δ-Catenin promotes E-cadherin processing and activates β-catenin-mediated signaling: implications on human prostate cancer progression

Abstract

δ-Catenin binds the juxtamembrane domain of E-cadherin and is known to be overexpressed in some human tumors. However, the functions of δ-catenin in epithelial cells and carcinomas remain elusive. We found that prostate cancer cells overexpressing δ-catenin show an increase in multi-layer growth in culture. In these cells, δ-catenin colocalizes with E-cadherin at the plasma membrane, and the E-cadherin processing is noticeably elevated. E-Cadherin processing induced by δ-catenin is serum-dependent and requires MMP- and PS-1/γ-secretase-mediated activities. A deletion mutant of δ-catenin that deprives the ability of δ-catenin to bind E-cadherin or to recruit PS-1 to E-cadherin totally abolishes the δ-catenin-induced E-cadherin processing and the multi-layer growth of the cells. In addition, prostate cancer cells overexpressing δ-catenin display an elevated total β-catenin level and increase its nuclear distribution, resulting in the activation of β-catenin/LEF-1-mediated transcription and their downstream target genes as well as androgen receptor-mediated transcription. Indeed, human prostate tumor xenograft in nude mice, which is derived from cells overexpressing δ-catenin, shows increased β-catenin nuclear localization and more rapid growth rates. Moreover, the metastatic xenograft tumor weights positively correlate with the level of 29kD E-cadherin fragment, and primary human prostate tumor tissues also show elevated levels of δ-catenin expression and the E-cadherin processing. Taken together, these results suggest that δ-catenin plays an important role in prostate cancer progression through inducing E-cadherin processing and thereby activating β-catenin-mediated oncogenic signals.

Figure 1. δ-Catenin overexpression induces increased multi-layer growth in CWR22Rv-1 cells

(A) Establishment of stable polyclonal CWR22Rv-1 cells overexpressing GFP (Rv/C) or GFP-δ-catenin (Rv/δ). Expression level of exogenous GFP epitope (left panel) and δ-catenin (right panel) were shown. Rv/C cells harbor only lower levels of endogenous δ-catenin while Rv/δ cells have robust exogenous δ-catenin. (B) Multi-layer cluster formation of Rv/δ cells. Cell growth morphologies of Rv/C and Rv/δ in culture were observed by phase contrast microscopy. High magnificent images of multilayer cluster region in boxed area were shown in right panel. (C) Membrane localization of δ-catenin in Rv/δ cells. GFP signal from Rv/δ cells was observed under confocal fluorescent microscopy. Z-X and Z-Y section analysis of multi-layer clusters revealed that on average five to six cells were stacked at a height of ~100 μm. High magnification image was shown in bottom left panel. (D–F) Immunoreactivity of membrane E-cadherin in Rv/δ cells. Paraformaldehyde-fixed and Triton X-100-permeabilized Rv/δ cells were incubated with E-cadherin (BD) antibody and subsequently with Rhodamine-conjugated mouse secondary antibody. While the majority of cells showed co-localization of δ-catenin with E-cadherin, some sort of cells showed an inverse correlation between δ-catenin and E-cadherin level. Arrow and arrowhead in bottom left panel indicate representative cell that showed high δ-catenin level with low E-cadherin immunoreactivity, and low δ-catenin level with high E-cadherin immunoreactivity, respectively (D). Rv/δ cells were independently immunostained with monoclonal E-cadherin antibody (Cell Signaling Technology, epitope: sequence surrounding 780^th amino acid of human E-cadherin) and Alexa Fluor 633 goat anti-mouse IgG (H+L) secondary antibody. Representative image was shown (upper panel, E). There are three types of junction be classified according to relative levels of δ-catenin and E-cadherin. ROI type 1 represents the junctions that δ-catenin and E-cadherin co-localizes with high intensity while ROI type 2 and type 3 represent the junctions which showed inverse correlation between δ-catenin and E-cadherin level (bottom panel, E). Red, E-cadherin; Green, GFP-δ-catenin. Quantitative numbers of each type of junction were shown (F). Note that δ-catenin and E-cadherin co-localized mainly at the cell-cell junction, while intracellularly, their location seems distinctive; δ-catenin localized only in a cytoplasm but E-cadherin was in both nuclear and cytoplasm. No significant co-localization of δ-catenin and E-cadherin was observed in intracellular region. Bar, 25 μm.

