Murine Gamma-Herpesvirus 68 Hijacks MAVS and IKKβ to Initiate Lytic Replication

Upon viral infection, the mitochondrial antiviral signaling (MAVS)-IKKβ pathway is activated to restrict viral replication. Manipulation of immune signaling events by pathogens has been an outstanding theme of host-pathogen interaction. Here we report that the loss of MAVS or IKKβ impaired the lytic replication of gamma-herpesvirus 68 (γHV68), a model herpesvirus for human Kaposi's sarcoma-associated herpesvirus and Epstein-Barr virus. γHV68 infection activated IKKβ in a MAVS-dependent manner; however, IKKβ phosphorylated and promoted the transcriptional activation of the γHV68 replication and transcription activator (RTA). Mutational analyses identified IKKβ phosphorylation sites, through which RTA-mediated transcription was increased by IKKβ, within the transactivation domain of RTA. Moreover, the lytic replication of recombinant γHV68 carrying mutations within the IKKβ phosphorylation sites was greatly impaired. These findings support the conclusion that γHV68 hijacks the antiviral MAVS-IKKβ pathway to promote viral transcription and lytic infection, representing an example whereby viral replication is coupled to host immune activation.


Introduction
Host cells activate innate immune signaling pathways to defend against invading pathogens. Pattern recognition receptors, including Toll-like receptors and cytosolic sensors (such as NOD-like receptors and RIG-I-like receptors), recognize pathogen-associated structural components and initiate signal transduction that leads to the biosynthesis and secretion of pro-inflammatory cytokines and interferons, thereby mounting a potent host immune response [1,2]. To survive within an infected host, viruses have evolved intricate strategies to counteract host immune responses. Herpesviruses and poxviruses have large genomes and therefore have the capacity to encode numerous proteins that modulate host immune responses.
Mitochonrial antiviral signaling (MAVS, also known as IPS-1, VISA, and CARDIF) protein serves as an adaptor to activate both the NFkB and interferon regulatory factor (IRF) pathways [3,4,5,6]. MAVS relays signals from RIG-I and MDA-5, cytosolic sensors that recognize viral dsRNA or ssRNA bearing 59triphosphate [7,8], to the IKKa/b/c and TBK-1/IKKe (also known as IKKi) kinase complexes [4,6]. IKKa/b, together with the scaffold protein IKKc, phosphorylates the inhibitor of NFkB (IkB) and promotes its subsequent ubiquitination and degradation by the proteasome, thereby unleashing NFkB that translocates into the nucleus to activate gene expression of pro-inflammatory cytokines [9,10]. By contrast, TBK-1 and IKKe directly phosphorylate a serine/threonine-rich sequence within the carboxyl termini of IRF3 and IRF7, leading to the dimerization and nuclear translocation of these transcription factors [11,12]. Together with NFkB and c-Jun/ATF-2, IRF3 and IRF7 bind to the interferon (IFN)-b enhancer and initiate the transcription of IFN-b [13,14]. Ultimately, these signaling events promote cytokine and interferon production, establishing an antiviral state in infected cells. Although it is not clear how MAVS activates these immune kinases, recent findings have established the vital roles of MAVS in host antiviral innate immunity [15]. Interestingly, the mitochondrial localization of MAVS is critical for its ability to activate downstream signaling events. As such, various RNA viruses, exemplified by human hepatitis C virus (HCV), encode proteases that cleave MAVS from the outer membrane of the mitochondrion, thereby disarming MAVS-dependent signaling cascades and the host antiviral innate immunity [6,16,17,18].
Murine gamma-herpesvirus 68 (cHV68 or MHV-68) is closely related to human Kaposi's sarcoma-associated herpesvirus (KSHV) and Epstein-Barr virus (EBV) [19]. KSHV and EBV are lymphotropic DNA viruses that are causally linked to malignancies of lymphoid or endothelial/epithelial origin, including lymphoma, nasopharyngeal carcinoma, and Kaposi's sarcoma [20,21]. Persisting within host immune cells, KSHV and EBV are known to evade, manipulate, and exploit host immune pathways [22,23]. Emerging studies suggest that c-herpesviruses may usurp host innate immune responses for their infection [24,25,26]. However, it is not known how human KSHV and EBV manipulate innate immune pathways in vivo. Such investigations are greatly hampered by the lack of permissive cell lines and animal models for both KSHV and EBV. By contrast, cHV68 infection in laboratory mice leads to a robust acute infection in the lung and a long-term latent infection in the spleen. For murine cHV68 and human KSHV, the replication and transcription activator (RTA, encoded by ORF50) is necessary and sufficient to initiate lytic replication from latently-infected cells, supporting the notion that RTA integrates diverse signaling pathways to initiate lytic replication [27,28,29]. Using cHV68 as a surrogate for human KSHV and EBV, we have unexpectedly discovered that cHV68 activated IKKb to phosphorylate RTA and promote RTA transcriptional activation, thereby increasing viral gene transcription and lytic replication. As such, RTA phosphorylation by IKKb couples cHV68 gene expression and lytic replication to host innate immune activation, representing the first example whereby a virus hijacks the antiviral MAVS-IKKb pathway to promote its lytic replication.
To determine whether cHV68 infection altered MAVS expression, we infected BL/6 mice intranasally with a high dose (1610 5 PFU) of cHV68, presumably permitting synchronized and maximal infection of lung epithelial cells. MAVS mRNA levels were determined by quantitative real-time PCR (qRT-PCR). The levels of MAVS mRNA were transiently increased at 2.5 and 5 d.p.i. in the lung and spleen, respectively ( Figure S2A). Interestingly, the up-regulation of MAVS mRNA preceded that of viral RTA mRNA ( Figure S2A and S2B), and that higher viral RTA mRNA levels tightly correlated with higher MAVS mRNA levels at 2.5 and 5 d.p.i., when MAVS mRNA levels peaked in the lung and spleen ( Figure S2C). Together with the reduced viral load in the lungs of MAVS 2/2 mice ( Figure 1A), these results suggest that MAVS is necessary for efficient lytic replication in mice and that the transiently induced MAVS expression by cHV68 infection may facilitate viral lytic replication in vivo.
To investigate the roles of MAVS in cHV68 infection, we then assessed the effects of MAVS-deficiency on cHV68 lytic replication ex vivo. Mouse embryonic fibroblasts (MEFs) were infected with a GFP-marked recombinant cHV68 (cHV68 K3/ GFP) and viral replication was examined by fluorescence microscopy and plaque assays. Surprisingly, cHV68 displayed delayed replication kinetics in MAVS 2/2 MEFs compared to MAVS +/+ MEFs at multiplicities of infection (MOI) of 0.01 and 0.1 ( Figure 1B, 1C and S3). To quantitatively determine the effect of MAVS on cHV68 lytic infection, we examined cHV68 lytic replication in MAVS +/+ and MAVS 2/2 MEFs by plaque assays. In fact, cHV68 formed approximately four-fold more plaques in MAVS +/+ MEFs than those in MAVS 2/2 MEFs, indicative of reduced initiation of lytic replication in MAVS-deficient MEFs ( Figure 1D, S4A, and S4B). Interestingly, the plaque size of cHV68 was equivalent in MAVS +/+ and MAVS 2/2 MEFs ( Figure S4C and S4D). To test whether MAVS 2/2 MEFs are defective in supporting viral lytic replication in general, we examined the lytic replication of vesicular stomatitis virus (VSV), a prototype RNA virus, with a plaque assay. Consistent with an antiviral activity of MAVS against RNA viruses, VSV formed 10fold more plaques in MAVS 2/2 MEFs than those in MAVS +/+ MEFs ( Figure 1D). The diminished lytic replication of cHV68 in MAVS-deficient MEFs is consistent with the reduced acute infection observed in the lung. To test whether exogenously expressed MAVS is able to restore cHV68 lytic replication, we generated lentivirus in 293T cells and MEFs stably expressing human MAVS (hMAVS) was established with puromycin selection ( Figure 1E). As shown in Figure 1F and 1G, exogenous hMAVS restored cHV68 lytic replication by a plaque assay and multi-step growth curves. Nevertheless, these results together support the conclusion that MAVS is necessary for efficient cHV68 lytic replication in vivo and ex vivo.

