Effects of Dickkopf-1 (DKK-1) on Prostate Cancer Growth and Bone Metastasis

Abstract

Osteoblastic bone metastases are commonly detected in patients with advanced prostate cancer (PCa) and are associated with an increased mortality rate. Dickkopf-1 (DKK-1) antagonizes canonical WNT/β-catenin signaling and plays a complex role in bone metastases. We explored the function of cancer cell-specific DKK-1 in PCa growth, metastasis, and cancer-bone interactions using the osteoblastic canine PCa cell line, Probasco. Probasco or Probasco + DKK-1 (cells transduced with human DKK-1) were injected into the tibia or left cardiac ventricle of athymic nude mice. Bone metastases were detected by bioluminescent imaging in vivo and evaluated by micro-computed tomography and histopathology. Cancer cell proliferation, migration, gene/protein expression, and their impact on primary murine osteoblasts and osteoclasts, were evaluated in vitro. DKK-1 increased cancer growth and stimulated cell migration independent of canonical WNT signaling. Enhanced cancer progression by DKK-1 was associated with increased cell proliferation, up-regulation of NF-kB/p65 signaling, inhibition of caspase-dependent apoptosis by down-regulation of non-canonical WNT/JNK signaling, and increased expression of epithelial-to-mesenchymal transition genes. In addition, DKK-1 attenuated the osteoblastic activity of Probasco cells, and bone metastases had decreased cancer-induced intramedullary woven bone formation. Decreased bone formation might be due to the inhibition of osteoblast differentiation and stimulation of osteoclast activity through a decrease in the OPG/RANKL ratio in the bone microenvironment. The present study indicated that the cancer-promoting role of DKK-1 in PCa bone metastases was associated with increased growth of bone metastases, reduced bone induction, and altered signaling through the canonical WNT-independent pathway. DKK-1 could be a promising therapeutic target for PCa.

Keywords: DKK-1; canine; dog; osteoblastic bone metastasis; prostate cancer.

Figure 1

Probasco or Probasco + DKK-1 cells were injected into the left cardiac ventricle of athymic nude mice. Mice were euthanized at week 5. (a) Representative bioluminescent images of mice showing the metastatic sites at week 5. (b) The quantification of the total number of metastasis sites based on bioluminescent signals at week 5 (signals detected from the chest region were due to the growth of cancer cells that leaked during intracardiac injections and were not counted as metastatic sites). Data were displayed as scatter plots with mean lines. (c) Quantification of bioluminescent intensity from metastatic bone sites at week 5. Bioluminescent intensity is proportional to the number of viable tumor cells and reflects tumor size. Data were displayed as box and whisker plots. (d) H&E section of the non-tumor tibia (i,ii) and tumor-bearing tibia. Probasco formed smaller tumors and caused intramedullary woven bone formation (iii,iv). Probasco + DKK-1 had larger tumors that grew inside and outside the bone and induced predominant periosteal new bone formation (v,vi) (Enlarged images of black rectangular regions in (i,iii,v) were shown in (ii,iv,vi); T, tumor; BM, bone marrow; WB, tumor-induced woven trabecular bone; arrows, osteoblasts; bars, (i,iii,v) = 500 µm; (ii,iv,vi) = 100 µm). n = 7 for Probasco group, n = 6 for Probasco + DKK-1 group. Data were analyzed using an unpaired t-test. ns: no significance, * p < 0.05. For all figures, significant differences are marked with asterisks.

Figure 2

Probasco or Probasco + DKK-1 cells were injected into the proximal tibias of athymic nude mice (n = 9 for Probasco group, n = 8 for Probasco + DKK-1 group). Mice were euthanized at week 4. The growth of tumors was monitored by bioluminescence, and the structure of bones was assessed by µCT (micro-computed tomography). (a) Representative bioluminescent images of tumors at weeks 2, 3, and 4 post-injection. (b) Quantification of tumor growth based on bioluminescent intensity. Two-way ANOVA with Sidak’s multiple comparisons test, *** p < 0.001. Data were displayed as box and whisker plots. (c) Representative µCT-scanned images of the whole hindlimb and proximal tibial metaphysis regions (2.5 mm volume of interest starting 1 mm below the proximal tibial metaphysis) showed the 3D-reconstructed structure of tumor-bearing tibias. Quantification of µCT demonstrated the bone volume fractions of cortical bone (d) and periosteal new bone (e). Data were analyzed using an unpaired t-test. *** p < 0.001, **** p < 0.0001. Data were displayed as mean + SD.

