A functional screening identifies five micrornas controlling glypican-3: role of mir-1271 down-regulation in hepatocellular carcinoma

Authors

  • Marion Maurel,

    1. Univ. Bordeaux, Physiopathologie du cancer du foie, U1053, F-33000 Bordeaux, France
    2. INSERM, Physiopathologie du cancer du foie, U1053, F-33000 Bordeaux, France
    Search for more papers by this author
  • Sandra Jalvy,

    1. Univ. Bordeaux, Physiopathologie du cancer du foie, U1053, F-33000 Bordeaux, France
    2. INSERM, Physiopathologie du cancer du foie, U1053, F-33000 Bordeaux, France
    Search for more papers by this author
    • These authors contributed equally to this work.

  • Yannick Ladeiro,

    1. Université Paris Descartes, Labex Immuno-oncology, Sorbonne Paris Cité, Faculté de Médecine, Paris, France
    2. INSERM, Altération génétique des tumeurs hépatiques, U674, IUH, F-75010 Paris, France
    Search for more papers by this author
    • These authors contributed equally to this work.

  • Chantal Combe,

    1. Univ. Bordeaux, Physiopathologie du cancer du foie, U1053, F-33000 Bordeaux, France
    2. INSERM, Physiopathologie du cancer du foie, U1053, F-33000 Bordeaux, France
    Search for more papers by this author
  • Laetitia Vachet,

    1. Univ. Bordeaux, Physiopathologie du cancer du foie, U1053, F-33000 Bordeaux, France
    2. INSERM, Physiopathologie du cancer du foie, U1053, F-33000 Bordeaux, France
    Search for more papers by this author
  • Francis Sagliocco,

    1. Univ. Bordeaux, Physiopathologie du cancer du foie, U1053, F-33000 Bordeaux, France
    2. INSERM, Physiopathologie du cancer du foie, U1053, F-33000 Bordeaux, France
    Search for more papers by this author
  • Paulette Bioulac-Sage,

    1. Univ. Bordeaux, Physiopathologie du cancer du foie, U1053, F-33000 Bordeaux, France
    2. INSERM, Physiopathologie du cancer du foie, U1053, F-33000 Bordeaux, France
    Search for more papers by this author
  • Vincent Pitard,

    1. Univ. Bordeaux, CIRID, UMR 5164, F-33000 Bordeaux, France
    2. CNRS, CIRID, UMR5164, F-33000 Bordeaux, France
    Search for more papers by this author
  • Hélène Jacquemin-Sablon,

    1. Univ. Bordeaux, Physiopathologie du cancer du foie, U1053, F-33000 Bordeaux, France
    2. INSERM, Physiopathologie du cancer du foie, U1053, F-33000 Bordeaux, France
    Search for more papers by this author
  • Jessica Zucman-Rossi,

    1. Université Paris Descartes, Labex Immuno-oncology, Sorbonne Paris Cité, Faculté de Médecine, Paris, France
    2. INSERM, Altération génétique des tumeurs hépatiques, U674, IUH, F-75010 Paris, France
    Search for more papers by this author
  • Benoît Laloo,

    1. Univ. Bordeaux, Physiopathologie du cancer du foie, U1053, F-33000 Bordeaux, France
    2. INSERM, Physiopathologie du cancer du foie, U1053, F-33000 Bordeaux, France
    Search for more papers by this author
  • Christophe F. Grosset

    Corresponding author
    1. Univ. Bordeaux, Physiopathologie du cancer du foie, U1053, F-33000 Bordeaux, France
    2. INSERM, Physiopathologie du cancer du foie, U1053, F-33000 Bordeaux, France
    • Université Bordeaux Segalen, Physiopathologie du cancer du foie, INSERM U1053, 146 Rue Léo Saignat, F-33076 Bordeaux cedex, France
    Search for more papers by this author
    • fax: +33 (0)556 514 077


  • Potential conflict of interest: Nothing to report.

  • Supported by the Agence Nationale pour la Recherche (ANR)-Programme Jeunes Chercheurs (Paris, France) (Grant no: JC07_184264 to C.G.), by La Ligue Nationale Contre le Cancer, by the Biointelligence consortium (OSEO) and INCa (PAIR-CHC). M.M and Y.L. were recipients of fellowships from the Ministère de l'Enseignement Supérieur et de la Recherche (MESR) and ANRS, respectively.

