Dickkopf-related protein 3 promotes pathogenic stromal remodeling in benign prostatic hyperplasia and prostate cancer

BACKGROUND Compartment-specific epithelial and stromal expression of the secreted glycoprotein Dickkopf-related protein (Dkk)-3 is altered in age-related proliferative disorders of the human prostate. This study aimed to determine the effect of Dkk-3 on prostate stromal remodeling that is stromal proliferation, fibroblast-to-myofibroblast differentiation and expression of angiogenic factors in vitro. METHODS Lentiviral-delivered overexpression and shRNA-mediated knockdown of DKK3 were applied to primary human prostatic stromal cells (PrSCs). Cellular proliferation was analyzed by BrdU incorporation ELISA. Expression of Dkk-3, apoptosis-related genes, cyclin-dependent kinase inhibitors and angiogenic factors were analyzed by qPCR, Western blot analysis or ELISA. Fibroblast-to-myofibroblast differentiation was monitored by smooth muscle cell actin and insulin-like growth factor binding protein 3 mRNA and protein levels. The relevance of Wnt/β-catenin and PI3K/AKT signaling pathways was assessed by cytoplasmic/nuclear β-catenin levels and phosphorylation of AKT. RESULTS Knockdown of DKK3 significantly attenuated PrSC proliferation as well as fibroblast-to-myofibroblast differentiation and increased the expression of the vessel stabilizing factor angiopoietin-1. DKK3 knockdown did not affect subcellular localization or levels of β-catenin but attenuated AKT phosphorylation in PrSCs. Consistently the PI3K/AKT inhibitor LY294002 mimicked the effects of DKK3 knockdown. CONCLUSIONS Dkk-3 promotes fibroblast proliferation and myofibroblast differentiation and regulates expression of angiopoietin-1 in prostatic stroma potentially via enhancing PI3K/AKT signaling. Thus, elevated Dkk-3 in the stroma of the diseased prostate presumably regulates stromal remodeling by enhancing proliferation and differentiation of stromal cells and contributing to the angiogenic switch observed in BPH and PCa. Therefore, Dkk-3 represents a potential therapeutic target for stromal remodeling in BPH and PCa. Prostate 73: 1441–1452, 2013. © 2013 The Authors. Prostate published by Willey-Blackwell. This is an open access article under the terms of the Creative Commons Attribution-Non-Commercial-NoDerivs Licence, which permits use and distribution in any medium, provided the original work is properly cited, the use is non-commercial and no modifications or adaptations are made.


