If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Department of Pathology, Bicêtre University Hospital, University of Paris-Saclay, Assistance Publique-Hôpitaux de Paris, Le Kremlin-Bicêtre, F-94275, France
Dual ß-catenin knocked-down HepG2 model allows studying independently structural and transcriptional activities of β-catenin
•
WT and mutated ß-catenins play antagonistic functions in hepatoblastoma cells
•
Fascin-1 is a target of mutated ß-catenin in tumor hepatocytes
•
Fascin-1 expression is high in ß-catenin-activated undifferentiated tumors in mice
•
Fascin-1 expression is a marker of immature cells in human hepatoblastomas
Abstract
BACKGROUND & AIMS
ß-catenin is a well-known effector of the Wnt pathway, and a key player in cadherin-mediated cell adhesion. Oncogenic mutations of ß-catenin are very frequent in pediatric liver primary tumors. Those mutations are mostly heterozygous, which allows the co-expression of wild-type (WT) and mutated ß-catenins in tumor cells. We investigated the interplay between WT and mutated ß-catenins in liver tumor cells, and searched for new actors of the ß-catenin pathway.
METHODS
Using an RNAi strategy in ß-catenin-mutated hepatoblastoma (HB) cells, we dissociated the structural and transcriptional activities of β-catenin, which are carried mainly by WT and mutated proteins, respectively. Their impact was characterized using transcriptomic and functional analyses. We studied mice that develop liver tumors upon activation of ß-catenin in hepatocytes (APCKO and ß-cateninΔexon3 mice). We used transcriptomic data from mouse and human HB specimens, and used immunohistochemistry to analyze samples.
RESULTS
We highlighted an antagonistic role of WT and mutated ß-catenins with regard to hepatocyte differentiation, as attested by alterations in the expression of hepatocyte markers and the formation of bile canaliculi. We characterized Fascin-1 as a transcriptional target of mutated ß-catenin involved in tumor cell differentiation. Using mouse models, we found that Fascin-1 is highly expressed in undifferentiated tumors. Finally, we found that Fascin-1 is a specific marker of primitive cells including embryonal and blastemal cells in human HBs.
CONCLUSIONS
Fascin-1 expression is linked to a loss of differentiation and polarity of hepatocytes. We present Fascin-1 as a previously unrecognized factor in the modulation of hepatocyte differentiation associated to ß-catenin pathway alteration in the liver, and as a new potential target in HB.
LAY SUMMARY
FSCN1 gene, encoding Fascin-1, was reported to be a metastasis-related gene in various cancers. Herein, we uncover its expression in poor-prognosis hepatoblastomas, a pediatric liver cancer. We show that Fascin-1 expression is driven by the mutated beta-catenin in liver tumor cells. We provide new insights on the impact of Fascin-1 expression on tumor cell differentiation. We highlight Fascin-1 as a marker of immature cells in mouse and human hepatoblastomas.
CG and SS were supported by postdoctoral fellowships from, respectively, La Ligue Nationale contre le Cancer and the Fondation pour la Recherche Médicale. L. Piquet and L. Paysan were supported by PhD fellowships from respectively the SIRIC BRIO and the Région Aquitaine. L. Dif is supported by a PhD scholarship from the French Ministry of Research (MENESR). This work was supported by grants from La Ligue contre le Cancer (comité régional) (to VL), La Ligue Nationale contre le Cancer “Equipe labellisée 2016” (to VM and FS), and from Institut National du Cancer (PLBIO-INCa2014-182 to VM and TRANSLA-INCa2013-209 to CFG).
Author Contributions
Study design: VM, Generation of experimental data: CG, SS, VN, LD, LP, LP, TR, NDS, RL, NAC, DD, VL. Analysis and interpretation of data: CG, SS, LD, DD, PBS, BLB, VL, SC, VM. Providing biological samples from HB patients: CFG, AR. Drafting of the manuscript: FS, VL, SB, VM.
Data availability statement
The transcriptomic data are archived in the public GEO data repository under the GEO accession number GSE144107.
Introduction
ß-catenin is an evolutionary conserved protein that plays a dual role in cells. It is the key effector of the canonical Wnt pathway, acting as a transcriptional cofactor with the lymphoid enhancer factor/T-cell factor (LEF/TCF)
. In addition, ß-catenin plays a central role in cadherin-mediated cell adhesion. In epithelial cells, in the absence of Wnt signaling, ß-catenin is associated to E-cadherin at cell-cell junctions, and is maintained at low cytoplasmic levels through its destruction complex of Axin, adenomatosis polyposis coli (APC), and glycogen synthase 3ß kinase (GSK-3ß). GSK-3ß phosphorylates ß-catenin, causing its degradation by the proteasome. The destruction complex is disrupted by the Wnt signal, whereupon the cytoplasmic stabilized ß-catenin translocates in the nucleus, where it drives the transcription of target genes. Thus, ß-catenin is endowed with two main functions: a structural function at cell-cell adhesion, and a transcriptional function in the nucleus. An imbalance in the signaling properties of ß-catenin may lead to deregulated cell growth, adhesion and migration, resulting in tumor development and metastasis. However, given the dual function of ß-catenin, it is difficult to distinguish whether the it is specifically the structural function or the transcriptional function of ß-catenin, which is involved in cellular processes.
