOR kinase inhibitors to tease out the independent effects of mTOR signaling downstream of PI3K, we find that simultaneous inhibition of mTORC1 and mTORC2 activity attenuates prostatic branching and is sufficient to phenocopy combined PI3K/mTOR inhibition. Surprisingly, however, mTORC1 inhibition alone has the reverse effect, increasing the number and length of prostatic branches. Finally, simultaneous activation of PI3K and downstream mTORC1/C2 via epithelial PTEN loss-of-function also results in decreased budding reversible by mTORC1 inhibition, suggesting that the effect of mTORC1 on branching is not primarily mediated by negative feedback on PI3K/mTORC2 signaling. Taken together, our data point to an important role for PI3K/mTOR signaling in prostatic epithelial invasion and migration and implicates the balance of PI3K and downstream mTORC1/C2 activity as a critical regulator of prostatic epithelial morphogenesis.

Introduction

The prostate gland develops from the urogenital sinus (UGS), an endodermal sac derived from the hindgut. In males and females, the UGS remains morphologically identical until E17.5 in mice, at which point branching morphogenesis commences in males under the influence of androgens produced by the fetal testes (Thomson, Marker. 2006). During prostatic branching, the urogenital sinus epithelium (UGE) invades the surrounding mesenchyme (UGM) forming epithelial buds that eventually ramify into a network of interconnected tubules. Classical studies in the 1970s demonstrated that androgen-mediated signaling is both necessary and sufficient for this process (Cunha, Lung. 1978; Takeda, et al. 1986). Further, tissue recombination experiments established that the effects of androgens are principally mediated by androgen receptor in the UGM rather than the UGE (Cunha. 1973; Cunha, Lung. 1978). Although these experiments suggest that paracrine signaling by the UGM to the UGE regulates prostatic epithelial invasion, the secreted molecules (“andromedins”) and downstream signaling pathways responsible for this process remain unclear (Prins, Putz. 2008; Pritchard, Nelson. 2008). Members of the fibroblast growth factor family, specifically FGF10 and FGF7, were early candidates for andromedins because they are secreted by the UGM and act as chemoattractants for the migrating prostatic epithelial cells (Sugimura, et al. 1996; Lu, et al. 1999; Thomson, Cunha. 1999; Thomson. 2001; Donjacour, et al. 2003). However, a number of recent studies have suggested that androgen-induced up-regulation of FGFR2 (the preferred receptor for FGF7 and FGF10) in the UGE may underlie the differential responsiveness to FGF signaling in males and females (Huang, et al. 2005; Lin, et al. 2007; Schaeffer, et al. 2008).

Although multiple signaling pathways are known to be activated downstream of androgen and FGF, it remains unclear how many of these are required for prostatic development. Recent work has focused on MAPK (mitogen activated protein kinase) signaling, as FGFR2 inhibition or loss of function inhibits MAPK signaling in vivo, and pharmacologic Mek/Erk1,2 inhibition attenuates prostatic branching in vitro (Huang, et al. 2005; Kuslak, Marker. 2007; Zhang, et al. 2008). However, several lines of evidence suggest that PI3K/mTOR (phosphoinositide-3-kinase/mammalian target of rapamycin) signaling may be an additional important regulator of prostate development. First, androgen can directly activate PI3K signaling in androgen-sensitive benign epithelial cells by interaction with the regulatory p85 subunit of PI3K (Baron, et al. 2004). Second, gene expression studies have documented that androgen induces expression of a number of regulatory members of the PI3K and mTOR signaling pathways, including Pik3r3 and Rheb in embryonic prostate tissue (Schaeffer, et al. 2008). Third, androgen indirectly activates PI3K signaling in the prostate via FGF signaling since PI3K signaling is also compromised in the prostates of mice with genetic inactivation of FGFR2 (Zhang, et al. 2008). Finally, and perhaps most importantly, PI3K/mTOR signaling is commonly aberrantly activated in prostate cancer and a number of recent gene expression studies have suggested that the signaling and transcriptional programs operative during prostatic tumorigenesis and embryonic development are strikingly similar (Schaeffer, et al. 2008; Pritchard, et al. 2009).

The PI3K and mTOR signaling pathways are intricately interconnected and modulate a number of cellular processes critical for embryonic development and tumorigenesis. Upon activation, PI3K phosphorylates PIP₂ (phosphatidylinositol [4,5]-bisphosphate) to PIP₃ (phosphatidylinositol 3,4,5]-trisphosphate) allowing the recruitment of a number of PH-domain containing signaling effectors to the cell membrane, including the kinase PDK1 and its substrate AKT. Importantly, PI3K activity is opposed by lipid phosphatases, the best characterized of which is PTEN (phosphatase and tensin homologue). Although AKT is partially activated following phosphorylation on the Thr308 residue by PDK1, for full activation, AKT must be independently phosphorylated on Ser473 by mTOR (Guertin, et al. 2006). This makes mTOR signaling simultaneously upstream and downstream of AKT because mTOR kinase exists in two competing complexes, mTORC1 (indirectly activated by AKT) and mTORC2 (which phosphorylates AKT at Ser473) (Bhaskar, Hay. 2007; Guertin, Sabatini. 2007). Once phosphorylated at both Thr308 and Ser473, AKT is fully activated and may phosphorylate a number of substrates, resulting in diverse cellular outcomes, including cell proliferation, apoptosis and migration (Guertin, Sabatini. 2007).

Despite its established role in numerous cellular processes critical for embryonic development and epithelial tumorigenesis, relatively few prior studies have looked at the role of PI3K/mTOR signaling in vertebrate branching morphogenesis and no studies have specifically examined its role in prostate development. In the kidney, inhibition of PI3K/mTOR completely blocks epithelial branching in organ cultures and similarly blocks the GDNF-dependent emergence of ectopic ureteric buds in vitro, likely by inhibiting GDNF-stimulated directed cell migration, as seen in kidney cell line model systems (Tang, et al. 2002). In the lung, PI3K/mTOR inhibitors decrease epithelial bud number and length by increasing apoptotic activity and decreasing proliferative activity (Wang, et al. 2005). In the submandibular salivary gland, PI3K/mTOR inhibition dramatically attenuates epithelial clefting in organ culture and mesenchyme-free epithelial cultures (Larsen, et al. 2003). This phenotype does not appear to be mediated by effects on proliferation, however the cellular mechanism has not been fully elucidated. Finally, in the Mullerian duct, PI3K/mTOR signaling is required for elongation of the duct tip and cellular proliferation, but is not required for cell migration (Fujino, et al. 2009). The most important theme emerging from this work is that PI3K/mTOR signaling is required for epithelial budding in a number of different systems. However, the cellular mechanisms responsible for this phenotype are varied and tissue-specific. Furthermore, because most experiments have utilized pharmacologic inhibitors that simultaneously target both PI3K and mTOR kinase, it remains unclear which of these interconnected signaling modules mediates this phenotype.

In the present study, we take advantage of a novel mesenchyme-free prostate epithelial culture system and newly available specific mTOR kinase inhibitors to investigate the roles of PI3K and mTOR signaling in prostatic branching morphogenesis. We find that PI3K/mTOR activity is up-regulated and required in the urogenital sinus epithelium for prostatic bud invasion into the surrounding mesenchyme. Consistent with a conserved role for this pathway in the regulation of cellular motility, inactivation of PI3K/mTOR signaling does not significantly impact cellular proliferation or apoptosis, but rather decreases the efficiency and speed of epithelial cell migration in response to growth factor stimulation. Using mTOR kinase inhibitors to dissect out the contribution of downstream mTOR signaling to prostatic branching, we find that while combined inhibition of mTORC1 and mTORC2 phenocopies the effects of PI3K/mTOR inhibition, inhibition of mTORC1 alone enhances prostatic branching. Importantly, simultaneous activation of PI3K and downstream mTORC1 and mTORC2 by early embryonic PTEN loss-of-function also results in decreased prostatic epithelial budding. Taken together, these data suggest that the balance of PI3K and downstream mTORC1/mTORC2 activity plays a key role in modulating prostatic branching morphogenesis.

