Negative Regulation of Virus-triggered IFN-β Signaling Pathway by Alternative Splicing of TBK1*

Induction of Type I IFNs is a central event in antiviral responses and must be tightly controlled. The protein kinase TBK1 is critically involved in virus-triggered type I IFN signaling. In this study, we identify an alternatively spliced isoform of TBK1, termed TBK1s, which lacks exons 3–6. Upon Sendai virus (SeV) infection, TBK1s is induced in both human and mouse cells and binds to RIG-1, disrupting the interaction between RIG-I and VISA. Consistent with that result, overexpression of TBK1s inhibits IRF3 nuclear translocation and leads to a shutdown of SeV-triggered IFN-β production. Taken together, our data indicate that TBK1s plays an inhibitory role in virus-triggered IFN-β signaling pathways.


Induction of Type I IFNs is a central event in antiviral responses and must be tightly controlled. The protein kinase TBK1 is critically involved in virus-triggered type I IFN signaling. In this study, we identify an alternatively spliced isoform of TBK1, termed TBK1s, which lacks exons 3-6. Upon Sendai virus (SeV) infection, TBK1s is induced in both human and mouse cells and binds to RIG-1, disrupting the interaction between RIG-I and VISA. Consistent with that result, overexpression of TBK1s inhibits IRF3 nuclear translocation and leads to a shutdown of SeV-triggered IFN-␤ production. Taken together, our data indicate that TBK1s plays an inhibitory role in virus-triggered IFN-␤ signaling pathways.
Innate immunity is the first line of defense against viral and microbial pathogens. There are at least two pattern-recognition receptors (PRRs) that detect the presence of viral doublestranded RNA (dsRNA) 4 (1,2): a subfamily of TLRs (TLR3, -7, -8, -9) (3-6); and retinoic acid-inducible gene I (RIG-I)-like helicases (RLHs), which include RIG-I and melanoma differentiation-associated gene 5 (MDA5) (7,8). TLR3, which detects extracellular viral dsRNA internalized into endosomes, recruits a TIR domain-containing adaptor-inducing IFN-␤ (TRIF) to the receptor upon ligand stimulation (9 -12). RIG-I and MDA-5 are cytoplasmic sensors of virally derived dsRNA sharing caspase recruitment domains (CARDs) at the N terminus followed by a DexD/H-box helicase domain at the C terminus. Using knock-out mice, it has been shown that RIG-I detects various RNA viruses including Sendai virus (SeV), vesicular stomatitis virus (VSV), influenza virus, hepatitis C virus (HCV), and Japanese encephalitis virus (JEV), whereas MDA5 recognizes poly(I:C) and is essential for triggering the host response to the picornavirus encephalomyocarditis virus (EMCV) (7,8). Both RIG-I and MDA5 interact with viral dsRNA in the cytosol via the helicase domain and initiate downstream signaling cascades via the CARDs by association with virus-induced signaling adaptor (VISA, also known as MAVS, IPS-1, or CARDIF) (13)(14)(15)(16). Engagement of any of these receptors triggers rapid production of type I interferon (IFN-␣/␤) and consequently establishes the innate immune status against virus replication (1,2). TANK binding kinase 1 (TBK1) (17)(18)(19), which was originally identified in the context of regulation of NF-B activity, is activated downstream of TRIF and VISA in response to viral dsRNA. TBK1 can phosphorylate IFN-regulatory factor (IRF)-3 and IRF-7 in vitro (20 -22). The phosphorylated IRF3 and IRF7 in turn form homodimers or heterodimers, translocate into the nucleus and induce the expression of type I IFN as well as IFNinducible gene. Knockdown assays and gene-targeting studies have shown that TBK1 is essential for type I interferon production in TLR3 and RLH signaling pathway (22)(23)(24)(25).
Although it is essential for provoking innate immune response and enhancing adaptive immunity against virus, host antiviral response must be tightly controlled to prevent harmful effects resulting from excessive activation. In this study, we identify an alternatively spliced isoform of TBK1, termed TBK1s, which lacks exons 3-6, and negatively regulates virus-triggered IFN-␤ signaling pathway.