The IKKb/c Complex is an Effector Downstream of MAVS in cHV68 Lytic Replication
Two known pathways, the IKKa/b/c-NFkB and TBK-1/ IKKe-IRF3/7 pathway, have been characterized downstream of MAVS ( Figure 2A) [3,4]. We therefore used MEFs deficient in key components of aforementioned pathways to identify downstream effectors of MAVS that are critical for cHV68 lytic infection. Plaque assays and multi-step growth curves of cHV68 lytic infection showed that deficiency in TRAF6, IKKc, and IKKb, but not deficiency in the closely related IKKa, recapitulated phenotypes of MAVS deficiency ( Figure 2B and 2C). Notably, TRAF6 is necessary for MAVS to activate IKKb that requires IKKc, a scaffold protein for both IKKa and IKKb [5]. By contrast, deficiency in type I IFN receptor (IFNAR) and double

Author Summary
Innate immunity represents the first line of defense against pathogen infection. Recent studies uncovered an array of sensors that detect pathogen-associated molecular patterns and induce antiviral cytokine production via two closely related kinase complexes, i.e., the IKKa/b/c and TBK-1/IKKe. To counteract host immune defense, herpesviruses have evolved diverse strategies to evade, manipulate, and exploit host immune responses. Here we report that infection by murine gamma-herpesvirus 68 (cHV68), a model gamma-herpesvirus for human Kaposi's sarcomaassociated herpesvirus and Epstein-Barr virus, activated the IKKb kinase and IKKb was usurped to promote viral transcriptional activation. As such, uncoupling IKKb from transcriptional activation by biochemical and genetic approaches impaired cHV68 lytic replication. Our study represents an example whereby viral lytic replication is coupled to host innate immune activation and sheds light on herpesvirus exploitation of immune responses.  The initiation of cHV68 lytic replication in wild-type (WT) MEFs and MAVS 2/2 , IKKa 2/2 , IKKb 2/2 , IKKc 2/2 , TRAF6 2/2 , IFNAR 2/2 , and IRF3 2/2 IRF7 2/2 (double knockout) MEFs was assessed by a plaque assay. Data represent the mean 6 SEM of three independent experiments. (C) Multi-step growth properties of cHV68 (MOI = 0.01) in wild-type MEFs and IKKb 2/2 , IKKc 2/2 , and IKKa 2/2 MEFs were examined by plaque assays. Data represents three independent experiments. (D to F) Wild-type, MAVS 2/2 , and IKKb 2/2 MEFs were respectively infected with control lentivirus (Vec) or lentivirus containing the Flag-tagged IKKb (IKKb), and selected with puromycin. (D) IKKb expression was confirmed by immunoprecipitation and immunoblot with anti-Flag antibody (top). cHV68 plaque assays were performed as in (B). (E) Reconstituted MEFs of indicated genotypes were used for cHV68 plaque assays as in (B) with increasing doses of cHV68. Data represent the mean deficiency in IRF3 and IRF7 had no discernable effect on the plaque numbers of cHV68 in MEFs, indicating that the IRF-IFN signaling pathway is not critical for the initiation of cHV68 lytic replication ( Figure 2B). Furthermore, the exogenous IKKb expression reconstituted by lentivirus restored the lytic replication of cHV68 as determined by a plaque assay and multi-step growth curves ( Figure 2D, 2E, and 2F). Interestingly, the expression of IKKb in MAVS 2/2 did not increase cHV68 lytic replication by a plaque assay ( Figure 2E), suggesting that the MAVS-dependent activation of IKKb, rather than the absolute expression level of IKKb, is crucial for efficient cHV68 lytic replication. Additionally, exogenous IKKb did not increase cHV68 plaque numbers in MAVS +/+ MEFs ( Figure 2E), implying that endogenous IKKb is sufficient to support efficient cHV68 lytic replication. Of note, lentivirus infection reduces the difference of cHV68 plaque forming capacity in wild-type MEFs and in MEFs deficient in MAVS and IKKb ( Figure 1F and 2D). Collectively, these data indicate that the MAVS-dependent IKKb activation is critical for efficient cHV68 lytic replication.