Material and Methods

Mouse lines

Animal protocols were approved by the Johns Hopkins University Animal Care and Use Committee (protocol #MO08M367). All mouse lines were maintained on a C57BL/6 background. PH-Akt-GFP mice were a kind gift of Takehiko Sasaki (Sasaki, et al. 2007). mT/mG reporter mice were obtained from Jackson Laboratory (Muzumdar, et al. 2007, Bar Harbor, ME) and crossed with R26ERCre mice (Badea, et al. 2003, Jackson Laboratory) to determine timing and localization of Cre expression during in vitro culture with 4-OHT (4-hydroxytamoxifen) exposure. Mice with inducible PTEN loss of function were generated by crossing R26CreER and PTEN^loxp/loxp mice (Lesche, et al. 2002, Jackson Laboratory) to obtain R26CreER;PTEN^loxp/loxp mice (referred to in the text as PTEN −/− following in vitro culture in 4-OHT) and PTEN^loxp/loxp littermates (referred to in the text as PTEN +/+ following in vitro culture in 4-OHT).

Inhibitors and reagents

PI3K or mTOR inhibitors were dissolved in DMSO and added to organ culture media (described below) at concentrations spanning the IC₅₀ values as follows: LY294002 (10 uM or 25 μM; Sigma-Aldrich; St. Louis, MO); wortmannin (5 μM; Sigma-Aldrich), PI-103 (100 nM; Axon MedChem, Netherlands); rapamycin (200 nM; Invitrogen; Camarillo, CA), DMK-1 (also known as compound 13; Griffin, et al. 2005; Ballou, et al. 2007; 80 μM; a generous gift from JD Powell) and torin1 (Thoreen, et al. 2009; 500 or 1000 nM; a kind gift from DM Sabatini). Antibodies utilized for immunoblotting and immunohistochemistry were as follows: anti-110α, p110β, p-AKT(T308), p-AKT(S473), p-p70S6K, AKT, p70S6K, PTEN, cleaved caspase-3 (all rabbit monoclonals from Cell Signaling Technologies; Danvers, MA); anti-BrdU (sheep polyclonal from Fitzgerald Industries; Concord, MA); anti-CK14 (mouse monoclonal from Millipore; Billerica, MA); anti-AE1/AE3 (mouse monoclonal from Ventana; Tucson, AZ); anti-NKX3.1 (rabbit polyclonal; Chen, et al. 2005; a kind gift of CJ Bieberich); anti-p63 (mouse monoclonal from Santa Cruz Biotechnology; Santa Cruz, CA); and anti-K8 (rat polyclonal developed by P. Brulet and R. Kemler; obtained from the Developmental Studies Hybridoma Bank developed under the auspices of the NICHD and maintained by The University of Iowa, Department of Biology, IA).

Organ culture

Mouse pregnancies were timed according to scheduled 10-hour male-female pairings. Embryos were dissected at indicated times (E15.5-P4) and the sex determined by gonadal inspection. Following dissection in ice-cold DMEM/F12 (1:1) media (Invitrogen), male urogenital sinuses (UGS) were cultured for 1–14 days as indicated at 37°C on 0.4 μm membranes (Millipore; Bedford, MA) overlying completely defined serum-free organ culture media [DMEM/F12 (1:1) supplemented with 1% of nonessential amino acids (Cellgro; Manassas, VA), 0.01 mg/ml of apo-transferrin (Sigma-Aldrich), 0.01 mg/ml of insulin (Sigma-Aldrich), 1×10⁻⁸M dihydrotestosterone (DHT) (Sigma-Aldrich), 1 g/L of D-glucose (Sigma-Aldrich), 2mM L-glutamine (Invitrogen), 1% of 10ug/mL penicillin/streptomycin (Invitrogen)]. Media containing inhibitors or DMSO vehicle were generally replaced every 24–48 hrs during the culture period. To evaluate the role of androgen on branching, E15.5 female UGSs were cultured in standard organ culture media with or without DHT for 48 hr and processed for immunoblotting as below. For experiments involving inducible PTEN loss of function, control (PTEN+/+) and experimental (PTEN−/−) UGSs were cultured in organ culture media lacking DHT with 6 μM 4-OHT for 48 hours followed by standard organ culture media with DHT and 6 μM 4-OHT for the subsequent 5–12 days. In all experiments, a minimum of 3–5 UGSs were treated per condition. Following organ culture, UGSs were imaged on a Motic SMZ 168 stereomicroscope (Motic Optics; Richmond, British Columbia) equipped with a Moticam 2300, 3.0 Megapixel digital color camera. Branches were counted from photomicrographs and branch length was measured for each branch using a scaled reference measurement.

PIP₃ localization in vivo

For histology, UGS tissue from P4 PH-Akt-GFP mice was immediately fixed in 4% paraformaldehyde for 6–8 hours and cryopreserved in 20% sucrose in PBS overnight, embedded in OCT media and snap-frozen at −80°C. 4 μm sections were prepared by standard cryosectioning technique, rinsed in PBS and nuclei were stained with DAPI-containing mounting media (Invitrogen). Membranous GFP localization was visualized on a Zeiss LSM510 Meta laser scanning confocal microscope (Carl Zeiss; Oberkochen, Germany).

Immunoblotting

Whole UGS tissues were homogenized and lysed in ice-cold modified RIPA buffer (1× PBS, 1% IGEPAL CA-630 [v/v], 0.1 % SDS [w/v], 0.5 %Na Deoxycholate [w/v]) supplemented with PMSF (1 mM), aprotinin (1 TIU/ml), NaVO₄ (1 mM), NaF (1 mM) and one protease inhibitor tablet (Roche Diagnostics; Mannheim, Germany) in 7 ml buffer for 10 min on ice. Protein concentrations were quantified using the Micro BCA Protein Assay Kit (Pierce; Rockford, IL), and 15ug of protein were loaded per lane on a 1.5-mm on a 7.5% Tris–HCl SDS-PAGE gel (BioRad; Hercules, CA). Protein was transferred to nitrocellulose membrane (Amersham Bioscience; Buckinghamshire, UK). Membranes were allowed to block for 1h at RT in 5% nonfat milk in 1XTBS-T and then incubated overnight with a primary antibody diluted in 1% BSA. Antibody dillutions were as follows: p-AKT(T308) (1:1000), p-AKT(S473) (1:1000), p-p70S6K (1:1000), p110α (1:1000), p110β (1:1000), pan-AKT (1:1000), p70S6K (1:1000), β-actin (1:4000) (all from Cell Signaling Technologies). The secondary antibodies used were anti-rabbit or anti-mouse immunoglobulin as appropriate (Cell Signaling) and diluted at 1:2000 in 1% BSA. Gel loading was assessed by blotting for β-actin. Blots were developed using a chemiluminescent development solution (Super Signal West Femto; Pierce) and bands were imaged on a chemiluminescent imaging system (Alpha Innotech; Santa Clara, CA). Digital images were quantified using background correction on the Alpha Innotech system and all bands were normalized to their respective β-actin levels.

Immunohistochemistry/Immunofluorescence

Following fixation in 10% neutral buffered formalin (Sigma-Aldrich) and standard tissue processing, embedding, and sectioning at 4 μm, slides were deparaffinized and rehydrated and equilibrated briefly in water. Antigen unmasking was performed by steaming in citrate buffer (pH 6.0) for 25 minutes for all antibodies except NKX3.1, which was unmasked in EDTA (pH 9) for 45 minutes. Endogenous peroxidase activity was quenched by incubation with peroxidase block for 5 minutes at room temperature. Non-specific binding was blocked by incubating in 1% bovine serum albumin (BSA) in TBST for 20 minutes at room temperature. Sections were incubated with each antibody overnight at 4°C. Antibodies were diluted in 1% BSA as follows: p110α (1:400), p-AKT(S473) (1:50), BrdU (1:2000), cleaved caspase3 (1:200), K14 (1:100), AE1/AE3 (prediluted), NKX3.1 (1:1500). For immunohistochemistry, a horseradish peroxidase–labeled polymer (ImmPress; Vector Laboratories, Burlingame, CA or PowerVision; Leica Microsystems, Bannockburn, IL for NKX3.1) was applied for 30 minutes at room temperature. Signal detection was performed using 3,3′-diaminobenzidine tetrahydrochloride (DAB) as the chromagen (Vector Laboratories). Slides were counterstained with hematoxylin, dehydrated, and mounted. For immunofluorescence, Alexafluor-594 anti-mouse (Invitrogen) or DyLight-549 anti-rat (Jackson Immunoresearch, West Grove, PA) secondary antibodies were applied at 1:200 for 30 minutes. Coverslips were mounted with Prolong Antifade containing DAPI (Invitrogen).