In the liver, the Wnt/ß-catenin pathway plays a number of important roles in the regulation of embryonic and postnatal development, zonation, metabolism and regeneration
. This pathway is also strongly involved in hepatocarcinogenesis. Its aberrant activation arises from mutations in the CTNNB1 gene encoding ß-catenin or in components of the degradation complex, such as AXIN and APC. CTNNB1 alterations are identified in up to 80% of human hepatoblastomas (HBs), the primary hepatic malignancy in children, and in 30 to 40% of hepatocellular carcinomas (HCCs)
, molecular analyses nonetheless described two subtypes: the fetal C1 subtype with favorable outcome, and the proliferative poorly differentiated C2 subtype characterized by an intense nuclear staining of ß-catenin
. The majority of ß-catenin mutations affect exon3 at GSK3ß phosphorylation sites, constitutively activating the Wnt pathway. Interstitial deletions in this exon are highly prevalent in HBs, while point mutations are more common in HCCs
. Interestingly, most of mutations in exon3 are monoallelic, leaving a wild-type (WT) allele in tumor cells. We therefore attempted to address the interplay between WT and mutated ß-catenins in liver tumor cells. To do so, we used the human HB HepG2 cell line, which exhibits a heterozygous deletion of 348 nucleotides in exon3, resulting in an abundant truncated form of ß-catenin and a smaller amount of WT ß-catenin
. The large deletion removes both the GSK-3ß phosphorylation sites and the binding site for α-catenin, which is the ß-catenin partner in the E-cadherin-mediated cell adhesion. Thus, both WT and mutated (Δaa25-140) forms of ß-catenin co-exist in these cells. The interplay between the two has never been explored. We therefore designed an RNA interference approach to specifically knock-down WT and/or mutated ß-catenin, and to address their reciprocal role in hepatocyte differentiation. Using this approach, we dissociated the structural and transcriptional activities of β-catenin, which are carried mainly by the WT and the mutated proteins, respectively. Transcriptomic and cellular analyses revealed that both functions play an antagonistic role in tumor hepatic cell differentiation. For the first time, molecular characterization revealed FSCN1 (fascin-1) as a target of ß-catenin, controlling tumor hepatocyte differentiation state and proliferation. We also found that fascin-1 expression is associated to undifferentiated ß-catenin-mutated tumors in mice, which are close to human HBs. Using human samples, we showed that fascin-1 is specifically expressed in the embryonal component of HBs. Thus, we described FSCN1 as a ß-catenin target gene associated with hepatic tumors of poor outcome, such as poorly differentiated HBs.
Materials and Methods
Cell culture
Human HB cell line, HepG2, and human HCC cell lines, Hep3B, Huh7 and SNU398, were purchased from American Type Culture Collection (Supplementary CTAT Table). HB cell line Huh6 was generously provided by C. Perret (Paris, France). All cell lines were cultured as previously described
. Cell line authentication was performed using short tandem repeat analysis, and absence of mycoplasma contamination in cell culture media was tested every week.
Mouse samples
We collected tumoral and non tumoral livers from mouse transgenic models with hepatic β-catenin activation. All animal procedures were approved by the ethical committee of Université de Paris according to the French government regulation. The Apcfs-ex15 and β-cateninΔex3 mouse tumors were obtained from compound Apcflox/flox/TTR-CreTam and β-cateninex3-flox/flox respectively, injected with 0.75 mg Tamoxifen or with 5 x 108 ip Cre-expressing adenovirus, as previously described
. All the mice were maintained at the animal facility with standard diet and housing. They were followed by ultrasonography every month until tumor detection, thereafter ultrasound imaging was continued every 2 weeks. Table S1 described the cohort of mice used for this study.
Patient samples
Liver tissues were immediately frozen in Isopentane with Snapfrost and stored at −80 °C until used for molecular studies. Samples were obtained from the Centre de Ressources Biologiques (CRB)-Paris-Sud (BRIF N°BB-0033-00089) with written informed consent, and the study protocol was approved by the French Government and the ethics committees of HEPATOBIO (HEPATOBIO project: CPP N°CO-15-003; CNIL N°915640). Liver samples were clinically, histologically, and genetically characterized (Table S2). Among the 20 cases, 12 were classified as C1, 3 as C2A and 5 as C2B in a C. Grosset’s previous study
Data were reported as the mean ± SEM of at least three experiments. Statistical significance (P < 0.05 or less) was determined using a Student’s t-test or analysis of variance (ANOVA) as appropriate and performed with GraphPad Prism software. P values are indicated as such: * P < 0.05; ** P < 0.01; *** P < 0.001; **** P <0.0001; ns, non significant.
