Figure 1.
Identification of functional IR1 site(s) in NDRG2.
A, Schematic representation of the NDRG2 gene with the IR1 site(s) located in the first intron. Human, mouse and rat intronic sequences with IR1 elements (in bold) are shown below (IR1 elements start at positions +1560, +757 and +1394 respectively). A version of the human NDRG2 response element with mutated IR1 sites is shown below. B, Luciferase reporter assays from HEK 293 cells co-transfected with pTReX-Dest30-hFXRα2 and a luciferase reporter plasmid containing either one copy of the wild type IR1 element (IR1) or the mutated element (IR1mut) from the human NDRG2 gene or 2 copies of the human IBABP-RE (IBABP). The fold induction in presence of 0.5 µM Px20350 (Px) versus DMSO control (D) of firefly luciferase activities (normalized against renilla luciferase) of triplicate experiments is shown. C, Amplification of a 196 bp DNA fragment encompassing the IR1 sequences from chromatin prepared by ChIP from HepG2 or HepG2-FLAG-FXR cells. PCR reactions from input DNA and DNA derived from the immunoprecipitation with the M2 anti-FLAG antibody are indicated D, Localization of murine FXRα and RXRα by ChIP-seq to a site corresponding to the predicted IR1 recognition element (star) in the first intron of the mouse Ndrg2 gene on chromosome 14 in mouse liver. The FXRα and RXRα UCSC genome browser tracks shown for the NDRG2 gene were retrieved from genome wide ChIP-seq datasets published by Thomas et al. [22] and Boergesen et al. [41], respectively. E, Chemical structures of FXR agonists. Among different structural changes, the unfavourable stilbene linker present in GW4064 was replaced by an amino-methylene linker in Px20350 [40]. The further modified Px20606 [53] yielded improved in vivo pharmakokinetic properties compared to Px20350. A characterization of FXR agonists in in-vitro assays is presented in Table 1.
Figure 2.
Induction of NDRG2 in human hepatoma cell lines.
A, Ectopic expression of FXR in HepG2-FXR5 compared to HepG2 parental cells. Positions of bands representing FXR and GAPDH proteins in the western blots from aliquots of whole cell extracts are indicated. B, Dose dependent induction of NDRG2 mRNA in HepG2 and HepG2-FXR5 cells by the FXR agonist PX20350. HepG2 or HepG2-FXR5 cells were grown for 18 h in medium containing either DMSO, 0.5 µM of PX20350 or 0.5 µM of PX20606 in 96 well plates as indicated and relative fold induction of NDRG2 mRNA over the DMSO control was determined by RT-qPCR using the ΔΔCt method. C, Induction of NDRG2 protein by the FXR agonists Px20350 and PX20606 in HepG2-FXR5 cells. HepG2-FXR5 cells were grown to a density of around 60% and further cultivated for 18 h in medium supplemented with either DMSO as a vehicle, 0.5 µM PX20350 or 0.5 µM PX20606 as indicated and aliquots of whole cell extracts analysed by western blots. Relative positions of size marker protein bands and the position of NDRG2 and GAPDH proteins are indicated. D, Stable overproduction of FXR in HuH-7-37 compared to HuH-7 cells. Positions of bands representing FXR and GAPDH proteins analysed by western blots from aliquots of whole cell extracts are indicated. E, Dose dependent induction of NDRG2 mRNA in HuH7 and HuH7-37 cells by FXR agonist Px20350. HuH7 or HuH7-37 cells were grown for 18 h in medium containing either DMSO or increasing concentrations of PX20350 in DMSO in 96 well plates and relative fold induction over DMSO control of NDRG2 mRNA was determined by RT- qPCR using the ΔΔCt method. F, NDRG2 is a direct transcriptional target of FXR. HepG2-FXR5 cells, grown in quadruplicates in 12 well plates at a density of around 60%, were further incubated for 4 hours in growth medium supplemented with either DMSO, 0.5 µM PX20350 (PX) or 0.5 µM PX20350 together with 10 µM Cycloheximide (CHX). The relative fold induction of NDRG2- and SHP- mRNAs over DMSO as control were determined by RT- qPCR using the ΔΔCt method and a Student's t-test was performed (*** = p<0.001).