Abstract

Hepatocellular carcinoma (HCC) is the major primary liver cancer. Glypican-3 (GPC3), one of the most abnormally expressed genes in HCC, participates in liver carcinogenesis. Based on data showing that GPC3 expression is posttranscriptionally altered in HCC cells compared to primary hepatocytes, we investigated the implication of microRNAs (miRNAs) in GPC3 overexpression and HCC. To identify GPC3-regulating miRNAs, we developed a dual-fluorescence FunREG (functional, integrated, and quantitative method to measure posttranscriptional regulations) system that allowed us to screen a library of 876 individual miRNAs. Expression of candidate miRNAs and that of GPC3 messenger RNA (mRNA) was measured in 21 nontumoral liver and 112 HCC samples. We then characterized the phenotypic consequences of modulating expression of one candidate miRNA in HuH7 cells and deciphered the molecular mechanism by which this miRNA controls the posttranscriptional regulation of GPC3. We identified five miRNAs targeting GPC3 3′-untranslated region (UTR) and regulating its expression about the 876 tested. Whereas miR-96 and its paralog miR-1271 repressed GPC3 expression, miR-129-1-3p, miR-1291, and miR-1303 had an inducible effect. We report that miR-1271 expression is down-regulated in HCC tumor samples and inversely correlates with GPC3 mRNA expression in a particular subgroup of HCC. We also report that miR-1271 inhibits the growth of HCC cells in a GPC3-dependent manner and induces cell death. Conclusion: Using a functional screen, we found that miR-96, miR-129-1-3p, miR-1271, miR-1291, and miR-1303 differentially control GPC3 expression in HCC cells. In a subgroup of HCC, the up-regulation of GPC3 was associated with a concomitant down-regulation of its repressor miR-1271. Therefore, we propose that GPC3 overexpression and its associated oncogenic effects are linked to the down-regulation of miR-1271 in HCC. (HEPATOLOGY 2013)

Hepatocellular carcinoma (HCC) is the most common form of primary liver cancer.1 It usually develops in an affected liver with cirrhosis due to viral infection (hepatitis B virus, HBV; hepatitis C virus, HCV), alcohol abuse, metabolic disorders, or a carcinogenic agent.1-3 HCC is a very heterogeneous class of tumors characterized by multiple types of genomic damages associated with its various etiologies.2-4 This tumor diversity arises from multistep hepatocarcinogenic processes requiring sequential genetic and epigenetic alterations including gene mutations and/or chromosome instability.1-5

Glypican-3 (GPC3), one of the numerous genes abnormally expressed in HCC, participates in hepatocarcinogenesis.4, 6, 7 The GPC3 protein belongs to the family of heparan-sulfate proteoglycans. Glypicans act as coreceptors and control signaling pathways by regulating growth factor/cell-surface receptor interactions.6, 7 In adult liver, GPC3 is generally not detectable. Nevertheless, this oncofetal protein plays a critical role in liver regeneration and hepatic growth during embryogenesis.6, 7 As shown in the X-linked genetic Simpson-Golabi-Behmel syndrome, loss-of-function mutations of GPC3 cause postnatal overgrowth with multiple congenital anomalies, highlighting its implication in control of cell proliferation and tissue growth.7 GPC3 is strongly expressed in most tumoral hepatic tissues in which it potentiates the malignancy of hepatic cells through the canonical Wnt/β-catenin pathway.6, 7 In HCC, its expression is associated with a poor histological tumor differentiation and a high-proliferative state of cancerous hepatic cells,3, 8 as well as with poor prognosis and short overall survival.9 These observations suggest that GPC3 expression is strongly linked to aggressive behavior of the tumors and are in agreement with its increased expression in several embryonic cancers.6, 7 Therefore, GPC3 clearly represents a relevant molecular target in several cancers, including HCC.6, 7

Besides deregulation occurring as a result of gene amplification, gene mutation, or transcriptional variations during carcinogenesis, there is now clear evidence that microRNAs (miRNAs) actively participate in gene misexpression.3, 5, 10, 11 MiRNAs are small noncoding RNAs that control gene expression by modulating stability and/or translation of messenger RNA (mRNA)12, 13 through interactions with specific sequences located in either the coding or the untranslated regions (UTR).14 MiRNAs are intricately involved in human diseases and actively participate in carcinogenesis as oncogenes or tumor suppressor genes.10, 11 Although the effects mediated by miRNAs on any particular target are modest, the simultaneous regulation of a broad array of targets by one miRNA can lead to a profound gene reprogramming and cell-phenotype changes.10, 11, 15 In the cancerous context, restoring high levels of a tumor suppressor miRNA or inhibiting the biological activity of an oncogenic miRNA constitutes a very promising avenue of investigation in anticancer therapy.15

