Viperin is an iron-sulfur protein that inhibits genome synthesis of tick-borne encephalitis virus via radical SAM domain activity

Authors


Summary

Viperin is an interferon-induced protein with a broad antiviral activity. This evolutionary conserved protein contains a radical S-adenosyl-l-methionine (SAM) domain which has been shown in vitro to hold a [4Fe-4S] cluster. We identified tick-borne encephalitis virus (TBEV) as a novel target for which human viperin inhibits productionof the viral genome RNA. Wt viperin was found to require ER localization for full antiviral activity and to interact with the cytosolic Fe/S protein assembly factor CIAO1. Radiolabelling in vivo revealed incorporation of 55Fe, indicative for the presence of an Fe-S cluster. Mutation of the cysteine residues ligating the Fe-S cluster in the central radical SAM domain entirely abolished both antiviral activity and incorporation of 55Fe. Mutants lacking the extreme C-terminal W361 did not interact with CIAO1, were not matured, and were antivirally inactive. Moreover, intracellular removal of SAM by ectopic expression of the bacteriophage T3 SAMase abolished antiviral activity. Collectively, our data suggest that viperin requires CIAO1 for [4Fe-4S] cluster assembly, and acts through an enzymatic, Fe-S cluster- and SAM-dependent mechanism to inhibit viral RNA synthesis.

Introduction

Viperin (virus-inhibitory protein, endoplasmic reticulum-associated, interferon-inducible) is a type I interferon (IFN-α/β)-induced protein with a broad spectrum of antiviral activity (Fitzgerald, 2011; Mattijssen and Pruijn, 2011; Seo et al., 2011a). It was shown to affect the multiplication of several important human pathogens such as influenza A virus (FLUAV), human cytomegalovirus (HCMV), hepatitis C virus (HCV), dengue virus (DENV), West Nile virus (WNV), human immunodeficiency virus (Chin and Cresswell, 2001; Rivieccio et al., 2006; Wang et al., 2007; Jiang et al., 2008; 2010; Szretter et al., 2011), and others (Zhang et al., 2007; Carlton-Smith and Elliott, 2012; Teng et al., 2012; McGillivary et al., 2013). The critical steps affected by viperin are depending on the virus, and include progeny particle budding (FLUAV), particle assembly and maturation (HCMV), or RNA genome replication (HCV). Viperin also plays a virus-independent role in immunoregulation (Hata et al., 2009; Qiu et al., 2009), promotes the production of type I IFNs in plasmacytoid dendritic cells (Saitoh et al., 2011), and is enslaved by an HCMV protein to reduce cellular ATP levels (Seo et al., 2011b).

The antiviral mechanism of viperin is ill-defined, but the studies published so far proposed it is based on the sequestration of viral or cellular factors which are essential for infection (Chin and Cresswell, 2001; Wang et al., 2007; 2011; Jiang et al., 2008; Helbig et al., 2011; 2013). Human viperin (also termed Cig5 or RSAD2) is composed of 361 amino acids and contains three major domains; an N-terminal amphipathic α-helix which is mediating localization to the endoplasmic reticulum (ER) and to lipid droplets (Hinson and Cresswell, 2009b), a central radical S-adenosyl-l-methionine (SAM) domain co-ordinating a [4Fe-4S] cluster in vitro (Duschene and Broderick, 2010; Shaveta et al., 2010), and a C-terminal domain with unknown function (Mattijssen and Pruijn, 2011; Seo et al., 2011a). Based on mutational analyses all three domains were shown to be necessary for the antiviral activity, with the weakly conserved N-terminus having a minor influence, and the strongly conserved radical SAM domain and C-terminus having a major influence (Jiang et al., 2008; 2010; Helbig et al., 2013). However, the relationship between the function of a particular domain and the antiviral mechanism of viperin has remained elusive, except for the C-terminus which for HCV was shown to interact with the viral NS5A protein (Helbig et al., 2011) and to deplete the pro-viral host cell factor hVAP-33 (Wang et al., 2011). Moreover, it is currently unclear whether the enzymatic activity of the central radical SAM domain, which was demonstrated in vitro to mediate the cleavage of SAM to 5′-deoxyadenosine (Duschene and Broderick, 2010), indeed has importance for viperin's antiviral function.

Here, we describe that human viperin exhibits a strong activity against Tick-borne encephalitis virus (TBEV; family Flaviviridae), the medically most important arthropod-borne virus in Europe and Russia (Charrel et al., 2004; Lindquist and Vapalahti, 2008). We exploited this extraordinarily sensitive virus/viperin system to investigate the antiviral mechanism. Our results demonstrate that ER localization mediated by the N-terminus contributes to viperin function and that the antiviral activity critically depends on the activity of the radical SAM domain with the central [4Fe-4S] cluster. Moreover, we show that the C-terminus of viperin is important for interaction with CIAO1, a key component of the cytosolic iron-sulfur protein assembly (CIA) machinery (Stehling et al., 2013), and that CIAO1 interaction is required for viperin stability and hence Fe-S cluster formation. Our results thus indicate that the inhibitory effect of viperin against TBEV is based on its Fe-S cluster-dependent enzymatic activity.

Results

Viperin displays strong activity against TBEV

We employed tetracycline (Tet)-inducible HEK293 cell lines (FLP-IN T Rex) (Jiang et al., 2008) to screen a library of (N-terminally FLAG-tagged) IFN-stimulated human host factors capable of inhibiting TBEV multiplication. Overexpressed viperin stood out as being highly active, reducing production of progeny particles 100-fold more than exogenously added 100 U ml−1 IFN (Fig. 1A). Likewise, FLAG-viperin strongly diminished levels of viral RNA (Fig. 1B), and TBEV envelope (E) protein (Fig. 1C). Other prominent antiviral IFN effectors, e.g. ISG20 or PKR, did not exhibit comparable anti-TBEV activity (data not shown).

Figure 1.

TBEV is highly sensitive to viperin. FLP-IN T Rex cell lines inducibly expressing CAT or FLAG-tagged viperin were either treated for 24 h with 1 μg ml−1 Tet and/or for 16 h with 100 U ml−1 of IFN-α, and then infected with TBEV strain Hypr at an moi of 1. TBEV replication was monitored 24 h p.i. using plaque assay (A), real-time RT-PCR (B) and Western blot analysis (C). White bars represent uninduced control cells (CTRL) and black bars represent Tet-treated cells. Mean values and standard deviations were calculated from three independent experiments. *P = 0.03 (A); **P = 0.0032 (B), unpaired t-test.

