Adenosine triphosphatase pontin is overexpressed in hepatocellular carcinoma and coregulated with reptin through a new posttranslational mechanism^†

Liver Samples and Real-Time Reverse-Transcription Polymerase Chain Reaction (RT-PCR).

Samples came from resected or explanted livers with HCC of patients treated in Bordeaux from 1992 to 2005. Fragments of fresh tumor and nontumor liver tissues (taken at a distance of at least 2 cm from the tumor) were either snap-frozen in liquid nitrogen and stored at −80°C or fixed with formalin and embedded in paraffin. In all, 104 HCC samples (Supporting Table 1) were used for real-time RT-PCR analysis. Eighteen nontumor liver samples were used as a control group. RNA extraction and real-time RT-PCR were performed as described.2, 20 Predeveloped sequence detection reagents specific for human RUVBL1 gene (Applied Biosystems, Courtaboeuf, France) were used as described20 using the 2^–ΔΔCT method.21 Gene expression results were normalized to internal control ribosomal 18S.

Immunohistochemistry.

This was done as described2 using a mouse monoclonal Pontin antibody22 diluted to 1.6 μg/μL.

Transient Transfection of Small Interfering RNA (siRNA).

We used two targeting siRNAs for each Reptin (R1 and R22) and Pontin messenger RNAs (mRNAs) (P1 and P2, Supporting Table 2). Controls were either scrambled R2 and P2 sequences, or the GL2 siRNA targeting Firefly luciferase (MWG, Ebersberg, Germany). siRNAs were transfected at a concentration of 125 nM with lipofectamine (Invitrogen, Cergy Pontoise, France).

Cell Proliferation Assay and Caspase 3 Activity Measurement.

Cells were counted with a Coulter counter (Beckman Coulter, Villepinte, France) in duplicate wells. DNA synthesis was measured by the quantification of bromodeoxyuridine (BrdU) incorporation, and caspase 3 activity with a colorimetric assay.2

Western Blot.

Cell extracts were prepared in RIPA buffer.23 We used Reptin, β-catenin (BD Biosciences, Pharmingen, Le Pont de Caix, France), FLAG-M2, β-actin (Sigma-Aldrich, Saint-Quentin Fallavier, France) mouse monoclonal, and Pontin rabbit polyclonal (ProteinTech, Chicago, IL) antibodies. Primary antibodies were detected by horseradish peroxidase-conjugated or infrared dye-labeled secondary antibodies (LI-COR, Lincoln, NE). Detection was achieved with the ECL kit (GE Healthcare, Saclay, France) or the Odyssey IR imaging system (LI-COR).

Polyribosome Fractionation.

KGL2 and KR2 cells lines stably expressing a short hairpin RNA (shRNA) targeting Firefly luciferase or Reptin, respectively, in a doxycycline-dependent manner (Supporting Methods), were cultured with or without doxycycline. Cycloheximide (100 μM) was added 10 minutes before cell lysis. Sucrose-gradient fractionation and polysome-associated RNA purification were as described.24 RNAs were analyzed by complementary DNA (cDNA) synthesis and PCR amplification.

Metabolic Labeling and Immunoprecipitation.

KGL2 and KR2 cells were stably transduced with a lentiviral vector coding hemagglutinin (HA)-Pontin resulting in the KGL2-HAP and KR2-HAP cell lines (Supporting Methods). Similarly, KP2 cells expressing the P2 Pontin shRNA (Supporting Methods) were transduced with Flag-Reptin,2 resulting in KP2-FR cells. Cells were incubated in methionine/cysteine-free medium for 30 minutes before pulse labeling with 150 μCi/mL EXPRE³⁵S³⁵S Protein Labeling Mix (Perkin Elmer, Courtaboeuf, France) for 15 minutes at 37°C. Cells were washed, then scraped in lysis buffer (50 mM Tris-HCl pH 7.5, 150 mM NaCl, 4 mM EDTA [ethylene diamine tetraacetic acid], 1% Triton, 1% SDS) supplemented with protease inhibitor cocktail (Roche, Meylan, France). The amount of radiolabeled TCA-precipitated material was measured by scintillation counting.

For pulse-chase experiments, cells were labeled with 100 μCi/mL EXPRE³⁵S³⁵S Protein Labeling Mix for 1 hour, washed with medium supplemented with 2 mM cysteine/methionine, and cultured for various times in this chase medium supplemented with the indicated agents before harvesting.

