Human Herpesvirus 6B U26 Inhibits the Activation of the RLR/MAVS Signaling Pathway

ABSTRACT U26 is one of the roseolovirus unique genes with unknown function. Human herpesvirus 6B (HHV-6B) pU26 is predicted to be an 8-transmembrane protein containing a mitochondrion location signal. Here, we analyzed U26 function during HHV-6B infection and find that (i) HHV-6B U26 is expressed at a very early stage during HHV-6B infection, and knockdown of it results in a significant decrease of HHV-6B progeny virus production; (ii) U26 inhibits the activation of the retinoic acid-inducible gene I (RIG-I)-like receptor (RLR)/mitochondrial antiviral signaling protein (MAVS) signaling pathway, an important anti-HHV-6B infection innate immune response, by targeting MAVS protein for degradation; and (iii) a portion of U26 locates to the mitochondria, which could affect the mitochondrial membrane potential and finally leads to MAVS degradation. These findings indicate that HHV-6B U26 is a novel antagonistic viral factor against host innate antiviral immunity.

organ transplantation, and AIDS and other immunosuppressed patients (13). Recently, it was reported that HHV-6 could be one of the pathogenic agents of Alzheimer's disease (14).
The innate immune response is the first line of host defense against incoming microbial infection (15,16). After virus infection, viral nucleic acid can be detected by various pattern recognition receptors (PRRs) (17). Following recognition, the antiviral immune response will be activated, which results in the production of type I interferons (IFNs) and other cytokines to achieve the inhibition and/or elimination of the virus infection. Generally, among PRRs, some Toll-like receptors (TLRs) and/or retinoic acidinducible gene I (RIG-I)-like receptors (RLRs) recognize viral RNAs during many virus infections (18). RLRs, which include RIG-I, melanoma differentiation-associated gene 5 (MDA5), and the Laboratory of Genetics and Physiology 2 gene (LGP2), recruit and activate mitochondrial antiviral signaling protein (MAVS) localized at the mitochondrial outer membrane. MAVS plays an important role in antiviral immunity and autoimmunity (19,20). There are two pathways to activate the downstream reaction of MAVS. The first one involves MAVS recruitment of downstream TANK-binding kinase 1 (TBK1), followed by TBK1 phosphorylation of interferon regulatory factor 3 (IRF3), which eventually stimulates the production of proinflammatory cytokines such as IFN-b and interleukin-6 (IL-6). The second pathway involves MAVS liberation of NF-κB from the IκB kinase a (IKKa)/IKKb/NEMO complex. NF-κB enters the nucleus and turns on the expression of a cluster of genes, which is essential for immune and inflammatory responses (21).
During viral infections, different antiviral DNA and RNA responses via PRRs are activated (17). In the case of infection of host cells by many DNA viruses, DNA sensors such as TLR9 and cyclic GMP-AMP synthase (cGAS) are activated. Subsequently, signals are transduced via downstream adaptors such as myeloid differentiation factor 88 (MYD88) and stimulator of interferon genes (STING), which is an endoplasmic reticulum (ER)-resident adaptor (22,23). Most RNA viruses are recognized by a specific set of RNA sensors. RLRs are such sensors, and they play a vital role during RNA virus infection (24)(25)(26). Interestingly, during several DNA virus infections, an antiviral role of the RLR signaling pathway has also been reported.
Kaposi's sarcoma-associated herpesvirus (KSHV) is recognized by both RIG-I and MDA5 via sensing of host RNAs during virus infection (27). During human cytomegalovirus (HCMV) infection, the RIG-I/MAVS signaling pathway is important for antiviral immune responses (28). It has been reported that HHV-6 infection could modulate CD4 1 T cells' intracellular TLR9 expression and TLR9-mediated signaling for the development of an effective immune response (29). It is still unknown whether RNA sensors function during HHV-6 infection.