We previously showed that posttranscriptional dysfunctions associated with 3′-UTRs of genes are deregulated in two examples of human pathological disorders.13, 16, 17 Here we report that the 3′-UTR-mediated posttranscriptional regulation of GPC3 is altered in HCC cells and that the 3′-UTR favors GPC3 expression in HCC. In an attempt to identify miRNAs involved in this dysfunction, we developed a methodology, called dual-fluorescence functional, integrated, and quantitative method to measure posttranscriptional regulations (DF-FunREG), which allows the functional screening of miRNA libraries and the systematic identification of miRNAs controlling one gene. We report the regulation of GPC3 expression by five miRNAs. Among them, two in silico-predicted miRNAs acted negatively on GPC3 expression,18 whereas three acted positively. Finally and importantly, we report that expression of one of these miRNAs is broadly decreased in HCC tumors and show that its down-regulation contributes to GPC3 overexpression and HCC-cell growth.

Abbreviations

AFP, alpha-fetoprotein; AM1271, antimiR-1271; ATP, adenosine triphosphate; CT, control 3′-UTR; DF-FunREG, dual fluorescence-FunREG; FACS, fluorescence-activated cell sorting; FunREG, functional, integrated, and quantitative method to measure posttranscriptional regulations; GFP, green fluorescent protein; GLO, β-globin; HBV, hepatitis B virus; HCC, hepatocellular carcinoma; HCV, hepatitis C virus; M, GFP mRNA; miRNA, microRNA; m.o.i., multiplicity(ies) of infection; mRNA, messenger RNA; NTL, nontumoral liver; P, GFP protein; PCR, polymerase chain reaction; qPCR, quantitative PCR; qRT-PCR, quantitative reverse transcription PCR; siRNA, small interfering RNA; TCN, transgene copy number; TOM, tdTomato; UTR, untranslated region.

Materials and Methods

Plasmids.

Plasmids are as described in the Supporting Information.

Cell Lines and Primary Hepatocytes.

Fresh primary human hepatocytes (Biopredic, Rennes, France), as well as HCC-derived HuH7 and SNU398 cell lines, were grown as described.13, 18 TGG and TG HuH7 cells expressing the pL-Tomato+pL-GFP-GPC3 and the pL-Tomato+pL-GFP transgenes, respectively, were developed by lentiviral transduction (multiplicity(ies) of infection [m.o.i.] = 3) and cell sorting.

Liver Samples and Clinical Data.

Liver tissues were immediately frozen in liquid nitrogen and stored at −80°C until used for molecular studies. All patients were recruited in accordance with French law and institutional ethical guidelines. Liver samples were clinically, histologically, and genetically characterized. A first series of 133 liver samples (112 HCC and 21 nontumoral liver [NTL] samples) was collected from 118 patients surgically treated at French University Hospitals (Supporting Table S1). These HCCs were classified according to the clinical, pathological, and genetic features as described.4 A second series of 38 liver paired samples (19 HCC and their corresponding NTL samples) was collected from 19 patients surgically treated at French University Hospitals (Supporting Table S2).

Lentiviral Production and Cell Transduction.

Production and titration of infectious lentiviral particles, as well as cell transduction, were as described elsewhere.13, 18

Small RNAs, miRNA Library, Cell Transfection, miRNA Quantification, FunREG and DF-FunREG Analyses, Western Blotting, Cellular Assays, and Statistical Analyses.

All materials and methods are as described in the Supporting Information.

Results

GPC3 Expression Is Posttranscriptionally and Differentially Regulated in HCC Cells Compared to Primary Hepatocytes.