Proliferation and apoptosis assays

For 5-bromo-2′deoxyuridine (BrdU) labeling studies, E15.5 UGSs were cultured in standard media with 25 μM LY294002 or DMSO (vehicle). On day 4 of culture, fresh media containing 10 μM BrdU (BD Bioscience, San Jose, CA) was added. Following a two hour incubation period, samples were immediately fixed overnight in 10% neutral buffered formalin and processed for immunohistochemistry. BrdU immunohistochemistry was scored manually by counting the proportion of positively stained nuclei in each ductal branch. At least 7 UGSs were analyzed for each condition.

To assess the rates of apoptosis, E15.5 UGSs were dissected and cultured in standard media with 25 μM LY294002 or DMSO (vehicle). Cultures were fixed in 10% neutral buffered formalin overnight on day 4 of culture and processed for caspase 3 immunohistochemistry. At least 7 UGSs were analyzed for each condition.

Mesenchyme-free culture

Male UGS tissue from R26ERCre */0; mT/mG*/0 mice was dissected at E15.5 and incubated in 2 U/ml dispase (Invitrogen) in HBSS at 37° for 15 minutes. Following washing in ice-cold HBSS, the mesenchyme was manually removed with fine surgical instruments under a dissecting microscope. The remaining intact urogenital sinus epithelium (UGE) was embedded in growth-factor reduced Matrigel (BD Biosciences) in a 4-well coverglass-bottom tissue culture dish (ThermoScientific, Rochester, NY) and submerged in DMEM/F12 media supplemented with 1×10-8 M DHT, 1% penicillin/streptomycin, 1% insulin-transferrin-selenium solution (Sigma Aldrich), 500 ng/mL FGF10 and 200 ng/mL FGF7 (Peprotech, Rocky Hill, NJ). Tissues were immediately incubated on the heated-CO₂ controlled stage of a Zeiss AxioObserver inverted microscope with fluorescence and phase contrast and equipped with an AxioCam digital camera (Carl Zeiss). Differential interference contrast (DIC) and epifluorescence images were collected at 10 μm intervals using a long-working distance 20x objective every 60 minutes for 50–60 hours. Following imaging, tissues were fixed in 10% neutral buffered formalin and submitted for histologic sectioning followed by standard hematoxylin and eosin (H&E) staining.

3D cell motility analysis

Movies collected from mesenchyme-free cultures were analyzed using Imaris 6.3 image analysis software (Bitplane, Zurich, Switzerland). EGFP-expressing cells were marked automatically and tracked in three dimensions over time. For each cell, track length, net displacement and mean speed were calculated. A minimum of 8 UGSs were analyzed per condition.

Results

PI3K is up-regulated and active in the developing prostate following androgen stimulation

To determine whether PI3K is present and active in the developing prostate, we took advantage of a well-characterized organ culture system to initiate prostate budding from male or female urogenital sinus (UGS) tissues following in vitro androgen exposure (Trowell. 1959; Martikainen, Suominen. 1983; Lipschutz, et al. 1997; Doles, et al. 2005). Androgen-naïve E15.5 female UGSs were cultured in vitro for 48 hours in the presence or absence of dihydrotestosterone (DHT), a potent androgen (Fig. 1A). Androgen exposure increased p-AKT (T308) levels, an indirect measure of PI3K activity, and resulted in variable up-regulation of the p110α and p110β catalytic subunits of PI3K by immunoblotting (Fig. 1B; n = 3 independent experiments, 3 UGS/condition/experiment). In contrast to total levels in the mesenchyme and epithelium measured by immunoblot, immunohistochemistry revealed prominent enrichment of p110α specifically in the invading epithelial bud in vivo in E17.5 and E18.5 male embryos (Fig. 1C, top left panel). Interestingly, p-AKT(S473), a measure of PI3K and downstream mTORC2 activity, was also up-regulated in the invasive prostatic buds in E18.5 male embryos (Fig. 1C, top right panel). Because p-AKT levels provide only an indirect measure of PI3K activity, we took advantage of a mouse transgenic for the PIP₃ biosensor AKT-PH-GFP to assess in vivo PIP₃ levels directly (Sasaki, et al. 2007). The emerging prostatic buds in P4 embryos from these mice showed increased membranous GFP signal compared to the surrounding urethral sinus epithelium (Fig. 1C, bottom panels), indicating that PI3K is active in the epithelium. From these experiments, we conclude that PI3K and mTORC2 are enriched and active in the invading prostatic buds, indicating that this signaling pathway plays an important role in epithelial invasion during prostatic branching morphogenesis.

Figure 1. PI3K is up-regulated and active during prostatic branching morphogenesis.

(A) Immunoblotting of E15.5 female urogenital sinus (UGS) tissues cultured with or without dihydrotestosterone (DHT) for 48 hours reveals mild up-regulation of p-AKT (T308), an indirect measure of PI3K signaling, with a variable increase in levels of the p110α catalytic subunit of PI3K in response to androgen exposure. (B) Quantification of immunoblot in (A); n = 3 independent experiments, 3 UGS/condition/experiment. (C) Immunohistochemistry of E17.5 and E18.5 male UGS tissues demonstrates enrichment of p110α and downstream p-AKT(S473) proteins in the invasive prostatic epithelial buds (top panels, arrows, 200x magnification) compared to surrounding urethral epithelium. Mice transgenic for the PIP₃ biosensor, AKT-PH-GFP, show enrichment of membranous PIP₃ in the emerging prostatic bud epithelium at P4 (bottom panels, arrows, 100x and 400x magnification).

PI3K/mTOR activity is required for prostatic budding in vitro

Next, we sought to determine whether PI3K/mTOR activity is required for prostatic morphogenesis. PI3K consists of multiple catalytic and regulatory subunits and homozygous deletion of either p110α or p110β is embryonic lethal (Bi, et al. 1999; Bi, et al. 2002). Further complicating the situation, loss of one isoform often alters expression of another, making genetic experiments resulting in conditional PI3K loss-of-function technically challenging (Ueki, et al. 2002; Ueki, et al. 2003; Brachmann, et al. 2005; Zhao, et al. 2006). Thus, to determine whether there is a requirement for PI3K signaling during prostatic branching morphogenesis, we took advantage of three pharmacologically distinct inhibitors for this pathway. LY294002, wortmannin and PI-103 are well-characterized inhibitors of PI3K, however because of catalytic site homology, these inhibitors also block downstream mTOR kinase activity (Brunn, et al. 1996; Knight, et al. 2006). To assess the effects of combined PI3K/mTOR signaling blockade, urogenital sinus (UGS) tissues from E15.5 mice were cultured with inhibitor or vehicle in the presence of androgen for 7 days. UGS tissues exposed to 10–25 μM LY294002 showed a striking, consistent and dose-dependent attenuation in prostatic branching with minimal phenotypic variability (Fig. 2A, Supplementary Figure 1). Interestingly, the UGS epithelium of LY294002-treated samples did expand over the culture period relative to its size at day 0, but without apparent epithelial budding. Immunoblots demonstrated that these concentrations of inhibitor resulted in a marked decrease in p-AKT(T308), a measure of PI3K signaling, as well as p-AKT(S473) which reflects mTORC2 and PI3K signaling (Fig. 2B, C). Phospho-p70S6K levels were also decreased, reflecting decreased downstream mTORC1 signaling as well. Wortmannin, an irreversible inhibitor of PI3K and mTOR kinase, similarly attenuated prostatic branching (Fig. 2A) and had a nearly identical profile on immunoblotting (data not shown). Finally, PI-103, an additional newly available and specific inhibitor of PI3K and mTOR had a similar phenotypic effect to LY294002 and wortmannin (Fig. 2A). Although experiments with pharmacologic inhibitors must always be interpreted with caution, the fact that we observed identical results with three pharmacologically distinct inhibitors strongly suggests that PI3K/mTOR signaling is required for prostatic branching morphogenesis.

Figure 2. PI3K/mTOR signaling is required for prostatic branching morphogenesis.