Results
The dual ß-catenin knocked-down HepG2 model
An exon3 region is deleted on one allele of the CTNNB1 gene in HepG2 cells
. A high amount of a truncated form of 76kDa is therefore co-expressed with a 92kDa full-length ß-catenin (Fig. 1). We designed small interfering RNAs (siRNAs), named “sißcat-WT”, “sißcat-mut” and “sißcat-both”, to respectively and specifically target the WT, the mutated form of ß-catenin, or both (Fig. 1A). Each siRNA efficiently knocked-down their targeted protein (Fig. 1B and S1A). At the mRNA level, sißcat-WT and sißcat-mut reduced the amount of ß-catenin transcripts by half. As expected, sißcat-both led to a near full elimination of ß-catenin in HepG2 cells (Fig. 1C).
Fig. 1The HepG2 dual ß-catenin KD model. (A) Schema showing the different siRNAs used. (B-E) HepG2 cells were transfected with indicated siRNAs. (B) Protein extracts were analyzed using anti-ß-catenin and GAPDH antibodies. Note that due to the W25-I140 deletion, mutated ß-catenin migrates faster and is more abundant than the WT ß-catenin. (C) ß-catenin mRNA expression was analyzed by qRT-PCR. (D) Promoter activity was evaluated by luciferase reporter assays with TCF responsive reporter. Shown is the mean relative luciferase activity, normalized to Renilla luciferase and compared to siRNA control transfected cells. (E) mRNA levels of indicated genes were analyzed by qRT-PCR. Shown is the relative mRNA level compared to control transfected cells. (C-E) Each graph shows the quantification of three independent experiments. Error bars indicate s.e.m (n=3). * P < 0.05; ** P < 0.01; *** P < 0.001; **** P < 0.0001 by one way ANOVA. (F) HepG2 cells transfected with indicated siRNAs were seeded in 96-well plates and total biomass, reflecting the number of cells, was assayed every day. Each time point was performed in triplicates. Error bars indicate s.e.m (n=3). **** P < 0.0001 by one way ANOVA. (G) HepG2 cells transfected with indicated siRNAs were fixed and stained with anti-E-Cadherin antibodies (green) and Hoescht (blue). Scale bar: 10 μm.
Using the TOP-flash reporter system, we characterized the transcriptional activity of ß-catenin relative to the silencing of both alleles. Whereas silencing the WT allele did not impact TCF/LEF reporter activity, silencing the expression of the mutated allele or of both alleles strongly inhibited it (Fig. 1D). Moreover, expressions of positive targets of ß-catenin, such as GPR49, Axin2 and cyclinD1, were strongly inhibited by the application of sißcat-mut and sißcat-both (Fig. 1E). The opposite result was consistently obtained for negative targets such as Arg1 (Fig. 1E). The data therefore suggest that, in HepG2 cells, the TCF-dependent transcriptional activity of ß-catenin is mainly carried by the mutated form of ß-catenin.
Because of this impact on ß-catenin targets, and specifically on CyclinD1, HepG2 cell growth was found to be highly dependent on a mutated ß-catenin expression, while being insensitive to the KD of the WT form of ß-catenin (Fig. 1F). These data were confirmed on spheroids (Fig. S1B), showing that the mutated ß-catenin is required for HepG2 cell growth in 2D and 3D environments. The mutated form of ß-catenin would thus appear to be a strict requirement for HepG2 cell growth, which is consistent with the idea of oncogene addiction.
The large deletion present in HepG2 cells also removes the α-catenin binding site of ß-catenin. We thus found that the silencing of the mutated allele of ß-catenin did not impact E-cadherin localization at cell-cell junctions. Interestingly, we found that in an obverse manner WT ß-catenin KD strongly affected E-cadherin engagement at adherens junctions in HepG2 cells (Fig. 1G). Consequently, we observed that HepG2 cells acquire a more migrating and invasive phenotype upon silencing of WT ß-catenin (Figs. S1C–D). In cells lacking the expression of both ß-catenins, E-cadherin staining appears to be strongly affected with a more intracellular localization (Fig. 1G). These results suggest that the structural role of ß-catenin is altered only upon WT ß-catenin KD. This conclusion is further supported by our observations of ß-catenin localization (Fig. S1E). In control HepG2 cells, ß-catenin localized at cell-cell junctions, in the cytoplasm and in nuclei. Those stainings which were lost when cells were transfected with sißcat-both. In cells transiently transfected with sißcat-mut, remaining ß-catenin, i.e. WT ß-catenin, was enriched at cell-cell junctions, but failed to localize in the cytoplasm and in nuclei. Obversely, in cells transfected with sißcat-WT, remaining ß-catenin, i.e. mutated ß-catenin, continued to be cytosolic and less at the plasma membrane (Fig. S1E). Thus, in HepG2 cells, WT β-catenin appears to be mainly involved in adherent junctions, whereas mutated β-catenin appears to be preferentially involved in the regulation of gene expression. Based on these observations, this dual ß-catenin KD HepG2 model allows the two functions of ß-catenin in the same cellular background to be distinguished: the membrane/structural activity mediated by the degradable WT ß-catenin and the nuclear/transcriptional activity mediated by the mutated β-catenin. Thus, this model provides a suitable method of addressing the structural and the transcriptional functions of ß-catenin independently.
