Cardif: A protein central to innate immunity is inactivated by the HCV NS3 serine protease


  • Potential conflict of interest: Nothing to report.


Antiviral immunity against a pathogen is mounted upon recognition by the host of virally associated structures. One of these viral ‘signatures’, double-stranded (ds) RNA, is a replication product of most viruses within infected cells and is sensed by Toll-like receptor 3 (TLR3) and the recently identified cytosolic RNA helicases RIG-I (retinoic acid inducible gene I, also known as Ddx58) and Mda5 (melanoma differentiation-associated gene 5, also known as Ifih1 or Helicard 1.) Both helicases detect dsRNA, and through their protein-interacting CARD domains, relay an undefined signal resulting in the activation of the transcription factors interferon regulatory factor 3 (IRF3) and NF-κB. Here we describe Cardif, a new CARD-containing adaptor protein that interacts with RIG-I and recruits IKK-α, IKK-β and IKK-ϵ kinases by means of its C-terminal region, leading to the activation of NF-κB and IRF3. Overexpression of Cardif results in interferon-β and NF-κB promoter activation, and knockdown of Cardif by short interfering RNA inhibits RIG-I-dependent antiviral responses. Cardif is targeted and inactivated by NS3-4A, a serine protease from hepatitis C virus known to block interferon-β production. Cardif thus functions as an adaptor, linking the cytoplasmic dsRNA receptor RIG-I to the initiation of antiviral programmes.

Meylan E, Curran J, Hofmann K, Moradpour D, Binder M, Bartenschlager R, Tschopp J. Cardif is an adaptor protein in the RIG-I antiviral pathway and is targeted by hepatitis C virus. Nature 2005; 437: 1167-1172. Available at (Reprinted with permission.)


One of the most interesting hypotheses in HCV biology stipulates that the viral protease encoded by the NS3-4A genes not only processes viral non-structural (NS) proteins, but also cleaves cellular proteins known to play a role in the induction of interferon beta (IFN-β) and other genes involved in the induction of innate immunity. The hypothesis was supported by results demonstrating that Huh7 cells expressing HCV replicons or NS3-4A alone failed to induce IFN-β following infection with Sendai virus.1 Consistent with these findings, HCV protease inhibitors could reverse the apparent inhibition, an observation that was interpreted by some to explain why patients responded to protease inhibitors with a rapid, massive drop in viral RNA titers. However, one of the key questions about the nature of the target(s) of the protease could not be answered because some components of the transduction pathways leading to the expression of IFN-β remained elusive.

Recognition of pathogen associated molecular patterns (PAMPs), such as double-stranded RNA (dsRNA) occurs through cytoplasmic receptors, such as the protein encoded by the RNA helicase retinoic acid inducible gene I (RIG-I,2 Fig. 1). RIG-I contains two N-terminal caspase recruitment domains (CARD), known to form homotypic interactions with other CARD domains, and a C-terminal DExD/H box RNA helicase domain. The latter can recognize PAMPs leading to the activation of the IκB kinase (IKK)-related kinase IKKϵ and TANK-binding kinase 1 (TBK1) by unknown mechanisms. The kinases then activate the latent transcription factor IRF3. In parallel, RIG-I can activate IKKα/β and, in turn, NF-κB. Thus, the RIG-I pathway results in the activation of two transcription factors that are both required for the transcription regulated by the IFN-β promoter and exhibits some redundancy with the previously characterized Toll-like receptor 3 (TLR3) pathway (for review see references 3, 4). As expected, the search for cellular NS3 targets focused on these pathways and resulted in the identification of a potential protease recognition site in the TLR3 adaptor protein TRIF. Indeed, some TRIF cleavage could be detected in HeLa cells replicating HCV subgenomes.5 However, since TRIF is not expressed in Huh7 cells, it was postulated that the protease cleaves a yet elusive adaptor that binds to the CARD domain of RIG-I and provides the link to the IKKϵ, TBK1, and IKKα/β kinases in the IRF3 and NF-κB pathways. The hunt began.

Figure 1.

Signal transduction pathways mediated by RIG-I and Cardif. Pathogen associated molecular patterns (PAMP), e.g., dsRNA, bind to the helicase domain of RIG-I, which then can undergo a CARD-CARD interaction with Cardif tethered to the mitochondrial membrane. In turn, Cardif recruits several different effector kinases, activating IRF3 and NFκB and, as a consequence, expression of IFN-β and several other genes.