EXPERIMENTAL PROCEDURES
cDNA Constructs and Reagents-Mouse TBK1 and mouse TBK1s sequences were amplified by PCR using cDNA from SeV-infected bone marrow-derived dendritic cells (BM-DCs) and confirmed by sequencing. PCR primers were based on Cell Culture and Transfection-HEK293T cells, TBK1 Ϫ/Ϫ MEFs and Huh7 cells were cultured in Dulbecco modified Eagle's medium supplemented with 10% fetal bovine serum (Invitrogen), 100 units/ml penicillin, and 100 units/ml streptomycin under humidified conditions with 5% CO 2 at 37°C. L929 cells were grown in RPMI 1640 supplemented with 10% fetal calf serum, 2 mM L-glutamine, 100 units/ml penicillin, and 100 g/ml streptomycin. The method for generation of bone marrow dendritic cells (GM-CSF-DC) has been described previously (26). Transient transfection of HEK293T cells was done with Lipofectamine (Invitrogen). TBK1 Ϫ/Ϫ MEFs were transfected using Lipofectamine 2000 (Invitrogen) according to the manufacturer's instructions.
Infection-293T, L929, Huh7, TBK1-1 MEF, and GM-CSF-DC cells were infected with the indicated concentration of SeV in serum-free medium. After incubation for 2 h, cells were added with the same volume medium containing 20% fetal calf serum. After infection, cells were collected and used for various experiments.
Immunoprecipitation and Immunoblot Analysis-Cultures of HEK293T cells in 6-well plates were transfected with various combinations of plasmids. Immunoprecipitation and immunoblot analysis were performed as described before (27). For endogenous co-immunoprecipitation experiments, L929 cells (1 ϫ 10 7 ) were infected with 10 HAU/ml SeV for the indicated time, while HEK293T cells (1 ϫ 10 7 ) were infected with 10 HAU/ml SeV after transfection with vector, HA-TBK1s, or TBK1s mutant 1-371 plasmid (4 g) for 24 h. Cells were then lysed in lysis buffer, and the lysates were incubated with 2 l of RIG-I antibody or control IgG. The subsequent procedures were carried out as described above.
Reverse Transcription and Quantitative Real-time PCR-Total RNA was extracted from cultured cells with TRIzol (Invitrogen), and cDNA was prepared as described previously (28). Quantitative real-time PCR (Q-PCR) analysis was performed using the ABIPRISM 7900HT(ABI). The primers used for PCR are listed in Table 1. Data were normalized according to the level of GAPDH expression in each sample.
Nuclear Extracts-Nuclear extracts were prepared as previously described (28).
Reporter Assays-Cells (1 ϫ 10 5 /well in 24-well plates) were transfected with the indicate amount of expression plasmids combined with 100 ng of indicated reporter genes and 40 ng of pRL-TK (Clontech). The total DNA concentration was kept constant by supplementing with empty vector pcDNA3.0. In some experiments, cells were infected with SeV after transfection. In all experiments, cells were lysed, and reporter activity was analyzed with the Dual-Luciferase Reporter Assay System (Promega). The values represented the average of three independent experiments with variability shown by the error bars.
Statistics-All measurements were performed in at least three independent experiments, and the means Ϯ S.D. were calculated. The Student's t test was used to compare two independent groups. For all tests, values of p Ͻ 0.05 were considered statistically significant.