Using the FunREG method,13 we investigated whether GPC3 expression is posttranscriptionally regulated in hepatic cells and whether this regulation is altered in HCC compared to normal cells. Infectious lentiviral particles were used to deliver a GFP-reporter transgene bearing a control 3′-UTR (CT) fusioned to either the GPC3 or the rabbit β-globin (GLO) 3′-UTR (Fig. 1A) into HCC-deriving HuH7 and SNU398 cells, and primary hepatocytes. One week later, the P/TCN (GFP protein/transgene copy number) ratio, which is indicative of a posttranscriptional regulation, was calculated.13 Compared to the referent GFP-GLO transgene, GPC3 3′-UTR increased GFP expression by 25% in HuH7 and SNU398 cells (Fig. 1B). In contrast, its presence strongly reduced GFP expression in normal hepatocytes. Consequently, normalized expression of the GFP-GPC3 transgene was 4-fold higher in HCC cells than in primary hepatocytes (Fig. 1B). These results showed that the GPC3 3′-UTR-mediated regulation is differentially controlled in normal and tumoral hepatic cells and that the 3′-UTR promotes GPC3 expression in the HCC context.

Figure 1.

The posttranscriptional regulation mediated by GPC3 3′-UTR is altered in HCC cells. (A) Schematic representations of the GFP-GLO (top) and GFP-GPC3 (bottom) transgenes. (B) HuH7, SNU398 cells, or hepatocytes were transduced once with lentiviruses expressing the indicated transgene and then analyzed by FunREG.13 After 1 week the GFP protein expression (P) and transgene copy number (TCN) were determined by FACS and quantitative polymerase chain reaction (qPCR), respectively. The P/TCN ratio is indicative of global posttranscriptional regulation (analysis of variance [ANOVA]: P < 0.0001; n = 3).13 Otherwise stated, in this figure and following ones, bars represent means, error bars represent standard deviations (SD) and the ANOVA test was followed by a Bonferroni's multiple comparison post-test. ***P < 0.001.

Systematic Identification of MiRNAs Targeting the GPC3 3′-UTR in HCC Cells by DF-FunREG.

Given the major role of miRNAs in gene regulation and liver carcinogenesis,11 we hypothesized that these small RNAs could be responsible for the posttranscriptional dysfunction associated with GPC3 3′-UTR in HCC cells. As in silico miRNA target predictions sometimes lack accuracy,18-20 we used a systematic approach in order to identify miRNAs controlling GPC3 expression through its 3′-UTR. Based on FunREG technology,13, 16 we developed an in cellulo system that allowed us to functionally screen a library of 876 miRNAs. This system, called DF-FunREG, is based on the dual expression of the green fluorescent protein (GFP) and the tdTomato (TOM). Using infectious lentiviruses, we established two HuH7 cell lines. The TGG cell line expressed the reference TOM transgene that bears the CT 3′-UTR and the test GFP transgene bearing the GPC3 3′-UTR upstream of the CT 3′-UTR (see Fig. 2A; Primary Screen box), whereas the TG cell line expressed TOM and a GFP transgene only bearing the CT 3′-UTR (see Fig. 2A, Secondary Screen box). According to our pipeline strategy (Fig. 2A), we performed two successive in cellulo screens. First, TGG cells were transfected by a control miRNA, miR-96 as a positive control for cell transfection (not shown) or each of the 876 human miRNAs from the library. Three days later, expressions of GFP and TOM were measured by fluorescence-activated cell sorting (FACS) and the GFP/TOM ratio was calculated in each case. Hits corresponding to miRNAs of interest were selected on the base of a significant change of GFP/TOM ratio compared to the negative-control (three independent experiments). As shown in Fig. 2A,B, the primary screen led to the selection of 16 miRNAs. Second, TG cells (lacking the GPC3 3′-UTR) were transfected with the 16 selected miRNAs and analyzed by FACS. The GFP/TOM ratio was calculated for each miRNA and compared to the one obtained with TGG cells (Fig. 2C). We reasoned that if an miRNA specifically targets GPC3 3′-UTR, the GFP/TOM ratio will differ between TG and TGG cells. Conversely, if the GFP/TOM ratio does not differ, this means that the miRNA recognizes the reporter transgenes outside the sequence of interest. As shown in Fig. 2A,C, eight miRNAs fulfilled this criterion and significantly changed the GFP/TOM ratio. MiR-96, already known as a GPC3 regulator,18 was among them, supporting the validity of our screening approach. Finally, these eight miRNAs were further characterized for their ability to control expression of the endogenous GPC3 in HuH7 cells. As shown in Fig. 2D, five miRNAs were validated as true modulators of GPC3 protein expression. Although miR-96 and its paralog miR-1271 significantly down-regulated GPC3 protein expression, miR-129-1-3p, miR-1291, and miR-1303 had a stimulatory effect. The functional effect of these five miRNAs on GPC3 expression was also demonstrated at the level of mRNA (Fig. 2E), suggesting mechanisms acting on mRNA stability. Presently, it is not clear why miR-200c*, miR-941, and miR-1973 had no effect on expression of endogenous GPC3. This could be due to the presence of other cis-regulatory sequences into the GPC3 mRNA located outside the 3′-UTR, or to differential processing of GPC3 and GFP transcripts along steps of mRNA maturation, export, or cytoplasmic localization.

