key: cord-252671-uf96jgig
authors: Wang, Yi; Liu, Li
title: The Membrane Protein of Severe Acute Respiratory Syndrome Coronavirus Functions as a Novel Cytosolic Pathogen-Associated Molecular Pattern To Promote Beta Interferon Induction via a Toll-Like-Receptor-Related TRAF3-Independent Mechanism
date: 2016-02-09
journal: mBio
DOI: 10.1128/mbio.01872-15
sha: 
doc_id: 252671
cord_uid: uf96jgig

Most of the intracellular pattern recognition receptors (PRRs) reside in either the endolysosome or the cytoplasm to sense pathogen-derived RNAs, DNAs, or synthetic analogs of double-stranded RNA (dsRNA), such as poly(I:C). However, it remains elusive whether or not a pathogen-derived protein can function as a cytosolic pathogen-associated molecular pattern (PAMP). In this study, we demonstrate that delivering the membrane gene of severe acute respiratory syndrome coronavirus (SARS-CoV) into HEK293T, HEK293ET, and immobilized murine bone marrow-derived macrophage (J2-Mφ) cells significantly upregulates beta interferon (IFN-β) production. Both NF-κB and TBK1-IRF3 signaling cascades are activated by M gene products. M protein rather than M mRNA is responsible for M-mediated IFN-β induction that is preferentially associated with the activation of the Toll-like receptor (TLR) adaptor proteins MyD88, TIRAP, and TICAM2 but not the RIG-I signaling cascade. Blocking the secretion of M protein by brefeldin A (BFA) failed to reverse the M-mediated IFN-β induction. The antagonist of both TLR2 and TLR4 did not impede M-mediated IFN-β induction, indicating that the driving force for the activation of IFN-β production was generated from inside the cells. Inhibition of TRAF3 expression by specific small interfering RNA (siRNA) did not prevent M-mediated IFN-β induction. SARS-CoV pseudovirus could induce IFN-β production in an M rather than M(V68A) dependent manner, since the valine-to-alanine alteration at residue 68 in M protein markedly inhibited IFN-β production. Overall, our study indicates for the first time that a pathogen-derived protein is able to function as a cytosolic PAMP to stimulate type I interferon production by activating a noncanonical TLR signaling cascade in a TRAF3-independent manner.

In addition to the TLR, which can be defined as a membraneassociated PRR, another set of PRRs is localized at the cytoplasm and mainly includes RIG-like receptors (RLRs) and NOD-like receptors (NLRs) to sense viral dsRNAs and bacterial cell wall components, respectively (2, 7) . The RLRs consist of at least three members, including RIG-I, MDA5, and LGP2. RIG-I recognizes 5=-triphosphate RNA and short dsRNA (4, 8) , while MDA5 senses long dsRNA (9) . An adaptor protein, MAVS, is required for the activation of the RIG-I/MDA5 signaling pathway. The association of viral nucleic acids with MAVS promotes the aggregation of MAVS on the mitochondrial membrane (10) . The "ligation" of TRAF3 with the aggregated MAVS may promote the phosphorylation of IRF3 that is required for IFN-␤ production (11) . A recent study also shows that an endoplasmic reticulum (ER)-derived adaptor protein, STING, could also function downstream of MAVS to promote IRF3 phosphorylation and the subsequent IFN-␤ response (12) .

Pathogen-derived proteins such as virus-encoded proteins are frequently documented as negative regulators in subverting type I interferon (IFN-I) induction by interfering with a certain key component(s) of IFN-I activation signaling cascades. Viral evolution may develop a unique strategy to inhibit host innate immunity by generating virus-derived antagonists to some key signaling molecules. The vaccinia virus encodes two Toll/interleukin-1 (IL-1) receptor (TIR) domains containing proteins A46R and A52R, which can negatively regulate TLR signaling by two distinct mechanisms (13) . The vaccinia virus A46R inhibits TLR signaling by physically interacting with the BB loop of TIR containing adaptor proteins such as MyD88 adaptor-like (MAL) and TRIF-related adaptor molecule (TRAM) to disrupt receptor-adaptor (e.g., TLR4-MAL and TLR4-TRAM) interactions (14, 15) . Differently, the A52R protein may function as a dominant negative MyD88 to directly interact with TRAF6 and IRAK2 (16, 17) . On the other hand, the vaccinia virus N1L protein, another protein homologous to A52R, employs a different anti-IFN-I strategy by targeting both the TBK1/IB kinase (IKK) and IKK␣/IKK␤ complexes to inhibit IRF3 and NF-B signaling, respectively (18) . Alternatively, virus may invade the cells to target the retinoic acid-inducible gene I (RIG-I)-like receptor signaling pathways for the prevention of IFN-I induction. For example, the influenza virus nonstructural protein NS1 can sequester either the dsRNA or 5=triphosphate RNA products of viral infection which can be sensed by or directly bound to the RNA helicase sensor RIG-I to inhibit RIG-I-mediated IFN-␤ production (8, 19, 20) . The paramyxovirus V protein inhibits IFN-␤ induction through the blockage of MDA5, another RIG-I-homologous cytosolic dsRNA sensor (21) . A recent study revealed that the transcriptional factor IRF3 might be alternatively targeted and inhibited by the paramyxovirus V protein to impede IFN-␤ gene transcription (22) .

It has been demonstrated in some cases that viral proteins may function as extracellular PAMPs to activate the IFN-I immune response, most often through TLR (such as TLR2 and TLR4) signaling pathways (23) (24) (25) . However, evidence is lacking in regard to whether or not a virus-derived protein can function as a cytosolic PAMP.