Inhibition of TBEV by viperin is not dependent on farnesyl pyrophosphate synthase

The pronounced effect of viperin against TBEV encouraged us to investigate its antiviral mechanism. First, we characterized the viperin-sensitive step in the TBEV infection cycle. Viperin is known to bind and inhibit farnesyl pyrophosphate synthase (FPPS) (Wang et al., 2007), a central enzyme of the mevalonate pathway of cholesterol synthesis (Edwards and Ericsson, 1998). Cholesterol is a major constituent of lipid raft microdomains (Charlton-Menys and Durrington, 2008). The inhibition of FPPS by viperin leads to the disruption of lipid rafts on the plasma membrane, thus interfering with the budding of FLUAV particles (Wang et al., 2007). For flaviviruses, lipid rafts are not required for budding, but rather for particle binding and entry into the cytoplasm (Medigeshi et al., 2008; Das et al., 2010). Therefore, we investigated whether viperin would affect these steps in the multiplication cycle of TBEV, using established assays (Aizaki et al., 2008). To test TBEV particle attachment, cells were infected with TBEV for 1 h on ice, unbound virus was washed away, total cellular RNA was extracted, and levels of TBEV RNA were measured by real-time RT-PCR. When FLP-IN T Rex cells were induced to express FLAG-viperin and subsequently compared to uninduced cells, no difference in TBEV RNA levels could be observed (Fig. 2A, left panel). By contrast, when cholesterol was depleted by treatment with methyl-B-cyclodextrin (MBCD), there was a clear inhibition of TBEV attachment (Fig. 2A, right panel). To measure the entry of virus into the cytoplasm, cells were infected for 1 h on ice, unbound particles were washed away, and cells were incubated in pre-warmed cell culture medium for another 2 h at 37°C. Then, cell surface-bound virions were destroyed by trypsin treatment, total RNA was extracted, and TBEV RNA was analysed by real-time RT-PCR. As shown in Fig. 2B (left panel), viperin expression had no influence on TBEV entry, whereas cholesterol depletion had (Fig. 2B, right panel). Similar (but less pronounced) effects on TBEV binding and entry were obtained with the cholesterol synthesis inhibitor mevinoline (data not shown). The cholesterol of the plasma membrane in these cells was also visualized with filipin staining. No visible reduction in cholesterol levels could be seen in viperin-expressing cells, unlike in cells treated with MBCD (Fig. 2C). These results indicate that cholesterol and lipid rafts are important for TBEV binding and uptake into cells. For the viperin effect, however, an involvement of FPPS or the cholesterol metabolism could not be demonstrated. In line with this, the FPPS inhibitor ibandronate had no apparent effect on TBEV growth (supplementary Fig. S1A ). Also, overexpression of FPPS could not relieve the antiviral effect of viperin on TBEV (supplementary Fig. S1B ), which is in contrast to the effect of viperin on FLUAV (Wang et al., 2007).

Figure 2.

Effects of viperin expression and cholesterol depletion on TBEV binding and entry. The sensitive stage of early TBEV infection was assayed for viperin expressed by Tet-induced FLP-IN T Rex cells (left panels) in comparison to the cholesterol depletion agent MBCD (10 mM, added 1 h before virus binding; right panels).

A. Virus attachment assay. Cells were inoculated with TBEV strain Hypr (moi 10) for 1 h on ice, washed with PBS, and the cellular RNA extracted. Virus bound onto the cells was measured by real-time RT-PCR detecting TBEV RNA, and these values were normalized to γ-actin mRNA. The TBEV amount bound to the Tet-induced viperin cells was normalized to the uninduced cells. *P < 0.05, unpaired t-test.

B. Entry assay. After infection of cells for 1 h on ice, the virus inoculum was removed, pre-warmed medium was added to the cells, and the cells were incubated for another 2 h. Then, cells were washed and trypsinated for 10 min to remove bound but not entered virus before RNA extraction and real-time RT-PCR analysis. Data sets are mean values and standard deviations from three independent experiments. **P < 0.005, unpaired t-test.

C. Cholesterol levels on the surface of non-permeabilized cells after the different treatments were visualized by filipin staining. VeroB4 cells had to be used for the cholesterol depletion since 293 do not tolerate MBCD treatment.

Viperin inhibits synthesis of viral positive-sense RNA

Several studies have shown that viperin targets the RNA replication step of flaviviruses (Jiang et al., 2008; 2010; Helbig et al., 2011; Wang et al., 2011). We started investigating this for TBEV by measuring the levels of viral RNA in viperin-expressing cells at the early time point of 5 h post-infection (p.i.). As control we used the translation inhibitor cycloheximide (CHX), since flavivirus RNA synthesis requires a de novo synthesized RNA polymerase. The quantitative RT-PCR analysis of total TBEV RNA levels from the early time point shows a clear reduction in the viperin-expressing cells, comparable to CHX inhibition (Fig. 3A). As CHX completely blocks de novo RNA production by TBEV, the residual RNA signal which is obtained after CHX treatment or viperin expression represents the genome of the input virus. We also differentiated between positive-sense and negative-sense RNA of TBEV, using strand-specific primers (Schwaiger and Cassinotti, 2003). For the positive-sense RNA, both viperin and CHX had a comparable and strong inhibitory effect, only leaving the viral input RNA (Fig. 3B, left panel). Interestingly, however, the negative-sense RNA levels measured at 5 h p.i. were not substantially affected by viperin, whereas CHX completely abrogated the signal (Fig. 3B, right panel). We undertook the same kind of analysis at a late (24 h) time point of TBEV infection and again found that viperin reduces positive-sense RNA to a greater extent than negative-sense RNA (Fig. 3C). Negative-sense RNA is however also affected, most likely because it needs positive-sense RNA as template. These results demonstrate that, for TBEV, viperin acts on the level of viral RNA replication. Moreover, the preferential reduction of positive-sense (i.e. genomic) RNA indicates that a specific step in RNA synthesis is targeted, incapacitating the formation of progeny particles.

Figure 3.

