Severe Acute Respiratory Syndrome Coronavirus M Protein Inhibits Type I Interferon Production by Impeding the Formation of TRAF3·TANK·TBK1/IKKϵ Complex*

Severe acute respiratory syndrome (SARS) coronavirus is highly pathogenic in humans and evades innate immunity at multiple levels. It has evolved various strategies to counteract the production and action of type I interferons, which mobilize the front-line defense against viral infection. In this study we demonstrate that SARS coronavirus M protein inhibits gene transcription of type I interferons. M protein potently antagonizes the activation of interferon-stimulated response element-dependent transcription by double-stranded RNA, RIG-I, MDA5, TBK1, IKKϵ, and virus-induced signaling adaptor (VISA) but has no influence on the transcriptional activity of this element when IRF3 or IRF7 is overexpressed. M protein physically associates with RIG-I, TBK1, IKKϵ, and TRAF3 and likely sequesters some of them in membrane-associated cytoplasmic compartments. Consequently, the expression of M protein prevents the formation of TRAF3·TANK·TBK1/IKKϵ complex and thereby inhibits TBK1/IKKϵ-dependent activation of IRF3/IRF7 transcription factors. Taken together, our findings reveal a new mechanism by which SARS coronavirus circumvents the production of type I interferons.

Severe acute respiratory syndrome (SARS) 2 coronavirus causes a highly lethal infectious disease in humans characterized by an aberrant immune response (1). The production and action of type I interferons, which are major components of antiviral innate immunity (2,3), are inhibited at multiple levels by SARS coronavirus (4,5). This inhibition is thought to be mediated through viral structural and nonstructural proteins N, ORF3b, ORF6, nsp1, and papain-like protease (6 -12).
The signaling pathways through which viruses induce the production of type I interferons have been well characterized (13)(14)(15). In response to double-stranded RNA (dsRNA) produced during viral replication, endosomal Toll-like receptor 3 (TLR3) and cytoplasmic retinoic acid-inducible gene I (RIG-I) trigger the activation of two different pathways adapted to downstream kinases through TRIF (TLR adaptor inducing interferon ␤) and VISA, respectively. These pathways converge on the formation of TRAF3⅐TANK⅐TBK1/IKK⑀ complex, which catalyzes the phosphorylation of IRF3 and IRF7 transcription factors, leading to the activation of type I interferon promoters (15)(16)(17).
SARS coronaviral proteins counteract the production of type I interferons at multiple steps. Although IRF3 phosphorylation was inhibited in cells expressing ORF3b, ORF6, or N protein (7), papain-like protease could physically interact with IRF3 and prevent its phosphorylation and nuclear translocation in a protease-independent manner (11). In addition, nsp1 suppressed the synthesis of host proteins including interferons by inducing mRNA degradation (6,12). Meanwhile, viral proteins such as nsp1 and ORF6 were multifunctional (10,18,19) and could also inhibit interferon signaling. For example, both nsp1 and ORF6 inhibited the activity of STAT1, a key regulator of interferon-responsive genes (8,9). Whereas nsp1 attenuated phosphorylation of STAT1 (9), ORF6 modulated nuclear import and blocked nuclear translocation of STAT1 by sequestering nuclear import factors in the endoplasmic reticulum and Golgi apparatus (8,19).
SARS coronavirus M protein is a glycosylated structural protein with three membrane-spanning domains (20 -23). M protein predominantly localizes to the Golgi complex and is essential for the assembly of viral particles (24,25). Whereas mutations in M protein of an animal coronavirus named transmissible gastroenteritis virus led to significantly reduced induction of type I interferons (26), an adaptive mutation in SARS coronavirus M protein was found to enhance viral replication and/or infectivity in cultured human cells (27). Additionally, SARS coronavirus M protein was also suggested to modulate apoptosis and the expression or activity of cellular proteins such as NFB and Akt (28 -30).
In our study of the modulation of cellular function by SARS coronavirus structural proteins (31,32), we found that M protein suppressed type I interferon production more potently than did N protein or influenza A virus NS1 protein. We went on to dissect the signaling pathway targeted by M protein and demonstrated its inhibition of the formation of TRAF3⅐TANK⅐TBK1/IKK⑀ complex. Our work reveals a new coronaviral countermeasure against host innate immunity.
Protein Analysis and Reporter Assay-Immunoprecipitation, Western blotting, and dual luciferase assay were carried out as described (41,42). Relative luciferase activity in arbitrary units was calculated by normalizing firefly luciferase activity to Renilla luciferase activity.