Our initial study indicates that delivering the membrane gene into HEK293 cells markedly induces type I interferon (IFN-I) production (26) . To our knowledge, there are limited reports regarding the induction of IFN-␤ expression directly by viral structural genes. Therefore, it is intriguing to know which mechanism is responsible for the severe acute respiratory syndrome coronavirus (SARS-CoV) M gene-mediated IFN-␤ response. In this study, we demonstrate that SARS-CoV M protein, rather than its mRNA, activates IFN-␤ and NF-B responses through TLR-related TRAF3-independent signaling cascades. The driving force for M-mediated IFN-␤ induction was most likely generated from the inside of the cells. Using SARS-CoV pseudovirus as an infectious agent, we further show that single point mutation at the valine 68 residue of M protein markedly inhibits virus-induced IFN-␤ production. Overall, SARS-CoV M protein may stand out as a novel cytosolic PAMP in mediating the IFN-␤ immune response.

The SARS-CoV M gene stimulates beta interferon gene expression in the human embryonic kidney 293T cell line. The overexpression of the SARS-CoV M gene has been shown to upregulate the transcriptional level of IFN-␤ (26) . To further confirm the result, using either enhanced green fluorescent protein (EGFP) or poly(I:C) as a negative or positive control, respectively, we demonstrated that M gene products specifically promoted IFN-␤ production, since cotransfection with M small interfering RNA (siRNA) completely abolished M-mediated IFN-␤ induction at both protein (Fig. 1A , comparing lanes 3 and 4) and mRNA ( Fig. 1B ) levels. Moreover, after a 48-h transfection, cell supernatants were collected and assayed for the presence of IFN-␤ by enzyme-linked immunosorbent assay (ELISA). Figure 1C clearly demonstrated that delivering pCMV-Myc-M into HEK293T cells specifically and significantly promoted the secretion of IFN-␤ into cell culture medium. In addition, the promoter sequence of IFN-␤ was placed upstream of the firefly luciferase reporter to generate the pGL3-IFN-␤-Luc construct. To test the specificity of M-mediated IFN-␤ induction, other viral envelope-associated genes such as the spike (S) and envelope (E) protein genes as well as the M mutant [M(V68A)] from the GZ50 isolate were also included (27) . The result of a dual-luciferase assay using the Renilla luciferase gene as a transfection control demonstrated that the SARS-CoV M gene rather than the S and E genes markedly increased IFN-␤ promoter activity (Fig. 1D) , whereas the valineto-alanine alteration at residue 68 of M protein completely abolished this induction, indicating that the specificity of M gene products played a role in this process. Consistent with these results, Western blotting and ELISA further validated the above observation ( Fig. 1E and F). To detect if the SARS-CoV M gene has a direct effect on NF-B activation, pCMV-Myc-M was cotransfected with pNFB-luc, which contained five copies of NF-B recognition sites, into HEK293T cells. The results of the dualluciferase assay revealed that the SARS-CoV M gene specifically and dramatically upregulated NF-B activity compared with the controls (Fig. 1G) . Moreover, M could mediate IFN-␤ induction in both dose-and time-dependent manner ( Fig. 1H and I) . Overall, the data strongly indicated that the SARS-CoV M gene product was sufficient to promote IFN-␤ gene expression.

The SARS-CoV M gene product activates the IFN-␤ signaling pathway at or upstream of TBK1. To further confirm the above results, increased doses of the pCMV-Myc-M gene were transiently transfected into HEK293ET cells. The cell lysates prepared from the transfection were examined for the activation of the downstream modulator and/or effector molecules, such as TBK1, IRF3, and NF-B. Figure 2A clearly demonstrates that SARS-CoV M gene products not only enhanced the phosphorylation level of TBK1 but also promoted the activation of both NF-B p65 and IRF3, indicating that M gene products may stimulate IFN-␤ activation by promoting its enhanceosome activity. To further define the activation level of M-mediated IFN-␤ induction, specific siR-NAs that selectively targeted either TBK1 ( Fig. 2B and C) or IRF3 ( Fig. 2E and F) mRNAs were generated. Individually diminishing either TBK1 or IRF3 mRNA expression by siTBK1 or siIRF3 sig-nificantly reversed M-mediated IFN-␤ induction ( Fig. 2D and G, respectively), indicating that M-mediated IFN-␤ induction functions at a level at or above the signaling molecule TBK1. The SARS-CoV M gene product preferentially activates IFN-␤ production through Toll-like-receptor-related signaling pathways in HEK293ET cells. RLR and TLR are two main PRRs

