SPATIALLY-RESTRICTED STROMAL WNT SIGNALING RESTRAINS PROSTATE EPITHELIAL PROGENITOR PROLIFERATION THROUGH DIRECT AND INDIRECT MECHANISMS

Summary

Cell-autonomous Wnt signaling has well-characterized functions in controlling stem cell activity, including in the prostate. While niche cells secrete Wnt ligands, the effects of Wnt signaling in niche cells per se are less understood. Here we show that stromal cells in the proximal prostatic duct near the urethra, a mouse prostate stem cell niche, not only produce multiple Wnt ligands but also exhibit strong Wnt/b-catenin activity. The non-canonical Wnt ligand Wnt5a, secreted by proximal stromal cells, directly inhibits epithelial prostate stem/progenitor proliferation whereas stromal cell-autonomous canonical Wnt/b-catenin signaling indirectly suppresses prostate stem/progenitor activity via the TGFb pathway. Collectively, these pathways restrain the proliferative potential of epithelial cells in the proximal prostatic ducts. Human prostate likewise exhibits spatially-restricted distribution of stromal Wnt/b-catenin activity, suggesting a conserved mechanism for tissue patterning. Thus, this study shows how distinct stromal signaling mechanisms within the prostate cooperate to regulate tissue homeostasis.

Graphical Abstract

eTOC

Wei et al. show that human and mouse prostate stromal cells exhibit region-specific differences in Wnt ligand expression and Wnt/b-catenin activity. Canonical and non-canonical Wnt signaling cooperate to restrain prostate stem/progenitor activity through direct and indirect, TGFb-dependent pathways.

Introduction

Wnt signaling is a fundamental growth control pathway (Clevers et al., 2014; Nusse and Clevers, 2017). It has been shown to play critical roles in regulating stem cell self-renewal, pluripotency, proliferation, and regeneration in various organ systems (Korinek et al., 1998; Nguyen et al., 2009; Reya et al., 2003; Silva et al., 2008; van Amerongen et al., 2012). Wnt ligands are hydrophobic, tethered to cell membranes, and act as short-range morphogens (Nusse and Clevers, 2017). Therefore, Wnt signaling in stem cells is usually activated through the ligands produced by their adjacent niche cells, such as Paneth cells and telocytes in the small intestine (Sato et al., 2011; Shoshkes-Carmel et al., 2018), dermal papilla in the skin (Reddy et al., 2001), and stromal cells in the lung and mammary gland (Miller et al., 2012; Zhao et al., 2017) etc. However, much attention has been focused on understanding how stem cell-autonomous Wnt signaling affects its biology, whereas little is known regarding whether niche cells also possess active Wnt signaling and how this may affect the biology of the stem cells (Yamane et al., 2001). We seek to employ the prostate as an organ system to investigate how Wnt signaling in niche cells affects prostate stem cell biology.

Prostate epithelial cells turn over slowly, but they possess extensive regenerative potential when stimulated by cycles of androgen deprivation and replacement (Isaacs, 1985). There are three types of epithelial cells in the prostate: the predominant secretory luminal cells, the basal cells that sit between the luminal cells and basement membrane, and the rare neuroendocrine cells that are less well characterized. In vivo lineage tracing studies have demonstrated that the basal and luminal cells in adult mice are independently sustained by their respective progenitors (Choi et al., 2012; Liu et al., 2011; Lu et al., 2013; Wang et al., 2013). Whether the human prostate epithelial lineage hierarchy is completely the same remains to be determined. But the basal and luminal progenitors in humans and mice are both functionally plastic and exhibit facultative capability for at least bipotent differentiation under various experimental and pathological conditions (Chua et al., 2014; Collins et al., 2001; Goldstein et al., 2010; Karthaus et al., 2014; Kwon et al., 2013; Leong et al., 2008; Xin et al., 2005).

The mouse prostate surrounds the urethra at the base of the bladder. It contains 4 different lobes that consist of tubules budding outward away from the urethra. The tubular structures near the urethra are termed proximal prostatic ducts, whereas those at the tips are called distal ducts. The epithelial cells in the proximal ducts in adult mice, including both the basal and luminal cells, have been shown to retain BrdU or histone H₂B-GFP labeling, thereby are considered quiescent (Tsujimura et al., 2002; Zhang et al., 2018). We and others showed that the stem cell activities of both the basal and luminal cell lineages are enriched in the proximal prostatic ducts (Burger et al., 2005; Kwon et al., 2015; Xin et al., 2005; Yoo et al., 2016).Therefore, the proximal prostatic duct has been considered a prostate stem cell niche in mice. But how the signaling in this niche regulate the prostate stem/progenitor activity is understudied.

Epithelial cell autonomous Wnt/β-Catenin signaling has been shown to regulate prostate stem cell specification, proliferation, and homeostasis. Wnt/β-Catenin signaling is essential for prostate epithelial lineage specification during development (Francis et al., 2013; Lee et al., 2015; Simons et al., 2012) and for basal progenitors to generate luminal cells in adult mice (Lu and Chen, 2015). Although Wnt/β-Catenin signaling is not necessary for the survival or proliferation of epithelial cells in adults (Francis et al., 2013; Simons et al., 2012), elevated epithelial Wnt/β-Catenin signaling induces proliferation both in vitro and in vivo (Francis et al., 2013; Karthaus et al., 2014; Shahi et al., 2011; Simons et al., 2012; Yu et al., 2011). In contrast, little is known about whether active Wnt/β-Catenin signaling is present in other cell lineages in the prostate stem cell niche and how it may affect prostate stem/progenitor activity. In this study, we show that Wnt/β-Catenin signaling is active in the stromal cells at the prostate stem cell nice and demonstrate an important role of this stromal Wnt/β-Catenin signaling in regulating prostate homeostasis.

Results

Stromal cells in proximal prostatic ducts express Wnt ligands and possess active Wnt/β-Catenin signaling

The mouse prostate stem cell activity is enriched at the proximal prostatic duct near the urethra, a prostate stem cell niche in mice (Supplementary Fig. 1A) (Burger et al., 2005; Kwon et al., 2015; Xin et al., 2005; Yoo et al., 2016; Zhang et al., 2018). Both basal and luminal cells in this region in adult mice are mitotically quiescent (Tsujimura et al., 2002). They express some lineage-specific antigens as expected (Keratin 5/14 and P63 for basal cells; Keratin 8 for luminal cells) (Supplementary Fig. 1B). However, the proximal luminal cells differ from distal luminal cells in that they express Sca-1 but not Nkx3.1 (Kwon et al., 2015) (Supplementary Fig. 1B). Using flow cytometric analyses, we found that even at the age of 5 weeks when the prostate is still undergoing active morphogenesis, BrdU was incorporated at a lower level in all three major cell lineages (basal, luminal, and stromal cells) at the proximal ducts than at the distal ducts (Fig. 1). This result indicates that the lower proliferative potential is a persistent property of the proximal cells and led us to hypothesize that a unique signaling in the microenvironment at the proximal prostatic ducts suppresses the proliferative potential of the stem/progenitor cells.

Fig. 1: Mouse proximal prostate cells are mitotically quiescent compared to those at distal prostatic ducts. See also Figure S1.

(A) Schematic illustration of experimental design. (B) FACS plots of BrdU⁺ cells in Lin⁻ CD49f⁺Sca-1⁺ basal cells, Lin⁻CD49f⁺Sca-1⁺ stromal cells, Lin⁻CD49f^LowSca-1⁻ distal luminal cells, and Lin⁻CD49f^LowSca-1⁺ proximal luminal cells. (C) Dot graphs show means ± s.d. of percentage of BrdU⁺ cells in individual cell lineages from three independent experiments.

To identify such signaling, we performed an RNA-seq analysis using FACS-isolated Lin⁻CD24⁻CD49f⁻Sca-1⁺ stromal cells from adult mouse proximal and distal prostatic ducts. Fig. 2A shows that only 506 genes are differentially expressed by at least 1.4-fold. Interestingly, among the 309 genes that are upregulated in the proximal stromal cells, six are the core components of the Wnt signaling (Wnt5a, Lgr5, Sfrp2, Sfrp3, Sfrp4, and Sfrp5), which is further confirmed by qRT-PCR analysis (Fig. 2B). Some Sfrps were positively regulated by the Wnt/β-Catenin signaling (Lescher et al., 1998). This prompted us to investigate whether the stromal cells possess active Wnt/β-Catenin signaling. We performed a Wnt pathway-focused PCR array to compare the expression of 84 Wnt-related genes in FACS-sorted Lin⁻CD24⁻CD49f⁻Sca-1⁺ stromal cells from proximal and distal prostatic ducts. Fig. 2C shows that 46 genes are significantly upregulated in the proximal stromal cells, including Porcn, an essential enzyme for Wnt ligand secretion, 10 Wnt ligands (Wnt2, 2b, 3, 4, 5a, 5b, 6, 8a, 9a, and 10a), 9 receptor complex components (Fzd2-9, Dvl2, and Lrp6) and 9 typical Wnt downstream target genes (Axin2, Sfrp2, and Lgr5 etc.). The differential expression of several major genes is shown in Fig. 2D and Supplementary Table 1.