Table 1.
Activities of the different FXR agonists in different assay systems.
Figure 3.
Ndrg2 mRNA is induced in wt mice by FXR agonists, reduced in livers of FXR−/− versus wt mice and both FXR and NDRG2 mRNAs are reduced in human HCC versus normal liver.
A, Ndrg2 mRNA induction in livers of C57Bl6 mice by FXR agonists. C57Bl6 mice on normal chow received a bolus of vehicle (n = 9), or 30 mg/kg of Px20606 or 30 mg/kg of GW4064 in vehicle (n = 6 each) by oral gavage. Mice were sacrificed after 4 h and liver RNA isolated for RT-qPCR analysis. Ndrg2 and Shp mRNA expression levels were calculated according to the ΔΔCt method using normalization to Tbp as a house keeping transcript and the mean level of Ndrg2 and Shp expression in vehicle treated mice were set to 1.0. Data are presented as the mean ± standard error of the mean (±SEM). Statistical significance in a nonpaired Student's t-test: *, ** and *** = p<0.05, <0.01 and <0.001. B, Quantitative determination of mRNA's in livers of wild type (wt) and FXK−/− (KO) mice. Relative mRNA levels of Ndrg2, Shp, Cyp7a1 and Cyclophilin E were determined by RT-qPCR from wt (n = 5) and FXR−/− (n = 5) mice as indicated. Tbp-normalized relative mRNA levels of wt mice were set to 1.0 respectively. C, Decreased expression of FXR and the FXR target genes NDRG2 and SHP in human HCC. FXR, NDRG2 and SHP mRNA levels were quantified by RT- qPCR in 8 normal, 34 HCC from different stages (7 samples stage I, 8 samples of stages II and IIIA, 3 samples of stage IV) and 12 non-HCC liver disease (LD) derived cDNA's. Relative folds of mRNA expression levels between normal livers and the respective stages of HCC or non-HCC liver disease were calculated according to the ΔΔCt method using normalization to TBP. The mean levels of FXR, NDRG2 and SHP expression in normal livers were set to 1.0 respectively and relative mRNA levels are expressed as mean ± SEM. Statistical significance in a nonpaired Student's t-test: *, P<0.05; **, P<0.01; ***, P<0.001.
Figure 4.
In vitro characterization of SK-Hep-1 and SK-GI-18 cells.
A, Expression of FXR in SK-Hep-1 and the FXR overexpressing SK-GI-18 cells. The signals in western blot analysis from aliquots of whole cell extracts representing FXR and GAPDH are indicated. B, Increased expression and inducibility of NDRG2 mRNA by Px20350 in SK-GI-18 cells compared to parental SK-Hep-1 cells. SK-Hep-1 cells and SK-GI-18 cells were grown in 24-well plates (n = 4) in presence of DMSO or 0.5 µM Px20350 for 72 hours. Total RNA was isolated from cells at around 60% confluency. The diagram shows RT-qPCR data for NDRG2 mRNA drawn as relative fold increase over the TBP corrected values for DMSO treated SK-Hep-1 cells using the ΔΔCt method. Statistical significance in a nonpaired Student's t-test: *, ** and *** = p<0.05, <0.01 and <0.001. C, Influence of PX20350 on proliferation of SK-Hep-1 and SK-GI-18 cells in vitro. SK-Hep-1 or SK-GI-18 cells were seeded in a 96 well plate and grown for 4 days in presence of DMSO or increasing concentrations of FXR ligand Px20350. Relative proliferation was measured by flurescence determination of DNA content using the CyQuant Direct Proliferation assay. D. Inhibition of proliferation by PX20606 is blunted in SK-GI-18-sh3-14 cells stably expressing shRNA directed against NDRG2. SK-Hep-1, SK-GI-18 NC-16 expressing a negative control shRNA and SK-GI-18-3-14, expressing an NDRG2 shRNA were seeded at 3000 cells/well in a 96 well plate and grown for 3 days in presence of DMSO or increasing concentrations of PX20606. Relative proliferation was measured by flurescence determination of DNA content using the CyQuant Direct Proliferation assay. E, Migration of SK-GI-18 and SK-Hep-1 cells in presence or absence of Px20350 (1 µM). Cells were seeded at 7000 cells per 96 well (in quadruplicates) of a Oris Fibronectin Coated Plate (Platyplus) and seeding stoppers removed after 14 h and cells grown in medium with DMSO as a vehicle or 1 µM Px20350 in DMSO for another 72 hours before quantification of migrated cells using the Cyquant Direct Cell Proliferation Assay.