Figure 2. δ-Catenin overexpression promotes E-cadherin processing

(A) Enhanced E-cadherin processing in Rv/δ cells. E-Cadherin from whole cell lysates of Rv/C and Rv/δ cells were probed with an anti-E-cadherin (BD) antibody. Shorter exposure image in upper right panel shows a concomitant reduction of full-length E-cadherin. (B) E-Cadherin processing in PC3 cells overexpressing δ-catenin. PC3 cells were transiently transfected with expression plasmid of δ-catenin with GFP tagging and harvested after 48 h incubation. Whole cell lysates were subjected to Western blot. (C) Reduction of E-cadherin processing in Rv/δ cells by knock-down of δ-catenin using δ-catenin siRNA in a dose dependent manner. Cells were transfected with two different δ-catenin-specific siRNAs, and whole cell lysates were subjected to Western blot. Shorter exposure image was shown in upper right panel. (D) Detection of E-cadherin processing in membrane fraction of Rv/δ cells. Rv/δ cells were fractionated into cytoplasmic (C), nuclear (N) and membrane (M). β-Tubulin was used to exclude contamination between nuclear and cytoplasmic fractions. Shorter exposure image was shown in right panel. (E) Detection of plasma membrane E-cadherin by biotin-labeling purification using sulfo-NHS-SS-biotin. Purified membrane E-cadherin was probed using E-cadherin antibody either purchased from BD Biosciences (epitope: 773–791 of E-cadherin cytoplasmic region, left panel) or Santa Cruz Biotechnology (epitope: 600–707 of E-cadherin extracellular region, right panel). Shorter exposure image was shown in right panel. Note that more evident patterns of E-cadherin processing were observed after purification, and that due to the different binding epitope of each antibody, the size and pattern of detected E-cadherin bands were different. Rv/C cells also underwent E-cadherin processing that was further potentiated in Rv/δ cells. (F) Blockage of endocytosis increases E-cadherin processing. Cells were treated with E-64 (lysosomal protease inhibitor) for 24 h or sucrose (endocytosis blocker) for 2 h prior to harvest, followed by biotin-labeling purification of membrane proteins.

Figure 3. E-Cadherin processing induced by δ-catenin is serum-dependent and requires MMP- and PS-1/γ-secretase-mediated activities

(A) Effects of several growth factors on E-cadherin processing in Rv/δ cells. Rv/δ cells were exposed to serum deprivation for 8 h and then further treated with FBS (10%), HGF, insulin or EGF at either 20 ng/ml or 40 ng/ml concentrations for 16 h, and then followed by biotin-labeling purification of membrane proteins. Shorter exposure image was shown in upper right panel. HGF, hepatocyte growth factor; EGF, epidermal growth factor. (B) Effects of exhausting culture media of extracellular serum factor(s) on E-cadherin processing in Rv/δ cells. Rv/δ cells were cultured in the same culture medium for up to four days, followed by biotin-labeling purification of membrane proteins. Shorter exposure image was shown in upper right panel. (C) Effects of MMP (GM6001) and γ-secretase inhibitors on E-cadherin processing in Rv/δ cells. Cells were treated with either GM6001 (10 μM) or γ-secretase inhibitor (20 μM) for 16 h, followed by biotin-labeling purification of membrane proteins. Shorter exposure image was shown in upper right panel. (D) Effects of wild-type or D257A mutant (dominant negative form) of presenilin-1 on E-cadherin processing. Rv/C cells were transfected with either WT or D257A mutant of presenilin-1 (PS-1), and then followed by biotin-labeling purification of membrane proteins. Bottom panel shows expression of transfected presenilin-1. Non-specific background of HA antibody was overlapped with presenilin-1 band.

Figure 4. δ-Catenin deletion mutant lacking ability to bind E-cadherin or to recruit PS-1 to E-cadherin shows neither E-cadherin processing nor multi-layer growth pattern