Multiple immune escape strategies are employed by individual viruses to evade detection by the host immune system. HCMV US9 locates to the mitochondria and inhibits the activation of IRF3 by inducing MAVS leakage from the mitochondria (28). The herpes simplex virus 1 (HSV-1) tegument protein UL46 interacts with both STING and TBK1 via separate domains and possibly disrupts their functions to block this DNAsensing pathway (30). HSV-1 US11 interacts with endogenous RIG-I and MDA5 through its C terminus and inhibits the downstream activation of the RLR signaling pathway (31). An HHV-6 gene cluster, located in U20s of the HHV-6 genome, encodes viral proteins responsible for immune evasion by the virus. U20 is necessary for HHV-6-induced inhibition of tumor necrosis factor (TNF) receptor-dependent signaling and apoptosis (32). An HHV-6 U21 homolog is present in HHV-7, and both of them could divert major histocompatibility complex (MHC) class I molecules to a lysosomal compartment, which contributes to recognition escape by cytotoxic T lymphocytes (CTLs) (33,34). U24 inhibits CD3 recycling to the cell surface and mediates T cell receptor (TCR) at early endosomes, which prevents the activation of T cells (35). It is still unknown whether there is a viral anti-nucleic acid sensing factor responsible for HHV-6 immune evasion.
Here, we have analyzed HHV-6B U26 function in detail and found that a portion of the HHV-6B U26 protein (pU26) was expressed in mitochondria and that pU26 inhibited the phosphorylation of IRF3 and the subsequent production of type I interferon by targeting MAVS protein to the proteasome degradation pathway. Knockdown of U26 expression resulted in the activation of the RLR/MAVS signaling pathway and decreased the production of progeny HHV-6B. In conclusion, we have identified a novel anti-HHV-6 infection pathway in host cells and also revealed that HHV-6B pU26 counteracts the antiviral RLR/MAVS signaling pathway immune response.

RESULTS
Identification of the full-length HHV-6B U26 transcript. In order to explore the role of U26 during HHV-6B infection, we first cloned the U26 open reading frame (ORF) according to the sequence deposited in the NCBI database (GeneID 1497028) from a cDNA library constructed by using HHV-6B-infected cells. To confirm the 39 and 59 ends of the U26 mRNA, we used 39 rapid amplification of cDNA ends (RACE) and 59-RACE analyses, respectively, according to the manufacturer's protocol. The results confirmed that the full-length U26 ORF was identical to that in the database. The U26 protein was predicted to have 8 transmembrane domains according to the SOSUI website (http://harrier.nagahama-i-bio.ac.jp/sosui/) ( Fig. 1A and B). We also analyzed the expression of U26 protein in HHV-6B-infected cells. An ;25-kDa protein was detected (Fig. 1C).
U26 is required for HHV-6B propagation. The expression kinetics of U26 during HHV-6B infection were analyzed by reverse transcription-quantitative PCR (RT-qPCR). U26 mRNA was detected at the early stage of HHV-6B infection (even at 8 h postinfection) and was dramatically increased following virus propagation ( Fig. 2A).
As there is still no effective gene knockout system for HHV-6B gene function analysis, we transduced HHV-6B-permissive cells (MT4 cells) with U26 short hairpin RNA (shRNA)-expressing lentivirus or its control lentivirus, followed by HHV-6B infection. RT-qPCR analyses confirmed the efficacy of U26 knockdown because U26 expression was significantly downregulated compared with the control shRNA-expressing cells (Fig. 2B). We also analyzed whether virus entry was affected by U26 knockdown by measuring incoming virus genomes in the infected cells at 2 h postinfection. The results showed similar entry efficiencies in U26 knockdown cells and control cells (Fig. 2C). Furthermore, we investigated virus propagation by measuring HHV-6B genome copies in the cell culture medium. We found that a retarded viral genome increases in U26 shRNA-expressing cells compared with that in control cells (Fig. 2D). We also detected viral protein expression in HHV-6B-infected cells and found that the expression of the immediate early 1 (IE1) protein and glycoprotein gB was reduced in U26 shRNA-expressing cells ( Fig. 2E and F). These results indicated that U26 plays an important role in HHV-6B infection.