Figure 2.

Five miRNAs regulate GPC3 expression through its 3′-UTR. (A) Screening strategy. Primary screen: schematic representation of the pL-GFP-GPC3 and pL-Tomato transgenes expressed in TGG HuH7 cell line. Secondary screen: schematic representation of the pL-GFP and pL-Tomato transgenes expressed in TG HuH7 cell line. Validation: The ability of eight miRNA candidates to control GPC3 expression was assessed by molecular approaches. (B) Primary screen using the TGG cell line and the library of 876 miRNAs (ANOVA: P < 0.0001; n = 3). (C) Secondary screen using the TG cell line and the 16 selected miRNA candidates, and comparison with data from the primary screen (ANOVA: P < 0.0001; n = 3). B,C: The negative small noncoding RNA control (not shown) was normalized to 1. (D,E) Validation experiments. Relative expression of GPC3 protein (D) and mRNA (E) in cells transfected by the indicated RNAs (ANOVA: P < 0.0001; n = 7). A representative western blot experiment is shown on top of (D). *P < 0.05; **P < 0.01; ***P < 0.001.

Although bioinformatics yield important insights into miRNA biology, the different available algorithms predicted that a large number of miRNAs should target GPC3 3′-UTR.18 Using miRWalk (a computational approach that compares its own miRNA:target predictions with those of nine established algorithms21), 200 miRNAs were predicted to target GPC3 through its 3′-UTR. Among them, 24 showed a higher probability of targeting with 4 to 6 positive predictions over 10 (Supporting Table S3). MiR-96 and miR-1271 paralogs appeared among the top ranked, with 6 and 4 positive predictions over 10, respectively. The three miRNAs having a positive effect on GPC3 expression (miR-129-1-3p, miR-1291, and miR-1303) were not predicted by any algorithms.

MiR-1271 Expression Is Decreased in HCC and Correlated with GPC3 mRNA in HBV-HCC Subgroup.

To investigate the relevance of the five GPC3-regulating miRNAs that we identified in HuH7 cells, we measured their relative expression as well as that of GPC3 mRNA in 112 HCC tumors and 21 NTL. As shown in Fig. 3A, GPC3 mRNA was highly and significantly overexpressed in all HCC including those from patients infected (HBV-HCC) or not (non-HBV-HCC) by HBV, as well as in the more aggressive tumors classified in G1 to G3 subgroups as defined by a transcriptome analysis.4 Concerning the miRNAs, whereas miR-129-1-3p was not detected in liver tissues, miR-1303 expression remained unchanged in HCC compared to NTL (Supporting Fig. S1A). Expression of miR-1291 significantly increased (3-fold, P < 0.05) in the G3 subgroup (Supporting Fig. S1B), whereas that of miR-96 was increased (201-fold, P < 0.05) in all HCC and more specifically in the G1 and G3 subgroups (Supporting Fig. S1C).22 Finally, compared to NTL, miR-1271 was found significantly decreased (−2.4-fold; P < 0.001) in all HCC groups (Fig. 3B,D). Importantly, miR-1271 expression inversely and strongly correlated with that of GPC3 mRNA in the HBV-HCC subgroup (Fig. 3C; P = 0.0008), whereas no correlation was found in any groups with any other miRNAs.

Figure 3.

miR-1271 expression is down-regulated in HCC and negatively correlated with GPC3 expression in HBV-HCC tumors. (A,B): Relative expression of GPC3 mRNA (A) and miR-1271 (B) in 112 HCC and 21 NTL samples (ANOVA: P < 0.0001; mean + standard error of the mean). Expressions in HCC subgroups are as indicated.4 (C) Correlative analysis of miR-1271 and GPC3 mRNA expression in the HBV-HCC subgroup (Spearman r = −0.72; ***P [two-tailed] = 0.0008). (D) Absolute expression of miR-96, miR-1271, miR-21, and miR-122 in 19 HCC and their corresponding adjacent NTL samples (ANOVA: P < 0.0001). *P < 0.05; **P < 0.01; ***P < 0.001.