Inhibition of Type I Interferon Production by SARS Coronavi-
rus M Protein-SARS coronavirus N protein has been shown to antagonize the production of type I interferons (7). In our study of the gene regulatory function of N protein (32), we included M protein as a control and found serendipitously that M protein was a more potent inhibitor of interferon ␤ promoter when compared with N protein (Fig. 1). Considered together with the ability of M protein to interact with host factors and to modu-late expression of cellular genes (25)(26)(27)(28)(29), we set out to characterize M protein-mediated inhibition of type I interferon production in detail.
As a first step, we expressed M protein transiently in HEK293 and HeLa cells. The M protein-expressing cells did not show any growth defect, and no signs of apoptosis were observed (Fig. 1A). M protein was abundantly found in transfected cells (Fig. 1B). Consistent with a predominant Golgi localization shown in previous reports (21,23), M protein was found to co-localize substantially with GM130 (Fig. 1A), a marker of the Golgi apparatus (45). Expression of M protein did not suppress transcription driven by the long terminal repeats of human T-cell leukemia virus type 1, which was stimulated by viral transactivator Tax (Fig. 1C). Hence, M protein is unlikely a general inhibitor of RNA polymerase IIdependent transcription. Next, we assessed the influence of M protein expression on the induction of interferon ␤ gene transcription using pIFN␤-Luc, a luciferase reporter construct driven by interferon ␤ promoter (33). Notably, expression of M protein led to significant and dose-dependent suppression of dsRNA-induced activation of interferon ␤ promoter (Fig.  1D). This suppression by M protein was more pronounced than the effect of two previously reported viral antagonists of interferon production, SARS coronavirus N protein and influenza A virus NS1 protein (7,39). In contrast, E or ORF7a protein of SARS coronavirus did not modulate the activation of interferon ␤ promoter by dsRNA (Fig. 1D). When we repeated the luciferase assay with a reporter construct driven by canonical ISRE enhancer elements, which were bound to IRF3 and IRF7 (36,46,47), the same pattern of inhibitory activity mediated by M, N, and influenza A virus NS1 proteins was observed (Fig. 1E). These results suggested that SARS coronavirus M protein is a potent inhibitor of the activity of IRF-responsive elements in the interferon ␤ promoter.
Interestingly, expression of M protein suppressed ISREdriven transcriptional activity induced by RIG-I, MDA5, VISA, TBK1, and IKK⑀ but not by IRF3 or IRF7 (Fig. 2). In these assays, SARS coronavirus N, E, or ORF7a protein was unable to counteract any ISRE activator. Thus, although both N and M proteins inhibited interferon production (7), they plausibly acted through different mechanisms. Consistent with previous findings (48 -50), influenza A virus NS1 protein inhibited the stimulatory activity of RIG-I, MDA5, TBK1, and IKK⑀ (Fig. 2, A ϳ  D). However, unlike SARS coronavirus M protein, NS1 had no influence on the activity of VISA (Fig. 2E). These results suggested that M protein suppressed interferon production in a manner distinct from that of NS1.
Association of SARS Coronavirus M Protein with TBK1/IKK⑀-M protein inhibited TBK1/IKK⑀-dependent activation of interferon production but was unable to counteract the action of transcription factors IRF3 and IRF7 (Fig. 2). This pattern of inhibitory activity suggested that M protein might function at a step before nuclear activation of IRF3 and IRF7. In light of this, we next investigated whether M protein physically interacted with the cytoplasmic transducer proteins. We overexpressed M and individual transducer proteins RIG-I, MDA5, TBK1, IKK⑀, TANK, TRAF3, and TRAF6 in HEK293 cells and performed reciprocal co-immunoprecipitation to assess their interactions (Fig. 3). Although both M protein and the transducer proteins were expressed abundantly in HEK293 cells (Fig. 3A), M protein was found to co-precipitate with RIG-I, TBK1, IKK⑀, and TRAF3 but not with MDA5, TANK, or TRAF6 (Fig. 3, B and C). Notably, precipitations with two different antibodies yielded similar results ( Fig. 3B compared with Fig. 3C), suggesting that M protein consistently formed a complex with RIG-I, TBK1, IKK⑀, and TRAF3. In