Viperin inhibits rabies virus replication via reduced cholesterol and sphingomyelin and is regulated upstream by TLR4

Viperin (virus inhibitory protein, endoplasmic reticulum-associated, IFN-inducible) is an interferon-inducible protein that mediates antiviral activity. Generally, rabies virus (RABV) multiplies extremely well in susceptible cells, leading to high virus titres. In this study, we found that viperin was significantly up-regulated in macrophage RAW264.7 cells but not in NA, BHK-21 or BSR cells. Transient viperin overexpression in BSR cells and stable expression in BHK-21 cells could inhibit RABV replication, including both attenuated and street RABV. Furthermore, the inhibitory function of viperin was related to reduce cholesterol/sphingomyelin on the membranes of RAW264.7 cells. We explored the up-stream regulation pathway of viperin in macrophage RAW264.7 cells in the context of RABV infection. An experiment confirmed that a specific Toll-like receptor 4 (TLR4) inhibitor, TAK-242, could inhibit viperin expression in RABV-infected RAW264.7 cells. These results support a regulatory role for TLR4. Geldanamycin, a specific inhibitor of interferon regulatory factor 3 (IRF3) (by inhibiting heat-shock protein 90 (Hsp90) of the IRF3 phosphorylation chaperone), significantly delayed and reduced viperin expression, indicating that IRF3 is involved in viperin induction in RAW264.7 cells. Taken together, our data support the therapeutic potential for viperin to inhibit RABV replication, which appears to involve upstream regulation by TLR4.

Therefore, we explored the capacity of viperin to function as an antiviral molecule against RABV and the mechanistic interaction between RABV and viperin in RAW264.7 cells. Viperin could inhibit both attenuated and street RABV replication and release by hindering viral budding and disrupting cholesterol/sphingomyelin in the RAW264.7 cell membrane. Additionally, the upstream regulation of viperin is regulated by Toll-like receptor (TLR) 4. These findings not only furthered our functional understanding of viperin but also provided evidence in support of this molecule as a new therapeutic target against rabies.