Fig. 2: Mouse proximal prostatic stromal cells express Wnt ligands and possess active Wnt/β-Catenin signaling. See also Figure S2. and Table S1.

(A) Heatmap of RNA-seq analysis of FACS-isolated stromal cells from adult mouse proximal and distal prostatic ducts. (B) qRT-PCR analysis of 6 representative Wnt pathway-related genes in FACS-isolated proximal and distal prostate stromal cells. Data represent means ± s.d. from five independent experiments. (C) Volcano graph shows fold differences and p-value of expression levels of 84 Wnt-related genes between proximal and distal prostatic stromal cells determined by a Wnt pathway-focused RT² PCR profiler array. (D) qRT-PCR analysis of 6 representative Wnt pathway-related genes identified in RT² PCR profiler array in FACS-isolated proximal and distal prostate stromal cells. Data represent means ± s.d. from three independent experiments. Pro: proximal stromal cells, Dis: distal stromal cells. (E) RNA-in-situ staining of Axin2, Wnt2b and Wnt5a in anterior prostate of 8-week-old C57BL/6 mice. E: epithelial cells, S: stromal cells. Bars = 20μm. (F) Co-immunostaining of GFP/Sca-1 and GFP/K5 in anterior prostate of 8-week-old Tcf/Lef:H₂B-GFP mice. Yellow arrows denote GFP⁺ cells at inter-glandular areas at proximal ducts. White arrows show GFP⁺ epithelial cells at distal ducts. Bars = 50μm. Dot graph shows means ± s.d. of percentage of GFP⁺ cells in proximal and distal prostate stromal cells from 3 mice. Each dot represents result from one image. (G) RNA-in-situ analysis of Axin2 in E15 and E18 UGS, and anterior prostate of 2, 5, 8-week-old C57BL/6 mice. E: epithelia; S: Stroma. Yellow arrows and asterisk point to Axin2 expression in stroma and epithelia, respectively. Bars = 20μm.

The expression levels of these Wnt signaling components are much lower in the basal cells (Supplementary Table 1). This result indicates that the proximal stromal cells not only produce Wnt ligands but also possess active Wnt/β-Catenin signaling.

We performed RNA-In-Situ analyses to corroborate the spatially-restricted expression patterns of three representative genes: Wnt2b, Wnt5a, and Axin2. Fig. 2E shows that they are all highly expressed in the proximal stromal cells than in distal stromal cells. Interestingly, Axin2 is expressed at a higher level in the distal epithelial cells than proximal epithelial cells, indicating a distinct spatially-restricted pattern for the Wnt/β-Catenin activity in both the epithelial and stromal cells.

Finally, we used a TCF/LEF:H₂B-GFP Wnt activity reporter mouse line to corroborate active Wnt/β-Catenin signaling in the proximal stromal cells. This transgenic model harbors a nuclear-localized GFP-expressing cassette driven by a 7xTCF/LEF promoter (Ferrer-Vaquer et al., 2010). We showed previously that only the luminal cells at proximal ducts, but not those at distal ducts, express Sca-1 (Kwon et al., 2015). Co-immunostaining of GFP with Sca-1 and cytokeratin5 (K5), respectively, shows that 12% of the cells in the inter-glandular areas at the proximal ducts express GFP (yellow arrows, Fig. 2F), whereas only 5% is GFP⁺ at the distal ducts. These GFP⁺ cells are mostly fibroblasts because they express vimentin and very little α-smooth muscle actin, but not the markers for leukocytes (CD45) and endothelial cells (CD31) (Supplementary Fig. 2A). Note that there are more GFP⁺ Wnt active epithelial cells in distal ducts (white arrows, Fig. 2F), which is consistent with the RNA-In-Situ result (Fig. 2E) showing that Axin2 is expressed at a higher level in the distal epithelial cells.

We further investigated the dynamics of Axin2 expression during prostate development. Fig. 2G shows that Axin2 is strongly and almost exclusively expressed in the urogenital sinus mesenchyme (UGSM) at E15. By E18, UGS epithelia (UGSE) start to bud into the surrounding UGSM. Interestingly, the expression of Axin2 is lower in the mesenchyme surrounding the epithelial buds (yellow arrows in E18 panel, Fig. 2G). By 2 weeks of age, it is already clear that Axin2 is expressed at a higher level in the stromal cells at the proximal ducts than distal ducts, and this pattern remains the same at the 5 weeks and 8 weeks of age (yellow arrows, Fig. 2G).

On the other hand, Axin2 is not expressed in UGSE at E15, but starts to be expressed at E18 in the epithelial cells especially those at the tips of epithelial buds (asterisk in E18 panel, Fig. 2G), but the expression is still significantly lower than that in the mesenchyme. In 2-week-old mice, Axin2 is predominantly expressed in the distal epithelial cells, but the expression declines at 5 and 8 weeks of age. The active stromal Wnt/β-Catenin signaling during prostate development is also corroborated using the TCF/Lef:H₂B-GFP Wnt activity reporter mice (Supplementary Fig. 2B). Collectively, these studies reveal that the proximal stromal cells not only highly express Wnt ligands but also possess stronger Wnt/β-Catenin activity than the distal stromal cells.

Non-canonical Wnt signaling suppresses epithelial proliferation at proximal prostatic ducts

Wnt/β-Catenin signaling can promote prostate epithelial proliferation (Karthaus et al., 2014; Shahi et al., 2011). Paradoxically, although the cells in the proximal ducts are exposed in a microenvironment with abundant Wnt ligands, their proliferative potential is lower. Some Wnt ligands (Wnt5a and Wnt4) expressed by the proximal stromal cells mediate non-canonical Wnt signaling. Wnt5a can suppress prostatic epithelial budding during early development(Huang et al., 2009). In addition, non-canonical Wnt signaling can antagnize the canonical signaling (Sato et al., 2010). Therefore, we reasoned that the lower proliferative potential of epithelial cells in proximal region is partly due to the non-canonical Wnt signaling.

We first examined the expression of the non-canonical Wnt receptors (Ror1, Ror2 and Ryk) in epithelial cells by qRT-PCR. Among them, Ror2 is not only the most highly expressed but also expressed at a higher level in the proximal ducts than in distal ducts (Fig. 3A and Supplementary Fig. 3A). We performed co-immunostaining of Ror2 with K5 and Nkx3.1, respectively. Nkx3.1 is expressed by the luminal cells at distal ducts but not in those at proximal ducts (Kwon et al., 2015). Fig. 3B corroborates a higher expression of Ror2 in both the proximal basal and luminal cells. We reasoned that the higher Wnt5a level at the proximal ducts mediates a higher non-canonical activity in the basal cells. To test this hypothesis, we compared the gene expression profiles of FACS-isolated Lin⁻CD49f⁺Sca-1⁺ basal cells in adult mouse proximal and distal prostatic ducts using RNA-seq analyses. We found 765 genes that are differentially expressed by at least 1.4-fold between the two groups (Fig. 3C). Fig. 3D shows that several gene ontology biology processes related to the non-canonical Wnt signaling are enriched in the proximal basal cell group, including calcium ion binding, Rho protein signaling transduction, intermediate filament organization, apical plasma membrane and regulation of establishment of planar polarity. The differential expressions of several representative genes that have been associated with noncanonical Wnt signaling in these Gene Ontology groups including Camk2, Cdc42, Rhoj, Plce1, Wnt7b (Zheng et al., 2013), and Tnfrsf11a (Maeda et al., 2012) were confirmed by qRT-PCR analysis (Supplementary Fig. 3B). These results support an enriched non-canonical Wnt activity in the proximal epithelial cells.

Fig. 3: Non-canonical Wnt signaling suppresses epithelial proliferation at proximal ducts. See also Figure S3.

(A) qRT-PCR of Ror2 in FACS-isolated prostate basal and luminal cells at proximal and distal prostatic ducts. Data represent means ± s.d. from 3 independent experiments. (B) Immunostaining of Ror2, Nkx3.1, and Keratin 5 (K5) in anterior prostates of 8-week-old C57BL/6 mice. Bars = 50 μm. (C) Heatmap of RNA-Seq analysis of FACS-isolated basal cells from proximal and distal prostatic ducts. (D) Gene ontology analysis of RNA-Seq of FACS-isolated basal cells from proximal and distal prostatic ducts. (E) Sphere-forming units and sphere sizes of FACS-isolated prostate basal cells with and without 100ng/ml Wnt5a. Data show means ± s.d. from 3 independent experiments. Data of sphere size show results of one representative experiment unless otherwise noted. (F) Organoid-forming units and organoid sizes of FACS-isolated prostate Sca-1⁺ luminal cells with and without 100ng/ml Wnt5a. Data show means ± s.d. from 5 independent experiments. n.s.: not significant. (G-H) Immunostaining of Ki67 and Cleaved caspase 3 (CC3) in spheres treated with Wnt5a and w/o Wnt5a (CTR). Each dot represents result from one sphere.

