Potential conflict of interest: Nothing to report.
An unresolved question regarding the physiopathology of hepatitis C virus (HCV) infection is the remarkable efficiency with which host defenses are neutralized to establish chronic infection. Modulation of an apoptotic response is one strategy used by viruses to escape immune surveillance. We previously showed that HCV proteins down-regulate expression of BH3-only Bcl2 interacting domain (Bid) in hepatocytes of HCV transgenic mice. As a consequence, cells acquire resistance to Fas-mediated apoptosis, which in turn leads to increased persistence of experimental viral infections in vivo. This mechanism might participate in the establishment of chronic infections and the resulting pathologies, including hepatocellular carcinoma. We now report that Bid is also down-regulated in patients in the context of noncirrhotic HCV-linked tumorigenesis and in the HCV RNA replicon system. We show that the nonstructural HCV viral protein NS5A is sufficient to activate a calpain cysteine protease, leading to degradation of Bid. Moreover, pharmacological inhibitors of calpains restore both the physiological levels of Bid and the sensitivity of cells toward a death receptor–mediated apoptotic signal. Finally, human HCV-related tumors and hepatocytes from HCV transgenic mice that display low Bid expression contain activated calpains. Conclusion: Calpains activated by HCV proteins degrade Bid and thus dampen apoptotic signaling. These results suggest that inhibiting calpains could lead to an improved efficiency of immune-mediated elimination of HCV-infected cells. (Hepatology 2009.)
Persistent infection with hepatitis C virus (HCV) is among the most common infectious causes of chronic liver disease. The majority of patients fail to clear the virus and become chronic carriers, with a persistent presence of detectable virus in the serum.1 Patients with chronic hepatitis C are at risk for hepatic fibrosis, frequently culminating in hepatic cirrhosis and hepatocellular carcinoma (HCC).2
FL-N/35 transgenic mice, with hepatocyte-targeted expression of the entire open reading frame of the genotype 1b HCV, are at risk for steatosis and hepatocellular adenoma and carcinoma.3 We previously showed that the FL-N/35 hepatocytes are resistant to apoptosis induced by the Fas/CD95 death receptor stimulation. The lack of sensitivity to apoptotic stimulation was related to decreased expression of BH3-only Bcl2 interacting domain (Bid), a BH3-only member of the Bcl-2 family of apoptosis regulators.4 Importantly, Bid-deficient hepatocytes are refractory to T lymphocyte–induced cell death, suggesting that apoptosis contributes to HCV persistence and, as a consequence, to liver pathologies characteristic of chronic HCV infection.4
Here, we report that Bid is also down-regulated in the context of the subgenomic and genome-length HCV replicons and in a subset of HCV-linked human HCC. Mechanistically, the nonstructural viral protein NS5A is sufficient for Bid degradation, which occurs via the activation of a calpain, a cellular cysteine protease. Pharmacological inhibition of calpains restores the physiological levels of Bid, suggesting that calpains may be interesting molecular targets for therapies aimed at increasing the efficiency of immune elimination of HCV-infected cells.
GAPDH, glyceraldehyde-3-phosphate dehydrogenase; HBV, hepatitis B virus; HCC, hepatocellular carcinoma; HCV, hepatitis C virus; mRNA, messenger RNA; NS5A, nonstructural protein 5A; PBS, phosphate-buffered saline; RT-qPCR, real-time quantitative PCR; SEM, standard error of the mean; shRNA, short hairpin RNA; TNF-α, tumor necrosis factor alpha; z-VAD-fmk, benzyloxycarbonyl-Val-Ala-Asp (OMe) fluoromethylketone.
Patients and Methods
FL-N/35 transgenic animals3 were bred and maintained according to the institutional guidelines. Nine-month-old males were used in these experiments.
Patients and Tissue Samples.
Twenty-three HCC and corresponding nontumoral tissues were obtained from resected specimens of patients treated at the University Hospital, Bordeaux, France. Small liver pieces from tumoral and nontumoral livers were immediately frozen in liquid nitrogen and stored at −80°C until use. In parallel, whole-liver specimens were sampled and representative sections were processed for diagnostic purposes. In this study, the series included 17 patients infected with HCV and six patients infected with HBV. Their clinical data such as sex and age, as well as fibrosis stage of the nontumoral liver according to Metavir score,5 are reported in Table 1. Informed consent was obtained according to the institutional regulations.
