DKK1 promotes hepatocellular carcinoma cell migration and invasion through β-catenin/MMP7 signaling pathway

3-(4, 5-dimethylthiazol)-2, 5-diphenyltetrazolium bromide (MTT), Hoechst33342, G418 and LiCl were purchased from Sigma (St. Louis, MO, USA). RPMI 1640 and fetal calf serum (FCS) were purchased from Gibco (Grand Island, NY, USA). The sources of primary antibodies used for Western blot: DKK1 (cat. sc-25516), LRP6 (cat. sc-17982), GSK3β (cat. sc-53931), β-catenin (cat. sc-7199) and MMP7 (cat. sc-101566) as well as the corresponding horseradish peroxidase-conjugated second antibodies were all purchased from Santa Cruz Biotechnologies (Santa Cruz, CA, USA). Cy3-labeled goat anti-mouse second antibody was purchased from Beyotime Institute of Biotechnology (cat. A0521, Shanghai, China). All other chemicals used in this study were commercial products of reagent grade.

Human HCC (30 cases of primary, 41 cases of metastasis) and adjacent non-tumor liver tissues were collected from 71 patients undergoing resection of HCC at the Huai-He Hospital, Henan University (Kaifeng, China). All tissues were classified according to the tumor-node-metastasis classification system of the union for International Cancer Control. Informed consent was obtained from each patient, and the study was approved by the Institute Research Ethics Committee at the hospital.

Cell lines, derivative from human HCC (HepG2, SMMC7721, Huh7 and Bel7402) and normal liver cell line (QSG-7701) were used in this study. All of them were purchased from the cell bank of the Chinese Academy of Science (Shanghai, China), these cells were maintained in RPMI1640 medium containing 10% fetal calf serum, 100 units/mL penicillin, and 100 μg/mL streptomycin, at 37°C in a humidified incubator containing 5% CO₂.

The full-length coding region (801 bp) of DKK1 was amplified from human genomic DNA by reverse transcription-polymerase chain reaction (RT-PCR) using the following primers: DKK1-sense 5′-CCG CTC GAG ATG ATG GCT CTG GGC GCA GCGGG-3′, and antisense 5′-CGG GAT CCG CTG GTT TAG TGT CTC TGA CAAGT-3′, then the PCR products were digested with XhoI/BamHI and were inserted into the pIRES2-EGFP vector (Clontech).The recombinant construct was verified by direct DNA sequencing.

For the construction of siRNA expression plasmids, A siRNA sequence targeted human DKK1 transcript (accession no. NM_012242.2; sense 5′-GGAATAAGTACC AGACCATTG-3′) was selected for RNA interference (RNAi). This sequence showed no homology with other known human genes. A scrambled sequence (sense 5′-GGAATAAGACCATGACCATTG-3′), which is no homologous to any human DNA sequence, served as a negative control. Two complementary oligonucleotides that contain both sense and antisense siRNA sequences, the loop sequence (5′-TTCAAGAGA-3′) and the flanking Bam H I and Eco R I sites were synthesized chemically. Then these two complementary oligonucleotides were annealed and ligated into the linearized pSIREN-Shuttle (Clontech). Each recombinant construct was sequenced to confirm the right sequence of insert. The construct contained siDKK1 sequence was designated as pSIREN-Shuttle-siDKK1, while the plasmid with scrambled sequence was named as pSIREN-Shuttle-Control.

For transfection, HepG2 and Bel7402 cells were seeded in 24-well plates at 5 × 10⁴ per well and incubated at 37°C in a humidified incubator containing 5% CO₂. The next day, when these cells were about 80% confluence, cells were transfected with TurboFect™ in vitro Transfection Reagent according to the manufacturer’s instructions. These transfected cells were selected with 0.1 mg/mL G418 for at least 2 weeks, and then the stable plasmid-transfected clones were generated by using limiting dilution analysis in 96-well plates. The clones derived from HepG2 cells stably transfected with pIRES2-EGFP-DKK1 or pIRES2-EGFP vector were classified as DKK1 and Vector respectively; whereas the clones derived from Bel7402 cells stably transfected with pSIREN-Shuttle-siDKK1 or pSIREN-Shuttle-Control were classified as shDKK1 and shControl respectively. For β-catenin and MMP7 transfections, tumor cells transiently transfected with human beta-catenin pcDNA3 (plasmid 16828; Addgene, Cambridge, MA), pcDNA3 (Invitrogen), pCMV6-XL5-MMP7 and pCMV6-XL5 (OriGene, USA) using TurboFect™ in vitro Transfection Reagent for 36 h, then were applied to other expriements.

