HNF1α inhibition triggers epithelial-mesenchymal transition in human liver cancer cell lines

HepG2 and Hep3B cells were obtained from the American Type Culture Collection and were cultured in Dulbecco's Modified Eagle Medium with high glucose (Invitrogen) supplemented with 10% fetal calf serum, penicillin 100 IU/ml and streptomycin 100 μg/ml. SiRNA transfections were performed, as decribed previously [17], according to the manufacturer's protocol, in 6 well-plates using the lipofectamine RNAiMax reagent (Invitrogen) with siRNA duplexes targeting HNF1A (NM_000545) (Ambion) with sequence: GGUCUUCACCUCAGACACUtt (exon 8-9 3544). Block-iT Alexa Fluor Red Fluorescent Oligo siRNA (Invitrogen) was used as a double-stranded RNA negative control. In most experiments 10 nM of each siRNA was transfected in triplicate, except for dose-effect studies, where several siRNA concentrations were tested (0, 0.01, 0.05, 0.1, 0.2, 0.4, 0.6, 0.8, 1, 5, 10 and 50 nM) in order to obtain different levels of HNF1A expression. Cells were prepared for analyses either 3 or 7 days after transfection for HepG2 cells but only after 3 days for Hep3B cells, because HNF1α inhibition could not be maintained until 7 days in this cell line. The absence of cross-reaction of the HNF1α-siRNA duplexes with the HNF1B sequence was checked by comparing the expression level of HNF1A transcript in cells transfected with siRNA targeting HNF1A with the control siRNA-transfected cells.

Quantitative RT-PCR (qRT-PCR) was performed in duplicate as previously described [19] using pre-designed primers and probe sets from Applied Biosystems (Additional file 1). Ribosomal 18S (R18S) was used for the normalization of expression data and the 2^-ΔΔCT method was applied. The final results were expressed as the fold differences of target gene expression in HNF1α siRNA compared with control siRNA in cell lines or in tested samples compared with the mean expression value of normal tissues for tumor analysis.

Western blot analyses were performed as previously described [18] using the primary antibodies specific for E-Cadherin (Cell Signaling Technology, diluted 1:100), HNF1α, Vimentin and N-Cadherin (Santa Cruz Biotechnology, 1:500, 1:200 and 1:200); Polyclonal rabbit anti-actin (Sigma, 1:3000) was used as loading control.

Cells were grown on slides for 3 or 7 days and fixed with 4% formaldehyde in phosphate-buffered saline (PBS) 1X for 15 min. After washing with PBS, cells were permeabilized with 0.1% triton for 15 mn, washed with PBS, then, cells were incubated with primary antibody overnight. After three washes with PBS, cells were incubated with secondary antibodies for 1 h. The slides were washed, then mounted with VECTASHIELD^® Mounting Medium with DAPI (Vector Laboratories). Immunofluorescence images were obtained using a Carl Zeiss Axiophot microscope. All images within one experiment were collected using 63x objective and the same exposure time. The antibodies used were: rabbit anti-E-cadherin (Santa Cruz Biotechnology, 1:100), rabbit anti-N-cadherin (Santa Cruz Biotechnology, 1:100), rabbit anti-Fibronectin (Sigma, 1:100), and the secondary antibodies were anti-mouse and anti-rabbit (GE Healthcare, 1:100, 1:100). Actin was stained by incubating cells for 1 h with Alexa Fluor 488 phalloidin (Molecular Probes, 1:300).

Boyden chamber migration assays were performed 72 h after transfection using 24-well migration inserts (BD Biosciences). 1,5 × 10⁵ cells were plated in the upper chamber of the migration insert and they were left to migrate towards medium with serum for 16 h. Cells on the upper side of the insert membrane were removed with a cotton swab, whereas cells that had migrated to the underside of the insert membrane were fixed with 4% formaldehyde in phosphate-buffered saline (PBS) for 15 min. After washing with PBS, cells were permeabilized with 0.1% triton for 15 min, washed with PBS, and stained with hematoxylin. Cells were counted under 300x magnified field, 10 fields were counted for each condition and each condition was done in triplicates.