(A) Treatment of E15.5 male urogenital sinus (UGS) tissues with LY294002, wortmannin and PI-103 (PI3K/mTOR inhibitors) for 7 days in organ culture results in a decrease in prostatic epithelial budding (arrowheads) compared to vehicle control. Scale bar = 1 mm. (B) Immunoblotting of UGS lysates following 24 hours of treatment with vehicle or LY294002 reveals a decrease in PI3K signaling (indirectly measured by p-AKT [T308] levels), mTORC1 signaling (p-p70S6K levels) and mTORC2 signaling (p-AKT [S473] levels). (C) Quantification of immunoblot results from (B); n = 3 independent experiments, 3 UGS/condition/experiment.

To assess whether the morphologic effects of PI3K/mTOR inhibitors on prostatic branching were due to toxicity, we performed drug washout experiments. Exposure to 25 μM LY294002 for 24 hours followed by washout and replacement with vehicle-containing media resulted in prostatic branching similar to vehicle treated controls, confirming that this concentration of LY294002 does not result in non-specific epithelial or mesenchymal toxicity (Fig. 3A). Histologic examination of UGS tissues treated with vehicle or LY294002 for 7 days verified that both the epithelial and mesenchymal compartments appeared viable (Fig. 3B). However, tissues treated with PI3K/mTOR inhibitors had a near total absence of the invasive finger-like epithelial buds interspersed with strands of mesenchymal tissue seen in the vehicle control samples. Instead, PI3K/mTOR-inhibited tissues showed a broad-based, pushing epithelial border with the mesenchyme, highlighted by pan-keratin immunostaining (Fig. 3B). Quantitation of bud number and size revealed a nearly four-fold decrease in epithelial bud number with PI3K/mTOR inhibition and a significant decrease in bud length for the few buds that were seen (Fig. 3C; n = 10–11 UGS/condition). Histologic examination of the prostate buds also suggested that the epithelial nuclei were more crowded in the samples treated with PI3K/mTOR inhibitor and quantitative analysis revealed that there were significantly more epithelial nuclei per unit area in the LY294002-treated samples compared to vehicle control, resulting the appearance of epithelial disorganization (Fig. 3D; n = 10–11 UGS/condition). Taken together, these data indicate that PI3K and/or mTOR signaling is required for epithelial duct invasion into surrounding mesenchyme as well as for duct elongation during prostatic branching morphogenesis.

Figure 3. PI3K/mTOR inhibition is not toxic and results in fewer and shorter invasive prostatic buds.

(A) Treatment of male E15.5 urogenital sinus (UGS) tissue with 25 uM LY294002 for 24 hours followed by drug washout restores prostatic branching compared to UGS tissues treated for 7 days and vehicle controls. Scale bar = 1 mm. (B) Histologic examination of E15.5 UGS tissues treated with LY294002 for 7 days in organ culture reveals a near total absence of the invasive finger-like epithelial buds interspersed with strands of mesenchyme seen in the vehicle control samples, and instead shows broad-based, pushing protrusions into the mesenchyme using hematoxylin and eosin (H&E) staining (buds indicated by black outlines, 100x magnification). Buds are highlighted by pan-keratin (AE1/AE3) immunostaining (200x magnification). Epithelial nuclei appear more crowded in the LY294002-treated samples (400x magnification). (C and D) Quantification of mean bud number, bud length and nuclear density in LY294002-treated samples reveals a statistically significant decrease in bud number and length and increase in nuclear density (n=10–11 UGS per condition, error bars = standard error of the mean (SEM), p-values using Student’s t-test for unpaired samples with unequal variance are indicated).

PI3K/mTOR activity is not required for prostate epithelial cell proliferation, apoptosis or specification

Next, we sought to determine the cellular mechanism by which PI3K/mTOR inhibition attenuates prostatic branching. Although both PI3K and mTOR signaling have established roles in cell proliferation and inhibition of apoptosis (Bhaskar, Hay. 2007; Guertin, Sabatini. 2007), work in multiple organ systems suggests that the cellular mechanism by which PI3K/mTOR signaling inhibits branching morphogenesis is varied and tissue-specific (Tang, et al. 2002; Larsen, et al. 2003; Wang, et al. 2005; Fujino, et al. 2009). To determine whether the attenuation in prostatic budding with PI3K/mTOR inhibition was due to decreased epithelial proliferation, we used BrdU labeling to quantify proliferating cells on day 4 of in vitro urogenital sinus culture. In the presence of both vehicle and inhibitor, proliferating cells were essentially confined to the urogenital sinus epithelium in the region of the emerging prostatic buds (Fig. 4A, left panel). Interestingly, quantification of the percentage of epithelial cells with BrdU labeling did not reveal a significant difference in the presence of PI3K/mTOR inhibitors (Fig. 4B; n = 7–9 UGS/condition). Apoptosis rates were assessed by immunohistochemical staining for cleaved caspase 3 at day 4 of culture. In both vehicle control- and LY294002-treated samples, apoptotic cells were virtually absent from the urogenital sinus epithelium, although rare apoptotic cells were seen in the surrounding mesenchymal tissue in both cases and provided an internal positive control (Fig. 4A, middle panel). Thus, we conclude that decreases in proliferation or increases in apoptotic activity do not primarily account for the effects of PI3K/mTOR inhibitors on prostatic epithelial branching.

Figure 4. PI3K/mTOR activity is not required for epithelial proliferation, apoptosis or specification.

(A) E15.5 male and female UGS tissues were cultured for 4 days in vehicle or 25 μM LY294002 and treated with BrdU for 2 hours prior to fixation. BrdU-positive epithelial nuclei are localized to epithelial buds in both LY294002- and vehicle-treated specimens (left panels, 200x magnification). Immunohistochemistry for cleaved caspase-3 using the same specimens reveals a total absence of apoptotic epithelial cells in both conditions, although rare apoptotic mesenchymal cells are present in vehicle- and LY294002-treated controls and serve as an internal positive control (middle panels, arrows, 400x magnification). Finally, immunohistochemistry for NKX3.1, an androgen-induced homeobox domain-containing transcription factor expressed during early prostatic development, shows positive nuclei confined to the emerging prostatic buds in both vehicle- and LY294002-treated specimens, suggesting that appropriate prostatic epithelial cell specification occurs in the presence of PI3K/mTOR inhibition (right panels, 200x magnification). (B) Quantification of proportion of BrdU-positive cells/epithelial bud in vehicle- and LY294002-treated specimens reveals no statistical difference between the two conditions (n=7–9 UGS/condition, error bars = SEM, p-value using Student’s t-test for unpaired samples with unequal variance).

Genetic loss-of-function models that inhibit prostatic epithelial cell specification during development result in a similar attenuated branching phenotype to that seen with PI3K/mTOR inhibitors (BS and DMB, unpublished data). To investigate whether the decrease in prostatic ductal branching might be due to a lack of prostatic epithelial differentiation, we performed immunostaining for NKX3.1, an androgen-regulated homeodomain-containing transcription factor. Up-regulation of NKX3.1 is one of the earliest known molecular markers of prostatic epithelial specification (Bieberich, et al. 1996; Bhatia-Gaur, et al. 1999). Interestingly, both vehicle control and PI3K/mTOR-inhibited tissues showed robust nuclear NKX3.1 expression confined to the emerging or abortive prostatic epithelial buds (Fig. 4A, right panel). Thus, we conclude that PI3K/mTOR activity is not required for prostatic epithelial specification.

