USP49 negatively regulates cellular antiviral responses via deconjugating K63-linked ubiquitination of MITA

Mediator of IRF3 activation (MITA, also known as STING and ERIS) is an essential adaptor protein for cytoplasmic DNA-triggered signaling and involved in innate immune responses, autoimmunity and tumorigenesis. The activity of MITA is critically regulated by ubiquitination and deubiquitination. Here, we report that USP49 interacts with and deubiquitinates MITA after HSV-1 infection, thereby turning down cellular antiviral responses. Knockdown or knockout of USP49 potentiated HSV-1-, cytoplasmic DNA- or cGAMP-induced production of type I interferons (IFNs) and proinflammatory cytokines and impairs HSV-1 replication. Consistently, Usp49-/- mice exhibit resistance to lethal HSV-1 infection and attenuated HSV-1 replication compared to Usp49+/+ mice. Mechanistically, USP49 removes K63-linked ubiquitin chains from MITA after HSV-1 infection which inhibits the aggregation of MITA and the subsequent recruitment of TBK1 to the signaling complex. These findings suggest a critical role of USP49 in terminating innate antiviral responses and provide insights into the complex regulatory mechanisms of MITA activation.

Introduction The innate immune system is the first line of defense against invading pathogens and innate immune response to microbial species is initiated by the recognition of pathogen-associated molecular patterns (PAMPs) by germline-encoded pattern-recognition receptors (PRRs) [1]. Viral nucleic acid including RNA, DNA and RNA-DNA hybrid constitutes classical PMAPs that are detected by Toll-like receptors (TLRs), retinoic acid-inducible gene I protein (RIG-I)like receptors (RLRs) and a variety of cytoplasmic DNA sensors [2][3][4][5]. Upon binding to the PAMPs, these PRRs recruit adaptor proteins or catalyze second messenger molecules for adaptor proteins activation, trigger a series of signaling cascades and eventually induce expression of an array of downstream genes such as type I interferons (IFNs) and proinflammatory cytokines to elicit antiviral immune responses.
Mediator of IRF3 activation (MITA, also known as STING and ERIS) is an adaptor protein essential for cytoplasmic DNA-triggered signaling and host defense against HSV-1 and Listeria monocytogenes [6][7][8][9][10]. Structural and biochemical studies demonstrate that MITA is activated by binding to cyclic dinucleotides such as cyclic di-GMP which is generated by invading bacteria and cyclic GMP-AMP (cGAMP) which is generated by cGAMP synthase (cGAS) upon binding to cytoplasmic DNA either from invading DNA viruses and retroviruses or from selfdamaged genomic or mitochondrial DNA [11][12][13][14][15][16][17][18]. In addition, several studies have shown that an array of cytoplasmic DNA sensors such as DAI, IFI16 and DDX41 directly recruit and activate MITA in a ligand and/or cell-type specific manner and the mechanisms are less clear [19][20][21]. Activated MITA undergoes oligomerization and traffics to the ER-Golgi intermediate compartment (ERGIC) to activate the transcription factors IRF3 and NF-κB, in which iRhom2 plays an essential role [22,23]. Activated IRF3 and NF-κB enter nucleus to initiate transcription of a large number of downstream genes.
In addition to mediating immune responses against viruses or bacteria, MITA is also implicated in autoimmunity and tumor immunity. Various gain-of-function mutants and noncoding SNPs of MITA are found in patients with MITA-associated vasculopathy with onset in infancy (SAVI) and systemic inflammatory or autoimmune diseases characterized by elevated expression of type I IFNs, IFN-stimulated genes (ISGs) and proinflammatory cytokines [24][25][26][27]. Ablation of MITA in Trex1 -/mice or Rnaseh2 loss-of-function mutation knock-in mice significantly reduces systemic inflammation and lethality [28][29][30]. In contrast, inflammatory responses are significantly elevated by deficiency of MITA in the model of MRL. Fas lpr lupus and 2,6,10,14-tetramethylpentadecane (TMPD)-mediated peritonitis as a result of diminished MITA-mediated expression of suppressive molecules [31], indicating that proper activity and availability of MITA are essential for the balance of protective immunity and excessive autoimmunity. In addition, it has been observed that MITA is activated by tumor DNAs in APCs to promote type I IFN induction, tumor antigen cross presentation and CD8 + T cell activation and therapeutic activation of MITA results in plausible tumor regression [32]. However, agonist-mediated abnormal activation of MITA may also contribute to induction of ISGs in tumor cells previously linked with chemoresistance [33][34][35]. Thus, the activity of MITA should be strictly regulated under physiological conditions and normalization of MITA activity may contribute to improved outcome of anti-tumor immunity or autoimmunity.
Deubiquitinating enzymes counteract with ubiquitination by cleaving poly-or mono-ubiquitin from target proteins and play essential roles in various physiological processes [50]. In an unbiased screening of MITA-interacting DUBs, we have found that USP49 interacts with MITA [45]. USP49 has been reported to deubiquitinate H2B, FKBP51, p53 and DUSP1 and function in pre-mRNA splicing, tumorigenesis and ischemia-reperfusion-induced cell viability [51][52][53][54]. In this study, we found that USP49 deconjugates K63-linked polyubiquitin chains from MITA and thereby inhibits the oligomerization and subsequent recruitment of TBK1 after HSV-1 infection. Consistently, USP49 deficiency potentiates HSV-1-or cytoplasmic DNA-triggered signaling and impairs replication of HSV-1 in vitro and in vivo. These findings uncover a complex regulatory mechanism of MITA by USP49 to prevent excessive immune responses and uncontrolled inflammation.