As reported,23 miR-96 and miR-1271 are paralogs with functional similarities in cellulo. However, they are not expressed in the same tissues. Although miR-96 is expressed in sensory organ, miR-1271 is expressed in a variety of human tissues including liver, suggesting distinctive functions in vivo.23 To obtain insight into the opposite deregulation of miR-96 and miR-1271 in HCC and into the lack of correlation between expression of GPC3 mRNA and that of miR-96 in any subgroups, we measured the absolute copy number of each miRNA in 19 HCC and the corresponding NTL samples. As a comparison, we concomitantly measured the absolute copy number of two liver miRNAs, miR-122 and miR-21. As previously reported,11 miR-122 was highly expressed in liver and its expression significantly decreased in HCC, whereas that of miR-21 increased. Moreover, miR-1271 expression was similar to that of miR-21 in NTL, suggesting that miR-1271 is rather abundant in hepatic tissues (Fig. 3D). Finally, miR-1271 was 2,770-fold and 319-fold more expressed than miR-96 in the NTL and HCC tissues, respectively (Fig. 3D). Therefore, we concluded that, compared to miR-1271, the much lower expression of miR-96 in both NTL and HCC prevents drawing conclusions concerning its regulatory role in GPC3 overexpression. Altogether, our data suggest that the overexpression of GPC3 mediated by the posttranscriptional up-regulation of its 3′-UTR (Fig. 1B) results at least partly from the marked underexpression of miR-1271 in HCC cells compared to NTL.

MiR-1271 Directly Targets GPC3 3′-UTR and Accelerates Its mRNA Degradation.

Based on the above results, we focused on miR-1271. We first assessed the specificity of the miR-1271-mediated effect on GPC3 expression. As shown in Fig. 4A, miR-1271 down-regulated GPC3-protein expression in HuH7 cells by roughly 30%. Its specific antimiR (AM1271) abrogated miR-1271-induced effects and led to a slight increase in GPC3-protein expression. MiR-1271-mediated effects on GPC3 expression were also observed at the mRNA level (Fig. 4B). Using FunREG (Fig. 4C-E), we further showed that miR-1271 acts as a GPC3-mRNA destabilizing factor. Similar results were obtained with SNU398 cells (data not shown). Finally, introduction of mutations in the sequence complementary to miR-1271 seed (“mut GPC3” transgene) or the base complementary to its 8th base (“G>U GPC3” transgene) completely abolished miR-1271 activity (Fig. 4F). These data demonstrated that miR-1271 directly targets GPC3 3′-UTR and induces GPC3 mRNA degradation through a seed+base 8 recognition, as does its paralog miR-96.18

Figure 4.

Molecular basis of the posttranscriptional regulation mediated by miR-1271 on GPC3. (A,B) Relative expression of GPC3 protein (A) and mRNA (B) in cells transfected with the indicated RNAs (ANOVA: P < 0.0001; n = 5). A representative western blot experiment is shown on top in (A). (C-E) GFP-GPC3-expressing HuH7 cells (Fig. 1A; TCN value known) were transfected with the indicated RNAs. Then the GFP protein (P) and mRNA (M) expressions were analyzed following the FunREG method.13 (C) Global posttranscriptional regulation (ANOVA: P < 0.0001; n = 5). (D) mRNA stability (ANOVA: P < 0.0001; n = 5). (E) Translation efficiency (ANOVA: P = NS; n = 5). (F) HuH7 cells expressing the indicated transgene (TCN value measured) were transfected with the indicated RNAs and analyzed as described in Fig. 1B (ANOVA: P < 0.0001; n = 3). Top panel shows miR-1271 pairing with the GPC3 3′-UTR sequence in its wildtype or mutated versions. *P < 0.05; **P < 0.01; ***P < 0.001.

MiR-1271 Inhibits HCC Cell Growth in a GPC3-Dependent Manner and Induces Apoptosis.