INTRODUCTION
Expression patterns of the secreted glycoprotein Dickkopf-related protein 3 (Dkk-3) are altered in benign prostatic hyperplasia (BPH) and prostate cancer (PCa). In the normal prostate, Dkk-3 is predominantly expressed in the epithelial compartment, whereas in the diseased prostate, Dkk-3 is reduced in the epithelial but elevated in the stromal compartment, especially endothelial cells [1]. We recently demonstrated that in patients with PCa seminal plasma Dkk-3 levels are significantly elevated [2] despite the reduced expression of Dkk-3 in secretory epithelial cells [1,3,4] indicating that Dkk-3 derived from tumor neovasculature/stroma is able to diffuse through the diseased tissue into the prostatic fluid while in normal/BPH tissue an intact basement membrane prevents leakage from the stroma to the epithelial compartment and vice versa.
The functional significance of elevated Dkk-3 in the diseased prostatic stroma is unknown. Dkk-3 is the most divergent member of the human Dickkopf family of Wnt/b-catenin signaling antagonists [5,6]. In contrast to other family members, Dkk-3 does not interact with LDL-receptor-related protein 6 or Kremen proteins and thus is not considered a Wnt/b-catenin signaling antagonist [7,8]. Nevertheless, Dkk-3 has been reported to either antagonize [9,10] or enhance [11] Wnt/b-catenin signaling in a cell-specific manner.
Dkk-3 has been proposed to represent a novel tumor suppressor since gene expression is downregulated in various tumor cells [3,[12][13][14][15] and hypermethylation of its promoter correlates with cancer occurrence [16,17]. Several studies reported anti-proliferative or pro-apoptotic effects upon DKK3 overexpression [3,13,18,19], However, these effects appeared to be caused by endoplasmatic reticulum stress (unfolded protein response) [18,19], which is commonly induced by overexpression of highly-glycosylated secreted proteins, such as Dkk-3, and thus might not reflect the biological role of endogenous Dkk-3. Indeed, addition of exogenous recombinant Dkk-3 uniformly failed to reduce proliferation or induce apoptosis of malignant and nonmalignant cells [1,19,20]. Moreover, in the human pancreatic carcinoma cell line PANC-1 over-expression of DKK3 did not alter cellular proliferation, while knockdown of DKK3 resulted in significant reduction of cellular proliferation and concomitant induction of pancreatic epithelial cell differentiation markers, indicating that Dkk-3 is required to maintain a highly dedifferentiated and proliferative state in these cells [21].
BPH and PCa are both associated with changes in the stromal microenvironment (stromal remodeling) that actively promote disease development. In particular, the BPH and PCa-adjacent stroma are characterized by increased extracellular matrix deposition, capillary density, and differentiation of fibroblasts into myofibroblasts, the mitogenic secretome of which promotes proliferation, angiogenesis, and tumorigenesis [22][23][24][25]. TGFb1 is considered to be a key inducer of pathogenic stromal reorganization, and others and we have demonstrated that TGFb1 induces prostatic fibroblast-to-myofibroblast differentiation [26][27][28][29][30].
Enhanced angiogenesis is also a key feature of the remodeled stroma. The angiogenic switch is a ratelimiting step in tumor progression [31] that is associated with a shift in the ratio of the vessel stabilizing angiopoietin-1 (ANGPT1) to the destabilizing factor angiopoietin-2 (ANGPT2) in favor of ANGPT2. Consequently, the angiogenic switch renders the tumor vasculature amenable to vessel sprouting [32]. Besides the prostate elevated Dkk-3 expression has also been shown in vessels from other tumors for example in colorectal cancer, glioma, non-Hodgkin lymphoma, melanoma, and pancreatic adenocarcinoma whereas vessels from normal tissue express low/not detectable Dkk-3 levels [33][34][35]. Dkk-3 has been shown to support tube formation in primary endothelial colonyforming cells and DKK3 overexpression reduced ANGPT1 expression in a murine B16F10 melanoma model [34]. Moreover, Dkk-3 and ANGPT2 were inversely regulated in human umbilical vein endothelial cells after knockdown of Axl [36], suggesting a role of Dkk-3 in tumor angiogenesis.
This study aimed to investigate the functional significance of elevated stromal Dkk-3 in BPH and PCa by lentiviral-delivered overexpression and shRNA-mediated knockdown of DKK3 in primary prostatic stromal cells and analysis of the downstream effects on proliferation, TGFb1-induced fibroblast-to-myofibroblast differentiation and expression of angiogenic factors.

Cell Culture and Fibroblast-to-Myof|broblast Differentiation
Human primary prostatic stromal cell (PrSC) and prostatic basal epithelial cell (PrEC) cultures were established as described previously [1]. PrSC were cultured in stromal cell growth medium (Quantum 333, PAA Laboratories), PrEC on collagen I-coated plates in prostate epithelial cell growth medium (PrEGM, Clonetics). All experiments were performed with primary cells from at least three independent donors. Fibroblast-to-myofibroblast differentiation was induced by 1 ng/ml TGFb1 (R&D Systems) in RPMI 1640 (PAA Laboratories) containing 1% charcoal treated fetal calf serum (HyClone) and 1% penicillin/ streptomycin (PAA Laboratories) as described [28]. Control cells were treated with 1 ng/ml human basic fibroblast growth factor (bFGF; Sigma-Aldrich) as control to maintain the fibroblast phenotype.

