The herpesvirus accessory protein γ134.5 facilitates viral replication by disabling mitochondrial translocation of RIG-I

RIG-I and MDA5 are cytoplasmic RNA sensors that mediate cell intrinsic immunity against viral pathogens. While it has been well-established that RIG-I and MDA5 recognize RNA viruses, their interactive network with DNA viruses, including herpes simplex virus 1 (HSV-1), remains less clear. Using a combination of RNA-deep sequencing and genetic studies, we show that the γ134.5 gene product, a virus-encoded virulence factor, enables HSV growth by neutralization of RIG-I dependent restriction. When expressed in mammalian cells, HSV-1 γ134.5 targets RIG-I, which cripples cytosolic RNA sensing and subsequently suppresses antiviral gene expression. Rather than inhibition of RIG-I K63-linked ubiquitination, the γ134.5 protein precludes the assembly of RIG-I and cellular chaperone 14-3-3ε into an active complex for mitochondrial translocation. The γ134.5-mediated inhibition of RIG-I-14-3-3ε binding abrogates the access of RIG-I to mitochondrial antiviral-signaling protein (MAVS) and activation of interferon regulatory factor 3. As such, unlike wild type virus HSV-1, a recombinant HSV-1 in which γ134.5 is deleted elicits efficient cytokine induction and replicates poorly, while genetic ablation of RIG-I expression, but not of MDA5 expression, rescues viral growth. Collectively, these findings suggest that viral suppression of cytosolic RNA sensing is a key determinant in the evolutionary arms race of a large DNA virus and its host.

The RIG-I protein consists of two caspase activation and recruitment domains at its N terminus, a helicase domain and a C-terminal domain [3]. Whereas RIG-I is in a closed, inactive conformation in uninfected cells, it adopts conformational changes upon activation by RNA viruses; this, then triggers RIG-I dephosphorylation, its K63-linked ubiquitination and mitochondrial translocation [2,4]. In this process, protein phosphatase 1, the ubiquitin ligases TRIM25 and Riplet, and the chaperone protein 14-3-3ε cooperatively allow RIG-I to active downstream targets, resulting in an antiviral state [5][6][7][8]. Emerging evidence also pinpoints to a role of RIG-I in the detection of DNA viruses [2,9]. RIG-I recognizes multiple RNAs of Kaposi's sarcoma-associated herpesvirus (KSHV) and host-derived 5'-ppp-vtRNAs, which triggers IFN production and limits KSHV reactivation from latency [10][11][12]. In latently infected B cells, RIG-I binds to small noncoding RNA encoded by Epstein-Barr virus (EBV) to drive IL-10 production [13,14]. Remarkably, herpes simplex virus 1 (HSV-1) triggers RIG-I activation via RNA polymerase III that generates 5'-ppp RNA species, including host 5S ribosomal pseudogene transcripts due to virus-mediated depletion of specific RNA-binding proteins [15,16].
To unravel key mechanisms of viral replication, we performed global gene expression and genetic analyses. We show that HSV-1 subverts RIG-I-mediated cytosolic RNA sensing via the γ 1 34.5 protein. Mechanistically, γ 1 34.5 targets RIG-I, which precludes the assembly of RIG-I and the chaperone 14-3-3ε into a translocon complex necessary for RIG-I translocation from the cytosol to mitochondria. As such, HSV-1 γ 1 34.5 inactivates RIG-I, which promotes effective viral replication.