Because GPC3 promotes HCC cell growth,6, 7 we wondered whether miR-1271, which represses GPC3 expression, could have antitumor effects on these cells. MiR-1271 overexpression strongly inhibited the growth of HuH7 cells by decreasing their proliferation and adenosine triphosphate (ATP) content, which is an indicator of cell metabolic activity (Fig. 5A-C). Using a variety of assays, we further found that miR-1271 slightly increased the percentage of cells in G0/G1 phase and decreased that in S phase (Supporting Fig. S2). Concomitantly, miR-1271 markedly induced apoptosis (Fig. 5D,E) and necrosis (Supporting Fig. S3). Importantly, all these effects were abrogated by the concomitant transfection of its anti-miR AM1271. Counteracting the activity of the endogenous miR-1271 with AM1271 stimulated the growth and proliferation of HCC cells (Fig. 5A,B). Because AM1271 had no significant effects on cell cycle phases, nor cell death (Fig. 5D,E; Supporting Figs. S2, S3), we hypothesized that AM1271 increases the proliferation of HCC cells by shortening their division time. This model is supported by the higher ATP content of HCC cells transfected with AM1271 (Fig. 5C). Some of these results were obtained with the SNU398 cells (Supporting Fig. S4). Interestingly, ectopic expression of a GPC3 transgene lacking its 3′-UTR in HuH7 cells overexpressing miR-1271 partially rescued the miR-1271-mediated inhibition of HuH7 cell growth (Fig. 5F). Altogether these results demonstrated that miR-1271 inhibits the growth of HCC cells and promotes their death, at least for a part, by lowering GPC3 expression. We therefore concluded that the specific down-regulation of miR-1271 in HCC tumors contributes to GPC3 overexpression and expansion of cancerous hepatic cells.

Figure 5.

MiR-1271 inhibits HCC-cell growth and induces cell death. In this figure, HuH7 cells were transfected by small RNAs and/or plasmids (expressing or not GPC3) as indicated in each panel. (A,F) Cell growth (total cellular proteins; ANOVA: P < 0.0003; n = 6). (B) Cell proliferation (ANOVA: P = 0.0001; n = 3). (C) ATP content measurement (ANOVA: P = 0.0001; n = 5). (D) Cell apoptosis (annexin V-positive cells; ANOVA: P = 0.0001; n = 3). (E) Caspases 3/7 activity (ANOVA: P = 0.0055; n = 3). A representative western blot experiment is shown on the bottom in (F). *P < 0.05; **P < 0.01; ***P < 0.001.

Discussion

In this work we report that GPC3 is posttranscriptionally regulated through its 3′-UTR. More important, we show that the regulation mediated by GPC3 3′-UTR favors GPC3 expression in HCC cells compared to primary hepatocytes (Fig. 1B). These results further support the notion that posttranscriptional regulations are part of the events participating in HCC-associated gene deregulations and indirectly in liver carcinogenesis.5, 13

Given the role of miRNAs in tumorigenesis5, 11, 22 and the relevance of GPC3 in HCC,6, 7 we investigated their implication in the deregulation of GPC3 expression mediated by its 3′-UTR in HCC cells. Because of the poor accuracy of bioinformatic predictions,18, 20 we opted for a blinded screening strategy in order to systematically identify miRNAs regulating GPC3 expression. We developed a dual reporter system deriving from FunREG,13, 16 which we termed DF-FunREG. The main difficulty encountered with miRNA screening is the pleiotropic effects of miRNAs on multiple gene targets and their subsequent cellular impacts on cellular processes (cell survival, growth, death). This led us to pay particular attention to eliminate as much as possible false-positive events. We first used a library of mature miRNAs in order to avoid problems linked to the differential processing of precursor miRNAs from a cell type to another one.24, 25 Second, by performing analyses 3 days after transfection, we minimized the effects due to miRNA overexpression on cell integrity (growth arrest, cell death) and therefore decreased the risk of losing miRNA candidates. Third, by establishing cell lines expressing two fluorescent reporter transgenes with a high degree of similarities (Fig. 2A, top panel), we get rid of false-positive events due to miRNAs either affecting transgene promoter activity by modulating expression of general transcription factors, or targeting the transgenes outside the sequence of interest. With these stringent conditions, we could not avoid eight false-positive events that were revealed by performing a secondary screen with HuH7 cells expressing the TOM and a GFP transgene lacking the GPC3 3′-UTR (Fig. 2A,C).

Using DF-FunREG as a three-step strategy including two successive screens and a set of validation experiments (see Fig. 2A), we tested the biological effect of 876 mature miRNAs and identified five miRNAs, which all regulate GPC3 expression in a 3′-UTR-dependent manner (Fig. 6). We therefore concluded that DF-FunREG is a robust and accurate screening approach for identifying functional miRNAs. To our knowledge, this is one of the most exhaustive miRNA screens described so far using a gene-reporter system.20, 26, 27 Among the five miRNAs, two down-regulated GPC3 expression: miR-96 that we previously described as a negative GPC3 regulator18 and its paralog miR-1271.23 As expected, both miRNAs interacted with the GPC3 mRNA through its 3′-UTR at the expected seeding 5′-GUGCCAA256-3′ site. We further showed that miR-96/1271 repress GPC3 expression by triggering the decay of GPC3 mRNA (Fig. 4).18

Figure 6.

