Involvement of the IκB Kinase (IKK)-Related Kinases Tank-Binding Kinase 1/IKKi and Cullin-Based Ubiquitin Ligases in IFN Regulatory Factor-3 Degradation¹

In response to pathogens, infected cells activate multiple signaling cascades involved in the induction of latent transcription factors. These transcription factors are responsible for the induction of a repertoire of genes known to impede pathogens’ survival in the host (reviewed in Ref. 1). One of these transcription factors, IFN regulatory factor 3 (IRF-3)³ is essential for the normal host response to pathogens (reviewed in Ref. 2). Several reports have now documented the activation of IRF-3 following infection with RNA viruses, as well as DNA viruses (3, 4, 5, 6, 7, 8). These infectious particles activate several kinases in the host including the recently described IκB kinase (IKK) homologs, IKKE (9), also called IKKi (10), and Tank-binding kinase 1 (TBK1) (11). These kinases target a Ser-Thr rich cluster located in the C-terminal end of IRF-3 (3, 8, 12, 13) (for review, see Ref. 2). Phosphorylation of IRF-3 induces its homodimerization and accumulation into the nucleus where it induces gene transcription through recognition of specific DNA response elements located in the promoters of genes encoding the chemokines IL-15, IFN-γ-inducible protein 10 and RANTES, as well as cytokines such as the type I IFN (12, 14, 15, 16).

It has been suggested that following virus infection, IRF-3 is targeted for degradation by the proteasome based on the observation that MG-132, a proteasome inhibitor, abrogates its degradation (17, 18). The ubiquitin-proteasome proteolytic pathway plays key roles in regulating the levels of many proteins involved in diverse cellular processes. Proteins targeted for degradation are first tagged with a polyubiquitin chain in a three-step cascade reaction involving ubiquitin activation (catalyzed by the ubiquitin-activating enzyme (E1)), ubiquitin transfer (catalyzed by a ubiquitin carrier protein or conjugating enzyme (E2)), and ubiquitin ligation (catalyzed by a ubiquitin-protein ligase enzyme (E3)) (see Refs. 19 and 20 for reviews). There are two major groups of E3s classified according to a common motif shared by one of the enzyme components: the homology to the E6-associated protein C terminus domain-containing E3s and the really interesting new gene (RING) domain-containing E3s (reviewed in Ref. 19). Other RING-like domain-containing E3 ligases have also been identified and include the protein inhibitors of activated STATs family small ubiquitin-related modifier ligases, the plant homeodomain domain-containing E3s and the U-box E3s. The RING family represents the largest class of E3s, which are found in single subunits or multicomponent protein complexes (see Refs. 21 and 22 for reviews). The best characterized of these multisubunit complexes consist of the three invariable subunits Skp1, Cullin1 (Cul1), and the RING finger protein Rbx1/Roc1 and a variable component known as the F-box protein. Together, they formed a ubiquitin-protein ligase complex termed SCF (Skp1-Cul1-F-box) (21, 22). Whereas Rbx1/Roc1 proteins are thought to provide a docking site for the E2 enzyme, the F-box proteins act as receptors and are responsible for substrates recognition and specificity. Members of the SCF E3 ligases family generally polyubiquitinate substrates phosphorylated at specific sites such as IκBα, p27^Kip1, and c-Myc and therefore play a key role in the regulation of the cell cycle, signal transduction and transcriptional activation (23, 24, 25, 26). Neural precursor cell expressed developmentally down-regulated protein 8 (NEDD8) is a mammalian member of ubiquitin-like proteins, which modify proteins in a manner similar to ubiquitin (reviewed in Ref. 27). The ability of SCF E3 ligases to ubiquitinate their substrates is enhanced by covalent modification of Cul1 proteins with NEDD8. This NEDDylation of Cul1 on arginine 720 is thought to result in an increased affinity of SCF E3 ligases for some E2 enzymes (28). The NEDDylation reaction requires the coordinated action of amyloid precursor protein-binding protein 1 (APP-BPI)/Uba3 (a heterodimeric E1-like enzyme) and UBC12 (an E2-like enzyme) (29, 30).