USP49 interacts with MITA and inhibits MITA-mediated signaling
MITA is a critical adaptor protein involved in innate immune signaling, autoimmunity and tumor immunity and the ubiquitination of MITA critically regulates its activation and stabilization [55]. We thus speculated that DUBs-mediated deubiquitination of MITA was as equally important. To test our hypothesis, we conducted an unbiased screening of MITA-interacting DUBs by cotransfection and immunoprecipitation assays and found that USP49 interacted with MITA [45]. Endogenous immunoprecipitation analysis further confirmed that there was an association between USP49 and MITA in human monocytic THP-1 and U937 cells and human foreskin fibroblast (HFF) after HSV-1 infection or transfection of interferon-stimulating DNA (ISD) or cGAMP (Figs 1A and S1A). Domain mapping analysis suggested that the UCH domain of USP49 and the second transmembrane domain (aa41-110) of MITA are responsible for their association (Fig 1B). Results from reporter assays suggested that overexpression of USP49 inhibited MITA-or cGAS and MITA-but not TBK1-or IRF3-mediated activation of ISRE, whereas knockdown of USP49 potentiated MITA-but not IRF3-mediated activation of ISRE (S1B and S1C Fig), indicating that USP49 might function at the level of MITA to regulate innate antiviral signaling.
To further confirm the regulatory role of USP49 in cellular antiviral signaling, we generated USP49 KO THP-1 cells by CRISPR/Cas9-mediated gene editing. As shown in Fig 1C and 1D, knockout of USP49 in THP-1 cells significantly potentiated the expression of IFNB, IFNA4, CCL5 and IL6 and the phosphorylation of IRF3, IκBα and TBK1 after HSV-1 infection. In contrast, SeV-induced IFNB, IFNA4, CCL5 and IL6 and phosphorylation of IRF3, IκBα and TBK1 were not affected by knockout of USP49 in THP-1 cells (Fig 1C and 1D). These data together suggest that USP49 interacts with MITA and inhibits MITA-mediated signaling.

Knockout of USP49 potentiates HSV-1-triggered signaling
To further investigate the role of USP49 in antiviral signaling in vivo, we generated USP49-deficient mice by CRISPR/Cas9-mediated genome editing (S2A Fig). Results from sequencing showed that there was a deletion of 182 bp of the third exon of Usp49 gene, which caused a reading frame shift and led to an early translational termination of USP49 (S2B Fig). The Usp49 -/mice bred normally with the Mendelian inheritance ratio and did not show any developmental defect until 16-week old compared to the wild-type littermates (S2B Fig). The numbers and percentages of various immune cells in thymus, spleen or peripheral lymph nodes were comparable between the Usp49 +/+ and Usp49 -/mice (S2C- S2E Fig). The differentiation of Usp49 -/bone marrow cells into BMDCs and BMDMs was similar to the wild-type counterparts in the presence of GM-CSF and M-CSF, respectively (S2F Fig), indicating that USP49 is dispensable for development and homoeostasis of immune cells.