Viperin is highly induced in RABV-infected macrophage RAW264.7 cells. Viperin is highly
induced in RABV-infected, TLR3-positive human neurons 4 . Viperin can be categorized as an antiviral protein [14][15][16] . We hypothesized that viperin might preferentially inhibit RABV replication in RAW264.7 cells. To evaluate this possibility, Western blot analyses were performed to detect viperin expression upon RABV infection in cell lines. Fortunately, we unexpectedly found that high levels of viperin were induced in RAW264.7 cells infected with attenuated rRC-HL at 24 hours post-inoculation (hpi), 16-fold higher than that in NA, BHK-21 and BSR cells, in which viperin was either weakly detected or not expressed at all (Fig. 1A,B).
In another experiment, we generated a stably viperin-expressing BHK-21 cell line by transfecting the viperin-expressing plasmid peGFP-viperin. The viperin-expressing BHK-21 cells presented a bright green-fluorescence of enhanced green fluorescent protein (eGFP) fused with viperin that appeared to be localized within the cytoplasm (data not shown). The stably viperin-expressing BHK-21 cells were infected with rRC-HL at a multiplicity of infection (MOI) of 0.01, and the cell supernatant was used to titrate the virus titre. Viral replication was significantly inhibited following viperin expression. The titres of rRC-HL in GFP-viperin-expressing BHK-21 cells were 3.31 × 10 4 FFU/mL, 3.36 × 10 5 FFU/mL and 1.00 × 10 4 FFU/mL at 24, 36 and 48 hpi; in contrast, the titres of rRC-HL in GFP-expressing BHK-21 cells reached 2.00 × 10 6 FFU/mL, 1.00 × 10 7 FFU/mL and 7.59 × 10 5 FFU/mL ( Fig. 2A). rRC-HL N, P and M protein expression levels were significantly reduced as a result of increased viperin expression (Fig. 2B,C). Meanwhile, RABV genomic RNA (vRNA) and N gene mRNA levels were also significantly reduced in the stably viperin-expressing cells (Fig. 2D).  Figure 2F were measured using Li-Cor Odyssey 3.0 analytical software version 29. (D) RNA expression levels of viperin. rRC-HL vRNA and N mRNA expression levels were detected by qRT-PCR at 24, 36, and 48 hpi. Viperin-expressing BHK-21 cells were infected with rRC-HL at an MOI of 0.01. Data were normalized to β -actin expression and are presented as relative fold expression values to each control cell population infected with rRC-HL. In addition, viperin also inhibited the replication of street RABV GX01 (Fig. S1A-D) and GXN119 (Fig. S1E-G) strains, which are representative isolates of groups I and III, respectively, from an epidemic in Guangxi, China 17,18 .
Based on previous studies [19][20][21] , the N-terminal amphipathic α -helical domain and the radical S-adenosylmethionine (SAM) domain (C83A/C87A/C90A) of viperin were also identified to have critical roles in inhibiting RABV replication (Fig. S2). To confirm whether the viperin-mediated inhibition of RABV infection occurs through the interaction between viperin and RABV proteins, pcDNA-N and pViperin, pcDNA-P and pViperin, pcDNA-M and pViperin plasmids were co-expressed in HEK293 cells, followed by co-immunoprecipitation assays with anti-viperin and anti-N, P and M monoclonal antibodies (MAbs). Viperin did not interact with N, P and M, specifically (Fig. S3). These results suggested that the viperin-mediated inhibition of RABV replication did not occur directly through binding to RABV proteins.
To understand if inhibition of RABV is a function of viperin expression levels, viperin was expressed in BSR cells by transfecting with 0.5-2.0 μ g of pViperin plasmid for 12 h and then infecting the cells with rRC-HL at an MOI of 0.001. Intracellular RABV N, P and M protein expression levels were analysed by Western blotting. The results showed a linear relationship between the increase in pViperin plasmid transfection and decreased expression levels observed for the RABV N, P, and M proteins, indicating that viperin inhibits RABV replication in a dose-dependent manner (Fig. S4A,B).
Viperin reduces cholesterol and sphingomyelin on the cellular membrane. Cholesterol and sphingolipids are the main components of lipid rafts, the specific membrane microdomains required for viral entry, assembly, and budding of various viruses 22,23 . Therefore, we evaluated if the cholesterol and sphingolipids on the cellular membrane were impaired by viperin. The changes in cellular cholesterol and sphingomyelin were analysed by transfecting with pViperin. BSR cells were transfected with pViperin and pcDNA3.0, and the cholesterol and sphingomyelin levels were then assessed using a colorimetric assay. Both cholesterol and sphingomyelin levels were reduced to 85.19% and 82.24% in the BSR cells transfected with pViperin plasmid compared to normal cells (Fig. 3). However, the amounts of cholesterol and sphingomyelin in the BSR cells transfected with an empty vector pcDNA3.0 were reduced to 93.73% and 94.72%, respectively, compared to normal cells, suggesting that viperin reduced cholesterol and sphingomyelin levels on the cellular membranes of the BSR cells.