Table 1. Clinical and Pathological Features of Patients with HCV and/or HBV
Cells were cultured in Dulbecco's modified Eagle medium supplemented with 10% fetal bovine serum, 100 ng/mL streptomycin, and 100 U/mL penicillin and 400 μg/mL of G418 for cells harboring the Nneo/C-5B and Nneo/3-5B replicons. When appropriate, cells were treated with lactacystin (50 μM), benzyloxycarbonyl-Val-Ala-Asp (OMe) fluoromethylketone (z-VAD-fmk; 10 μM), E64D, (50 μM), or MDL-28170 (100 μM) for 24 hours and staurosporine (1 μM), tumor necrosis factor-alpha (TNF-α) (0.5 ng/mL), and actinomycin D (0.5 μg/μL) for 14 hours. All drugs were from Sigma (St. Louis, MO).
Subconfluent cultures were treated with TNF-α (0.5 ng/mL) and actinomycin D (0.5 μg/mL). Where appropriate, MDL-28170 (100 μM) was added 6 hours before apoptotic inducers. Apoptosis was assayed after 14 hours on pooled adherent and floating cells using the Annexin V-Cy3 Apoptosis Detection Kit (Abcam, Cambridge, UK) according to the manufacturer's instruction. The cells were analyzed on a FACSCalibur flow cytometer using CellQuestPro software (BD Bioscience).
Zymograms were run according to previously described methods6, 7 with minor modifications. Resolving gels contained 0.2% of casein (Hammerstein, USB Corp., Cleveland, OH). Cells were homogenized in 150 μL of extraction buffer (20 mM Tris-HCl [pH 7.5], 5 mM ethylene diamine tetraacetic acid (EDTA), 5 mM ethylene glycol tetraacetic acid, 1 mM dithiothreitol) and centrifuged for 15 minutes at 10,000g at 4°C. Protein concentrations were measured using the bicinchoninic (BCA) assay. Samples (25 μg) were electrophoresed at 100 V for 4 hours at 4°C, the gels were rinsed and incubated for 18 hours at 20°C in Ca2+ containing buffer, stained for 2 hours with Coomassie brilliant blue, then placed in boiling water for 8 minutes. Protease activity is visualized as a clear band on the gel.
In Vitro Bid Degradation Assay.
Recombinant human Bid (50 ng) was incubated with 50 μg of cellular extract for 3 hours at room temperature in 50 μL calpain assay buffer (50 mM Tris-HCl, 100 mM NaCl [pH 7.5], 20 mM dithiothreitol, 5 mM CaCl2). Protease inhibitors (lactacystin (50 μM), z-VAD-fmk (10 μM) or MDL-28170 (100 μM) were included, as indicated. For tBid control assays, 50 ng of human rBid was incubated for 45 minutes at 37°C in the presence or absence of 10 ng recombinant human active caspase 8 in the buffer containing 25 mM HEPES [4-(2-hydroxyethyl)-1-piperazine ethanesulfonic acid; pH 7.4], 5 mM dithiothreitol, 50 mM NaCl, 5% sucrose, 0.5 mM EDTA.
Experiments were performed at least three times. Data are presented either from a representative experiment or as mean ± standard error of the mean (SEM). Comparisons between groups were analyzed by Student t test. Significance values: P ≤ 0.01 (*), P ≤ 0.001 (**). The statistical significance of data in Fig. 1 was calculated by the test of Bernoulli.8
Bid Decrease in HCV-Related HCCs.