Knockdown and Overexpression of DKK3 by Lentiviral Particles
Production of lentiviral particles was carried out according to the manufacturer's protocol (Addgene) as described previously [21] using the lentiviral pLKO.1-TRC short hairpin system (Addgene) for knockdown and full-length cDNA of DKK3 subcloned into the pLenti6 vector (Invitrogen) for overexpression, respectively. The scramble shRNA vector (Addgene plasmid 1864) and the empty pLenti6 vector were used as controls. For viral transduction, cells were seeded in appropriate vessels and left to adhere overnight. Thereafter, medium was replenished and supplemented with virus-containing supernatant at MOI 4 (knockdown) and MOI 0.5 (overexpression), respectively.

Cell Proliferation Assay
Two thousand cells were seeded in triplicate into 96-well plates (Nunc) in 100 ml culture medium and left to adhere overnight. Thereafter, fresh medium was supplemented with lentivirus particles to transduce cells or the phosphatidylinositol 3-kinase (PI3K) inhibitor LY294002 (Calbiochem) at the indicated concentration. Proliferation was determined by relative quantification of DNA synthesis using a bromodeoxyuridine (BrdU) cell proliferation ELISA (Roche Applied Science) according to the manufacturer's instructions at indicated times post-transduction.
Quantitative Real-Time PCR mRNA extraction, cDNA synthesis and quantitative PCR (qPCR) were performed as described elsewhere [28]. Primer sequences are given in Table I. cDNA concentrations were normalized by the housekeeping gene hydroxymethylbilane synthase (HMBS).

Dkk-3 and Angiopoietin Quantif|cation in Cell
Culture Supernatants PrSCs were seeded at a density of 1 Â 10 5 per 6 cm dishes and left to adhere overnight. Subsequently, medium was replaced and cells were transduced with lentiviral particles. After 72 hr, medium was replaced with fresh medium containing bFGF, TGFb1 and/or LY294002 as indicated, and conditioned for 24 hr (for determination of Dkk-3) and 72 hr (for determination of angiopoietin-1 and angiopoietin-2), respectively. Secreted Dkk-3 was quantified by immunoenzymometric assay (IEMA) as previously described [37,38]. Secreted angiopoietin-1 and angiopoietin-2 levels were analyzed by the RayBio ® Human angiopoietin-1 ELISA Kit and RayBio ® Human angiopoietin-2 ELISA Kit (RayBiotech) according to manufacturer's instructions, respectively. In order to account for different cell proliferation angiopoietin levels were normalized using corresponding relative BrdU-incorporation ELISA values.

Western Blot Analysis
Total cell extracts were prepared and analyzed by Western blot as described previously [1]. Subcellular fractionation was performed using the Pierce NE-PER nuclear and cytoplasmic extraction reagents according

Immunofluorescence
Immunofluorescence for smooth muscle cell a-actin (SMA) was performed as described previously [27].

Statistics
Results are expressed as mean values AE SEM. Statistical differences between treatments were calculated by paired Student's t-test and considered significant when P < 0.05 ( Ã P < 0.05, ÃÃ P < 0.01, ÃÃÃ P < 0.001).

Eff|cient Overexpression and Knockdown of Dkk-3 in PrSCs
Primary prostatic stromal cells (PrSCs) were used to investigate the functional significance of Dkk-3 in the stromal compartment in vitro. Consistent with the predominant expression of Dkk-3 in the epithelial compartment of the benign prostate [1], Dkk-3 was more abundant in cell lysates from primary prostatic epithelial cells (PrECs) than PrSCs at the protein level as determined by Western blot analysis (Fig. 1A), however PrSCs secreted Dkk-3 at significant levels ( Fig. 1C).
To efficiently modify the expression of Dkk-3 in PrSCs in vitro, lentiviral-delivered DKK3 overexpression and knockdown systems were established. Stable