HSV-1 regulates host genes linked to cytosolic RNA recognition
To gain insight into HSV replication, we examined global gene expression in response to virus infection by RNA deep sequencing. Among diverse host genes, we observed a range of infected, infected with wild type HSV-1 or Δγ 1 34.5 (5 pfu/cell). At 8 h postinfection, RNAs were extracted for RNA deep sequencing. Data from triplicate samples, processed as described in the Materials and Methods, are presented as Scatterplot. X axis denotes log2 fold change in HSV-1 to mock ratios. Y axis indicates log2 fold change in Δγ 1 34.5 to mock ratios. Dots are single genes. A few representative genes upregulated more by Δγ 1 34.5 than by wild type HSV-1 are highlighted. (B) GSEA hallmark analysis on genome wide gene expression. Pathways enriched in gene sets upregulated by Δγ 1 34.5 relative to wild type HSV-1 are ranked based on the normalized enrichment score (NES). False discovery rate (FDR) q value < 0.25 is defined as significantly enriched. Nominal p values are also indicated for top ranked pathways. (C) Heatmap visualization of RNA transcripts linked to the IFN response. The map shows 46 genes including IFN-stimulated genes, intracellular DNA sensors and RNA sensors. G1, G2 and G3 are distinct experimental replicates. The data represents the Log2FC (Fold Changes). (D) Effect of γ 1 34.5 on antiviral gene expression. MEF cells infected as in (A) were subjected to quantitative PCR analysis to test the expression of Ifit1, Ccl5, Ifi204, Ddx58, Dhx58, Ddx60, and Ifih1, Mx2, Oas1b, Oas2, Oas3, and Oasl1. The results were expressed as fold activation relative to 18S ribosomal RNA, with standard deviations among triplicate samples. The data were statistically analyzed by one-way ANOVA ( �� , P < 0.01).
Given the normalized enhancement scores closely coupled with the IFN pathways (Figs 1B and S1), we surveyed innate immune genes. This revealed unique patterns of RNA transcripts that were differentially accumulated (Fig 1C). We found that both viruses upregulated many IFN-related genes relative to the mock control, which reflects the cellular antiviral response. A more detailed comparative analysis of the genes upregulated by wild type vs mutant HSV-1 revealed several distinct features. The γ 1 34.5 null mutant, compared to wild type virus, highly stimulated a spectrum of IFN effector molecules, including Ifit1 (Isg56), Isg15, Sp100, Gbp5, Oasl2, Mx1, and Ifgga3. Moreover, As compared to the wild type virus, the γ 1 34.5 null mutant strongly increased transcript abundance of several DNA sensors (e.g. Ifi203, Ifi204, and Ifi205) that mediate antiviral gene induction in response to DNA ligands [36][37][38]. This data is in line with the fact that HSV-1 inactivates STING, a critical adaptor downstream of many DNA sensors, via the γ 1 34.5 gene product [34]. Importantly, the γ 1 34.5 null mutant exhibited propensity to induce more transcripts of Ddx60, Ddx58 (RIG-I), Dhx58 (LGP2), Ifih1 (MDA5) and Zbp1, which are prominent components of cytosolic RNA sensing pathways that regulate cytokine induction and necroptosis [1,39]. We also noted that wild type virus and the γ 1 34.5 null mutant comparably induced a subset of IFN-stimulated genes such as Mx2, Oas3 and Oasl1. These heterogeneous responses are likely attributable to a complex regulation of individual IFN-stimulated genes. We validated the RNAseq results by performing qPCR analysis of representative genes, which confirmed increased antiviral gene expression in the absence of γ 1 34.5 ( Fig 1D). These results raised the possibility that besides intracellular DNA recognition, γ 1 34.5 might modulate cytosolic RNA sensing.