We further investigated how Wnt5a affects the prostate stem/progenitor activity. A prostate sphere assay (Xin et al., 2007) and a prostate organoid assay (Karthaus et al., 2014) have been widely used to measure the prostate basal and luminal stem/progenitor cell activity, respectively. In these assays prostate stem/progenitor cells can form clonogenic spheres or organoids in defined media. Fig 3E shows that Wnt5a does not affect the sphere-forming activity but reduces the sphere size by 15%. The luminal cells (Sca-1⁺) in proximal ducts also express a higher level of Ror2 than those at distal ducts (Fig. 3A). The luminal cells are separated from the Wnt5a-generating stromal cells by only a thin layer of basal cells, so Wnt5a may diffuse in short distance to reach luminal cells. In addition, the proximal basal cells also express Wnt5a more than distal basal cells, although the expression level is much lower than in stromal cells (Supplementary Fig. 3C). Therefore, we also investigated how Wnt5a affects organoid growth from the Sca-1⁺ proximal luminal cells. Fig. 3F shows that Wnt5a also reduced the organoid size by 14% although the organoid-forming activity was not affected significantly (Fig 3F). The reduction of sphere/organoid size is due to decreased cell proliferation but not elevated apoptosis (Figs. 3G and 3H).

The non-canonical Wnt signaling can antagonize the canonical Wnt signaling in other organ systems (Roarty et al., 2015; Sato et al., 2010). We confirmed that Wnt5a can suppress Wnt3a-stimulated Axin2 expression in the prostate basal sphere cells (Supplementary Fig. 3D) and partially inhibit the Wnt3a-stimulated TCF/LEF reporter activity (Supplementary Fig. 3E). In addition, Wnt5a is less effective in reducing the size of prostate organoids in the absence of R-Spondin (Supplementary Fig. 3F). These results support that the lower proliferative potential of the prostate stem/progenitor cells in the proximal ducts is partially caused through suppression of the Wnt/β-Catenin signaling by Wnt5a.

Stromal Wnt/β-Catenin signaling suppresses proliferation of prostate epithelial cells

We then investigated how the stromal Wnt/β-Catenin signaling affects the activity of prostate basal stem/progenitor cells. In the prostate sphere assay, embryonically derived urogenital sinus mesenchymal cells (UGSM) can stimulate sphere-forming activity of basal cells (Xin et al., 2007). We infected UGSM cells with a control RFP-expressing lentivirus and a lentivirus expressing both RFP and a stabilized S37A β-Catenin mutant separately (Supplementary Fig. 4A). Dissociated single mouse prostate epithelial cells were cocultured with the engineered UGSM cells separately. Fig. 4A shows that the sphere forming activity and size were reduced by 17% and 15%, respectively, in the S37A β-Catenin group. Expression of a dominant negative β-Catenin in the S37A β-Catenin-expressing UGSM cells ablated their capacity to suppress prostate sphere growth and formation (Fig. 4A). The S37A β-Catenin-expressing and control stromal cells survived at the same rate in the assay, excluding the possibility that the reductions in sphere formation and size were caused by different stromal cell densities (Supplementary Fig. 4B). When the prostate spheres in the two groups were passaged with the control stromal cells separately, they formed secondary spheres of the same size and at the same frequency (Supplementary Fig. 4C). Collectively, these results suggest that the signaling mediated by the S37A β-Catenin-expressing stromal cells suppress the proliferative potential of prostate sphere cells but not their capacity for self-renewal. The growth inhibitory effect of S37A β-catenin is not unique to UGSM cells as we obtained the same results using primary adult mouse prostate stromal cells (Supplementary Fig. 4D).

Fig. 4: Stromal Wnt/β-Catenin signaling suppresses proliferation of prostate basal stem cell in vitro. See also Figure S4.

(A) Sphere-forming units and sphere sizes of prostate epithelial cells cultured w/o UGSM, with control wild type UGSM cells, with UGSM cells expressing S37A-βCatenin, and with UGSM cells expressing dominant negative (DN) βCatenin, respectively. Data show means ± s.d. from 3 independent experiments. (B) Transillumination image of regenerated prostate tissues. Bar = 5 mm. Dot graph shows means ± s.d. of weight of xenografts. N=6. (C) Co-immunostaining of BrdU and K5 in regenerated prostatic tissues derived from prostate epithelial cells stimulated by control UGSM and UGSM expressing S37A-BCatenin. Bars = 50 μm. Dot graph shows means ± s.d. of percentage of BrdU⁺ epithelial cells in regenerated prostatic tissues. Each dot represents result from one image. Data collected from 4 pairs of independent samples. (D) Schematic illustration of experimental design. Control: WT littermate mice; Col1a2-CA: Col1a2-CreER^T2;Ctnnb1^lox(ex3) mice. Tmx: tamoxifen. AP: anterior prostate. (E) Transillumination images of prostatic xenografts after tamoxifen treatment. Dot graph shows quantification of weight of 10 pairs of xenografts analyzed by paired t-test. (F) Co-immunostaining of BrdU and K5 in prostatic xenografts. Bars = 50 μm. Dot graph shows means ± s.d. of percentage of BrdU⁺ epithelial cells in prostatic xenografts. Each dot represents result from one image. (G) Schematic illustration of experimental design. Tmx: tamoxifen. (H) Co-immunostaining of BrdU and K5 in anterior prostates of Col1a2-CreER^T2;Ctnnb1^loxp/loxp (Col1a2-KO) and littermate C57Bl/6 (control) mice. Bars = 50 μm. Dot graph shows means ± s.d. of percentage of BrdU⁺ epithelial cells. Each dot represents result from one image. Data collected from 4 pairs of mice.

We sought to corroborate the impact of the stromal Wnt signaling on prostate stem/progenitor cell activity in vivo using a prostate regeneration assay (Xin et al., 2003). Dissociated adult mouse prostate epithelial cells were mixed with the control and S37A β-Catenin-expressing UGSM cells separately and incubated under the renal capsules of immunodeficient male mice for 2 months. Fig. 4B shows that the S37A β-Catenin-expressing UGSM cells are 50% less efficient than the control RFP-expressing UGSM cells in promoting prostate regeneration. Immunostaining shows that there were 31% less BrdU⁺ proliferating epithelial cells in the regenerated tissues in the S37A β-Catenin group than in the control group (Fig. 4C), whereas apoptotic index was not different (Supplementary Fig. 4E). This study confirms that higher stromal Wnt/β-Catenin signaling suppresses the proliferative potential of prostate stem/progenitors.

We further extended this study using genetically engineered mouse models. A Col1a2-CreER^T2 model and a Ctnnb1^lox(ex3) model (Harada et al., 1999) were used to genetically activate Wnt/β-Catenin signaling in the prostate stromal cells. Using an eYFP reporter line, we demonstrated that the Col1a2-CreER^T2 line can specifically and efficiently target prostate stromal cells (Supplementary Fig. 4F). Five-week-old Col1a2-CreER^T2;Ctnnb1^lox(ex3) bigenic mice were treated with tamoxifen to stabilized β-catenin in the prostate stromal cells (Supplementary Figs. 4G–I). Stabilization of β-catenin caused a reduction in prostatic weight and epithelial proliferative index (Supplementary Figs. 4J–K). However, stabilization of β-catenin in multiple organs through the Col1a2 promoter led to a 10% reduction in body weight, thickened skin, and smaller testis (Supplementary Fig. 4L), although the androgen receptor (AR) signaling is not altered as indicated by the strong epithelial AR nuclear staining (Supplementary Fig. 4M). To further corroborate that the reduced prostate size was a direct consequence of β-catenin activation in the prostate stromal cells, we transplanted anterior prostate tissues of 3-week-old Col1a2-CreER^T2;Ctnnb1^lox(ex3) and Col1a2-CreER^T2 mice under the renal capsules of immunodeficient male hosts. The hosts were treated with tamoxifen to turn on the stromal expression of the stabilized β-Catenin in the xenografts. The xenografts were collected 4 weeks later (Fig. 4D). Fig. 4E shows that the average weight of the xenografts in the Col1a2-CreER^T2;Ctnnb1^lox(ex3) group is only 85% of that in the control group. The proliferative index of epithelial cells in the Col1a2-CreER^T2;Ctnnb1^lox(ex3) group was 53% less than that of the control group (Fig. 4F). Collectively, these results demonstrate that activation of Wnt/β-Catenin signaling in the stromal cells suppresses prostate epithelial proliferation.

Conversely, we utilized the Col1a2-CreER^T2 and Ctnnb1^loxp/loxp models (Huelsken et al., 2001) to disrupt Wnt/β-Catenin signaling in the prostate stromal cells (Fig. 4G). Quantitative RT-PCR analyses confirmed that β-Catenin was specifically and efficiently knocked out in approximately 70% of the prostate stromal cells (Supplementary Figs. 4N–P). Immunostaining shows that the proliferative index of epithelial cells in the Col1a2-CreER^T2;Ctnnb1^loxp/loxp mice was 3-fold higher than that in the control group (Fig. 4H). Collectively, these results demonstrate that stromal Wnt/β-Catenin activity suppresses prostate epithelial proliferation.