PI3K/mTOR activity is specifically required in prostatic epithelial cells during branching morphogenesis

Many of the morphogenic signals regulating prostatic epithelial development are paracrine signals from the urogenital sinus mesenchyme (Prins, Putz. 2008), so we considered the possibility that PI3K/mTOR inhibition attenuated prostatic epithelial branching by inhibiting mesenchymal signaling. To examine the specific effects of PI3K/mTOR inhibition on developing prostatic epithelial cells, we developed a mesenchyme-free embryonic epithelial culture system that supports prostatic epithelial branching, similar to systems previously described for the study of salivary gland, lung and mammary morphogenesis (Hahm, et al. 1990; Nogawa, Takahashi. 1991; Nogawa, Ito. 1995). Using a combination of enzymatic and manual dissection, we dissociated the urogenital sinus epithelium from the mesenchyme in E15.5 embryos and embedded the intact epithelial structure in laminin-rich extracellular matrix (Matrigel). While media containing only androgen led to cystic atrophy of the epithelium, addition of androgen with FGF10 and FGF7 supported prostate branching in the absence of mesenchymal tissue over a 48 hour period (Fig. 5A, 5B). These mesenchyme-free epithelial explants expressed NKX3.1, consistent with prostatic differentiation (Fig. 5B), and maintained minimally overlapping basal and luminal cell layers (expressing p63 and K8, respectively; Fig. 5C), similar to mature prostate tissue. When viewed by time-lapse differential interference contrast (DIC) imaging, the Matrigel-embedded urogenital sinus epithelium branched in an identical pattern in 5 independent experiments, with distinct anterior and ventral epithelial lobes appearing within the first 24 hours after exposure to FGF10 and FGF7 (Fig. 5D, arrows). In contrast, PI3K/mTOR inhibition in this system resulted in only small, abortive epithelial branches without evident prostate lobe formation as observed by time-lapse DIC imaging (Fig. 5D, Movie 1A and 1B). Consistent with our finding that PI3K is up-regulated and expressed in the emerging epithelial buds (Fig. 1C), these data support a specific requirement for PI3K/mTOR activity in the prostatic epithelium during branching.

Figure 5. PI3K/mTOR activity is required specifically in epithelial cells.

(A) A combination of enzymatic and manual dissection allows dissociation of intact urogenital sinus epithelium (UGE) from mesenchymal tissue and epithelial branching following 48 hours of culture in Matrigel with androgen (DHT, 1 × 10⁻⁸ M) and FGF10 (500 ng/mL) and FGF7 (200 ng/mL). (B) Hematoxylin and eosin (H&E) staining of UGE cultures shows individual epithelial buds surrounded by a prominent basal cell layer, with most cells expressing NKX3.1, a prostatic differentiation marker (200x; inset at 400x magnification). (C) Immunofluorescence for K8 demonstrates that most interior cells are luminal cells, while exterior cells predominantly express basal marker p63 with minimal overlap between these populations in merged images (600x magnification). (D) Time-lapse DIC microscopy of mesenchyme-free UGE cultures demonstrates the emergence of large prostatic buds in the distribution of the anterior and ventral prostate lobes (arrows) in the vehicle control (arrows, top panels), while only small and abortive buds are visible in the 25 μM LY294002-treated specimens (arrowheads, bottom panels). Scale bars=100 μm. Images shown are representative of one experiment (n=5).

PI3K/mTOR activity is required for epithelial motility during branching morphogenesis

PI3K signaling has a well-established role in the regulation of single cell motility and chemotaxis (Kolsch, et al. 2008) and an equally important role is emerging for mTORC2 in these systems (Cai, et al. 2010; Charest, et al. 2010; Liu, et al. 2010), so we considered the possibility that PI3K/mTOR inhibition might decrease prostatic epithelial migration during branching. The fact that prostate epithelial proliferation is unaffected by PI3K/mTOR inhibition, while epithelial cell density is increased also suggested that PI3K/mTOR activity may be required for epithelial motility. To examine three-dimensional prostatic epithelial motility during morphogenesis in intact tissues, we performed time-lapse imaging experiments using transgenic mouse tissues mosaic for membranous EGFP expression. Fortuitously, Cre expression in the R26ERCre mouse line is tightly regulated, and when crossed to the mT/mG reporter line, R26ERCre;m/TmG mice show a low level of Cre expression in the absence of 4-OHT exposure. Accordingly, we observed a small population of EGFP-expressing cells (<1%) in the urogenital sinus epithelium at E15.5 without 4-OHT. Timelapse epifluorescence imaging of mesenchyme-free urogenital sinus epithelial cultures from these mice revealed abundant epithelial cell motility during prostatic branching, with visible extension and retraction of cytoplasmic protrusions during this process (Fig. 6A, top panels and Movie 2A). In the presence of PI3K/mTOR inhibition, prostatic epithelial cells assumed an elongated morphology compared to vehicle control cells, and though they exhibited cytoplasmic protrusions, their motility appeared significantly impaired (Fig. 6A, bottom panels and Movie 2B).

Figure 6. PI3K/mTOR activity is required for efficient epithelial cell migration in mesenchyme-free epithelial cultures.

(A) Urogenital sinus epithelial (UGE) tissues from R26ERCre;mT/mG transgenic mice show mosaic expression of membranous EGFP allowing three-dimensional tracking of epithelial cell migration over time using time-lapse epifluorescence imaging of mesenchyme-free cultures. While individual cells in vehicle-treated controls demonstrate efficient motility with cytoplasmic protrusions in the direction of migration (top panels, arrows) as well as readily apparent cell divisions (top panels, arrowheads), LY294002-treated epithelial cells assume an elongated shape with extension and retraction of protrusions, but are relatively immobile (bottom panels, arrows). (B) Positions of individual epithelial cells can be tracked over time in three dimensions (arrows indicate the net displacement over a 30 hour time period). Individual cell tracks in LY294002-treated UGEs are visibly shorter than those in vehicle controls. Scale bars = 100 um. (C) Net track displacement and mean velocity are significantly decreased with LY294002 treatment. Data shown are from one representative experiment out of 5 (n= 16–35 cells/condition, error bars = SEM, p-value using Student’s t-test for unpaired samples with unequal variance).

To quantify epithelial motility, we used image analysis software to track the position of individual prostatic epithelial cells in three dimensions over time during mesenchyme-free morphogenesis (Fig. 6B). This analysis revealed that PI3K/mTOR inhibition results in a statistically significant decrease in epithelial motility as measured by net displacement and mean epithelial cell speed (Fig. 6C). Compared to net displacement, total distance traveled was less affected in LY294002-treated samples, indicating that the efficiency of epithelial motility is relatively more compromised than overall motility in PI3K/mTOR-inhibited samples. Taken together, these data indicate that regulation of cellular migration contributes to the cellular mechanism by which PI3K/mTOR activity regulates prostatic branching.

mTOR kinase activity is required for prostatic branching

Because LY294002, wortmannin and PI-103 all inhibit PI3K as well as mTOR kinase, these data do not distinguish whether the effects of these drugs are primarily modulated by PI3K or by downstream mTOR kinase inhibition. Given the emerging role of mTORC2 in single cell motility (Cai, et al. 2010; Charest, et al. 2010; Liu, et al. 2010), we hypothesized that specific inhibition of mTOR signaling, without inhibition of upstream PI3K signaling, might be sufficient to abrogate prostatic branching. To address this question, we took advantage of a number of recently described ATP-competitive inhibitors that block mTOR kinase function without inhibiting PI3K (Griffin, et al. 2005; Ballou, et al. 2007; Thoreen, et al. 2009). These inhibitors block mTORC1 and mTORC2 signaling simultaneously, as both complexes require mTOR kinase for catalytic activity. Treatment of embryonic urogenital sinuses with torin1 or DMK-1 resulted in markedly decreased branching, phenotypically replicating the results seen with combined PI3K/mTOR inhibition by LY294002 (Fig 7A; images are representative of 3 independent experiments; concentrations used were 1000 nM for torin1 and 80 μM for DMK-1). Upon histologic sectioning, urogenital sinuses treated with mTOR kinase inhibitors showed abortive branches without invasion of the surrounding mesenchymal tissues, nearly identical to specimens treated with LY294002 (Fig. 7B). Interestingly, in contrast to LY294002, these inhibitors did not show consistently graded phenotypic effects with decreased dosages (eg, 500 nM for torin1 or 40 μM for DMK-1), but rather showed a significant decrease in branching only at doses coinciding with ~50% inhibition of both mTORC1 (as measured by p-p70S6K levels) and mTORC2 activity (as measured by p-AKT[S473]), without effects on PI3K activity (indirectly measured by p-AKT[T308] levels) (1000 nM for torin1 and 80 μM for DMK-1; Fig. 7C, 7D). This may reflect the fact that long term treatment with lower doses of mTOR kinase inhibitors (eg, less than 250 nM for torin1 in cell line experiments) does not effectively inhibit mTORC2 activity and may even increase PI3K activity (Guertin, Sabatini. 2009). The fact that inhibition of mTOR kinase alone is sufficient to phenocopy the effects of dual PI3K/mTOR kinase inhibition implies that mTOR kinase may be a critical downstream effector of PI3K signaling during branching morphogenesis.