Dkk-3 Promotes Proliferation of PrSCs
We next investigated the influence of DKK3 overexpression and knockdown on the proliferation of PrSCs in vitro. Consistent with previous observations using adenovirus-delivered transient DKK3 overexpression [1], stable overexpression of DKK3 did not influence proliferation of PrSCs (Fig. 1D). On the other hand, lentiviral-delivered DKK3-shRNA significantly reduced cellular proliferation of PrSCs by approximately 23% compared with the SCR-shRNA ( Fig. 1D; P ¼ 0.007). To exclude potential off-target effects of the lentiviral knockdown system the effect of DKK3-shRNA on proliferation was additionally investigated in PC3 prostate cancer and HT-29 colon carcinoma cell lines with low endogenous DKK3 expression (Fig. 1E). DKK3-shRNA did not significantly affect cellular proliferation of both cell lines compared with SCR-shRNA control cells (Fig. 1F).
Knockdown of DKK3 has been reported to induce apoptosis and increase levels of BAX, p53 and p21 CIP1 in H460 lung cancer cells [39]. Thus, levels of these markers and the cyclin-dependent kinase inhibitor p27 KIP1 were analyzed in DKK3-shRNA PrSCs. Neither BAX nor phospho-p53 protein levels were significantly altered compared with SCR-shRNA, indicating that the lentiviral knockdown of DKK3 did not induce apoptosis in PrSCs (Fig. 1G). Consistent with reduced proliferation of DKK3-shRNA PrSCs, CDKN1A (p21 CIP1 ) and CDKN1B (p27 KIP1 ) mRNA levels were significantly elevated compared to SCR-shRNA (Supplemental Fig. 1). However, at the protein level only p27 KIP1 was elevated in DKK3-shRNA PrSCs, while p21 CIP1 protein levels were found decreased (Fig. 1G).
To exclude potential off-target effects of the lentiviral-delivered DKK3-shRNA construct a set of three different DKK3-targeted siRNA duplexes was investigated. As observed with DKK3-shRNA, siRNA-mediated DKK3 knockdown in PrSCs attenuated TGFb1induced fibroblast-to-myofibroblast differentiation as determined by SMA and IGFBP3 mRNA levels compared with scrambled control siRNA-treated cells (Fig. 2E).

DKK3 Knockdown Attenuates PI3K/AKT Signaling But Does Not Affect Wnt/b-Catenin Signaling
We next analyzed potential molecular pathways by which Dkk-3 mediates its effects. Since Dkk-3 has been related to Wnt/b-catenin signaling, the effect of DKK3-shRNA on b-catenin localization was analyzed using subcellular fractions. However, DKK3-shRNA neither affected cytosolic nor nuclear b-catenin levels in PrSCs (Fig. 4A), indicating that the effects of DKK3-shRNA are unlikely to be mediated via Wnt/b-catenin signaling.
Subsequently, we investigated the PI3K/AKT signaling pathway, a known mediator of proliferation and angiogenesis. DKK3-shRNA reduced basal AKT phosphorylation in PrSCs compared to SCR-shRNA treated control cells and additionally strongly attenuated induction of AKT phosphorylation in response to TGFb1 treatment (Fig. 4B), raising the possibility that the aforementioned modulatory effects of DKK3-shRNA on differentiation and angiogenic markers were due to attenuation of PI3K/AKT signaling.
We additionally analyzed phosphorylation of the TGFb-signaling mediator Smad2 and c-jun N-terminal kinase (JNK) which is essential for myofibroblast differentiation [29]. However, DKK3-shRNA did not significantly affect phosphorylation of both, Smad2 or JNK (Fig. 4B).