To assess whether the kinase activity of IKKb is important for cHV68 lytic infection, we performed plaque assays with or without the specific IKKb inhibitor, Bay11-7082 (Bay11). This experiment revealed that Bay11 reduced the plaque number of cHV68 in a dose-dependent manner ( Figure 3A). Whereas treatment with 1 mM of Bay11 at 0.5 h before infection reduced cHV68 plaque number by 52%, the same treatment at 7 h post-infection (h.p.i.) reduced the plaque number by 29%, emphasizing the important roles of IKKb during early cHV68 infection ( Figure 3A). We further examined IKKb activity by an in vitro kinase assay with IKKb precipitated from MAVS +/+ and MAVS 2/2 MEFs infected with cHV68. The IKKb kinase activity was transiently and moderately increased in MAVS +/+ MEFs, however, it was drastically diminished in MAVS 2/2 MEFs after cHV68 infection ( Figure 3B). The activation of IKKb was further supported by the rapid degradation of IkBa concurrent to IKKb activation by cHV68 infection in MAVS +/+ MEFs, but not in MAVS 2/2 MEFs ( Figure 3C). To test whether UV-inactivated virus is able to trigger IKKb activation, we examined the levels of IKKb kinase activity and IkBa in MAVS +/+ MEFs by in vitro kinase and immunoblot assays, respectively. Interestingly, UV-inactivated cHV68 activated IKKb and reduced IkBa protein levels, although less efficiently than live cHV68 ( Figure 3D and 3E). This observation suggests that cHV68 lytic replication is necessary to activate the MAVS-IKKb pathway. Alternatively, UV treatment may damage or disrupt viral structural components whose integrity is necessary to activate the MAVS-IKKb pathway.
MAVS activation by RNA viruses is known to increase the expression of pro-inflammatory cytokines and interferons. However, cHV68 appears to be a poor inducer for these antiviral molecules, suggesting that cHV68 evades signaling events downstream of the MAVS adaptor. Indeed, cHV68 infection failed to upregulate the expression of IFN-b ( Figure S5A). In agreement with this observation, cHV68 RTA, similar to KSHV RTA [30], is sufficient to reduce IRF3 expression ( Figure S5B). Meanwhile, it was previously shown that cHV68 infection did not significantly activate NFkB during early infection [31], suggesting that cHV68 uncouples NFkB activation from activated IKKb. Taken together, these results support the conclusion that cHV68 infection selectively activates IKKb to promote viral lytic replication.

The MAVS-IKKb Pathway is Implicated in cHV68 Transcriptional Activation
To discern the molecular mechanisms underlying the requirement of the MAVS-IKKb pathway in cHV68 lytic infection, levels of cHV68 genomic DNA and mRNA were assessed by PCR or reverse transcription followed by real-time PCR analyses, respectively. At a low MOI (0.01), analyses by PCR ( Figure 4A) and realtime PCR ( Figure 4B) revealed comparable levels of viral genomes in MAVS +/+ and MAVS 2/2 MEFs early after de novo infection, suggesting comparable viral entry into MAVS +/+ and MAVS 2/2 MEFs. Interestingly, levels of viral mRNA transcripts representing immediate early (RTA, ORF73, and ORF57) and early (ORF60 and ORF9) gene products in MAVS +/+ MEFs were higher than those in MAVS 2/2 MEFs as determined by reverse-transcriptase PCR ( Figure 4C). Real-time PCR analyses with cDNA showed approximately 4-to 16-fold higher levels of cHV68 mRNA transcripts in MAVS +/+ MEFs compared to those in MAVS 2/2 MEFs at 2 and 3 d.p.i. ( Figure 4D). It has been shown that TRAF6 is necessary for MAVS to activate IKKb [5] and exogenous TRAF6 is sufficient to activate IKKb. To further examine the effects of the MAVS-IKKb pathway on levels of cHV68 mRNA transcripts, a bacterial artificial chromosome (BAC) containing the cHV68 genome and a plasmid expressing TRAF6 were transfected into 293T cells. The effects of exogenous TRAF6 (that activates IKKb) on viral transcription were determined by reverse transcription and real-time PCR. At 28 h post-transfection, a time point when immediate early and early genes are transcribed, exogenous TRAF6 efficiently increased the mRNA levels of cHV68 RTA, ORF57, ORF60, and ORF73, without discernable effect on levels of viral genomic DNA ( Figure 4E and 4F). These results, obtained under conditions of loss of function (MAVS 2/2 MEFs) and gain of function (TRAF6 expression), indicate that the activated IKKb increases the levels of cHV68 mRNA transcripts.