keeping with this notion, M protein was found to co-localize substantially with IKK⑀ and TRAF3 in HeLa cells to discrete cytoplasmic subdomains that were compatible with the Golgi complex (Fig. 4). Particularly, IKK⑀ was sporadically found in multiple cytoplasmic dots in the absence of M protein (Fig. 4, panels 1-3). In contrast, IKK⑀ was more concentrated to the M protein-containing region in cells expressing M protein (Fig. 4, panels 7-9). SARS Coronavirus M Protein Impedes TRAF3⅐TANK⅐TBK1/ IKK⑀ Complex Formation and IRF3 Phosphorylation-The association of M protein with TBK1/IKK⑀ (Fig. 3) raised two  possibilities that might explain the inhibition of interferon production. First, M protein might inhibit catalytic activity of TBK1/IKK⑀ to phosphorylate IRF3/IRF7. Second, M protein might interfere with the formation of TRAF3⅐TANK⅐TBK1/ IKK⑀ complex required for phosphorylation of IRF3/IRF7 in the cytoplasm. To test these possibilities, we assessed the impact of M protein on the kinase activity of TBK1/IKK⑀ both in vitro and in vivo. We expressed M protein and TBK1/IKK⑀ in HEK293 cells (Fig. 5A) and immunoprecipitated TBK1/IKK⑀containing protein complex (Fig. 5B). The precipitates were then incubated with recombinant IB␣ or recombinant IRF3. Phosphorylated IB␣ and IRF3 were analyzed by autoradiography (Fig. 5, C and D). Notably, the TBK1/IKK⑀ complex purified from M protein-expressing cells contained M protein (Fig. 3C). However, the TBK1/IKK⑀ complexes containing or not containing M protein showed similar activity to phosphorylate IB␣ or IRF3 (Fig. 5, C and D On the other hand, phosphorylated IRF3 was detected in cultured HEK293 cells expressing TBK1/IKK⑀ alone but not in cells simultaneously expressing TBK1/IKK⑀ and M protein, whereas comparable amounts of IRF3 were found in all different groups of cells (Fig. 6, lane 2 compared with lane 3 and lane  4 compared with lane 5). These results were consistent with the notion that M protein inhibits in vivo phosphorylation of IRF3 by TBK1/IKK⑀.
To verify the relevance of M protein-mediated inhibitory effects to SARS coronavirus infection, we also examined phosphorylation of IRF3 and production of interferon ␤ transcript in cells infected with SARS coronavirus. Both phosphorylated IRF3 and elevated expression of interferon ␤ mRNA were detected in cells infected with Sendai virus (Fig. 7, A and C). In contrast, although IRF3 was abundantly found in cells infected with SARS coronavirus, neither phosphorylation of IRF3 nor induced expression of interferon ␤ transcript was observed. These results were consistent with previous findings (4 -12).
The formation of TRAF3⅐TANK⅐TBK1/IKK⑀ complex in the cytoplasm is an essential step in the activation of IRF3/IRF7 (15)(16)(17)34). M protein physically interacted with TBK1/IKK⑀ and TRAF3 (Fig. 3). In addition, they also colocalized to the cytoplasmic subdomains that were likely the Golgi apparatus (Fig. 4). Plausibly, M protein might impede the formation of a functional TRAF3⅐TANK⅐TBK1/IKK⑀ complex by masking the interaction domains or retaining some of the subunits in the Golgi complex. Indeed, when we co-expressed TBK1, TRAF3, and M in HEK293 cells, the interaction between TBK1 and TRAF3, which was evident in cells expressing TBK1 and TRAF3, was not detected (Fig. 8A, lane 2 compared with lane  1). Likewise, the interaction between IKK⑀ and TRAF3 was not observed in M protein-expressing cells (Fig. 8A, lane 4 compared with lane 3). Furthermore, the formation of TRAF3-TANK complex was also blocked in the presence of M protein (Fig. 8A, lane 6 compared with lane 5). Hence, M protein impeded the formation of a functional TRAF3⅐TANK⅐TBK1/ IKK⑀ complex. Importantly, the formation of this complex was also impaired in SARS coronavirus-infected cells (Fig. 8B). As such, the interaction of TRAF3 with TBK1, IKK⑀, and TANK was impeded in infected cells where N protein was expressed (Fig. 8B, lanes 2, 4, and 6 compared with lanes 1, 3, and 5,  respectively). Although the roles of other viral proteins could not be excluded, the same pattern for the impeded formation of TRAF3⅐TANK⅐TBK1/IKK⑀ complex in cells expressing M protein alone and cells infected with SARS coronavirus supported the notion that M protein contributes significantly to the inhibition of type I interferon production during viral infection.