The RIG-I-γ 1 34.5 axis influences the innate immune response
Several lines of evidence show that HSV-1 triggers RIG-I to initiate antiviral signaling [2,9,15]. As wild type virus, unlike Δγ 1 34.5, favorably attenuated the IFN response, we reasoned that γ 1 34.5 might modulate innate immunity mediated by RIG-I. To probe this, we examined the induction of cytokines and ISGs by wild type HSV-1 and the γ 1 34.5 null virus in the presence or absence of RIG-I. In Rig-I +/+ mouse embryonic fibroblast (MEF) cells, wild type virus modestly induced the expression of IFN-β, Ifit1, Ifit2 (Isg54) and Ccl5 (RANTES) as measured by qPCR (Fig 2A). This was in stark contrast to the γ 1 34.5 null virus, which robustly induced the transcript expression of those genes, suggesting that the γ 1 34.5 protein acts to dampen RIG-Idependent innate immune responses. Importantly, in infected Rig-I -/-MEFs, viral induction of antiviral genes was greatly diminished, which attests a critical role of RIG-I in HSV-1 sensing as previously shown [15,16,40]. We further confirmed that a recombinant HSV, in which the γ 1 34.5 gene was restored, behaved like wild type virus, ruling out the possibility that the observed phenotypes were due to an irrelevant mutation(s) elsewhere in the virus genome (S2 Fig). To assess whether γ 1 34.5 functioned similarly in human cells, we determined cytokine expression in human lung fibroblasts infected with either wild type HSV-1 or the γ 1 34.5 null virus (Fig 3A and 3B). Albeit with a different magnitude, the γ 1 34.5 null virus readily induced expression of IFN-β, Ifit1, Ifit2 (Isg54) and Ccl5 (RANTES), relative to the wild type virus. RIG-I depletion by shRNA profoundly impaired such response to the HSV-1 variants. These results suggested that γ 1 34.5 dampens the antiviral response mediated by RIG-I.
To verify an effect of γ 1 34.5 on RIG-I signaling, we examined IRF3 phosphorylation, a hallmark of innate immune activation [1,41]. In Rig-I +/+ MEFs the γ 1 34.5 null virus, but not wild type virus, readily induced the phosphorylation of IRF3 (Fig 2B). Although viral infectivity, as measured by ICP27 expression, was comparable, phosphorylation of IRF3 occurred only with the virus devoid of γ 1 34.5 expression. The recombinant HSV-1 with repaired γ 1 34.5 inhibited IRF3 phosphorylation similarly to wild type HSV-1 (S2 Fig). In Rig-I -/-MEFs virus-induced phosphorylation of IRF3 was abolished, further indicating a requirement of RIG-I in activating IRF3 during HSV-1 infection (Figs 2B and S2). Infection by the γ 1 34.5 null virus, but not wild type virus, also readily triggered IRF3 phosphorylation in human lung fibroblasts ( Fig 3C). Taken together, these data show that the γ 1 34.5 gene product functions as a previously unrecognized herpesviral inhibitor of RIG-I-induced innate immune signaling.

The γ 1 34.5 gene product interacts with and inhibits RIG-I
To determine whether γ 1 34.5 interacts with RIG-I, we performed immunoprecipitation using an anti-RIG-I antibody ( Fig 4A). We found that γ 1 34.5 was precipitated with endogenous RIG-I in cells infected with wild type virus. Neither RIG-I nor γ 1 34.5 was precipitated by control IgG. The γ 1 34.5-RIG-I interaction was further verified in reciprocal immunoprecipitation with an anti-γ 1 34.5 antibody (Fig 4B). To determine whether γ 1 34.5 can bind RIG-I in the absence of other HSV proteins, we tested this interaction in 293T cells co-expressing Flagγ 1 34.5 or Flag-mCherry (control) together with Myc-RIG-I. As shown in Fig 4C, HSV-1 γ 1 34.5, but not irrelevant mCherry, precipitated with Myc-RIG-I by IP with anti-Myc antibody. Conversely, RIG-I was specifically precipitated with γ 1 34.5 by IP with anti-Flag antibody ( Fig 4D). Moreover, γ 1 34.5 interacted with the 2CARD domain of RIG-I and inhibited The γ 1 34.5 protein blocks the mitochondrial translocation of RIG-I To initiate antiviral immunity, RIG-I undergoes K63-linked ubiquitination by the E3 ligase tripartite motif-containing protein (TRIM25) and subsequently moves from the cytoplasm to mitochondria [1]. We noted that Sendai virus (SeV) effectively induced K63-linked ubiquitination of RIG-I ectopically expressed in 293T cells. Whereas influenza A virus NS1, a viral TRIM25 antagonist [42], potently diminished the K63-linked ubiquitination of RIG-I, HSV-1 γ 1 34.5 exhibited little inhibitory effect on RIG-I polyubiquitination (Fig 5A and 5B). Congruently, the γ 1 34.5 protein did not interrupt the interaction of TRIM25 and RIG-I (Fig 5C), suggesting a different mechanism is in operation. To test this, we examined the subcellular localization of RIG-I by fractionation analysis. In uninfected control 293T cells, RIG-I was seen primarily in the cytoplasm, as expected ( Fig 6A). SeV infection markedly increased the abundance of RIG-I at the mitochondria; however, overexpression of γ 1 34.5 substantially reduced the abundance of RIG-I in the mitochondrial fraction, which correlated with reduced IFN-promoter activation ( Fig 6B). These results demonstrate that the γ 1 34.5 protein interrupts the translocation of RIG-I from the cytoplasm to the mitochondria induced by RNA virus infection.