We have previously reported that Bid protein expression is down-regulated in hepatocytes of the FL-N/35 mice,4 which express roughly homogenous amounts of all HCV proteins in each cell.9 In contrast, a minor and variable proportion of liver parenchymal cells stain positive for viral proteins in chronically infected human patients.10, 11 Because changes in Bid expression are expected to be specific to infected cells, biopsies from patients with hepatitis C infection are not suitable for biochemical validation of the mouse data in human pathology. In the search for a human in vivo clonal population derived from an HCV-infected cell, we concentrated on HCV-related tumors. Whereas a vast majority of HCV-linked HCCs develop on necroinflammatory, cirrhotic livers, a small proportion appear on quasi-healthy tissues.12 We reasoned that these rare HCCs were likely to have arisen from an infected cell, in which the initial steps of oncogenesis would have been driven by the virus, rather than by the tumor-prone cirrhotic environment.13, 14 We analyzed Bid protein expression levels in 23 surgical specimens of HCC and the corresponding nontumoral tissues from patients suffering from chronic infection with HCV or HBV (Table 1). Seventeen patients were HCV-positive, nine had liver cirrhosis, and the remaining eight had varying fibrosis scores. Interestingly, the only group of patients in whose tumors Bid was consistently down-regulated (eight of eight patients) was the HCV-positive group bearing tumors on noncirrhotic livers (Fig. 1A, see also Fig. 5B). Similar to the transgenic mouse model, Bid protein level did not correlate with the messenger RNA (mRNA) expression, suggesting a posttranslational effect (Fig. 1B). As predicted by our hypothesis, only a minor fraction (one of nine patients) of HCV-linked tumors that appeared on cirrhotic livers expressed a low level of Bid (Fig. 1C). Finally, in the limited number of HBV-positive tumors analyzed, there was a tendency for an increased Bid expression, independently of the underlying liver cirrhosis (Fig. 1D). These data suggest that the HCV-linked tumors arising in noncirrhotic liver backgrounds are molecularly distinct from the majority of HCCs. Indeed, the probability that the difference of phenotypes observed in HCV-linked HCC in noncirrhotic versus cirrhotic livers was due to chance is low (P = 2 × 10−8). Although not directly proving it, these results support the idea that similarly to the mouse model, Bid is regulated by the HCV proteins in infected human hepatocytes in vivo.
Bid Decrease in the Replicon System.
Human hepatoma Huh7 cells harboring the full-length genome (Nneo/C-5B) or the subgenomic (Nneo/3-5B) HCV RNA replicons express, respectively, the full complement of viral proteins or only the nonstructural proteins (NS3 through NS5B) of a genotype 1b viral isolate.15 Both replicon-containing cell lines expressed less Bid than the parental Huh7 cells (Fig. 2A-C), the decrease being more pronounced for the full-length replicon. This effect appears to be specific for Bid, because several other members of the Bcl-2 family were expressed similarly in the three cell lines (Supporting Fig. 1). Importantly, as in human samples (Fig. 1B) and in the transgenic mouse model,4 Bid mRNA levels were not affected by the presence of viral gene products (Fig. 2D).
In the course of apoptosis, full-length Bid undergoes a proteolytic cleavage; the truncated protein, tBid, is the active form that initiates the mitochondrial step of apoptotic signaling.16 We found no tBid accumulation in the replicon-containing cells (Fig. 2E), suggesting that in this case, activating Bid cleavage was unlikely to be responsible for the disappearance of the full-length protein.
Inhibitors of Calpains Restore Bid Expression.
Increased protein turnover could explain the decreased level of Bid observed in the presence of the HCV proteins. We therefore treated the parental and the replicon-containing cells with a series of cell-permeable pharmacological protease inhibitors and looked for the rescue of the Bid down-regulation phenotype. Proteasome and caspase inhibitors, lactacystin and z-VAD, respectively, had no effect on Bid levels (Supporting Fig. 2). In contrast, Bid expression was restored in the replicon cell lines grown in the presence of calpain inhibitors E64D and MDL-28170 (Fig. 3A,B). Of note, inhibition of calpains had no discernible effect on Bid expression in the parental Huh7 cells.
To test if the augmented protease activity was sufficient to account for Bid degradation, we incubated purified recombinant Bid (r-Bid) protein with soluble extracts of Huh7 and Nneo/C-5B replicon cells (Fig. 3C,D). As control, we used recombinant active caspase 8, which, as expected, acted on the full-length r-Bid to give rise to tBid under these conditions. Whereas exogenous Bid was stable in the presence of Huh7 cytosol throughout the course of the experiment, the Nneo/C-5B extract contained a proteolytic activity which led to r-Bid degradation. It was prevented by a calpain inhibitor, but not by proteasome or caspase inhibitors (Supporting Fig. 3), suggesting that Bid is a direct target of a soluble active calpain present in cells expressing HCV proteins.