HepG2 cells were seeded and transfected in 6-well plates at the density of 5 × 10⁵ cells per well. After 48 h, a scratch was made through confluent cells with a pipette tip and cells were washed with PBS, and medium without serum was added. Picture&+s were taken just after the scratch was made and at 24, 48 and 72 h afterwards, to monitor cell movements. The experiment was reproduced three times.

HepG2 cells transfected with Control or HNF1α siRNA for 3 days in glass-bottom dishes were imaged using 20x objective and Biostation IM at Nikon Imaging Centre at Institut Curie, Paris. Cells were incubated overnight (during 16-18 h) in the Biostation IM and Images were collected every 10 minutes during 16-18 h. The experiment was repeated three times. Data were analysed using MetaMorph image analysis software.

Liver tissues were collected in nine French surgery departments from 1992 to 2004. They were immediately frozen in liquid nitrogen and stored at -80°C until used for molecular studies. The whole series of HCA used for the different molecular analyses included 35 H-HCA previously described [8, 9, 17], and 23 normal livers taken from patients resected with primary liver tumors developed in the absence of cirrhosis. All the patients were recruited in accordance with French law and institutional ethical guidelines. The study was approved by the ethical committee of Hôpital Saint-Louis, Paris, France.

All the values reported are mean ± SD. Statistical analyses were performed using GraphPad Prism version 5 software and significance was determined using either the nonparametric Mann-Whitney test for unpaired data or the two-tailed t-test. Difference was considered significant at P < 0.05. In all graphs, *, **, *** indicate difference between groups at P < 0.05, 0.01 and 0.001, respectively.

HepG2 et Hep3B cell lines were transfected with siRNA targeting exon 8-9 of HNF1A or control siRNA, as previously described [17]. HNF1A mRNA inhibition reached 98% and was maximal at 72 h after transfection, as well as the expression of its transactivated gene FABP1 [17]. Silencing of HNF1α lasted until 7 days in HepG2, but was not maintained beyond 3 days in Hep3B. Expression of HNF1α homologue, HNF1β, was not diminished by HNF1α siRNA at 24 and 48 h after transfection, assessing that HNF1α siRNA did not target HNF1β mRNA (Additional File 2A).

Cells transfected with HNF1α siRNA had a different phenotype from cells transfected with control siRNA. On phase contrast microscopy, they looked elongated and had lost cell-cell contacts (Figure 1A). This phenotype was maintained until at least 7 days after transfection in HepG2 cells. Phalloidin labelling revealed reorganized actin cytoskeleton with development of actin structures looking like lamelipodia and filopodia in both cell type (Figure 1B). Time-lapse microscopy of HepG2 cells transfected with HNF1α siRNA showed that the cytoplasmic protrusions observed in those cells were dynamic structures protruding from the cell (Figure 1C).

Expression of albumin, a liver-specific gene, and of transcription factors involved in hepatocyte differentiation, assessed by quantitative RT-PCR (qRT-PCR), was diminished 3 days after transfection in both cell type, and was maintained low until at least 7 days after transfection in HepG2 (Figure 1D and 1E). Particularly, HNF4α expression, which has been shown to be regulated by HNF1α [20, 21], was decreased early after transfection and this decrease was strongly correlated to HNF1α expression, which was modulated by using several concentrations of siRNA (Additional files 2A and 2B). These results revealed dedifferentiation of cells transfected with HNF1α siRNA.

Epithelial-mesenchymal transition (EMT) is defined by loss of epithelial cell polarity, disappearance of differentiated junctions, reorganization of the cytoskeleton and changes in migration abilities [22‐25]. During this process, epithelial markers such as E-cadherin are under expressed and mesenchymal markers are over expressed. In HepG2 cells transfected with HNF1α siRNA, E-cadherin is strongly under expressed at the transcription level as well as at protein level (Figure 2A and 2B). Immunostaining of E-cadherin showed presence at cell-cell junctions in control-siRNA-transfected cells whereas cells transfected with HNF1α siRNA showed no staining at cell borders, suggesting loss of adherens junction in those cells (Figure 2C). Interestingly, the decrease of E-cadherin mRNA was significantly correlated to HNF1α mRNA decrease, when it was modulated using a range of siRNA (Additional file 3). Moreover, zonula occludens-1 (ZO-1), a tight-junction protein, was also under expressed at transcriptional level (Figure 2A). In HNF1α-inhibited HepG2 cells, the mesenchymal markers vimentin and fibronectin were over expressed both at RNA and protein levels (Figure 2A, B and 2C). Several proteins involved in bassement membrane degradation, metalloproteinases (MMP) 2, 3 and 9, were also over expressed in HepG2 cells transfected with HNF1α siRNA (Figure 2A). These characteristics of EMT were observed at 3 days after transfection and were mostly maintained until 7 days.

