TLR8 Couples SOCS-1 and Restrains TLR7-Mediated Antiviral Immunity, Exacerbating West Nile Virus Infection in Mice Free

This article is featured in In This Issue, p.4181

West Nile virus (WNV) is a mosquito-borne ssRNA flavivirus that has caused significant morbidity and mortality in North America (1). In humans, WNV can cause a wide range of debilitating illnesses, from febrile-like illness to viral encephalitis, paralysis, and even death (2). However, the pathogenesis of WNV is still not clearly defined, and there is no approved WNV vaccine or specific antiviral therapeutic available for human use.

After mosquito inoculation, WNV initially infects skin Langerhans cells and macrophages and then further replicates in draining lymph nodes, spleen, and other peripheral organs, generating transient viremia. Prior to the development of specific humoral or T cell–mediated immune responses, WNV may enter the spinal cord and brain, leading to symptomatic neuronal dysfunction. Therefore, early control of WNV infection of neurons relies heavily on innate immunity. TLRs are a family of innate pattern recognition receptors (PRRs) that are located on plasma membranes or within endosomal membranes of host cells. WNV is reported to be recognized by TLR3 (3) and TLR7 (4), which play important roles during antiviral immunity by initiating a variety of cellular signal-transduction cascades, including the MyD88-dependent and -independent cascades, to control infection (3–8) and initiate the expression of IFN-stimulated genes (Isg) that can inhibit viral replication and transcription/translation of viral proteins through various antiviral mechanisms (9, 10). In contrast, TLR signaling can be compromised when single nucleotide polymorphisms are present in Tlr2, Tlr3, Tlr4, Tlr7, Tlr8, and Tlr9 genes that have been associated with defective antiviral immunity in HIV (11, 12), HSV type 2 (13), and Rift Valley fever virus (14) infections.

Human TLR8 can recognize viral ssRNA, but mouse TLR8 was described as nonfunctional (15–17), and its natural ligand remains unknown. This may be due to a deletion of 5 aa in the leucine-rich repeat ectodomain of TLR8 in mice, which is a region critical for recognition of viral ssRNA (18, 19). Although somewhat controversial (20), one study suggested that mouse TLR8 could recognize specific DNA motifs in vaccinia virus (21), whereas another report suggested that TLR8 recognizes a combination of imidazoquinoline and poly-T oligodeoxynucleotides (22); however, the natural ligand of mouse TLR8 remains elusive. Interestingly, overexpression of murine TLR8 does not activate IFN regulatory factor (IRF)-3 or IFN-α in HEK293T cells, suggesting that TLR8 may inhibit the type I IFN pathway (23). In addition, TLR8-knockout (Tlr8^−/−) mice develop lupus-like autoimmunity that is due to increased TLR7 function (24–26). Therefore, the function of TLR8 in mice is complicated, and its role during antiviral immunity needs to be investigated further.

Signal-transduction pathways that are triggered following cytokine and PRR engagement must be tightly regulated to prevent aberrant immune responses. Regulation is maintained through protein tyrosine phosphatases, protein inhibitors of activated STATs, and suppressor of cytokine signaling (SOCS) proteins (27, 28). Socs genes are expressed at relatively low levels in an inactivated state but are rapidly transcribed following TLR or cytokine engagement (28). For instance, Socs-1 and Socs-3 are induced following WNV infection, possibly acting as neuroprotective responses within the brain to regulate aberrant inflammation (29). SOCS-1 was shown to inhibit STAT-1 signaling of IFN-α (30), suggesting that SOCS-1 regulates type I IFNs. Antiviral immunity involves the secretion of type I IFNs, such as IFN-α and IFN-β, from viral-infected cells that act through a paracrine or autocrine mechanism, which engages the JAK/STAT signal-transduction pathway to induce a multitude of antiviral molecules that directly or indirectly inhibit viral infection (31). Isg-56 (ISG-56/IFIT-1) was shown to be induced by active STAT-1/2 and IRF-3/5/7/9 molecules following type I IFN signaling transduction (32), and it protects neurons from WNV infection (33, 34).

In this article, we report that TLR8 partners with SOCS-1 to control TLR7-mediated antiviral immunity in the CNS of mice during WNV infection.

All animal experimental procedures were reviewed and approved by the Institutional Animal Care and Use Committees at The University of Southern Mississippi (USM) and Yale University. All of the in vitro experiments and animal studies involving live WNV were performed by certified personnel in biosafety level 3 laboratories following standard biosafety protocols approved by USM and Yale University Institutional Biosafety Committees.

WNV isolate (CT2741) was kindly provided by Dr. John F. Anderson (Connecticut Agricultural Experiment Station). To prepare virus stocks, WNV was propagated and titered in Vero cells (ATCC CCL-81) by a plaque assay, as previously described (35). Tlr8^−/− mouse breeding pairs were provided by Dr. R.A. Flavell, and wild-type (WT; C57BL/6J) control mice were purchased from the Jackson Laboratory (Bar Harbor, ME). Seven-week-old WT and Tlr8^−/− mice were inoculated i.p. with 2000 PFU of WNV in 1% gelatin for survival analysis and tissue collection, according to previous publications (36–38).