Usp49 -/mice exhibit increased resistance to lethal HSV-1 infection
To investigate the function of USP49 in host defense against virus infection in vivo, we monitored survival of Usp49 +/+ and Usp49 -/mice after intravenous (i.v.) injection of HSV-1. As shown in Fig 3A, Usp49 -/mice exhibited a later onset of death and a higher survival rate compared to the wild-type littermates. All the wild-type mice exhibited severe lethargy and died within 10 days. In contrast, 25% (4 out of 16) Usp49 -/mice survived after injection ( Fig 3A). Consistently, the concentrations of IFN-β and IL-6 in the sera of Usp49 -/mice were significantly increased compared to those in the Usp49 +/+ littermates 12 hours after HSV-1 infection ( Fig 3B). Furthermore, we observed that the expression of Ifnb, Il6, and Ifnan was potentiated and the expression of HSV-1-UL30 gene was inhibited in the lungs from Usp49 -/mice compared to Usp49 +/+ mice at 24 hours after HSV-1 infection (Fig 3C). In addition, we analyzed the expression of cytokines and viral titers in the brains 4 days after HSV-1 infection. The results showed that the expression of Ifnb and Ifna4 was increased, whereas the expression of HSV-1-UL30 gene and the replication of HSV-1 was suppressed in the brains from Usp49 -/mice compared to those from Usp49 +/+ mice ( Fig 3D). Collectively, these data demonstrate that loss of USP49 protects mice from HSV-1 infection by promoting the induction of type I IFNs and proinflammatory cytokines.

USP49 deconjugates K63-linked ubiquitination and aggregation of MITA
Since USP49 is a MITA-interacting DUB and its deubiquitinating enzymatic activity is required for regulating antiviral signaling, we hypothesized that USP49 may eliminated polyubiquitin chains from MITA. As expected, we found that USP49 but not the enzymatic inactive mutant USP49 (C262A) catalyzed deubiquitination of MITA in cells or in vitro (Figs 5A and  S5A). To examine the type of ubiquitin linkage on MITA that was regulated by USP49, we transfected MITA together with wild-type ubiquitin, or their mutants either retaining a single lysine residue (KO) or retaining all but one lysine residues (KR) in the presence or absence of USP49 followed by deubiquitination assays. The results showed that USP49 removed K63Oor K48R-linked but not K27O-, K48O-or K63R-linked polyubiquitin chains from MITA in cells (S5B and S5C Fig), indicating that USP49 primarily removes K63-but not K48-linked ubiquitination of MITA. Consistently, we observed that USP49 inhibited K63-linked ubiquitination of MITA in cells and in vitro by using an anti-Ub(K63-specific linkage) antibody (Figs 5A and S5A). We have previously demonstrated that USP18 recruits USP20 to remove K48-linked polyubiquitin chains from MITA [45]. Interestingly, we found that knockdown of USP18 did not affect USP49-mediated deubiquitinaton of MITA (S5D Fig), suggesting different regulatory mechanisms of K48-and K63-linked ubiquitination of MITA. We next adopted a strategy called Tandem Ubiquitin Binding Entity (TUBE) to examine the effects of USP49 on ubiquitination of MITA [44,[57][58][59][60]. As shown in Fig 5B, Pan-Ub-TUBE-and K63-Ub-TUBE-pulldown MITA were decreased by overexpression of USP49 but not USP49(C262A), suggesting that K63-linked polyubiquitin-modified MITA was deonjugated by USP49.
To further substantiate the conclusion, we next examined virus-triggered ubiquitination of MITA in the absence of USP49 and found that HSV-1 infection enhanced K63-linked ubiquitination of MITA more profoundly in Usp49 -/-MLFs than in the wild-type MLFs (Fig 5C and  5D). Furthermore, we found that reconstitution of USP49 but not USP49 (C262A) into Usp49 -/-MLFs inhibited HSV-1-induced K63-linked ubiquitination of MITA (Fig 5E). It has been reported that E3 ligases promote aggregation of signaling adaptor proteins through K63-linked ubiquitin [61]. Because USP49 removed K63-linked polyubiquitin from MITA, we examined whether USP49 affected the aggregation of MITA. As expected, USP49 deficiency in MLFs promoted the aggregation of MITA and reconstitution of USP49 but not USP49 (C262A) into Usp49 -/-MLFs impaired the aggregation of MITA after HSV-1 infection ( Fig  5F). The aggregation of MITA provides a platform to recruit kinase TBK1 which is essential for phosphorylation of MITA and subsequent recruitment of IRF3 [36]. Consistent with this notion, USP49 deficiency led to increased TBK1-MITA association and phosphorylation of MITA after HSV-1 infection (Fig 5G). Together these data demonstrate that USP49 deconjugates K63-linked polyubiquitin chains from MITA and inhibits MITA aggregation and the recruitment of TBK1 to the signaling platform.

