Ebolavirus VP24 Binding to Karyopherins Is Required for Inhibition of Interferon Signaling

Abstract

The Ebolavirus VP24 protein counteracts alpha/beta interferon (IFN-α/β) and IFN-γ signaling by blocking the nuclear accumulation of tyrosine-phosphorylated STAT1 (PY-STAT1). According to the proposed model, VP24 binding to members of the NPI-1 subfamily of karyopherin alpha (KPNα) nuclear localization signal receptors prevents their binding to PY-STAT1, thereby preventing PY-STAT1 nuclear accumulation. This study now identifies two domains of VP24 required for inhibition of IFN-β-induced gene expression and PY-STAT1 nuclear accumulation. We demonstrate that loss of function correlates with loss of binding to KPNα proteins. Thus, the VP24 IFN antagonist function requires the ability of VP24 to interact with KPNα.

In the order Mononegavirales, Ebolavirus and Marburgvirus belong to the family Filoviridae (12). Ebolaviruses are responsible for outbreaks in central Africa of severe hemorrhagic fever in humans and nonhuman primates, with human fatality rates of up to 90% (16). Currently, there are no licensed vaccines or approved treatments available for filovirus infections. Ebolavirus counteracts synthesis of alpha/beta interferon (IFN-α/β) and cellular responses to IFN-α/β/γ (7, 8, 10, 11, 23). VP35 blocks production of IFN-α/β by inhibiting the activation of interferon regulatory factors by blocking interferon regulatory factor 3 (IRF-3) kinases IKKepsilon and TBK-1 (1, 2, 21) and by increasing SUMOylation of IRF-7 (3). VP24 impairs cellular responses to IFN-α/β/γ by blocking nuclear accumulation of tyrosine-phosphorylated STAT1 (PY-STAT1). Nuclear translocation of PY-STAT1 is essential for transcriptional activation of numerous IFN-responsive genes and occurs through interaction with karyopherin α-1 (KPNα1) and perhaps other members of the NPI-1 subfamily of KPNα proteins (14, 15, 18, 19, 22, 25). Correspondingly, VP24 interacts with the KPNα proteins that mediate PY-STAT1 nuclear accumulation (23, 24). Other pathogenic viruses also impair innate immune signaling via interaction with KPNα proteins, suggesting that this may be a common immune evasion strategy. For example, the severe acute respiratory syndrome coronavirus (SARS-CoV) ORF6 protein tethers a karyopherin α-2/karyopherin β complex to the endoplasmic reticulum (ER)/Golgi membrane to prevent PY-STAT1 nuclear import complex formation (6), and Hantaan virus nucleocapsid protein inhibits NF-κB activation through interaction with KPNα (26). Our previous work suggests that the interaction of VP24 with select KPNα proteins is responsible for the ability of VP24 to inhibit IFN signaling (23, 24).

In this study, mutations that impair VP24 inhibition of IFN-induced gene expression were identified. Plasmids expressing wild-type or mutated VP24 proteins were cotransfected into 293T cells together with plasmids expressing Renilla luciferase (expressed from a constitutive cytomegalovirus [CMV] promoter) and firefly luciferase (expressed from an IFN-inducible ISG54 promoter). At 1 day posttransfection, the cells were treated with 1,000 U/ml of human IFN-β (PBL Interferon Source) for 16 h, and then cells were lysed and luciferase levels were quantified. Firefly luciferase levels were normalized to Renilla luciferase levels. Using this assay to screen the activity of amino-terminal truncation mutants, we found that deletion of amino acids 1 to 26 of VP24 did not significantly affect the ability of VP24 to inhibit IFN-β-induced gene expression. However, mutants with larger deletions, encompassing residues 1 to 50, lost the ability to efficiently inhibit IFN-β-induced gene expression (Fig. 1A). This pointed to residues 26 to 50 as important for VP24 function. Alanine-scanning mutagenesis of clusters of 5 amino acids from position 25 to position 50 was performed, and this restricted the effect to residues 36 to 45 (data not shown). Alanine-scanning mutagenesis of this region was then performed and identified residue 42 as critical for inhibition of the reporter gene expression (Fig. 1B). Specifically, a W42A VP24 mutant (mut1) inhibited ISG54 reporter activation to 30% relative to the level for the empty-vector, IFN-β-treated sample, whereas wild-type VP24 almost completely inhibited reporter expression (Fig. 1C and D).