Bone marrow–derived dendritic cells (BMDCs) were isolated from WT or Tlr8^−/− mice (3–6 mo old) and cultured as previously described (39). Briefly, mouse bone marrow cells were collected from femurs and grown in DMEM supplemented with 10% FBS, 2% plasmacytoma cell medium containing GM-CSF (J588L), 1% penicillin–streptomycin, 1% l-glutamine, and 50 μM 2-ME until maturation (11 d). Mature BMDCs were infected with WNV or stimulated with the TLR7 ligands CL264 and loxoribine (both from Invitrogen).

Murine primary mixed neuronal cultures were isolated from WT and Tlr8^−/− mice (6–12 mo old), as previously described with some modifications (40). Briefly, whole brains were isolated in ice-cold HEPES-buffered saline (HBS), minced, triturated in papain (2 mg/ml) in HBS, and incubated for 15 min at 37°C. Following incubation, cells were plated on polyornithine-pretreated plates for 20 min at 37°C, followed by a gentle wash with HBS to remove cellular debris. Cells were cultured in DMEM:F12 (1:1) medium (Thermo Fisher Scientific) supplemented with 1% penicillin–streptomycin, 10% FBS, 1% l-glutamine, and glucose (4.5 g/l). On day 11, supernatant was removed and replaced with Neurobasal-A Medium supplemented with 2% B-27 (both from Life Technologies), 10% FBS, 1% l-glutamine, and 1% penicillin–streptomycin. The mature neurons were infected with WNV (multiplicity of infection [MOI] = 1) for 24 or 48 h.

The murine Neuro-2a cell line (CCL-131) and the murine macrophage cell line RAW 264.7 (TIB-71) were purchased from the American Type Culture Collection and maintained in DMEM containing 10% FBS and 1% penicillin–streptomycin at 37°C with 5% CO₂.

TLR7 and TLR8 ligands CL264, CL075, loxoribine, and Poly(dT) (all from InvivoGen) were used at the indicated concentrations and lengths of time.

Bioactive type I IFN in culture supernatant was analyzed by a previously described method (41) that measured protection against encephalomyocarditis virus (EMCV) in a susceptible cell line (L929; American Type Culture Collection) that was pretreated with culture supernatant containing IFN. Briefly, culture supernatant collected from WT and Tlr8^−/− mice BMDCs that were infected in vitro with WNV for 24 h (MOI = 5) were inactivated with UV light (10 min at 120 mJ/s). The crude, UV-inactivated supernatant was added to monolayers of L929 cells (cultured in DMEM supplemented with 10% FBS and 1% penicillin–streptomycin) in 96-well flat-bottom plates. Following incubation for 14 h at 37°C, medium was removed, and cells were infected with EMCV (MOI = 10) for 7 h. ECMV-mediated cell death was measured using a CellTiter 96 Aqueous cell proliferation assay kit (Promega) and an ELx808 Ultra Microplate Reader (BioTek Instruments). The percentage of protected cells was calculated as described (41), according to the following formula: (OD at 492 nm [OD₄₉₂] of supernatant-treated EMCV-infected cells/OD₄₉₂ of non–EMCV-infected cells × OD₄₉₂ of EMCV-infected cells)/(OD₄₉₂ of non–EMCV-infected cells) × 100%.

Mouse tissues, blood, or cultured cells were collected for total RNA extraction with TRI Reagent (Molecular Research Center) and converted into first-strand cDNA using the iScript cDNA Synthesis Kit (Bio-Rad). Quantitative PCR (qPCR) assays were performed using iTaq Universal Probes Supermix for probe-based assays or iQ SYBR Green Supermix polymerase (both from Bio-Rad). WNV-envelope (WNV-E) gene and mouse gene primer and probe sequences were adapted from previous publications: WNV-E (4), β-actin (42), Tlr7 (24), Irf-7 (43), Ifn-α (43), Isg-56 (43), Isg-54 (44), Isg-49 (44), Ifn-β (43), and Socs-1 (45). Primers were designed for murine Bcl2-associated X protein (Bax): forward 5′-TGCTAGCAAACTGGTGCTCA-3′ and reverse 5′-TAGGAGAGGAGGCCTTCCCAG-3′. Data are presented as relative fold change by the 2^−ΔΔCT method, using β-actin as a housekeeping gene, or as a ratio of target gene/β-actin copy numbers. All of the primers and probes were synthesized by Integrated DNA Technologies or Applied Biosystems.

Small interfering RNAs (siRNAs) were designed, using the Thermo Fisher Scientific siRNA designing tool (siDESIGN Center), targeting murine Isg-56 (10 nM, 5′-GUAAGUAGCCAGAGGAAGGUGAUGCUU-3′) or a scrambled sequence (5′-ACUACUUCAGGUGUGAGCUAAUAUACC-3′) and were transfected with Lipofectamine RNAiMAX reagent into Neuro-2a cells in Opti-MEM (both from Life Technologies) for 20 min. DMEM containing 2% FBS was added, and cells were cultured for 24 h. Following incubation, media were removed, and cells were infected with WNV (MOI = 5) for 48 h. Then cells were collected for qPCR and flow cytometric analyses.

