Hepatocellular carcinoma (HCC) is the most frequent primary liver cancer. Despite the discovery of several carcinogenetic pathways, the mechanisms of liver carcinogenesis are still unclear.1 It is also apparent that the underlying mechanisms may be different according to the etiology of the HCC.2, 3 Because of its high incidence in occidental countries, hepatitis C is responsible for a strong increase in HCC incidence. Like several other groups,4, 5 we have recently applied proteomics technology to discover new markers or targets in hepatitis C virus (HCV)–related HCC. This was done with 2-dimensional electrophoresis and mass spectrometry and led to the discovery of many regulated proteins,6 including the recently uncovered protein RuvB-like 2 (RUVB2), which is also known as TIP49b,7 TIP48,8 Reptin52,9 Rvb2,10 TAP54β,11 ECP-51,12 and TIH2p.13 RUVBL2 contains the Walker A and Walker B motifs, which are found in proteins that bind and hydrolyze adenosine triphosphate.14 RUVBL2 also has limited homology to the bacterial RuvB adenosine triphosphate–dependent DNA helicase. Human RUVBL2 has indeed been shown to have ATPase activity,7, 11 whereas its DNA helicase activity has been disputed.7, 11 RUVBL2 is found in several nuclear high-molecular-weight chromatin-remodeling complexes.10, 11 Recently, it has been found that RUVBL2 is able to interact with β-catenin, resulting in altered T-cell factor/lymphoid enhancer-binding factor–mediated transcription.9, 15 RUVBL2 also interacts with a domain of cellular v-myc myelocytomatosis viral oncogene homolog of the manuscript (c-myc) required for oncogenic transformation,8 and its expression is controlled by c-myc itself.8, 16 Given that β-catenin and c-myc are 2 key actors of liver carcinogenesis, our study was designed to assess the role of RUVBL2 in liver carcinogenesis.
Using a proteomic analysis of human hepatocellular carcinoma (HCC), we identified the overexpression in 4 tumors of RuvB-like 2 (RUVBL2), an ATPase and putative DNA helicase known to interact with β-catenin and cellular v-myc myelocytomatosis viral oncogene homolog (c-myc). RUVBL2 expression was further analyzed in tumors with quantitative reverse-transcription polymerase chain reaction analysis and immunohistochemistry; in addition, RUVBL2 expression in a HuH7 cell line was silenced by small interfering RNA or increased with a lentiviral vector. RUVBL2 messenger RNA overexpression was confirmed in 72 of 96 HCC cases, and it was associated with poorly differentiated tumors (P = 0.02) and a poor prognosis (P = 0.02) but not with β-catenin mutations or c-myc levels. Although RUVBL2 was strictly nuclear in normal hepatocytes, tumoral hepatocytes exhibited additional cytoplasmic staining. There was no mutation in the coding sequence of RUVBL2 in 10 sequenced cases. Silencing RUVBL2 in HuH7 HCC cells reduced cell growth (P < 0.001) and increased apoptosis, as shown by DNA fragmentation (P < 0.001) and caspase 3 activity (P < 0.005). This was associated with an increased expression of several proapoptotic genes and with an increased conformational activation of Bak-1 and Bax. On the other hand, HuH7 cells with an overexpression of RUVBL2 grew better in soft agar (P < 0.03), had increased resistance to C2 ceramide–induced apoptosis (P < 0.001), and gave rise to significantly larger tumors when injected into immunodeficient Rag2/γc mice (P = 0.016). Conclusion: RUVBL2 is overexpressed in a large majority of HCCs. RUVBL2 overexpression enhances tumorigenicity, and RUVBL2 is required for tumor cell viability. These results argue for a major role of RUVBL2 in liver carcinogenesis. (HEPATOLOGY 2007.)