Inhibitors of cholesterol and sphingomyelin affect RABV budding and release.
In this experiment, Mβ CD, a cholesterol inhibitor, and myriocin, a sphingomyelin biosynthesis inhibitor, were used to first understand the effects on cell growth. Mβ CD and myriocin were dissolved in Dulbecco's modified Eagle's medium (DMEM) supplemented with 2% foetal calf serum (FCS) at 0.5-32 mM and 0.01-100 μ M, respectively. The BSR cells were incubated with Mβ CD and myriocin for 24 h, respectively, and the viability of the treated BSR cells was determined by 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay. Both Mβ CD and myriocin could reduce BSR cell viability in a dose-dependent manner. When the concentration of Mβ CD reached 8 mM, it could significantly reduce BSR cell viability to 86.57% (Fig. 4A), and when the concentration of myriocin reached 50 μ M, it could significantly reduce BSR cell viability to 92.13% (Fig. 4E).
To investigate if Mβ CD and myriocin affected RABV replication, BSR cells were pre-treated with 0.5-8 mM Mβ CD and 0.5-50 μ M myriocin, respectively. At 2 h post-treatment at 37 °C, the cells were collected to prepare the lysates for assessment of cholesterol and sphingomyelin, respectively. The contents of both cholesterol and sphingolipids decreased in a dose-dependent manner. Treatment with 1 mM Mβ CD decreased the cholesterol to 43.22% the level in the control cells (Fig. 4B), and treatment with 50 μ M myriocin decreased sphingolipids to  were measured using Li-Cor Odyssey 3.0 analytical software version 29. The error bars were calculated from at least 3 independent inhibition tests. (E) Effect of myriocin on BSR cell viability. BSR cells were seeded in 96-well microplates and either untreated or pre-treated with 0.01-100 μ m myriocin for 1 h. The supernatants were removed and washed twice with PBS. An MTT assay was then performed. (F) Effects of myriocin on sphingomyelin content and RABV replication. Cells seeded in six-well microplates were infected with rRC-HL at an MOI of 0.001 and then treated with 0.5-50 μ m myriocin for 24 h at 37 °C. The supernatants were harvested for viral titration. The cells were assayed for sphingomyelin content using a sphingomyelin colorimetric assay kit (10009928). (G) Effect of myriocin on RABV replication. Based on D and E, another set of cells was used to prepare lysates that were then subjected to Western blotting analysis for RABV N and β -actin protein expression. (H) The N protein/actin ratios in Fig. 3G were measured using Li-Cor Odyssey 3.0 analytical software version 29. or myriocin for 24 h. The infected BSR cell cultures were used to measure the contents of cholesterol and sphingomyelin and were simultaneously used to detect RABV N protein by Western blotting; their supernatants were used to titrate RABV titres. RABV titres decreased as Mβ CD and myriocin concentrations increased (Fig. 4B,F). Furthermore, RABV N protein expression in the infected BSR cells cultures decreased as Mβ CD and myriocin levels increased (Fig. 4C,D,G,H). These results revealed that inhibitors of cholesterol and sphingolipids affected RABV budding and release.
RABV induces TLR4 signalling pathways to regulate viperin in RAW264.7 cells. TLR4 was identified as recognizing lipopolysaccharides to activate interferon regulatory factor 3 (IRF3), resulting in IFNβ production 24 . In addition, TLR4 can also recognize various glycoproteins in viral envelopes, such as the fusion (F) protein of respiratory syncytial virus (RSV) 25,26 . Furthermore, the envelope proteins (EnV) of the retroviruses murine mammary tumour virus (MMTV) and Moloney murine leukaemia virus both interact with TLR4 27 . To determine whether RABV induced IFN-α /β expression through the TLR4 signal pathway, RAW246.7 cells were incubated with UV-inactivated purified RABV rRC-HL virion or were infected with RABV rRC-HL strain. TLR4 expression in the cells was subsequently analysed by flow cytometry at 24 hours post-incubation. The uninfected RAW264.7 cells incubated with fluorescein isothiocyanate (FITC)-labelled normal mouse IgG were used to normalisation (actual expression of 0.08%). The respective TLR4 expression levels in uninfected, UV-inactivated purified RABV rRC-HL virion and RABV-infected RAW246.7 cells were 16.16%, 47.68% and 54.52%. TLR4 expression in the RAW264.7 cells infected with RABV was 3.37-fold higher than that of the uninfected cells (Fig. 5A). In addition, TLR4 mRNA was up-regulated in RAW264.7 cells infected with RABV by 2.24-, 3.37-and 2.78-fold at 12, 24, and 36 hpi, respectively (data not shown), suggesting that TLR4 is involved in the process of the RABV infection-related induction of viperin expression in this cell line.
Furthermore, the TLR4-specific inhibitor TAK-242, a cell-permeable cyclohexenecarboxylate interacting with the adaptor molecules TIRAP and TRAM via directly binding to the intracellular Cys747 residue of TLR4 28,29 , was used to block the TLR4 signalling pathway. RAW264.7 cells were subsequently infected with rRC-HL at an MOI of 0.1. Viperin levels were obviously reduced by the TLR4-specific inhibitor TAK-242 (Fig. 5B,C). In addition, RAW246.7 cells inoculated with UV-inactivated purified RABV rRC-HL virion showed that RABV virion could be acted as a ligand binding to TLR4 and induced viperin. Furthermore, viperin levels induced by UV-inactivated purified RABV rRC-HL virion were obviously reduced by the TLR4-specific inhibitor TAK-242 (Fig. 5D,E). These findings suggested that RABV induces TLR4 signalling pathways to regulate viperin expression in RAW264.7 cells.

Identification of other cellular factors in the TLR4 signal transduction pathway involved in RABV-induced viperin regulation in RAW264.7 cells.
The above data confirmed that RABV induced IFN-α /β and viperin through TLR4. At the same time, the signalling factors, such as MyD88, IRF3 and nuclear factor κ B (NF-κ B), of the TLR4 transduction pathway were also monitored during RABV infection. Of these factors, MyD88 and IRF3 were up-regulated following RABV infection, which was consistent with RABV-induced IFN-α /β and viperin up-regulation ( Fig. 6A-C).