Murine siRNAs targeting murine Socs-1 (Santa Cruz Biotechnology) were transfected into RAW 264.7 cells (6 × 10⁵ cells per milliliter) following the manufacturer’s recommendation, with some minor changes. Briefly, lipoplexes were prepared in Opti-MEM (Life Technologies) by mixing siRNA (25 nM) and transfection reagent (Santa Cruz Biotechnology) for 30 min. Cells and lipoplexes were mixed in 12-well plates and incubated in Opti-MEM for 24 h. DMEM containing 2% FBS was added to the cells, followed by infection with WNV (MOI = 0.1), and cells were cultured for an additional 24 h. Following infection, cells were collected and prepared for qPCR analysis.

Murine BMDCs were isolated, plated at 3 × 10⁵ cells per well, and infected with WNV at day 11. Infected cells were fixed with 4% paraformaldehyde (PFA) in PBS for 15 min at room temperature (RT). Cells were washed with PBS, blocked with 2% normal goat serum (Life Technologies) containing 0.4% Triton-X for 1 h at RT, and probed with monoclonal mouse anti-flavivirus glycoprotein E IgG Ab (4G2, ATCC D1-4G2-4-15 HB-112) overnight at 4°C. Then cells were washed with PBS and probed with goat polyclonal anti-mouse-HRP IgG (KPL) for 2 h at RT. Immunopositive cells were developed with TrueBlue peroxidase substrate (KPL). Images were taken using an Axiostar plus light microscope (Zeiss), and mean pixel intensity was quantified using ImageJ (version 1.48), as previously described (46).

Similarly, primary neurons were isolated from mice, infected with WNV, and fixed as described above. After a 1-h blocking step at RT with 2% normal goat serum containing 0.4% Triton-X, neurons were probed with mouse monoclonal anti–WNV-E (1:50; Abcam) and rabbit polyclonal anti–ISG-56 (1:100) Abs overnight at 4°C. Then cells were washed with PBS and probed with polyclonal goat Anti-Mouse IgC FITC (eBioscience) and polyclonal goat anti-rabbit DyLight 594 IgG (Thermo Fisher Scientific) for 2 h at RT. Cells were washed with PBS, mounted using VECTASHIELD containing DAPI, and imaged as above.

For brain immunohistochemistry, WNV-infected WT and Tlr8^−/− mice were euthanized (day 6 postinfection [p.i.]) and perfused with ice-cold PBS. Half brain was fixed overnight in 4% PFA, followed by frozen tissue cyroprotection with daily changes in 10, 20, and 30% sucrose in PBS. Brain tissues were frozen in Tissue-Plus O.C.T. buffer (Fisher Healthcare), and midsagittal sections (10 μm) were cut using a Tissue-Tek Cryo3 microtome/cryostat (Sakura) and mounted on precleaned Superfrost Plus microscope slides (Thermo Fisher Scientific). Apoptotic measurement of brain tissue was detected by a TACS 2 TdT Fluoroscein in situ apoptosis detection kit (Trevigen), following the manufacturer’s recommendations, and images were acquired using a confocal LSM 510 microscope (Zeiss).

Mouse whole brains were prepared with lysis buffer containing 50 mM Tris-HCl, 150 mM NaCl, 0.25% NaDodSO₄ (SDS), 0.25% sodium deoxycholate, 1 mM EDTA, and 1% proteinase inhibitor mixture (P8340; Sigma). Protein concentration in lysates was quantified by a Bradford Assay (Bio-Rad), and cell lysates were mixed with 2× Laemmli buffer (Bio-Rad) containing 0.1% 2-ME (Sigma), boiled at 95°C for 5 min, and rapidly spun down. Whole-cell protein lysates (5–10 μg per well) were separated by PAGE and transferred to a nitrocellulose membrane (Bio-Rad). The membrane was blocked with 5% BSA for 1 h and probed with rabbit polyclonal anti-TLR7 (Cell Signaling Technology), rabbit polyclonal anti–ISG-56 (Pierce Antibodies), rabbit monoclonal anti–STAT-1 (Cell Signaling Technology), and rabbit monoclonal anti–IRF-7 (Abcam) Abs in 5% BSA overnight at 4°C. The immunolabeled membrane was probed with secondary HRP-conjugated Goat Anti-Rabbit IgG (Jackson ImmunoResearch Laboratories) for 2 h. β-Tubulin–HRP (Cell Signaling Technology) was used as a loading control. Membranes were developed with SuperSignal West Pico Chemiluminescent Substrate (Thermo Fisher Scientific), and images were acquired using a ChemiDoc XRS⁺ System (Bio-Rad).

For immunoprecipitation, Neuro-2a cell lysates were prepared as described above, mixed with rabbit polyclonal anti-TLR7 (1:50; Cell Signaling Technology) or rabbit polyclonal anti-TLR8 (1:50; Sigma) Abs for 2 h at RT, washed in 1× TBS containing 0.05% Tween 20 and 0.5 M NaCl, and mixed with Dynabeads Protein G (Life Technologies) for an additional hour. Samples were washed and resuspended in 2× Laemmli buffer containing 0.1% 2-ME, boiled at 95°C for 5 min, and rapidly spun down. Proteins from whole-cell lysates were separated by PAGE and transferred to a nitrocellulose membrane. Membranes were probed with a rabbit polyclonal anti–SOCS-1 Ab (1:1000; Sigma) in 5% BSA overnight at 4°C, followed by probing with a secondary peroxidase-conjugated goat anti-rabbit IgG (Jackson ImmunoResearch Laboratories) for 2 h, and developed as above. Loading control input bands were detected following back incubation with the rabbit polyclonal anti-TLR8 Ab (Sigma).