A fraction of U26 is located in mitochondria. In order to study the U26 function, we first analyzed the predicted intracellular localization of HHV-6B U26 using tools available at https://ihg.gsf.de/ihg/mitoprot.html. We found that U26 had a probability score of 55% for its location in mitochondria, which was much higher than those of the other control proteins, HHV-6 gH, gL, and gB, known as viral envelope glycoproteins, and higher than that of the HCMV US9 protein reported to be located in mitochondria (28). GRP75, a mitochondrial marker, had the highest score, 96% (Fig. 3A). To confirm the prediction result, we performed confocal microscopy analysis and found that a portion of U26 was indeed enriched in mitochondria and overlapped a mitochondrial marker, Tom20, although much of U26 was expressed in the cytosol ( Fig. 3B and C). To further confirm its subcellular localization, we isolated cytosolic and mitochondrial fractions from U26-expressing HEK293T cells, analyzed each fraction with antibodies (Abs) for actin (cytosol) and Tom20 (mitochondria) by Western blotting, and found that pU26 was mainly detected in the mitochondrial fraction. Interestingly, two forms of U26 were expressed in U26-transfected cells (Fig. 3D, left) but not in HHV-6B-infected cells ( Fig. 1C and Fig. 3D, right), and the relatively large size of U26 was almost completely expressed in the mitochondrial fraction (Fig. 3D, left). How these two sizes of U26 formed still needs to be elucidated. These results indicated that pU26 is expressed and may function in mitochondria.
U26 downregulates MAVS expression. In previous studies, it has been reported that many viral proteins located in mitochondria affect MAVS expression, which in turn inhibits the RLR/MAVS signaling pathway and ultimately regulates host innate immunity (e.g., HCMV US9 and influenza A virus protein M2) (25,28). We examined whether U26 functions in a similar way during HHV-6B infection by transfecting HEK293T cells with the plasmid for either hemagglutinin (HA)-tagged MAVS or Myc-tagged STING expression alone or together with U26. We assessed MAVS and STING expression by immunoblot assays and found that U26 dramatically reduced MAVS expression but not HHV-6B U26 Blocks RLR/MAVS Signaling ® STING expression (Fig. 4A). To explore whether MAVS was an antiviral factor during HHV-6 infection, we knocked out MAVS in MT4 cells by using CRISPR-Cas9 technology and analyzed HHV-6 infection in these cells. MAVS knockout was confirmed by Western blotting (Fig. 4B, left). We infected MAVS knockout cells and control cells with HHV-6B and measured virus propagation by detecting viral genome replication and viral protein expression. We found that both virus genome replication and viral protein synthesis were significantly upregulated when MAVS was knocked out compared to those in control cells (Fig. 4B, right). We also measured incoming virus genomes in the infected cells at 2 h postinfection to analyze whether virus entry was affected by MAVS knockout. Our results showed that knockout of MAVS has little effect on virus entry compared with that in control cells (data not shown). To further analyze how U26 affected MAVS expression, we treated U26-expressing cells with either MG132 (a proteasome inhibitor) or chloroquine (a lysosome inhibitor) and found that MAVS expression could be rescued by MG132 treatment but not chloroquine treatment. These results indicated that MAVS reduction occurs mainly through the proteasome pathway rather than the lysosomal pathway (Fig. 4C). These results suggested that MAVS was ubiquitinated for degradation when it was coexpressed with HHV-6B U26. To further investigate U26-dependent ubiquitination of MAVS, we coexpressed MAVS (tagged with Flag), U26, and different types of ubiquitin (wild and K63 and K48 tagged with HA) and then prepared cell lysates for coimmunoprecipitation analysis with anti-Flag antibody. We analyzed the immunoprecipitants with anti-HA antibody and found that MAVS was mainly modified with K63-linked but not K48-linked ubiquitin (Fig. 4D). We also examined whether HHV-6B U26 expression affected endogenous MAVS expression by transfecting HEK293T cells with the U26 expression plasmid alone. We found that U26 inhibited MAVS expression in U26-expressing cells by Western blotting (Fig. 4E). Furthermore, when we used shRNA-expressing lentivirus to knock down U26 during HHV-6B infection, endogenous MAVS expression was significantly upregulated (Fig. 4F).