Many viruses execute multiple immune-evasive activities in infected cells by targeting the type 1 IFN signaling pathway (5, 31). Notably, it was recently reported that a viral product from rotavirus, the nonstructural protein 1, induces a rapid degradation of IRF-3 through a proteasome-dependent pathway (32). In contrast, uncontrolled IRF-3 activation is detrimental for the host since reports have demonstrated a role of activated IRF-3 in septic shock syndrome, ischemia-reperfusion injury of the liver, as well as apoptosis (33, 34, 35, 36, 37). Thus, IRF-3 activity needs to be strictly controlled. In this context, the host cell-mediated degradation of IRF-3 following virus infection may play an important role in the termination of an IRF-3-mediated response. This study was undertaken to characterize the cellular mechanisms involved in IRF-3 degradation following virus infection.

MG-132 and lactacystin were purchased from Boston Biochem. Doxycyclin was obtained from Sigma-Aldrich. Poly(I:C) (Amersham Pharmacia) was reconstituted in PBS at 2 mg/ml, denatured at 55°C for 30 min, and allowed to anneal to room temperature before use. N,N,N′,N′,-tetrakis(2-pyridylmethyl)ethylenediamine (TPEN) and N-ethylmaleimide were obtained from Sigma-Aldrich. Commercial Abs were from the following suppliers: anti-IRF-3 Abs specific for human and rodent species were from Immuno-Biological Laboratories (IBL) and Zymed Laboratories, respectively; anti-IKKE Ab (IMG-270A) (that recognize as well TBK1) was from Imgenex; anti-ubiquitin mAb (clone P4D1) and mAb to MYC were from Santa Cruz Biotechnology; mAbs to hemagglutinin (HA) (clone HA-7) and Flag epitopes and β-actin (clone AC-74) were from Sigma-Aldrich. Plasmids encoding for Flag-wtIRF-3, Flag-IRF3 5D, Flag-IRF-3 J2D, Flag-IRF-3 7D, Flag-IRF-3 5A, and Myc-wtIRF-3 have been described (17, 38). Plasmids encoding for Flag-IKKiwt and the dominant-negative version Flag-IKKi K38A were gifts of Dr. R. Lin (McGill University, Montreal, Quebec, Canada). Flag-Cul1 was a gift from Dr. M. Pagano (New York University School of Medicine, New York, NY). pMT123-HA-ubiquine has been described (39).

The ts20 and ts41 were used as previously described (39, 40). TBK1 wild-type and knockout murine embryonic fibroblasts (MEFs) have been described (8) and were immortalized using the 3T3 protocol (41). Immortalized MEFs were maintained in MEM containing 10% FBS, 2 mM glutamine, 0.1 mM nonessential amino acids. The Tet-inducible cell line HEK 293 expressing a dominant-negative version of Cul1 (Flag-Cul1 N252) was a gift from M. Pagano (New York University School of Medicine) and cultured in DMEM containing 10% FBS tetracycline-free FBS. Human diploid fibroblasts (Hel 299), HeLa, and human embryonic Kinney 293T cell lines (293T) were obtained from American Type Culture Collection (ATCC) and cultured in DMEM containing 10% FBS. 293T cells were transfected with the calcium phosphate coprecipitation method. HeLa cells were transfected with Lipofectamine 2000. When specified, cells were also transfected with 12.5 μg of poly(I:C) in 30 μl of Lipofectamine 2000. Human CMV (HCMV) Towne strain was obtained from ATCC and propagated as previously described (3). Sendai virus (SeV) was obtained from specific pathogen-free avian supply.

Cells were infected with HCMV Towne strain at a multiplicity of infection of 1.0 PFU/cell or with SeV at 100 HA units/10⁶ cells for 2 h in serum-free medium. Then the serum-free medium was replaced with complete medium for the rest of the kinetic.

Preparation of whole cell extracts, immunoprecipitation, native-PAGE, and immunoblot analysis were performed as described previously (3, 42).

To monitor the polyubiquitination of IRF-3, cells were lysed in a radioimmunoprecipitation assay (RIPA) buffer (50 mM Tris-HCl, pH 7.4, 100 mM NaCl, 5 mM EDTA, 50 mM sodium fluoride, 40 mM β-glycerophosphate, 1 mM sodium orthovanadate, 1% Triton X-100, 0.1% SDS, 1% sodium deoxycholate, 2 mM N-ethylmaleimide, and protease inhibitors mixture (Sigma)). Cells lysates were transferred to microcentrifuge tubes and passed through a 25-gauge needle five times and then centrifuged at 14,000 rpm for 20 min at 4°C. Immunoprecipitated IRF-3 (epitope tagged or endogenous) was washed 5 times with RIPA buffer before electrophoresis on 7.5% acrylamide gels. In some experiments, cell lysates were boiled for 7 min at 90°C in 1% SDS for removal of noncovalently attached proteins prior to immmunoprecipitation. Proteins were electrophoretically transferred to Hybound-C nitrocellulose membranes and polyubiquitinated IRF-3 was detected by immunoblotting using monoclonal anti-HA or anti-ubiquitin Abs.