Cell extracts were diluted in lysis buffer without SDS and incubated for 2 hours at 4°C with monoclonal anti-HA-Agarose or anti-Flag M2 beads (Sigma-Aldrich). The beads were washed five times with lysis buffer and eluted with Laemmli sample buffer. Eluates were separated on 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). In some cases, Pontin was precipitated with a polyclonal anti-Pontin antibody.

Signals from radiolabeled protein bands were acquired in an Instant-Imager (Packard, Perkin-Elmer). The protein degradation rate is expressed as half-life (t_1/2), the time where 50% of the protein is degraded. The results are expressed as mean ± standard error of the mean (SEM) of five to six independent determinations.

Statistical Analysis.

Statistical analyses were performed using the 2-tailed Student's t test or one-way analysis of variance (ANOVA) when comparing multiple means. Statistical significance was set at P less than 0.05. Correlations between mRNA levels of expression and qualitative variables were calculated with the nonparametric Kruskal-Wallis test with STATA software (College Station, TX). The ages of patients and diameters of the tumors were partitioned with a median. For survival analysis, expression data were logarithm 10-transformed and dichotomized according to the median. Survival comparisons were done by log-rank test. Survival curves were obtained by the Kaplan-Meier method. Multivariate analyses were performed using the Cox model. All survival analyses were performed using STATA software v. 8.2.

Pontin Is Overexpressed in Human HCC.

There was a 3.9-fold increase in Pontin transcripts in HCC as compared to nontumor livers (Fig. 1A, P < 0.0004). Pontin mRNA levels were highly correlated with those of Reptin mRNA (r² = 0.78, P < 0.0001, Fig. 1B). Immunohistochemistry confirmed the overexpression of Pontin in HCC with the same pattern as Reptin, and showed that, similar to Reptin,2 Pontin was partly localized to the cytoplasm of tumor cells (Fig. 1C-E). Pontin mRNA levels were correlated with features of poor prognosis: levels were higher in large tumors (>60 mm, P = 0.026), in tumors with vascular embolism (P = 0.047), portal invasion (P = 0.008), or chromosomal instability (fractional allelic loss >0.128, P = 0.02), in patients with early relapse following curative surgery (P = 0.001), or who died within 3 years of surgery (P = 0.001). A significantly shorter disease-free survival in patients with a high Pontin level was observed as compared to those with a low level (Fig. 1F, P = 0.0022). Most important, using a Cox model multivariate analysis, high Pontin mRNA levels remained associated with shorter disease-free survival after adjustment for age, gender, tumor diameter, and the presence of vascular embolism (hazard ratio = 3.9, 95% confidence interval = 1.9-8.0, P = 0.003).

Pontin Silencing Reduced Cell Proliferation and Induced Apoptosis.

Pontin mRNA level was significantly reduced with P1 (70% ± 13%) and P2 (77% ± 12%) siRNAs (Fig. 2A). The siRNAs also greatly decreased Pontin protein expression (78.6% ± 6.9% with P1 and 87.7% ± 9.0% with P2) (Fig. 2B).

Starting from Day 3 after transfection, the growth of cells transfected with Pontin siRNAs strikingly decreased (Fig. 2C). This was associated with both reduced DNA synthesis, evidenced by decreased BrdU incorporation, and increased apoptosis, demonstrated by an increased caspase 3 activity (Fig. 2D).

Concomitant Down-regulation of Reptin and Pontin.

Because findings with Pontin silencing were reminiscent of those we obtained previously upon Reptin silencing, we evaluated Reptin expression after transfection of Pontin siRNAs. Surprisingly, Reptin protein levels were markedly reduced after transfection of P1 or P2 compared to control siRNA. The reciprocal observation was done in cells transfected with the Reptin-specific R1 and R2 siRNAs, where Pontin protein levels were also decreased (Fig. 3A,B). Because of the significant homology between Reptin and Pontin, we performed extensive control experiments that allowed excluding crossreactivity of antibodies (Supporting Fig. 1).

To rule out off-target effects of siRNAs, we used a rescue strategy. Cell lines resistant to siRNAs were established by transduction of lentiviral vectors bearing either HA-Pontin or Flag-Reptin cDNA in which silent mutations were introduced in the sequence targeted by the P2 or R2 siRNAs (Supporting Fig. 2). Reptin protein expression was not decreased by P2 siRNA in P2-resistant cells, whereas it was reduced, as expected, upon transfection of P1 siRNA (Fig. 3C). Similarly, Pontin expression was not affected in cells expressing FLAG-Reptin resistant to the R2 siRNA (Fig. 3D). Altogether, these results showed that the down-regulation of Pontin was a direct consequence of the silencing of Reptin and vice versa.