Identification of a Splice Variant of TBK1, termed TBK1s, Which Lacks Exons 3-6-
The mouse Tbk1 gene is organized into 21 exons and 20 introns (Fig. 1A). Exons 2-6 (corresponding to amino acids 1-234) encode the S_TKc domain, which mediates the phosphorylation of IRF3 and IRF7. RT-PCR for TBK1 on RNA isolated from BM-DC infected with 10 HAU/ml Sendai virus reveals two cDNA species of 2190 and 1567 bps, respectively (Fig. 1B). Sequence analysis shows that the larger band is identical to the published sequence of full-length Tbk1, while the smaller band represents an alternative splicing isoform of Tbk1, lacking exons 3-6 and subsequently referred to as Tbk1s. Excision of exons 3-6 results in translation from the second ATG and leads to an in-frame deletion of the kinase domain (amino acids 1-234) (Fig.  1A). Moreover, analysis of genomic locus that contains the human TBK1 sequence suggests a similar splice acceptor-donor site that could lead to a splice variant similar to the mouse TBK1s isoform. To further investigate the existence of the Tbk1 spliced isoform in human, we designed human Tbk1 isoformspecific primers and Tbk1s isoformspecific primers based on the difference of splicing site (human Tbk1 NCBI GenBank TM Accession Number: NM_013254). Human hepatoma Huh7 cells were infected with Sendai virus and the expression of human TBK1 and TBK1s mRNA were measured by RT-PCR using human Tbk1 isform-specific primers and human Tbk1s isoformspecific primers, respectively. As shown in Fig. 1C, a Tbk1s-specific PCR product (ϳ60 bp) was detected in cells infected with SeV. To further confirm the existence of the TBK1 spliced isoform, we analyzed TBK1s protein expression. To assess this, we used rabbit polyclonal antibodies against amino acids 355-729 of mouse TBK1 (Santa Cruz Biotechnology) to detect the presence of TBK1s protein, and cell lysates from 293T cells transfected with TBK1s plasmid served as positive control. In addition to an endogenous ϳ75-kDa band (TBK1 protein), an endogenous ϳ55-kDa protein (TBK1s protein), which was consistent with exogenous expression of TBK1s in 293T was recognized in BM-DC infected with SeV (Fig. 1D). Furthermore, TBK1s protein could be induced in Huh7 cells treated with SeV or IFN␤ (Fig. 1E). Compared with the constitutive expression of TBK1, the expression of TBK1s was inducible and lower. The tissue distribution of TBK1s expression, assessed by immunoblot analysis, showed that brain, heart, kidney, lymph node, and thymus had relative high expression of TBK1s protein (Fig. 1F). Intriguingly, TBK1s protein was more abundant than TBK1 protein in heart and kidney. Our data not only demonstrated the existence of the Tbk1 spliced isoform in both mouse and human cells, but also led us to further investigate the function of TBK1s.
The Kinase Domain of TBK1 Is Critical for the Induction of IFN-␤-Initially, to investigate the function of TBK1s, TBK1 Ϫ/Ϫ MEF cells were reconstituted with plasmids encoding TBK1 or TBK1s, and cells were infected with SeV after . The ratio of TBK1-TBK1s is shown as indicated. C, RT-PCR analysis on RNA extracted from Huh7 infected by SeV with human Tbk1 isoform-specific primers, human Tbk1s isoformspecific primers, Ifnb primers, and GAPDH primers listed in Table 1. D, BM-DCs were infected with 10 HAU/ml SeV for the indicated time. E, Huh7 were treated with 20 HAU/ml SeV or 2000 units/ml IFN-␤ for 12 h. After treatment, cells were subjected to immunoblot assay using anti-TBK1 antibody. The ratio of TBK1-TBK1s is shown as indicated. F, immunoblot of mouse tissues prepared and quantified by bicinchoninic acid assay (50 g of protein loaded per sample), analyzed with anti-TBK1 antibody. The ratio of TBK1-TBK1s is shown as indicated. Arrows indicated TBK1 and TBK1s. KDa, kilodaltons.
transfection. Expression of the transfected protein in TBK1 Ϫ/Ϫ MEF cells were first confirmed by immunoblot assay. As shown in Fig. 2A, TBK1 and TBK1s protein were expressed at similar levels in the transfected cells. Luciferase assays showed that expression of TBK1 but not TBK1s could induce IFN-␤ promoter reporter activity upon SeV infection (Fig. 2B). Consistently, the expression of TBK1 but not TBK1s in these cells strongly induced Ifnb and IP-10 mRNA (Fig. 2C). Previous studies had shown that TBK1 mediates its effects via IRF3 (20, 23), we therefore analyzed nuclear translocation of IRF3 upon SeV infection. To assess this, TBK1 Ϫ/Ϫ MEF cells were transfected with IRF3-GFP plasmid in addition, and nuclear translocation of IRF3 was examined by immunofluorescence microscopy after SeV infection. As shown in Fig. 2D, SeV infection resulted in translocation of IRF3 into the nucleus in cells transfected with TBK1 plasmid. In contrast, GFP-IRF3 remained in the cytoplasm in cells transfected with TBK1s plasmid. Together, these data indicate that the kinase domain of TBK1 is essential for its mediated signaling pathway.