The MTT assay was used to detect the proliferation rate of tumor cells. Briefly, 2000 cells per well were plated in 96-well plates and incubated 1, 2, 3, 4, 5, 6 and 7d, respectively. At indicated time point, the process was performed as described before [18]. Briefly, 50 μl of MTT reagent (1 mg/mL) was added and incubated for 4 h at 37°C in a humidified incubator containing 5% CO₂. Supernatants were removed from the wells, and then 100 μl DMSO was added to solubilize the crystal products at room temperature for 10 min. The absorbance (OD) was measured with a microplate reader (Bio-Rad) at a wavelength of 570 nm.

For PCR, total RNA was extracted from sub-confluent using TRIzol reagent (Invitrogen). Two microgram of total RNA was subjected to DNase I digestion (1 U/μL, Fermentas, Hanover, MD) at 37°C for 30 min, and then the DNase I was heated inactivation at 70°C for 15 min, followed by reverse-transcription using PrimeScript™ Reverse Transcriptase (Takara). Semiquantitative RT-PCR was performed using primers listed in Additional file 1: Table S1. All PCR reactions were done using the following conditions: 95°C 5 min, 95°C 45 s, annealing at different temperatures for each gene respectively 45 s, extension 72°C 1 min for 30 cycles, and a final extension at 72°C for 10 min. All PCR products were separated by electrophoresis on 1.0% agarose gels.

The colony formation assay was performed as previously described with some modification [19]. Briefly, a total of 400 cells every well were seeded into a fresh 6-well plate and incubated in RPMI1640 containing 10% FCS, cell medium was changed every 3 d for 15 d until visible colonies formed. Colonies were fixed with methanol and stained with 0.1% crystal violet in 20% methanol for 20 min. Microscopic colonies composed of more than 50 cells were counted in each well.

The wound scratch assay was performed as described previously [15]. Briefly, cells were plated in a 24-well plate and grown to a confluence cell monolayer overnight prior to serum starvation for 24 h. After scratching with a pipette tip (20 μL); in order to eliminate the floating cells, culture medium was removed and wells were washed three times with PBS. The width of the wound was measured and recorded as t = 0. The cells were then allowed to migrated back into the wounded area, 24 h latter, the width of the open area was measured. Cell migration was expressed as the percentage of the gap (t = 24 h) relative to the primary width of the open area (t = 0 h). All experiments were performed in triplicate.

The migration and invasion assays were performed as in a 24-well Boyden chamber with 8 μm pore size polycarbonate membrane (Corning, NY, USA), as previously described [20]. For migration assay, 200 μL of serum-free medium (containing 1 × 10⁵ cells) was added to the upper compartment of the chamber, while the lower compartment was filled with 600 μL of RPMI 1640 supplemented with 10% FBS. After incubation at 37°C for 24 h, the tumor cells remaining inside the upper chamber were removed with cotton swabs. The cells on the lower surface of the membrane were stained with 0.1% crystal violet after fixation with methanol, and then counted under a light microscope. The invasion assay was done by the same procedure, except that the membrane was coated with Matrigel to form a matrix barrier, and 3 × 10⁴ tumor cells were added to the upper chamber.