Discussion
Viperin inhibits the replication of different viruses via different mechanisms 14 . In this study, we found that RABV replication was inhibited in macrophage RAW264.7 cells. Further studies indicated that RABV was inhibited by viperin, an IFN-inducible protein (Figs 2 and 3). Therefore, viperin was identified as the first immune response factor to inhibit RABV, although it was previously reported to inhibit the budding and release of influenza A virus 13 , human immunodeficiency virus (HIV) 31 and hepatitis C virus (HCV) 12 . We demonstrated that natural viperin in RAW264. 7

cells inhibited rRC-HL and that the viperin expressed in transiently transfected BSR cells or in stably transfected BHK-21 cells also inhibited rRC-HL.
Further experimental results showed that cholesterol and sphingomyelin levels were reduced in transfected pViperin BSR cells (Fig. 3). This hypothesis was also supported by the use of inhibitors of cholesterol and sphingomyelin. Mβ CD, an inhibitor of cholesterol, and myriocin, an inhibitor of sphingomyelin, could reduce the respective cholesterol and sphingomyelin levels and also inhibited RABV replication (Fig. 4B,F). However, inhibition by Mβ CD and myriocin was not effective in pre-treated cells (Fig. S5), indicating that Mβ CD and myriocin did not efficiently affect RABV adsorption. Furthermore, RAW246.7 cells were inoculated with UV-inactivated purified RABV rRC-HL virion and demonstrated that RABV virion acted as a ligand of TLR4 to induce viperin (Fig. 5).
To understand how RABV induces viperin production in RAW264.7 cells, innate immune response receptor expression on RAW264.7 cells during RABV infection was assessed. Using a TLR4 inhibitor (TAK-242), TLR4 was identified to be involved in viperin regulation during RABV infection. Previous investigations found that IRF3 was an important transcriptional regulator of the antiviral immune response, mediating the expression of type I IFN and ISGs 32,33 . Thus, IRF3 was determined to be induced in RAW264.7 cells during RABV infection. Furthermore, the immune response factors MyD88, IRF3, and Hsp90 were identified as important components of the TLR4 signal transduction pathway and mediated the antiviral action of viperin (Figs 6 and 7A,B). Although NF-κ B expression increased only slightly during RABV infection (Figs 6A and 7D), viperin expression was essentially not affected when the NF-κ B inhibitor BAY11-7082 was administered (Fig. 7D-F), suggesting that NF-κ B is not involved in viperin regulation.
In this study, a multi-functional protein; i.e., viperin, was identified to inhibit RABV replication. To date, viperin is the first protein to be reported to affect RABV replication, even though it has been previously shown to affect influenza A virus, HIV and HCV. The function of viperin-mediated RABV inhibition in RAW264.7 cells was determined to operate via cholesterol and sphingomyelin reducing on the cellular membrane. Furthermore, TLR4 was verified as an upstream regulator in the signal transduction pathway. The novel data from this study are hypothesized in Fig. 8. These findings will be helpful for the development of new antiviral strategies using viperin in the future.

Materials and Methods
Ethics statement. All   BHK -GFP or BHK -viperin-GFP cells were constructed by transfecting peGFP-N1 or peGFP-viperin plasmids into BHK-21 cells, respectively, and were screened by G418 at least three times. These stably GFP-or viperin-GFP-expressing BHK cells were maintained in DMEM supplemented with 10% FCS and 500 μ g/mL of G418.
The rabies virus rRC-HL was rescued from an infectious cDNA clone pRC-HL (kindly provided by Professor Minamoto, Gifu University, Japan) based on the fixed RABV RC-HL strain used to vaccinate animals in Japan 34 . The street rabies virus isolates GX01 and GXN119 were obtained from the brains of dogs in Guangxi, China and inoculated into 4-week-old mice; the strains were then prepared from mouse brains in our laboratory 17,18 .
RABV virions were purified by referencing the Sokol's method 35 . Briefly, monolayer cultures of BSR cells were prepared and infected with RABV rRC-HL strain at a multiplicity of infection (MOI) of 0.01. The cells supernatant was harvested at 48h post-inoculation, and then the cells supernatant was centrifuged at 8000 rpm for 30 min at 4 °C, the sediment was removed. The RABV virions were precipitated by adding zinc acetate to a final concentration of 0.02 M to the cells supernatant at 4 °C for 1 h, and harvested by centrifugation at 8000 rpm for 1 h, and   Scientific RepoRts | 6:30529 | DOI: 10.1038/srep30529 the pellet was suspended in a saturated solution of ethylenediaminetetraacetate (EDTA)-Tris, pH8.0. The suspension containing RABV virions was centrifuged at 1000 rpm for 5 min and the sediment was removed. The suspension was centrifuged to precipitate RABV virions in 30000 rpm for 3 h at 4 °C, and the pellet was resuspended with NTE buffer (0.13 M NaCl, 0.05 M Tris, 0.001 M EDTA). And then the resuspension containing RABV virions was layered onto 10-60% (w/v) sucrose density gradient in NTE buffer for centrifugation at 30000 rpm for 90 min. The RABV virion band was collected and dissolved in NTE buffer, and recovered by centrifugation at 30000 rpm for 3 h at 4 °C. Finally, the purified RABV virions were suspended in NTE buffer.