WT and mutated ß-catenin have distinct gene expression patterns
To study the involvement of each ß-catenin allele in gene expression, we performed a transcriptional analysis of HepG2 KD cells (Tables S3–5). The expression of ß-catenin and known-ß-catenin targets (Fig. S2A) was altered in a way similar to that previously observed by qRT-PCR (Fig. 1C and 1E), thus validating our data. Global analysis demonstrated that WT ß-catenin KD cells and mutated ß-catenin KD cells have distinct gene expression patterns. The silencing of both alleles led to a gene expression pattern closer to the mutated than the WT ß-catenin KD, suggesting a dominant impact of the oncogenic ß-catenin on gene expression in HepG2 cells (Fig. S2B). A venn diagram of our observations shows that only a small proportion of genes were altered by both the silencing of WT and of mutated ß-Catenin (Fig. S2C). Hoping to understand how WT ß-catenin may regulate gene expression, we performed gene set enrichment analysis (GSEA). GSEA revealed that the genes that up-regulated upon WT ß-catenin depletion showed significant overlap with genes associated with the TGFß and Hippo pathways (Fig. S2D). Indeed, the transcription factors TEAD1 and TEAD2, as well as the target genes of the Hippo pathway, were found to be up-regulated by the silencing of WT ß-catenin (Figs. S2E–F). These data are consistent with Stefano Piccolo’s work which shows that depletion of the sole WT pool of ß-catenin is sufficient to promote TAZ activity in HepG2 cells
. We also used FuncAssociate 3.0 to analyze alterations in biological functions (Table S6). Removal of the WT ß-catenin led to a decrease in the expression of genes involved in metabolic processes. Strikingly, we observed the opposite effect upon removal of the oncogenic ß-catenin. Given that metabolic functions are key features of differentiated hepatocytes, these alterations led us to explore the impact of both ß-catenin functions on cell differentiation.
Alteration of hepatocyte differentiation and polarity upon ß-catenin knock-downs
Differentiated hepatocytes are characterized by the expression of specific markers, including xenobiotic-metabolizing enzymes, transporters, transcription factors and bile canaliculi molecules. Interestingly, we observed a mirror image alteration upon the silencing of either the WT or the mutated allele of ß-catenin. Hepatocyte markers were found to be up-regulated by mutated ß-catenin silencing and down-regulated by WT ß-catenin silencing (Fig. 2A). Given that HNF4α (hepatocyte nuclear factor-4 alpha) is the major transcription factor involved in hepatocyte differentiation, we pursued our exploration of the impact of ß-catenin KD on HNF4α signaling. Using qRT-PCR analysis, we observed only a slight decrease of HNF4α expression upon WT ß-catenin KD, whereas it is not significantly altered by the silencing of the oncogenic ß-catenin (Fig. 2B). However, we observed a significant increase of HNF4α transcriptional activity when the oncogenic ß-catenin was silenced. This observation was made using an ApoC3 promoter reporter assay (Fig. 2C) and expression analysis of HNF4α transcriptional positive targets (Figs. S3A–C and Fig. 2D). The removal of WT ß-catenin slightly decreased the expression of HNF4α target genes. As described earlier
, this result confirms our belief that transcriptional activity of ß-catenin may repress the hepatocyte differentiation program of HNF4α. Taken as a whole, these results demonstrate that the structural function of ß-catenin, supported mainly by the WT ß-catenin, is necessary to maintain a differentiated state of hepatocytes, and that the inhibition of the transcriptional activity of oncogenic ß-catenin reverses the dedifferentiation program of HepG2 cells.
Fig. 2Antagonistic impact of WT and mutated ß-catenins. (A) The graph shows the relative expression of differentiated hepatocyte markers, upon silencing of both alleles of ß-catenin in HepG2 cells, extracted from the transcriptomic analysis. (B) Indicated siRNA were transfected in HepG2 cells, and HNF4A relative mRNA level was analyzed. Shown is the relative mRNA level compared to control transfected cells. (C) HepG2 cells transfected with indicated siRNAs were transfected with HNF4a responsive luciferase reporter. Shown is the mean relative luciferase activity, normalized to Renilla luciferase and compared to siRNA control transfected cells. (D) Relative mRNA levels of APOC3 and APOM, both positive transcriptional targets of HNF4a were analyzed by qRT-PCR. (E) Alteration of the expression of polarity markers upon silencing of both alleles of ß-catenin in HepG2 cells. The graph shows the relative expression of the indicated genes analyzed by qRT-PCR. (B-E) Each graph shows the quantification of three independent experiments. Error bars indicate s.e.m (n=3). * P < 0.05; ** P < 0.01; *** P < 0.001; **** P < 0.0001; ns, non significant by one way ANOVA.