PLOS PATHOGENS
We next analyzed the cytosol-to-mitochondria re-localization of RIG-I in HSV-1-infected cells. RIG-I predominantly localized to the cytoplasmic fraction in mock-infected MEF cells ( Fig 6C). Infection with the γ 1 34.5 null virus drastically increased the abundance of RIG-I in the mitochondrial fraction. However, this increase was not detectable in wild type HSV-1 infected cells. Taken in combination, these results show that the γ 1 34.5 protein prevents the mitochondrial translocation of RIG-I during HSV-1 infection.

HSV-1 γ 1 34.5 prevents assembly of RIG-I and 14-3-3ε into a functional translocation complex
Cellular 14-3-3 proteins are essential components of cytosolic RNA sensing machineries [8,43]. In response to RNA virus infections, 14-3-3ε forms a complex with RIG-I to facilitate its mitochondrial translocation, activating downstream signaling. To further probe the mechanism of γ 1 34.5 action, we assessed the effect of γ 1 34.5 on RIG-I-14-3-3ε complex formation. Myc-RIG-I detectably bound to endogenous 14-3-3ε in mock-infected cells, and infection with SeV enhanced such interaction ( Fig 7A). The weaker band in the sample with vector cotransfection likely reflected non-specific binding. However, the presence of γ 1 34.5 inhibited the binding of RIG-I to 14-3-3ε, as did dengue NS3, a viral antagonist of 14-3-3ε [44]. Interestingly, dengue NS3 bound to 14-3-3ε whereas γ 1 34.5 failed to do so ( Fig 7B). We reasoned that the γ 1 34.5 protein may specifically target RIG-I, which prevents its assembly into the 14-3-3 translocation complex. To further test this, we analyzed 14-3-3ε-RIG-I binding in HSV-1-infected cells. As illustrated in Fig 7C, only a small amount of 14-3-3ε was precipitated with RIG-I in mock infected cells; however, infection with the γ 1 34.5 null virus increased the amount of 14-3-3ε that co-precipitated with RIG-I. Crucially, infection with wild type HSV-1 nearly eliminated the binding of 14-3-3ε to RIG-I, which coincided with γ 1 34.5 bound to RIG-I.
To determine the consequence of γ 1 34.5 expression on downstream signaling, we assessed IRF3 activation with respect to the distribution of 14-3-3ε and RIG-I (Fig 7D). We observed  that in mock-infected cells, RIG-I and 14-3-3ε were predominantly localized to the cytoplasm. On the other hand, MAVS was seen at the mitochondria, as expected. Infection with the γ 1 34.5 null virus significantly increased the levels of 14-3-3ε and RIG-I at the mitochondria. However, this was not detectable with wild type virus, indicative of a block in mitochondrial translocation of RIG-I. Inhibition of RIG-I mitochondrial translocation also occurred upon ectopic expression of γ 1 34.5 (S5 Fig), indicating that γ 1 34.5 is directly responsible for this effect. Further analysis showed that unlike wild type HSV-1, the γ 1 34.5 null virus induced phosphorylation of IRF3 (Fig 7D). Thus, the γ 1 34.5 protein functionally disengaged RIG-I from 14-3-3ε to halt downstream signaling. This demonstrates that HSV-1 γ 1 34.5 specifically interrupts a key step of RIG-I activation.