Figure 5.
Time course of tumor growth and metastasis.
A and B, Nude mice were inoculated with either 5×106 SK-Hep-1 (A) or SK-GI-18 cells (B), randomized into groups and animals were treated by daily gavage starting on day 4 with vehicle, Px20606 (10 mg/kg/d) or Sorafenib (100 mg/kg/d) as indicated. In vivo imaging pictures for representative animals from each group that received SK-Hep-1 (A) or SK-GI-18 (B) cells at the indicated days are displayed. C, Time course of tumor development in the liver. The in vivo fluorescence intensities at day 7 were set to 100% for each group and data plotted for data gathered at days 7, 14, 21, 28, 35 and 56. The sharp rise of fluorescence in the SK-Hep-1/Px20606 group from day 35 to day 56 is due to sharp increases in tumor sizes in just three individual animals out of a total of nine animals (fluorescence data from these animals on day 56 are circled in Fig. 6A).
Figure 6.
The effect of treatments on tumor cell growth in liver or lymph nodes.
A, Tumor load in the liver (primary tumor) at the end of the experiment on day 56. Scatter plots of fluorescence intensities with horizontal bars as mean are shown. Group sizes at beginning of the experiments were 9 animals each, with the exception of SK-Hep-1/Sorafenib (n = 7) and SK-GI-18/Sorafenib (n = 8). For statistical analysis non-paired Student's t-tests were performed as indicated (ns = >0.05; * = <0.05; ** = <0.01; *** = <0.001). Three animals in the SK-Hep-1/Px20606 group showed a sharp increase in fluorescence from day 35 to day 56 and are highlighted by a circle. B, Lymphnode metastases at the end of the experiment on day 56. Scatter plots of fluorescence intensities localized to lymph nodes with horizontal bars as mean are shown. Group sizes at beginning were 9 animals each, with the exception of SK-Hep-1/Sorafenib (n = 7) and SK-GI-18/Sorafenib (n = 8). For statistical analysis non-paired Student's t-tests were performed as indicated (ns = >0.05; * = <0.05; ** = <0.01; *** = <0.001). C, Combination of PX20606 and Sorafenib does not further reduce growth of tumors derived from SK-GI-18 cells in the liver. Nude mice were injected with 5×106 SK-GI-18 cells and in vivo imaging performed on day 4 before randomization of animals into the four treatment groups vehicle (V), 10 mg/kg/d PX20606 (P), Sorafenib 50 mg/kg/d (S) or 10 mg/kg/d PX20606+Sorafenib 50 mg/kg/d (P+S). Scatter plots of fluorescence intensities in the liver at termination of the experiment on day 51 with horizontal bars as mean are shown. Group sizes were 9 animals each for vehicle and Sorafenib and 10 animals each for PX20606 and PX20606+Sorafenib. For statistical analysis non-paired Student's t-tests against were performed as indicated (ns = >0.05; * = <0.05; ** = <0.01;*** = <0.001). D, Combination of PX20606 and Sorafenib further reduces metastasis into lymph nodes. The same animals as in Fig. 6C were analysed for lymph node metastases. Scatter plots of fluorescence intensities at sites of lymph nodes with horizontal bars as mean are shown. Group sizes were 9 animals each for vehicle and Sorafenib and 10 animals each for PX20606 and PX20606+Sorafenib. For statistical analysis non-paired Student's t-tests against were performed as indicated (ns = >0.05; * = <0.05; ** = <0.01; *** = <0.001).
Table 2.
Number of animals with metastases in lymph nodes (A) and bone (B) and the total numbers of animals in each group surviving till day 56 (in parentheses) are shown.