IKKb Phosphorylates cHV68 RTA and Promotes cHV68 Transcription
MAVS is an adaptor that activates IKKb and the MAVSdependent IKKb increases cHV68 mRNA levels. We thus postulated that MAVS influences cHV68 transcription via its downstream IKKb on RTA, because RTA, the master transcription activator, is critical for cHV68 lytic replication. To test this hypothesis, we examined whether IKKb phosphorylates cHV68 RTA. IKKb was purified from 293T cells and bacterial GST fusion proteins containing the RTA internal region (RTA-M, aa 335-466) or the RTA C-terminal transactivation domain (RTA-C, aa 457-583) were purified from E.coli ( Figure 5A). In the presence of [ 32 P]cATP, IKKb efficiently transferred the phosphate group to GST-RTA-C. By contrast, GST was not phosphorylated and GST-RTA-M was weakly phosphorylated by IKKb. Furthermore, the kinase domain deletion variant of IKKb (IKKbDKD) failed to phosphorylate GST-RTA-C and GST-RTA-M ( Figure 5A), and IKKa had only residual kinase activity toward RTA-C ( Figure S6). To confirm the MAVS-and IKKb-dependent phosphorylation of RTA, RTA phosphorylation in cHV68-infected cells was analyzed by autoradiography and immunoblot. We found that MAVS-and IKKb deficiency reduced RTA phosphorylation by 50% and 85%, respectively, while reconstituted IKKb expression restored RTA phosphorylation to that of RTA in MAVS +/+ MEFs    Figure 5B). To assess the roles of phosphorylation of RTA in transcription regulation, luciferase reporter assays were carried out with plasmids containing RTA-responsive promoters of RTA, ORF57, and M3. As shown in Figure 5C, the transcription activity of RTA on all three promoters was significantly increased by exogenous TRAF6 and IKKb, but not by the kinase dead variant IKKbDKD, supporting the notion that IKKb promotes RTA transcription activation via phosphorylation. When expressed to similar levels of IKKb, IKKbDKD had no significant effect on RTA transcriptional activation ( Figure S7). Given that RTA is a substrate for IKKb, we sought to examine whether RTA can physically associate with the IKKa/b/c complex. However, we were unable to detect interaction between RTA and any of the three subunits of IKKa/b/c by co-immunoprecipitation (data not shown), suggesting that the RTA interaction with the IKKa/b/c complex is transient or mediated via additional cellular proteins.
To identify IKKb phosphorylation sites, series of truncations from the C-terminus of RTA were constructed and purified as GST fusion proteins for in vitro kinase assays with IKKb. These experiments demonstrated that the IKKb phosphorylation sites were located within the region containing residues 540 through 567 ( Figure S8). Given that IKKb is a serine/threonine kinase, clusters of various serine/threonine residues were changed to alanines and RTA phosphorylation was assessed similarly. Two clusters of mutations, replacement of S 550 T 552 S 556 (STS/A) and T 561 T 562 S 564 (TTS/A) by alanines, reduced the phosphorylation levels of RTA-C by approximately 72% and 45%, respectively ( Figure 5D and S8). These results indicate that the STS and TTS sequences represent two major IKKb phosphorylation sites within the transactivation domain of RTA.
To further examine the roles of IKKb phosphorylation in regulating RTA transcription activity, reporter assays with plasmids containing wild-type RTA, the STS/A and TTS/A variants were carried out with exogenously expressed IKKb. The STS/A and TTS/A variants had lower basal activity to activate promoters of RTA, ORF57, and M3. Moreover, exogenous IKKb failed to further stimulate the transcription activities of the STS/A and TTS/A variants to activate promoters of RTA and ORF57 ( Figure 5E). Interestingly, the STS/A variant activated M3 promoter to the level of wild-type RTA with or without IKKb, indicating that the STS site is dispensable for IKKb to promote RTA transcriptional activity on the M3 promoter ( Figure 5E). It is noteworthy that the STS/A and TTS/A variants were expressed at higher levels than wild-type RTA, the transcription activities of the STS/A and TTS/A variants were approximately 50% and 20% of that of wild-type RTA, respectively, when luciferase activity was normalized against protein levels ( Figure 5F). Collectively, these results demonstrated that IKKb promotes RTA transcriptional activation via phosphorylation of the TTS and STS sites within the transactivation domain.