FIG. 1.

VP24 mutants with impaired inhibition of IFN-β-induced gene expression. At 24 h posttransfection, 293T cells were mock treated or treated with 1,000 U/ml IFN-β for 16 h. Cells were then lysed, and lysates were assessed for firefly luciferase activity using a dual luciferase assay (Promega). All reporter gene assays were performed in triplicate (A, B, and D). (A) Inhibition of an ISG54 promoter-driven firefly luciferase gene by Z-VP24 and VP24 truncation mutants. 293T cells were transfected with plasmids pRLTK and pISG54FFluc plus either pCAGGS-empty, -HAVP24, -HAVP24(26-251), -HAVP24(51-251), -HAVP24(61-251), or -HAVP24(71-251). (B) Inhibition of an ISG54 promoter-driven firefly luciferase gene by Z-VP24 and VP24 single alanine mutants. 293T cells were transfected with plasmids pRLTK and pISG54FFluc plus either pCAGGS-empty; Z-VP24; or VP24-Q36A, -G47A, -W38A, -K39A, -V40A, -Y41A, -W42A, -G44A, or -I45A. (C) Schematic representation and nomenclature of VP24 mutants. Amino acid 42 and amino acids 142 to 146 of Z-VP24 were replaced with alanine residues to create VP24 mutants mut1 and mut2, respectively. Mutant VP24 proteins containing both mut1 and mut2 substitutions were designated mut3. (D) Inhibition of an ISG54 promoter-driven firefly luciferase gene by Z-VP24 and VP24 mutants. 293T cells were transfected with plasmids pRLTK and pISG54FFluc plus either pCAGGS-empty, -HA-Z-VP24, -HAmut1, -HAmut2, or -HAmut3. Whole-cell extracts were analyzed by Western blotting using antitubulin and anti-hemagglutinin (anti-HA) antibodies (right panel).

Interestingly, amino acids 142 to 146 share some similarities with STAT1 amino acids 410 to 413, which are involved in KPNα1 binding (5). Mutation of residues 142 to 146 to alanines (mut2) also reduced the ability of VP24 to inhibit ISG54 activation by IFN-β (Fig. 1C and D). When W42A and the mutations at residues 142 to 146 were combined (mut3), ISG54 reporter expression was increased up to 90% relative to the level for the empty vector, thus drastically reducing VP24 activity (Fig. 1C and D). These are the first mutations within VP24 demonstrated to affect the VP24 IFN antagonist function.

IFN-β triggers intracellular activation of STAT1, which consequently accumulates in the nucleus to induce expression of interferon-stimulated genes (ISGs). VP24-mediated inhibition of IFN-induced gene expression was previously correlated with the ability of VP24 to block STAT1 nuclear accumulation (23). We therefore determined whether VP24 mutants exhibit altered ability to block nuclear accumulation of STAT1, using a human STAT1 protein fused to green fluorescent protein (STAT1-GFP). While a recent publication has shown that GFP tagging affects the nucleocytoplasmic shuttling kinetics of unphosphorylated STAT1 (20), this assay is commonly used to monitor STAT1 localization and provides a straightforward method for assaying VP24 function in unfixed cells (4, 13, 23). Vero E6 cells were cotransfected with plasmids encoding STAT1-GFP and either wild-type VP24, mut1, mut2, or mut3. At 24 hours posttransfection, cells were treated with 1,000 U/ml IFN-β for 45 min, fixed, and analyzed for STAT1-GFP and VP24 by using an anti-VP24 antibody and indirect immunofluorescence (Fig. 2A). Quantification of STAT1 subcellular localization was then performed, and the percentage of VP24 positive cells in which the GFP signal was clearly concentrated in the nucleus was determined. In the absence of a clear nuclear GFP signal, the cells were considered to have cytoplasmic STAT1-GFP (Fig. 2B). In nontreated cells, STAT1-GFP was mainly localized in the cytoplasm, with weak expression in the nucleus (100% cells; n = 74). However, upon IFN-β treatment, STAT1-GFP was relocalized to the nuclei of the cells (86% cells; n = 72). In the presence of wild-type VP24, STAT1-GFP nuclear accumulation was almost completely blocked (n_VP24 = 85). Strikingly, 83% of cells expressing mut3 showed a strong nuclear accumulation of STAT1-GFP (n_mut3 = 84). Mutants mut1 and mut2 (n_mut1 = 51; n_mut2 = 79) did not reveal a clear difference from wild-type VP24, although mut2 showed a small percentage of cells with nuclear accumulation of STAT1-GFP. The latter observation can be explained by partial destruction of the VP24 activity and also by possible variation in the level of the protein expression. The fact that VP24 mut3 possesses an obvious defect in its ability to block IFN-β-induced nuclear accumulation of STAT1-GFP was consistent with the ISG reporter data (Fig. 1).