The RIG-I-γ 1 34.5 interaction affects viral growth
Lastly, we assessed the impact of the RIG-I-γ 1 34.5 interaction on viral growth. Fig 8A shows that wild type HSV-1 replicated robustly in both Rig-I +/+ and Rig-I -/-MEF cells, with titers reaching 2 x 10 6 and 4 x 10 6 pfu/ml, respectively. In contrast, the γ 1 34.5 null virus replicated poorly in Rig-I +/+ MEF cells, with a titer of 2 x 10 2 pfu/ml. The growth defect of the γ 1 34.5 null virus was dramatically restored in Rig-I -/-MEF cells (1 x 10 4 pfu/ml). We further examined the kinetics of viral growth (Fig 8B). In Rig-I +/+ MEF cells, wild type virus grew steadily as infection progressed, with a titer increasing to 1 x 10 6 pfu/ml at 48 h post-infection. However, the PLOS PATHOGENS γ 1 34.5 null mutant barely replicated in Rig-I +/+ cells throughout infection, with a titer of <1 x 10 2 pfu/ml. In Rig-I -/cells, both viruses replicated more efficiently than each virus did in Rig-I +/+ cells, with a faster growth kinetics. Similarly, knockdown of RIG-I by shRNA enhanced HSV replication in human lung fibroblasts in the absence γ 1 34.5 (Fig 8C and 8D). In contrast, virus replication was comparable in the presence and absence of the related RNA sensor MDA5 (S6 Fig). Collectively, these results demonstrate that RIG-I functions to limit HSV-1 replication, where the γ 1 34.5 protein serves to overcome the RIG-I-mediated virus restriction.