Materials and Methods
Samples came from resected or explanted livers with HCC of patients treated in Bordeaux from 1992-2005. Fragments of fresh tumoral and nontumoral liver tissues (taken at a distance of at least 2 cm from the tumor) were immediately snap-frozen in liquid nitrogen and stored at −80°C. DNA and RNA were extracted as described.17 The proteomic study included 4 male patients (mean age = 77, range = 72-82) with HCC developed on chronic viral hepatitis C (stage F3 for 3 patients and stage F2 for 1 patient according to the METAVIR classification18). In addition, HCC samples from 96 patients, related to various etiologies, were used for quantitative reverse-transcription polymerase chain reaction (RT-PCR) analysis and immunohistochemistry. The main clinical and pathological characteristics are indicated in Table 1. Six normal liver samples taken at a distance from benign liver tumors were used as controls in quantitative RT-PCR and immunohistochemistry. The survival analysis included 45 patients treated with a complete tumor resection before 2002; patients that were treated with liver transplantation or incomplete resection or died in the postoperative period were excluded.
|Age (mean ± standard deviation)||64.6 ± 10.6 years|
|Alpha fetoprotein < 20 ng/ml||63%|
|Cirrhosis or chronic active hepatitis||60%|
|Diameter (mean ± standard deviation)||7.1 ± 5 cm|
|Edmondson grade III or IV||57%|
|Microscopic vascular embolus||36%|
Four tumoral samples were compared to matched nontumor samples with 2-dimensional electrophoresis as described in detail.6
One microgram of the total RNA was reverse-transcribed with a high capacity archive kit and random hexamers (Applied Biosystems), and 5 μl of complementary DNA (cDNA) corresponding to 10 ng of reverse-transcribed RNA was analyzed by TaqMan polymerase chain reaction (PCR) analysis, in duplicate, with the ABI-Prism 7900HT system (Applied Biosystems). Predeveloped sequence detection reagents specific for the human RUVBL2 gene (Applied Biosystems) were used as described17 according to the 2−ΔΔCT method.19 Briefly, the gene expression results were first normalized to internal control ribosomal 18S. Then, the results for HCC samples were expressed as a ratio with respect to the mean expression level in nontumor samples. We also measured the messenger RNA (mRNA) level of expression for c-myc and glutamate-ammonia ligase (GLUL; encoding glutamine synthase) and G protein–coupled receptor 49 (GPR49; encoding an orphan nuclear receptor), 2 well-known β-catenin target genes, as described.20 Finally, mutations in the catenin β-1 (CTNNB1) gene coding for β-catenin were searched as described.2
The total RNA was extracted with Trizol reagent (Invitrogen). One microgram of the total RNA was reverse-transcribed with the Superscript II reverse transcriptase (Invitrogen) and random hexamers (Roche). Five microliters of cDNA corresponding to 5 ng of reverse-transcribed RNA was analyzed by SYBR Green PCR analysis with the Mx4000 Multiplex Quantitative PCR system (Stratagene). PCR mixes were made so that each 25-μl reaction contained 12.5 μl of iQ Sybr Green Supermix (Bio-Rad), 5 μl of diluted cDNA template, and 0.3 μM sense and antisense gene primers. Primer sequences are shown in Supplementary Table 1. The reactions were incubated at 95°C for 3 minutes, and this was followed by 40 cycles at 95°C for 15 seconds and at 65° for 15 seconds. Data analysis was performed with Mx4000 software (version 4.2, Stratagene). Gene expression results were first normalized to internal control RLP0. Relative levels of expression were quantified by the calculation of 2−ΔΔCt. The results were expressed as a ratio with respect to the expression level in nontransfected cells.
Twenty samples of HCC and their corresponding nontumoral livers were evaluated by immunohistochemistry, together with 6 normal liver samples. Formalin-fixed, paraffin-embedded sections were used. Following dewaxing and dehydration, sections were submitted to antigen retrieval with a steam cooker in a citrate buffer (0.01 M, pH 7) for 10 minutes and then permeabilized with methanol for 1 hour. The immunostaining procedure was carried out in an autostainer (Dako). Endogenous peroxidase was inhibited with 3% H2O2 in methanol, and the sections were treated with 2 N HCl for 30 minutes at 37°C. Nonspecific sites were saturated with 3% BSA, and the sections were incubated with a mouse monoclonal anti–RUVBL2 antibody (BD Biosciences) diluted to 1:50. After they were washed, the signal was amplified with the Envision reagent coupled to horseradish peroxidase. The staining was revealed with diaminobenzidine.