Given that differentiated hepatocytes are polarized epithelial cells endowed with the capacity to produce and excrete bile into a specialized structure called bile canaliculus (BC), we further analyzed the effect of ß-catenin KD on BC markers. Similar to the alteration described above for hepatocyte markers, we observed a mirror image alteration of the expression of junction- and polarity-associated genes (Figs. S3D–E), such as JAM-A (F11R gene), connexin-32 (GJB1 gene) and claudin-1 (CLDN1 gene) (Fig. 2E, S3D-E). Given that HepG2 cells retain the ability to form BC in culture
Further cellular investigation of the human hepatoblastoma-derived cell line HepG2: morphology and immunocytochemical studies of hepatic-secreted proteins.
, we addressed the influence of ß-catenin on their maintenance, using confocal microscopy by staining F-actin and radixin (Fig. 3A). As for cells exhibiting BC, the depletion of mutated β-catenin induced an increase in their number, whereas the depletion of WT β-catenin induced a decrease in their number (Fig. 3A-B). We performed super-resolution STED microscopy on HepG2 cells stained with phalloidin, in order to visualize BC with a resolution better than confocal (Fig. 3C and Fig. S4A). Quantification of BC features demonstrated that mutated ß-catenin KD increases their size in terms of area, perimeter and diameter, whereas their circularity remains unchanged (Fig. 3D), with the number of cells engaged in the formation of each canaliculus also being found to increase (Fig. S4B). Finally, we examined their ability to translocate CFDA, a fluorescent substrate of ABC transporters, into the apical lumen. The removal of the mutated form of ß-catenin increased the percentage of CFDA positive BC, demonstrating that they were fully functional (Fig. 3E). Thus, WT and mutated ß-catenins can be said to act antagonistically on hepatocyte BC formation. These results highlight the fact that WT ß-catenin continues to act as a gatekeeper of differentiation and polarity in tumor hepatocytes.
Fig. 3Alteration of bile canaliculi formation upon silencing of ß-catenin in HepG2 cells. (A) HepG2 cells transfected with indicated siRNAs were fixed and stained with phalloidin (red), anti-radixin antibodies (green) and Hoescht (blue). Scale bar: 15 μm. (B) Quantification of the percentage of cells forming BC in the conditions described in (A). Graph shows the quantification of four independent experiments, where at least 100 cells were observed per experiment. (C) siRNA transfected HepG2 cells were fixed, stained with phalloidin-ATTO and observed by STED microscopy. Scale bar: 5 μm. (D) BC features (area, perimeter, circularity, min Feret (diameter)) were quantified by imageJ on STED images performed as described in (C). Each dot corresponds to one BC. (E) Quantification of BC functionality by using CFDA incorporation. Life-Act (red) expressing HepG2 cells were treated with CFDA (green). Scale bar: 25 μm. Graph shows the quantification of three independent experiments. (B, D-E) Error bars indicate s.e.m (n=4 for B, n=3 for D-E). * P < 0.05; ** P < 0.01 by one way ANOVA.
Fascin-1, as a target of ß-catenin, involved in hepatocyte dedifferentiation
Searching for genes that may impact hepatocyte differentiation and polarity, we focused on FSCN-1, which was found to up-regulate upon WT ß-catenin KD, and to down-regulate upon mutated ß-catenin KD (Fig. S5A). FSCN-1 encodes fascin-1, an actin-bundling protein, which is normally not expressed in epithelial cells. Unlike villin, an actin-bundling protein associated with BC microvilli, fascin-1 is absent in mature hepatocytes. Moreover, fascin-1 is expressed in tumors including HCC
. First, we analyzed fascin-1 protein expression in HB and HCC cell lines with different ß-catenin status (Fig. S5B). We found a greater expression of fascin-1 in cell lines bearing-CTNNB1 deletion (HepG2) or mutations (SNU398 and Huh6), as compared to non-mutated (Huh7, Hep3B) cell lines. The level of fascin-1 protein was positively correlated to the ß-catenin protein level, which accumulates upon mutations (Fig. S5C). As a ß-catenin transcriptional target, fascin-1 mRNA expression was strongly inhibited upon treatment of HepG2 cells with siRNAs sißcat-mut and sißcat-both (Fig. 4A). As previously observed for Axin2 and CyclinD1 (Fig. 1E), fascin-1 expression was slightly but significantly up-regulated upon removal of WT ß-catenin. These alterations were also detected at the protein level (Fig. S5D). Using a luciferase assay with fascin-1 promoter, we demonstrated that this regulation occurs at the transcriptional level, with an antagonistic regulation of promoter activity which occurs upon the silencing of either WT or mutated ß-catenin (Fig. 4B). Thus, FSCN1 shows itself to be a target gene of β-catenin in tumor hepatocytes. We then considered the possibility that fascin-1 expression might play a role in the alteration of hepatocyte differentiation. The silencing of fascin-1 in HepG2 cells led to increased ApoC3, claudin1 and E-cadherin mRNA expressions (Fig. 4C). The same tendency was observed in HCC cell lines, Hep3B and Huh7 (Figs. S6A–B). We also found that fascin-1 KD (Fig. S6C) led to a two- to three-fold increase in the number of BC in HepG2 cells (Fig. 4D). Moreover, the depletion of fascin-1 upon inhibition of WT β-catenin expression has a restorative effect on the formation of BC (Fig. 4E), demonstrating that fascin-1 is one of the effectors responsible for the impact of ß-catenin on BC formation. In the same way, the overexpression of fascin-1 induced a decrease in BC formation, as compared to control (Fig. S6D). Finally, consistent with the gain of differentiation observed with fascin-1 KD, we observed a strong inhibition of cell growth upon fascin-1 silencing (Fig. 4F). In parallel, in these conditions, we also found a slight decrease of the invasive properties of fascin-1 KD HepG2 cells (Fig. S6E). Taken together, these results demonstrate that the level of fascin-1 regulated by ß-catenin alters hepatocyte differentiation status.