Impaired Lytic Replication of Recombinant cHV68 Carrying Mutations within the IKKb Phosphorylation Sites
To further investigate the roles of RTA phosphorylation, we assessed the effects of the STS/A and TTS/A mutations on cHV68 lytic replication. Taking advantage of the cHV68containing BAC with a transposon insertion that inactivates RTA (ORF50 Null) [32], a recombination-based strategy [33] was employed to generate viruses carrying wild-type RTA (Null Rescued, designated NR), the STS/A allele, or the TTS/A allele ( Figure 6A). Whereas we easily obtained recombinant cHV68 containing wild-type RTA (cHV68.NR) or the TTS/A allele (cHV68.TTS/A), the STS/A variant failed to support cHV68 recombination in multiple independent experiments. This observation suggests an essential role for the phosphorylated STS sequence in cHV68 lytic replication. To confirm the integrity of viral genomic DNA, we performed restriction digestion with KpnI and EcoRI, and analyzed with agarose gel electrophoresis. As expected, the removal of the Kanamycin cassette within RTA alleles reduced the 9-kb fragment to 7.5-kb counterpart released by KpnI digestion (Figure 6B), and abolished an EcoRI site within the Kanamycin cassette ( Figure 6C). To assess the transcriptional activity of RTA derived from BAC DNA, BAC DNA and the M3p luciferase reporter plasmid were transfected into 293T cells and RTA transcriptional activity was assessed by luciferase reporter assay. The activity of wild-type RTA to activate M3 promoter was approximately 6-fold higher than that of the TTS/A mutant ( Figure 6D). Using 293T cells transfected with the cHV68 BAC containing the TTS/A allele and a plasmid expressing TRAF6, we assessed the effects of TRAF6 (that activates IKKb) on cHV68 gene expression. In contrast to what was observed for the cHV68 BAC containing wild-type RTA ( Figure 4F), exogenous TRAF6 had marginal effects on the levels of viral mRNAs transcribed from cHV68 BAC containing the TTS/A allele ( Figure 6E). These findings are consistent with the observation that IKKb failed to further promote the transcription of the TTS/A variant ( Figure 5E), supporting the conclusion that the TTS residues constitute an IKKb phosphorylation sequence by which RTAdependent transcription is positively regulated.
Next, we examined whether recombinant cHV68.TTS/A recapitulates the defects of wild-type cHV68 lytic replication in MEFs deficient in MAVS and IKKb (plaque assays and multi-step growth curves). To assess the effects of the TTS/A mutation on cHV68 transcription activation, we normalized viral genomes immediately after cHV68 de novo infection of MEFs by qRT-PCR. With equal number of viral genomes, cHV68.NR displayed approximately 32-fold higher of RTA mRNA than recombinant cHV68.TTS/A in MAVS +/+ MEFs at 30 h.p.i. (Figure 7A). This is consistent with the observation that RTA activates its own promoter to facilitate viral lytic replication ( Figure 5C and 5E). Furthermore, multi-step growth curves (at an MOI of 0.01) demonstrated that cHV68.TTS/A had delayed replication kinetics and produced .3 orders of magnitude less virion progeny in MAVS +/+ MEFs ( Figure 7B). To test whether RTA phosphorylation and the MAVS-IKKb pathway are functionally redundant, we examined the replication kinetics of recombinant cHV68.NR and cHV68.TTS/A in wild-type, MAVS 2/2 , and IKKb 2/2 MEFs. Consistent with our previous observations ( Figure 1C, 2C, and S3B), cHV68.NR showed delayed lytic replication in MAVS 2/2 and IKKb 2/2 MEFs (Figure 7B  in wild-type, MAVS 2/2 , and IKKb 2/2 MEFs, suggesting that the MAVS-IKKb pathway functions on RTA to promote viral lytic replication ( Figure 7B and 7C). However, these replication defects of recombinant cHV68 carrying the TTS/A mutation are much more pronounced than the phenotypes of wild-type cHV68 in MAVS 2/2 and IKKb 2/2 MEFs, implying that additional kinases may influence RTA transcriptional activation via phosphorylation of the TTS site. Taken together, we conclude that the TTS site of RTA is likely phosphorylated by IKKb and is crucially important for cHV68 lytic replication.