Construction of a Cell Line with an Overexpression of RUVBL2
We used the human HCC cell line HuH7. A Flag-tagged RUVBL2 cDNA was placed under the control of the MND (Myeloproliferative sarcoma virus enhancer, Negative control region deleted, d1587rev primer-binding site substituted) promoter as follows. The MND promoter (myeloproliferative sarcoma virus enhancer, negative control region deleted, dl587rev primer-binding site substituted) was removed from a lentiviral construct.21 pRRLSIN.cPPT-PGK-WPRE, a self-inactivating human immunodeficiency virus type 1–based vector containing the central polypurine and termination tract and the woodchuck hepatitis virus posttranscriptional regulatory element, was provided by D. Trono (Geneva, Switzerland). In short, we replaced the phosphoglycerate kinase (PGK) promoter with the MND promoter. Then, cDNA coding human RUVBL2 with an N-terminal FLAG tag8 was cloned downstream of the MND promoter, leading to the TMRepW vector. As a control, we constructed the TMEW vector by cloning the enhanced green fluorescent protein (EGFP) cDNA instead of the RUVBL2 sequence. Vesicular stomatitis virus G protein (VSV-G) pseudotyped lentivectors were produced by triple-transient transfection of 293T cells.22 Titers were determined by the transduction of 293T cells through the serial dilution of the lentiviral supernatant and were analyzed for EGFP expression 5 days later. Finally, HuH7 cells were transduced with lentiviral vectors at a multiplicity of infection of 20 and/or 10 in Dulbecco/Vogt modified Eagle's minimal essential medium (DMEM)/10% fetal bovine serum (FBS) medium with 8 μg/ml protamine sulfate (Sigma, St. Louis, MO). The infection efficiency was checked through testing for EGFP expression by flow cytometry.
Transient Transfection of Small Interfering RNA (siRNA)
We used siRNAs targeting 2 separate sequences from RUVBL2 mRNA. The sequence 5′-GATGATTGAGTCCCTGACCAA-3′ corresponds to nucleotides 552-572 from the reference sequence (gi:9367026) and is called siR1, whereas the sequence 5′-GAAGATGTGGAGATGAGTGAG-3′ corresponds to nucleotides 1153-1173 and is called siR2. As a control, we used a scrambled sequence from siR2, 5′-GGATGTAAGTGGGAAAGTGGA-3′. The corresponding oligonucleotides were bought from Eurogentec (Searing, Belgium) as duplexes and were transfected into HuH7 cells with Lipofectamine (Invitrogen, Cergy Pontoise, France) at a concentration of 125 nM.
Cell extracts and tissue samples were prepared in a radioimmunoprecipitation buffer as described.23 The protein concentration was measured with a Bio-Rad assay. The samples were analyzed by a western blot with either a rabbit polyclonal8 or mouse monoclonal antibody (BD Biosciences Pharmingen) to RUVBL2 (#612482) or an anti–Flag M2 antibody (Sigma). Normalization was achieved by rehybridization with an anti–β-actin antibody. Signals were acquired on a Macintosh computer with a Kodak Digital Science DC 120 digital camera and quantified with National Institutes of Health Image software.
The proliferation of cells transfected with siRNAs was assessed through the counting of adherent cells at various times with a Coulter counter (Beckman Coulter) in duplicate wells.