Discussion
Productive herpesvirus infection involves viral blockade of translation arrest by the dsRNAdependent protein kinase, PKR [26,[45][46][47]. Whether and how herpesviruses interact with the networks of cytosolic RNA sensors has been unresolved. Here we identified a previously unrecognized mechanism by which HSV-1 inhibits cytosolic RNA recognition by RIG-I. This activity is dependent on the virulence factor γ 1 34.5, which prevents assembly of RIG-I and 14-3-3ε into a 'translocon' complex, thereby impairing subsequent IRF3 phosphorylation and antiviral immunity. This work further underscores the importance of RIG-I in HSV-1 restriction [15,16,40].
As a large DNA virus, HSV-1 replication proceeds temporally, generating various virusand host-derived stimulatory RNA species during the course of infection [15,17] that likely trigger several RNA sensing pathways simultaneously or sequentially. First, onset of HSV DNA replication is thought to activate PKR via dsRNA and shuts off protein synthesis [19]. Yet, viral γ 1 34.5, in cooperation with Us11, functionally inhibits PKR at discrete phases of HSV infection [48]. This involves dephosphorylation of the eukaryotic translation initiation factor eIF2 alpha by γ 1 34.5, which facilitates neuroinvasion in vivo [26,49]. Published work also indicates that herpesviruses instigate RIG-I through RNA polymerase III, a distinct pathway to initiate antiviral immunity [16,50]. HSV-1 infection causes translocation of host 5S ribosomal pseudogene transcripts (in particular RNA5SP141; also generated by RNA polymerase III) from the nucleus to the cytoplasm, and their subsequent unmasking by HSV-1-mediated depletion of RNA5SP141-binding proteins leads to activation of RIG-I [15]. RIG-I activation imposes another barrier to HSV-1 which would necessitate the immune-evasive actions of one or more viral genes to facilitate infection. As γ 1 34.5 is expressed early as well as late in infection [51-53], it may serve to control different RNA sensing machineries, which ensures the progression of HSV-1 replication.
We recently reported that the γ 1 34.5 protein interferes with the DNA sensing pathway through STING inactivation [34]. STING acts downstream of several DNA sensors, including cyclic GAMP synthase (cGAS), IFI16 and DDX41 that detect and limit HSV infection [36,54-56]. HSV γ 1 34.5 directly targets STING, and this interaction depends on the N-terminal domain of γ 1 34. 5 [34]. This is different from its regulation of RIG-I, where full-length γ 1 34.5 is required for RIG-I antagonism as indicated by our data that showed that deletion of either the N-terminal or C-terminal domain of γ 1 34.5 abolished its activity against RIG-I. This suggests that the γ 1 34.5 protein functions to regulate two major signaling proteins in innate sensing (RIG-I and STING) using distinct interacting modes. In this context, it is notable that crosstalk between RIG-I and STING has been reported. For example, RIG-I activation by synthetic or viral agonists induces STING expression [57]. Conversely, STING deficiency leads to diminished IFN production in response to dsRNA or RNA virus infection [54,58]. Upregulated RIG-I can also participate in STING degradation [59]. Furthermore, temporally distinct roles have been reported for cGAS and RIG-I in the sensing of HSV-1 infection and subsequent cytokine induction [15]. Thus, a complex interplay between RIG-I and STING exists, and how γ 1 34.5 coordinately controls these two antiviral pathways throughout the HSV life cycle is an important question that awaits further investigation. Considering host selective pressures, our results suggest that the γ 1 34.5 gene of HSV-1 may have evolved to cope with RIG-I in addition to PKR and STING. This model may explain, at least in part, why the γ 1 34.5 protein functions as an HSV virulence factor in vivo [60], which warrants further investigation.
The mechanisms of RIG-I regulation are under intensive investigation [1,2]. Accumulating studies show that herpesviruses activate RIG-I via both viral and host-derived RNAs [10][11][12]15,16]. The mechanism by which HSV-1 γ 1 34.5 inactivates RIG-I has been unknown. We found that HSV-1 γ 1 34.5 displayed no inhibitory effect on the K63-linked ubiquitination of RIG-I mediated by TRIM25, an essential step in RIG-I activation [1,2]. Instead, γ 1 34.5 prevented the assembly of the RIG-I-14-3-3ε complex and its re-localization from the cytoplasm to the mitochondria. This illustrates a powerful mechanism by which HSV-1 avoids MAVS activation, IRF3 phosphorylation and subsequent cytokine expression. As intact γ 1 34.5 is required to interact with the CARDs of RIG-I, we infer that γ 1 34.5 may compete with 14-3-3ε for the binding site on RIG-I. Alternatively, HSV-1 γ 1 34.5 may alter the conformation of RIG-I that is required for access by 14-3-3ε, which then halts mitochondrial translocation of RIG-I. We propose that while RNA ligands induce conformational changes and posttranslational modifications of RIG-I, γ 1 34.5 serves to selectively disable the cytosol-to-mitochondria translocation of RIG-I, which ultimately inhibits immune activation.
Our work reveals that the γ 1 34.5-RIG-I interaction influences HSV-1 replication. In the presence of RIG-I, wild type virus replicated efficiently whereas deletion of the γ 1 34.5 gene crippled viral replication. This is linked to the ability of γ 1 34.5 to block the mitochondrial translocation of RIG-I, which phenotypically resembles the antagonistic activity by the NS3 proteins of dengue and Zika viruses [44,61]. Unlike the flavivirus NS3 proteins, HSV-1 γ 1 34.5 uniquely targets RIG-I instead of 14-3-3. Many other RNA viruses perturb the RIG-I signaling pathway [62]. For example, influenza A virus NS1 inhibits the K63-linked ubiquitination of RIG-I mediated by TRIM25 [42]. The NS3/4A protease of hepatitis C virus cleaves MAVS [63,64] and also abolishes RIG-I ubiquitination by Riplet [65], while the 3C proteins from certain picornaviruses cleave RIG-I [66]. Other studies suggest that the BPLF1 protein of Epstein-Barr virus (EBV) inhibits the K63-linked ubiquitination of RIG-I through recruitment of 14-3-3 to sequester and inactivate TRIM25, whereas ORF64 of KSHV directly cleaves K63-polyubiquitin chains of RIG-I [67]. HSV-1 UL37 deaminates and inhibits RIG-I sensing [68,69]. On the other hand, Us11 of HSV-1 impairs RIG-I via PACT inactivation [70,71]. Additional modulation of RIG-I by γ 1 34.5 indicates a complex regulatory circuit that might be relevant to temporal replication of herpesviruses. Strikingly, genetic ablation of RIG-I, but not MDA5, markedly reversed the HSV-1 growth defect in the absence of γ 1 34.5. Such specificity may reflect the requirement of a separate HSV function for restricting MDA5, or it may imply that MDA5 is not required for the restriction of HSV-1 replication. Further work is required to clarify this issue.
It is noteworthy that inborn errors in type I IFN-mediated immunity contribute to HSV encephalitis in humans. This is illustrated by mutations in TBK1, IRF3, or Toll-like receptor 3 (TLR3) in individual patients [72][73][74][75]. Consistently, in murine HSV encephalitis models, deficiency in MAVS or in TRIF that is a TLR3 adaptor, increases mortality rates [76]. It is tempting to speculate that inhibition of RIG-I by γ 1 34.5 may favor HSV-mediated pathology. Further development of relevant in vivo models will be required to address this question. Moreover, genetically modified HSV that lacks γ 1 34.5 is avirulent and approved for melanoma therapy in humans [21,24]. Recent work further suggested that its combination with immune checkpoint blockade enhances therapeutic efficacy [23]. Therefore, it would be intriguing to investigate whether RIG-I activation by γ 1 34.5 null oncolytic HSV primes antitumor immunity. Further characterization may lead to the development of next generation therapeutic agents.