Discussion
Here we provide evidence that murine cHV68 hijacks the antiviral MAVS-IKKb pathway to promote its lytic replication. The MAVS adaptor is important for host defense against invading pathogens, including various DNA and RNA viruses. For example, mice lacking MAVS were severely compromised in innate immune defense against VSV infection, leading to an elevated peak viral load and prolonged acute viral infection [34]. The antiviral effects of MAVS have been observed against the infection of a number of RNA and DNA pathogens [35,36,37]. To our surprise, cHV68 viral load in the lungs of MAVS 2/2 mice was significantly lower than that in the lungs of MAVS +/+ mice at 10 d.p.i. The reduced viral load of cHV68 in MAVS 2/2 mice is counter-intuitive to the presumed antiviral function of the MAVS adaptor in promoting innate immune responses. Although type I interferons in cHV68-infected mice were undetectable [38], mice deficient in type I IFN receptor had higher viral loads and succumbed to cHV68 infection [39]. We surmise that the effects of MAVS deficiency on cHV68 acute infection is likely underestimated, providing that MAVS is critical for interferon production in response to viral infection. Thus, the viral load of cHV68 acute infection in MAVS 2/2 mice likely represents a ''neutralized'' phenotype, in which reduced cHV68 lytic replication is compensated by the lack of type I interferon inhibition. Moreover, the observation that viral RTA mRNA levels correlates tightly with the MAVS mRNA levels during early cHV68 acute infection suggests that MAVS is necessary for cHV68 lytic replication ( Figure S2). Although we have not formally excluded the contribution of host immune responses against cHV68 infection to the reduced viral load at 10 d.p.i. in MAVS 2/2 mice, our experiments with cHV68 replication ex vivo demonstrated critical roles of the MAVS-IKKb pathway in facilitating cHV68 lytic infection.
During early stages of viral infection, cHV68 activated IKKb in a MAVS-dependent manner, a signaling event that is likely triggered by a variety of pathogens. The MAVS-dependent activation was supported by elevated IKKb kinase activity and accelerated IkBa degradation, signature signaling events downstream of the MAVS adaptor. Although the up-regulation of IKKb kianse activity appears modest, cHV68 may direct IKKb kinase activity to efficiently modify cellular and viral components that are critical for cHV68 infection, such as RTA. Consequently, cHV68 can harness activated IKKb without inducing NFkB activation that may be resulted from massive IKKb activation. Indeed, it was reported that cHV68 infection does not induce NFkB activation during early infection [40], suggesting that modest IKKb activation is beneficial for cHV68 infection and that cHV68 may uncouple NFkB activation from IKKb activation. Interestingly, cHV68 appears to block the interferon limb of the MAVS-dependent innate immune pathway. In fact, we found that cHV68 infection failed to induce the expression of IFN-b ( Figure  S5A). Consistent with this observation, cHV68 RTA, similar to KSHV RTA [30], is sufficient to reduce IRF3 protein ( Figure  S5B), potentially abrogating the production of interferons that otherwise would potently thwart cHV68 replication. Moreover, ORF36 was reported to deregulate the phosphorylated form of IRF3 and inhibit interferon production [41]. These observations suggest that cHV68 selectively activates the MAVS-IKKb pathway to promote viral lytic replication.
Within this report, we have identified one requisite role of the MAVS-IKKb pathway in cHV68 lytic replication with MEFs deficient in key components of this pathway. Phenotypically, cHV68 displayed similar replication defects in MEFs deficient in MAVS, IKKb, and IKKc, although the replication defects in IKKb 2/2 and IKKc 2/2 MEFs were more pronounced than those in MAVS 2/2 MEFs ( Figure 1C, 2B, and 2C). This result supports the corollary that IKKb, with the scaffold protein IKKc, functions downstream of MAVS and likely integrates additional signaling emanating from other innate immune pathways including Toll-like receptors. It is worthy to point out that our result does not exclude the antiviral activity of the IRF-IFN pathway in cHV68 lytic replication, although deficiency of IRF3 and IRF7 or IFNAR did not appear to impact the initiation of cHV68 lytic infection as assessed by plaque assays ( Figure 2B). It is possible that the IRF-IFN pathway may inhibit molecular events other than the initiation of lytic replication and reduce viral yield during cHV68 infection. Mechanistically, we identified cHV68 RTA, the master viral replication transactivator, as one of the IKKb kinase substrates. Phosphorylation of RTA by IKKb increases RTA transcriptional activity and consequently viral mRNA production. Indeed, cHV68 had lower levels of various mRNA transcripts that correlated with reduced lytic replication in MAVS 2/2 MEFs (Figure 1 and 4). Conversely, exogenous TRAF6 potentiated RTA transcriptional activity and substantially increased the levels of viral mRNA transcripts ( Figure 4F and 5C). Additionally, exogenously reconstituted expression of MAVS and IKKb restored RTA phosphorylation ( Figure 5B) and restored cHV68 lytic replication (Figure 1 and 2). Moreover, lytic replication of recombinant cHV68 viruses carrying mutations within the IKKb phosphorylation sites was greatly impaired, displaying phenotypes that are more pronounced than those of wild-type cHV68 in MEFs deficient in components of the MAVS-IKKb pathway. Conceivably, other kinases and signaling pathways may converge to modulate RTA transcriptional activation via phosphorylation within these identified IKKb sites. For example, virus-encoded kinases, such as the functionally conserved ORF36, may amplify the phosphorylation cascade that is initiated by the MAVS-IKKb pathway [42]. Most importantly, RTA auto-activates its own promoter and increases RTA protein