Effects of viperin expression on viral RNA synthesis. TBEV RNA species in FLAG-viperin-expressing FLP-IN T Rex cells. Cells were treated with Tet, with the protein synthesis inhibitor cycloheximide (100 μg ml−1), or left untreated.

A and B. Levels of total TBEV RNA (A) and positive and negative-sense TBEV RNA (B) were determined at 5 h post infection. The symbol § marks the total absence of negative-sense TBEV RNA under CHX treatment.

C. Levels of positive and negative-sense TBEV RNA at 24 h post-infection.

Mean values and standard deviation are calculated from six independent experiments. *P = 0.011 (A); *P = 0.026 (B); **P = 0.0023, ***P = 0.0004, ****P <0.0001 (C), NS = not significant, unpaired t-test.

Mutational analysis of viperin

Viperin has three major domains; an N-terminal amphipathic alpha helix which is responsible for the ER localization of the protein (Hinson and Cresswell, 2009b), a central radical SAM domain (Duschene and Broderick, 2010; Shaveta et al., 2010), and a C-terminal domain with a yet unknown function. All three domains were found to be involved in antiviral activity (Jiang et al., 2008; 2010; Helbig et al., 2011; Wang et al., 2011). We determined which domain is important for the inhibition of TBEV, using a set of FLAG-tagged deletion mutants and point mutants expressed in the Tet-inducible FLP-IN T Rex cell system (Jiang et al., 2008).

First, we deleted the N-terminal 50 amino acid residues of viperin (mutant TN50) which encompass the amphipathic alpha-helix responsible for ER localization (Fig. 4A). Double immunofluorescence analysis confirmed that, in contrast to wild-type (wt) viperin, the TN50 mutant has lost colocalization with the ER marker calnexin (Fig. 4B, upper and middle panels). When the influence on early viral RNA replication (5 h p.i.) was measured, no difference in antiviral activity could be detected between the wt and the TN50 mutant version (Fig. 4C). However, at the late (24 h) time point of infection, wt viperin decreased the TBEV load about 1000-fold, while the TN50 mutant was 100-fold less effective and achieved only a 10-fold reduction (Fig. 4D). This indicates that the N-terminus is involved in the long-term anti-TBEV activity of viperin, but is not absolutely essential for it. To clarify the role of the N-terminal 50 amino acids, we replaced them with the 30-aa-long ER localization signal of the HCV NS5a protein (Hinson and Cresswell, 2009a). As expected, the chimeric viperin (denoted NS5a-TN50) localized to the ER compartment (Fig. 4B, lower panel). Interestingly, the artificially ER-targeted viperin mutant has regained anti-TBEV activity to a level comparable to wt viperin (Fig. 4E). Thus, in agreement with the fact that TBEV replication occurs on ER-derived membranes (Overby et al., 2010), the ER localization of viperin seems to be directing rather than causing the antiviral activity. The striking independence of the early antiviral activity from the N-terminal domain may suggest that the initial establishment of TBEV infection occurs independently of the ER.

Figure 4.

Mutation of the N-terminal domain.

A. Immunoblot analysis of Tet-inducible FLP-IN T Rex cells expressing full-length FLAG-tagged viperin and a mutant with the 50 first aa deleted (TN50).

B. Immunofluorescence analysis of A549 cells transfected with wt viperin, the TN50 mutant, or the NS5a-TN50 chimera (red channel) and the ER marker calnexin (green channel).

C and D. Levels of total TBEV RNA at 5 h (C) and 24 h (D) after infection.

E. Comparison of the antiviral activity of transfected GFP, viperin, TN50 and chimera NS5a-TN50 by immunoblot analysis.

*P = 0.036 (C), **P = 0.0086 (C), **P = 0.0013 and 0.0018 (D), unpaired t-test.

Second, we studied the influence of the central radical SAM domain. Viperin contains four conserved motifs in common with the large family of radical SAM proteins. The first motif, CxxxCxxC, is supposed to bind a [4Fe-4S] cluster which, together with the three other motifs, is involved in binding and cleavage of the cofactor SAM into methionine and a 5′-deoxyadenosyl radical intermediate (Duschene and Broderick, 2010; Shaveta et al., 2010). There are conflicting reports with respect to the importance of the radical SAM domain in the antiviral activity of viperin (Jiang et al., 2008; Helbig et al., 2011). Therefore, we mutated all four conserved radical SAM motifs (Jiang et al., 2008; Helbig et al., 2011) individually (Fig. 5A, mutants denoted M1–M4) and studied the effect on TBEV replication. All mutant proteins were expressed to detectable levels (Fig. 5B), but completely lost their inhibitory activity (Fig. 5C). To ensure that the point mutants are not inactive due to their slightly lower expression levels, we performed a dose–response experiment, using mutant M1 as an example. As shown in Fig. 5D, even at expression levels exceeding those of wt viperin, mutant M1 is unable to inhibit TBEV replication, suggesting that all four radical SAM domain motifs are required for antiviral function.

Figure 5.

Mutation of the central SAM radical domain and the C-terminus.

A. Schematic representation of wt viperin and of point mutants with exchanges in the four motifs of the SAM radical domain (termed M1–M4).

B. Western blot analysis of FLAG-tagged wt viperin and mutants M1–M4 expressed in FLP-IN T Rex cells.

C. TBEV RNA levels of the various viperin lines compared to CAT control cells.

D. Antiviral dose–response experiment for mutant M1 in comparison to wt viperin. HeLa cells were transiently transfected with expression constructs for GFP (1 μg), wt viperin (0.5 μg), or increasing amounts of mutant M1 (0.25 μg, 0.5 μg or 1 μg). All plasmid amounts were kept at 1 μg by addition of the GFP plasmid. At 24 h post-transfection, cells were infected with TBEV at an moi of 0.1, and virus replication was analysed by Western blot 24 h later.

E and F. FLP-IN T Rex cells expressing FLAG-tagged wt viperin, mutants with C-terminal deletions of 60 aa (TC60), 40 aa (TC40), 20 aa (TC20), or 1 aa (TC1), or with an exchange of the C-terminal tryptophan to alanine (W361A).

E. Western blot analysis.

F. TBEV RNA levels of the various viperin lines compared to CAT control cells.

*Non-specific band.