(A) Establishment of polyclonal CWR22Rv-1 cells (Rv/Δ) harboring ΔC787 δ-catenin mutant lacking the C-terminal 787 amino acids. Expression level of GFP epitope in Rv/C, Rv/δ and Rv/Δ cells were shown using GFP antibody. (B) Loss of membrane localization of ΔC787 δ-catenin mutant in Rv/Δ cells. GFP signal from Rv/Δ cells were observed under fluorescent microscopy. Note that ΔC787 δ-catenin mutant is mainly localized at the cytoplasm. (C) ΔC787 δ-catenin mutant lacks binding to E-cadherin. Whole cell lysates were subjected to immunoprecipitation with GFP antibody and bound E-cadherin was probed by E-cadherin (upper panel). Levels of δ-catenin and E-cadherin in lysates were shown in bottom panel. Arrowhead, GFP or GFP-tagged δ-catenin; arrow, E-cadherin. (D) ΔC787 δ-catenin mutant unable to recruit PS-1 to E-cadherin. Rv cell lines were transfected with HA-presenilin-1, and whole cell lysates were subjected to immunoprecipitation with HA antibody followed by immunoblotting with E-cadherin antibody for bound E-cadherin. Faint level of bound E-cadherin was observed due to the endogenous δ-catenin in both Rv/C and Rv/Δ cells. Evident recruitment of presenilin-1 to E-cadherin was seen only in Rv/δ cells, which harbor full-length δ-catenin. Arrow, E-cadherin. (E) Loss of E-cadherin processing in Rv/Δ cells. Plasma membrane E-cadherin processing was measured by biotin-labeling purification using sulfo-NHS-SS-biotin. Shorter exposure image was shown in upper right panel, respectively. (F) Loss of multi-layer cluster formation of Rv/Δ cells. Cell growth morphology of Rv/Δ cells in culture was observed by phase contrast microscopy. High magnificent images were shown in bottom panel. Bar, 25 μm.

Figure 5. δ-Catenin overexpression alters subcellular β-catenin distribution and activates its downstream effectors

(A) Increased β-catenin nuclear localization in Rv/δ cells. Nuclear and cytoplasmic fractionation of β-catenin was performed in Rv/C and Rv/δ cells. Moderate increase in total β-catenin level was also observed in Rv/δ cells. α-Tubulin was used as a cytoplasmic marker and α-histone H3 was used as a nuclear marker. (B) Knock-down of δ-catenin from Rv/δ cells reverts increased β-catenin level into that of Rv/C cells. Cells were transfected with either control or δ-catenin-specific siRNA, and the level of β-catenin was measured from whole cell lysates after 48 h incubation. (C) Increased TOPFLASH activity in Rv/δ cells. Cells were transfected with LEF-1 together with TOPFLASH reporter plasmid, and luciferase activity was measured after 48 h incubation. *, p<0.001 compared to TOPFLASH in Rv/C. (D and E) Augmentation of β-catenin target gene level in Rv/δ cells. Levels of several target genes for β-catenin/LEF-1 were analyzed by RT-PCR (D) and Western blot (E) in Rv/C and Rv/δ cells. GAPDH, cyclin-dependent kinase 4 (CDK4), and α-histone H3 were used as loading controls. (F) Greater androgen-stimulated androgen receptor (AR) activation in Rv/δ cells. Cells were transfected with pGL-ARE4-luc, and R1881 was treated 24 h post-transfection. After additional 24 h incubation, luciferase activity was measured. Each bar represents the mean ± S.D. In these lysates, the levels of endogenous androgen receptor in Rv/C and Rv/δ cells were shown (panel G). No significant induction was observed.

Figure 6. Effects of δ-catenin overexpression on prostate cancer cell expansion

(A) Rapid growth of tumor xenografts derived from Rv/δ cells. Tumor xenografts were generated after s.c. injecting either Rv/C or Rv/δ cells into BALB/c nude mice, and tumor volumes were measured on the days indicated (n=4 for each). (B) Correlation between metastatic tumor weight and 29 kD E-cadherin fragment. Metastatic tumors were formed around the neck of mice after tail-vein injection of Rv/δ cells. Upon analyzing these tumors for E-cadherin status by Western blot, larger tumors tended to have more processed E-cadherin fragments, especially 29 kD. (C) Increased β-catenin nuclear accumulation in xenografts tissue from Rv/δ cells. Xenograft tissues were subjected to immunohistochemistry for β- and δ-catenin. In xenograft tumor tissues, those that were generated by injecting Rv/δ cells either subcutaneously or intravenously showed clear and frequent nuclear β-catenin localization. Arrows indicate tumor cells displaying nuclear accumulation of β-catenin. Bar, 25 μm.

Figure 7. Increased δ-catenin expression and E-cadherin processing in human prostate cancer specimens

Human prostate cancer tissues and matched adjacent normal tissues were analyzed for E-cadherin status and δ-catenin expression. N, normal; T, Tumor. Arrow indicates upper 75 kD E-cadherin band. The T/N ratio of δ-catenin levels was obtained from quantified δ-catenin level in tumor and normal tissues and is normalized by corresponding quantified actin band.

Figure 8. A proposed model of E-cadherin processing

A proposed model showing that δ-catenin activates MMP-dependent E-cadherin processing and recruits PS-1/γ-secretase complexes in order to activate another protease that further cleaves several site of E-cadherin ectodomain. MMP and γ-secretase inactivate an unknown protease (symboled as ‘?’), which normally cleaves near the transmembrane region of E-cadherin. β; β-catenin