Fig. 4Fascin-1, a target of β-catenin, alters hepatocyte differentiation status. (A) Fascin-1 mRNA expression upon depletion of β-catenin analyzed by RT-qPCR in HepG2 cells. (B) Activity of Fascin-1 promoter studied by reporter luciferase assay. Shown is the mean relative luciferase activity, normalized to Renilla luciferase and compared to control siRNA transfected cells. (C) mRNA levels of FSCN1, APOC3, CLDN1 and CDH1 upon treatment of HepG2 cells with siRNA targeting Fascin-1 (siFascin1#1, #2, #3 and #4). (D) siRNA transfected HepG2 cells were fixed and stained with phalloidin (red), anti-radixin antibodies (green) and Hoescht (blue). Scale bar: 15 μm. The graph shows the quantification of three independent experiments where the number of BC formed for 100 cells are indicated. (E) Experiments were performed as described in (D) with co-transfection of indicated siRNAs. (F) HepG2 cell growth was monitored upon KD of Fascin-1 using the indicated siRNAs. (A-F) Graphs show the quantification of at least three independent experiments. Error bars indicate s.e.m (n=4 for A, n=5 for B, and n=3 for C-F). * P < 0.05; ** P < 0.01; *** P < 0.001; **** P < 0.0001; ns, non significant by one way ANOVA.
High fascin expression in undifferentiated tumors in mice
Next, we sought to confirm fascin-1 as a ß-catenin target in liver tumors in mice. For this, we used mouse models that mimic ß-catenin dependent tumorigenesis, such as the APC loss-of-function and the Δexon3 ß-catenin models. These models have been shown to lead to the development of phenotypically undistinguishable liver tumors within about 10 months
. The well-differentiated tumors were characterized by hepatocyte-like tumor cells that maintain a β-catenin-induced expression of glutamine synthetase (GS) and in which nuclear β-catenin is present. The undifferentiated tumors were composed of small cells with basophilic nuclei, and showed a strong expression of nuclear β-catenin
. RNA-Seq analysis showed that these tumors lose the expression of GS, which is reminiscent of a loss of differentiation (Fig. S7A). We therefore analyzed fascin-1 expression in this mouse cohort (Table S1). We found that fascin-1 expression increased in both well-differentiated and undifferentiated types of tumors, as compared to normal liver (Fig. 5A and S7B). Doing so, we again highlighted that fascin-1 expression was higher in undifferentiated samples than in differentiated ones (Fig. 5B). Moreover, fascin mRNA expression was found to correlate positively with various markers of undifferentiated tumors such as MMP2, VIM, HIF1A and YAP1, and was found to correlate negatively with markers of differentiated tumors such as HNF4a, APOC3 and GJB1 (Fig. S7C). Finally, as a bona fide ß-catenin target, fascin-1 expression correlated positively with the levels of CTNNB1, LEF1 and TCF4 (Fig. S7C). These new findings were confirmed by immunohistochemistry performed on both types of murine tumors obtained from a new cohort of APC KO mice. As described previously
, the well-differentiated tumors were characterized by the nuclear localization of β-catenin, along with a high expression of GS (Fig. 5C), while the undifferentiated tumors were characterized by a GS expression ranging from lower to near-absent(Fig. 5D). Fascin-1 staining, which is absent from the non-tumoral hepatocytes, was found to have increased in both types of tumors. However, whereas its expression is low in differentiated tumors, fascin expression was high in undifferentiated tumor cells, with a cytoplasmic and membranous staining. In mice, fascin-1 expression may therefore be said to be a marker of ß-catenin-induced undifferentiated tumors. Given that those murine tumors were found to be transcriptionally close to human mesenchymal HBs
, we were prompted to explore fascin-1 expression in human HBs.
Fig. 5Fascin-1 expression in murine ß-catenin-mediated tumors. (A-B) Data were extracted from RNAseq performed on mouse hepatic tumors induced either from APC KO or ß-catenin Δexon3 expression in livers. FSCN-1 expression is shown in non-tumoral samples (n=12) and in differentiated (n=13) and undifferentiated (n=6) tumors. Levels of significance: * P < 0.05; **** P < 0.0001; by one way ANOVA test and Kruskal-Wallis post-test. (C) Representative IHC images of ß-catenin, Fascin-1, and glutamine synthetase (GS) in differentiated (HCC-type) (C), and undifferentiated (HB-type) (D) murine tumors. Boxed regions are enlarged in the zoom images.