In a third set of experiments, we truncated the C-terminus of viperin by 60, 40 or 20 amino acids (mutants TC60, TC40, TC20) (Jiang et al., 2008). Moreover, the last residue (W361), which is essential for the inhibition of HCV (Jiang et al., 2008), was either deleted (mutant TC1) or exchanged to alanine (mutant W361A). All mutant proteins were expressed to wt levels after Tet induction (Fig. 5E), yet entirely lost their ability to inhibit TBEV replication (Fig. 5F). Thus, the C-terminal region, in particular the endmost residue W361, is essential for the antiviral activity of viperin against TBEV.

Taken together, these results suggest a strict requirement for both the radical SAM domain and the C-terminus of viperin for its anti-TBEV activity, whereas the N-terminus can be replaced with an unrelated ER-targeting sequence.

The cytosolic Fe-S protein assembly factor CIAO1 is a physical and functional interactor of viperin

To obtain mechanistic insights into the antiviral activity of viperin, we aimed at the identification of its cellular interaction partners, using co-immunoprecipitation combined with mass spectrometric analysis. A Tet-inducible FLP-IN T Rex cell expressing a viperin variant tagged with a tandem affinity purification (TAP) tag was established. After induction of TAP-viperin expression, tandem affinity purification and mass spectrometry analysis were performed according to previous protocols (Pichlmair et al., 2011; 2012). Several peptides corresponded to the cellular protein CIAO1 (data not shown) which is a member of the WD40 protein family and required for cytosolic Fe-S protein assembly in human cells (Stehling et al., 2013). CIAO1 is a structural and functional orthologue of yeast CIA1 (Srinivasan et al., 2007) as its heterologous expression in yeast can replace CIA1 as a key component of the cytosolic Fe-S protein assembly (CIA) machinery (Balk et al., 2005; Srinivasan et al., 2007). To confirm the viperin–CIAO1 interaction, we induced FLAG-viperin expression in FLP-IN T Rex cells by addition of tetracycline, and immunoprecipitated the protein. The precipitates contained endogenous CIAO1 as early as 6 h after viperin induction, whereas no CIAO1 was precipitated from uninduced cells (Fig. 6A). Likewise, immunoprecipitation of endogenous viperin after IFN treatment of HeLa cells resulted in the co-purification of endogenous CIAO1 (Fig. 6B).

Figure 6.

Viperin interacts physically and functionally with CIAO1, the human homologue of the yeast Fe-S cluster assembly factor CIA1.

A and B. Co-immunoprecipitation analysis. Viperin was immunoprecipitated with a monoclonal antibody either from extracts of FLP-IN T Rex cells expressing FLAG-tagged wt viperin which were induced with Tet for 0, 6, 12 and 18 h (A), or from HeLa cells treated with IFN-α (1000 U) for 24 h (B). Whole-cell lysates and immunoprecipitates were assayed with specific antibodies against viperin and CIAO1. Cell lysates were also tested for actin levels as loading control.

C–E. siRNA knock-down of CIAO1.

C. Western blot analysis of CIAO1 protein levels (top panel) and real-time RT-PCR analysis of CIAO1 transcript levels (bottom panel). ***P = 0.0003, ****P <0.0001, unpaired t-test.

D. TBEV RNA levels in viperin-expressing FLP-IN T Rex cells treated with siRNAs. ****P <0.0001, unpaired t-test.

E. TBEV titres in viperin-expressing FLP-IN T Rex cells treated with siRNAs. ***P = 0.0004, unpaired t-test.

To evaluate if CIAO1 is required for the antiviral activity of viperin, siRNA knock-down studies were performed. FLP-IN T Rex cells inducible for viperin expression were transfected with an siRNA against the CIAO1 mRNA or a control siRNA (Allstar) and grown for 48 h. The knock-down was verified on the protein level by Western blot analysis, and on the transcript level by real-time RT-PCR (Fig. 6C). To determine the effect of CIAO1 siRNA knock-down on TBEV replication, the siRNA-transfected viperin cells were infected with TBEV for 24 h. The reduction of CIAO1 mRNA levels impaired viperin-mediated suppression of intracellular viral RNA levels (Fig. 6D) as well as viral progeny particle production (Fig. 6E) by at least one order of magnitude.

Thus, we have identified CIAO1 as a novel interactor of viperin which takes part in the antiviral activity against TBEV.

The C-terminus of viperin is required for both the interaction with CIAO1 and the binding of iron

As a next step we mapped the region at which viperin is interacting with CIAO1, employing our viperin mutants described above. Co-immunoprecipitation analyses using a viperin-specific antibody showed that mutants TN50 and M1 (see Figs 4 and 5) still interacted with CIAO1 (Fig. 7A). This indicates that the N-terminus and the conserved CxxxCxxC motif of viperin are not involved in the interaction with CIAO1. By contrast, truncating the C-terminal part of viperin abolished this interaction. Thus, the C-terminal tryptophan, which is conserved in viperin proteins, seems to be the minimal requirement for the interaction with CIAO1.

Figure 7.

CIAO1 interaction and 55Fe incorporation.

A. Mapping of the viperin domain which binds to CIAO1. FLP IN T Rex cell lines expressing FLAG-tagged wt viperin and different deletion and point mutants were tested for the viperin–CIAO1 interaction, using co-immunoprecipitation analysis as outlined for Fig. 6.

B–D. 55Fe incorporation into viperin as a measure of Fe-S cluster assembly. HEK FLP-IN T Rex cells expressing the indicated FLAG-tagged viperin variants or carrying an empty vector (CTRL) were grown in the presence of 55Fe-labelled transferrin for 2 days. Cells were harvested, lysed, and lysates cleared by centrifugation. Protein amounts were measured by the Bradford assay (B) and the 55Fe contents were determined by scintillation counting (C). For absolute numbers of protein content and radioactivity see main text. Lysates were subjected to FLAG-tag affinity purification. 55Fe incorporation into FLAG-tagged viperin, normalized to wt viperin, was determined by immunoprecipitation (D). Mean values and standard deviations from three independent experiments are shown.