Virus infection assay
Cells were infected with viruses at the indicated multiplicity of infection. After adsorption for 2 h, the monolayers were overlaid with DMEM supplemental with 1% FBS and incubated at 37˚C. For viral titer determination, samples were harvested at 48 hours postinfection and viruses, released by three cycles of freezing and thawing, were titrated on Vero cells [53].

RNA sequencing and data analysis
Total RNA from Mock, HSV-1(F) and Δγ 1 34.5 infected MEFs was extracted using a RNeasy Plus mini kit (Qiagen) and then subjected to RNA-deep sequencing (RNA-seq) analysis (Novogene). Sequencing libraries were generated using NEBNext Ultra RNA Library Prep Kit for Illumina following the manufacturer's recommendations, and index codes were added to attribute sequences to each sample. The clustering of the index-coded samples was performed on a cBot Cluster Generation System using TruSeq PE Cluster Kit v3-cBot-HS (Illumina). After cluster generation, the library preparations were sequenced on an Illumina HiSeq platform, and 125 bp/150 bp paired-end reads were generated.
Raw reads were aligned to the mouse reference genome in a splice-aware manner using the STAR aligner [85]. ENSEMBL gene and transcript annotations, which include non-coding RNAs in addition to mRNAs, were used. With Feature Counts [86], gene expression was first quantified as raw read counts and then normalized to reads-per-million for direct comparison between samples. Pair-wise differential expression statistics (fold-change and p-value) were computed using edgeR [87,88]. p-values were adjusted for multiple testing using the false discovery rate (FDR) correction of Benjamini and Hochberg [89].
To perform GSEA analysis on RNA-seq datasets, the log fold changes of all genes in edgeR result output were used to generate a ranked list for GSEA preranked analysis using the Molecular Signatures Database v5.2 (H: hallmark gene sets) [35]. Specifically, the differences in log2 fold changes of all genes in virus-infected cells relative to the mock group were used to generate the ranked list. Enriched gene sets ranked by GSEA normalized enrichment score (NES) were visualized using ggplot2 package in R. Gene sets with a nominal p value < 0.05 and false discovery rate (FDR) < 0.25 were defined as significantly enriched. Heat maps were produced from the primary data (the normalized expression value) using the R package pheatmap v1.0.8.