The immunofluorescence staining was carried out as previously described with some changes [20]. Briefly, tumor cells were washed three times with ice-cold PBS, fixed in 4.0% paraformaldehyde. After permeabilization with 0.1% Triton X-100, cells were blocked of nonspecific binding by incubation with 5% normal goat serum in PBS. Afterwards, cells were incubated with the rabbit polyclonal β-catenin antibody overnight at 4°C. The staining was completed with Cy3-labeled goat anti-rabbit second antibody (1:500, Beyotime, China) incubation for 1 h at room temperature, and nuclei was counterstained with Hoechst33342 (10 μg/mL, Sigma, USA) for 30 min at dark. All photographs were taken by using High content screening (HCS) (Thermo Scientific Cellomics ArrayScan Vti, Cellomics, Inc., Pittsburgh, PA).

Western blot analysis was performed as our described previously [18]. Briefly, cell pellets were washed thrice with ice-cold phosphate buffered saline (PBS) and were lysed with RIPA buffer (Beyotime, China). The total protein concentration was determined using a BCA assay kit (Beyotime, China). Samples were denatured in 5 × SDS-sample buffer at 95°C for 5 min. Equal amounts of total proteins were separated using 12% SDS-PAGE, and then transferred onto PVDF membranes. Membranes were then blocked by 5% dried skimmed milk in PBST at room temperature for 1 h. After blocking, membranes were incubated with corresponding primary antibodies overnight at 4 °C. After washed three times with PBST, the membranes were incubated with appropriate HRP-conjugated secondary antibody, and then washed three times with PBST. Proteins were detected by using the ECL plus reagents (Beyotime, China).

Formalin-fixed and Paraffin-embedded tissue sections were de-paraffinized in xylene, rehydrated through graded ethanol, quenched for endogenous peroxidase activity in 0.3% hydrogen peroxide, and processed for antigen retrieval in citrate buffer (pH 6.0) for 10 min by microwave heating. Following overnight incubation at 4°C using rabbit mAb against human DKK1 (1:200) or β-catenin (1:100). Immunostaining was performed with an optimal amount of HRP-conjugated secondary antibody for 1 h at room temperature and DAB was used as chromogen. Subsequently, sections were counterstained with hematoxylin.

DKK1 and β-catenin expression were evaluated under a light microscope at a magnification of 200×. For each specimen, five images of representative areas were acquired and tumor cells were counted. For human samples, IHC scoring was performed using a modified Histo-score (H-score) as described by Fang et al.[21].

Data in this study were expressed as mean ± SEM from at least three separate experiments. Unless otherwise noted, the differences between groups were analyzed using Student’s t-test (two groups) or ANOVA (≥ three groups). All statistical analyses were performed using SPSS 13.0 software (Chicago, IL, USA). All tests employed were two-sided, and p <0.05 was set to be considered statistically significant.

To determine the possible role of DKK1 in human HCC, the levels of DKK1 mRNA in four different hepatocellular carcinoma cell lines were compared with that in the normal liver cell line by using RT-PCR. We observed that the mRNA expression of DKK1 in Huh7, SMMC7721, and Bel7402 cell lines were much higher in comparison with that in control normal QSG7701 hepatocytes (Figure 1A). We further investigated the protein expression of DKK1 in these cells by western blotting. As shown in Figure 1B, the higher protein level of DKK1 was confirmed in Huh7, SMMC7721, and Bel7402 HCC cell lines, compared with those of QSG7701 cells. Of note, Bel7402 hepatocellular carcinoma cells exhibited the highest level of DKK1 in both mRNA and protein level. In contrast, the level of DKK1 mRNA and protein in HepG2 cells was comparable to that in the normal liver cell line (QSG-7701). Therefore, HepG2 and Bel7402 cells were selected for further study.