Activation of Calpains in the Replicon System and in HCV-Related HCC.
Modulation of calpain activity could be due either to changes in their expression, their activation, or both. We found no variation in expression of two ubiquitously expressed calpains, calpain 1 or calpain 2, in the replicon-containing cell lines compared to the parental Huh7 cells (Supporting Fig. 4A). These data were consistent with our quantitative reverse transcription polymerase chain reaction (RT-qPCR) analysis of the 14 known and putative catalytic and two regulatory calpain subunits.17 Nine catalytic and both regulatory subunits had detectable levels of expression in the Huh7-derived cells; none of them were significantly altered in the presence of HCV proteins (Supporting Fig. 4B). Next, we tested the hypothesis that the viral proteins influence calpain enzymatic activity. Zymography assays performed under conditions compatible with calpain activity detected a single protease band (Fig. 4A). A weak signal was detected in the parental Huh7 cells, and both replicon lines exhibited a strong increase of the proteolytic activity. To assay calpain activation in the Nneo/C-5B and the Nneo/3-5B cells using an alternative approach, we next looked for the presence of calpastatin, an endogenous calpain inhibitor that is cleaved upon calpain activation.18 As expected, cells harboring either replicon displayed diminished calpastatin levels, consistent with calpain activation by the viral proteins (Fig. 4B).
We next investigated if calpain activation was also associated with viral protein expression in vivo. In FL-N/35 mice, where Bid down-regulation by the HCV proteins was originally described,4 calpastatin degradation accompanied the presence of the transgene and the lower Bid expression level (Fig. 5A). Next, we analyzed calpain activity in extracts of tumors and surrounding tissues obtained from human liver biopsies. All of the HCV-positive tumors from noncirrhotic livers displayed diminished calpastatin levels, in parallel to the decrease of Bid expression (Fig. 5B). A single tumor of our collection of HCV-positive HCCs in the cirrhotic environment had a low level of calpastatin (Fig. 5C), which in this case did not correlate with low Bid levels (Fig. 1C).
Our data from the transgenic mouse model, the subgenomic and genomic replicons and from the analysis of human patient samples, all point to the calpains as HCV-activated mediators of proteolytic degradation of Bid.
NS5A Activates Calpains and Lowers Bid Expression.
HCV codes for three structural and seven nonstructural proteins. In vivo, they are organized in an endoplasmic reticulum–associated multiprotein complex.19 Nevertheless, isolated viral proteins maintain some activities that may be relevant to the physiopathology of the viral infection. To determine if the phenotype of calpain activation and Bid down-regulation could be ascribed to a specific viral protein, we established stable populations of Huh7 cells in which the expression of a single protein, or a combination of two proteins, was driven by a heterologous promoter. Concentrating on viral proteins previously described as major regulators of host cell physiology,13 we tested the effect of core, NS3, NS4A, and NS5A. Retrovirus-driven expression of viral proteins was controlled by immunoblotting (Supporting Fig. 5). Within the limits of the combinations tested, only NS5A mimicked the entire complement of viral proteins, both in its effect on Bid and on calpain activity (Fig. 6A,B), without affecting other members of the Bcl-2 family (Supporting Fig. 6). Similarly to the data obtained in the replicon model (Fig. 2E), decreased Bid expression was not associated with Bid activation (Fig. 6C). In addition, a pharmacological inhibitor of calpains, the MDL 28170 compound, partially restored Bid expression (Fig. 6B). Calpains 1 and 2 were probably not responsible for the observed phenotype, because their partial (capn 2) or highly efficient (capn 1) silencing by specific short hairpin RNAs (shRNAs) restored neither the Bid nor the calpastatin levels in the NS5A-expressing cells (Fig. 6D). To determine if Bid down-regulation participated in the resistance to death receptor–mediated cell death, we assayed the effect of calpain inhibition on apoptosis. TNF-α–induced apoptosis was significantly reduced in the presence of NS5A (Fig. 6E; Supporting Fig. 7). Importantly, down-regulation of Bid expression by shRNA (Supporting Fig. 8) gave rise to a similar protection (Fig. 6E). As expected from the different modes of action leading to Bid down-regulation, pharmacological inhibition of calpains restored the apoptotic response of the NS5A-expressors, but not of the shRNA-treated cells.