Inhibition of PI3K Mimics the Effects of DKK3 Knockdown
The specific PI3K inhibitor LY294002 was used to investigate whether attenuation of AKT phosphorylation in DKK3-shRNA PrSCs is responsible for the modulatory effects of DKK3 knockdown on angiogenesis and fibroblast-to-myofibroblast differentiation marker expression. PI3K inhibition significantly attenuated cellular proliferation in a dose-dependent manner ( Fig. 5A; 0 vs. 10 mM: P ¼ 0.010; 10 vs. 20 mM: P ¼ 0.044). Moreover, similar to DKK3-shRNA, the reduction in proliferation upon PI3K inhibition was associated with elevated CDKN1B/p27 KIP1 mRNA and protein levels and reduced p21 CIP1 protein levels while CDKN1A mRNA levels were increased ( Fig. 5B and Supplemental Fig. 1). Dkk-3 levels were unaffected by LY294002 (Fig. 5B).

Based on elevated Dkk-3 expression in BPH and
PCa-reactive stroma [1], the influence of Dkk-3 on remodeling of the tumor adjacent stroma was analyzed in vitro. Therefore lentiviral-delivered overexpression and shRNA-mediated knockdown of DKK3 was applied to PrSCs. In agreement with our previous finding using transient DKK3 overexpression [1], stable overexpression of DKK3 did not affect proliferation of PrSCs. These findings are in line with previous reports by ourselves and others that DKK3 overexpression or addition of exogenous purified Dkk-3 protein failed to reduce proliferation or induce apoptosis in malignant cells [1,[19][20][21]. Moreover, these findings further support the hypothesis that reported anti-proliferative or pro-apoptotic effects of DKK3 overexpression as a result of the unfolded protein response are in vitro artifacts that do not reflect the biological role of the endogenous protein.
We report herein that shRNA-mediated knockdown of DKK3 significantly attenuated proliferation of PrSCs, a finding consistent with our previous observations in PANC-1 cells [21]. While siRNA-mediated knockdown of DKK3 in H460 lung cancer cells has been recently shown to cause apoptosis and increased levels of p53, p21 CIP1 and BAX [39], we demonstrate that DKK3-shRNA mediated knockdown had no effect on phospho-p53 and BAX levels but increased p27 KIP1 and reduced p21 CIP1 levels. These effects could be mimicked by PI3K/AKT inhibition and are consistent with studies demonstrating stabilization of p21 CIP1 by PI3K/AKT signaling at the protein level [40,41]. Of note, both p27 KIP1 and p21 CIP1 mRNA levels were significantly elevated by DKK3-shRNA or PI3K/AKT inhibition (Supplemental Fig. 1), further suggesting elevated p21 CIP1 protein degradation.
As determined by knockdown of endogenous DKK3, Dkk-3 supported fibroblast-to-myofibroblast differentiation, a central process of stromal remodeling that promotes the development of BPH and PCa. Dkk-3 has also been shown to support differentiation of other cell types. For example, depletion of Dkk-3 disrupted acinar morphogenesis of the prostate epithelial cell line RWPE-1 [4,42]. Moreover, Dkk-3 supported capillary formation of peripheral blood-derived endothelial colony-forming cells [34]. On the other hand, DKK3-shRNA induced expression of differentiation markers in PANC-1 cells [21].
In vitro PrSCs abundantly expressed Dkk-3 whereas in the stromal compartment of the normal prostate Dkk-3 was not abundantly detected by immunohistochemistry [1]. However, normal prostate tissue homeostasis is associated with slow turnover and low proliferation index of epithelial and stromal cells [43][44][45]. Thus, it is conceivable that given the proproliferative activity of Dkk-3, in vitro out-growth of stromal cells from prostatic organoids selects for cells that (re-)express Dkk-3. Knockdown of DKK3 in PrSCs might therefore reflect the quiescent homeostatic state of stromal cells associated with low proliferation and differentiation.