Wnt/β-Catenin signaling transcriptionally upregulates TGFβ ligands in prostate stromal cells

Using a chamber assay, we demonstrated that direct contact between the stromal and basal cells is not necessary for the S37A β-Catenin-expressing stromal cells to suppress prostate sphere formation and growth (Supplementary Fig. 5A–C). This indicates that the inhibitory effect is mediated via paracrine factors. To identify the factors, we compared the gene expression profiles of the primarily cultured adult mouse prostate stromal cells that express S37A β-Catenin and the control cells that only express RFP. We identified 783 genes that were differentially expressed by at least 1.2-fold (Fig. 5A). Fig. 5B summarizes the differentially enriched Gene Ontology biological processes. As expected, genes associated with the Wnt receptor signaling pathway is enriched in the S37A β-catenin group. Of note, we did not observe upregulation of Sfrps by β-Catenin (Supplementary Fig. 5D), suggesting that the higher expression of Sfrps in the proximal stromal cells may not reflect a negative feedback of the Wnt/β-Catenin signaling.

Fig. 5: β-Catenin transcriptionally upregulates Tgfβ ligands in prostate stromal cells. See also Figure S5.

(A) Heatmap of RNA-Seq analysis of control and β-Catenin S37A-expressing primary prostate stromal cells. The heat map depicts fold changes in experiment versus control. Each gene is centered on average of control. (B) Gene ontology analysis of RNA-Seq analysis of control and β-Catenin S37A-expressing prostate stromal cells. (C) qRT-PCR analysis of Tgfβ1/2/3 in UGSM cells that express red fluorescent protein only (control), β-Catenin S37A, and dominant negative (DN) β-Catenin. Dot graphs represent means ± s.d. from 5 independent samples. (D) qRT-PCR analysis of Tgfβ1/2/3 in FACS-isolated stromal cells from Tmx-treated Col1a2-CreER^T2;Ctnnb1^loxp/loxp (KO) and WT littermate control mice. Dot graphs represent means ± s.d. from 3 pairs of mice. (E) qRT-PCR analysis of Tgfβ1/2/3 in FACS-isolated stromal cells from proximal and distal prostatic ducts. Dot graphs represent means ± s.d. from 4 pairs of mice. (F) Schematic illustration of loci of TCF/LEF binding sites at promoters of Tgfβ2 and Tgfβ3. Two adjacent TCF/LEF binding sites at −4kb of Tgfβ3 promoter are represented in box 1. TSC: translational start codon. (G) Luciferase reporter assays determine activity of Tgfβ2/3 luciferase reporters in prostate stromal cells expressing red fluorescent protein only (control) and S37A β-Catenin. Dot graphs represent means ± s.d. from 3 experiments. (H) ChIP analysis of binding of anti-β-Catenin antibody and IgG control to promoters of Tgfβ2 and 3 in β-Catenin S37A-expressing prostate stromal cells. Two sites at Chr.15 and Gapdh are used as controls for no binding, whereas a site at Axin2 is used as positive control for binding. Histograms show means ± s.d. of relative enrichment from 3 independent experiments.

PDGF receptor and TGFβ receptor binding are the other two GO entries enriched in the S37A β-Catenin group. Although Pdgfc is upregulated in the S37A β-Catenin group, blocking its activity did not alleviate the growth inhibitory effect of the S37A β-Catenin-expressing stromal cells on the prostate spheres (Supplementary Fig. 5E–F). We therefore focused on the TGFβ signaling. QRT-PCR analysis confirms that Tgfβ2 and Tgfβ3 were upregulated by 2.3- and 1.4-fold, respectively, in the S37A β-catenin-expressing prostate stromal cells, but the expression of Tgfβ1 was not significantly different (Fig. 5C). These observations suggest that Tgfβ2 and Tgfβ3 may be upregulated by Wnt signaling in the prostate stromal cells. Consistently, both ligands were downregulated in the FACS-isolated prostate stromal cells from tamoxifen treated Col1a2-CreER^T2;Ctnnb1^loxP/loxP mice than in those of the control Col1a2-CreER^T2 mice, although the reduction of Tgfβ3 was not statistically significant (Fig. 5D). Conversely, in the FACS-isolated stromal cells from proximal prostatic ducts where Wnt signaling is higher, Tgfβ2 is expressed at 3.5-fold higher than in those from distal ducts (Fig. 5E). Tgfβ3 expression is slightly higher but the difference is not statistically significant.

We investigated how Wnt signaling upregulates Tgfβ2 and 3. There are 2 TCF/LEFs consensus binding sites [(A/T)(A/T)CAAAG] (YOCHUM ET AL., 2008) in a 2kb fragment upstream of the translational start codon (TSC) of Tgfβ2, whereas 5 sites are presented 4kb upstream of TSC of Tgfβ3 (Fig. 5F). We cloned these 2 fragments into the pGL3 luciferase reporter and transfected them into primarily cultured mouse prostate stromal cells. Fig. 5G shows that S37A β-catenin increased the luciferase activity by 1.9- and 1.7–fold, respectively, whereas the induction of the luciferase activity was abolished when these TCF/LEFs consensus binding sites were deleted or mutated. Chromatin immunoprecipitation analysis demonstrated that β-Catenin directly binds to two of these sites in primary mouse prostate stromal cells (Fig. 5H). Collectively, these results show that β-Catenin transcriptionally activates Tgfβ2/3 in the prostate stromal cells.

Wnt/β-Catenin-active prostate stromal cells suppress basal cell proliferation via TGFβ

Since Tgfβ suppresses prostate basal stem/progenitor activity (Valdez et al., 2012), we reasoned that stromal Wnt/β-Catenin signaling suppresses basal stem cell proliferation via Tgfβ. To test this hypothesis, we introduced a dominant negative TgfβRII (dnTgfβRII) into the prostate basal stem cells to disrupt their capability to respond to all the Tgfβs (Supplementary Fig. 6A). Briefly, FACS-isolated basal stem cells were infected with a lentivirus that expresses both dnTgfβRII and GFP and a control GFP lentivirus, separately. Infected basal cells were cocultured with the control and S37A β-catenin-expressing primary prostate stromal cells separately in the sphere assay. Consistent with the result shown in Fig. 4A, compared to the RFP-expressing stromal cells, the S37A β-catenin-expressing stromal cells suppressed the sphere formation and growth of prostate basal cells (groups 2 vs 3, Figs. 6A–B). In contrast, the sphere forming activity and size were not significantly different when dnTgfβRII-expressing basal cells were cocultured with the S37A β-catenin-expressing and RFP-expressing stromal cells, respectively (groups 5 vs 6, Figs. 6A–B). Since not all basal cells were infected by the dnTgfβRII-GFP lentivirus, only about 70% of the prostate spheres in this group expressed GFP. The mean size of the GFP⁺ spheres derived from dnTgfβRII-expressing basal cells was 1.3-fold larger than that of the GFP⁻ spheres derived from the uninfected basal cells (groups 6 vs 6a, Fig. 6B). This further demonstrates that disrupting Tgfβ signaling desensitizes basal cells from the growth inhibitory effect of stromal Wnt/β-Catenin signaling. Similarly, when a Pan-Tgfβ blocking antibody was added in the sphere assay to neutralize the activity of Tgfβs, the growth inhibitory effect of the S37A β-catenin-expressing stromal cells on prostate spheres was ablated (Supplementary Fig. 6B). This further corroborates that Tgfβ plays a critical role in the growth inhibition of prostate spheres mediated by the stromal Wnt/β-Catenin signaling.

Fig. 6: Wnt/β-Catenin-active prostate stromal cells suppress basal cell proliferation via Tgfβ See also Figure S6.

(A-B) Dot graphs quantify sphere-forming units (A) and sphere sizes (B) derived from prostate epithelial cells co-cultured with and without UGSM cells. Epithelial cells are infected by lentivirus expressing GFP (CGW-control) or dominant negative (DN) TgfβRII, whereas UGSM cells are infected by lentivirus that express RFP (CRW-control) or S37A β-Catenin. Data show means ± s.d. from three independent experiments. (C) Transillumination images of regenerated prostate tissues derived from prostate epithelial cells infected with lentivirus expressing GFP (control) and dominant negative (DN) TgfβRII, respectively. Bar = 5 mm. Dot graph shows means ± s.d. of xenograft weight. N=5. (D) Co-immunostaining of BrdU and K5 in regenerated prostatic tissues. Bars = 50 μm. Dot graph shows means ± s.d. of percentage of BrdU⁺ epithelial cells in regenerated prostatic tissues. Each dot represents result from one image. Data were collected from 4 pairs of regenerated prostatic tissues. n.s.: not significant.

Finally, we employed the prostate regeneration assay to corroborate the finding in vivo. The control and dnTgfβRII-expressing prostate epithelial cells were incubated with S37A β-catenin-expressing UGSM cells separately under the renal capsules of immunodeficient male mice for 2 months. Fig. 6C shows that the mean weight of regenerated tissue in the dnTgfβRII group is 2-fold heavier, and these tissues contain more regenerated glands (Supplementary Fig. 6C). The proliferative index of epithelial cells was 1.8-fold higher in the dnTgfβRII group (Fig. 6D). Collectively, these studies demonstrate that Tgfβ plays a critical role in the growth inhibitory effect mediated by the Wnt active prostate stromal cells.

Wnt/β-Catenin signaling is higher in stromal cells of human transition zone versus peripheral zone prostate.