Next, we asked for the functional role of DKK1 in tumor cells. After stably transfected cells were individually selected, the DKK1 and the housekeeping gene GAPDH mRNA levels were measured using RT-PCR. Compared with the cells which without transfection or transfected with Vector, the levels of DKK1 mRNA were greatly up-regulated in HepG2 cells transfected with DKK1 (Figure 2A). However, the tumor cells derived from shDKK1-tranfectants showed a much lower level of DKK1 mRNA in comparison with that derived from shControl or nontransfected Bel7402 cells (Figure 2B). Consistent with our RT-PCR data, Western blot analysis also revealed that DKK1 protein levels were also obviously increased in HepG2 cells that transfected with DKK1 (Figure 2C), while decreased in Bel7402 cells that transfected with shDKK1, compared with their control groups respectively (Figure 2D). The empty vector or the control scrambled shRNA did not substantially affect the endogenous DKK1 expression in both mRNA and protein levels. Collectively, these findings indicated that DKK1 can be stably overexpressed in HepG2 cells and shDKK1 could effectively suppress the expression of DKK1 in Bel7402 cells. Then we investigated whether DKK1 regulates cell proliferation. When DKK1 was up-regulated in HepG2 cells, no change of proliferation rates occurred compared with that without transfection or transfected with Vector (Figure 2E). Next, we aimed to investigate if loss of DKK1 may influence tumor cell proliferation. Therefore, DKK1 was depleted using one DKK1-specific shRNA and proliferation was assessed in Bel7402 cells, which expressed high levels of DKK1 endogenously. In these cells, remain no obvious change of proliferation rates were observed (Figure 2F). Using colony formation assay, although DKK1-transfectants displayed a tendency of increased colonies (Figure 2G), and shDKK1-transfectants showed a reduced tendency (Figure 2H), however, the difference did not reach statistical significance, compared with their control cells respectively.

Given that DKK1 did not influence tumor cell proliferation or colony formation, then we investigated whether DKK1 regulated HCC cell migration and invasion, which are two of the most important features of malignant cell behavior. First, we examined the role of DKK1 in tumor cell migration by using wound scratch assay. As illustrated in Figure 3A, enforced expression of DKK1 significantly promoted wound closure compared with that without transfection or transfected with Vector in HepG2 cells, vice versa, down-regulation of DKK1 remarkably attenuated the wound closure (Figure 3D). To further identify these observations, we examined the effects of DKK1 on HCC cell migration using the Boyden chamber transwell without Martrigel. Indeed, tumor cells derived from DKK1-transfectants displayed a higher ability of migration, compared with that derived from Vector or nontransfected cells (Figure 3B). In contrast, tumor cells derived from shDKK1-transfectants displayed a lower ability of migration (Figure 3E). Next, we determine the role of DKK1 in tumor cell invasion. We performed these experiments using transwell chamber with Matrigel. As illustrated in Figure 3C, up-regulation of DKK1 expression significantly enhanced the number of HepG2 cells that invaded through Matrigel in comparison with that derived from Vector or nontransfected cells, however, the number of Bel7402 cells that invaded through Matrigel was sharply reduced after DKK1 was silenced by a specific shRNA (Figure 3F). Consistently, in the human HCC tissues, the primary tumors showed slightly higher DKK1 expression, compared to those in the adjacent non-tumor liver tissues, moreover, the metastatic tumors showed much higher DKK1 expression in comparison with that in the primary tumors (Figure 3G and H). Collectively, these data demonstrate that DKK1 promotes HCC cell migration and invasion in vitro and tumor metastasis in vivo.

Then we investigate a potential mechanism for DKK1-mediated HCC cell migration and invasion. It is well known that DKK1 is a typical inhibitor of Wnt/β-catenin signaling pathway, hence we examined that whether DKK1 was able to inhibit the expression of β-catenin. Firstly, we analyzed β-catenin expression using Immunohistochemical staining in 71 cases of human HCC tissues. Of great interest, the DKK1 level was positively correlated with β-catenin expression (Figure 4A and B). Consistently, introduction of DKK1 did not inhibit β-catenin expression, but remarkably increased the mRNA and protein level of β-catenin in HepG2 cells, compared with those of control cells (Figure 4C). In addition, by using immunofluorescence staining, we also observed that up-regulation of DKK1 expression noticeably promoted the cytoplasmic/nuclear β-catenin accumulation in HepG2 cells (Figure 4D). Furthermore, dual-luciferase reporter analysis also showed that restoration of DKK1 significantly promoted the activity of firefly luciferase that carried wildtype but not mutant TCF binding sites (Figure 4E). In contrast, the blockage of DKK1 dramatically attenuated the mRNA and protein level of β-catenin in Bel7402 cells (Figure 5A). Moreover, the activity of firefly luciferase that carried wildtype but not mutant TCF binding sites was dramatically attenuated when DKK1 was silenced in Bel7402 cells (Figure 5B). These findings indicated that β-catenin may be an important target downstream of DKK1.

