Metastatic progression of prostate cancer and e-cadherin regulation by zeb1 and SRC family kinases

Abstract

Expression of E-cadherin is used to monitor the epithelial phenotype, and its loss is suggestive of epithelial-mesenchymal transition (EMT). EMT triggers tumor metastasis. Exit from EMT is marked by increased E-cadherin expression and is considered necessary for tumor growth at sites of metastasis; however, the mechanisms associated with exit from EMT are poorly understood. Herein are analyzed 185 prostate cancer metastases, with significantly higher E-cadherin expression in bone than in lymph node and soft tissue metastases. To determine the molecular mechanisms of regulation of E-cadherin expression, three stable isogenic cell lines from DU145 were derived that differ in structure, migration, and colony formation on soft agar and Matrigel. When injected into mouse tibia, the epithelial subline grows most aggressively, whereas the mesenchymal subline does not grow. In cultured cells, ZEB1 and Src family kinases decrease E-cadherin expression. In contrast, in tibial xenografts, E-cadherin RNA levels increase eight- to 10-fold despite persistent ZEB1 expression, and in all ZEB1-positive metastases (10 of 120), ZEB1 and E-cadherin proteins were co-expressed. These data suggest that transcriptional regulation of E-cadherin differs in cultured cells versus xenografts, which more faithfully reflect E-cadherin regulation in cancers in human beings. Furthermore, the aggressive nature of xenografts positive for E-cadherin and the frequency of metastases positive for E-cadherin suggest that high E-cadherin expression in metastatic prostate cancer is associated with aggressive tumor growth.

Figure 1

E-cadherin expression in metastatic prostate cancers. A: E-cadherin protein in 40 patients. Bars represent average expression from all metastatic sites in a patient, and SD demonstrates the variability of expression. B: E-cadherin expression in bone versus soft tissue metastases. Circles represent mean intensity scores from duplicate cores. Bone metastases are from 40 patients, and soft tissue metastases from 23 patients. The difference between bone and soft tissue metastases is statistically significant (*P < 0.01). C: Contrast of E-cadherin expression between metastatic sites in the same patient. Solid circles represent bone metastases, and open circles represent lymph node metastases (top panel) or soft tissue metastases (bottom panel). Lines connect metastases in the same patient. D: E-cadherin RNA and protein expression in prostate cancer xenografts. RNA data are from oligonucleotide arrays (Agilent Technologies, Inc., Santa Clara, CA), and protein expression data are from a xenograft TMA.

Figure 2

Isolation of isogenic sublines from DU145 prostate cancer cells based on expression levels of E-cadherin (E-cad). A: Generation of spindle (S-DU145), round (R-DU145), and tight (T-DU145) sublines. Parent DU145 cells were sorted using FACS based on E-cadherin expression. Cells with the lowest 10% and highest 10% E-cadherin expression were placed in three-dimensional Matrigel, and resulting spheres were picked individually and propagated in two dimensions. Scale bars = 50 μm. B: RNA expression of E-cadherin and E-cadherin repressors SNAIL1, SLUG, ZEB1, and ZEB2. Quantitative PCR cycle numbers were normalized to RPL13A. C: Protein expression of E-cadherin and E-cadherin repressors. Fifty micrograms of whole-cell lysates was analyzed for E-cadherin, vimentin, and cytokeratin 18 (CK18), and 25 μg nuclear protein was analyzed for ZEB1 and SNAIL. D: Expression of SFKs in DU145 sublines. Fifty micrograms of membrane (mem) fraction or whole-cell lysates (WCL) from S-DU145 (S), R-DU145 (R), and T-DU145 (T) was analyzed for expression of pSFK. The blot was reprobed to determine SFK and β-tubulin expression. WCL from VCAP (V) cells were used as a positive control. E: Regulation of E-cadherin expression by SFK. R-DU145, S-DU145, and T-DU145 were treated with the SFK inhibitor PP2 or with the proteosome inhibitors E64 or MG132. WCL were analyzed for E-cadherin protein expression.