Since human CIAO1 is required for the assembly Fe/S clusters on cytosolic/nuclear target proteins, we examined whether viperin may hold a cluster by following the in vivo incorporation of 55Fe into the protein as a measure of Fe-S cofactor assembly at the radical SAM domain. To this end, FLP-IN T Rex cells expressing FLAG-tagged viperin were supplemented with 55Fe-loaded transferrin, lysed 2 days later, and viperin was immunoprecipitated via its FLAG tag. The 55Fe-content of immunoprecipitates was determined by scintillation counting. In addition to wt viperin, we tested 55Fe incorporation of the viperin mutants TC20 and W361 as well as the radical SAM motif mutant M1. An empty vector FLP-IN T Rex cell line was used as negative control. The lysates of all cell lines contained similar amounts of protein (ranging from 1300 to 1700 μg, Fig. 7B) and 55Fe (ranging from 450 to 650 cpm μg−1, Fig. 7C). After affinity purification, however, only wt viperin, but none of the mutants, contained significant amounts of 55Fe (with respect to the total protein input about 1350 cpm μg−1 for wt viperin, and up to 220 cpm μg−1 for the other samples; Fig. 7D). These data reveal not only that viperin harbours an Fe-S cluster in vivo, but also that both the C-terminal tryptophan and the three co-ordinating cysteine residues of the radical SAM motif are essential for assembly of the cluster. Moreover, the lack of 55Fe binding to the mutants TC20 and W361 suggested that CIAO1 is involved in the maturation process.

CIAO1 is required for the maturation of viperin

In order to examine the involvement of CIAO1 in the maturation of viperin more directly, we performed knock-down experiments. CIAO1-directed siRNAs were transfected twice into FLP-IN T Rex cells at a 3-day interval, and after the second round of transfection cells were supplemented with 55Fe-loaded transferring. Expression of FLAG-tagged wt viperin was induced by doxycyclin. Two days later, the 55Fe content of immunoprecipitated viperin was determined as described above. While total protein (Fig. 8B) and 55Fe content (Fig. 8C) of the CIAO1-depleted cells were comparable to the control samples, the 55Fe content of the viperin immunoprecipitates was severely decreased (Fig. 8D), indicating a requirement of CIAO1 for the maturation of viperin. However, Western blot analysis revealed that viperin protein levels were also substantially lowered upon CIAO1 depletion (Fig. 8A), suggesting a critical importance of CIAO1 for viperin stability. In fact, such a stabilizing effect is in agreement with the function of CIAO1 as an Fe-S cluster assembly factor, as its ablation results also in the destabilization and degradation of several other Fe-S proteins (Stehling et al., 2008; 2012; 2013). Consequently, our results are in line with a requirement of CIAO1 for the maturation and stability of viperin.

Figure 8.

55Fe binding and stability of viperin in CIAO1-depleted cells.

A. Western blot analysis of FLP IN T Rex cells expressing FLAG-tagged wt viperin after treatment with the indicated siRNAs and the tetracycline analogue doxycycline, as indicated. Staining for CIAO1 demonstrates the efficiency of RNAi. Tubulin served as loading control.

B–D. CIAO1-dependent 55Fe incorporation into viperin. The cells from (A) were grown in the presence of 55Fe-labelled transferrin and further processed as described for Fig. 7. Protein amounts were determined by the Bradford assay (B) and 55Fe contents were analysed by scintillation counting (C).

D. 55Fe incorporation into FLAG-tagged wt viperin was determined after immunoprecipitation. Mean values and standard deviations from four independent experiments are shown. Values were normalized to mock transfected cells induced with doxycycline.

Antiviral action of viperin is based on the radical SAM domain activity

All viperin mutants which lost their interaction with CIAO1 and failed to incorporate 55Fe also lost their antiviral activity. This may indicate that the C-terminus is not directly involved in antiviral activity, but is rather required for the assembly of Fe-S clusters via its interaction with CIAO1. Consequently, the anti-TBEV mechanism of viperin might be entirely based on the enzymatic, Fe-S-dependent radical SAM activity. Consistently, our viperin mutants that harbour point mutations in the radical SAM domain lack antiviral activity and fail to incorporate 55Fe. However, the inactivation of the radical mutants could also be a non-specific consequence of protein misfolding, as it is known that viperin point mutants lacking the central [4Fe-4S] cluster are unstable and prone to aggregation (Haldar et al., 2012). We therefore sought independent evidence for the contribution of the radical SAM activity to the function of viperin and inhibited cellular SAM synthesis by cycloleucine (Caboche and Hatzfeld, 1978). The FLP-IN T Rex cells inducibly expressing CAT or viperin were treated with cycloleucine and after 48 h infected with TBEV. Total RNA was extracted 18 h p.i., and TBEV RNA levels were determined by real-time RT-PCR. In the absence of transgenic viperin (CAT cells and viperin cells without Tet induction), cycloleucine treatment resulted in a slight increase (1.5-fold) in TBEV RNA levels, and in the presence of viperin in a fourfold rescue of TBEV RNA levels (Fig. 9A).

Figure 9.

SAM is necessary for the antiviral effect.

A. Viperin effect under conditions of SAM synthesis inhibition. FLP IN T Rex cells expressing CAT or wt viperin were treated for 48 h with Tet and cycloleucine in different combinations, infected with TBEV (moi 0.1) for 18 h, and assayed for TBEV RNA levels using real-time RT-PCR. Mean values and standard deviation are calculated from three independent experiments. **P < 0.05, unpaired t-test.

B. Viperin effect under conditions of SAM destruction. 293T cells were co-transfected with plasmids encoding FLAG-viperin, the T3 bacteriophage SAMase, and FLAG-ΔMx (as negative control) in different combinations. At 24 h post-transfection, cells were infected with TBEV (moi of 0.1), and 24 h later intracellular virus RNA was measured using real-time RT-PCR. Mean values and standard deviation from three independent experiments are shown. ****P < 0.0001, unpaired t-test.

C. Immunoblot analysis of cells transfected with FLAG-ΔMx, FLAG-viperin and SAMase using antisera specific for the FLAG epitope (staining FLAG-tagged ΔMx and viperin), or actin.