Finally, we investigated whether the effect of viral proteins on calpains and, as a consequence, on Bid expression, also operated in the context of a bona fide viral infection. Several tests performed on Huh7.5.1 cells productively infected with either the JFH1 isolate virus20, 21 or the Con1/C3 chimera22 showed no Bid phenotype (data not shown). Because the NS5A sequence is highly variable among HCV genotypes, we hypothesized that this result could be due to a functional difference between NS5A genotypes. To address this point, we compared the effects of expression in Huh7 cells of these three NS5A genotypes: 1b, 1a, and 2a, the latter amplified from the JFH1 complementary DNA. In contrast to both genotype 1 sequences, the JFH1-derived protein had no effect on either the calpain activity or on Bid expression (Fig. 6F).
These results prompted us to reanalyze the data shown in Fig. 1A. Interestingly, the three patients characterized by the most profound decline of Bid expression (patients 6741, 3218, and 1866) were all infected with genotype 1 HCV.
Viruses are under a strong evolutionary pressure to develop strategies of resistance to their host's innate and acquired immune responses. HCV is particularly successful in avoiding elimination: it is estimated that up to 80% of acute infections progress to chronicity.1 On top of several mechanisms of interference with the type 1 interferon production and response, antigen presentation, and cell-mediated immune response,23, 24 HCV also modulates apoptosis of the infected cell.25 We previously reported that the HCV viral proteins decrease the expression of the proapoptotic protein Bid.4 Bid is a ubiquitously expressed BH3-only member of the Bcl-2 family.16 In some cell types, including hepatocytes, it is an essential component of the apoptotic signaling pathway originating at the so-called death receptors at the cell surface. In the FL/N-35 immunocompetent mice, decreased Bid expression and the associated apoptotic defect lead to an increased persistence in the liver of an unrelated virus, suggesting that the resistance to cell death opposes the T cell–mediated viral clearance, thus favoring chronic infection.4 Now, we report that Bid deficiency, and presumably its consequences, is involved in at least some cases of human pathology related to HCV infection.
Because very little of the viral proteins is present in a chronically infected human hepatocyte, it is difficult to accurately estimate the fraction of infected cells in patient liver biopsies.10, 11 Consequently, extracts from biopsies obtained from livers of patients with HCV cannot be used for tracking the effect of viral proteins on Bid. Because biochemical assays are the most reliable way of quantifying both RNA and proteins, we attempted to identify a cellular population in vivo that is homogenous with respect to its exposure to the virus. We reasoned that tumors derived from an infected cell might correspond to this definition. Although there is no doubt that the inflammatory environment, with its associated oxidative stress, and cirrhosis, which are the context of a vast majority of HCV-related HCCs, provide favorable conditions for an oncogenic transformation,14, 26 the presence of HCV is an independent risk factor for HCC.27 Moreover, despite the lack of strong oncogenes among the HCV proteins, their direct impact on hepatocyte cell cycle and apoptosis regulatory mechanisms could significantly increase the probability of tumor initiation in infected cells.13 It follows that a non-negligible proportion of HCV-related tumors probably derive from infected cells. We hypothesized that tumors of such origin were likely to be particularly abundant in the minor fraction of HCV-linked HCCs that appear in livers that are not cirrhotic and, even more infrequently, have a low fibrotic index. The data presented here are consistent with such a hypothesis. Indeed, in contrast to HCV-positive HCCs arising on cirrhotic livers, but similarly to cells from the HCV-positive transgenic mouse livers and from the replicon model, tumors from the patients without cirrhosis had a decreased expression of Bid. This phenotype is thus clearly not a general feature of HCC. However, it does appear to be specific for HCV, because it was not observed either in the HBV-related tumors (independently of liver cirrhosis, see Fig. 1D) or in tumors unrelated to viral infection (not shown). Thus, whereas we do not think that these data shed light on the process of tumorigenesis, we believe that they support the contention that the HCV affects Bid expression in hepatocytes of chronically infected human patients.