DNA synthesis was measured by the quantification of the incorporation of bromodeoxyuridine (BrdU). Briefly, cells were incubated for 2 hours with 15 mM BrdU. They were fixed with methanol, treated with 2N HCl, and incubated with a rat anti–BrdU antibody (Serotec) followed by an Alexa fluor 488–labeled secondary antibody (Invitrogen Molecular Probes). Nuclear DNA was stained with 4′,6-diamidino-2-phenylindole (DAPI). BrdU-positive cells were counted under a fluorescence microscope in at least 15 nonoverlapping fields, and the results were expressed as the percentage of the total cells.
We also measured the percentage of cells expressing the proliferation antigen Ki67.24 Briefly, cells were fixed with 0.4% paraformaldehyde, permeabilized with 0.02% saponin, then incubated with a fluorescein isothiocyanate–labeled Ki67 antibody (Dako), and analyzed by flow cytometry.
Anchorage-independent growth was assayed as follows. Flag-RUVBL2–expressing cells or control cells (10,000) were suspended in DMEM containing 10% FBS and 0.2% agar (overlay) on top of an agar underlay (DMEM/10% FBS/0.5% agar) in a 60-mm-diameter dish. Cells were fed twice a week with 2 ml of the overlay medium, and the area of the colonies was measured after 2 weeks with Image J software (http://rsb.info.nih.gov/ij/). Thirty different fields (5×) were scored. The experiments were performed in triplicate.
Assay for Cell Viability
Cells were seeded at a density of 6000 per well in 96-well plates. After 7 hours, they were treated with the indicated agonists for 17 hours. The cell viability was assessed by incubation with DMEM containing 1 mg/ml 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) for 2 hours at 37°C as described.25 The linearity of the assay over the range of cell densities was checked in preliminary experiments.
Cells were fixed with 4% paraformaldehyde in PBS at 4°C for 10 minutes and then permeabilized with 0.5% Triton X-100. They were stained with DAPI (Sigma; 0.5 μg/ml) together with a rabbit anti–human active caspase 3 antibody (R&D Systems; 0.25 μg/ml). After being washed, the cells were incubated with goat anti–rabbit Alexa 488 (Molecular Probes) and examined under a Zeiss Axioplan microscope.
The extent of DNA fragmentation was quantified with flow cytometry as described.26 Briefly, trypsinized cells were resuspended in 0.1% sodium citrate containing 0.1% Triton X-100, 50 μg/ml propidium iodide, and 100 μg/ml ribonuclease. The cell DNA content was measured with a FACSCalibur (BD Biosciences, San Jose, CA).
We also quantified the caspase 3 activity, using a colorimetric kit from Chemicon (Temecula, CA) according to the instructions from the supplier. Briefly, cleared cell extracts were incubated with the caspase 3 substrate, and the optical density was read at 405 nm in a Dynatech microplate reader (MTX Lab Systems, Inc., Vienna, VA).
Finally, we assessed Bak-1 (Bcl2-antagonist killer 1) and Bax (Bcl2-associated x protein) activation, using antibodies (Bak-1: AM03, Calbiochem, VWR International, Fontenay-sous-Bois, France; Bax: 6A7, BD Pharmingen, Le Pont de Claix, France) that recognize the N-terminus of the proteins made accessible following the conformational changes upon activation. This was done with flow cytometry as described by others.27, 28
Twelve-week-old Rag2/γc mice (a gift from J. Di Santo29) were housed under germ-free conditions. The study was performed in accordance with the European Community Standards on the Care and Use of Laboratory Animals. Mice were injected subcutaneously with 2 × 106 cells in the flank. The tumor size was monitored with calipers. The tumor volume was estimated as (D2 × d)/2, where D is the large diameter and d is the small diameter of the tumor. At the end of the experiment, the tumors were harvested, and a part was snap-frozen in liquid nitrogen.
Differences between means were assessed with the Mann-Whitney test. When comparing multiple means, we used an analysis of variance (ANOVA). A P value less than 0.05 was considered significant. Correlations between the mRNA level of expression and qualitative variables were calculated with the nonparametric Kruskal-Wallis test with STATA software (Stata Corp., College Station, TX). The ages of the patients and diameters of the tumors were partitioned with a median. Survival analyses were performed with the Cox model to fit maximum-likelihood proportional hazard models with STATA software.