To further evaluate the requirement of SAM as a cofactor of viperin, we reduced the intracellular SAM levels by means of a SAM-cleaving enzyme, SAMase (also called S-adenosyl-l-methionine hydrolase or AdoMetase). The cDNA sequence of the Escherichia coli bacteriophage T3 (Hausmann, 1967; Krueger et al., 1975; Hughes et al., 1987) was chemically synthesized and cloned into the eukaryotic expression vector pI.18 (see Experimental procedures). 293T cells were transiently transfected with a FLAG-viperin plasmid combined with the SAMase construct or an inactive control (a C-terminal deletion mutant of the antiviral Mx protein; ΔMx). After 24 h of incubation, transfected cells were superinfected with TBEV for 24 h, and then analysed for levels of viral RNA (Fig. 9B). Transfection of the ΔMx control alone or in combination with the T3 SAMase had no apparent effect on TBEV replication. Transfection of viperin substantially reduced levels of TBEV RNA, as expected. Strikingly, however, RNA levels of TBEV were completely restored if the SAMase was expressed along with viperin. Also the virus titres recovered upon coexpression of the SAMase and viperin (data not shown), and Western blot analyses demonstrated that the SAMase had no influence on viperin levels (Fig. 9C).

Taken together, the interaction of viperin with CIAO1, the failure of viperin mutants to incorporate 55Fe, and the lack of viperin activity upon SAM withdrawal demonstrate that the inhibitory function of viperin against TBEV is strictly dependent on the enzymatic activity of the Fe-S cluster-containing radical SAM domain.

Discussion

Viperin is an IFN-stimulated antiviral host factor and immunoregulator, and its importance is more and more recognized. It is highly upregulated upon IFN-α/β stimulation (Chin and Cresswell, 2001; Chan et al., 2008), and capable of inhibiting a wide range of viruses from different families and with different lifestyles. We show here that viperin exerts a strong antiviral effect against the medically important TBEV. A strong inhibition was also shown for FLUAV in case of murine but not of human viperin (Wang et al., 2007; Tan et al., 2012), and for human viperin against Bunyamwera virus (Carlton-Smith and Elliott, 2012), whereas inhibition of other viruses tested so far seems weaker (Helbig et al., 2011; 2013; Seo et al., 2011a; Teng et al., 2012; McGillivary et al., 2013).

The mechanism of viperin action is ill-defined, and the hitherto published studies suggest virus-specific effects (Fitzgerald, 2011; Seo et al., 2011a). For FLUAV it was reported that viperin inhibits progeny particle budding by sequestering FPPS, an enzyme necessary for lipid raft formation on the plasma membrane (Wang et al., 2007). For HCMV, viperin affected particle assembly and maturation by reducing the expression of structural proteins (Chin and Cresswell, 2001). In the case of HCV, by contrast, viperin was shown to interrupt the intracellular step of RNA replication and to interact with the proviral factors hVAP-33 and VAP-A, and the viral NS5A (Jiang et al., 2008; Helbig et al., 2011; Wang et al., 2011). In line with reports for other flaviviruses, we show that for TBEV it is the step of positive-strand RNA synthesis in the ER rather than the involvement of lipid rafts at the plasma membrane, since we were unable to show an involvement of the cholesterol metabolism or of FPPS in viperin action.

Our results indicate that the subcellular localization of viperin is important to inhibit the late phase of TBEV infection, but not the early phase. The N-terminal amphipathic α-helix targets viperin to lipid droplets (Hinson and Cresswell, 2009a), ER (Hinson and Cresswell, 2009b) and mitochondria (Seo et al., 2011b). However, the importance of intracellular localization for the antiviral activity has remained unclear. For WNV (Jiang et al., 2010) and TBEV (this study), removing the amphipathic α-helix reduced but did not abolish the antiviral activity, while a more dramatic effect was shown for HCV (Helbig et al., 2011). For TBEV, the N-terminal deletion mutant TN50 was as active as wt viperin early in infection, but at 24 h p.i. the mutant was two orders of magnitude less potent. We could restore the antiviral activity of TN50 by attaching a foreign ER-localization signal. Although highly speculative, this could mean that only an established TBEV infection is confined to ER membranes, whereas the first round of RNA replication occurs in the cytoplasm. In any case this implies that the impact of the infection phases has to be taken into account when studying effects of antiviral proteins.

Recent reports provided in vitro evidence for the predicted SAM cleavage activity and for the presence of an Fe-S cluster in E. coli-expressed viperin (Duschene and Broderick, 2010; Shaveta et al., 2010). The significance of the central radical SAM domain for the antiviral activity, however, has been debated. Mutating the CxxxCxxC motif abolishes the antiviral activity against WNV, DENV (Jiang et al., 2010) and HCV (Jiang et al., 2008), but the mutations introduced (AxxxAxxA) were predicted to destabilize the whole protein structure due to the lack of the central [4Fe-4S] cluster (Haldar et al., 2012). This renders the interpretation of the mutant data difficult. Also in the case of TBEV the CxxxCxxC to AxxxAxxA mutant has lost all inhibitory potency. We manipulated intracellular SAM levels without mutational alterations of viperin and found its activity markedly decreased. This clearly indicates that the antiviral function of viperin is dependent on the activity of the radical SAM domain.

The catalytic activity of radical SAM enzymes is mediated by [4Fe-4S] centres. Assembly of such metalloclusters on cytosolic proteins requires a dedicated set of maturation factors including the WD40 protein CIAO1 (Balk et al., 2005; Srinivasan et al., 2007; Stehling et al., 2013). In the present study, we have identified the host cell protein CIAO1 as an interaction partner of viperin, and our 55Fe labelling approach indicated that the protein is indeed holding an Fe-S cluster in vivo. Mutation of the cysteine ligands within the radical SAM domain abrogated 55Fe binding and diminished anti-TBEV activity, suggesting that the radical SAM domain is the site of cluster co-ordination and important for protein function. The binding to CIAO1 is mediated by viperin's extreme C-terminus as its ablation or mutation completely abrogates this protein–protein interaction. Moreover, incorporation of 55Fe and anti-TBEV activity were severely impaired in these viperin mutants, consistent with a function of CIAO1 as a viperin maturation factor. In many Fe-S proteins the cofactor exerts a double function: On the one hand it is required for the activity of the protein, and on the other hand it contributes to protein stability (Stehling et al., 2008; 2013). We observed that an siRNA knock-down of CIAO1 mRNA reduced not only the levels of viperin and 55Fe binding, but also viperin's antiviral activity against TBEV. Apparently, depletion of CIAO1 resulted in a much more pronounced destabilization of viperin than impaired cluster binding upon mutation of cystein ligands or C-terminal ablation, indicating that CIAO1 and possibly other cytosolic Fe-S protein assembly factors (Stehling et al., 2012; 2013) contribute to the stability of viperin by yet undefined means. The striking correlation between antiviral activity, CIAO1 binding and Fe-S cluster incorporation is in line with a radical SAM-dependent mechanism of virus inhibition. In this context, the extreme C-terminus of viperin might fulfil multiple functions. On the one hand, CIAO1 binding and Fe-S cluster incorporation were even abrogated when only the C-terminal amino acid residue (W361) was mutated. On the other hand, it is known that the C-terminal domain is one of the most variable regions in radical SAM enzymes, and responsible for binding to cellular substrates (Frey et al., 2008). In line with this, the C-terminal domain of viperin was reported to interact with the cellular cofactor of HCV, hVAP-33 (Wang et al., 2011), but also with the viral protein NS5A (Helbig et al., 2011). Apparently, the C-terminal domain of viperin is binding to different cellular (hVAP-33, CIAO1) and viral (NS5A) partners, suggesting an adapter function for this region.