Discussion
Ubiquitination and deubiquitination are reversible posttranslational modifications involved in various biological or pathological processes [62,63]. MITA is an adaptor protein critically mediating innate immune signaling in response to cytoplasmic DNA challenge. The ubiquitination of MITA is essential for its function and stability which should be properly controlled to elicit protective immunity and avoid excessive harmful immunity [55]. AMFR promotes K27-linked ubiquitination and activation of MITA [43], whereas RNF5, TRIM30α and TRIM29 induce K48-linked ubiquitination and proteasomal degradation of MITA [39][40][41][42], which is counteracted by USP13 and USP18, USP21 or EIF3S5, respectively [23,[44][45][46]. Multiple E3 ligases including TRIM56, TRIM32 and MUL1 have been reported to promote K63-linked ubiquitination and aggregation of MITA [47][48][49]. However, the counteracting deubiquitination process is unclear. In this study, we demonstrated that USP49 interacted with MITA and deconjugated K63-linked polyubiquitin chains from MITA after HSV-1 infection, thereby downregulating MITA activation and turning down immune responses (Fig 6). Consistent with this notion, we found that HSV-1-induced K63-linked ubiquitination and aggregation of MITA and subsequent recruitment of TBK1 to MITA was substantially enhanced in USP49 deficient cells. USP49 deficiency also resulted in resistance to lethal HSV-1 infection, enhanced phosphorylation of TBK1, IRF3 and IκBα, increased production of type I IFNs and proinflammatory cytokines and compromised HSV-1 replication after HSV-1 infection. These findings together suggest an essential role of USP49 in the regulation of innate antiviral signaling by modulating K63-linked ubiquitination of MITA.
MITA has been implicated in defense against RNA viruses in human cell lines such as HEK293 and HeLa cells in a manner dependent on TRIM56-and/or TRIM32-mediated K63-linked ubiquitination of MITA [47,48]. However, gene knockout studies show that neither TRIM56 nor TRIM32 deficiency in mice has any obvious effects on SeV-triggered signaling [64,65]. Whether TRIM56 and TRIM32 function redundantly to ubiquitinate MITA and regulate RNA virus-triggered signaling remains unclear. Furthermore, infection with RNA viruses in telomerase-immortalized human foreskin fibroblasts (hTERT-BJ1) or MEFs fails to induce ubiquitination or activation of MITA [49], indicating that K63-linked ubiquitination of MITA is not required for innate immune signaling in response to RNA virus infection. Consistent with this notion, we found that knockout of USP49 had no effect on SeV-induced phosphorylation of IRF3 or IκBα and expression of downstream genes in mouse cells. In this context, it has been shown that MITA restricts RNA virus replication through inhibition of viral mRNA translation or promotion of membrane fusion-stimulated IFN production but independent of cGAMP-mediated dimerization or oligomerization [66,67].
Although the direct genetic evidence of E3s that are responsible for K63-linked ubiquitination of MITA is still lacking, our data suggest that the K63-linked ubiquitination of MITA is essential for innate antiviral responses against HSV-1 infection. In this context, it has been reported that HSV-1 infection or cytoplasmic DNA challenge strongly induces K63-linked ubiquitination of MITA prior to its phosphorylation and translocation in both human and mouse cells [36]. Consistently, we found that knockout of USP49 potentiated HSV-1-induced K63-linked ubiquitination, aggregation and phosphorylation of MITA. Our data support and add a previously uncharacterized component in the step-wise model of MITA full activation: (i) cGAMP binding to MITA initiates dimerization of MITA; (ii) MUL1-or USP49-mediated balanced K63-linked ubiquitination of MITA modulates its aggregation and migration to ERGIC to from puncta; (iii) TBK1 is recruited to the MITA aggregate platform where TBK1 phosphorylates MITA; and (iv) such a phosphorylation leads to recruitment of IRF3 and subsequent phosphorylation by TBK1. However, it is so far unclear why USP49 interacts and counteracts K63-linked ubiquitination of MITA at early time after HSV-1 infection. A simplest explanation for this is that MITA is hyperactivated at the very early time in such a system in our study and USP49 functions as a corrector to erase the overubiquitination on MITA as a protective mechanism.
MITA-mediated signaling plays important roles in autoimmunity and tumor immunity [35,68]. Gain-of-function mutations of MITA are found in patients with various systemic inflammatory or autoimmune diseases. A recent study on exome sequencing indicates that USP49 is one of the frequently mutated genes associated with obesity which is recognized as a chronic inflammatory disease related to cardiovascular and respiratory diseases, type II diabetes, and cancers [69]. MITA activation by tumor DNA elicits anti-tumor immunity by type I  [32], whereas MITA-mediated inflammation promotes tumorigenesis in an inflammation-induced skin cancer model [70]. Interestingly, an SNP of a large linkage disequilibrium block containing USP49 gene is associated with breast cancer survival and USP49 functions as a tumor suppressor in pancreatic cancer [53,71]. The Usp49 -/mice did not exhibit any autoimmune symptoms at 8 months of age which might be due to a lack of inducer of MITA activation in our housing conditions. In this context, we observed that USP49-MITA association was induced by HSV-1 infection and the basal expression of type I IFNs or proinflammatory cytokines was comparable between the wildtype and Usp49 -/cells. Whether and how the USP49-MITA axis regulates inflammatory diseases such as colitis and tumorigenesis such as colon cancer is of great interest and requires future investigations.