. Doing so, we found that fascin-1 mRNA is specifically expressed in the C2-subtype of HBs that corresponds to poor-prognosis tumors, with no obvious difference between C2A and C2B subtypes (Fig. 6A). As observed in mice, fascin-1 mRNA expression correlated negatively with markers of differentiated hepatocytes such as HNF4a, APOC3, GJB1 and CLDN1 (Fig. S8A). Moreover, in the cohort used by Sumazin et al.
to classify HBs in three risk-stratifying molecular subtypes, we found that fascin-1 is significantly more expressed in high-risk and intermediate-risk tumors than in low-risk tumors, indicating that high-fascin expression is associated to poor-prognosis subgroups (Fig. 6B). We then used immunohistochemistry to confirm the expression of the fascin-1 protein in HBs. Whereas fascin-1 was not expressed in normal hepatocytes, and was restricted to sinusoidal cells in normal and peri-tumoral tissues (Fig. 6C and Fig. S8B), it was expressed in a specific contingent of tumor cells in HBs (Fig. 6C). This fascin-1 staining is highly consistent with our mouse data in showing that fascin-1 is expressed in GS-negative cells and in cells, which are highly ß-catenin positive (Fig. 6C and S8C). These cells may correspond to the embryonal contingent of undifferentiated small cells with basophilic nuclei. However, given that chemotherapy is known to induce alteration of hepatoblastoma cell subtypes, we further used a human sample from a patient that was not treated by chemotherapy before surgery (Fig. S8D). We observed that fascin-1 positive tumor cells are also present as small cuboid or round cells forming solid nests or glands. They look as poorly differentiated epithelial cells, with intermediate phenotype between embryonal and small undifferentiated cells, since anti-hepatocyte is always negative, CK19 unfrequently positive (slightly positive on glands) and vimentin faintly positive. The mesenchymal cells with strong vimentin stain were conversely negative for fascin-1 (Fig. S8E). Fetal areas strongly stained with anti-hepatocyte and anti-GS antibodies were also negative for fascin-1 (Fig. S8F). Taken together, fascin-1 expression can therefore be confirmed as a specific marker of ß-catenin-induced primitive cells including embryonal and blastemal cells.
Fig. 6Fascin-1 expression in human HBs. (A) Data were extracted from RNAseq performed on human HBs by Hooks et al.
. FSCN-1 expression is shown in non-tumoral samples (n=30) and in C1 (n=20) and C2A or C2B (n=9) tumors. Levels of significance: *** P < 0.001; **** P < 0.0001; ns, non significant by one way ANOVA test. (B) Data were extracted from RNAseq performed on human HBs by Sumazin et al.
. FSCN-1 expression is shown in high-risk and/or intermediate-risk or low-risk HB tumors. Levels of significance: * P < 0.05; ** P < 0.01; ns, non significant by one way ANOVA test (left-hand graph) or unpaired t-test (right-hand graph). (C) Representative IHC images of ß-catenin, Fascin-1 and GS and HES staining in a C2B HB pediatric case. Boxed regions are enlarged in the zoom images. NT, non-tumoral; T, Tumoral.
ß-catenin performs a structural function at cell-cell junctions and a transcriptional function in the nucleus, but these functions are difficult to dissociate and therefore to study individually. Here, we developed the dual ß-catenin KD HepG2 model, which allowed us to perform allele-specific KD, and to address the interplay between the WT and the mutated ß-catenin in liver tumor cells. We found that the mutated form of ß-catenin is mostly dedicated to the transcriptional function, whereas the WT ß-catenin, without Wnt stimulation, is more endowed with its adhesive function in HepG2 cells. Our model is therefore enables the structural and transcriptional functions of ß-catenin to be studied independent of each other. Our transcriptomic analysis revealed that the disruption of each function alters gene expression. Consistent with the transcriptional activity of ß-catenin, genes that were dysregulated upon removal of mutated ß-catenin were coherent with a Wnt/ß-catenin signature. By contrast, genes altered upon the silencing of the WT allele were endowed with different signaling pathways, including the Hippo pathway. In fact, the structural function of ß-catenin, considered transcriptionally irrelevant, had previously been shown to be a potent repressor of the TAZ transcriptional program
. We also described both the transcriptional and the adhesive activities of β-catenin as playing antagonistic roles in tumor hepatocytes. Whereas the structural function of ß-catenin is necessary for the maintenance of a differentiated state of hepatocytes, the transcriptional activity of ß-catenin induces the dedifferentiation program of HepG2 cells. The KD of each allele specifically reverts those programs. As previously published
, we found that this antagonism is due in part to the regulation of the transcriptional activity of HNF4α, which is required to maintain a hepatocyte differentiation program. Thus, the dual ß-catenin KD model recapitulates many features previously described for ß-catenin-dependent pathways, and is therefore an appropriate means of furthering our understanding of the role of ß-catenin in liver carcinogenesis.