To examine the stability of the different IRF-3 phosphomimetic point mutants, transiently transfected HeLa cells in 60-mm petri dishes were pulse-labeled for 2 h with 170 μCi/ml [³⁵S]methionine and [³⁵S]cysteine and then chased for the indicated times in complete medium containing excess methionine and cysteine. The cells were then washed twice with ice-cold PBS and lysed in Triton X-100 lysis buffer (50 mM Tris-HCl (pH 7.4), 100 mM NaCl, 5 mM EDTA, 50 mM sodium fluoride, 40 mM β-glycerophosphate, 1 mM sodium orthovanadate, 1% Triton X-100, and protease inhibitor mixture (Sigma-Aldrich)). Lysates (300 μg of protein) were precleared for 1 h with 2 μg of normal mouse serum and the resulting supernatants were incubated with protein G-Sepharose beads preabsorbed with 2 μg of anti-Flag for 16 h at 4°C. Immune complexes were washed five times with Triton X-100 lysis buffer. Proteins were eluted by heating at 95°C for 5 min in denaturing sample buffer and analyzed by SDS-gel electrophoresis on 10% acrylamide gels. The IRF-3 proteins were detected by fluorography and visualized using a gel documentation device (Typhoon scanner 9410; Amersham Biosciences) for densitometric analysis. For qualitative measurement of the stability of the different IRF-3 phosphomimetic point mutants, transiently transfected HeLa cells in 60 mm were treated with 100 μg/ml cycloheximide (cycloheximide chase) for up to 8 h. Cell extracts were prepared and subjected to immunoblotting analysis.

IRF-3 activation has been observed with both DNA and RNA enveloped viruses. In addition to its activation, a number of studies have shown that IRF-3 is degraded following infection with RNA viruses such as vesicular stomatitis virus, Newcastle disease virus, measles virus, and SeV (6, 7, 8, 17, 18, 38, 42, 43). In contrast, DNA viruses such as the Herpesvirus family member HSV-1 do not induce IRF-3 degradation (6). Because HCMV, another Herpesvirus family member, was recently shown to induce IRF-3 activation and degradation (3, 44), we wanted to revisit the possibility that both RNA as well as DNA viruses target IRF-3 to the proteasome. Infection of Hel 299 fibroblasts with SeV resulted in a dramatic reduction in the steady-state level of IRF-3 (Fig. 1,A, lanes 3–5). Native gel electrophoresis demonstrated that it is the activated form of IRF-3 that is subjected to degradation as observed by the reduction in the activated dimeric forms of IRF-3 (Fig. 1,B, lanes 3–5). As shown in Fig. 1,A, treatment of cells with small concentration (1 μM) of MG-132 significantly diminished the degradation of IRF-3 following virus infection and induced the accumulation of C-terminal hyperphosphorylated forms of IRF-3 (compare lanes 3–5 with lanes 8–10), which migrate more slowly on SDS-gel (17, 38). Higher concentrations of MG-132 totally blocked IRF-3 degradation but also inhibited the activation of the latter following virus infection (data not shown). The concentration of MG-132 used in this study did not affect IRF-3 activation as observed by its dimerization state (Fig. 1,B). Treatment with lactacystin, another structurally unrelated proteasome inhibitor, also resulted in the accumulation and stabilization of activated forms of IRF-3 following virus infection (Fig. 1). Similar results were observed in HCMV-infected cells (data not shown). These results further substantiate the importance of the proteasome in IRF-3 degradation following infection by with both DNA and RNA viruses and also reiterate a potential role for the C-terminal phosphorylation of IRF-3 as a destabilization signal (17).