Because HCV had no effect on the levels of Bid mRNA in our experimental systems, we turned our attention to protein stability. Many proteases can participate in the turnover of cellular proteins. Moreover, Bid is a target of several proteolytic enzymes.16 The full-length Bid has a long half-life,28 which is probably why we detected little effect of any protease inhibitors on the level of the protein in parental Huh7 cells. In contrast, in cells harboring either the subgenomic or the full-genome-length HCV replicons, Bid expression was restored by pharmacological inhibition of calpains. Although it is notoriously difficult to eliminate the possible cross-specificities in the use of pharmacological enzyme inhibitors, the fact that two chemically unrelated calpain inhibitors29, 30 had a similar activity in our assay argues for these proteases' involvement in Bid degradation.
NS5A, an HCV protein involved in regulation of many cellular signal transduction pathways,31 has been reported to activate calpains.32 We have confirmed and extended these results: while a strong increase of calpain-like activity occurred upon the expression both of the full complement of viral proteins, the nonstructural subset, or specifically the NS5A of the 1b and the 1a HCV genotypes, the 2a genotype protein had no effect either on calpain activity or on the stability of the Bid protein. Interestingly, these viral genotypes with documented divergence in the NS5A sequence have been associated with pathologies of different prognosis.33 Because a diminished sensitivity to apoptotic stimuli is a hallmark of a transformed cell, the association of the genotype 1 with HCV-related tumors34 might be of particular interest in the context of these results.
Several proteases, including calpains, have been reported to process Bid into its active form, the tBid.35 In contrast to caspase 8, extracts from replicon-harboring cells did not give rise to processed tBid in our experimental set-up. Moreover, in agreement with our previous report,4 we now show that, similarly to the phenotype of the shRNA-mediated Bid knockdown, the presence of HCV proteins (and in particular of NS5A) and the associated calpain activation correlate with a decreased sensitivity to death receptor stimulation. Importantly, whereas inhibition of calpains restores Bid activity and apoptosis sensitivity in NS5A-expressing cells (Figs. 3, 6A; Supporting Fig. 7), it has no effect on cell death when Bid is down-regulated by shRNA. Therefore, we believe that Bid degradation is a physiologically relevant consequence of HCV-related calpain activation. In this context, it is worth noting that calpains 1 and 2, which have been described as Bid activators, do not appear to be involved in the Bid degradation initiated by HCV (Fig. 6F). Pharmacological inhibition of calpains partially restores Bid levels and the cell's sensitivity to apoptotic signaling initiated by TNF-α treatment. These results strongly argue that, in addition to the previously described effect of NS5A on the initial steps of TNF receptor signal transduction,36 Bid is a physiologically relevant target of this viral protein.
Although high protease activities are found in many solid tumors,37 the use of a stringent assay of calpastatin degradation and the limiting of our analysis to HCCs arising on noncirrhotic livers allowed us to correlate calpain activation with a Bid decrease in human HCC. Although more clinical samples need to be analyzed to substantiate our interpretation, these data are consistent with the idea that the activation of calpains by HCV and its consequences on the apoptotic signaling pathways, observed in several animal and cellular models, are also relevant to human pathology.
If true, these results might open new perspectives for therapeutic strategies for control of HCV infection and its consequences. A strong prediction of our model is that inhibition of the virus-activated calpains should lead to the restoration of Bid in infected cells and therefore to the recovery of their sensitivity to apoptotic stimulation. This in turn would facilitate their elimination in the course of the physiological immune response. Calpains have already been investigated as possible therapeutic targets in human disease, mainly in the context of neurological and muscular disorders.38, 39 Our data suggest that it might be worthwhile to pursue the identification of calpain isoforms activated by the HCV and to search for their specific inhibitors to boost the efficiency of the immune elimination of the infected cells.
We are indebted to Stanley M. Lemon for the gift of the FL-N/35 transgenic mice and the N/neoC-5b and N/neo3-5b cell lines, to Dr. Takaji Wakita for the gift of the clones of the JFH1 and Con1/C3 viruses, and Drs. Mark Harris and Robert Lanford for clones of different genotypes of NS5A. We thank Thierry Gostan for help with the statistical analysis, Krishna Damodar for technical assistance, Charles Balabaud and Jean Michel Pawlotsky for their interest and support, and all members of the U. Hibner lab for comments and discussions.