HEK293T cells. After 0, 6, 12, and 24 h of transfection, dual-luciferase assays were performed to detect IFN-␤ expression. Each value represents the mean Ϯ standard deviation from three independent tests. *, P Յ 0.05; **, P Յ 0.01. In all data presented above, the relative luciferase activity was determined as firefly luciferase activity divided by Renilla luciferase activity. The effect of TBK1 siRNA (siTBK1) on the expression of endogenous TBK1 by semiquantitative RT-PCR. The increased doses of pBS/U6-siTBK1 plasmid DNAs (0, 0.5, and 2 g) were transiently transfected into HEK293ET cells. Total RNAs or whole-cell lysates were isolated or harvested at 48 h posttransfection. One-step RT-PCR (RT) was conducted to detect the TBK1 mRNA expression with specific primers (upper panel), while Western blotting (WB) was performed to detect TBK1 protein expression using specific anti-TBK1 antibody (lower panel). The expression of ␤-actin served as a loading control. The relative band intensity was quantitated with the Image J program in comparison with the ␤-actin control. The result is representative of at least 2 identical experiments. (C) Effect of siTBK1 on the expression of TBK1 mRNAs by real-time qRT-PCR analysis. Total RNAs isolated in panel B were subjected to qRT-PCR analysis using specific TBK1 primers. Each value represents the mean Ϯ standard deviation from three reactions. The result is representative of at least 2 identical experiments. (D) Effect of siTBK1 on M-mediated IFN-␤ induction. Plasmid pGL3-IFN-␤-luc reporter was cotransfected with 2 g of pCMV-Myc-M or 2 g of pCMV-Myc-M plus increased doses of siTBK1 (0, 0.5, and 2 g) into HEK293ET cells grown on a 12-well plate. At 48 h posttransfection, the dual-luciferase assay was conducted to assay fold induction of M-mediated IFN-␤ expression. Each value represents the mean Ϯ standard deviation from three independent tests. (E) Effect of IRF3 siRNA (siIRF3) on expression of endogenous IFR3 by semiquantitative RT-PCR. The increased doses of pBS/U6-siIRF3 plasmid DNAs (0, 0.5, and 2 g) were transiently transfected into HEK293T cells. Total RNAs or whole-cell lysates were isolated or harvested at 48 h posttransfection. One-step RT-PCR was conducted to detect IRF3 expression with specific primers (upper panel), while Western blotting was performed to detect IRF3 protein expression (lower panel). The expression of ␤-actin served as a loading control. The result is representative of at least 2 identical experiments. (F) Effect of siIRF3 on M-mediated IFN-␤ mRNA expression by real-time qRT-PCR analysis. Real-time qRT-PCR was performed to detect the IRF3 mRNA (isolated in panel E) expression. Each value represents the mean Ϯ standard deviation from three reactions. The result is representative of at least 2 identical experiments. (G) Effect of siIRF3 on M-mediated IFN-␤ induction by luciferase assay. Plasmid pGL3-IFN-␤-luc reporter was cotransfected with 2 g of pCMV-Myc-M or 2 g of pCMV-Myc-M plus increased doses of siIRF3 (0, 0.5, and 2 g) into HEK293ET cells grown on a 12-well plate. At 48 h posttransfection, the dual-luciferase assay was conducted to assay fold induction of M-mediated IFN-␤ expression. Each value represents the mean Ϯ standard deviation from three independent tests. recognizing the majority of extracellular and intracellular PAMPs. Upon the ligation of a PRR with its specific PAMP, both RLR and TLR pathways transmit the signal to a common class of adaptors called tumor necrosis factor (TNF) receptor-associated factors (TRAFs) including TRAF2/TRAF5, TRAF3, and TRAF6 (28, 29) . To test the effect of M on RLR and TLR signaling as well as TRAF expression, an increased dose of pCMV-Myc-M constructs was first transiently transfected into HEK293ET cells. The results in Fig. 3A demonstrate that the increased delivery of pCMV-Myc-M into HEK293ET cells markedly enhanced TRAF6 but not TRAF2 and TRAF3 expression, indicating that TRAF6-mediated signaling transduction might contribute to the upregulation of IFN-␤ production. To further address how TRAF expression is associated with RLR and/or TLR signaling pathways, the M genetransfected HEK293ET cells were also assayed for the expression of upstream sensors and/or signaling molecules of TRAFs. Figure 3B demonstrates that no significant alteration was observed in the expression of RIG-I, MDA5, and MAVS after exogenously delivering M genes into HEK293ET cells, indicating that the RLR signaling pathway might not be targeted by M gene products. In contrast, three adaptor proteins (MyD88, TRAM/TICAM2, and TIRAP) associated with TLRs were all upregulated (Fig. 3C) , while the adaptor protein TRIF failed to be upregulated in responding to M gene overexpression (Fig. 3D) . Overall, the results indicate that TLR signaling pathways are mainly targeted by the SARS-CoV M gene product for the induction of IFN-␤ expression.

The SARS-CoV M gene product promotes IFN-␤ production through Toll-like-receptor-related signaling pathways in immortalized murine bone marrow-derived macrophage cells. To further confirm the above results, the pCMV-Myc-M construct was also transiently transfected into J2-M cells, an immortalized murine bone marrow-derived macrophage cell line established with J2 virus (30, 31) . The delivery of the M gene product is effec-tive in stimulating the activation of both IFN-␤ and NF-B in murine J2-M cells (Fig. 4A , B, and C). Figure 4D demonstrates that increased delivery of M gene product into J2-M cells indeed promotes IFN-␤ induction through the phosphorylation of IRF3, NF-B p65, and TBK1. In accord with the results in HEK293ET cells, the increased delivery of pCMV-Myc-M into J2-M cells markedly enhanced TRAF6 but not TRAF2 and TRAF3 expression (Fig. 4E) . Moreover, the M gene product did not significantly increase the protein levels of RIG-I, MAVS, STING, and MDA5 ( Fig. 4F) , indicating that the RIG-I signaling pathway might not be activated in responding to exogenous delivery of the M gene product into J2-M cells. In contrast, the increased delivery of the M gene into J2-M cells markedly enhanced MyD88 and TRAM/ TICAM2 but not TRIF expression (Fig. 4G ), indicating that TLRrelated signaling pathways might be mainly associated with M-mediated IFN-␤ induction. Overall, our data in both HEK293ET and J2-M cells strongly indicate that M-mediated induction of IFN-␤ expression is likely associated with the activation of TLR-related signaling pathways.

The SARS-CoV M gene product functions at the protein level to induce IFN-␤ production. The next question that we tried to ask was at which level (mRNA or protein) the M-mediated IFN-␤ induction occurred. To address this issue directly, we created an M-stop construct by replacing the start codon AUG with three in-frame tandem stop codons at the 5= end of the M gene (Fig. 5A ). This expression construct can generate only mRNA and no protein due to the translation failure of the mRNA substrates. Western blot analysis shows that the M protein synthesis was completely blocked in the M-stop construct but not the wild-type M construct (Fig. 5B ). Real-time quantitative reverse transcription PCR (qRT-PCR) analysis shows that the M-stop construct did not induce IFN-␤ production in HEK293T cells, indicating that M-mediated IFN-␤ production is dependent on M protein rather than M mRNA (Fig. 5C) .