RUVBL2 Expression in HCC
Proteome analysis was performed with 2-dimensional gel electrophoresis, comparing 4 HCCs related to HCV infection with their matched nontumor tissues. One of the 850 protein spots detected was increased in all 4 cases of HCC with a mean tumor/nontumor ratio of 4.1 (range = 2.8-6.9; Fig. 1A). It was identified as RUVBL2 on the basis of the detection of 2 peptides (GLGLDDALEPR and LLIVSTTPYSEK) and detailed tandem mass spectrometry analysis as described.6 The overexpression of RUVBL2 was confirmed by a western blot in these 4 cases (Fig. 1B).
The cellular localization of RUVBL2 was assessed by immunohistochemistry in 20 samples of HCC from various etiologies, in the corresponding nontumoral livers, and in 4 cases of histologically normal livers. In a normal liver, RUVBL2 expression was found in a variable fraction of hepatocytes, in which it was faint and restricted to the nucleus; in addition, RUVBL2 staining was intense in bile duct cells, in which it was found in the nucleus and cytoplasm, and was also observed in some lymphocytes, either sinusoidal or portal (Fig. 2A). The same staining pattern, in terms of the distribution and intensity, was seen in all samples of peritumoral livers, whether cirrhotic or not (Fig. 2C,F). In sharp contrast, RUVBL2 was overexpressed in all analyzed HCCs (Fig. 2C), in which it also showed definite cytoplasmic staining that varied from faint to intense (Fig. 2D,E); in addition, the nuclear staining of tumor cells was in most cases stronger than that of peritumoral hepatocytes (Fig. 2D,E). It should be noted, however, that the staining pattern was often heterogeneous within a given tumor (Fig. 2C). To search for detailed clinicopathological correlations, we measured RUVBL2 mRNA expression in a larger series of 96 well-annotated HCCs related to various risk factors. We first confirmed in this series the increased expression of RUVBL2 transcripts in 72 of 96 HCCs, ranging from 1.5-6.6, in comparison with that of nontumor liver tissues, which ranged from 0.42-1.5 (P = 0.002; Fig. 3A). RUVBL2 expression was significantly lower in hepatitis B virus (HBV)–related HCC (Fig. 3B), whereas it was significantly higher in poorly differentiated tumors proportionally to the Edmondson grade (Fig. 3C). In contrast, the other clinical and pathological characteristics mentioned in Table 1 were not significantly associated with the RUVBL2 expression level. Using a Cox model analysis without partition, we showed a significant relationship between a high level of RUVBL2 in tumors and shorter overall survival of patients following surgery (P = 0.02, relative risk = 1.4, standard error = 0.2, 95% confidence interval = 1.05-1.85). In contrast, we did not find a difference in the RUVBL2 expression in β-catenin–activated HCC in comparison with CTNNB1 nonmutated HCC (Fig. 3D). Accordingly, in the 96 HCC samples, we did not observe any correlation between the expression levels of RUVBL2 and β-catenin target genes (GLUL and GPR49, data not shown). In addition, the c-myc transcript levels were measured in a subset of 42 patients, and no correlation was found with the RUVBL2 levels (r2 = 0.01, P = 0.5, not shown).
Finally, the whole coding sequence of RUVBL2 was PCR-amplified in 10 HCC samples, and no mutation leading to a change in the amino acid sequence was identified (not shown).
Thus, because RUVBL2 overexpression was found in the vast majority of HCCs and was related to the aggressiveness of the tumors, we explored the role of RUVBL2 in tumorigenesis by modulating its expression in the HCC cell line HuH7.