U26 inhibits the RLR/MAVS/IRF3 signaling pathway. Since U26 downregulated MAVS expression, we hypothesized that mitochondrion-located U26 could negatively regulate the MAVS-mediated signal pathway, thereby inhibiting the production of IFNs and other cytokines. To test this hypothesis, we expressed U26 in HEK293T cells and measured the expression of MAVS-mediated antiviral factors by RT-qPCR. We found that IFN-b and IL-6 cytokine expression was dramatically reduced (Fig. 5A, left). In HHV-6-infected cells, U26 knockdown also resulted in the upregulation of the expression of these two cytokines (Fig. 5A, right). These results indicate the HHV-6B U26 can significantly inhibit the activation of the MAVS-mediated signal pathway. There are two major signal pathways for MAVS to induce cytokine expression. One is the IKK/NF-κB pathway, and the other is the TBK1/IRF3 pathway (Fig. 5B). To further examine which signaling pathway was affected by U26, we analyzed the phosphorylation form of key molecules in these pathways by immunoblot assays. We transfected HEK293T cells with the plasmid for Flag-tagged MAVS expression alone or together with U26 and tested the phosphorylation form of IKKa, IKKb, and NF-κB in these cells. We found that U26 had minor effects on these proteins' phosphorylation, which meant that U26 might have a slight effect on this signaling pathway (Fig. 5C). However, when checking the other signaling pathway, we found that U26 dramatically reduced the expression and phosphorylation of TBK1 and IRF3 (Fig. 5D). To confirm U26 inhibition of the TBK1/ IRF3 signal pathway, we used U26 shRNA-expressing lentivirus to knock down U26 expression in MT4 cells during HHV-6B infection and then analyzed the phosphorylation of these proteins. The results showed that the phosphorylation of TBK1 and IRF3 was upregulated in U26 knockdown cells compared with that in control cells (Fig. 5E). These data indicated that U26 negatively regulated the RLR/MAVS signal cascade mainly through the TBK1/IRF3 pathway.
U26 affects mitochondrial membrane potential. Previous studies have reported that mitochondrial membrane potential (DWm) correlates with MAVS-mediated mitochondrial antiviral responses (36). Viral proteins that aggregate in mitochondria can fragment mitochondria. Mitochondrial fragmentation leads to DWm loss (25,37). The decrease of DWm further accelerates mitochondrial fragmentation and MAVS translocation (28,36). We hypothesized that U26 may affect DWm to leak MAVS from mitochondria and then suppress the host innate immune response. To verify this hypothesis, we investigated the mitochondrial membrane potential using a DWm-sensitive dye (tetramethylrhodamine ethyl ester [TMRE]) in U26-expressing cells and U26 knockdown cells infected with HHV-6B. The result showed that DWm was reduced in U26-expressing cells compared with the control cells (Fig. 6, top). In U26 knockdown cells, DWm obviously increased (Fig. 6, bottom). These data indicated that U26 was involved in DWm disruption.

DISCUSSION
As a mitochondrial membrane protein, MAVS functions as a critical adaptor for the RLR signaling pathway in which it links upstream recognition to downstream signal activation (38,39). Antiviral functions of RLR for many RNA viruses have been reported (40,41). Recently, several studies showed that RLR also functions during DNA virus infections (25,42,43). Targeting MAVS for viral immune evasion has also been reported during a set of virus infections (28,44,45). Here, we report a novel HHV-6B U26-targeted MAVS signaling pathway for inhibition of innate antiviral responses. We elucidated U26 function and analyzed the role of MAVS during HHV-6B infection.
We started our study from HHV-6B U26 functional analyses. The U26 gene is unique to the Roseolovirus genus with unknown function in the betaherpesvirus subfamily. Our kinetics analyses revealed that U26 was expressed at an early stage during HHV-6B infection (even at 8 h postinfection) ( Fig. 2A), indicating that it should be classified as an immediate early (IE) gene.
Herpesvirus IE genes could function either in nuclei as transcription-associated factors promoting viral gene expression or in the cytosol as disruptors of host antiviral responses during incoming virus propagation (46)(47)(48)(49). Our analysis revealed that U26 was expressed in the cytosol and mitochondria (Fig. 3). Several viral proteins expressed in mitochondria could affect the RNA-induced RLR/MAVS signaling pathway (50,51). We explored whether U26 functioned in a similar way in mitochondria. We found that U26 expression induced MAVS degradation by the proteasome pathway ( Fig. 4D and E). Furthermore, we also found that TBK1/IRF3 but not NF-κB activation was affected by U26 expression (Fig. 5).