Schematic representation of the five miRNAs differentially regulating GPC3 expression through its 3′-UTR.

The presence of miR-96 among the selected miRNAs clearly demonstrated the reliability of our screening strategy and its specificity. Unexpectedly, none of the numerous other miRNAs predicted to target the GPC3 3′-UTR were found in our screen.18, 21 As highlighted elsewhere,18, 20, 28, 29 such discrepancies could partly be explained by the high error rate of prediction programs. Alternatively, the capacity of some miRNAs from the library to target GPC3 3′-UTR could depend on the cellular context (presence of other miRNAs and/or RNA-binding proteins). Interestingly, we also identified miR-129-1-3p, miR-1291, and miR-1303 as positive and 3′-UTR-dependent regulators of GPC3 expression (Fig. 6). Based on miRWalk,21 none of these three miRNAs were predicted to interact with GPC3 3′-UTR. It is thus likely that these miRNAs operate indirectly, by down-regulating the expression of negative posttranscriptional regulators of GPC3 mRNA (Fig. 6). Of interest, several RNA-binding proteins and ribonucleases involved in mRNA degradation are potential targets of these three miRNAs. We are currently working at understanding how these three miRNAs positively control GPC3 expression. Finally, our work contributes to highlight the fascinating notion of miRNA-associated systems biology.19 In our opinion, systematic and blinded screening approaches such as DF-FunREG described here should, in the future, greatly contribute to the understanding of “miRNA:target” networks, as well as the individual or collective role of miRNAs in gene alteration and human pathologies.

In the context of HCC, we found that miR-1271 expression is reduced in 92% of the HCC samples analyzed compared to NTL. It would therefore be of interest to understand the molecular origin of this underexpression in HCC. Interestingly, miR-1271 expression inversely correlated with that of GPC3 mRNA in the HBV-HCC subgroup, a result consistent with the regulatory effect of miR-1271 on the GPC3 transcript (Fig. 4). Up to now, it is not clear why this correlation is only observed in this particular subgroup, but context-dependent correlations have been reported for miR-21 and Serpini1 in gastric cancer.30 This could be due to the low number of samples in other subgroups, the differentiation state of these particular tumors, the influence of HBV, or a dominant role of miR-1271 over other trans-acting regulatory factors in this subgroup.2, 3 Our work also revealed that in the G1-G3 HCC subgroups (which highly express GPC3; Fig. 3A), miR-1271 down-regulation was associated with an increase in plasma alpha-fetoprotein (AFP) (Supporting Fig. S5A,B), a major biomarker of HCC in clinics. Interestingly both AFP and GPC3 are oncofetal proteins associated with the undifferentiated state and aggressiveness of hepatic tumoral cells.3, 8, 9 This suggests that miR-1271 expression could be linked to the differentiation status of hepatic cells. In terms of expression, we showed that miR-1271 is rather abundant in liver (Fig. 3D). These results are somewhat surprising compared to others,23 but these discrepancies could be explained by the different approaches used to quantify miRNA amounts in liver tissues. Finally, we showed that miR-1271 acts as an antiproliferative and proapoptotic factor in HCC cells (Fig. 5). These results are in accordance with the tumor suppressive role of its paralog miR-96 reported in pancreatic cancer.31 More important, we demonstrated that the antiproliferative effect of miR-1271 on HCC cells depends, at least partly, on GPC3 expression, reinforcing the notion that GPC3 is a key tumor driver gene in liver.6, 7

In conclusion, we identified five miRNAs controlling GPC3 expression in HCC cells using a functional screening approach. Among them, miR-1271 was found reproducibly underexpressed in HCC and displayed antitumoral properties in cellulo. MiR-1271 overexpression has been reported in head and neck cancers.32 However, this is the first demonstration of its direct implication in human disease. Collectively, our data suggest that the underexpression of miR-1271 participates in the overexpression of GPC3 in HCC and favors HCC cell expansion.

Acknowledgements

We thank Véronique Guyonnet-Duperat, Yuh-Shan Jou, and Roger Y. Tsien for providing experimental materials.

Ancillary