Interestingly, overexpression of DKK3 did not affect TGFb1-induced myofibroblast differentiation. Thus, Dkk-3 appears to be required as a permissive factor for efficient differentiation as well as proliferation, while its overexpression has no noticeable effect on PrSCs. This is consistent with a previously reported permissive role of Dkk-3 in TGFb signaling during Xenopus mesoderm induction [46]. In contrast, in RWPE-1 cells silencing of Dkk-3 increased TGFb-signaling/phosphorylation of Smad-2 indicating that in epithelial cells Dkk-3 is not required as a permissive factor but rather limits TGFb-signaling [42]. However, in PrSCs phospho-Smad2 levels were unaffected by DKK3 knockdown strongly suggesting that in stromal cells Dkk-3 does not directly regulate TGFb/Smad-signaling. Dkk-3 expression was unaffected by TGFb1-treatment, additionally excluding the possibility that Dkk-3 represents a downstream target of TGFb1-signaling. In DKK3-shRNA PrSCs, suppression of TGFb1-induced differentiation correlated with attenuated phosphorylation of AKT and inhibition of PI3K significantly attenuated myofibroblast differentiation, mimicking the effect of DKK3-shRNA. These findings suggest that Dkk-3 represents a permissive factor that supports proliferation as well as fibroblast-to-myofibroblast differentiation potentially via modulation of PI3K/ AKT signaling. DKK3-shRNA did not significantly affect TGFb-induced phosphorylation of Smad2 or cjun N-terminal kinase, indicating that the effects of DKK3 knockdown were specific to and mediated via AKT and not due to blocking upstream TGFb1 signaling for example by attenuating global activation of the TGFb receptor. However, the detailed mechanism how Dkk-3 enhances AKT phosphorylation remains unclear and future studies will focus on identifying Dkk-3-interacting partners and whether DKK3 knockdown attenuates AKT phosphorylation upstream via PI3K or alternative kinases/phosphatases. Given the conflicting data on a potential role of Dkk-3 in Wnt/b-catenin signaling [7][8][9][10][11] we investigated intracellular b-catenin levels that upon activation of the canonical Wnt signaling accumulates in the cytoplasm and is translocated into the nucleus. How-ever, DKK3-shRNA did not affect b-catenin levels or subcellular localization, indicating that in PrSCs Dkk-3 does not act as a modulator of Wnt/b-catenin signaling.
In addition to inhibiting proliferation and differentiation, DKK3-shRNA or PI3K inhibition induced the expression of ANGPT1 and ANGPT2 mRNA and angiopoietin-1 but not angiopoetin-2 protein levels in PrSCs. ANGPT2 has been shown to be induced by inhibition of PI3K/AKT signaling in endothelial cells [47]. While ANGPT1 is known to stimulate the PI3K/AKT pathway [48], our data indicate that ANGPT1 expression is downregulated, potentially as a feedback loop, in response to PI3K/AKT. These findings raise the possibility that Dkk-3 is a co-factor in the initiation of the angiogenic switch observed in BPH and PCa that is associated with a shift in the ANGPT1/ANGPT2 ratio in favor of ANGPT2. We hypothesize that the loss of the vessel stabilizing factor ANGPT1 (that is highly expressed in the absence of  3). B: Western blot analysis of Dkk-3, p21 CIP1 and p27 KIP1 after 24 hr incubation with LY294002. b-actin served as loading control.C: Effect of PI3K inhibition using10 mMLY294002 and/or lentiviraldelivered shRNA-mediated DKK3 knockdown (DKK3-shRNA) on fibroblast-to-myofibroblast differentiation of PrSCs as determined by mRNA levels of the marker genes SMA and IGFBP3 after stimulation with 1 ng/ml TGFb1 (differentiation) or bFGF (control) for 24 hr.Gene expression levels were normalized using the housekeeping gene HMBS and are shown relative to scrambled (SCR-)shRNA and bFGF-treated controls. Bars represent mean AE SEM of three independent experiments. D: LY294002 induced ANGPT1 and ANGPT2 in a dose-dependent manner at mRNA levels within 4 hr of treatment (n ¼ 3). E: Secreted angiopoietin-1 levels as determined in conditioned media of SCR-/ DKK3-shRNA-treated PrSCs after incubationwithout/with10 mMLY294002 for 72 hr (n ¼ 4).
Dkk-3) due to elevated local Dkk-3 levels in endothelial cells and the surrounding stroma, leads to vessel destabilization that favors angiogenic sprouting.
Taken together our findings indicate that elevated Dkk-3 levels in the stromal compartment of BPH and PCa patients enhances (i) fibroblast proliferation and (ii) myofibroblast differentiation, both hallmarks of stromal remodeling and (iii) contribute to the angiogenic switch via suppression of vessel stabilizing angiogenic factors like ANGPT1.