Down-Regulation of RUVBL2 Expression Reduced Cell Growth and Increased Apoptosis
The down-regulation of RUVBL2 expression was obtained with transient transfections of siRNA duplexes. Quantitative RT-PCR showed that as soon as 24 hours after transfection, RUVBL2 transcripts were reduced to 33.5% ± 9.8% with siR1 and 29.5% ± 14.4% with siR2 versus 104.4% ± 15.4% with the scrambled sequence (n = 6, P = 0.02). In addition, as shown in Fig. 4A, the 2 targeting siRNAs led to a large decrease in RUVBL2 protein expression that reached more than 90% with siR1 and almost 80% with siR2. In both cases, the decrease was stable for at least 6 days.
We next evaluated the effect of RUVBL2 down expression on cell growth. Following a slight reduction in the cell numbers observed at days 3-4 after transfection with either scrambled or RUVBL2 siRNAs (siR1 and siR2), a marked and significant reduction in the cell numbers was observed in siR1-transfected and siR2-transfected cells until day 6 in comparison with nontransfected or scrambled-transfected cells (P < 0.001; Fig. 4B); the reduction was more drastic with siR1 than siR2. This reduction was linked to decreased cell proliferation, as evidenced by the measure of BrdU incorporation: at day 4 after transfection, there was an average of 37.0% ± 1.5% labeled cells with the scramble oligonucleotide versus 21.7% ± 1.0% with siR1. The mean decrease with siR1 was 41.0% ± 4.4% (n = 3, P < 0.04). Similarly, the percentage of cells positive for Ki67 was significantly decreased by both siR1 and siR2 (Fig. 4C).
In parallel, we observed that the down regulation of RUVBL2 expression was associated with a high number of cells floating in the supernatant (not shown), and this suggested that the cells may undergo apoptosis. This was confirmed by the staining of nuclear DNA with DAPI, which showed characteristic pictures of fragmented nuclei in RUVBL2 siRNA-transfected cells (Fig. 4E) in comparison with scrambled ones (Fig. 4D). The simultaneous staining of the cells with an antibody specific for activated caspase 3 demonstrated that many cells transfected with RUVBL2 siRNAs showed an activation of this apoptotic pathway (compare Fig. 4F,G). DNA fragmentation was quantified with flow cytometry, which showed a significantly higher percentage of fragmented DNA at day 6 after transfection in cells transfected with siR1 or siR2 than in scrambled-transfected cells (ANOVA, P < 0.001; Fig. 4J). Similarly, the quantification of the caspase 3 activity with a colorimetric assay showed significantly increased activity with both siR1 and siR2, as shown in Fig. 4K (ANOVA, P < 0.005).
To gain insight into the mechanisms of apoptosis, we measured the activation of the proapoptotic Bcl-2 (B-cell lymphoma protein-2) family members Bax and Bak-1. These 2 proteins undergo a conformational activation that can be evidenced with conformation-sensitive antibodies in flow cytometry. As shown in Fig. 5A,B, the number of cells showing activation of either Bax or Bak-1 was significantly increased upon the down-regulation of RUVBL2. In addition, we also show a significantly increased expression of the transcripts of Bak-1 and the other proapoptotic proteins, Bad (Bcl2 antagonist of cell death) and Bid (BH3-interacting domain death agonist), at days 3 and 4 after the transfection of RUVBL2 siRNA and of Bcl-xS at day 4. There was also a trend of an increase for Bax and PUMA, whereas no changes were seen for the antiapoptotic gene Bcl-xL (Fig. 5C,D).
Up-Regulation of RUVBL2 Expression Promotes In Vitro Anchorage-Independent Cell Growth and In Vivo Tumorigenesis
Here, Flag-tagged RUVBL2 was overexpressed under the control of the MND promoter in a lentiviral vector. Flag-tagged RUVBL2 could be easily detected and discriminated from endogenous RUVBL2 on western blots because of its slower gel mobility. The factor of the overexpression of the RUVBL2 protein, assessed from the sum of the intensities of the endogenous and Flag-tagged RUVBL2 bands, was only 1.2 in 2 independent transductions. This was due at least in part to a decrease in endogenous RUVBL2 levels in Flag-RUVBL2–infected cells (Fig. 6A). Several preliminary experiments allowed us to check that Flag-tagged RUVBL2 behaved similarly to endogenous RUVBL2: a western blot of nuclear and cytosolic extracts showed that both forms segregated in the same way in these 2 compartments. The immunoprecipitation of Flag-tagged RUVBL2 pulled down expected partners (RUVBL1 and β-catenin9), and this indicated that the tagged protein was correctly incorporated into supramolecular complexes (data not shown).