Because previous reports have suggested that some proteins may be targeted to the proteasome without ubiquitination (45, 46, 47), we first verified whether IRF-3 was polyubiquitinated following infection with SeV. Cotransfection of 293T cells with a HA-tagged ubiquitin construct together with a Flag-tagged-IRF-3 construct revealed that IRF-3 is polyubiquitinated in intact cells as noted by the appearance of a high molecular mass HA signal (Fig. 2,A, lane 4). Reciprocally, in the presence of HA-tagged ubiquitin, a subpopulation of IRF-3 is also represented by a significant high molecular mass Flag-signal (Fig. 2,B, lane 4). Importantly, polyubiquitination of endogenous IRF-3 increased following SeV infection and was only observed in the presence of MG-132 (Fig. 2,C, upper panel). This increase in polyubiquitination clearly correlated with the stabilization of the hyperphosphorylated forms of IRF-3 (Fig. 2,C, lower panel). Boiling the cell extracts prior to immunoprecipitation resulted in the same signal in Western blot analysis confirming polyubiquination of IRF-3 and not coimmunoprecipitated proteins (data not shown). To address the role of ubiquitination in the modulation of the steady state abundance of IRF-3 following SeV infection, we used the temperature-sensitive cell line ts20, in which the ubiquitin-activating enzyme E1 is active at 34°C but inactive at 39°C (48). Fig. 2,D shows that the expression level of IRF-3 declined significantly at 8 and 12 h postinfection when cells were infected at the permissive temperature (see lanes 3 and 4, a). However, shifting the cells at the nonpermissive temperature resulted in the stabilization of IRF-3 (Fig. 2,Da, lanes 7–8). This effect was not related to a lower infectability of the cells at 39°C since the expression of viral proteins was comparable between the two conditions of infection (Fig. 2 Dc). These data demonstrate that ubiquitination of IRF-3 is involved in its degradation following SeV infection.

Because we observed the stabilization of phosphorylated forms of IRF-3 in the presence of proteasome inhibitors (Fig. 1), we then asked whether a SCF complex was involved in IRF-3 degradation. Rbx1/Roc1 is a RING finger protein that binds Zn²⁺. TPEN is a Zn²⁺ chelator that has been shown to inhibit the activity of purified RING domain-containing E3 ligases, presumably by removing the Zn²⁺ that is normally complexed with the RING domain (49, 50, 51). However, high concentrations of TPEN are often used, which can lead to nonspecific effects. Interestingly, the use of only 5 μM TPEN on Hel 299 fibroblasts did not prevent IRF-3 activation as judged by the accumulation of the dimeric forms, but totally blocked its degradation following virus infection (Fig. 3,A). Addition of ZnCl₂ completely antagonized the effect of TPEN. Next, given that Cul1 protein is one of the invariable components of SCF complexes, we specifically targeted this subunit to inactivate the complex. First, we used the temperature-sensitive cell line ts41, in which the APP-BPI subunit of the NEDD8-activating enzyme E1 is inactivated at the nonpermissive temperature (39°C) (52), thus enabling the use of these cells as a model for study of Cul1-dependent protein degradation (53). Infection of ts41 cells at the permissive temperature (34°C) resulted in a significant loss of IRF-3 at 12 h postinfection, whereas no degradation of IRF-3 was observed when cells were infected at the nonpermissive temperature (Fig. 3,Ba, compare lanes 6 and 12) under similar conditions of infection (Fig. 3,Bc). Infection of the parental cell lines at the nonpermissive temperature (39°C) did not affect the degradation of IRF-3 following virus infection (data not shown). We also tested whether IRF-3 was recruited to Cul1 following virus infection. Coimmunoprecipitation experiments in 293T cells overexpressing IRF-3 and Cul1 clearly demonstrated the recruitment of IRF-3 to Cul1 following SeV infection (Fig. 3,C). To further substantiate the hypothesis of a possible involvement of Cul1 in host cell-mediated IRF-3 degradation following SeV infection, we took advantage of a deletion mutant version of Cul1 (Cul1-N252), which lacks the docking sites for Rbx1/Roc1 but is able to bind and sequestrate Skp1 (54). When overexpressed in cells, it acts in a dominant-negative fashion to prevent degradation of known SCF E3 ligases substrates such as p27^Kip1, cyclin E, β-catenin, p105, and IκBα (Ref. 54 and data not shown). Induction of Cul1 mutant by doxycycline resulted in a net increase in the stability of phosphorylated forms of IRF-3 following SeV infection of HEK 293 cells (Fig. 3,Da, compare lanes 3–5 with lanes 8–10). Interestingly, this increase in the stability of the hyperphosphorylated forms of IRF-3 was also associated with a sustained activation of IRF-3 as verified by the presence of dimers or its association to CREB coactivator after infection with SeV (Fig. 3 E). These data strongly suggest that upon C-terminal phosphorylation, IRF-3 is recognized by a Cullin-based ubiquitin ligase, belonging to the SCF complex family, thereby leading to its polyubiquitination and targeting to the proteasome.