Figure 3

Histological assessment of intratibial-injected bones. (a) Representative low-magnification and high-magnification (rectangular regions) images of tibia bone sections stained with H&E (i–iv) and TRAP (v–viii) (Enlarged images of black rectangular regions in (i,iii,v,vii) were shown in (ii,iv,vi,viii); WB, tumor-induced woven bone; T, tumors; arrow, osteoblasts; arrowhead, osteoclasts; bars, (i,iii,v,vii) = 1 mm; (ii,iv,vi,viii) = 50 µm). (b) Quantitative histomorphometric analysis displayed the ratio of the intramedullary woven bone area to the tumor area. (c) The ratio of periosteal new bone area to the tumor area. n = 9 for Probasco group, n = 8 for Probasco + DKK-1 group. t-test was used for data analysis. *** p < 0.001. Data were displayed as mean + SD.

Figure 4

Effects of Probasco or Probasco + DKK-1 conditioned medium (CM) on primary murine osteoclasts and osteoblasts. (a) mRNA expression of Mmp9 and Ctsk in control (ctrl) and CM-treated osteoclasts. (b) Representative images and quantification of TRAP-stained osteoclasts. TRAP-positive osteoclasts were stained with a red to purple color, and the number of multinucleated osteoclasts was quantified as osteoclast number per area (mm²). Bar = 200 µm. n = 5 to 6 for each osteoclast group. (c) mRNA expression of Rankl and Opg in osteoblasts. (d) von Kossa staining for the mineralization of osteoblast cultures, and the mineralized area was quantified. n = 3 for each osteoblast group. Data were analyzed using one-way ANOVA with Tukey’s multiple comparison test. ns: no significance, * p < 0.05, ** p < 0.01. Data were displayed as mean ± SD.

Figure 5

In vitro morphology, growth and migration, and gene expression of Probasco and Probasco + DKK-1 cells. (a) Phase contrast microscopy of cells in culture. (b) In vitro cell growth. n = 3 for each group, two-way ANOVA with Sidak’s multiple comparisons test, *** p < 0.001, **** p < 0.0001. (c) Representative images and quantification of in vitro cell migration. n = 4 for each group, two-way ANOVA with Sidak’s multiple comparisons test, **** p < 0.0001. (d) mRNA expression of human DKK-1 and epithelial–mesenchymal transition (EMT)-related genes (CDH1, SLUG, SNAIL, TWIST1) in Probasco + DKK-1 cells compared to Probasco cells. n = 3 for each group, t-test, ns: no significance, * p < 0.05, ** p < 0.01, *** p < 0.001. Data were displayed as mean ± SD.

Figure 6

Alteration of signaling pathways by autocrine effects of DKK-1 in Probasco cells. (a) Western blots were performed on total protein lysates from cell lines for detecting (b) the overexpressed human DKK-1, (c) canonical WNT signaling; non-phospho β-catenin (active form) and β-actin (loading control), (d) non-canonical WNT/JNK signaling; phospho-JNK (active form) and total JNK, (e) apoptosis; cleaved caspase-9 (active form) and caspase-9, and (f) NF-kB/p65 signaling; phosphor p65 (active form) and total p65. Protein levels in DKK-1 cells were analyzed by densitometry and compared to levels in parental Probasco cells. (g,h) mRNA expression of NF-kB/p65 downstream target genes COX2 and VEGFA. n = 3 for each group, t-test, ns: no significance, * p < 0.05, ** p < 0.01, *** p < 0.001. Data were displayed as mean ± SD.