In Hep3B cell line, epithelial markers were also under expressed and E-cadherin staining was not found at cell border in cells transfected with HNF1α siRNA (Figure 2D, E and 2F). However, the mesenchymal markers over expressed in Hep3B were not the same than in HepG2 cell line. Vimentin and fibronectin remained unchanged whereas N-cadherin was up regulated at RNA and protein levels in Hep3B cells transfected with HNF1α siRNA (Figure 2D and 2E). Overexpression of N-cadherin was not obvious by immunofluorescence analysis, but N-cadherin was mostly found at cell borders and cell-cell contacts were diminished in HNF1α-siRNA-transfected cells (Figure 2F). Finally, metalloproteinase 9 was also significantly over expressed in Hep3B cells transfected with HNF1α siRNA (Figure 2D).

Overall, liver cancer cells transfected with HNF1α siRNA lost expression of epithelial and tight junction markers and over expressed proteins usually expressed in mesenchymal cells, defining an epithelial-mesenchymal transition in those cells.

Several transcription factors have been involved in the establishment of epithelial-mesenchymal transition, and in particular, in the repression of E-cadherin expression. These transcription factors are usually up regulated during EMT [22‐24]. Among these proteins, the Snail family members (Snail1 and Snail2, also known as Snail and Slug) play a key role in EMT. Snail1 was up regulated in HepG2 cells transfected with HNF1α siRNA compared with control siRNA, at 3 days after transfection and until 7 days (Figure 2A). Snail2 was slightly under expressed at 3 days after transfection but it was importantly over expressed at 7 days. The transcription factors of the ZEB family, and particularly ZEB2, were over expressed in HepG2 cells transfected with HNF1α-siRNA at 3 and 7 days after transfection (Figure 2A). Up regulation of all these transcription factors was also observed in Hep3B cells, along with the overexpression of Twist1, another transcription factor involved in EMT (Figure 2D).

We performed several experiments to assess the ability of migration of HNF1α-inhibited HepG2 cells. First, we put the cells transfected with HNF1α siRNA deprived with serum in a migration insert and let them migrate four 16 h towards medium with serum. More cells were able to migrate when they where transfected with HNF1α siRNA than with control siRNA (Figure 3A).

In a wound healing assay, the scratch caused in cells tranfected with control siRNA did not close completely, even after 72 h (Figure 3B). In HepG2 cells transfected with HNF1α siRNA, the wound did not close completely either but HNF1α-inhibited cells were able to move at the center of the wound unlike control cells (Figure 3B).

Those results showed that HepG2 cells transfected with HNF1α siRNA developed greater migration abilities than control cells.

Many proteins can trigger epithelial-mesenchymal transition [23, 24]. Among them, we found that TGFβ1 was over expressed in HepG2 and Hep3B cells transfected with HNF1α siRNA (Figure 4A and 4B). Moreover, the overexpression of TGFβ1 mRNA was inversely correlated to HNF1α expression in HepG2 cells (Figure 4C). We then studied the transcriptomic expression of two genes that are known to be induced by TGFβ/Smad pathway: SMAD7, an inhibitor of TGFβ pathway that is Smad-regulated and is induced by TGFβ in an early response [26, 27], and TGFβ-induced (TGFBI), an extracellular matrix protein which plays a role in cell-collagen interactions [28]. SMAD7 and TGFBI were up-regulated at 3 and 7 days after transfection in HepG2 (Figure 4A) and in Hep3B cell lines (Figure 4B). These results suggest that TGFβ pathway is activated in HNF1α-inhibited cells and could participate to the EMT observed in these cells.

Interestingly, we found an overexpression of TGFβ1 in H-HCA compared to normal livers by quantitative RT-PCR (Figure 4D). But we couldn't find any proof of TGFβ pathway activation in these tumors, since neither SMAD7 nor TGFBI were over expressed, nor any other known TGFβ pathway target genes (Figure 4D, E and data not shown).