Anatomically, the mouse and human prostate glands are different. The human prostate is defined into 4 different zones, of which the peripheral zone (PZ) and transition zone (TZ) are the two major zones (McNeal et al., 1988). PZ covers the dorsolateral and apical parts of the gland, whereas TZ is located between the proximal prostatic urethra and lateral parts of the PZ. Despite the anatomic difference, the mouse proximal prostatic ducts and human TZ share some similarities. They are both anatomically closer to the urethra compared to other prostatic regions. In addition, epithelial cells in mouse proximal prostatic ducts and human TZ are both mitotically inert compared to those in mouse distal ducts and human PZ, respectively (Colombel et al., 1998; Tsujimura et al., 2002). Based on these similarities, we hypothesized that the stromal cells in human TZ also possess higher Wnt/β-Catenin signaling than those in PZ.

We obtained 8 pairs of TZ and PZ stromal cells from formalin-fixed paraffin-embedded normal human TZ and PZ tissues by laser capture microdissection (Supplementary Fig. 7A), and performed a qRT-PCR analysis of AXIN2, WNT5A, TGFB1, TGFB2, and TGFB3. Fig. 7A shows that all these genes except TGFB1 are expressed at a higher level in the TZ stromal cells than in the PZ stromal cells. We attempted to confirm the results using an RNA-In-Situ analysis. The probes for WNT5A and TGFB2 stained too weak in human specimens to be quantitated convincingly. Fig. 7B shows representative staining of AXIN2 in TZ and PZ of the same donor. In TZ, AXIN2 staining is mostly observed in the stromal cells that are immediately adjacent to epithelial glandular structures. In comparison, staining in the stromal cells adjacent to epithelial glands in PZ is weaker. Quantification of pairwise comparison shown in Fig. 7C demonstrates that the difference is statistically significant. Interestingly, expression of AXIN2 appears higher in epithelial cells in PZ than in TZ (Supplementary Fig. 7B). The anatomically distinct expression patterns of AXIN2 between epithelial cells in TZ and PZ are very similar to those observed between mouse proximal and distal prostatic ducts (Fig. 2E). Similarly, the expression of TGFB3 in stromal cells of TZ is also significantly higher than that in stromal cells in PZ (Figs. 7D–E). These results support that the differential expression of these genes in distinct anatomic locations is a conserved mechanism for tissue morphogenesis among different species.

Fig. 7: Wnt signaling is higher in stromal cells of human transition zone versus peripheral zone prostate. See also Figure S7.

(A) qRT-PCR analysis of AXIN2, WNT5A, TGFB1, TGFB2, and TGFB3 in human TZ and PZ stromal cells obtained by laser capture microdissection. Data obtained from 8 pairs of formalin-fixed paraffin-embedded normal human TZ and PZ tissues. Statistical analysis performed by paired t-test. TZ: transition zone; PZ: peripheral zone. (B) RNA-in-situ analysis of AXIN2 in human TZ and PZ prostates. Bars = 20 μm. (C) Dot graph shows quantification of AXIN2 expression in human TZ and PZ. Data represent means ± s.d. from 9 independent human patient samples. Each dot represents result from one image. Staining areas of AXIN2 in images of each pair of TZ and PZ are normalized by the average staining area in PZ of the same patient. (D) RNA-in-situ analysis of TGFB3 in human TZ and PZ prostate. Bars = 20 μm. (E) Dot graph shows quantification of TGFB3 expression in human TZ and PZ. Data represent means ± s.d. from 9 independent human patient samples. Each dot represents result from one image. Staining areas of TGFβ3 in images of each pair of TZ and PZ are normalized by the average staining area in PZ of the same patient.

Discussion

Heterogeneity of tissue resident stromal cells has been widely appreciated. Our study reveals that the stromal cells in different anatomic prostatic regions express different levels of Wnt ligands and exhibit distinct levels of Wnt/β-Catenin activity, which influence prostate epithelial stem/progenitor cell activity, directly and indirectly (graphic abstract). It should be noted that these two mechanisms may not be totally independent since TGFβ can regulate expression of both canonical and noncanonical Wnt ligands (Placencio et al., 2008; Roarty and Serra, 2007). The two mechanisms are also not exclusive as Wnt2 was shown to regulate prostate growth probably by downregulating FGF10 in stroma (Madueke et al., 2018). Our study in the prostate provides another example of the phenotypic and functional heterogeneity of the tissue resident stromal cells and raises more questions. For example, how do the Wnt/β-Catenin active stromal cells turn over and do they possess mesenchymal stem cell potential? Future studies using the single cell analysis and lineage tracing approach should address their features in more details.

Molecular mechanisms that regulate expression of Wnt ligands and activation of Wnt signaling

We showed that Wnt ligands are predominantly expressed in the stromal cells in the mouse prostate. In addition, the expression levels of the Wnt ligands such as Wnt2 and Wnt5a are higher in the proximal stromal cells. The molecular mechanisms underlying this differential expression pattern remains unclear. Previous studies have shown that Wnt ligands can be regulated by various mechanisms. Firstly, growth factors and transcription factors such as Fgf, Hgf, Hox5, and Gli proteins can regulate expression of Wnt ligands (Hrycaj et al., 2015; Huguet et al., 1995; Minor et al., 2013; Mullor et al., 2001). Regulation of Wnt ligands by Shh is particularly relevant to the distinct expression pattern of Wnt ligands in the prostate. Mouse prostate stromal cells possess active Gli activity that is induced by the Shh ligands secreted by the basal cells (Peng et al., 2013). Interestingly, prostate basal cells are enriched in the proximal prostatic ducts. This may lead to an increased local Shh concentration, thereby an elevated Gli activity and Wnt ligand expression in the proximal prostatic stromal cells. Consistent with this theory, we found that when stromal cells isolated from the proximal prostatic ducts were cultured in vitro, the expression levels of Wnt ligands and Axin2 were reduced dramatically (Supplementary Fig. 7C). This observation implies that the interaction between the basal cells and stromal cells regulates stromal expression of Wnt ligands, although it is also possible that the Wnt-expressing stromal cells were selected against during the in vitro culture. Secondly, Wnt ligands are also regulated by hormones such as the testosterone (Placencio et al., 2008; Tanaka et al., 2016). It has been well accepted that testosterone level within the prostate tissues is heterogeneous. Prostate stromal cells express the androgen receptor. Therefore, it is not impossible that the androgen differentially regulates expression of Wnt ligands in the stromal cells at distinct anatomic regions. Finally, cell shape and confluence are also shown to regulate Wnt ligand expression (Huguet et al., 1995). We noticed that the morphologies and densities of stromal cells in human TZ and PZ are different (X.W. and L.X. unpublished observation). This may reflect the differences in extracellular matrix stiffness and composition between the two zones, which also actively regulate Wnt ligand expression.

We also discovered that Wnt/β-Catenin signaling is differentially activated in both prostate epithelial cells and stromal cells at different anatomic locations. Although the proximal epithelial cells reside in a microenvironment that is rich in various Wnt ligands, they possess a lower canonical Wnt activity because of the antagonistic effect of Wnt5a as well as the relative lower expression of canonical Wnt receptors. Interestingly, a similar spatial regulation of the Wnt activity was also reported in the mammary gland epithelial ducts (Roarty et al., 2015), suggesting that it represents a conserved mechanism for tissue patterning. On the other hand, Wnt/β-Catenin signaling is preferentially activated in the proximal stromal cells because they express both more Wnt ligands and receptors. In addition, R-spondins, the Wnt signaling activators, are also expressed at a higher level in proximal prostatic stromal cells, although the difference is not statistically significant. These observations highlight that the status of the Wnt activity reflects combined input of various signaling components.

Benign prostatic hyperplasia (BPH) is a heterogeneous disease that results from nonmalignant proliferation of both the prostate epithelial and stromal compartment (Roehrborn, 2008). Pathological regeneration of stem cells in prostate epithelia and mesenchyme has been proposed as a cause for the tissue enlargement (Isaacs, 2008). Interestingly, BPH only occurs in the transition zone prostate. Our study shows that prostatic stromal Wnt signaling suppresses the epithelial proliferation directly and indirectly. Based on these facts, it is tempting to hypothesize that the stromal Wnt signaling is attenuated during BPH initiation and progression. Future studies using BPH specimens will test this hypothesis.

Wnt signaling in other cell lineages in tissue microenvironment

Besides the epithelial and stromal cells, other cell lineages such as the immune, endothelial and nerve cells are also capable of responding to Wnt ligands, for example, Wnt signaling regulates T cell survival and lineage fate decisions (Gattinoni et al., 2010), dendritic cell-mediated immune tolerance (Swafford and Manicassamy, 2015), neurogenesis (Wang et al., 2012), and angiogenesis (Stenman et al., 2008). This implies that the differential expression pattern of Wnt ligands in distinct anatomic regions may result in different tissue microenvironment. Because of the complex and sometime contradictory roles of the Wnt signaling in different cell lineages, whether the overall tissue microenvironment is proinflammatory, immune suppressive, or tumor permissive will depend on the lineage composition in specific regions. It is tempting to hypothesize that such spatially restricted distinct tissue microenvironment determines region-specific prevalence or clinical behaviors of certain diseases, for example, why the frequency and aggressiveness of ductal versus alveolar adenocarcinoma are different in the same organ? This hypothesis is particularly interesting in the context of the prostate because the two major prostate-related diseases, prostate cancer and benign prostatic hyperplasia, occur preferentially in the peripheral and transition zones, respectively. It will be interestingly to investigate whether the different gradient of stromal Wnt ligands in the two zones plays a role in the zonal specific prevalence of the two diseases.