Blood samples were collected in EDTA-coated tubes from WNV-infected WT and Tlr8^−/− mice by retro-orbital bleeding, and RBCs were lysed by adding RBC lysis buffer (Sigma). Blood cells were washed two times to remove lysed RBCs and resuspended in flow cytometry buffer (PBS +2% FBS) at 5 × 10⁵ cells per milliliter. Cells were probed overnight at 4°C with rabbit polyclonal anti-TLR7 Ab (Cell Signaling Technology), washed two times with flow buffer, and probed with secondary goat anti-rabbit DyLight 594 IgG (Thermo Fisher Scientific) for 2 h at RT. Cells were washed two times and analyzed in a BD LSR Fortessa flow cytometer, and data were acquired using BD FACSDIVA version 7.0 (both from BD Biosciences). Cells probed only with secondary Ab were used as controls for fluorescence gating.

WT and Tlr8^−/− BMDCs were infected with WNV (MOI = 5) for 24 h, as described above. Cells were fixed with 2% PFA and permeabilized with PBS +0.05% Tween-20 (permeabilization buffer). Cells were incubated overnight with mouse monoclonal anti–WNV-E IgG2b (1:100; Abcam), rat monoclonal anti–IFN-α IgG1 (1:100; Abcam), or rabbit polyclonal anti-TLR7 (1:100; Cell Signaling Technology) Abs, followed by two washes with permeabilization buffer, and probed with a secondary goat anti-mouse IgG Ab conjugated to FITC (Santa Cruz Biotechnology), goat anti-rat IgG Ab conjugated to Alexa Fluor 555 (Molecular Probes), or goat anti-rabbit IgG Ab conjugated to DyLight 594 (Thermo Fisher Scientific), respectively, for 2 h at RT in dark. For WNV Ag detection, cells were washed two times with permeabilization buffer and resuspended in DAPI (100 μM) for 10 min, and FITC mean fluorescent intensity (MFI) was analyzed with a flow cytometer (BD LSR Fortessa). DAPI expression was analyzed with a microplate reader (BioTek Synergy H1) using Gen5 (version 2.07) software to determine the ratio of FITC⁺ MFI normalized to DAPI⁺ cells.

UV-inactivated supernatant collected from WT and Tlr8^−/− BMDCs was incubated with rat monoclonal anti–IFN-α Ab (1:50) overnight at 4°C on a rocker to form immune complexes. In a separate tube, Dynabeads Protein G (Life Technologies) were washed in buffer (TBS + 0.05% Tween-20, 0.5 M NaCl) and added to the supernatant containing IFN-α Abs for 1 h at RT. The beads were magnetically separated from the remaining supernatant and washed, and secondary anti-rat Ab conjugated to Alexa Fluor 555 (1:1000) was added. After incubation for 2 h at RT, the beads were detached from the immune complexes with an elution buffer (0.1 M glycine [pH 2]) and neutralized in Tris-HCl (pH 9) buffer. The MFI of the beads was analyzed by flow cytometry, as above.

Apoptosis was measured using a modified annexin V and propidium iodide (PI) apoptosis assay (47). Briefly, Neuro-2a cells were collected and resuspended in annexin V binding buffer, followed by staining with annexin V conjugated to Alexa Fluor 488 (Molecular Probes) and PI (Sigma). Cells were washed with annexin V binding buffer, fixed in 1% PFA for 10 min on ice, incubated in RNase A (50 μg/ml; Sigma) for 15 min at 37°C, washed twice in PBS, and analyzed by flow cytometry.

Data were compared with a Student t test or two-way ANOVA with Bonferroni post hoc analysis. Survival curves were analyzed using Kaplan–Meier analysis. All statistical analyses were performed using GraphPad Prism software (version 6.0).

Our previous report demonstrated that TLR7 signaling protects mice from lethal WNV infection (4) and Tlr7 expression is upregulated in TLR8-deficient (Tlr8^−/−) mice (24). To investigate the potential role of TLR8 during WNV infection in mice, we challenged Tlr8^−/− and WT mice i.p. with 2000 PFU of WNV and monitored mice twice daily for morbidity and mortality for up to 21 d. The Kaplan–Meier survival analysis shows that 52% of Tlr8^−/− mice and 26% of WT mice survived lethal WNV infection (Fig. 1A), indicating that TLR8 signaling facilitates WNV infection in mice. To further confirm this observation, we measured viral burden in the blood and brains of WNV-infected mice by qPCR. Consistent with the survival data, the qPCR results show lower expression of WNV-E transcripts in blood of Tlr8^−/− mice compared with WT controls at day 4 and in the brain at days 4 and 6 p.i. (Fig. 1B–D). These results indicate that TLR8 signaling in mice plays a negative role in WNV immunity by facilitating WNV replication. Type I IFNs play essential roles in viral clearance and control of WNV burden in mice (48, 49). Therefore, we assessed the expression of type I IFNs in blood samples collected from WNV-infected Tlr8^−/− and WT mice at days 1–4 p.i. by qPCR. The results showed that Tlr8^−/− mice have increased expression of Ifn-α at days 1–4 and Ifn-β at days 1 and 2 compared with WT controls following WNV infection (Fig. 1E, 1F). In addition, Irf-7 and Isg-56 measured in blood of WNV-infected Tlr8^−/− mice showed an increasing trend in gene expression (Fig. 1G, 1H). In summary, these results indicate that TLR8 facilitates WNV infection in mice possibly as a result of the downregulation of type I IFN–dependent antiviral responses.