As to how U26 induced MAVS degradation, we tested the possibility that U26 directly interacted with MAVS. However, the interaction was not confirmed (data not shown), thus offering other possible mechanisms. Several viral proteins jeopardize mitochondrion integrity when they are located in it, which results in the release of MAVS from mitochondria and the degradation of MAVS. For example, the influenza A virus protein PB1-F2 almost completely translocates into the mitochondrial inner membrane space via Tom40 channels, which accelerates mitochondrial fragmentation due to reduced mitochondrial inner membrane potential (37). The hepatitis C virus (HCV) NS3/4A protease cleaves MAVS rapidly, resulting in the release of MAVS from the mitochondrial membrane, destabilization, and loss of its interaction with RIG-I, which finally disables the antiviral signaling pathway (52). The KSHV tegument protein ORF33 interacts with MAVS, which results in the dephosphorylation of MAVS by a host factor, PPM1G, and leads to immunosuppression. The HCMV US9 protein accumulates in mitochondria, which could attenuate the mitochondrial membrane potential and lead to the promotion of MAVS leakage from the mitochondria. Our results showed that one of the mechanisms for U26-induced MAVS degradation is by affecting MAVS function by disrupting mitochondrial integrity (Fig. 6).
Activation of the RLR/MAVS signaling pathway needs the stimulation of pathogenassociated molecular patterns (PAMPs) or damage-associated molecular patterns (DAMPs) (40,53,54). However, the possible molecular mechanisms involving RLR/MAVS signaling pathway activation during HHV-6B still need further investigations. For DNA virus infections, several possible mechanisms exist. In some previous studies, it has been reported that RNA polymerase III (Pol III) can mediate double-stranded DNA (dsDNA) recognition by RNA intermediates, which in turn can be recognized by RIG-I and MAVS (42, HHV-6B U26 Blocks RLR/MAVS Signaling ® 43, 55). This is a potential mechanism by which RIG-I could respond to DNA virus infections, via RNA intermediates. During HSV-1 infection, 5S RNA pseudogene transcripts, particularly RNA5SP141, contribute to RIG-1 recognition and activation, possibly because of mislocation of these transcripts in the cytoplasm and downregulation of their binding partners in infected cells. In the case of lytic KSHV infection, 59-PPP-bearing vault RNAs accumulate because of the reduced level of DUSP11, which could remove two phosphates from the 59 end of vault RNAs. Moreover, it has been reported that the Epstein-Barr virus (EBV) genome has multiple regions encoding RNA molecules that function as potential virus ligands binding to RIG-I and stimulating RIG-I-dependent but RNA Pol III- independent IFN-b signaling (56). Further research is needed to determine what kind of RNA (cellular or viral) is recognized by RLR/MAVS in HHV-6B infection.
Many RNA and DNA viruses have developed mechanisms to perturb the host IFN response. Based on these findings, we suggest that HHV-6 U26 targets MAVS signaling to evade viral infection in host cells. The virus-host innate immune interaction represents a critical evolutionary battleground that shapes the outcome of viral infection. Therefore, an improved understanding of pU26 may facilitate the development of novel antiviral therapeutics for HHV-6 infection. The molecular mechanism of U26-influenced immune evasion is still unclear and needs to be investigated in future studies.
Antibodies. A RIG-I pathway antibody sampler kit (catalog number 8348) was purchased from Cell Signaling Technology. The antibody for Tom20 (catalog number AF1717) was purchased from Beyotime. Mouse monoclonal antibodies (MAbs) to Flag (catalog number F1804) and HA (catalog number H3663) tags were purchased from Sigma-Aldrich. To generate the antibodies against HHV-6B IE1and gB proteins, IE1and gB were purified from Escherichia coli transformed with an IE1 or gB expression plasmid as described previously (59,60). Mice were immunized with the IE1 protein, and the animal sera were used as the polyclonal antibody for IE1. Mice were immunized with gB protein, and MAbs for gB were produced. To generate the antibodies for U26, a plasmid (pGEX-4T-1U26NTD) for the expression of connected nontransmembrane domains in U26 was synthesized by General Biosystems (China), and the recombinant U26 protein was expressed and purified from Escherichia coli, which had been transformed with the plasmid pGEX-4T-1U26NTD. Mice were immunized with the U26 protein, and the animal sera were used as the polyclonal antibody for U26.
Western blotting. Samples were lysed with radioimmunoprecipitation assay (RIPA) buffer (Thermo Scientific), separated by SDS-PAGE, electrotransferred onto polyvinylidene difluoride (PVDF) membranes, and reacted with primary antibodies. Reactive bands were detected using a horseradish peroxidase-linked secondary conjugate (Santa Cruz) and visualized using enhanced chemiluminescence (ECL) reagents.