To directly compare M-and M-stop-induced IFN-␤ production levels, the increased doses of either M or M-stop constructs were cotransfected with IFN-␤ luciferase reporter into HEK293T cells. Figure 5D clearly demonstrates that M-stop does not induce IFN-␤ production, indicating that protein translation is necessary for M-induced IFN-␤ production. To confirm this result further, the chemical inhibitor cycloheximide (CHX) was used to block the protein biosynthesis in transfected HEK293T cells. Figure 5E shows that addition of CHX significantly inhibited and completely blocked M protein synthesis at 5 g/ml and above 20 g/ ml, respectively, which are directly correlated with the marked reduction and complete inhibition in IFN-␤ expression, indicating that M protein may function as a PAMP to induce IFN-␤ production.

If M protein indeed functions as a PAMP, blocking M protein synthesis should prevent the M-mediated activation of the IFN-␤ signaling pathway. Figure 5F clearly shows that M-stop reverses the M-mediated upregulation of the adaptor proteins MyD88 and TICAM2/TRAM in the initiating phase of TLR signaling pathways. Moreover, M-stop prevents the activation of the downstream modulator and key effectors of the IFN-␤ signaling pathway, such as TBK1, IRF3, and NF-B p65, in a dose-dependent manner (Fig. 5F ). Thus, blocking M protein translation could prevent M-mediated IFN-␤ induction by inactivating the TLRrelated signaling pathway.

SARS-CoV M protein may function as a novel intracellular PAMP to induce IFN-␤ production. One critical question remaining to be answered is whether the driving force for M protein-mediated IFN-␤ induction is generated intracellularly or extracellularly. To answer this question directly, the TRAP␥ gene, an endoplasmic reticulum (ER)-associated gene, was cotransfected with M into HeLa cells. Brefeldin A (BFA) was employed in the assay system to block M protein transport from the rough endoplasmic reticulum to the cell surface. Indeed, the addition of BFA effectively increased the retention of M proteins in the ER compartment (Fig. 6A) . Figure 6B shows that addition of BFA also effectively inhibited the secretion of IFN-␤ into the cell culture medium (right panel) but did not inhibit M-mediated IFN-␤ induction at either the mRNA level or the protein level (left panel), indicating that the driving force for M-mediated IFN-␤ induction was indeed derived from intracellular stimulation by M proteins.

Reports indicate that SARS-CoV M protein alone can form virus-like particles (VLPs) that can be secreted extracellularly. Therefore, the secreted M protein might be sensed by its own or other cell PRRs on the cell surface to activate IFN-I responses. To rule out this possibility, OxPAPC (a TLR2 and TLR4 inhibitor) was used to block the function of the accessory proteins CD14, LBP, and MD2 that are required for TLR2 and TLR4 signaling (32) . Figure 6C shows that addition of OxPAPC could effectively inhibit lipopolysaccharide (LPS)-mediated IFN-␤ production (right panel) but did not reverse the M-mediated IFN-␤ induction (left panel), indicating that the extracellular stimulation by M pro- SARS-CoV M protein promotes IFN-␤ induction independently of TRAF3. It has been well known that TRAF3 plays a critical role in TLR-mediated IFN-␤ induction, especially through TLR3 and TLR4 pathways (27, 29, 33) . Since M protein could activate the TLR pathway from inside the cells, it remained unclear whether or not this activation is in a TRAF3-dependent or -independent manner. To address this issue directly, a specific siRNA against TRAF3 was successfully constructed (Fig. 7A) . A dual-luciferase assay was conducted to assay the effect of siTRAF3 on M-mediated IFN-␤ induction in HEK293T cells. Figure 7B clearly shows that the increased delivery of siTRAF3 did not reverse the M-mediated IFN-␤ induction, indicating that TRAF3 might not be essential for this activation. To further confirm the above result, Western blot analysis was performed to detect IFN-␤ expression in responding to the increased delivery of siTRAF3 into HEK293T cells. The increased delivery of siTRAF3 into HEK293T Co-IP was conducted as shown in the left panel. About 10% input from each lysate preparation was subjected to Western blot analysis using anti-HA, anti-Flag, and anti-Myc antibodies as probes. The rest of the lysate was first immunoprecipitated with anti-Flag antibody conjugated with an affinity gel. Then, the reaction products were probed with anti-HA and anti-Myc antibodies. A reverse co-IP experiment was also conducted as shown in the right panel. The lysates were first immunoprecipitated with anti-Myc antibody. Then, the reaction products were individually probed with anti-Myc, anti-Flag, and anti-HA antibodies and subsequently subjected to Western blotting. The result is representative of at least 2 identical experiments. IB, immunoblotting; ns, nonspecific. (H) The M gene product does not suppress poly(I:C)-mediated IFN-␤ induction. About 2 g/ml of poly(I:C) was cotransfected with 2 g of either M or M(V68A) plasmid along with pGL3-IFN-␤luciferase reporter (100 ng) plus pRL-TK (10 ng) into HEK293T cells. After 48 h of transfection, dual-luciferase assays were performed to detect IFN-␤ expression (left panel). Each value represents the mean Ϯ standard deviation from three independent tests. The statistical difference was considered to be significant at a P value of Յ0.05. The IFN-␤ expression was also subjected to Western blotting (right panel). The relative band intensity was quantitated with the Image J program in comparison with the ␤-actin control. cells failed to inactivate the activation of IRF3 and NF-B p65 and did not affect the M-mediated IFN-␤ induction (Fig. 7C) , while a result was obtained consistent with the increased delivery of si-TRAF3 into J2-M cells (Fig. 7D) . In addition, a stable HEK293 cell line with TRAF3 knocked down by siTRAF3 was also established (Fig. 7E) . A dose-dependent increase in IFN-␤ production was observed with the increased delivery of M gene into 293 si-TRAF3 stable cells (Fig. 7F) . Thus, the above data strongly indicate that TRAF3 is not required for M-mediated IFN-␤ induction.