We then asked if our findings were specific to HuH7 cells. We examined the consequences of Reptin or Pontin silencing in another human liver cell line (HepG2), in human breast cancer cells (MCF7), in human prostatic cancer cells (LNCap), and in Hela cells. We also tested the coextinction of Pontin after transfection of the R2 siRNA in mouse (Hepa 1-6) or rat hepatoma cells (FAO). In every case we observed a concomitant reduction of Reptin and Pontin protein levels (Supporting Fig. 3). These results argue in favor of a general rather than a cell-specific process for regulation of Pontin and Reptin expression.

Reptin Silencing Does Not Decrease Pontin mRNA Level (and Vice Versa).

Reptin and Pontin interact with transcription factors and bind to gene promoters (reviewed4). We thus hypothesized that they could regulate reciprocally their expression through a transcriptional mechanism. However, transfection of Pontin siRNAs did not alter Reptin mRNA levels, whereas, as expected, Pontin mRNA decreased. Reciprocally, Reptin, but not Pontin, mRNA levels were efficiently silenced with Reptin siRNAs (Fig. 4A,B).

As an additional control, we tested whether Reptin produced from an mRNA devoid of its natural regulatory sequences would also be depleted upon transfection of a Pontin siRNA. We used HuH7 cells stably expressing N-term FLAG-tagged Reptin from the viral MND promoter.2 We found that the level of both endogenous and Flag-tagged Reptin was decreased after transfection of Pontin-specific siRNAs (Supporting Fig. 4A). We obtained similar results with HeLa cells expressing FLAG/HA-tagged Reptin or Pontin from the cytomegalovirus (CMV) promoter (Supporting Fig. 4B).

Altogether, these experiments allow concluding that the reduction of Reptin or Pontin protein levels upon silencing of the respective binding partner is not related to a transcriptional effect, and thus occurs by a posttranscriptional mechanism.

Reptin Silencing Does Not Alter Pontin mRNA Translation (and Vice Versa).

Given the pleiotropic role of Reptin and Pontin, we hypothesized that they might regulate their partner's translation. This was tested using polyribosome fractionation and RT-PCR analysis in HuH7 cell lines stably expressing shRNAs targeting either Firefly luciferase (KGL2) or Reptin (KR2) in a doxycycline-dependent manner (Supporting Fig. 5). The polysomal distribution of Pontin mRNA was analyzed after 4 days of culture with or without doxycycline (Fig. 5A,B). In KGL2 cells, Reptin and Pontin transcripts exhibited a similar profile and were found all along the gradient, with an enhancement in the heaviest fractions, associated with polyribosomes. As expected, in doxycycline-treated KR2 cells, Reptin mRNA became barely detectable, whereas the profile of Pontin mRNA was unchanged as compared to that in KGL2 cells, and its distribution over the fractions was similar to the profile of glyceraldehyde 3-phosphate dehydrogenase (GAPDH) mRNA. When cells were incubated with puromycin for 1 hour prior to extraction, the profile of Pontin and GAPDH mRNAs shifted from heavy fractions to the top of the gradient (Fig. 5B). Because puromycin is a polypeptide chain terminator that causes premature termination of translation, our results demonstrate a genuine association of Pontin mRNA with polyribosomes, even in conditions where Reptin is depleted.

To make sure that Pontin mRNA was indeed actively translated, we performed metabolic labeling after 4 days of culture with doxycycline using cells stably expressing HA-Pontin (KGL2-HAP and KR2-HAP). After a 15-minute pulse of radiolabeled methionine/cysteine, the level of³⁵S incorporation into newly synthesized proteins was not different between KGL2-HAP and KR2-HAP cells (78.8% ± 4.3% versus 71.6 ± 6.8%, respectively), suggesting no gross alteration in overall protein biosynthesis in Reptin silencing conditions. After immunoprecipitation with anti-HA antibodies, the label accumulated in a 55-kDa band confirmed to be the full-length HA-Pontin polypeptide because it was reactive with an antibody against Pontin (Fig. 5C). When Reptin shRNA expression was induced in KR2-HAP cells, the amount of labeled HA-Pontin immunoprecipitated after a 15-minute pulse that was 98% of the control. This experiment showed the decrease in HA-Pontin expression in cells expressing Reptin shRNA was not the consequence of a defective translation. Therefore, the association of Pontin mRNA with polyribosomes and the completion of HA-Pontin translation argued in favor of a posttranslational regulation that explains the codepletion of Pontin and Reptin. In reciprocal experiments, we also found that Pontin silencing did not alter Reptin translation, although Reptin silencing with R1 siRNA did reduce Reptin translation, as expected (Fig. 5D).