Figure 7.

Specific inhibition of mTOR kinase without PI3K inhibition abrogates prostatic branching and phenocopies combined PI3K/mTOR inhibition. (A) Treatment of male urogenital sinus (UGS) tissues with either of two different mTOR kinase inhibitors (1000 nM torin1 or 80 μM DMK-1) for 5–7 days results in decreased prostatic branching grossly identical to that seen after 25 μM LY294002 treatment. Arrowheads indicate prostatic branches in vehicle control, while seminal vesicle morphogenesis is grossly unaffected by any inhibitors (arrows). Images are representative of 3 independent experiments. (B) Histologic sectioning of urogenital sinuses treated with DMK-1 for 5 days reveals absence of invasive and elongated epithelial buds (arrows), as compared to vehicle-treated controls. (C) Immunoblotting of UGS tissues treated for 24 hours with mTOR kinase inhibitors demonstrates decreased mTORC1 signaling (represented by decreased p-p70S6K levels) and mTORC2 signaling (represented by decreased p-AKT[S473] levels), while PI3K signaling (as measured by p-AKT[T308] levels) remains unchanged. (D) Quantitation of immunoblot from (B); n = 3 independent experiments, 3 UGS/condition/experiment.

mTORC1 inhibition results in increased prostatic branching

By definition, mTOR kinase inhibitors block both mTORC1 and mTORC2 signaling. To determine whether mTORC1 or mTORC2 signaling is specifically required for prostatic branching, we took advantage of the fact that mTORC1 is preferentially sensitive to rapamycin inhibition (Guertin, Sabatini. 2009). Surprisingly, treatment of urogenital sinus cultures with rapamycin resulted in a consistent increase in prostatic branching, with more numerous and longer prostatic buds visible after 11 days of culture (Fig. 8A, 8B; n=4 UGS/condition). Recently, rapamycin has been reported to result in partial mTORC2 inhibition in some systems after prolonged exposure or at high concentrations (Sarbassov, et al. 2006; Shor, et al. 2008). In our system, using 200 nM rapamycin, we did not observe evidence of mTORC2 inhibition with rapamycin after 24 hours or 6 days of culture (Fig. 8 and data not shown). In fact, while rapamycin decreased p-p70S6K levels, consistent with mTORC1 inhibition, levels of p-AKT(S473) were mildly increased after rapamycin treatment, by immunoblotting after 24 hours (Fig. 8C, D) and by immunohistochemistry after 11 days (Fig. 8A). This reflects an increase in mTORC2 activity following mTORC1 inhibition, a finding that results from the release of an established negative feedback loop between S6K signaling and PI3K/mTORC2 signaling. (Manning. 2004; Manning, et al. 2005; O’Reilly, et al. 2006; Guertin, Sabatini. 2007). Thus, while these experiments indicate that mTORC1 activity attenuates prostatic branching, it remains possible that the effects of mTORC1 are indirectly mediated by negative feedback on PI3K/mTORC2 signaling.

Figure 8. Specific inhibition of mTORC1 increases the number and length of prostatic branches.

(A) Culture of E15.5 urogenital sinus (UGS) tissues in 200 nM rapamycin results in increased numbers of prostatic branches (top panels, arrowheads) by day 11 of culture. Excess branches in rapamycin-treated specimens are highlighted on cytokeratin 14 (CK14) immunostained histologic sections (middle panels, arrowheads, 200x magnification). Immunohistochemistry for pAKT(S473) after 11 days of culture in rapamycin reveals increased mTORC2 activity in prostatic buds relative to vehicle-treated control (bottom panels, 200x magnification). (B) Quantification of bud number and length in day 11 samples (n= 4 UGS/condition, error bars = SEM, p-value using Student’s t-test for unpaired samples with unequal variance.) (C and D) Immunoblots with quanitification of vehicle- and rapamycin-treated UGS tissues after 24 hours of culture reveals decreased mTORC1 activity (p-p70S6K levels), and slightly increased mTORC2 activity (p-AKT[S473]) levels, consistent with rapamycin-mediated relief of negative feedback between p70S6K and mTORC2 signaling; n = 3 independent experiments, 3 UGS/condition/experiment.

PTEN loss inhibits prostatic branching in an mTORC1-dependent fashion

To distinguish whether mTORC1 modulates prostatic branching directly or indirectly via feedback to PI3K/mTORC2, we took advantage of the fact that PI3K, mTORC1 and mTORC2 signaling are all simultaneously activated in the setting of PTEN loss, essentially abrogating the effects of negative feedback between mTORC1 and PI3K/mTORC2 (Manning, et al. 2005). If mTORC1 indirectly inhibits prostatic branching by inhibiting PI3K/mTORC2 signaling, branching should remain unchanged or even increase in the setting of PTEN loss, despite high mTORC1 activity. Alternatively, if the inhibitory effects of mTORC1 are independent of feedback to PI3K/mTORC2 activity, PTEN loss might be expected to attenuate prostatic branching. Although numerous conditional and prostate-specific PTEN loss-of-function models exist, all are driven by prostate-epithelial specific Cre expression which occurs typically at P14 or later, when the majority of prostatic branching morphogenesis is already completed (Brunn, et al. 1996; Backman, et al. 2004; Wang, et al. 2005; Ratnacaram, et al. 2008). To test whether early embryonic PTEN loss-of-function would alter prostatic branching, we generated a tamoxifen (4-OHT)-inducible in vitro PTEN loss-of-function model using the R26ERCre mouse line. To verify that robust Cre expression could be induced in vitro prior to prostatic branching, we crossed these mice to mT/mG reporter mice, allowing Cre expression to be tracked by visualization of a membranous EGFP label, while cells negative for Cre express a membranous RFP (tomato red). After 18 hours of organ culture with 6 μM 4-OHT and DHT, urogenital sinuses (UGS) from Cre-positive animals (as determined by genotyping) expressed EGFP and were readily distinguishable from Cre-negative littermates under a fluorescent dissecting microscope (data not shown). By day 4 of organ culture, the timepoint at which the first prostatic buds begin to emerge, we observed nearly uniform Cre expression in the UGS epithelium of cryosectioned tissues (visualized as membranous EGFP expression) (Fig. 9A). Relatively less Cre expression was visible in the surrounding mesenchyme (visualized as membranous tomato red expression). Importantly, although estrogen signaling is known to affect prostatic morphogenesis (Huang, et al. 2005), prolonged culture in 6 μM 4-OHT did not have any significant independent effects on in vitro prostatic branching (as demonstrated by the PTEN+/+ UGS in Fig. 9C), and 4-OHT only showed toxic effects at a dose of 100 μM or above (BS, unpublished observations).

Figure 9. PTEN loss-of-function decreases prostatic epithelial bud number and length in an mTORC1-dependent fashion.

(A) Cre is efficiently induced prior to extensive prostatic branching in E15.5 R26ERCre; mT/mG reporter mouse urogenital sinus (UGS) tissues cultured for 4 days in 6 μM 4-OHT. Vehicle-treated control UGS tissues express a membrane-bound tomato red fluorophore using the mT/mG reporter allele, consistent with the absence of inducible Cre expression (left panel, merged image), while 4-OHT-treated UGS epithelial cells express membrane-bound EGFP, consistent with Cre expression in the majority of cells (middle panel, merged image). Only small epithelial buds are visible in in vitro cultures at this early timepoint (right panel, merged image). The mesenchymal cells in the 4-OHT-treated specimens demonstrate a lower level of inducible Cre than the epithelial cells (middle and right panels, merged images). (B) Immunoblotting for PTEN and PI3K/mTOR signaling components in PTEN^fl/fl (PTEN+/+) and R26ERCre;PTEN^fl/fl (PTEN−/−) UGS tissues treated with 4-OHT for 7 days during in vitro organ culture. PTEN protein is dramatically decreased while PI3K and mTOR signaling is increased in UGS specimens with inducible PTEN loss (PTEN−/−). (C) Prostatic bud number and length are decreased by inducible PTEN loss-of-function and this effect is reversible by mTORC1 inhibition with rapamycin (n= 5–16 UGS/condition, error bars = SEM, * p<0.01; ** p < 0.0001; *** p < 0.01 using Student’s t-test for unpaired samples with unequal variance).