Because naive Tlr8^−/− mice express higher levels of Tlr7 (24), we measured the expression of TLR7 in blood leukocytes collected from WNV-infected WT and Tlr8^−/− mice by flow cytometry at day 1 post-WNV infection. The results revealed an increased expression of TLR7 in leukocytes of WNV-infected Tlr8^−/− mice compared with WT controls (Fig. 2A). To further investigate the potential role of TLR8 in regulating the expression of Tlr7 during WNV infection, we infected BMDCs from Tlr8^−/− and WT mice with WNV (MOI = 5) in vitro and measured the expression of Tlr7 by qPCR. In line with the in vivo results, the qPCR results showed a higher basal expression of Tlr7 in BMDCs from Tlr8^−/− mice compared with WT controls, which was further amplified following WNV infection, indicating overexpressed Tlr7 in the absence of TLR8 (Fig. 2B). Consistent with this, we also used flow cytometry to confirm that TLR7 expression on BMDCs from Tlr8^−/− mice was further increased at the protein level following WNV infection (Supplemental Fig. 1A). To assess the antiviral function of Tlr7 overexpression, we stimulated BMDCs from Tlr8^−/− and WT mice with the TLR7-specific ligand CL264 (5 μg/ml) and measured the expression of type I IFN by qPCR. We found higher levels of Ifn-α (Fig. 2C) and Ifn-β (Supplemental Fig. 1B) expression in BMDCs generated from Tlr8^−/− mice compared with WT controls at multiple time points, suggesting that overexpression of TLR7 in Tlr8^−/− mice may induce stronger antiviral immunity. To confirm this, we infected BMDCs from Tlr8^−/− and WT mice with WNV (MOI = 5) or an alphavirus, Chikungunya virus (CHIKV; MOI = 5), and assessed the transcript levels of Ifn-α and Ifn-β by qPCR at 24 h p.i. The results show higher expression of Ifn-α (Fig. 2D) and Ifn-β (Supplemental Fig. 1C) in BMDCs generated from Tlr8^−/− mice than in those from WT controls in response to both type of infections (WNV or CHIKV), suggesting a response that is not specific to WNV. In addition, intracellular and secreted IFN-α also were increased in Tlr8^−/− BMDCs compared with WT controls following WNV infection in a flow cytometric assay (Fig. 2E, 2F). Moreover, we measured gene expression of signaling proteins involved in the type I IFN response and found that WNV infection significantly induced the expression of Irf-7 and Isg-56 in Tlr8^−/− BMDCs compared with WT controls at 24 h p.i. (Fig. 2G, 2H). To further confirm higher expression of type I IFN in the absence of TLR8, we measured bioactive type I IFN in the cell culture supernatant of WNV-infected Tlr8^−/− and WT BMDCs, as previously described (41). The IFN bioassay further confirmed that BMDCs generated from Tlr8^−/− mice produced more IFN than did those from WT controls during WNV infection (Fig. 2I). Because BMDCs generated from Tlr8^−/− mice have increased antiviral immunity, we expected these cells to be more efficient in controlling WNV infection. Indeed, we found that BMDCs from Tlr8^−/− mice are more resistant to WNV infection in vitro compared with those from WT mice, as measured by reduced WNV Ag expression (Fig. 2J, 2K, Supplemental Fig. 1D, 1E). In summary, these results show that TLR8 signaling in mice downregulates the expression of Tlr7, which results in a reduced antiviral immune response against WNV infection.

WNV can cause severe CNS infection and can lead to injury and death of neurons in mice and humans. Because we observed reduced viral burden in the brains of WNV-infected Tlr8^−/− mice and increased expression of Tlr7 and other antiviral genes in Tlr8^−/− BMDCs infected in vitro with WNV, we asked whether TLR8 also regulates antiviral immunity in the brains of WNV-infected mice. On day 4 post-WNV infection (2000 PFU/mouse, i.p.), Tlr8^−/− and WT mice were euthanized and perfused with ice-cold PBS, and whole brains were collected for qPCR and Western blotting analyses of TLR7 and other antiviral molecules. Consistent with the results from BMDCs, uninfected Tlr8^−/− mice exhibited increased basal expression of Tlr7 in brains compared with WT controls, which was further increased following WNV infection (Fig. 3A). These results suggest that increased Tlr7 expression in brains of Tlr8^−/− mice may result in a better ability to recognize and respond to neuroinvasive WNV. To test this further, we used qPCR to measure the expression of Irf-7, Ifn-α, and Ifn-β in brain tissues collected from WNV-infected Tlr8^−/− and WT mice. The results revealed slightly higher expression of Irf-7 (Fig. 3B) and a trend toward increased expression of Ifn-α (Fig. 3C), whereas the expression of Ifn-β remained unaltered (data not shown). Because type I IFNs were not robustly increased in the brains of Tlr8^−/− mice, but viral burden was significantly reduced (Fig. 1C, 1D), we sought to examine whether Tlr8^−/− mice overexpress other antiviral genes. To test this, we examined the expression of Isg-56, Isg-54, and Isg-49. Interestingly, the expression of Isg-56 was increased (Fig. 3D) in Tlr8^−/− mice brains following WNV infection, whereas the expression of Isg-54 and Isg-49 was not significantly altered (data not shown). In addition, we confirmed higher expression of TLR7, ISG-56, IRF-7, and STAT-1 in brain lysates of WNV-infected Tlr8^−/− mice at the protein level by an immunoblotting assay (Fig. 3E). Together, these results confirm that Tlr8^−/− mice express higher levels of TLR7 in brain tissue, which may lead to strong antiviral responses via increased expression of ISG-56, IRF-7, and STAT-1.