Quantitative real-time PCR assay
Total RNA was harvested from cells using a RNeasy Plus mini kit (Qiagen). Genomic DNA was eliminated using gDNA Eliminator columns. cDNA was synthesized using a high-capacity cDNA reverse transcription kit (Applied Biosystems). Quantitative real-time PCR was performed using an Applied Biosystems ABI Prism 7900HT instrument with SYBR green master mix (Applied Biosystems). Gene expression levels were normalized to that of endogenous control 18S rRNA. Relative gene expression was determined as described previously [90]. All primers were listed in S1 Table. Western blot Cells were harvested, washed with phosphate-buffered saline (PBS), and lysed as described previously [91]. Samples were then subjected to electrophoresis on denaturing polyacrylamide gels, transferred to Polyvinylidene difluoride (PVDF) membranes, and reacted with indicated antibodies [83].

Immunoprecipitation and ubiquitination analysis
To detect protein interactions, immunoprecipitation was performed as described previously [91]. Briefly, cells were lysed, and cell extracts were incubated with the indicated antibodies and agarose conjugated with protein A/G (sc-2003, Santa Cruz Biotechnology) at 4˚C. The beads were washed three times with wash buffer (50 mm Tris-HCl, pH 7.4, 150 mm NaCl, 5 mm EDTA, 0.1% Triton X-100, and protease inhibitor mixture). The samples were then subjected to immunoblotting analysis. For detection of RIG-I ubiquitination, cells were lysed with lysis buffer (1% SDS, 150 mM NaCl, 10 mM Tris-HCl, pH 8.0) with 2mM sodium orthovanadate, 50 mM sodium fluoride, and protease inhibitors. Samples were precipitated and washed RIPA buffer containing 2M urea to remove nonspecific binding of other ubiquitinated proteins.

Lentiviral transduction
pLKO.1 Puro RIG-I Target shRNA Plasmid and pLKO.1 Puro Non-Target shRNA Control Plasmid (negative control) were purchased from shRNA (Sigma-Aldrich). The lentivirus was produced after transfection of shRNA plasmid together with package plasmids (pCMV-VSV-G, pMDLg/pRRE, and pRSV-REV) in HEK-293T cells. HEL were then infected with the collected lentivirus. At 16 h after infection, the medium was replaced with fresh medium. At 3 days after infection, the cells were selected by 3μg/ml puromycin (sc-205821, Santa-Cruz Biotechnology). Experiments were performed within 2 weeks after lentiviral transduction.

Mitochondrial fractionation analysis
MEFs were mock infected or infected with wild type HSV-1(F) and Δγ 1 34.5. At 8 h postinfection, the cells were harvested to prepare cytoplasmic and mitochondrial fractions using an EzSubcell Fraction Kit (ATTO, Tokyo, Japan). Samples were then analyzed by immunoblotting. For HEK-293T analysis, cells were transfected with Flag-γ 1 34.5 for 24 h, followed by treatment with SeV at the 100HA/ml for additional 24h. And the cells were harvested for cell fractionation analysis.

Statistical analysis
All data were presented as means ± SD and analyzed using GraphPad Prism software (version 6). One-way ANOVA with Dunnett's multiple comparisons or an unpaired two-tailed Student's t test was used as indicated in the legends. For the graphs, data were in general three biological replicates and reproduced in independent experiments as indicated in the legends.