We revealed that WT and mutated ß-catenins act in an opposite manner on hepatocyte polarity. Normal membrane polarity and the formation of BC are essential for hepatocyte function, and their alteration may lead to numerous diseases, including cholestasis-associated diseases. In the literature, however, the link between β-catenin and BC abnormalities has not been elucidated. On the one hand, cholestasis has been described as a feature of ß-catenin-mutated HCCs
. Whether this phenotype is due to the structural or the transcriptional activity of ß-catenin remains largely unknown. Several studies have reported the involvement of cell adhesion molecules, such as E-cadherin or α-catenin, in the maintenance of BC
. Here, we revealed that mutated ß-catenin plays a repressor role by decreasing the number and the size of BC, and that WT ß-catenin is important for their maintenance. As demonstrated for the development of bile ducts
, ß-catenin must be maintained at the right level, as its loss or overactivation is detrimental to BC formation and/or stabilization.
In addition to this, our work also highlights the involvement of fascin-1 downstream of ß-catenin activation. Fascin-1 has been reported as a transcriptional target of β-catenin/TCF signaling in colon cancer cells
. Obversely, fascin-1 has been found to induce epithelial-mesenchymal transition of cholangiocarcinoma cells, and to promote breast cancer stem cell function by regulating Wnt/ß-catenin signaling
Fascin Activates beta-Catenin Signaling and Promotes Breast Cancer Stem Cell Function Mainly Through Focal Adhesion Kinase (FAK): Relation With Disease Progression.
Downstream of ß-catenin, we found that fascin-1 expression alters hepatocyte polarity and differentiation status. In vitro, the silencing of fascin-1 up-regulates epithelial markers and increases BC formation. Fascin-1 is highly expressed in ß-catenin-mutated undifferentiated tumors both in mice and in human, where its expression correlates negatively with differentiated hepatocyte markers and correlates positively with mesenchymal markers. It has yet to be explored how epithelial and mesenchymal gene expression may be regulated by fascin-1, which is an actin-binding protein. One hypothesis is mechanotransduction, given that modulation of the actin cytoskeleton is known to alter gene expression. An alternative hypothesis points to the presence of fascin-1 in a large list of proteins found to bind mRNA, suggesting a potential role of fascin in post-translational regulation
Our study sheds light on the existence of various differentiation states of tumors upon ß-catenin activation. As has been recently described, the expression of mutated ß-catenin in mouse hepatocytes can generate either well-differentiated HCC-like tumors or undifferentiated HB-like tumors
. GS expression is the gold standard marker of CTNNB1-mutated HCC, i.e. well-differentiated CTNNB1-mutated tumor cells. In this study, we propose fascin-1 as a strong marker of undifferentiated CTNNB1-mutated tumor cells, negative for GS. The manner or mechanism by which the same ß-catenin mutation can generate different gene expression profiles and different tumor types, however, remains a matter of ongoing debate
. We have clearly demonstrated that WT ß-catenin acts as a gatekeeper of differentiation in tumor hepatocytes. Our data are consistent with the hypothesis that the expression of WT ß-catenin may blunt the oncogenic ability of ß-catenin mutations in hepatocytes. Specifically, according to this hypothesis WT ß-catenin removal favors the Hippo pathway target gene expression, which is also key in the development of HB
Finally, we revealed fascin-1 as a previously unrecognized marker of CTNNB1-mutated immature tumor cells, i.e. the embryonal contingent, in HBs. Even if fascin-1 is enriched in the C2 subclass of HBs, we also found fascin-1 staining in C1 samples, thus indicating the complexity of these tumors. Fascin-1 being mainly associated to HBs with bad prognosis, it may therefore be suitable to consider it as a new actionable target in these liver pediatric tumors.
Conflict of interest
The authors declare no conflict of interest.
Please refer to the accompanying ICMJE disclosure forms for further details.
Acknowledgements
We thank Dr D. Vignjevic (Curie Institute, Paris, France) for Fascin-1 DNA constructs. We thank the plateforme GenomEast of Strasbourg (Strasbourg, France). Microscopy was done in the Bordeaux Imaging Center, a service unit of the CNRS-INSERM and Bordeaux University, member of the national infrastructure France Bio Imaging, with the help of Dr. Philippe Legros for STED analysis. We thank Anne-Aurélie Raymond (Inserm U1053, Bordeaux, France) for her help in transcriptomic data analysis. We thank the department of Pathology and the Centre de Ressources Biologiques of the University Bordeaux Hospital for human HB sample stains. The graphical abstract was created with BioRender.com.
Appendix A. Supplementary data
The following is/are the supplementary data to this article:
Further cellular investigation of the human hepatoblastoma-derived cell line HepG2: morphology and immunocytochemical studies of hepatic-secreted proteins.
Fascin Activates beta-Catenin Signaling and Promotes Breast Cancer Stem Cell Function Mainly Through Focal Adhesion Kinase (FAK): Relation With Disease Progression.
00022-8&id=ga1.jpg){kind=link}
00022-8&id=gr1.jpg){kind=link}
00022-8&id=gr2.jpg){kind=link}
00022-8&id=gr3.jpg){kind=link}
00022-8&id=gr4.jpg){kind=link}
00022-8&id=gr5.jpg){kind=link}
00022-8&id=gr6.jpg){kind=link}