DISCUSSION
We demonstrated in this study that SARS coronavirus M protein inhibited the production of type I interferons through a new mechanism. M protein predominantly localized to the Golgi complex and potently suppressed dsRNA-induced expression of type I interferon genes (Fig. 1). M protein exerted its inhibitory effect on cytoplasmic dsRNA sensors RIG1 and MDA5 as well as other transducer proteins VISA, TBK1, and IKK⑀, but it could not counteract the activation mediated directly by IRF3 and IRF7 (Fig.  2). M protein interacted with RIG-I, TBK1, IKK⑀, and TRAF3 in the cytoplasm (Fig. 3) and likely sequestered some of them in discrete compartments that were consistent with the membrane-associated Golgi apparatus (Fig. 4). Although M protein did not directly inhibit the catalytic activity of TBK1/IKK⑀ kinase (Fig. 5), it impeded the formation of a functional TRAF3⅐TANK⅐TBK1/IKK⑀ complex (Fig. 8), thereby inhibiting the phosphorylation of IRF3 (Figs. 6 and 7). Importantly, formation of the TRAF3⅐TANK⅐TBK1/IKK⑀ complex, phosphorylation of IRF3, and production of interferon ␤ were also inhibited in SARS coronavirus-infected cells (Figs. 7 and 8).
Our findings suggested a new role for M protein in pathogenesis of SARS coronavirus. This is not surprising a since M protein was previously shown to modulate cellular signaling and apoptosis (28 -30). Interestingly, an adaptive mutant of M protein was found to enhance viral replication and/or infectivity (27). Because a reduction in interferon-inducing activity was shown to be associated with M protein mutations in transmissible gastroenteritis virus (26), M protein of SARS coronavirus might also influence viral replication through inhibition of interferon production. Thus, it will be of great interest to understand whether the particular mutation in SARS coronavirus M protein might influence its ability to inhibit the induction of type I interferons. A better understanding of the role of M protein in pathogenesis of SARS has important implications in the development of new antivirals and vaccines.
Other SARS coronaviral proteins including ORF3b, ORF6, N, nsp1, and papain-like protease were also known to inhibit interferon production and/or signaling (6 -12). However, their modes of action were thought to be different from that of M protein. Particularly, our detailed comparison of M and N proteins in their inhibition of various activators (Figs. 1 and 2) suggested that they targeted different steps. Although the requirement of M protein for virion assembly prevents us from studying interferon production with M protein deletion mutants of SARS coronavirus, the coordinated actions of different interferon-antagonizing viral proteins in infected cells merit further investigations. This also poses major technical challenges for future attempts to verify and distinguish the contributions from different viral proteins during infection. Identification of M protein mutants lacking the interferon-antagonizing activity alone might prove useful in future studies.
Viral countermeasures for interferon production and signaling have been well described (17). Compared with influenza A virus NS1 protein, an extensively studied viral antagonist of interferon (48 -51), SARS coronavirus M protein displayed a similar pattern of inhibitory activity on RIG-1, MDA5, TBK1, and IKK⑀ (Fig. 2). However, M protein inhibited the activity of  VISA, whereas NS1 did not (Fig. 2E). Additional experiments are required to elucidate the underlying cause of this difference between M and NS1 proteins. Both TRAF3 and TANK are known to be important for interferon response (34,52,53). Our findings that M protein associated with TRAF3 (Fig. 3) and impeded its interaction with TANK, TBK1, and IKK⑀ (Fig. 7) revealed a new mechanism for modulation of interferon production by SARS coronavirus. In this regard, NY-1 hantavirus Gn cytoplasmic tail was recently shown to co-precipitate with TRAF3 and inhibit the formation of TRAF3-TBK1 complex (54), whereas Ebola virus VP35 was able to impair the function of TBK1/IKK⑀ by preventing their interaction with VISA and IRF3/IRF7 (55). Plausibly, inhibition of the formation of a functional TRAF3-TBK1/IKK⑀-IRF3/IRF7 complex might be a common mechanism through which viruses inhibit interferon production.
Cytoplasmic modulators of interferon induction have just begun to be understood. Whereas NLRX1 and MULAN are mitochondrial proteins that interact with VISA to inhibit RIG-I activity (56,57), DDX3X interacts with TBK1 directly (58) and is targeted by vaccinia virus K7 protein (59). In this connection, our demonstration of a coronaviral inhibitor of interferon production not only adds a new member to the group of viral and cellular modulators of interferon response but might also reveal new targets and strategies in the design of new antiviral agents.