that, in turn, up-regulates the expression of numerous immediate early and early genes during cHV68 infection. Thus, the 50-80% reduction in RTA transcriptional activity of the STS/A and TTS/ A variants ( Figure 5F) likely translates into, through the aforementioned amplification cascades, the viral yields that are less than 0.1% of the recombinant cHV68.NR ( Figure 7B). Finally, it is noteworthy that deficiency in MAVS and IKKb and mutations within RTA exhibited distinct phenotypes (such as peak viral titers of multi-step growth curves), in addition to the shared reduction of cHV68 lytic replication. These differing effects on cHV68 infection are likely due to their unique hierarchical position within the MAVS-IKKb-RTA signaling axis. In essence, these experiments identified novel phosphorylation sites within RTA that couples cHV68 lytic replication to the antiviral IKKb kinase. These findings collectively demonstrate that the MAVSdependent IKKb kinase activity is critical for RTA transcriptional activation and cHV68 lytic replication. Interestingly, Gwack et al. reported that phosphorylation of the internal serine/threoninerich region of KSHV and cHV68 RTA inhibited RTA transcriptional activity and suppressed viral lytic replication [43]. Together with our findings, these results indicate that site-specific phosphorylation determines the transcriptional activity, and likely the promoter-specificity, of gamma-herpesvirus RTA.
Although it is well accepted that the NFkB pathway is crucial for gamma-herpesvirus latent infection [44], the roles of this pathway in gamma-herpesvirus lytic replication appear to be inconsistent. Particularly, Krug et al. reported that the recombinant cHV68 expressing the IkBa super suppressor replicated indistinguishably compared to wild type cHV68 [31]. Thus, the authors concluded that the NFkB pathway is dispensable for cHV68 lytic replication. By contrast, it was shown that RelA, the p65 subunit of an NFkB transcription dimer, inhibits cHV68 lytic replication through suppressing RTA transcription activity in 293T cells [45]. Finally, our current report indicates that the MAVS-IKKb pathway is necessary for efficient cHV68 lytic replication. However, the seemingly paradox can be explained by the differential effects of three distinct components of the NFkB pathway on cHV68 lytic replication. Although the IkBa super suppressor is commonly employed to inhibit the activation of the NFkB transcription factors, it is important to note that no significant NFkB activation was observed during early cHV68 infection (within the first 6 hours post-infection) [40], temporal phase in which the critical roles of IKKb was indentified by our genetic and biochemical experiments. Conceivably, the unphosphorylatable IkBa super suppressor may not impact IKKb kinase activity. By contrast, we have focused on the IKKb kinase and our study indicated that the ability of IKKb to promote viral lytic replication largely stems from IKKb kinase activity to phosphorylate RTA and increase RTA transcriptional activation. Apparently, neither IkBa, nor RelA can do so in replace of IKKb function. On the other hand, although RelA was shown to suppress cHV68 lytic replication [45], the lack of NFkB activation during early cHV68 infection implies that cHV68 uncouples NFkB activation from IKKb activation, which are otherwise tightly correlated. As such, cHV68 infection may selectively activate the IKKb kinase, while sparing the inhibition by preventing NFkB activation. Therefore, a scenario that potentially accommodates all three reports is that nuclear activated RelA is necessary to inhibit cHV68 lytic replication and cHV68 is capable of preventing RelA activation in an IkBa-independent manner. Crucial to this hypothesis is the mechanisms that cHV68 has evolved to thwart NFkB activation and future experiments are necessary to address this possibility.
It was previously reported that cHV68 was impaired for latency establishment and reactivation in MyD88-deficient mice, although the lytic replication of cHV68 appeared to be normal in these mice [26]. Moreover, agonists specific for TLR7/8, which activate downstream signaling events through MyD88, induced KSHV lytic gene expression and reactivated KSHV replication from latently-infected B cells [25]. The specific roles of MAVS in lytic replication and MyD88 in latent infection are consistent with their distinct functions in innate immune responses of epithelial cells and immune cells, respectively. Given that MyD88 also activates the IKKa/b kinase complex, it is possible that IKKb-dependent activation of RTA may contribute to cHV68 and KSHV latent infection as well. Finally, reduced lytic replication of human KSHV and cytomegalovirus has been observed under experimental conditions in which IKKb was inhibited by Bay11, implying that human KSHV and cytomegalovirus have evolved similar molecular mechanisms to facilitate lytic replication [46,47,48]. Taken together, the mechanism whereby an antiviral innate immune signaling pathway is exploited to promote viral lytic replication may be applied to other herpesviruses and viral reactivation from latency. This study thus has uncovered an intricate interplay between the viral replication transactivator, RTA, and the MAVS-IKKb pathway. To our best knowledge, this is the first example that illustrates how a virus hijacks an antiviral signaling pathway, downstream of cytosolic sensors, to initiate its lytic replication. Perhaps, co-evolution between the persistent herpesviruses and their hosts has selected viruses that exploited the inevitable innate immune activation by viral infection. Although our current study delineates the key signaling events downstream of MAVS and IKKb, it remains unknown what viral components and cellular factors activate the MAVS-IKKb pathway and whether these mechanisms are shared by the oncogenic KSHV and EBV to promote lytic replication or reactivation.