Figure 3

Functional characterization of DU145 sublines. A: Migration of cells through Transwell filters. To the upper chamber, 1 × 10⁵ cells in RPMI were added, and the lower chamber contained either 1% fetal bovine serum (FBS), 1000 U/mL hepatocyte growth factor (HGF), or serum-free conditioned medium from bone marrow stroma cells (BMSC-CM). For HGF migration, filters were coated with Matrigel, and cells were allowed to migrate for 6 hours, then were fixed, stained, and counted. Six fields were counted per filter and averaged, and the background spontaneous migration (no attractant) was subtracted. B: Soft agar colony formation assay. Left panel, Cells, 1 × 10⁶, were encased in top agar and grown for 2 weeks. Right panel, Plates were photographed at 4× magnification, and colonies were counted, with the large and small colonies differentiated at 0.5 cm in diameter. C: Matrigel assay. Left panel, Cells, 1 × 10³ per well of a four-well chamber slide, were grown for 10 days in Matrigel. Right panel, Wells were photographed at 10× magnification for counting.

Figure 4

Transcriptional regulation of E-cadherin in DU145 sublines. A: Knockdown of E-cadherin repressors. SNAIL, SLUG, ZEB1, or ZEB2 expression was reduced in R-DU145 cells using a SMARTpool of 4 siRNAs for each gene. Twenty-five micrograms of protein was probed for expression of E-cadherin (E-cad), ZEB1, SNAIL, or β-catenin. T-DU145 served as a positive control in lanes 1, 4, and 5. B: Morphologic analysis after transfection with siRNAs. S-DU145 cells were transfected and photographed after 72 hours. The transfection efficiency, evaluated with GFP, was >80% (data not shown). Scale bars = 100 μm.

Figure 5

Growth of DU145 sublines in mouse tibia. A: Radiographs of tibial tumors from S-DU145, R-DU145, and T-DU145 cells. Tumor growth was followed up at radiography. Numbers indicate animals with tumors per total animals injected. Ovals demonstrate areas of osteolysis. B: Quantification of bone destruction in T-DU145 and R-DU145 xenografts. Bone destruction was measured by the loss of pixels in radiographs of four mice. C: Cell proliferation. Proliferation rates were measured using Ki-67 IHC. The percentage of Ki-67–positive tumor cells from three separate xenografts is shown. The difference in proliferation index is insignificant (*P = 0.74). D: Apoptosis. Apoptotic cells were visualized via TUNEL staining. The mean percentage per 40× high-powered field from three separate T-DU145 or R-DU145 xenografts is shown. The difference in the apoptotic index between T-DU145 and R-DU145 is significant (*P = 0.03).

Figure 6

Co-expression of E-cadherin and ZEB1 in xenografts and patient metastases. A: RNA expression in preinjection R-DU145 and two tibial xenografts, mR-T4 and m-RT5. RNA for E-cadherin, ZEB1, and SNAIL was measured using human-specific primers. B: Expression of protein in preinjection R-DU145 and in cells cultured from three tibial xenografts. R-DU145 cells from three tibiae (mR-A, mR-B, and mR-C) were briefly cultured ex vivo. Cells were analyzed for E-cadherin, β-catenin, ZEB1, and SNAIL using Western blot analysis. C: E-cadherin and ZEB1 are co-expressed in R-DU145 xenografts. Top left panels, Parallel sections of tibial xenografts were stained with E-cadherin (E-cad), ZEB1, and Ki-67. Bottom left panels, The percentage of ZEB1- or Ki-67-positive cells was quantified by counting E-cadherin–positive and E-cadherin–negative areas. D: SFK activity in R-DU145 cells. Cultured R-DU145 cells before and after injection were analyzed for pSFK. The Western blot membrane was reprobed for total expression of SFK, vimentin, and β-actin. VCAP cells were used as a positive control for SKF. E: E-cadherin and ZEB1 are co-expressed in patient tumor metastases. The TMA used in Figure 1 was stained with the ZEB1 antibody (S1), and cores were scored for ZEB1 expression in the nucleus of cancer cells. Cores positive for ZEB1 expression are shown in blue. E-cadherin expression in the same core is indicated in the column adjacent to ZEB1. ZEB1-negative metastatic sites are shown in yellow.