FIG. 2.

Efficiency of STAT1-GFP nuclear accumulation in cells expressing wild-type and mutant Z-VP24 proteins. (A) Vero E6 cells were transfected with plasmids encoding STAT1-GFP plus either pCAGGS (empty vector), pCAGGS-HAVP24 (encoding wild-type VP24), pCAGGS-HAmut1, pCAGGS-HAmut2, or pCAGGS-HAmut3. At 16 h posttransfection, cells were mock treated or treated with 1,000 U/ml IFN-β for at least 30 min. Cells were then fixed with paraformaldehyde and stained for expression of VP24. Nuclei were stained with dapi (4′,6-diamidino-2-phenylindole). (B) The graph represents the percentages of cells showing nuclear STAT1-GFP (white) or cytoplasmic STAT1-GFP (black) in the presence of different VP24 proteins.

According to our previous data, VP24 selectively blocks nuclear accumulation of activated phosphorylated PY-STAT1 (23). Next, the impact of wild-type and mutant VP24 on nuclear accumulation of endogenous PY-STAT1 was addressed using Vero E6 cells. Treatment with 1,000 U/ml IFN-β for 30 min led to nuclear accumulation of PY-STAT1 (Fig. 3A). In the presence of Z-VP24, PY-STAT1 was retained in the cytoplasm. Interestingly, in mut1- and mut2-expressing cells, nuclear accumulation of PY-STAT1 was only partially blocked, an observation that correlates with the attenuated phenotype observed in the reporter gene assay. Strikingly, nuclear accumulation of PY-STAT1 was not impaired by VP24 mutant mut3, which also correlates with the inability of this mutant to inhibit IFN-β-induced gene expression. We previously demonstrated that VP24 binds a subset of KPNα proteins which are involved in nuclear translocation of STAT1 (24). To assess whether the loss-of-function phenotype of the VP24 mutants was due to impaired interaction with KPNα1, cells were transfected with plasmids encoding Flag-tagged KPNα1 and either wild-type or mutant VP24. In this and subsequent studies, 293T cells were used instead of Vero E6 cells in order to substantially increase the number of cells expressing VP24. Anti-Flag immunoprecipitations were then performed and the samples analyzed for the presence of Z-VP24 (Fig. 3B). Whereas Z-VP24 was coimmunoprecipitated with Flag-KPNα1, levels of mut1 and mut2 binding to Flag-KPNα1 were considerably reduced, and interaction was almost completely lost with mut3.

FIG. 3.

Efficiency of nuclear accumulation of tyrosine-phosphorylated STAT1 in cells expressing wild-type and mutant Z-VP24 proteins. (A) Vero E6 cells were transfected with pCAGGS (empty vector), pCAGGS-HAVP24 (encoding wild-type VP24), pCAGGS-HAmut1, pCAGGS-HAmut2, or pCAGGS-HAmut3. At 16 h posttransfection, cells were mock treated or treated with 1,000 U/ml IFN-β for at least 30 min. Cells were then fixed in paraformaldehyde and stained for expression of VP24 and PY-STAT1. This experiment was carried out three times with similar results (white arrows point to VP24-positive cells). (B) 293T cells were transfected with plasmids encoding Flag-KPNα1 and with either pCAGGS-HAVP24, pCAGGS-HAmut1, pCAGGS-HAmut2, or pCAGGS-HAmut3. At 16 h posttransfection, cells were lysed and lysates were incubated for at least 2 h with a slurry of EZ view red anti-Flag beads (Sigma). Whole-cell extracts (WCE) and immunoprecipitates (IP: Flag) were then analyzed by Western blotting for the presence of Flag-karyopherin α-1 and VP24 by using anti-Flag and anti-HA antibodies. (C) 293T cells were transfected with plasmids encoding Flag-karyopherin α-1 (Flag-KPNα1) and with pCAGGS-empty, pCAGGS-HAVP24, pCAGGS-HAmut1, pCAGGS-HAmut2, or pCAGGS-HAmut3. At 16 h posttransfection, cells were mock treated or treated with 1,000 U/ml IFN-β for at least 30 min before lysis in extract buffer. Lysates were clarified and incubated for at least 2 h with a slurry of EZ view red anti-Flag beads (Sigma). Whole-cell extracts (WCE) and immunoprecipitates (IP: Flag) were then analyzed by Western blotting for the presence of PY-STAT1, Flag-KPNα1, and VP24 by using anti-PY-STAT1, anti-Flag, and anti-HA antibodies.