Reptin Silencing Induces Pontin Destabilization.

Because Pontin and Reptin form a stoichiometric complex, we hypothesized that down-regulation of one of the two partners could disturb this stoichiometry and lead to posttranslational destabilization of the binding partner. We thus measured Pontin half-life using metabolic labeling-chase experiments. Newly synthesized HA-Pontin was less stable upon Reptin silencing, as shown by a significantly reduced half-life as compared to control (t_1/2 = 66.6 ± 7.6 versus 118 ± 21 min, P = 0.002) (Fig. 6A,C). Because the proteasome is the major intracellular proteolytic machinery in higher eukaryotic cells, we tested its involvement using specific inhibitors. Figure 6B,C shows that MG132 was able to prolong the half-life of HA-Pontin in both Reptin-repleted and depleted conditions. The effect of Reptin depletion was specific because it did not alter the half-life of calnexin tested as a control (Fig. 6D, E).

We then asked if proteasome inhibitors could restore the Pontin steady-state level in Reptin-depleted cells. We performed many experiments using a variety of experimental settings. For instance, Fig. 7A shows an experiment where, 3 days after induction of Reptin shRNA expression, HuH7 cells were treated with MG132 or epoxomycin for 4 hours. This did not modify the codepletion, although effective inhibition of the proteasome was confirmed by the accumulation of higher molecular weight forms of β-catenin, a proteasome target.25 Similar results were seen when using other MG132 concentrations, duration of treatment, or with other inhibitors such as clasto-Lactacystin-β-lactone (not shown). We hypothesized that if the stabilizing effect of proteasome inhibition was restricted to newly synthesized protein chains, our experiments might not be sensitive enough to detect an accumulation of proteins within the short time course of the experiments. We thus repeated them using HuH7 cells that stably express HA-Pontin. These cells have high levels of Pontin mRNA (not shown), and interestingly a higher rate of Pontin translation than control cells (Fig. 7B, left panel). They do, however, overexpress very little Pontin at the protein level, reminiscent of our previous results with Flag-Reptin2 (Fig. 7B, right panel). When these cells were treated with MG132 for 4 hours, we could detect a significant increase in Pontin protein levels both in basal conditions and when Reptin was previously depleted upon shRNA induction with doxycycline (Fig. 7C,D). On the other hand, the low level of Reptin consecutive to R2 shRNA induction in doxycycline-treated cells was not increased by MG132. Similar results were obtained when the proteasome was inhibited with 2.5 μM epoxomycin (not shown).

Altogether, these results suggested that the restoration seen with proteasome inhibition was restricted to the pool of freshly translated protein. This hypothesis was tested in experiments where, following Reptin depletion, proteasome inhibition with MG1342 or epoxomycin was done in the absence or in the presence of cycloheximide, which blocks new protein synthesis. As shown in Fig. 7E,F, cycloheximide treatment prevented the restoration of Pontin levels under these conditions.

Because in most cases proteins are targeted to the proteasome following polyubiquitylation, we examined the stability of newly synthesized Pontin in the presence of the ubiquitin activating enzyme inhibitor UBEI-41 (Biogenova). We found indeed that newly synthesized Pontin stability was enhanced in cells with Reptin depletion when they were treated with UBEI-41 (Fig. 8A, B). We also performed the reverse experiment. We first showed that the stability of newly synthesized Reptin was reduced upon Pontin silencing (Fig. 8D, compare lanes 2 and 6), an effect reversed in great part by proteasome inhibition (Fig. 8D). Furthermore, Reptin stability in Pontin-depleted cells was also largely restored following inhibition of ubiquitin activating enzyme (Fig. 8D). These data proved that a ubiquitylation step was involved in the destabilization of Pontin and Reptin in this setting.

We next attempted to detect whether Pontin and Reptin underwent polyubiquitylation upon silencing of their partner. Because endogenous ubiquitin might be limiting, and in order to maximize the sensitivity of detection, we transduced KR2 cells with adenoviral vectors (kindly provided by H. Wodrich, Bordeaux) expressing either HA-ubiquitin and green fluorescent protein (GFP), or GFP alone. After 3 days with or without doxycycline, the proteasome was inhibited with epoxomycin in order to allow accumulation of ubiquitylated proteins, and ubiquitin was precipitated using the HA tag. Western blot with a Pontin antibody failed to reveal ubiquitylated Pontin, whereas ubiquitylated β-catenin was indeed detected (Supporting Fig. 6A). The same results were obtained when looking for Reptin ubiquitylation in KP2-FR cells (Supporting Fig. 6B).