We then generated R26ERCre;PTEN^loxp/loxp embryos (PTEN−/−) and PTEN^loxp/loxp littermate controls (PTEN +/+). Culture of UGS from R26ERCre;PTEN^loxp/loxp embryos for 7 days in DHT after the addition of 6 μM 4-OHT resulted in near total loss of PTEN protein as demonstrated by immunoblotting, and dramatically increased levels of PI3K signaling (measured by p-AKT T308), mTORC1 signaling (measured by p-p70S6K) and mTORC2 signaling (measured by p-AKT S473) (Fig. 9B). Importantly, compared to littermate PTEN+/+ controls cultured in identical conditions with 4-OHT, we observed a significant decrease in mean bud number and length in PTEN−/− UGS (Fig. 9C, 9D; n= 5–16 UGS/condition). Because these experiments do not formally exclude the possibility that PTEN loss results in decreased prostatic branching specifically in the context of altered estrogenic signaling due to 4-OHT, we also cultured wildtype E15.5 UGS in a vanadate compound known to inhibit PTEN phosphatase activity, bpV(pic) (Schmid, et al, 2004) (Supplementary Figure 2). In these experiments conducted without 4-OHT, PTEN phosphatase inhibition also resulted in abrogated prostatic branching, suggesting that PTEN activity may be required for prostatic morphogenesis independent of estrogenic signaling status. Finally, to demonstrate that the effect of PTEN-inactivation on prostatic branching was specifically due to increased mTORC1 activity, we returned to our genetic inactivation system. We treated PTEN−/− samples with rapamycin and found that mTORC1 inhibition restored epithelial branching in the context of PTEN loss (Fig. 9C, D). Strikingly, PTEN−/− UGS samples treated with rapamycin not only recovered prostatic branching, but branched even more robustly than wildtype samples, an effect we hypothesize is attributable to the increased baseline level of PI3K/mTORC2 signaling in these samples. We conclude from these experiments that the inhibitory effects of mTORC1 activity on prostatic branching are independent of the feedback loop between mTORC1 and PI3K/mTORC2 signaling. Given that combined PI3K/mTOR inhibitors and specific mTOR kinase inhibitors similarly attenuate prostatic branching, our data are compatible with a model wherein PI3K/mTORC2 signaling is required for prostatic branching, while mTORC1 signaling negatively regulates the same process. This strongly suggests that the overall balance of mTORC1 and PI3K/mTORC2 signaling is a critical regulator of prostatic morphogenesis.

Discussion

In this study, we have shown that PI3K/mTOR signaling is activated in the invading epithelial buds during prostatic development and required for prostatic ductal morphogenesis. Consistent with a specific role in developing prostate epithelial cells, the p110α catalytic subunit of PI3K is up-regulated in response to androgen exposure in the emerging prostatic buds. Immunoblotting for p-AKT(T308), as well as imaging in mice transgenic for the PIP₃ biosensor, demonstrate that PI3K is active in the developing prostate and concentrated in the invasive epithelium. Similarly, p-AKT(S473) localization in the prostatic buds suggests a role for downstream mTORC2 signaling in prostatic morphogenesis. In accordance with the epithelial activation of PI3K/mTOR signaling during prostatic development, combined inhibition of PI3K and mTOR kinase results in a dramatic decrease in epithelial buds and ductal elongation in whole organ cultures, with minimal phenotypic variability. Instead of invasive, finger-like protrusions into the surrounding mesenchyme, samples exposed to PI3K/mTOR inhibitors show a broad, pushing epithelial border with the mesenchymal tissue. Similarly, in mesenchyme-free urogenital sinus epithelial cultures, PI3K/mTOR inhibition results in absent lobe formation with abortive branching. Thus, PI3K/mTOR signaling plays a critical role in regulating prostate epithelial invasion into the surrounding mesenchyme and extracellular matrix.

To our knowledge, this study is the first to demonstrate that PI3K/mTOR signaling regulates prostate epithelial motility during development, an intriguing finding given the frequent up-regulation of this signaling pathway in human prostate tumors (Taylor, et al. 2010). Surprisingly we found that the cellular mechanism mediating the effect of PI3K/mTOR inhibition on prostate epithelial invasion is not altered rates of proliferation, apoptosis or impaired epithelial specification. Rather, PI3K/mTOR signaling regulates prostate epithelial cell migration in response to growth factor stimulation. Using a novel mesenchyme-free culture system that supports prostate lobe formation and branching, we were able to measure three-dimensional epithelial cell motility during prostatic development. We found that LY294002-treated urogenital sinus epithelial cells assumed an elongated shape, and although they did display intermittent cytoplasmic protrusions, their net displacement and mean speed over time was significantly lower than controls. This decrease in the efficiency of epithelial motility may in part account for our observation that epithelial nuclei were more crowded in PI3K/mTOR-inhibited urogenital sinus samples (Fig. 3D). If epithelial cell proliferation continues at a similar rate in LY294002-treated samples, but the epithelial cells do not efficiently invade and migrate into the surrounding mesenchyme, epithelial cell density might be expected to increase.

Overall, our finding that PI3K/mTOR signaling regulates prostate epithelial migration after growth factor exposure is consistent with extensive data collected from single cell systems. In Dictyostelium, directed migration towards a chemotactic gradient of cAMP is accompanied by accumulation of PIP₃ (mediated by PI3K activity) on the leading plasma membrane, while PTEN is sequestered in the back of the cell (Iijima, Devreotes. 2002; Iijima, et al. 2002). Although chemotaxis still occurs in the presence of LY294002, or in cells lacking all five Dictyostelium PI3K genes, strong chemotactic gradients are required and directional migration is slower and less efficient (Kolsch, et al. 2008). This pathway is highly conserved, as mammalian neutrophils with genetic inactivation of PI3Kγ also migrate more slowly towards an fMLP gradient compared to controls (Ferguson, et al. 2007). Intriguingly, in several recent studies, TORC2 signaling has been shown to be an additional critical regulator of directed single cell migration, functioning in parallel with PI3K signaling modules (Cai, et al. 2010; Charest, et al. 2010; Liu, et al. 2010). Although it has been known for some time that Dictyostelium mutant for Pianissimo (the homologue of mammalian rictor, a required component of TORC2) fail to chemotax, two independent reports have detailed a PIP₃-independent pathway by which chemoattractant leads to the activation of TORC2 at the leading edge of the cell membrane, activating PKB-mediated downstream signaling events that are restricted to the cell’s leading edge even in the absence of PIP₃ (Cai, et al. 2010; Charest, et al. 2010). This pathway is independent of PI3K signaling, and may be especially important for cell migration under conditions of particularly low or high PIP₃ levels (Cai, et al. 2010). As with PIP₃-dependent chemotaxis, this pathway is also highly conserved and recent work has documented that it plays an important role in mammalian neutrophil chemotaxis (Liu, et al. 2010).

Although there is abundant evidence linking PI3K/mTOR signaling to directed cell migration in single cell systems, there are only limited data to suggest that the role of these pathways is conserved during migration of multicellular epithelial structures during organogenesis. In one of the first studies to address the question of PI3K involvement in epithelial chemotaxis during mammalian development, Tang et al found that MDCK cells migrating towards a gradient of GDNF activate PI3K and that LY294002 exposure blocks chemotaxis in this cell line system. To confirm this results using a multicellular tissue, they also demonstrated that directed migration of ureteric bud tissue towards a localized source of GDNF is abrogated by LY294002 exposure. Similarly, a number of studies have demonstrated that PIP₃ accumulation at the leading edge of the cell is conserved during kidney cell migration. Using MDCK cells in three dimensional cultures, Yu et al demonstrated that HGF exposure results in PIP₃ accumulation in cytoplasmic extensions that form prior to chain migration (Yu, et al. 2003). Along the same lines, an additional study found that Ret9 kidney cells localize PIP₃ to cytoplasmic extensions oriented towards a GDNF coated bead, and ectopic PTEN expression inhibits this directed cell migration (Tang, et al. 2002).