SARS-CoV M protein has been shown to destabilize the functional TRAF3-TBK1 complex formation (34) . However, it remains unknown whether or not M protein is able to associate with the key components of this complex. To clarify this issue directly, a coimmunoprecipitation (co-IP) experiment was conducted. Plasmid TBK1-Flag, TRAF3-HA, and Myc-M DNAs were transiently cotransfected into HEK293T cells. The cell lysates were first immunoprecipitated with anti-Flag antibody conjugated with an affinity gel and then subsequently probed with antihemagglutinin (anti-HA) and anti-Myc antibodies. The left panel of Fig. 7G indicates that M protein was able to disrupt the physical interaction between TBK1 and TRAF3, but M protein itself could not form a complex with TBK1. In the reverse immunoprecipitation (IP) experiment as shown in the right panel of Fig. 7G , M protein was unable to interact with both TBK1 and TRAF3 directly, indicating that M protein may modulate the TBK1-TRAF3 complex formation indirectly. Interestingly, in contrast to the inhibitory effect of M(V68A) on poly(I:C)-mediated IFN-␤ induction, the M gene product did not affect the IFN-␤ induction stimulated by poly(I:C) (Fig. 7H) , indicating that M has no effect on poly(I:C)induced IFN-␤ production while retaining the ability to destabilize the complex formation.

One critical question remaining to be answered is whether or not M-mediated IFN-␤ induction could occur in real virus infection. It has been shown that codelivering the M, N, and S genes of SARS-CoV into HEK293 cells readily produced SARS-CoV pseudovirus with the corona-like halo (35) . Earlier results demonstrate that valine-toalanine mutation at residue 68 [abbreviated as M(V68A)] inhibited IFN-␤ induction (34) . We employed a SARS-CoV pseudovirus system to mimic the real SARS-CoV infection. Cell lysate supernatants isolated from either M, N, and S cotransfection or M(V68A), N, and S cotransfection were first incubated with 293-ACE2 stable cells (Fig. 8A) for 4 h. Then, the cell culture medium was replaced with fresh medium and further incubated at 37°C for another 24 h before harvesting. The SARS-CoV pseudovirus VLP(S-M-N) markedly upregulated IFN-␤ expression at both RNA and protein levels (Fig. 8B) , whereas the point mutation at the valine 68 residue of M protein completely diminished SARS-CoV pseudovirus-mediated IFN-␤ induction, strongly indicating that the M protein residing on SARS-CoV virion could specifically induce IFN-␤ production upon infecting the targeted cells. Taken together, our data indicate for the first time that SARS-CoV M protein may function as a novel cytosolic PAMP to activate IFN-␤ induction through an intracellular TLR-related signaling pathway in a TRAF3-independent manner.

Pathogen-associated molecular patterns (PAMPs) are pathogenborne components that can be sensed by either transmembrane, endolysosomal, or cytosolic sensors known as pattern recognition receptors (PRRs) (36) . In this study, we demonstrate that the membrane protein of SARS-CoV significantly upregulates IFN-␤ production by activating both NF-B and TBK1-IRF3 signaling cascades. Our data show that M-mediated IFN-␤ production is induced by M protein rather than M mRNA, indicating that pathogen-derived protein might be able to serve as a cytosolic PAMP. In addition, a mechanism study indicates that M proteinmediated IFN-␤ induction may be mainly due to the selective activation of TLR-related signaling pathways rather than the RLR signaling pathway in a TRAF3-independent manner. Overall, the current study indicates for the first time that pathogen-derived protein may function as a novel cytosolic PAMP to initiate a TLRrelated TRAF3-independent signaling pathway that subsequently promotes type I interferon (IFN-I) production.

The membrane-associated PRRs, such as TLR2 and TLR4, can sense not only bacterial components but also viral coat proteins (37) . Kurt-Jones and colleagues first demonstrated that the innate immune response to the fusion (F) protein of respiratory syncytial virus (RSV) is mediated by TLR4 plus its cofactor CD14 on the plasma membrane, indicating that nonbacterial components can serve as extracellular PAMPs (24) . Although some viral structural proteins, such as the nucleocapsid (N) from measles virus (38) and viral ribonucleoprotein from vesicular stomatitis virus (39) , can activate IRF3 and TBK1/IKK␦, respectively, it remains to be determined how these viral products can function as PAMPs and where the driving forces for their activation come from. It has been shown that the SARS-CoV M protein alone not only can form virus-like particles (VLPs) in a viral RNA-independent manner (40) but also can induce cell apoptosis by inhibiting some key survival signal pathways, such as the Akt signaling pathway (41) . Thus, the M proteins derived from these infected and apoptotic cells might be eventually secreted and released into the extracellular compartments where the TLRs (such as TLR2 and TLR4) on the cell surface can be potentially recognized and subsequently activated by these extracellular PAMPs for the induction of the IFN-I response. To clarify this possibility, we employed several approaches to test whether the driving force for M proteinmediated IFN-␤ production is generated from inside or outside the cells. We used two types of inhibitors, BFA and OxPAPC, to block either the intracellular transport of M protein or the extracellular binding of M protein on TLR2 and TLR4, respectively. Our results reveal that M-mediated IFN-␤ induction is independent of M protein secretion as well as the extracellular stimulation of TLR2/TLR4 signaling, indicating that the driving force for the M-mediated IFN-␤ induction is likely generated from inside the cells.