STAR METHODS

Contact for Reagent and Resource Sharing

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Li Xin (xin18@uw.edu)

Experimental Model and Subject Details

Mice

All animals used in this study received humane care in compliance with the principles stated in the Guide for the Care and Use of Laboratory Animals, NIH Publication, 1996 edition, and the protocol was approved by the Institutional Animal Care Committee of Baylor College of Medicine and University of Washington. The C57BL/6 and SCID/Beige mice were purchased from Charles River (Wilmington, MA). Col1a2-CreER^T2, and B6.Cg-Gt(ROSA)26Sor^{tm3(CAG-EYFP)Hze}/J mice were purchased from the Jackson Laboratory (Bar Harbor, ME). TCF/LEF:H₂B-GFP reporter line, Ctnnb1^loxp/loxp, and Ctnnb1^lox(ex3) were described previously (Ferrer-Vaquer et al., 2010; Harada et al., 1999; Huelsken et al., 2001) and were obtained from Dr. Hoang Nugyen at the Baylor College of Medicine. Mice were genotyped by polymerase chain reaction using mouse genomic DNA from tail biopsy specimens. The sequences of genotyping primers and the expected band sizes for PCR are listed in Supplementary Table 2. PCR products were separated electophoretically on 1% agarose gels and visualized via ethidium bromide under UV light.

Human specimens

Human PZ and TZ prostate specimens used in this study were obtained from patients undergoing open simple prostatectomy for prostate cancer at the Michael E. DeBakey Veteran Medical Center with Institutional Review Board approval. Peripheral zone tissues were obtained from patients with relatively smaller localized tumors (prostate weights between 25-35g), and tissues were collected distant from major tumor nodules. Fresh specimens were fixed in 10% buffered formalin for 24 hours and processed for RNA-In-Situ analyses. Specimens were confirmed to be cancer free by histological and immunostaining of CK5 and P63.

Method Details

Tamoxifen and BrdU treatment

Tamoxifen (Sigma-Aldrich, St. Louis, MO) was dissolved in corn oil and administrated i.p. into experimental mice at the specified age (5 mg/40g/day for four consecutive days unless otherwise specified). BrdU (Sigma-Aldrich, St. Louis, MO) (80mg/kg/day) was administrated 3 days before mice were sacrificed unless otherwise specified.

Cell culture

8~12-week-old C57BL/6 mouse prostate tissues were digested and dissociated into single cells as described previously (Zhang et al., 2016). Dissociated single cells were cultured in Biocoat™ Collage I-coated plates (Corning, Corning, NY) in Bfs medium (5% Nu-Serum, 5% FBS, 1×Insulin/Selenium, 1×L-Glutamine, l×Penicillin/Streptomycin, and 1×10⁻¹⁰M DHT in DMEM medium) at 37°C with 5% CO₂. When the cell confluency reached around 90%, cells were trypsinized into single cells with 0.25% Trypsin-EDTA (Invitrogen, Carlsbad, CA). Cells were replated in Biocoat™ Collage I-coated plates (Corning, Corning, NY) for 30 min at 37°C/5% CO₂. Unattached cells were discarded, and remaining cells were cultured in Bfs at 37°C/5% CO₂ till 80-90% confluency. All the experiments in this study used fresh primary stromal cells within 3 weeks after single cell dissociation from prostates.

Flow cytometry and cell sorting

Dissociated single mouse prostate cells were incubated with florescence conjugated antibodies at 4°C for 30 minutes. Information for antibodies for FACS analysis and sorting is listed in Supplementary Table 3. FACS analyses and sorting were performed by using the BD LSR II, BD LSR Fortessa, Aria I (BD Biosciences, San Jose, CA).

RNA isolation, quantitative RT–PCR, and RT² Profiler™ PCR array

Total RNA was extracted using Nucleospin RNA extraction Kit (Macherey-Nagel, Bethlehem, PA). RNA was reverse transcribed to cDNA using iScript™ Reverse Transcriptase Kit (BioRad, Hercules, CA). QRT-PCR was performed using iTaq™ Universal SYBR Green Supermix (BioRad, Hercules, CA) and detected on a StepOne plus Real-Time PCR system (Applied Biosystems, Foster City, CA). Primers for target genes were listed in Supplementary Table 4. Analysis of the Wnt signaling components was performed on RT² Profiler™ PCR Array Mouse Wnt signaling plates following the manufacturer’s instruction (Qiagen, Valencia, CA).

qRT-PCR from Laser captured FFPE tissues

Slides of 10 μm were sectioned from FFPE blocks using a Leica RM 2253 microtome, mounted onto Arcturus PEN Membrane Frame Slides (Thermo Scientific, Waltham, MA), and dried overnight at room temperature. Sections were stained with Cresyl Violet (Acros Organic, New Jersey, NJ) and left at room temperature for one hour to dry. Adjacent sections of 5 αm were cut and stained with hematoxylin and eosin. Histology review and slide annotation was performed by pathologist. Areas of cancer, PIN, and extensive inflammation were marked if present and were excluded from the subsequent experiments. Areas of stroma within 120 microns of a benign gland or 240 microns between benign glands were captured using the Arcturus XT (Thermo Scientific, Waltham, MA) with CapSure Macro LCM Caps (Thermo Scientific, LCM0211) and RNA was extracted using the truXTRAC FFPE RNA microTUBE Kit (Covaris, Woburn, MA) according to manufacturer’s recommendations. RNA quality (DV200) and quantity were assessed using the TapeStation 4200 (Agilent Technologies, Santa Clara, CA) with High Sensitivity RNA Screentape (Agilent Technologies, Santa Clara, CA). RNA was reverse transcribed to cDNA using iScript™ Reverse Transcriptase kit (BioRad, Hercules, CA). cDNA was preamplified using SsoAdvanced™ PreAmp Supermix (BioRad, Hercules, CA). QRT-PCR was performed using iTaq™ Universal SYBR Green Supermix (BioRad, Hercules, CA) and detected on a Quantstudio Real-Time PCR system (Applied Biosystems, Foster City, CA).

Lentivirus preparation

CDNAs for S37A β-Catenin, dominant negative β-Catenin, and dominant negative TgfβRII were gifts from Dr. Roger Lo at UCLA, Dr. Pierre Mccrea at MD Anderson Cancer Center, and Dr. Lalage Wakefield at NIH, respectively. cDNA was cloned into the FU-CGW or FU-CRW lentiviral vector (Xin et al., 2003) at the EcoRI site. Lentivirus preparation, titering, and infection of UGSM cells or dissociated prostate cells were performed as described previously (Xin et al., 2003).

Prostate sphere assay

The prostate sphere assay was performed as described previously (Xin et al., 2007). Briefly, 1×10⁴ dissociated prostate cells were mixed in 1:1 Matrigel/PrEGM (Matrigel (BD Biosciences, San Jose, CA)/PrEGM (Lonza, Walkersville, MD)), plated in 12-well plates, and cultured in PrEGM medium. Prostate spheres were defined as spheroids with a diameter > 30 μm after a 6-day culture. When prostate basal cells were cocultured with stromal cells or urogenital sinus mesenchymal cells, the ratio of epithelial versus stromal cells is 1:2.

Prostate organoid culture

The organoid culture was performed as described previously (Karthaus et al., 2014). Briefly, dissociated prostate cells from 8-12 week-old C57B1/6 mice were cultured in DMEM/F12 supplemented with B27 (Life technologies, Grand Island, NY), 10 mM HEPES, Glutamax (Life technologies, Grand Island, NY), Penicillin/Streptomycin, and the following growth factors: EGF 50 ng/ml (Peprotech, Rocky Hill, NJ), 500 ng/ml recombinant R-spondin1 (Peprotech, Rocky Hill, NJ), 100 ng/ml recombinant Noggin (Peprotech, Rocky Hill, NJ), 200 μM TGF-β/Alk inhibitor A83-01 (Tocris, Ellisville, MO), and 10 μM Y-27632 (Tocris, Ellisville, MO). Dihydrotestosterone (Sigma, St. Louis, MO) was added at 1 nM final concentration. Cells were resuspended in growth factor reduced matrigel (Corning, Corning, NY) and plated in 96-well plates.

Prostate regeneration assay

8-12 week old mouse prostate tissues were dissociated into single cells by the procedure described previously (Kwon et al., 2015; Valdez et al., 2012). UGSM cell preparation and prostate regeneration assays were performed as described previously (Xin et al., 2003). Briefly, 1×10⁵ wild type or lentivirus-infected dissociated prostate cells were mixed with 1×10⁵ wild type or lentivirus-infected murine UGSM cells in type I collagen extracted from rat tails. Cell mixtures were grafted under the renal capsules of immunodeficient male SCID/Bg mice and incubated for 8 weeks.

RNA-seq

NucleoSpin RNA XS Kit (Macherey-Nagel, Bethlehem, PA) was used to purify RNAs from FACS-isolated proximal and distal mouse prostate basal and stromal cells. Reverse transcriptions were performed using SMART Seq™ v4 Ultra™ Low Input RNA Kit for Sequencing (Clontech Laboratories, Mountain View, CA). CDNA libraries were prepared using Nextera XT DNA Library Preparation Kit (Illumina, San Diego, CA).