Collectively, we demonstrate that the radical SAM domain with its associated enzymatic function is crucial for the antiviral activity of viperin against TBEV. It will be of high interest to clarify whether the in vitro product of the viperin enzyme, 5′-deoxyadenosine (Duschene and Broderick, 2010), is responsible for blocking the synthesis of TBEV genome RNA, potentially paving the way for future antiviral treatments.

Experimental procedures

Reagents, standard cells and viruses

Cycloleucine, Ibandronate, filipin III, methyl-B-cyclodextrin (MBCD), cycloheximide and Tet were purchased from Sigma. IFN-α (Multiferon) was purchased from Viragen. Simian Vero B4 and human lung carcinoma cells (A549) were grown in M199 (Invitrogen) and Dulbecco's modified Eagle's medium (DMEM) respectively and supplemented with 5% fetal calf serum (FCS). Human 293T FLP-IN T Rex cells inducibly expressing different N-terminally FLAG-tagged ISGs (Jiang et al., 2008) and FPPS (kindly provided by Ju-Tao Guo) were propagated in DMEM supplemented with 5% Tet-negative FCS (PAA). TBEV strain Hypr was propagated in Vero B4 cells under biosafety laboratory 3 (BSL-3) conditions.

Plasmid constructs

Plasmid pI.18-FLAG-ΔMx (Habjan et al., 2008) was used as negative control. Expression plasmids encoding the bacteriophage T3 SAMase (cDNA synthesized by MrGene), wt viperin, mutant M1, chimeric NS5a-TN50 and N-terminally FLAG-tagged wt viperin were constructed based on the eukaryotic expression vector pI.18 (kindly provided by Jim Robertson, National Institute for Biological Standards and Control, Hertfordshire, UK), using standard PCR cloning methods. KOD Hot Start polymerase (Novagen), restriction enzymes and T4 DNA ligase (Fermentas) were used according to the manufacturers’ recommendations. All plasmids were sequenced to ensure correctness, and oligonucleotide primer sequences are available upon request.

Virus titrations

Viral titres were determined by plaque assay (Overby et al., 2010). Briefly, VeroB4 cells were seeded into six-well dishes and infected with a 10-fold serial dilution of TBEV in a total volume of 200 μl of Opti-MEM. After 1 h incubation a 2 ml overlay containing 0.5% noble agar (Sigma), 2% FCS, 20 mM Hepes and 0.002% DEAE-Dextran (Sigma) in DMEM was added. The plates were incubated at 37°C for 5 days before the staining with crystal violet.

RNA extraction and real-time RT-PCR

Total cellular RNA was extracted at the indicated times p.i. by using the Nucleospin RNA II kit (Macherey-Nagel) according to the manufacturer's recommendations. Aliquots of 600 ng or 1 μg of RNA were used to synthesize cDNA with the Quantitect reverse transcription (RT) kit (Qiagen). All real-time RT-PCR reactions were performed with a LightCycler 1.0 5 (Roche). In the subsequent PCR reactions, the γ-actin mRNA was detected with QuantiTect primers QT00996415 and the QuantiTect SYBR Green RT-PCR Kit (Qiagen). TBEV RNA was detected using a previously described Taqman probe (Schwaiger and Cassinotti, 2003) with the QuantiFast probe PCR kit (Qiagen) and normalized to the γ-actin mRNA signal.

Viral infection, DNA and siRNA transfections

For virus infections, monolayers of cells grown in 12-well plates were incubated for 1 h at 37°C with viruses dissolved OptiMEM (Invitrogen). The virus inoculum was removed, 1 ml of DMEM–2% FCS was added, and the incubation was continued at 37°C. For the transfection of cells with DNA, 1 μg of DNA was prepared with 3 μl of Nanofectin (PAA) in 100 μl of nanofectin diluent according to the manufacturers’ instructions. After 15 min of incubation, the DNA-Nanofectin mixture was dropped onto cells without changing the medium.

For chemical siRNA transfections, approximately 200 000 viperin FLP-In T Rex cells were seeded in 12-well plates and transfected with a mix of 10 nM siRNA (Qiagen) and HiPerfect (Qiagen) transfection reagent, diluted in DMEM. siRNA against CIAO1 mRNA was purchased from Qiagen (Hs_CIAO1_6 siRNA FlexiTube siRNA, order number: SI04285778). Cells were incubated for 24 h before changing the medium to DMEM with or without Tet, and transfected a second time with 10 nM siRNA. After another 24 h the cells were infected with TBEV (moi 0.1) for 24 h, before RNA and protein analysis. In 55Fe labelling experiments, siRNAs were transfected by electroporation into 4 to 8 × 106 FLP-In T Rex cells. Transfections were carried out twice in a 3-day interval as described (Stehling et al., 2013), using a Bio-Rad GenePulser device (4 mm gap cuvette, 250 μl cells suspension volume, capacitance 500 μF, voltage 310 V, indefinite resistance). Subsequently, FLP-In T Rex cells were seeded into tissue culture flasks pre-coated with rat tail collagen I.