While these studies suggest that PI3K signaling may regulate the directed migration of multicellular epithelial tissues in much the same way as in single cell systems, none directly examined whether mTORC2 signaling may play a similar role. Our finding that specific mTOR kinase inhibitors block prostate budding as effectively as combined PI3K/mTOR inhibition strongly suggests that mTORC2 signaling may play an equally important role as PI3K in regulating epithelial migration during development. Although we did not explicitly address the question of whether mTORC2 inactivation can impair prostatic epithelial cell motility during development in the absence of PI3K and/or mTORC1 inhibition, future genetic inactivation studies may be useful to resolve this question. Germline mTORC2-loss-of-function in the mouse is embryonic lethal, but the recent availability of mice with a conditional null allele of rictor, a required component of the mTORC2 makes it now feasible to systematically examine the role of mTORC2 in epithelial morphogenesis in any number of systems (Shiota, et al. 2006).

Although PI3K/mTORC2 signaling is required for prostatic branching, our study also reveals a novel inhibitory role for mTORC1 signaling during branching morphogenesis. Our finding that specific mTORC1 inhibition leads to increased and longer prostatic buds is especially surprising given that mTOR kinase inhibition has the opposite effect. Because mTORC1 inhibition activates PI3K/mTORC2 signaling through eradication of a well-characterized negative feedback loop, we hypothesized that the apparent inhibitory effects of mTORC1 might be mediated by decreased PI3K/mTORC2 signaling. However, when we used a conditional PTEN loss-of-function model to simultaneously activate mTORC1 and PI3K/mTORC2 signaling, we still observed a mTORC1-dependent decrease in prostatic budding. Thus, we conclude that mTORC1 inhibits prostatic branching through a mechanism independent of the known negative feedback loop between mTORC1 and PI3K/mTORC2. Interestingly, a recent study of lung morphogenesis revealed that this inhibitory role of mTORC1 is likely conserved in multiple organ systems. Scott et al found that rapamycin treatment of fetal lung explants led to a dramatic increase in airway epithelial surface complexity, a measure of airway branching (Scott, et al. 2010), and we have had similar findings in a breast branching morphogenesis system (unpublished data). Although the mechanism by which mTORC1 suppresses epithelial branching is unclear at this point, mTORC1 activity is known to stimulate differentiation and/or exhaustive proliferation of stem/progenitor cells in multiple systems (Gan, et al. 2008; Chen, et al. 2008; Castilho, et al 2009; Sun, et al. 2010). It is tempting to speculate that mTORC1-driven progenitor cell depletion may play a similar role in limiting epithelial branching morphogenesis.

Finally, the fact that mTORC1 plays an inhibitory role in branching, while the inter-related PI3K/mTORC2 signaling pathway plays a permissive role might at first seem surprising. However a number of recent studies suggest that in many systems the outcome of mTOR activity may be entirely dependent on the signaling complex with which it is associated. For example, during T cell development, mTORC1 activity directs T(H)1 and T(H)17 differentiation while mTORC2 has the opposite effect, enabling T(H)2 differentiation (Delgoffe, et al. 2011). mTORC1 and mTORC2 likely exist in a state of mutual inhibition resulting from competition for mTOR kinase and numerous negative feedback loops (Hwang, et al. 2008). In this way, the balance of mTORC1 and mTORC2 activity may provide a natural homeostatic mechanism that modulates a number of developmental processes, ranging from differentiation to branching morphogenesis.

Taken together, our findings have important potential implications for understanding the signaling pathways that regulate prostate cancer invasion and motility. There is little doubt that embryonic prostate development and prostatic tumorigenesis, both processes in which the epithelium invades surrounding stroma, bear striking resemblance to one another, both on a morphologic and gene expression level (Schaeffer, et al. 2008; Pritchard, et al. 2009). In this way, studies of each process can inform work on the other. Importantly, PI3K/mTOR signaling is frequently aberrantly activated in prostate tumors, most commonly by PTEN inactivation (Taylor, et al. 2010) and many of the genes up-regulated during prostatic tumorigenesis in the PTEN−/− mouse model map to the branching morphogenesis stage of prostatic development (Pritchard, et al. 2009). Our data extend these findings to suggest that some of the same signaling pathways may be required for both prostatic morphogenesis and tumorigenesis.

Perhaps surprisingly given the frequency of PTEN loss in prostatic tumors, we found that PTEN loss by itself inhibited prostatic epithelial invasion into stroma during development. While our genetic recombination system precludes us from definitively determining whether this effect of PTEN loss occurs only in the presence of 4-OHT (which alters estrogenic signaling), we found that PTEN loss actually increased epithelial branching in the context of low mTORC1 activity. An important question for future research is whether this finding holds true during tumorigenesis. Does PI3K/mTORC2 activation in the absence of mTORC1 activity enhance the efficiency of prostate tumor cell invasion? Our data predict that PI3K and mTORC2 inhibitors might be particularly effective for blocking prostate tumor invasion. Along these lines, a recent study has established that mTORC2, but not mTORC1, activity is required for prostatic tumorigenesis in the PTEN−/− mouse model (Guertin, et al. 2009). Further, a number of clinical trials have suggested that pharmacologic inhibition of mTORC1 signaling is insufficient as a monotherapy for solid tumors (Guertin, Sabatini. 2009). Based on data presented here, we would predict that mTORC1 inhibitor treatment might not only be ineffective, but might have the paradoxical effect of enhancing tumor invasion in prostate cancer, particularly in the setting of PTEN inactivation. Future work will focus on testing these hypotheses in genetically engineered mouse tumor models and human tumor samples.

Supplementary Material

01. Movie 1A.

Time-lapse DIC microscopy of mesenchyme-free urogenital sinus epithelial (UGE) cultures (E15.5) demonstrates the emergence of the anterior and ventral prostate lobes in the vehicle control sample. Images taken every hour for 41 hours.

02. Movie 1B.

Time-lapse DIC microscopy of mesenchyme-free urogenital sinus epithelial (UGE) cultures (E15.5) demonstrates absent lobation and only small abortive buds in the LY294002-treated sample. Images taken every hour for 41 hours.

03. Movie 2A.

Vehicle control-treated urogenital sinus epithelial (UGE) tissues from R26ERCre;mT/mG transgenic mice show mosaic expression of membranous EGFP in the absence of 4-OHT addition, allowing three-dimensional tracking of epithelial cell migration over time using time-lapse epifluorescence imaging of mesenchyme-free cultures. Cytoplasmic protrusions extend in the direction of cell motility and individual cell divisions can be appreciated. Images taken every hour for 59 hours.

04. Movie 2B.

25 μM LY294002-treated urogenital sinus epithelial (UGE) tissues from R26ERCre;mT/mG transgenic mice show decreased cell motility. In contrast to vehicle control-treated cells (Movie 2A), LY294002-treated epithelial cells assume an elongated shape with extension and retraction of protrusions, but net migration is minimal. Images taken every hour for 50 hours.

05. Supplementary Figure 1.

(A) Dose-dependent attenuation of urogenital sinus branching by day 7 of culture in PI3K/mTOR inhibitor LY294002. Prostatic epithelial branches reach the edge of the surrounding mesenchymal tissues in vehicle control-treated tissues but only extend partly into the surrounding mesenchyme when treated with LY294002 at 10 uM. Treatment with 20 μM LY294002 completely abrogates prostatic branching, similar to the 25 uM dosage used in Figures 2, 3 and 4. (B) Corresponding dose- dependent attenuation in AKT phosphorylation by immunoblot after 24 hours of UGS culture.

06. Supplementary Figure 2.

(A) Culture of urogenital sinus tissues for 7 days in 10 uM bpV(pic), a vanadate compound that inhibits PTEN phosphatase, results in decreased prostatic branching. (B) Corresponding increase in AKT phosphorylation after 24 hours of drug incubation confirms PTEN enzymatic inhibition.

Highlights.

• We examine the roles of the inter-related PI3K/mTOR signaling pathways in prostatic branching morphogenesis.

• PI3K activity is up-regulated and required for prostatic budding.

• PI3K inhibition does not alter epithelial proliferation, apoptosis or specification, but instead modulates the efficiency of epithelial cell migration.

• While PI3K and/or downstream mTORC2 activity is required for epithelial budding, mTORC1 activity specifically inhibits prostatic branching.

• Thus, mTOR kinase activity has differential effects on prostatic branching depending on the signaling complex with which it associates.