Phosphorylation of IRF-3 always precedes its degradation (see Refs. 3 , 6 , 17 , 38 , 42 , and 43 and this study). In addition, phosphorylation of IRF-3 correlates with its polyubiquitination (Fig. 2,C). Therefore, we next addressed whether C-terminal phosphorylation of IRF-3 is essential for its degradation. To determine the effect of the C-terminal Ser/Thr cluster on the rate of IRF-3 turnover, pulse-chase experiments were conducted in HeLa cells transfected with different phosphomimetic point mutants (Fig. 4,A). Whereas wild-type IRF-3 and the phosphomimetic J2D were very stable over a 12 h period, the phosphomimetics IRF-3 5D and 7D were unstable (Fig. 4,B, a and b, and c). IRF-3 5D behaves as constitutive activated forms of IRF-3 when overexpressed in target cells (reviewed in Refs. 2 and 5) and the cluster ranging form Ser³⁹⁶ to Ser⁴⁰⁵ is thought to be directly phosphorylated by TBK1 and IKKi (3, 8, 13). As suspected, in the presence of IKKi, IRF-3 was significantly phosphorylated but also very unstable (Fig. 4 Bc and C). Quantification of the data revealed that the half-lives of IRF-3 5D, IRF-3 7D, and IRF-3 in the presence of IKKi were reduced to 7.2, 5.9, and 5.5 h, respectively, compared with that of wild-type IRF-3 and IRF-3 J2D (15.5 and 17.5 h, respectively). Similar results were obtained using cycloheximide chase experiments (data not shown).

We next directly examined the contribution of the IKK-related kinases in IRF-3 degradation by first using RNA interference silencing technology. Upon transfection of siRNA duplexes directed against IKKi and TBK1, the expression levels of both kinase isoforms were down-regulated by ∼70%. However, under these conditions, IRF-3 was still phosphorylated and degraded upon SeV infection (data not shown). Therefore, we switched to TBK1^−/− MEFs, in which the activation of IRF-3 was reported to be dramatically reduced in response to LPS, dsRNA and viral infection (15, 16), and observed a complete stabilization of IRF-3 following SeV infection (Fig. 5,A). The Ab used to verify the expression level of TBK1 also detected a lower migrating band below TBK1, which is likely the other isoform IKKi. However, IKKi was not significantly involved in IRF-3 activation as neither dimerization nor degradation was observed in native-PAGE assay (Fig. 5, B and D). Transfection of poly I:C into TBK1^+/+ MEFs also resulted in IRF-3 degradation (Fig. 5,C) and activation (Fig. 5 D) whereas both of these biochemical process were completely abolished in TBK1-deficient MEFs. All together, our data suggest that the phosphorylation of the phosphoacceptor sites (between amino acid 396–405) by the IKK-related kinases TBK1/IKKi create a signal that leads to the destabilization of IRF-3.

Over the last five years, the regulation of the innate immune response through activation of IRF-3 has been the subject of several reports. These studies have led to the characterization of new inducers of IRF-3 activity as well as the intracellular signaling pathways leading to IRF-3 activation (38, 42, 43, 55, 56). Notably, two IKK homologs, IKKi and TBK1, were shown to be involved in IRF-3 phosphorylation and activation (12, 13, 14, 15, 16, 57). Another important issue, which remains poorly understood, is the molecular mechanisms that trigger the degradation of IRF-3 following virus infection. Using a combination of pharmacological, biochemical, and genetic approaches, the results presented here strongly suggest that phosphorylation of the C-terminal phosphoacceptor sites by the IKK-related kinases TBK1/IKKi create a signal that leads to the recognition of IRF-3 by a Cullin-based ubiquitin ligase pathway, which then induces the polyubiquitination-dependent degradation of IRF-3. This conclusion is based on several observations. First, the two unrelated protease inhibitors MG-132 and lactacystin stabilized the hyperphosphorylated-activated forms of IRF-3. Second, the accumulation of high molecular mass ubiquitin conjugates in MG-132-treated cells following viral infection. Third, the stabilization of IRF-3 protein at the nonpermissive temperature in cells bearing a thermolabile allele of E1 or APP-BP1. Fourth, the ability of TPEN to stabilize IRF-3 following virus infection, suggesting the implication of a RING domain-containing E3 in the polyubiquitination-dependent degradation of IRF-3. Fifth, the recruitment of IRF-3 by Cul1 following virus infection and its stabilization in cells overexpressing the Cul1-N252 mutant. Finally, the half-lives of C-terminal phosphomimetic IRF-3 mutants were shorter than that of wild-type IRF-3. Overexpression of IKKi resulted in a net decrease in the half-life of IRF-3 and reciprocally, no degradation of IRF-3 was observed in TBK1^−/− mouse embryonic fibroblasts. Given that Cul1 is one of the invariable components of SCF complexes our data thus suggest an important role for this family of E3 ligases in the control of IRF-3 stability (see Fig. 6).