Plasmids
For protein expression in mammalian cells, all genes were cloned into pcDNA5/FRT/TO (Invitrogen) unless specified. For protein expression and purification in E.coli, the internal region (RTA-M, aa 335-466) and C-terminal transactivation domain (RTA-C, aa 457-583) of RTA were cloned into pGEX-4T-1 (Promega) with BamHI and XhoI sites.

Mice and Infections
All animal experiments were performed in accordance to NIH guidelines, the Animal Welfare Act, and US federal law. The experimental protocol (entitled: Innate immune pathways in cHV68 infection) were approved by the Institutional Animal Care and Use Committee (IACUC). All animals were housed in a centralized research animal facility that is accredited by the Association of Assessment and Accreditation of Laboratory Animal Care International, and that is fully staffed with trained husbandry, technical, and veterinary personnel.
To assess MAVS expression in the lung and spleen, BL/6 mice were intranasally infected with 1610 5 PFU cHV68. The lungs and spleens were harvested and homogenized in DMEM.

Plaque Assay
Viral titer of mice tissues or cell lysates was assessed by a plaque assay on NIH3T3 monolayers. After three rounds of freezing and thawing, 10-fold serially-diluted virus supernatants were added onto NIH3T3 cells and incubated for 2 hours at 37uC. Then, DMEM containing 2% NCS and 0.75% methylcellulose (Sigma) was added after removing the supernatant. Plaques were counted at day 6 post-infection. The detection limit for this assay is 5 PFU. To assess the infectivity of cHV68 on various MEFs, a similar plaque assay was carried out with the initial cell density of 5000 cells/cm 2 . In Bay11-7082 treatment assay, 0.5mM or 1mM Bay11-7082 was added at 0.5 h before infection or 7 h post-infection. Supernatant was removed after 30 min incubation at 37uC, and cells were washed with medium and incubated for plaque formation.

Antibodies
Commercial antibodies used in this study include: anti-Flag (Sigma), anti-GFP (Covance), anti-IKKb (H4), anti-IkBa (C20) (Santa Cruz Biotech.), anti-actin (Abcam.). To generate antibody to cHV68 RTA, the mixture of GST fusion proteins containing the RTA-M and RTA-C was used to immunize a rabbit and polyclonal antibodies were tested for the specificity with preimmune serum as control.

In Vitro Kinase Assay
Endogenous IKKb or exogenously expressed IKKb and IKKbDKD were used for in vitro kinase assays. The kinase reaction includes 0.5 mg GST or GST fusion proteins, 100 mCi [ 32 P]cATP, and approximately 250 ng kinase in 20 ml of kinase buffer. Reaction was incubated at room temperature for 25 min and denatured proteins were analyzed by SDS-PAGE and autoradiography.

Reverse Transcription (RT)-PCR and Quantitative Real-Time PCR (qRT-PCR) Analysis
To determine the relative levels of viral transcripts, total RNA was extracted from MEFs or mice tissues using TRIzol reagent (Invitrogen). To remove genomic DNA, total RNA was treated with RNase-free DNase I (New England Biolab) at 37uC for 1 hour. After heat inactivation, total RNA was re-purified with TRIzol reagent. cDNA was prepared with 1.5 mg total RNA and reverse transcriptase (Invitrogen). RNA was then removed by incubation with RNase H (Epicentre). Abundance of viral transcripts was assessed by qRT-PCR. Mouse b-actin was used as an internal control. Primers used in this study were summarized in Table S1.

Limiting-Dilution Ex Vivo Reactivation Analyses
Bulk splenocytes were re-suspended in DMEM, and plated onto primary MEF monolayers in 96-well plates in 2-fold serial dilutions (from 10 5 to 48 cells/well) as previously described [49]. Twelve wells were plated every dilution. Reactivation percentage was scored for cytopathic effects (CPE) positive wells on day 6. In order to measure preformed infectious virus, disrupted cells were plated onto primary MEF monolayers. This procedure destroys over 99% of the cells, but has minimal effect on preformed infectious virus, thus allowing distinction between reactivation from latency and persistent infection.

Limiting-Dilution Nested PCR (LDPCR) Detection of cHV68 Genome-positive Cells
The frequency of splenocytes harboring wild-type cHV68 genome was assessed by a single-copy-sensitive nested PCR analysis of serial dilutions of splenocytes as previously described [49]. Briefly, mice spleens were homogenized and re-suspended in isotonic buffer and subjected to 3-fold serial dilutions (from 10 4 to 41 cells/well) in a background of uninfected RAW 264.7 cells, with a total of 10 4 cells per well. Twelve replicates were plated for each cell dilution. After being plated, cells were subjected to lysis by proteinase K at 56uC for 8 hours. After inactivating the enzyme for 30 minutes at 85uC, samples were subjected to nested PCR using primers specific for cHV68 ORF72. Positive controls of 10, 1, and 0.1 copies of viral DNA and negative controls of uninfected RAW 264.7 cells alone were included on each plate. Reaction products were separated using 2.5% UltraPure agarose (Invitrogen) gels and visualized by ethidium bromide staining.

Statistical Analyses
Reactivation and LDPCR results were analyzed using Graph-Pad Prism software (GraphPad Software, San Diego, CA). The frequencies of genome-positive cells were statistically analyzed using the paired Student's t-test. The frequencies of viral genomepositive cells were determined from a nonlinear regression analysis of sigmoidal dose-response best-fit curve data. Based on a Poisson distribution, the frequency at which at least one event is present in a given population occurs at the point at which the regression analysis line intersects 63.2%. Pooled data of at least three independent experiments were used to calculate P values with the two-tailed, unpaired Student's t-test.

Luciferase Reporter Assay
293T cells (2610 5 cells/well) were seeded in 24-well plates 16 hours before transfection. A total of 377 ng of plasmid DNA per well was co-transfected by the calcium phosphate method (Clontech). The plasmid cocktail includes 75 ng of luciferase plasmid (RTAp_luc, ORF57p_Luc or M3p_luc), 200 ng of pCMV-b-galactosidase plasmid, 2 ng of pcDNA5_RTA and 100 ng of pcDNA5 containing TRAF6, IKKb or IKKbDKD. At 21 hours post-transfection, whole cell lysates were used to measure the firefly luciferase activity and b-galactosidase activity.

Generating Recombinant cHV68
The bacterial artificial chromosome (BAC) system was used to generate recombinant cHV68 similarly to what was described previously [33]. Briefly, wild-type RTA or the STS/A and TTS/A alleles were PCR amplified with overlapping PCR primers. Purified PCR products, along with the BAC clone 5.15 [32]