We next analyzed the expression of Tlr7 and antiviral genes in primary mixed neuronal cultures isolated from WNV-infected (MOI = 1, 24 h) WT and Tlr8^−/− mice whole brain homogenates in vitro. The qPCR analysis indicated that Tlr7, Irf-7, and Ifn-α were all increased following WNV infection in Tlr8^−/− neurons compared with WT controls (Fig. 3F–H), suggesting that the TLR7-mediated IFN signaling pathway may increase neuronal resistance against WNV infection in Tlr8^−/− mice. We further assessed whether the expression of ISGs was also altered in WNV-infected Tlr8^−/− neurons by qPCR. Consistent with in vivo brain tissue results (Fig. 3D), we found increased expression of Isg-56 (Fig. 3I), along with increased Isg-54 and Isg-49, in the Tlr8^−/− neurons infected in vitro with WNV compared with WT controls (data not shown). In line with increased antiviral immunity, immunofluorescence microscopy showed reduced WNV Ag and increased ISG-56 expression in WNV-infected Tlr8^−/− neurons compared with WT controls (Fig. 3J, 3K). In summary, these results indicate that TLR8 inhibits antiviral immunity against WNV infection, leading to enhance viral replication within the CNS.

One of the major cell death pathways that occurs in WNV neuroinvasive disease is apoptosis, in particular via the induction of the apoptotic-mediated gene Bax (50, 51). To examine the degree of WNV-mediated cellular apoptosis in Tlr8^−/− and WT brain tissue, whole brains were isolated at day 6 p.i. and analyzed by a TUNEL assay. It showed that Tlr8^−/− mice had significantly reduced cell death of WNV-permissive Purkinje neurons of the cerebellum (52–57) compared with WT controls (Fig. 4A), with no observable differences in any other regions of the brain. Interestingly, we found that the gene expression of Bax (Fig. 4B), along with that of antiviral genes Isg-56 and Socs-1, was significantly altered only in the spinal cords and cerebellar regions of Tlr8^−/− mice at day 4 p.i. (data not shown), suggesting that antiviral immunity confined viral-induced apoptosis to these regions of the brain. In line with this, previous reports suggested that the expression (56) and neuroprotective role of ISG-56 indeed localize to these specific region of the CNS (34). WT and Tlr8^−/− neurons infected with WNV (MOI = 1) for 24 h in vitro were also analyzed for Bax expression by qPCR, indicating increased neuron survival in Tlr8^−/− mice following WNV infection (Fig. 4C). Collectively, these results indicate that TLR8 signaling counteracts antiviral immunity, possibly by downregulation of Isg-56 expression, favoring WNV-induced neuronal death.

Because higher expression of ISG-56 in brains of Tlr8^−/− mice following WNV infection is associated with reduced Bax expression, we tested whether the Bax-mediated apoptosis is dependent on Isg-56 expression. For this, we used siRNAs to knockdown Isg-56 in mouse Neuro-2a cells and assessed Bax expression and apoptosis following WNV infection in vitro. Isg-56–specific siRNAs were transfected into Neuro-2a cells for 24 h, and the cells were infected with WNV (MOI = 5) for an additional 48 h. The efficiency of siRNA knockdown of Isg-56 expression was confirmed by qPCR in WNV-infected Neuro-2a cells (Fig. 4D). Interestingly, knockdown of Isg-56 leads to increased replication of WNV (Fig. 4E) and expression of Bax (Fig. 4F), suggesting that WNV replication may be inhibited by ISG-56 expression. To test whether apoptosis is the direct consequence of increased Bax expression in siRNA–Isg-56–transfected Neuro-2a cells, we stained WNV-infected Neuro-2a cells with annexin V and measured apoptosis by flow cytometry. The results confirmed that knocking down Isg-56 expression increased apoptosis of WNV-infected Neuro-2a cells (Fig. 4G, 4H). Collectively, these data indicate that ISG-56 is an essential antiviral molecule that controls WNV-induced neuronal apoptosis.