CONCLUSIONS
Dkk-3 supported proliferation and fibroblast-tomyofibroblast differentiation and suppressed expression of angiogenic factors in PrSCs. DKK3-shRNA mediated knockdown attenuated AKT-phosphorylation and inhibition of PI3K mimicked the effects observed by DKK3 knockdown, suggesting that Dkk-3 may represent a permissive co-factor of PI3K/AKT signaling in PrSCs. Collectively, these data suggest that altered Dkk-3 expression observed in BPH and PCa may support stromal proliferation and differentiation and the initiation of the angiogenic switch, all of which are key hallmarks of stromal remodeling in prostatic disease (Fig. 6). Therefore, Dkk-3 represents a potential therapeutic target for stromal remodeling in BPH and PCa.

ACKNOWLEDGMENTS
The authors thank Martin Heitz for his excellent technical support. The work of PB was supported by the Austrian Science Fund (FWF; NRN S9307-B05). NS was funded by an Elise Richter post-doctoral fellowship (Austrian Science Fund; FWF V216-B13).  6. Proposed model of Dkk-3 impact on stromal remodeling in BPH and PCa. In normal prostate tissue, Dkk-3 is predominantly expressed in the epithelium but not in the stroma, which is characterized by a high fibroblast/myofibroblast ratio. Angiogenic factors such as ANGPT1, which stabilize the vessels, are highly expressed in endothelial and surrounding stromal cells that produce low levels of Dkk-3. In the diseased prostate, Dkk-3 expression is elevated in the stromal compartment, especially endothelial cells and potentially acts as a permissive factor for PI3K/AKTsignaling, enhancing proliferation and differentiation of fibroblasts leading to stromal enlargement and elevated myofibroblast content. High Dkk-3 levels in vessels and surrounding stroma downregulate local expression of ANGPT1 shifting the ANGPT1/ ANGPT2 ratio in favor of ANGPT2 thatconsequentlyresultsinvessel destabilization and sproutingofmicrovesselsinto the stroma.

SUPPORTING INFORMATION
Additional supporting information may be found in the online version of this article at the publisher's website. Figure S1.A: CDKN1A (P ¼ 0.0046) and CDKN1B (P ¼ 0.0001) mRNA levels in primary prostatic stromal cells (PrSCs) were elevated upon DKK3 knockdown using lentiviral-delivered shRNA compared to srambled control (SCR) shRNA as determined 72 hr post-transduction (n ¼ 6). B: Treatment with 10 mM of the PI3K inhibitor LY294002 for 4 hr induces CDKN1A (P ¼ 0.0036) and CDKN1B (P ¼ 0.0067) mRNA levels in PrSCs (n ¼ 5). Gene expression levels were normalized using the housekeeping gene HMBS and are shown relative to controls. Figure S2. A: Basal DKK3, ANGPT1, and ANGPT2 mRNA levels in primary prostatic stromal cells (PrSCs; n ¼ 3) relative to the housekeeping gene HMBS. B: 72 hr post-viral transduction with DKK3-shRNA or scrambled control (SCR) shRNA PrSCs were stimulated with 1 ng/ml bFGF or TGF b AE 10 mM of the PI3K inhibitor LY294002 for 72 hr. Secreted angiopoietin-2 levels were not significantly affected by any treatment applied (n ¼ 3).