Mice
The Usp49 -/mice on the C57BL/6 background were generated by Nanjing Biomedical Research Institute of Nanjing University through CRISPR/Cas9-mediated gene editing. In brief, Cas9 mRNA and guide RNA (5'-AGACCACATGACTCGGAAGAGGG-3' and 5'-AGC CACGGAAGGCGGGAATCAGG-3') were in vitro transcribed followed by injection into the fertilized eggs that were transplanted into pseudopregnant mice. The tail DNA of F0 mice was amplified with PCR and sequenced and the chimeras were crossed with wild-type C57BL/6 mice to obtain the Usp49 +/mice. The F1 Usp49 +/mice were further crossed with wild-type C57BL/6 mice for at least three generations. The genotyping of the Usp49 -/mice was confirmed by sequencing of the PCR fragments amplified from the genomic DNA isolated from tails using the following primers: Forward 5 0 -CAGAGTTGTCAGTAAGGAGT -3 0 and Reverse 5 0 -ACCCAAGTTCACCTACACGG-3 0 . The age-and sex-matched Usp49 +/+ and Usp49 -/littermates were randomized into groups for animal studies.

Ethics statement
All mice were housed in the specific pathogen-free animal facility at Wuhan University and all animal experiments were in accordance with protocols were adhered to the Chinese National Laboratory Animal-Guideline for Ethical Review of Animal Welfare and approved by the Institutional Animal Care and Use Committee of Wuhan University (NO. 16040B and 16060F). The mice were euthanatized with CO 2 followed by various studies.

Reagents and antibodies
Mouse control IgG (Santa Cruz Biotechnology, sc-2025) and rabbit control IgG (Millipore

Quantitative RT-PCR and ELISA
Total RNA was extracted from cells or tissues using TRIzol (Invitrogen), and the first-strand cDNA was reversed-transcribed with All-in-One cDNA Synthesis SuperMix (Biotool). Gene expression was examined with a Bio-Rad CFX Connect system by a fast two-step amplification program with 2x SYBR Green Fast qPCR Master Mix (Biotool). The value obtained for each gene was normalized to that of the gene encoding β-Actin. The ELISA kits for IFN-β, IL-6 and CCL5 (BioLegend) were used to detect the indicated cytokines in the sera or cell culture supernatants.

Co-immunoprecipitation and immunoblot analysis
The experiments were performed as previously described [44,45,72,73]. In brief, cells were lysed in Nonidet P-40 lysis buffer containing 150 mM NaCl, 1 mM EDTA,1% Nonidet P-40, and 1% protease and phosphatase inhibitor cocktail (Biotool). Cell lysates were subjected to SDS-PAGE and immunoblot analysis was performed with the appropriate antibodies. For immunoprecipitation assays, the lysates were immunoprecipitated with IgG or the appropriate antibodies, and the precipitants were washed three times with lysis buffer containing 150 mM NaCl, followed by immunoblot analysis. The antibodies were diluted in 3-5% (wt/vol) fat-free milk (BD Biosciences) or 1% BSA (Sigma) in TBS (1:500-1:2,000).