Virus binding, entry and replication assay

FLP-IN T Rex cells inducibly expressing FLAG-viperin were seeded in six-well dishes and treated with Tet for 36 h. Cells were infected with TBEV at an moi of 10 for 1 h on ice to allow attachment but impede virus entry. After three washes with PBS, RNA was extracted to measure the amount of cell-bound virus. To assay the subsequent infection steps, the virus inoculum was removed after 1 h of binding on ice, cells were washed, pre-warmed medium was added, and the cells were incubated for another 2 h (entry) or 5 h (RNA synthesis) at 37°C. Cells were washed with PBS, trypsinated for 10 min, and washed again to remove any cell-associated virus which had not entered the cytoplasm before RNA extraction.

Immunofluorescence analysis

Cells were grown on coverslips to 30–50% confluence, transfected with plasmids expressing wt or mutated viperin and incubated for 24 h. Cells were fixed with 3% paraformaldehyde, permeabilized with 0.5% Triton X-100 dissolved in PBS, and washed three times with PBS containing 1% FCS. Primary antibodies were diluted in PBS containing 1% FCS. Viperin was detected by using mouse monoclonal anti-viperin (Abcam) diluted 1:250, and calnexin was detected with rabbit polyclonal antibody (Abcam) diluted 1:200. After incubation at room temperature for 1 h, the coverslips were washed with PBS and treated with donkey anti-mouse 555 and donkey anti-rabbit 488 (Alexa Fluor, Invitrogen) secondary antibodies conjugated at a dilution of 1:200, washed and mounted. Pictures were taken with a Nikon A1 confocal microscope. To stain for cholesterol, cells grown on coverslips were fixed with PFA, washed with PBS, and incubated with 0.25 mg ml−1 filipin III diluted in PBS for 1 h. The coverslips were washed and mounted before visualization with a Leica confocal laser scanning microscope.

Western blot analyses

Cells were lysed in radioimmunoprecipitation assay buffer [50 mM Tris (pH 7.4), 150 mM NaCl, 1% NP-40, 0.25% sodium deoxycholate, 0.05% SDS] containing protease inhibitors (Complete protease inhibitor; Roche). A total of 10 μg of protein was separated by SDS-PAGE and transferred onto an Immobilon-P PVDF membrane (Millipore), followed by incubation in saturation buffer (PBS containing 5% non-fat dry milk and 0.05% Tween). The membrane was first incubated for overnight with primary antibodies and washed three times with 0.05% PBS–Tween, followed by incubation with horseradish peroxidase-conjugated secondary antibodies (Pierce). After an additional three washing steps, detection was performed by using the SuperSignal West Pico or Femto kit (Pierce). Primary antibodies used were directed against TBEV E (mouse monoclonal 1493.1, diluted 1:5000; Niedrig et al., 1994), actin (rabbit polyclonal anti-actin, diluted 1:5000; Sigma), FLAG epitope (mouse monoclonal anti-FLAG M2; Stratagene, diluted 1:2500), viperin (rabbit polyclonal anti-viperin, diluted 1:1500; Abcam) and CIAO-1 (rabbit polyclonal anti-CIAO1, diluted 1:5000; Abcam).

Co-immunoprecipitation

FLP-IN T Rex cells inducibly expressing viperin variants were treated with Tet for the indicated times before cell lysis [50 mM Tris (pH 8), 150 mM NaCl, 1% Triton X-100 and protease inhibitor]. The viperin–CIAO1 complex was immunoprecipitated with monoclonal antibodies directed against viperin (Abcam). Protein A agarose beads (Millipore) were used to precipitate the antibody-protein complexes, and the proteins were resuspended in reducing Laemmli SDS-PAGE sample buffer before Western blot analysis.

55Fe incorporation into viperin

FLP-IN T Rex cells transgenic for FLAG-tagged wt viperin or the mutants TC20, W361A, or M1 were seeded in 25 cm2 tissue culture flasks. A cell line stably transfected with an empty vector was included as negative control. Viperin expression was induced by addition of 3 μg ml−1 doxycycline per day, and medium was supplemented with 55Fe-loaded transferrin to a final concentration of 900 pM, corresponding to a radioactivity of 2.5 μCi ml−1. In experiments where CIAO1 was depleted by RNAi, 55Fe and doxycyclin supplementation was started immediately after the second round of transfection. Two days after induction of FLAG-viperin expression, cells were subjected to anti-FLAG immunoprecipitation. Briefly, cells were harvested, analysed for total protein content, lysed in 400 μl TNEGT buffer (10 mM Tris-HCl pH 7.4, 150 mM NaCl, 1 mM EDTA, 10% Glycerol, 0.2% Triton X-100, 1 mM PMSF), and cleared by centrifugation at 13 000 g. An aliquot of 5 μl of each cleared lysate was removed for scintillation counting to determine total 55Fe uptake. The remaining lysate was incubated for 2 h with 15 μl of pre-washed slurry of sepharose anti-FLAG M2 beads (Sigma). Subsequently, flow-through (non-bound material) was removed, beads were washed three times in TNEGT buffer, and resuspended in 2 ml of scintillation fluid to determine 55Fe incorporation into the immunoprecipitate.

Acknowledgements

We thank Valentina Wagner and Ralf Rösser for excellent technical assistance, Otto Haller for constant support and advice, and Matthias Niedrig, Jim Robertson, Richard Lundmark and Mari Andersson for providing reagents. We are indebted to Rudi Hausmann who has brought the phage T3 SAMase to our attention many years ago. Work in our laboratories is supported by the Grants 01 KI 0711 from the Bundesministerium für Bildung und Forschung (BMBF) and SFB 593 from the Deutsche Forschungsgemeinschaft (DFG) (to F.W.) and by the Grants ICA 10-0059 from the Swedish foundation for strategic research, 2011-2795 from the Swedish research council, JCK-1120 from the Kempe Foundation and the laboratory for molecular infection medicine Sweden (MIMS) (to A.K.Ö.). Work in the lab of R.L. was supported by grants of DFG (SFB 593 and GRK 1216), von Behring-Röntgen Stiftung, LOEWE programme of state Hessen, and Max-Planck Gesellschaft.

Contributions

A.S.U., K.V., A.P., O.S., K.L.B. and A.K.Ö. performed experiments. G.D., J.-T.G., G.S.-F., R.L., A.K.Ö. and F.W. designed research and analysed data, and A.K.Ö. and F.W. supervised the project and wrote the manuscript.

Competing financial interests

The authors declare no competing financial interests.

Ancillary