RNA was extracted from the β-Catenin S37A-expressing mouse prostate stromal cells and RFP-expressing control stromal cells using MasterPure™ RNA Purification Kit (Illumina, San Diego, CA). TruSeq Stranded mRNA Sample Preparation Kit (Illumina, San Diego, CA) was used to prepare cDNA libraries, which were sequenced using HiSeq 2500 sequencer. Sequenced reads in FASTQ files were mapped to mm10 whole genome using Tophat2, and Fragments Per Kilobase of transcript per Million mapped reads (FPKM) were calculated using Cufflinks. Genes found differentially expressed (p<0.05 by t-test, and minimum fold change of 1.4 or 1.2) were evaluated for enrichment of Gene Ontology (GO) gene classes, using SigTerms software (Creighton et al., 2008). Data have been deposited at GEO.

ChIP assay

Briefly, primary mouse prostate stromal cells were crosslinked in cell culture media containing 1% methanol free formaldehyde (Polysciences, Warrington, PA) for 10 minutes at room temperature. Cross-linking was terminated by adding 125 mM Glycine for 5 minutes at room temperature. Cells were harvested in PBS containing protease inhibitors and PMSF. Subsequently, cells were resuspended in sonication buffer and sonicated. Input samples were taken following sonication. The remaining sonication product was divided into 2 halves incubated with 4μg antibodies against β-Catenin (610154, BD Biosciences, San Jose, CA) or normal rabbit IgG (sc-2027, Santa Cruz Biotech, Santa Cruz, CA) in ChIP Buffer 1 containing protease inhibitors and magnetic G beads overnight at 4°C in siliconized tubes. Beads were collected and washed with ChIP Buffer 1 and ChIP Buffer 2. Beads were then resuspended in elution buffer for 15 minutes at room temperature. Formaldehyde crosslinks were reversed by incubating samples in reverse cross-linking buffer at 65°C for 2.5 hours. Remaining proteins were digested with Proteinase K (50μg/μl) at 37°C for 1 hour. DNA was then isolated using the Purelink Quick Gel Extraction Kit (Invitrogen, Carlsbad, CA). qRT-PCR was performed to determine enrichment using the primers listed in Supplemental Table 5. Ct values from α-IgG and α-β-Catenin were normalized against Ct values generated from the input samples. The resulting value was then normalized by dilution and concentration to determine the value relative to the input.

Luciferase reporter assay

A 2 and 4 Kb genomic sequence upstream of the translational start codon of the mouse Tgfβ2 and 3, respectively, containing multiple TCF/LEFs binding sites confirmed by the ChIP analysis was PCR amplified from two BAC clones from the BACPAC Resources Center (Children’s Hospital Oakland Research Institute) containing the corresponding genomic region using LA Taq (Takara Bio Inc., Otsu, Shiga, Japan). The primers used are listed in Supplementary Table 6. The amplicon was cloned into the pGL3 luciferase vector (Promega, Madison, WI) via the KpnI and BglII restriction sites upstream of luciferase generating the pGL3-Tgfβ2-luc and pGL3-Tgfβ3-luc reporters. Mutations in the TCF/LEFs binding sites were performed by site-directed-mutagenesis using the Q5® Site-Directed Mutagenesis kit (New England Lab, Woburn, MA).

Primary mouse prostate stromal cells were infected with either FU-CRW or FU-β-catenin S37A-CRW. Two days later, cells were seeded in 6 well plates and co-transfected with 40 ng of pRL-CMV Renilla and 2ug of pGL3-Tgfβ2-luc, pGL3-Tgfβ3-luc, or corresponding mutant constructs, respectively, using lipofectamine 3000 according to the manufacturer’s instructions. Two days after transfection, the Bfs cell culture medium was replaced with 1:10 Opti-MED diluted Bfs medium. 24 hours later, luciferase activity was measured using the dual-Luciferase reporter assay system (Promega, Madison, WI). Firefly luciferase activity was normalized to CMV-enilla luciferase activity. Data were presented relative to CMV-Renilla readings and shown as mean ± s.d. Experiments were performed in triplicates.

RNA-Scope

TZ and PZ tissues collected from the same donors were collected at the same time and fixed by 10% neutral buffered formalin for 16~32 hours at room temperature. Samples were embedded in paraffin blocks and cut into 5um sections for staining. Freshly cut slides were air dried overnight at room temperature, then baked for 1 hour at 60°C. The RNA-Scope in situ hybridization was performed by using RNA-Scope 2.5 HD Detection Reagent Red Kit (Advanced Cell Diagnostics, Newark, CA) following the manufacturer’s standard protocol. 20–60 images were taken for each sample to cover all areas of the stained specimens. For analysis, we focused on the stromal cells adjacent to the epithelial compartment. We used the Image-Pro Plus version 6.3 by Media Cybernetics to include all inter-glandular areas that were within a range of 150 um away from the basement membrane between the epithelial and stromal compartments. Nuclei numbers and areas with staining of AXIN2 and TGFβ3 in the defined areas were determined by the count feature in the software. Total AXIN2 or TGFβ3 staining areas within the stromal cells were normalized by nucleus number of stromal cells on each image. Images of each pair of TZ and PZ samples were further normalized by the average value of the PZ staining area of the same patient so that data from different patients can be pooled for analysis shown in Fig. 7.

Western blots

Cells were lysed in RIPA buffer (20 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1 mM Na₂EDTA, 1 mM EGTA, 1% NP-40, 1% sodium deoxycholate, 2.5 mM sodium pyrophosphate, 1 mM β-glycerophosphate, 1 mM Na₃VO₄) with protease inhibitors and phosphatase inhibitors (Roche Applied Science, Indianapolis, IN). Protein concentrations were determined by a Bradford Assay kit (BioRad, Hercules, CA). Protein was separated by 10% SDS/PAGE and transferred onto a nitrocellulose membrane (Amersham Biosciences, Arlington Heights, IL). The membrane was blocked in 5% skim milk, and subsequently incubated with primary antibodies listed in Supplementary Table 7 at 4°C overnight followed by incubation with peroxidase-conjugated goat anti-mouse IgG or goat anti-rabbit IgG (Jackson ImmunoResearch, Inc., West Grove, PA), and developed with Pierce ECL reagent (Thermal Scientific, Rockford, IL).

Histology and Immunostaining

Prostate tissues were fixed by 10% buffered formalin and paraffin embedded. HE staining and immunofluorescence staining were performed with 5 μm sections. For hematoxylin and eosin staining and immunostaining, sections were processed as described previously (Choi et al., 2012). For immunostaining, sections were processed as described previously (Choi et al., 2012) and incubated with primary antibody in 3% of normal goat serum (Vector Laboratories, Burlingame, CA) overnight. Information for primary antibodies is listed in Supplementary Table 6. Slides then were incubated with secondary antibodies (diluted 1:250 in PBST) labeled with Alexa Fluor 488 and 594 (Invitrogen/Molecular Probes, Eugene, OR). Sections were counterstained with 4,6-diamidino-2-phenylindole (DAPI) (Sigma-Aldrich, St. Louis, MO). Immunofluorescence staining was imaged using an Olympus BX60 fluorescence microscope (Olympus Optical Co Ltd, Tokyo, Japan) or a Leica EL6000 confocal microscope (Leica Microsystems, Wetzlar, Germany). Images of IHC were analyzed by Image-Pro Plus version 6.3 by Media Cybernetics. Cell number was determined by using the count feature in the software which asks for the user to indicate the color that would be used to indicate a positive cell (For example: blue would be indicated to count nuclei and thus indicate total numbers of cells). Borders were created so that only stromal cells would be analyzed. Sections were counterstained with hematoxylin and mounted with Permount.

Statistical analyses

All experiments were performed using 3-10 mice in independent experiments. Data are presented as mean ± s.d. Student’s t test and one-way ANOVA with multiple comparisons were used to determine significance in two-group and multiple-group experiments, respectively. For all statistical tests, the two-tail p<0.05 level of confidence was accepted for statistical significance. More details are found in the figure legends.

Data availability

The accession numbers for the RNA-seq data of the stromal cells at proximal and distal prostatic ducts, basal cells at proximal and distal prostatic ducts, stromal cells expressing S37A β-Catenin and RFP in this paper are GEO: GSE115631, 72318, 115467, respectively. Detailed descriptions of data analysis and the software used can be found in Method Details.