Our results may contradict some previous reports related to M-mediated IFN-I response. Siu et al. demonstrated that the M protein of SARS-CoV negatively regulated the dsRNA-induced and signaling molecule-induced IFN-I production. They also showed that M protein alone was unable to activate the IFN-I promoter activity (34) . One study even showed that M protein negatively regulates the NF-B signaling pathway (42) . A possible explanation for these discrepancies might be related to the M gene itself. Early study has shown that M protein possesses a higher substitution rate than other structural proteins in SARS-CoV, and the outcome of these substitutions alters the biochemical and immunological properties of M proteins (43, 44) . One amino acid alteration from valine to alanine at residue 68 is indeed found in the M protein from the GZ50 isolate studied by Siu et al. compared with the isolate used in the current study. Our functional analysis provides strong evidence to demonstrate that the valine-toalanine change at residue 68 of M protein is sufficient to abolish M-mediated IFN-␤ induction at both the transient-transfection level and the viral infection level (Fig. 1D and E and 8B and C) . Therefore, this amino acid substitution in M proteins indeed affects the interaction between M protein and the PRR for the subsequent IFN-I induction.

It still remains elusive which cytosolic PRR is responsible for M-mediated IFN-␤ induction. The M protein might be a multifaceted molecule that physically interacts with diverse intracellular sensors and signaling factors (34, 42) . Our data indicate that the TLR-related signaling pathway rather than the RIG-I signaling cascade is responsible for M-mediated IFN-␤ induction. Interestingly, we observed consistent TRAF3 reductions in both HEK293T and J2-M cells as tested by the transiently intracellular overexpression of the M gene. TRAF3 is one of the key signaling molecules specifically responsible for IFN-I induction (28) . Data from the work of Siu and colleagues have shown that M(V68A) excludes the TRAF3 inclusion in the TRAF3.TANK.TBK1/IKK complex (39) . Although M(V68A)-mediated TRAF3 exclusion inhibits IFN-␤ induction, our study showed that M-mediated TRAF3 exclusion is independent of ligand stimulation and the disassociated TRAF3 and TBK1 could not complex with M protein, indicating that M protein might modulate the functions of TBK1 complex indirectly. Interestingly, we demonstrated that M-mediated IFN-␤ induction was associated with the upregulation of the adaptor proteins MyD88, TIRAP, and TICAM2 but not TRIF, which are all involved in the initiating phase of TLR signaling pathways. TRIF is required for both TLR3-and TLR4mediated IFN-␤ induction. In the activation of TLR3, TRIF is directly recruited to the TIR domain of dimerized TLR3, while in the activation of TLR4, TRIF is recruited to dimerized TLR4 indirectly through another TIR-containing adaptor protein, TICAM2 (also called TRAM). If TRIF is not required for M-mediated IFN-␤ induction, M-mediated IFN-␤ induction might be activated via a noncanonical TLR4-related signaling cascade independently of TRIF and TRAF3.

In summary, the current study demonstrates for the first time that the M protein of SARS-CoV is able to function as a cytosolic PAMP to promote IFN-␤ production by activating a TLR-related TRAF3-independent pathway. The driving force for M-mediated IFN-␤ induction is likely generated from inside the cells rather than the extracellular binding of M proteins with the defined cell surface PRRs, such as TLR4. Plasmid construction. Plasmids pCMV-Myc-M and pBS-U6-siM1 were constructed previously (26) . Plasmids pCMV-Myc-S and pCMV-Myc-E were constructed by inserting S and E into the EcoRI and KpnI sites of pCMV-Myc. The mutant of pCMV-Myc-M(V68A) was generated by using the site-directed mutagenesis kit (TaKaRa, Dalian, China). One copy of the IFN-␤ promoter sequence (5=-CTAAAATGTAAATGACATA GGAAAACTGAAAGGGAGAAGTGAAAGTGGGAAATTCCTCTGAAT AGAGAGAGGACCATCTCATATAAATAGGCCATACCCATGGAGAA AGGACATTCTAACTGCAACCTTTCGA-3=) was PCR amplified and subcloned into the KpnI and XhoI sites of the luciferase reporter pGL3basic (Promega, Madison, WI, USA) to generate the pGL3-IFN-␤-luc construct. All primers were synthesized by Sangon (Shanghai, China). For the construction of pBS/U6 siTBK1, the sense strand 5= TCGAGTTGCG AAGCCGGAAGTGTCCTAAGCTTAGGACACTTCCGGCTTCGCAAT TTTTG 3= and antisense strand 5= ACGCTTCGGCCTTCACAGGATTC GAATCCTGTGAAGGCCGAAGCGTTAAAAACTTAA 3= were annealed and then subcloned into the EcoRI and XhoI sites of pBS/U6. For the construction of pBS/U6 siIRF3, the sense strand 5= TCGAGCATCGGCT TTTGGGTCTGTTAAAGCTTTAACAGACCCAAAAGCCGATGTTTT TG 3= and antisense strand 5= CGTAGCCGAAAACCCAGACAATTTCG AAATTGTCTGGGTTTTCGGCTACAAAAACTTAA 3= were annealed and then subcloned into the EcoRI and XhoI sites of pBS/U6.

For construction of pSilencer-siTRAF3, the single-strand oligonucleotides 5= GATCCGCGAGAACTCCTCTTTCCCTCGAGGGAAAGAGG AGTTCTCGCAGA 3= and 5= AGCTTCTGCGAGAACTCCTCTTTCCC TCGAGGGAAAGAGGAGTTCTCGCG 3= were annealed to double strands before being subcloned into the HindIII and BamHI sites of pSilencer 4.1-CMV neo to form the pSilencer-siTRAF3 construct.