The growth rate on plastic of Flag-RUVBL2–expressing cells was compared to that of control EGFP-transfected cells. No differences in the cell numbers were seen after 1, 2, or 3 days in serum concentrations ranging from 0.5%-10% (data not shown). However, when anchorage-independent growth was measured in a soft agar assay, we found a large difference in the growth ability of Flag-RUVBL2–expressing cells, as shown by a significant increase in the surface of the colonies (Fig. 6B; P < 0.03, Kruskal-Wallis).
In subsequent experiments, HuH7 cells were challenged with an apoptotic stimulus, C2 ceramide,30 and cell survival was assessed with the MTT assay. At every concentration of C2 ceramide tested, cells expressing Flag-RUVBL2 were slightly less susceptible to cell death than control cells. The difference was statistically significant, except for the highest concentration of C2 ceramide, at which most cells were dead (Fig. 6C). Furthermore, when the extent of DNA fragmentation, as evidence for apoptosis, was examined, cells expressing Flag-RUVBL2 exhibited significant protection over a range of C2 ceramide concentrations in comparison with control cells (Fig. 6D). The fractional reduction in apoptosis in comparison with control cells was 27.2% ± 6.0%, 19.3 ± 8.5%, and 28.7% ± 3.3% at 20, 40, and 60 μM C2 ceramide, respectively.
Finally, cells transduced with either EGFP or Flag-tagged RUVBL2 were injected into immunocompromised mice (n = 20 in each group). Although there was no significant difference in the incidence of tumors in the two groups (85.0% versus 94.7% in the EGFP and RUVBL2 groups, respectively), animals injected with Flag-RUVBL2–expressing cells developed significantly larger tumors than control animals (Fig. 6E; P = 0.016 by ANOVA). Similar results were obtained in another experiment using cells from an independent transduction with the viral vectors. When, at the end of the experiment, tumors from surviving animals were removed and analyzed by a western blot, the persistence of Flag-RUVBL2 expression was confirmed in every tumor in the Flag-RUVBL2 group by the detection of a doublet upon blotting with the RUVBL2 antibody and by the detection of the Flag epitope (Fig. 6F). This also showed the persistent decrease in endogenous RUVBL2.
Here we show that RUVBL2 is up-regulated in a majority of HCCs. This overexpression is significantly associated with the differentiation grade of the tumors, and RUVBL2 overexpression can be considered a marker of poor prognosis. Moreover, our results strongly suggest that RUVBL2, a protein with no specific function yet ascribed in humans, may play an essential role in liver carcinogenesis. To the best of our knowledge, we are the first to report the deregulation of RUVBL2 expression in cancer. Our immunohistochemical data indicate that hepatocyte RUVBL2 expression appears to be increased in every HCC studied, although to various extents. It is likely that tumor/nontumor ratios obtained with RT-PCR are artifactually low because of the high-level expression of RUVBL2 seen in the ductular proliferation of the nontumoral, cirrhotic livers (Fig. 2F).