TBK1s Specifically Inhibits SeV-induced IFN-␤ Activation-
It has been reported that TBK1 is essential for Type I interferon production in virus-triggered signaling pathway. Because TBK1s is an alternative splicing isoform of TBK1, which lacks the kinase domain, we hypothesized that TBK1s could play an inhibitory role on this pathway. To test this hypothesis, 293T cells were transfected with TBK1s plasmid or the vector control, and infected with different doses of SeV after 24 h of transfection. In reporter gene assays, TBK1s inhibited SeV-triggered activation of ISRE and IFN-␤ promoter, whereas it slightly enhanced SeV-triggered activation of NF-B (Fig. 3A). Furthermore, we measured the effect of TBK1s on the induction of type I IFN-inducible genes including Ifnb, IP-10, and the NF-B-dependent gene IL-8 by quantitative real-time PCR. As shown in Fig. 3B, TBK1s inhibited expression of Ifnb and IP-10 mRNA induced by SeV, while TBK1s had no effect on expression of IL-8 mRNA. These data suggest that TBK1s specifically inhibits SeV-induced IFN-␤ activation. We also assessed the expression of transfected TBK1s protein. Immunoblot analysis with anti-TBK1 antibody showed that ectopic expression of TBK1s is about once higher than that of endogenous TBK1 protein (Fig.  3C). To further evaluate the effect of TBK1s on the activation and nuclear translocation of IRF3, nuclear extracts from SeVinfected 293T cells were analyzed for the presence of IRF3. The accumulation of IRF3 in the nucleus triggered by SeV was inhibited when 293T cells were transfected with TBK1s plasmid, whereas NF-B subunit p65 translocation was not affected (Fig. 3D). Together, these observations show that TBK1s nega-  tively regulates SeV-triggered IRF3 but not the NF-B activation pathway.
TBK1s Inhibits RIG-I-but Not VISA-or TBK1-mediated Activation of IFN-␤ Promoter-To elucidate the mechanism for the negative effect of TBK1s on SeV-triggered IFN-␤ signaling pathway, we first used reporter gene assays to locate the target of TBK1s. As TBK1s lacks a kinase domain compared with fulllength TBK1, we theorized that TBK1s might inhibit TBK1mediated activation. To address this, 293T cells were transfected with TBK1 plasmid, the indicated reporter, and increasing concentrations of TBK1s plasmid. Unexpectedly, our data revealed that TBK1s had no effect on ISRE or the IFN-␤ promoter activity induced by TBK1 (Fig. 4A, supplemental Fig. S1A). Because TBK1 is activated downstream of RIG-I and VISA in response to SeV, we then investigated the effect of TBK1s on RIG-I-and VISA-mediated activation of ISRE, the IFN-␤ promoter, and NF-B. RIG-I or VISA combined with the indicated reporter were transfected into 293T cells with increasing concentrations of TBK1s. As shown in Fig. 4B, TBK1s inhibited RIG-I-mediated activation of the IFN-␤ promoter in a dose-dependent manner. In contrast, TBK1s had no effect on ISRE or the IFN-␤ promoter activity induced by VISA (Fig. 4C, supplemental Fig. S1C). Furthermore, immunoblot analysis with anti-TBK1 antibody showed that TBK1s could inhibit RIG-I-mediated activation of the IFN-␤ promoter although ectopic expression of TBK1s protein was less than that of endogenous TBK1 protein (Fig. 4B). Consistent with previous data, TBK1s did not inhibit NF-B activity induced by RIG-I, VISA, or TBK1 (supplemental Fig. S1, A-C). These data suggest that TBK1s could target RIG-I.
TBK1s Binds RIG-I and Disrupts the Interactions of RIG-I with VISA-Because TBK1s inhibited RIG-I-but not VISA-or TBK1-mediated signaling, we speculated that TBK1s may sequester RIG-I in an inactive complex and block its interaction with VISA. To test this hypothesis, we first determined whether TBK1s could interact with RIG-I or VISA. Coimmunoprecipitation assays revealed that the splice variant TBK1s but not full-length TBK1 can interact with RIG-I (Fig. 5, A and B).