Native PAGE
The gel was pre-run with 25mm Tris and 192mm glycine, pH8.4 with 1% deoxycholate at 70 mA for 30 min. Samples in the native sample buffer (10 μg protein, 62.5 mm Tris-Cl, pH6.8, 15% glycerol and 1% DOC) were applied to the gel and electrophoresed for 30min at 70mA and 1h at 120 mA.

Deubiquitination assays
The experiments were performed as previously described [44,45,72]. For deubiquitination assays in cells, cells were lysed with the lysis buffer (100 μl) and the supernatants were denatured at 95˚C for 5 min in the presence of 1% SDS by lysates. The denatured lysates were diluted with lysis buffer until the concentration of SDS reduced below 0.1% followed by immunoprecipitation (denature-IP) with the indicated antibodies. The immunoprecipitants were subject to immunoblot analysis with anti-ubiquitin, or anti-K63-linked ubiquitin chains. For in vitro deubiquitination aasays, FLAG-tagged MITA and HA-tagged ubiquitin were cotransfected into HEK293 (human embryonic kidney 293, from ATCC) cells. Denature-IP was performed and the precipitants were eluted by 3 x FLAG peptide (sigma) to obtain ubiquitinmodified MITA. USP49 and USP49 (C262A) were obtained by a TNT in vitro transcription / translation kit (Promega). The ubiquitinated MITA were incubated with in vitro synthesized proteins at 37˚C for 2 hours followed by overnight incubation at 16˚C in the presence of 1 μM ATP. The mixture was analyzed by immunoblot with the indicated antibodies.

Cell culture
The Usp49 +/+ and Usp49 -/-MLFs were isolated and cultured as previously described [44,45,72,73]. Bone marrow cells were isolated from femurs of Usp49 +/+ and Usp49 -/mice. The cells were cultured in DMEM containing 10% (vol/vol) FBS, 1% streptomycin-penicillin and 10 μM β-mercaptoethanol. GM-CSF and M-CSF were added to the bone marrow culture for differentiation of BMDCs and BMDMs respectively. THP-1 (human monocyte; acute monocytic leukemia), U937 (human monocytic leukemia) and HEK293 cells were from the American Type Culture Collection, authenticated by STR locus analysis and tested for mycoplasma contamination. HFF (human forehead fibroblasts) cells were kindly provided by Dr. Yan-Yi Wang from Wuhan Institute of Virology, Chinese Academy of Sciences.

Transfection and reporter gene assays
HEK293 cells were transiently transfected with firefly luciferase reporter (100 ng) and TK Renilla luciferase reporter (20 ng) and indicated plasmids or empty vector (100 ng) using standard calcium phosphate precipitation. After 24 h, luciferase assays were performed with a dual-specific luciferase reporter kit (Promega). The activity of firefly luciferase was normalized by that of Renilla luciferase to obtain relative luciferase activity.

Viral infection
For qRT-PCR or immunoblot analysis, cells seeded into 24-well plates (2-5x10 5 cells per well) or six-well plates (1x10 6 -1x10 7 cells per well) were infected with various viruses for the indicated time points. For viral replication assays, cells (2-5x10 5 ) were infected with HSV-1 or H129-G4 [56]. One hour later, the supernatants were removed and cells were washed with prewarmed PBS (1 ml) twice followed by culture in full medium for 24 hours. Viral replication was analyzed by flow cytometry, fluorescent microscopy or qRT-PCR analysis. For mice infection, age-and sex-matched Usp49 +/+ and Usp49 -/were intravenously or intranasally injected with HSV-1 (2.5 x10 6 PFU per mouse) and the survival of animals was monitored every day. The lungs or brains were collected for qRT-PCR analysis or plaque assays at 24 hours or 4 days after infection, respectively.

Plaque assay
The supernatants of BMDCs or MLFs cultures and the homogenates of brains from infected mice (or the serial dilutions) were used to infect monolayers of Vero cells. One hour later, the supernatants or homogenates were removed and the infected Vero cells were washed with prewarmed PBS twice followed by incubation with DMEM containing 2% methylcellulose for 48 h. The cells were fixed with 4% paraformaldehyde for 15min and stained with 1% crystal violet for 30min before counting the plaques.

Statistical analysis
Differences between experimental and control groups were tested using Student's t-test or two-way ANOVA with Bonferroni post-test. P values less than 0.05 were considered statistically significant. For animal survival analysis, the Kaplan-Meier method was adopted to generate graphs, and the survival curves were analyzed with log-rank analysis.