Cells in which RUVBL2 was down-regulated showed a decreased proliferation rate, which was concomitant with a strong induction of apoptosis, as evidenced by the occurrence of DNA fragmentation and increased caspase 3 activity. In addition to HuH7, we observed a similar phenotype in 2 other HCC cell lines, Hep3B and FOCUS (Friendship of China and United States, not shown). Thus, RUVBL2 deficiency leads to liver cancer cell growth arrest and to apoptosis. RUVBL2 was previously shown to be also required for the growth and survival of yeast.7, 31, 32 Our data have shed some light on the mechanisms of apoptosis induced by RUVBL2 deficiency. We show that RUVBL2 down-regulation is associated with the increased conformational activation of the proapoptotic proteins Bak-1 and Bax. In quiescent cells, Bak-1 and Bax are maintained in an inactive state through interactions with the antiapoptotic proteins Bcl-xL and myeloid cell leukemia-1 (mcl-1).33 Their activation leads to the permeabilization of the outer membrane of mitochondria, resulting in cytochrome c release and caspase activation. In addition to protein activation, we also found that Bak-1 and Bax transcripts were up-regulated, together with those of Bid, a BH3 (Bcl2 homology domain 3)-only protein involved in Bak-1 and Bax activation,34, 35 whereas interestingly, Bcl-xL levels did not increase. These changes were seen as early as day 3, well before the occurrence of cell death, and this suggests that they may play a causal role. We thus propose that apoptosis induced by RUVBL2 depletion is linked to the activation of proapoptotic Bcl-2 family members. Such a mechanism could be amplified because of the transcriptional up-regulation of these proteins. Whether the related genes are direct targets of RUVBL2 remains to be determined.
We then transduced cells with a lentiviral construct driving the expression of Flag-RUVBL2. This led to only a modest increase in the total expression of RUVBL2, which was not the result of a low level of transduction or of insufficient promoter activity, because the RUVBL2 mRNA levels were 7.5 ± 2.2 times higher in transduced cells in comparison with control cells (not shown). The posttranscriptional regulation responsible for this is currently under study. Despite this modest 20% increase, Flag-RUVBL2 cells grew better in soft agar than control cells. Moreover, when Flag-RUVBL2–expressing cells were injected into immunocompromised mice, tumors grew eventually to a larger size. The mechanism for this effect remains to be evaluated in more depth but may be partly related to a better ability of Flag-RUVBL2–expressing cells for survival under stressful conditions. We indeed found that Flag-RUVBL2–expressing cells were less susceptible to cell killing induced by the proapoptotic agent C2 ceramide. Although the protection was limited, it was in the range of 20%-30% at various concentrations of C2 ceramide and thus commensurate with the 20% overexpression of RUVBL2.
RUVBL2 is known to physically interact with β-catenin.9 Several publications have shown that RUVBL2 behaves as an antagonist of the transcriptional activity of the β-catenin–T-cell factor/lymphoid enhancer-binding factor complex.9, 15 Because β-catenin is considered an oncogene in HCC, these results appear counterintuitive. However, recently, RUVBL2 was also found to be required for the repressing effect of β-catenin on the transcription of the antimetastatic gene KAI-1,36 which is frequently down-regulated in HCC.37 In our study of a large number of HCC samples, there were, however, no correlations between the levels of expression of RUVBL2 and β-catenin target genes. It is thus likely that additional mechanisms, independent of β-catenin, may also explain the oncogenic properties of RUVBL2. Indeed, RUVBL2 has been found in supramolecular complexes that do not include β-catenin,11 and β-catenin–independent functions of RUVBL2 have already been identified.15 It is also noteworthy that RUVBL2 is required for the viability of yeast,7, 31, 32 an organism devoid of β-catenin. In this context, 1 of the salient findings of this study was that in addition to its overexpression in HCC, RUVBL2 was also found in the cytoplasm of HCC cells. This suggests that RUVBL2 serves specific cancer-related functions in the cytoplasm of tumor cells.
We have thus identified a new protein up-regulated in HCC. In vitro and in vivo results suggest that RUVBL2 could play an important role in liver carcinogenesis. Because mutations in the ATPase domains of RUVBL2 abolish its biological activity,7, 31, 32 the targeting of its activity with specific inhibitors may offer a new therapeutic avenue in cancer management.
We thank Marc Bonneu for proteomic analysis; Christophe Laurent for providing human liver samples; Ivan Bièche for help with the primer selection for real-time PCR; and Vincent Pitard, Nathalie Dugot-Senant, and Gaëlle Cubel for their precious technical help.