Analysis of the human genome suggests that there are >70 genes encoding for F-box proteins in mammals (reviewed in Ref. 58). So far, only four SCFs have been characterized in details: SCF^{βTrCP1/Fbw1A}, SCF^{βTrCP2/Fbw1B}, SCF^Skp2, and SCF^hCDC4/Fbw7 (reviewed in Refs. 59, 60, 61). SCF^{βTrCP1/Fbw1A} specifically recognizes IκBα, IκBβ, IκBε, and β-catenin as substrates only when they are phosphorylated at both serine residues in the conserved DSGXXS motif (reviewed in Ref. 19). This motif is not found in IRF-3. SCF^{βTrCP2/Fbw1B} contributes also to IκBα ubiquitination (62), whereas SCF^Skp2 ubiquitinates various cell cycle regulators including p27^Kip1 (23), p21^Cip1 (63), and p130 (64). SCF ^hCDC4/Fbw7 was shown to target cyclin E (65). Interestingly, both SCF^Skp2 and SCF^hCDC4/Fbw7 interact and promote degradation of c-Myc transcription factor (25, 26). In addition to promoting c-Myc degradation, SCF^Skp2 increases its transactivation activity, suggesting that SCF^Skp2 is a transcriptional cofactor (25, 26). Given these interesting observations, we speculate that SCF complexes might also regulate IRF-3 transcriptional activity. Other posttranslational modifications such as the conjugation of small ubiquitin-related modifier SUMO to IRF-3 might also be involved in the regulation of its transcriptional activity (66).

Overexpression of Cul1-N252 does not completely prevent IRF-3 degradation following SeV infection (Fig. 3, D and E), therefore suggesting that ubiquitin ligases other than the SCF complex are also likely to be involved in IRF-3 degradation following virus infection. The nature of these other E3 ligases is presently unknown. However, because chelating the Zn²⁺ and blocking the NEDDylation pathway completely abrogated the degradation of IRF-3 following virus infection (see Fig. 3, A and B), this suggests that RING-type E3 ligases or more specifically members of Cullin-RING ubiquitin ligase family may be involved in IRF-3 degradation (21, 22).

To circumvent the innate immune response, several RNA and DNA viruses express viral proteins with antagonistic activities toward essential components of the innate immune system. For example, a recent study has shown that the immediate-early transcription factor RTA from HHV8 has an unconventional intrinsic Ub E3 ligase activity that targets both IRF-7 and IRF-3 for proteasome-mediated degradation. These authors also observed that RTA was associated with a homology to the E6-associated protein C terminus domain E3 ligase protein that also catalyzes poly-Ub conjugation of IRF-3 and IRF-7 (67). Rotavirus nonstructural protein 1 also induces IRF-3 degradation but the molecular mechanism by which this viral protein targets IRF-3 to the proteasome remains to identify (32). These two recent reports thus support the concept that viruses have evolved proteins to target IRF-3 for degradation. In contrast, several studies have shown that infection of target cells with replication-incompetent viruses (UV-inactivated viruses) or treatment with dsRNA alone lead to IRF-3 degradation (Refs. 8 and 44 and see Fig. 5, C and D). Therefore, these observations suggest a role of the host cell’s degradation machinery, part of which is the ubiquitin proteasome system, for the control of IRF-3 stability following its activation through C-terminal phosphorylation.

In conclusion, we show that, in addition to virus-encoded E3 ligases, infected cells use an endogenous Cullin-based ubiquitin ligase to regulate the steady state levels of IRF-3. Because IRF-3 activation is linked with cell death and endotoxin shock syndrome (33, 34, 35, 36, 37), IRF-3 activation needs to be finely controlled and we believe that in addition to A20, a negative regulator of IRF-3/NF-κB-signaling pathways (68, 69, 70), host cell-mediated IRF-3 degradation following virus infection plays a significant role in the termination of the IRF-3 response. While our paper was in revision, a related paper was published (71), indicating a role of the prolyl isomerase Pin1 in poly I:C-induced polyubiquitination of IRF-3.

We thank Rongtuan Lin and Michele Pagano for reagents used in this study and Valérie Chénard for technical assistance. We thank also Guy Servant for helpful discussions.

The authors have no financial conflict of interest.