Transient transfection. Cells (HEK293ET, HEK293T, and J2-M) were transiently transfected with the plasmid DNAs (pCMV-Myc-M, pCMV-Myc, pGL3-IFN-␤-luc, or pNF-B-luc) using VigoFect (Vigorous Biotechnology, Beijing, China) according to the manufacturer's instructions. For example, about 5 g plasmid DNAs was first added to 100 l of 0.9% NaCl. Then, 2.5 l VigoFect was resuspended into another 100 l of 0.9% NaCl solution. Then, the DNA-NaCl mixture was added to the VigoFect-NaCl mixture drop by drop with gentle vortexing. After a 15min incubation at room temperature, the reaction product was evenly distributed onto the cell culture surface of either 6-well plates or 35-mm 2 dishes and then continuously incubated for 48 h before harvesting.

Construction of the 293 siTRAF3 stable cells. Plasmid pSilencer-siTRAF3 DNAs (5 g) were transfected into HEK293 cells. After a 24-h transfection, the cell culture medium was replaced with fresh medium containing 1,200 g/ml G418. After 2 weeks of culture, cell colonies were picked up and expanded in a 24-well tissue culture plate. Finally, Western blot analysis was performed to detect TRAF3 expression.

Reverse transcription-PCR (RT-PCR) and qRT-PCR. Total RNAs were extracted from the cultured cells with TRIzol (Invitrogen, Carlsbad, CA, USA). The purified total RNAs were treated with DNase I (Qiagen, Düsseldorf, Germany). All primers used in the RT-PCRs are listed in Table 1 . The RT-PCR was carried out with the PrimeScript one-step RT-PCR kit (TaKaRa Biotechnology, Dalian, China) according to the manufacturer's instructions. The RT-PCR was carried out in a DNA Thermal Cycler (Applied Biosystems, Carlsbad, CA) under the following conditions: 50°C for 35 min for reverse transcription and 94°C for 5 min for denaturation. The PCR conditions were 94°C for 30 s, 50°C for 30 s, and 72°C for 50 s, repeated for 20 to 30 cycles; the reaction was extended at 72°C for 10 min before the reaction product was stored at 4°C. One-step real-time quantitative RT-PCR (qRT-PCR) (TaKaRa Biotechnology, Dalian, China) was also performed to monitor the targeted gene expression. The primers used in qRT-PCR were also listed in Table 1 . Real-time qRT-PCR was carried out with the CFX real-time PCR detection system (Bio-Rad Laboratories, Hercules, CA, USA) under the following conditions: 42°C for 5 min and 95°C for 10 s, and then 95°C for 5 s and 60°C for 10 s, repeated for 40 cycles. The dissociation of the reaction products was conducted from 55°C to 95°C as the temperature rose at 0.2°C per 10 s. 

Viral evasion and subversion of pattern-recognition receptor signalling

Induction and function of IFNbeta during viral and bacterial infection

Innate immunity to virus infection

Role of adaptor TRIF in the MyD88-independent Toll-like receptor signaling pathway

Innate antiviral responses by means of TLR7-mediated recognition of singlestranded RNA

TLR9 signals after translocating from the ER to CpG DNA in the lysosome

The interferon inducing pathways and the hepatitis C virus

5=-Triphosphate RNA is the ligand for RIG-I

Differential roles of MDA5 and RIG-I helicases in the recognition of RNA viruses

MAVS forms functional prion-like aggregates to activate and propagate antiviral innate immune response

Regulation of antiviral responses by a direct and specific interaction between TRAF3 and Cardif

The adaptor protein MITA links virus-sensing receptors to IRF3 transcription factor activation

Vaccinia virus protein A46R targets multiple toll-like-interleukin-1 receptor adaptors and contributes to virulence

Poxviral protein A46 antagonizes Toll-like receptor 4 signaling by targeting BB loop motifs in toll-IL-1 receptor adaptor proteins to disrupt receptor:adaptor interactions

Viral inhibitory peptide of TLR4, a peptide derived from vaccinia protein A46, specifically inhibits TLR4 by directly targeting MyD88 adaptor-like and TRIF-related adaptor molecule

The poxvirus protein A52R targets Toll-like receptor signaling complexes to suppress host defense

A46R and A52R from vaccinia virus are antagonists of host IL-1 and Toll-like receptor signaling

Poxvirus protein N1L targets the I-kappaB kinase complex, inhibits signaling to NF-kappaB by the tumor necrosis factor superfamily of receptors, and inhibits NF-kappaB and IRF3 signaling by Toll-like receptors

Inhibition of retinoic acid-inducible gene I-mediated induction of beta interferon by the NS1 protein of influenza A virus

Influenza C virus NS1 protein counteracts RIG-I-mediated IFN signalling

The V proteins of paramyxoviruses bind the IFNinducible RNA helicase, mda-5, and inhibit its activation of the IFN-beta promoter

Inhibition of interferon regulatory factor 3 activation by paramyxovirus V protein

Murine retroviruses activate B cells via interaction with Toll-like receptor 4

Pattern recognition receptors TLR4 and CD14 mediate response to respiratory syncytial virus

Reading the viral signature by Toll-like receptors and other pattern recognition receptors

Small interfering RNA effectively inhibits the expression of SARS coronavirus membrane gene at two novel targeting sites

The family of five: TIR-domain-containing adaptors in Toll-like receptor signalling

TRAF molecules in cell signaling and in human diseases

Expanding TRAF function: TRAF3 as a tri-faced immune regulator

Tumoricidal alveolar macrophage and tumor infiltrating macrophage cell lines

Interaction between Raf and Myc oncogenes in transformation in vivo and in vitro

Oxidized phospholipid inhibition of Toll-like receptor (TLR) signaling is restricted to TLR2 and TLR4: roles for CD14, LPS-binding protein, and MD2 as targets for specificity of inhibition