Enter Meylan, Tschopp, and colleagues. They performed a database search for proteins containing CARD domains and found a 540 amino-acid-long polypeptide that had previously been identified in a screen for activators of the NF-κB pathway.6 How did this protein fulfill the requirements necessary for an adaptor of RIG-I? Like RIG-I, it contains a CARD domain at its N-terminus, which can interact with the CARD domain of RIG-I. Ectopic expression of this protein resulted in increased transcription from NF-κB and IFN-β promoters demonstrating that it belonged to the RIG-IFN-β signal transduction pathway. Accordingly, the authors named it CARD adaptor inducing IFN-β (Cardif). Consistent with these results, biochemical studies showed that Cardif interacts with IKKϵ, IKKα, and IKKβ. Furthermore, siRNA-mediated knockdown of Cardif ablated the activity of RIG-I to induce the IFN-β promoter in response to Sendai virus infection and transfection of cells with poly(I:C), a surrogate for dsRNA.

Thus, from the results of Meylan et al., it can be concluded that Cardif represents the missing link in the RIG-IFN-β signal transduction pathway, but is it the elusive target for NS3? Interestingly, expression of HCV NS3/4A led to the cleavage of Cardif near its C-terminus, at a site that contains a sequence resembling protease cleavages in the NS region of the HCV polypeptide (Fig. 1). Similarly, expression of the entire HCV polyprotein either in a stable cell line or in infected Huh7 cells also led to the cleavage of Cardif. The authors demonstrated the specificity of the cleavage reaction with a single mutation of a cysteine residue (C508A), known to be required for processing by the HCV protease. Cardif with this mutation was resistant to cleavage and, importantly, retained its biological activity.

Concurrent with this work, three other laboratories also reported the identification of the elusive RIG-I adaptor.7–9 Not surprisingly, Cardif has three more names: IPS-1, VISA, and MAVS. Seth et al.8 showed that Cardif is embedded at the C-terminus in mitochondrial membranes, linking mitochondria to innate immunity for the first time. These authors also confirmed that the HCV protease could cleave and inactivate Cardif and hence prevent activation of the RIG-IFN-β pathway.10 Thus, it can be postulated that NS3/4A cleavage could release Cardif from mitochondria. Also, Kawai et al.7 showed that Cardif could directly interact with members of the death receptor pathway, RIP-1 and FADD, leading to the activation of the NF-κB pathway. This finding explains how Cardif could be involved in the activation of genes in the absence of IRF3 activation. However, the specificities of such interactions and what roles they may play in the signaling cascade are not yet known.

From all the above data, a model of dsRNA response and how HCV inhibits it can be constructed. RIG-I recognizes and binds PAMPs, such as dsRNA with its helicase domain, which then exposes the CARD domain, leading to direct binding with Cardif at the mitochondrial membrane. Cardif then activates two distinct pathways leading to the activation of the latent transcription factors IRF3 and NF-κB. Thus, by targeting Cardif, the HCV protease acts at the nexus of two very important pathways that normally mediate the induction of IFN-β and, as a consequence, the IFN-based innate immune response to viruses. This model clashes however with results demonstrating that HCV replication leads to an activation of the IFN-α/β signal transduction pathway.11 We can envision at least two scenarios that can explain the apparent paradox. First, non-parenchymal cells, such as plasmacytoid dendritic cells, might be producing IFN-α in response to cell death induced by the virus. Second, infected hepatocytes could still be able to produce IFN-β in response to HCV infection, but with a delay caused by the protease. Such a delay could be critical for the virus to better establish an infection.

Future research will have to address at least four major questions: First, is Cardif adequately cleaved during natural HCV infections? So far all experiments were performed with Huh7 cells expressing HCV replicons or an unusually robust HCV isolate, both of which produce relatively high levels of viral proteins compared to natural infections. Second, can other CARD-containing helicases such as Mda5 (Helicard) substitute for Cardif in hepatocytes, and if so, could this explain the induction of IFN-induced genes observed in HCV infected livers? Third, are there polymorphisms in the NS3/4A cleavage site of Cardif in the human population, and if so, are individuals with such mutations more resistant to HCV infections? Lastly, does treatment of patients with protease inhibitors induce IFN-β in hepatocytes as one would predict from the model based on the recent studies? Answers to these questions will further help to understand the pathogenesis of HCV and, perhaps, provide opportunities to develop improved antiviral therapies to treat it.