Socs-1 was shown to be induced by WNV (29), and its expression inhibits antiviral responses, such as type I IFNs (58). We found that the expression of Socs-1 was reduced significantly in brain tissue and neurons of Tlr8^−/− mice following WNV infection (Fig. 5A, 5B), suggesting that TLR8 regulates Socs-1 expression during WNV infection. To test whether TLR8 directly regulates SOCS-1 function, we performed an immunoprecipitation assay in Neuro-2a cells in the presence of various TLR7 and suspected mouse TLR8 ligands (22). Interestingly, we found that SOCS-1 coprecipitates with TLR8, but not with TLR7 (Fig. 5C), even in the presence of TLR7 and TLR8 ligation. These results suggest that TLR8 may directly control SOCS-1 function within neurons in mice. To test whether Isg-56 is regulated by Socs-1, we transfected a mouse macrophage cell line (RAW 264.7) with siRNA targeting Socs-1 for 24 h, infected these cells for an additional 24 h with WNV (MOI = 0.1), and used qPCR to analyze the results. We found that Socs-1 was slightly reduced following siRNA transfection in WNV-infected mouse RAW 264.7 cells (Fig. 5D), with increased Isg-56 (Fig. 5E) and Tlr7 (Fig. 5F) expression. These results indicate that SOCS-1 signaling downregulates Tlr7 and Isg-56 expression in mice. Because SOCS-1 associates with TLR8 directly (Fig. 5C) in mice, SOCS-1 may couple with TLR8 to inhibit the expression of Tlr7 and, subsequently, its downstream signaling molecules, such as Isg-56, facilitating WNV infection in mice (Fig. 6).

Human and mouse Tlr7 and Tlr8 share a high degree of structural and phylogenetic similarity and are located only 70 kb apart on the same X chromosome (17, 24). Functionally, human and mouse TLR7 and human TLR8 recognize viral ssRNA motifs. However, mouse TLR8 does not recognize viral ssRNA, which led to the belief that TLR8 may be nonfunctional in mouse in terms of sensing viral ssRNAs (16). It was since suggested that mouse TLR8 recognizes DNA motifs of vaccinia virus (21); however, this was contradicted by another study (20). It also was suggested that mouse TLR8 can recognize a combination of imidazoquinoline and poly-T oligodeoxynucleotides (22); however, the natural ligand for mouse TLR8 remains unknown. We (24, 25) and other investigators (26, 59) demonstrated that mice deficient in TLR8 (Tlr8^−/−) overexpress Tlr7 and manifest systemic autoimmunity as a result of the development of a subset of B cells that produce anti-small nuclear ribonucleoproteins, anti-ribonucleoproteins, and anti-DNA Abs. Furthermore, transgenic mice that overexpress human TLR8 have reduced expression of mouse Tlr7 (15). Based on these reports, it can be speculated that TLR8 in mice might play multiple roles in the regulation of immunity other than serving as a PRR. We reported that TLR7 plays an important role in the recognition and mitigation of WNV infection in mice (4); however, the role of mouse TLR8 during WNV pathogenesis was not previously studied.

Defective Tlr7 and type I Ifn genes are contributing factors to WNV susceptibility, as well as to enhanced WNV-induced disease (4, 60, 61). In the current study, we found that TLR8 deficiency in mice induced a strong antiviral immune response that facilitated efficient control of viral burden and increased survival of mice after lethal WNV challenge. In line with previous reports that showed that TLR8 negatively regulated TLR7 expression in mouse dendritic cells (24, 25), Tlr8^−/− mice had increased gene expression of Tlr7 in multiple tissues compared with WT controls, which was further amplified following WNV infection. Although the mechanism by which increased Tlr7 expression in the absence of viral infection was not explored further in this study, it has been suggested that deficiency in TLR signaling could be compensated for by other PRRs (62). As a result of their close loci proximity and redundancy in pattern recognition (17, 63), TLR7 signaling may overcompensate for a loss of TLR8, as the reciprocal was identified in TLR7 deficiency (64). Yet, in the context of viral infection, it may be possible that TLR8 directly or indirectly represses Tlr7 transcription in mice. Based on the evidence that TLR8 coprecipitates with SOCS-1, and knockdown of Socs-1 by siRNA leads to increased Tlr7 transcription in WNV-infected RAW 264.7 cells, we may conclude that TLR8 couples to SOCS-1 to suppress the expression of Tlr7 and its downstream signaling during WNV infection in mice. However, we cannot exclude the possibility that expression of Tlr7 is also indirectly enhanced by type I IFN responses induced by other PRRs, such as RIG-I and MDA5, in response to WNV infection (9). Although we only measured inactive IRF-7 expression, active phospho–IRF-7 was described to act as a master transcription factor for IFN-α (65, 66), which regulates the induction of Isg-56 and Tlr7 and, thus, requires further study (67, 68). Furthermore, there was an overall increase in the expression levels of TLR7, IRF-7, and ISG-56 in Tlr8^−/− mice, in particular within primary neurons; therefore, involvement of multiple regulatory pathways may contribute to hyperactive antiviral immunity to control viral infection. In line with this, it was described that functional polymorphisms in the human Tlr8 gene may predispose individuals to differential viral susceptibility (14, 69–72). For instance, TLR8C-A haplotypes are associated with dengue fever susceptibility, whereas TLR8G-G haplotypes are associated with dengue hemorrhagic fever susceptibility (71). Additionally, truncated Tlr8 protects against HIV, because rapid decay of Tlr8 mRNA results in induced TNF-α signaling (70). Therefore, TLR8 may suppress antiviral immunity through negative regulation of TLR7 signaling.