The role of β-catenin in DKK1-mediated phenotypes was then evaluated. First, we examined whether blockage of β-catenin could attenuated the effect of DKK1 expression. As expected, upon treatment with β-catenin inhibitor IWP-2, the promotive effect of DKK1 on migration and invasion was significantly attenuated in DKK1-transfectants, whereas there were no statistically significant differences in those of Vector-transfectants (Figure 4F and G). On the other hand, overexpression of β-catenin in shDKK1-transfectants restored the transcription of β-catenin (Figure 5C), and attenuated the inhibitory effect of shDKK1 on invasion (Figure 5D). As GSK3β is an important upstream target of β-catenin signaling, we further analyzed whether blockage of GSK3β could mimic the effect of β-catenin expression. Upon treatment with GSK3β inhibitor LiCl, the inhibitory effect of shDKK1 on migration (Figure 5E) and invasion (Figure 5F) was dramatically released.

It is well known that DKK1 antagonizes Wnt signaling by direct high-affinity binding to the extracellular domain of Wnt coreceptor lipoprotein receptor-related protein 6 (LRP6) and inhibiting the internalization of LRP6, which is responsible for the inhibition of GSK3β and activation of Wnt signaling cascade downstream. Therefore, two pairs of primers were selected to examine the mRNA expression of LRP6: one for the extracellular domain of LRP6, responsible for DKK1 binding, and another for the intracellular region, essential for inhibiting GSK3β [22]. Of great note, neither the extracellular domain nor the intracellular domain of LRP6 were changed, although the expression of β-catenin was dramatically altered upon DKK1 up-regulation (Figure 6A) or knockdown (Figure 6B). In addition, western blotting analysis also displayed that the protein level of LRP6 and GSK3β was not changed in comparison with those of control groups (Figure 6C and D). Whereas restoration of DKK1 significantly enhanced the expression of β-catenin, and the protein level of its downstream target c-Myc was also increased in HepG2 cells (Figure 6C). Vice versa, the blockage of DKK1 reduced the expression of β-catenin and c-Myc in Bel7402 cells (Figure 6D). Furthermore, upon β-catenin inhibitor IWP-2 treatment, the expression of β-catenin was attenuated (Figure 6E), which is consistent with the above observation on migration (Figure 4F) and invasion (Figure 4G), compared with the control. On the other hand, GSK3β inhibitor LiCl indeed recovered the expression of β-catenin in shDKK1-transfectants (Figure 6F), in agreement with the results of Figure 5E and F. These findings indicated that DKK1 may positively regulate the expression of β-catenin through a non-canonical Wnt signaling pathway.

Considering that MMP7 is one of the most important target genes downstream of β-catenin/TCF signaling, in addition, MMP7 has been shown to play a crucial role in promoting tumor cell migration and invasion [23]. We then investigate whether DKK1 regulated the expression of MMP7. As shown in Figure 7A and B, introduction of DKK1 significantly enhanced the mRNA and protein level of MMP7 in HepG2 cells (Figure 7A). Vice versa, when DKK1 was silenced by shRNA, the expression of MMP7 mRNA and protein were remarkably down-regulated in Bel7402 cells (Figure 7B). Moreover, upon treatment with β-catenin inhibitor IWP-2, the protein level of MMP7 was reduced in a dose-dependent manner (Figure 7C). Then we examined whether introduction of MMP7 could mimic the effect of β-catenin expression. As expected, HepG2 cells transiently transfected with MMP7 plasmids displayed a much invasive activity (Figure 7D and E), which phenocopied those of enhanced β-catenin expression. Taken together, our findings suggested that DKK1 promotes HCC cell migration and invasion at least partly by promoting β-catenin/MMP7 signaling.