However, the interaction between TBK1s and VISA could not be detected (data not shown). Furthermore, to map the binding regions for RIG-I and TBK1s, we generated a series of FLAG-tagged truncation mutants of RIG-I and HA-tagged truncation mutants of TBK1s. As shown in Fig. 5, C and D, the coiledcoil domain of TBK1s (amino acids 372-501) and the CARDs domain of RIG-I (amino acids 1-128) were found to be responsible for the interaction. It has been demonstrated that RIG-I initiates downstream signaling cascades via CARDs by association with VISA (13)(14)(15)(16). Because TBK1s interacts with RIG-I via the CARDs domain of RIG-I, we assessed whether TBK1s can disrupt the interaction of RIG-I with VISA. The vector, TBK1s plasmid or TBK1s mutant 1-371 were transfected into 293T cells, and then the cells were infected with SeV. It was found that the interaction between RIG-I and VISA could be detected upon SeV infection, while ectopic expression of wild-type TBK1s in 293T cells was able to block the endogenous interaction of RIG-I with VISA (Fig. 5E). Furthermore, the TBK1s mutant 1-371, which lacks the coiled-coil domain, did not efficiently inhibit the interaction of RIG-I with VISA as did wild-type TBK1s. Consistent with these data, the TBK1s mutant 1-371 was unable to effectively inhibit SeV-triggered activation of the IFN-␤ promoter compared with wild-type TBK1s (Fig. 5F). To illustrate the physiological role of TBK1s, we also analyzed the endogenous interaction of TBK1s with RIG-I. We performed co-immunoprecipitation experiments of lysates of L929 after SeV infection, which showed specific interaction between endogenous RIG-I and TBK1s (Fig. 5G). In contrast, the endogenous interaction between TBK1 and RIG-I could not be detected (Fig. 5G). In conclusion, the splice variant TBK1s negatively regulated SeV-triggered IFN-␤ signaling pathway through disrupting the interaction of RIG-I with VISA.

DISCUSSION
Negative control of Type I IFN production is an essential physiological process and can be achieved at multiple levels. For instance, LGP2, a RNA helicase protein, which lacks CARD domains at its N terminus, can be induced by virus and serves as a negative regulator by sequestering viral RNA from RIG-I and MDA5 (29 -32). RNF125, an E3 ubiquitin ligase, negatively regulates RIG-I-induced antiviral signaling through mediation of RIG-I degradation (33). In this study, we demonstrate a new mechanism of negative regulation of Type I IFN production by alternative splicing of TBK1 pre-mRNA.
Alternative splicing is a mechanism that allows for individual genes to express multiple mRNAs that encode proteins with diverse and even antagonistic function. Indeed, alternative splicing has been described for various adaptors and transcrip- tion factors involved in antiviral response, including IRF family members (34 -39). For instance, it has been reported that IRF-5 alternatively spliced isoforms are differentially regulated by at least two alternative promoters, while the splicing of IRF-3 and IRF-3a transcripts may be regulated in a tissue-specific manner (34,35). Here, we demonstrated that the expression of TBK1s, as compared with constitutive expression of TBK1, can be induced by SeV infection or IFN␤ treatment. Furthermore, expression of Tbk1s was significantly higher in peripheral blood mononuclear cell (PBMC) from the HCV-infected patients than those in the healthy controls, while it was down-regulated in PBMC from the HCV-infected patients treated with IFN-␣/ ribavirin whose HCV RNA turn to negativity (HCV Neg) (supplemental Fig. S2 and Table S1). Consistent with a previous report (40), the expression of Tbk1 was not found to alter significantly upon HCV infection. Such a regulated production of TBK1s would result in the controlled inhibition of type I IFN production. Intriguingly, the ratio of TBK1s protein expression versus TBK1 protein expression in heart and kidney were really high, indicating TBK1s could be regulated in a tissue-specific manner. However, differential isoform expression and regulation of TBK1 remain to be addressed.