Assembly and localization of Toll-like receptor signalling complexes

Severe acute respiratory syndrome coronavirus M protein inhibits type I interferon production by impeding the formation of TRAF3.TANK.TBK1/IKKepsilon complex

Generation of synthetic severe acute respiratory syndrome coronavirus pseudoparticles: implications for assembly and vaccine production

Pattern recognition receptors and inflammation

Pathogen recognition by the innate immune system

Recognition of the measles virus nucleocapsid as a mechanism of IRF-3 activation

Activation of TBK1 and IKKvarepsilon kinases by vesicular stomatitis virus infection and the role of viral ribonucleoprotein in the development of interferon antiviral immunity

Self-assembly of severe acute respiratory syndrome coronavirus membrane protein

The SARS-coronavirus membrane protein induces apoptosis through modulating the Akt survival pathway

The membrane protein of SARS-CoV suppresses NF-kappaB activation

The M protein of SARS-CoV: basic structural and immunological properties

Persistent replication of severe acute respiratory syndrome coronavirus in human tubular kidney cells selects for adaptive mutations in the membrane protein

The orphan nuclear receptor TR3/Nur77 regulates ER stress and induces apoptosis via interaction with TRAP␥

We thank Genhong Cheng and Hang-Zi Chen for providing the J2-M cells and TRAP␥, respectively. 

Western blot analysis. The transfected cells were lysed with a lysis buffer containing 1% NP-40, 50 mM Tris-HCl (pH 7.5), 120 mM NaCl, 200 M NaVO 4 , 1 g/ml leupeptin, 1 g/ml aprotinin, and 1 M phenylmethylsulfonyl fluoride (PMSF). About 15 g of cell lysate for each sample was resolved by 12% SDS-PAGE. After separation, the reaction products were transferred onto a Hybond nitrocellulose membrane (Pharmacia, St. Louis, MO, USA). The transferred membrane was first probed with a primary antibody. Then, a secondary antibody labeled with horseradish peroxidase was added to the reaction product and was finally visualized with an enhanced chemiluminescence (ECL) kit (Santa Cruz Biotechnology, Santa Cruz, CA, USA).Dual-luciferase assay. The dual-luciferase kit was purchased from Promega (Promega Corporation, USA). Cells were transfected with pGL3-IFN-␤-luc or pNF-B-luc plus pRL-TK at the ratio of 100:1 or 10:1 as suggested by the manual. The detection was done by following the assay protocol in the kit. The firefly and Renilla luciferase activities were read with a Modulus microplate multimode reader (Turner Biosystems, Sunnyvale, CA, USA).Coimmunoprecipitation (co-IP) assay. The transfected cells were lysed with a lysis buffer containing 50 mM Tris-HCl (pH 7.4), 150 mM NaCl, 1 mM EDTA, 0.2% Triton, 1 g/ml leupeptin, 1 g/ml aprotinin, and 1 mM PMSF. About 10% of the lysate was subjected to input analysis. Flag affinity gel (Sigma, USA) or anti-Myc and protein G plus agarose (Santa Cruz Biotechnology, Santa Cruz, CA, USA) were added to the rest of the lysate (90%) and rotated overnight at 4°C. The reaction products were washed 5 times with a wash buffer and then centrifuged at 5,000 rpm for 1 min. The IP products were resuspended in the loading buffer and subjected to Western blot analysis.ELISA. Cell culture supernatants were harvested 48 h after transfection. The human IFN-␤ ELISA kit was purchased from Bluegene (Shanghai, China). The reaction was carried out by following the manufacturer's instructions. The reaction products were detected with Synogen 4 (BioTek, Seattle, WA, USA) under 450 nm.Generation of SARS-CoV pseudovirus and assay of virus-induced IFN-␤ production. The generation of SARS-CoV pseudovirus followed the procedure described by Huang et al. (35) . Briefly, HEK293 cells were cotransfected with either the S, M, and N genes or the S, M mutant M(V68A), and N of SARS-CoV. After 48 h of transfection, the culture medium and the transfected cells were collected and subjected to freezing and thawing cycles at least four times. The reaction products were centrifuged at 12,000 rpm for 10 min. The supernatant containing VLPs was then applied to HEK293-ACE2 stable cells and incubated for 4 h. Then, the cell culture medium was replaced with fresh medium. After 24 h, the infected cells were subjected to total RNA isolation and preparation of whole-cell lysates as described elsewhere. SARS-CoV-mediated IFN-␤ expression was monitored by standard Western blotting, RT-PCR, and qRT-PCR.Immunostaining assay. HeLa cells were transiently transfected with Myc-M and Flag-TRAP␥ (45) . After 24 h of transfection, the cells were fixed with 4% formaldehyde for 20 min on ice and then incubated in 1ϫ phosphate-buffered saline (PBS) containing 10% bovine serum albumin (BSA) for 1 h. The cells were first incubated with primary antibody rabbit anti-Flag or mouse anti-Myc for 1 h. Then, the secondary antibody fluorescein isothiocyanate (FITC)-labeled goat anti-mouse or tetramethyl rhodamine isocyanate (TRITC)-labeled goat anti-rabbit antibody was added, and the mixture was incubated for another 1 h. Finally, the cells were incubated with 4=,6-diamidino-2-phenylindole (DAPI) and sealed with glycerol. The results were observed under an Olympus FV1000 confocal microscope.Statistical analysis. All values were calculated as means Ϯ standard deviations (SDs) from three independent experiments. The statistical difference between the assayed group and the standard group was subjected to Student's t test. The calculated difference was considered significant at the P value of Ͻ0.05.