Construction of viperin-or mutant-expressing plasmids.
To generate a plasmid for expressing viperin, total RNA was extracted from mouse brains and was used to amplify the viperin gene by RT-PCR with specific primers (Table S1). The PCR product was cloned into the pMD-18T plasmid to construct recombinant pMD-viperin. The ORF of the viperin gene was subsequently subcloned into the expression plasmids pcDNA3.0/ MCS and peGFP-N1 to construct pViperin and peGFP-viperin, respectively. Two other mutants containing the 43-361 amino acid region, the domain that mediates targeting of viperin to the cytosolic side of the endoplasmic reticulum (ER) and lipid droplets 20 , and the SAM domain, which is homologous to the MoaA motif present in the family of radical SAM enzymes, which uses SAM as a cofactor to bind to proteins containing iron-sulphur clusters via the CxxxCxxC motif 21 , were also constructed. In addition, the three expression plasmids, pcDNA-N, pcDNA-P, and pcDNA-M, which express the nucleoprotein (N), phosphoprotein (P) and matrix (M) proteins of the RABV rRC-HL strain, were constructed.
Total RNA extraction and quantitative RT-PCR (qRT-PCR). Total RNA was extracted using RNAeasy Mini Kits (Qiagen, Cat. No. 74104) according to the manufacturer's instructions. On-column DNase digestion was performed using the RNase-Free DNase Set (Qiagen, Cat. No. 79254). The integrity of total RNA was analysed using a Nanodrop 1000 spectrophotometer (Thermo, USA).
qRT-PCR was performed using the SYBR Green method, as previously described 36 . Briefly, 1 μ g of total RNA served as the template for the first-strand cDNA synthesis in a reaction using an oligo(dT) primer and Moloney murine leukaemia virus (MMLV) reverse transcriptase under the conditions described by the manufacturer. A LightCycler 480 PCR detection system (Roche Diagnostics Ltd.) was used for the quantitative assessment of RABV N and viperin mRNA under standard cycling conditions. β -actin gene expression was assessed as a control for all reactions. The primer sequences are listed in Table S1.
MTT assay. An MTT assay was performed as follows: BSR cells were seeded in 96-well microplates with six replicates, either untreated or pre-treated with Mβ CD (Sigma, USA) or myriocin (Sigma, USA), and incubated at 37 °C and 5% CO 2 . After incubation for 24 h post-treatment, the supernatants were discarded and washed three times with PBS, and the MTT reagent (5 mg/mL in PBS) was then added to each well and incubated at 37 °C for 3 h, after which the supernatant was removed. Subsequently, 200 μ L dimethyl sulphoxide (DMSO) was added and incubated at 37 °C for 30 min. Finally, the plates were read on an Imark Microplate Reader (Bio-Rad, USA). The data were calculated as coverage values of three repeated experiments.

Assessment of cholesterol and sphingomyelin content on the cellular membrane.
To quantify cholesterol and sphingomyelin on the membranes of cells either infected or uninfected with RABV, treated or untreated with Mβ CD (Sigma, USA) and myriocin (Sigma, USA), BSR cells were seeded in six-well microplates and either untreated or pre-treated with Mβ CD or myriocin for 1 h at 37 °C. After three washes with PBS, 10 6 cells were harvested and lysed and were then used to measure the content of cholesterol or sphingomyelin using a cholesterol quantification kit (40006; AAT Bioquest, USA) or a sphingomyelin colorimetric assay kit (10009928; Cayman Chemical, USA) according to the manufacturer's specifications. Experiments were repeated in triplicate.