Because TBK1s does not contain the kinase domain but does maintain the ability to bind IRF3 (data not shown), it was proposed that the mechanism by which TBK1s inhibits virus-triggered signaling pathway probably involves sequestration of TBK1 from IRF3, thereby preventing IRF3 activation. While the present study is in agreement with the negative regulatory role FIGURE 5. TBK1s binds to RIG-I and disrupts the interactions of RIG-I with VISA. A-E, immunoprecipitation (IP) and immunoblot (IB) analysis of lysates of HEK293T cells expressing various recombinant proteins (total amount: 2 g), using the indicated antibodies. A, FLAG-RIG-I together with HA-TBK1 or HA-TBK1s. B, HA-TBK1 or HA-TBK1s together with vector or FLAG-RIG-I. C, schematic diagram of TBK1s truncation mutants used in this study (above). ULD, ubiquitin-like domain. CC, coil-coiled domain. FLAG-RIG-I (WT) together with HA-TBK1s wild type or truncation mutants. D, schematic diagram of RIG-I truncation mutants used in this study (above). CARD, caspase recruitment domain. HA-TBK1s together with the indicated FLAG-RIG-I wild type or truncation mutants. Arrow indicated HA-TBK1s. E, HEK293T cells (1 ϫ 10 7 ) were transfected with an empty vector, HA-TBK1s expression vector or TBK1s mutant 1-371(4 g). After 24 h of transfection, cells were infected with 10 HAU/ml SeV for 4 h. F, HA-TBK1s or TBK1s mutant 1-371 was transfected into 293T cells together with pRL-TK and IFN-␤ promoter reporter. After 24 h of transfection, cells were infected with 10 HAU/ml SeV for 16 h and subjected to the dual-luciferase assay (above) and immunoblot assay with anti-HA antibody (below). Error bars indicate Ϯ S.D. between duplicates. **, p Ͻ 0.01. G, L929 cells (1 ϫ 10 7 ) were infected with 10 HAU/ml SeV for the indicated time. Cell lysates were incubated with 2 l of RIG-I antibody or control IgG for immunoprecipitation followed by immunoblotting, using the indicated antibodies. The ratio of TBK1-TBK1s was shown as indicated.
of TBK1s, differences were found in the proposed mechanism of inhibition. Actually, TBK1s inhibits RIG-I-but not TBK1mediated activation of ISRE and the IFN-␤ promoter, and TBK1s negatively regulates virus-triggered IFN-␤ signaling pathway by disrupting the interaction of RIG-I with VISA. It is reasonable to assume that TBK1s targets RIG-I instead of TBK1. First, the expression of TBK1s is lower than that of TBK1 even upon Sendai virus infection (Figs. 1 and 5G); second, IRF3 phosphorylation mediated by TBK1 is not blocked in the presence of TBK1s (data not shown).
In this study, we show that the coiled-coil domain of TBK1s is responsible for the interaction between RIG-I and TBK1s. Actually, full-length TBK1 also contains the coiled-coil domain. It is not yet clear why TBK1 does not bind to RIG-I but TBK1s does. Given the fact that kinase-defective TBK1 (K38A) can bind to RIG-I, we suggest that it would be related to the kinase activity (supplemental Fig. S2). Moreover, RIG-I binds to TBK1s via its CARDs. It has been reported the Lys-172 residue of RIG-I is critical for efficient TRIM25-mediated ubiquitination and for VISA binding (41). In fact, RIG-I interacts with TBK1s through the Lys-172 and Lys-193 residues of RIG-I (supplemental Fig. S3). These data further supported the idea that TBK1s negatively regulates virus-triggered IFN-␤ signaling pathway by disrupting the interaction of RIG-I and VISA.
It is notable that TBK1s negatively regulates virus-triggered IRF-3 but not NF-B activation pathways. It is most likely that TBK1s is similar to MyD88s, a splice variant of MyD88 that differentially modulates NF-B and AP-1-dependent gene expression in the TLR/IL-1R signaling pathway (42)(43)(44). We proposed other domains in TBK1s such as ubiquitin-like domain (ULD) might be responsible for differential regulation of the IRF3 and NF-B pathway by TBK1s. Considerable further experimentation will be necessary to clarify the precise mechanism. We tried to further explore the physiological role of TBK1s by siRNA (data not shown). Unfortunately, neither of the two pairs of siRNA we designed based on the difference of the splicing sites could specifically knockdown TBK1s expression. Nonetheless, the regulated expression of TBK1s, endogenous interaction of TBK1s with RIG-I, and higher expression of TBK1s in HCV-infected patients indicate that TBK1s play an important role in modulating the virus-triggered signaling pathway.