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Rat anti-Mouse CD31-eFluor 450 (Clone 390)	eBioscience	Cat# 48-0311-82; RRID:AB_10598807
Rat anti-Mouse CD45-eFluor 450 (Clone 30-F11)	eBioscience	Cat# 48-0451-82; RRID:AB_1518806
Rat anti-Mouse Ter119-eFluor 450 (Clone TER-119)	eBioscience	Cat# 48-5921-82; RRID:AB_1518808
Rat anti-Mouse CD49f-APC (Clone eBioGoH3)	eBioscience	Cat# 17-0495-82; RRID:AB_2016694
Rat anti-Mouse Sca1-PE (Clone D7)	eBioscience	Cat# 12-5981-83; RRID:AB_466087
Rat anti-Mouse CD24-PECy7 (Clone M1/69)	BD Biosciences	Cat# 560536; RRID:AB_1727452
Mouse anti-BrdU-FITC (Clone BU20A)	eBioscience	Cat# 11-5071-42; RRID:AB_11042627
Rabbit anti-Mouse K5 (Clone Poly19055)	Covance	Cat# PRB-160P; RRID:AB_2565050
Mouse anti-Mouse K8 (Clone 1E8)	Covance	Cat# MMS-162P; RRID:AB_2565043
Mouse anti-Mouse K14 (Clone LL002)	Santa Cruz Biotechnology	Cat# sc-58724; RRID:AB_784170
Mouse anti-Mouse P63 (Clone 4A4)	Abcam	Cat# ab735; RRID:AB_305870
Chicken anti-GFP	Abcam	Cat# ab13970; RRID:AB_300798
Rat anti-BrdU (Clone BU1/75)	Abcam	Cat# ab6326; RRID:AB_305426
Rabbit anti-Mouse Ki67	Leica Biosystems	Cat# NCL-Ki67-P; RRID:AB_442102
Rabbit anti-Mouse AR	Santa Cruz Biotechnology	Cat# sc-816; RRID:AB_1563391
Rabbit anti-Mouse Cleaved Caspase 3	Cell Signaling Technology	Cat# 9661S; RRID:AB_2341188
Rat anti-Mouse Sca-1 (Clone D7)	BD Pharmingen	Cat# 557403; RRID:AB_396686
Mouse anti-Mouse Smooth Muscle Actin (Clone 1A4)	Sigma-Aldrich	Cat# 202M; RRID:AB_1157937
Rabbit anti-Mouse Vimentin	Cell Signaling Technology	Cat# 5741S; RRID:AB_10695459
Mouse anti-Mouse E-cadherin (Clone 36/E-Cadherin)	BD Transduction Labs	Cat# 610181; RRID:AB_397580
Rabbit anti-Mouse Nkx3.1	Athena Enzyme Systems	Cat# 0315
Mouse anti-Mouse Ror2	Developmental studies hybridoma bank	Cat# Nt 2535-2835; RRID:AB_10804796
Rabbit anti-Mouse CD31	Abcam	Cat# ab28364; RRID:AB_726362
Rat anti-Mouse CD45 (Clone 30-F11)	BD Pharmingen	Cat# 550539; RRID:AB_2174426
Rabbit anti-Mouse pSmad3 (pS423/pS425)	Rockland	Cat# 600-401-919; RRID:AB_2192878
Rabbit anti-Mouse Smad2/3	Cell Signaling Technology	Cat# 5678; RRID:AB_10693547
Mouse anti-Mouse β-actin (Clone AC-74)	Sigma-Aldrich	Cat# A2228; RRID:AB_476697
Goat α-Mouse IgG(H+L) Alexa Fluor 488	Invitrogen	Cat# A11017; RRID:AB_2534084
Goat α-Mouse IgG(H+L) Alexa Fluor 594	Invitrogen	Cat# A11020; RRID:AB_2534087
Goat α-Rabbit IgG(H+L) Alexa Fluor 488	Invitrogen	Cat# A11034; RRID:AB_2576217
Goat α-Rabbit IgG(H+L) Alexa Fluor 594	Invitrogen	Cat# A11037; RRID:AB_2534095
Goat α-Rat IgG(H+L) Alexa Fluor 488	Invitrogen	Cat# A11006; RRID:AB_2534074
Goat α-Rat IgG(H+L) Alexa Fluor 594	Invitrogen	Cat# A11007; RRID:AB_10561522
Horse α-Mouse IgG(H+L) HRP	Vector Lab.	Cat# PI-2000; RRID:AB_2336177
Goat α-Rabbit IgG(H+L) HRP	Vector Lab.	Cat# PI-1000; RRID:AB_2336198
Bacterial and Virus Strains
Biological Samples
Human PZ and TZ prostate specimens	Michael E. DeBakey Veteran Medical Center, U.S.	N/A
Chemicals, Peptides, and Recombinant Proteins
Tamoxifen	Sigma-Aldrich	Cat# T5648
BrdU	Sigma-Aldrich	Cat# B5002-5G
Wnt3a	R&D	Cat#1324-WN-002
Wnt5a	R&D	Cat# 645-WN-010
Tgfβ1	R&D	Cat# 240-B-002
Tgfβ2	R&D	Cat# 302-B2-002
Tgfβ3	R&D	Cat# 8420-B3-005
Critical Commercial Assays
NucleoSpin RNA Kit	Macherey-Nagel	Cat# 740955
NucleoSpin RNA XS Kit	Macherey-Nagel	Cat# 740902
SsoAdvanced PreAmp Supermix	Bio-rad	Cat# 172-5160
WNT Signaling Pathway RT2 Profiler PCR Array	QIAGEN	Cat# 330231 PAMM-043Z
RT2 First Strand Kit	QIAGEN	Cat# 330401
RT2 PreAMP Wnt Pathway Primer Mix	QIAGEN	Cat# PBM-043Z
RT2 PreAMP cDNA Synthesis Kit	QIAGEN	Cat# 330451
RT2 SYBR Green ROX qPCR Mastermix	QIAGEN	Cat# 330522
BD Pharmingen BrdU Flow Kit	BD	Cat# 559619
SMART-Seq v4 Ultra Low Input RNA Kit	Clontech	Cat# 634888
Nextera XT DNA Library Preparation Kit	Illumina	Cat# FC-131-1024
TruSeq Stranded mRNA Library Prep Kit	Illumina	Cat# 20020594
Q5 Site-Directed Mutagenesis Kit	New England Lab	Cat# E0554S
RNA-Scope 2.5 HD Detection Reagent Red Kit	Advanced Cell Diagnostics	Cat# 322360
truXTRAC FFPE RNA microTUBE Kit	Covaris	Cat# 520161
MasterPure RNA Purification Kit	Illumina	Cat# MCR85102
Deposited Data
RNA-seq data of the stromal cells at proximal and distal prostatic ducts	This paper	GEO: GSE115631
RNA-seq data of the basal cells at proximal and distal prostatic ducts	This paper	GEO: GSE72318
RNA-seq data of the prostate stromal cells expressing S37A β-Catenin and RFP	This paper	GEO: GSE115467
Experimental Models: Cell Lines
Experimental Models: Organisms/Strains
Mouse: C57BL/6	Charles River	Strain code: 027
Mouse: SCID/Beige	Charles River	Strain code: 250
Mouse: Col1a2-CreERT2	Jackson Laboratory	JAX stock #029567
Mouse: B6.Cg-Gt(ROSA)26Sortm3(CAG-EYFP)Hze/J	Jackson Laboratory	JAX stock #007903
Mouse: TCF/Lef:H2B-GFP	Jackson Laboratory	JAX stock #013752
Mouse: Ctnnb1loxp/loxp	Baylor College of Medicine, U.S.	Kind Gift from Hoang Nugyen
Mouse: Ctnnb1lox(ex3)	Baylor College of Medicine, U.S.	Kind Gift from Hoang Nugyen
Oligonucleotides
Primers for genotyping of mouse lines, see Table 1	This paper	N/A
qPCR Primers for human genes, see Table 2	This paper	N/A
qPCR Primers for mouse genes, see Table 3	This paper	N/A
qPCR Primers for ChIP assay, see Table 4	This paper	N/A
Primers for Tgfβ 2/3 Luciferase Reporter Cloning, see Table 5	This paper	N/A
Recombinant DNA
Plasmid: FU-CGW	Xin et al., 2003	N/A
Plasmid: FU-CRW	Xin et al., 2003	N/A
cDNA: S37A β-Catenin	UCLA, U.S.	Kind Gift from Roger Lo
cDNA: dominant negative β-Catenin	MD Anderson Cancer Center, U.S.	Kind Gift from Pierre Mccrea
cDNA: dominant negative TgfβRII	NIH, U.S.	Kind Gift from Lalage Wakefield
Software and Algorithms
Leica Application Suite X	Leica Microsystems GmbH	RRID:SCR_013673 URL: https://www.leica-microsystems.com/products/microscope-software/details/product/leica-las-x-ls/
FlowJo	Tree Star	RRID:SCR_008520 URL:https://www.flowjo.com/solutions/flowjo
GraphPad Prism 7	GraphPad Software	RRID:SCR_002798
Image-Pro Plus version 6.3	Media Cybernetics	RRID:SCR_007369
Fiji	ImageJ	RRID: SCR_002285
Adobe Photoshop CC	Adobe Systems	RRID: SCR_014199
Adobe Illustrator CC	Adobe Systems	RRID: SCR_010279
Other
FISH probe: human AXIN2	Advanced Cell Diagnostics	N/A
FISH probe: human TGFβ3	Advanced Cell Diagnostics	N/A

Highlights.

• stromal cells in mouse proximal prostatic ducts highly express multiple Wnt ligands

• proximal duct stromal cells exhibit high Wnt/b-catenin activity

• prostate stromal Wnt/b-catenin activity inhibits epithelial stem cell activity via TGFb

• Spatially-restricted stromal Wnt/b-catenin signaling is conserved across species