ISG-56 (IFIT-1), along with other IFIT family members, including ISG-54 (IFIT-2) and ISG-49 (IFIT-3), protects neurons from WNV infection (33, 34, 73). These molecules are enhanced by type I IFNs during viral infection (9, 56) and function by binding to the 5′-PPP end of viral RNA, mRNA, or the translation initiation factor eIF-3 to inhibit the initiation of viral/host protein translation (33, 67, 74). WNV-infected Tlr8^−/− mice expressed higher levels of type I IFNs and Isg-56 in various tissues compared with WT controls. Moreover, increased expression of Isg-54 and Isg-49 was also observed in WNV-infected Tlr8^−/− mice neurons, suggesting a vast array of enhanced antiviral immunity. Consistent with the previous report (75) that IFITs are vital antiviral proteins that work in concert to control viral infection within the CNS, we further verified the importance of Isg-56 during WNV infection in Tlr8^−/− mice by knocking down its expression. This resulted in increased WNV-induced neuronal apoptosis mediated by Bax, highlighting an important role for ISG-56 during WNV-induced neuronal death.

Tissue and cellular tropism of WNV is regulated by antiviral gene localization (56, 76), and IFIT molecules play a crucial role in controlling viral spread within neurons (33). Additionally, the cerebellum and choroid plexus express Isg-56 following WNV infection (33), and IFN-β treatment induces a higher expression of Isg-56 in cerebellar cells (56). In line with this, we observed a significant increase in Isg-56 only within the spinal cord and cerebellar regions of Tlr8^−/− mice brains, together with reduced morbidity, suggesting that neuropathogenesis of WNV may be limited by antiviral control to these regions of the CNS. Overexpression of Tlr7 in the CNS was linked to the induction of IL-6–dependent dendrite retraction (64), and it leads to cell death by induction of a CNS-specific TIR adaptor protein called sterile alpha and HEAT/Armadillo motif (SARM) (77). Yet, in the presence of WNV infection, SARM is necessary to reduce WNV burden and prevent cell death, because Sarm^−/− mice are susceptible to WNV infection (78). In addition, the TLR7 agonists imiquimod and ssRNA were described to induce neuronal cell death (79). However, no phenotypic or behavioral defects are observed in naive Tlr8^−/− mice, and the increased percentage of survival and reduced TUNEL signals observed within Tlr8^−/− mice brains following lethal WNV challenge suggest that increased TLR7 does not cause any apparent neuronal damage in our experimental conditions. Furthermore, the expression of TLR7 in the brains of Tlr8^−/− mice is relatively mild, which suggests that only dramatically increased expression/signaling by TLR7 may lead to neuronal damage.

SOCS-1, a negative regulator of IFN signaling, can be induced following viral challenge (29, 80–82). SOCS-1 suppresses a multitude of signaling molecules, such as the Mal adaptor protein in TLR4 signaling (83), IRAK-4 (84), TRAF6 (85), type I IFNs (30, 58), JAKs (86), and transcriptional promoters, including STAT-1 (58, 87), IRF7 (88), and p53 (89). Importantly, Socs-1 knockdown in Japanese encephalitis virus–infected macrophages resulted in reduced viral load and increased ISGs, suggesting that SOCS-1 may directly regulate expression of ISGs (90). Because Tlr8^−/− mice had a hyperactive IFN response and reduced Socs-1, we hypothesized that TLR8 signaling may regulate Socs-1 expression. It was reported that TLR7/8 signaling induces the expression of Socs-1 in HEK293 cells, yet these studies did not determine whether this signaling is dependent on TLR7 or TLR8 alone or is an effect of combined TLR7 and TLR8 ligation (91). We demonstrated that SOCS-1 partnered with TLR8, but not TLR7, suggesting that SOCS-1 uses TLR8 as an adaptor molecule for its regulation. This interaction is possible because there are 13 phospho-tyrosine residues located on the cytoplasmic domain of TLR8, but not on TLR7 (92), which provides ample docking sites for the SH2 regions of SOCS-1. SOCS proteins directly associate with and inhibit adaptor proteins and their signaling pathways via protein degradation (83). However, we observed a stable interaction between TLR8 and SOCS-1 24 h poststimulation in Neuro-2a cells, suggesting a noncanonical role for TLR8 SOCS-1 regulation that requires further investigation.

In conclusion, TLR8 couples with SOCS-1 to exacerbate neuroinvasive WNV infection in mice, by negative regulation of TLR7-mediated antiviral immunity. The identification of a novel role for SOCS-1 in TLR8-mediated immune regulation may have broad implications in understanding antiviral immunity in mice and humans. Although we did not assess human TLR8 regulation of TLR7 with SOCS-1, the therapeutic potential of inhibiting the TLR8 pathway may provide an alternative antiviral strategy to combat WNV infection in humans that requires further investigation.

We thank Dr. John F. Anderson for providing the WNV isolate (CT2741), the USM animal facility for excellent care of the mice, and Mississippi INBRE IDeA Network of Biomedical Research Excellence for the use of the facility’s equipment.

The authors have no financial conflicts of interest.