Previously, we identified arm domain 10 of KPNα1 as critical for binding to both PY-STAT1 and VP24, and we proposed that VP24 may compete with PY-STAT1 for binding to KPNα proteins in order to block PY-STAT1 nuclear import (23, 24). In this study, we addressed the latter point by performing a KPNα1-binding competition assay. Using Western blot analysis of whole-cell extracts, we demonstrate that PY-STAT1 was detected only in IFN-β-treated 293T cells and that the levels of expression of wild-type and mutated VP24 proteins were comparable (Fig. 3C). In the absence of VP24, PY-STAT1 was precipitated with KPNα1, but the presence of wild-type VP24 affected this interaction and clearly reduced the amount of precipitated PY-STAT1. All three VP24 mutants were impaired in their ability to affect the interaction between KPNα1 and PY-STAT1. Similarly, each of the mutants again exhibited reduced capacity for binding to KPNα1. As expected, mut3 showed the least capacity to interfere with PY-STAT1-KPNα1 binding, and this correlated with the lowest level of binding to KPNα1 (Fig. 3C).

Because the combination of W42A with mutations at residues 142 to 146 (mut3) decreased interaction with KPNα1, alanine scanning of individual residues between positions 142 and 146 was performed in the background of the W42A mutation. These mutants were then assessed for KPNα1 binding by using coimmunoprecipitation with a Flag-tagged KPNα1 protein (data not shown). Of these, only K142A resulted in complete loss of binding (Fig. 4A). When assessed for its ability to inhibit IFN-β-induced gene expression, the W42A K142A double mutant showed an even lower level of activity than did W42A (Fig. 4B). Finally, to address the role of the mutations in ebolavirus replication, we used a reverse genetics approach (28). Several plasmid constructs carrying full-length cDNA of ebolavirus were generated, each encoding mutations in VP24 at position 42 and also other mutations in different viral gene products, including those also in VP24 that were associated with guinea pig adaptation (17, 27). Multiple attempts to rescue these VP24 mutant viruses failed, while in control experiments, viruses carrying wild-type VP24 were successfully rescued (data not shown). Importantly, a reverse of the VP24 mutation at position 42 to the wild-type amino acid sequence led to recovery of all of these recombinant viruses. These results further emphasize the importance of VP24 for viral replication and suggest that the VP24 region involved in KPNα binding is critical for virus replication.

FIG. 4.

Interaction of wild-type and mutant Z-VP24 with karyopherin α-1. (A) 293T cells were transfected with plasmids encoding Flag-KPNα1 and with either phCMV-VP24wt, phCMV-W42A, or phCMV-W42A-K142A. At 16 h posttransfection, cells were mock treated or treated with 1,000 U/ml IFN-β for at least 30 min before lysis in extract buffer. Lysates were clarified and incubated for at least 2 h with a slurry of EZ view red anti-Flag beads (Sigma). Whole-cell extracts (WCE) and immunoprecipitates (IP: Flag) were analyzed by Western blotting for the presence of Flag-karyopherin α-1 and VP24 by using anti-Flag and anti-VP24 antibodies. (B) ISG54-FFluc reporter gene assay. 293T cells were transfected to express similar amounts of Z-VP24, VP24-W42A, or VP24-W42A-K142A. At 24 hours posttransfection, cells were mock treated or treated with 1,000 U/ml of IFN-β for 16 h. Cells were then harvested, and samples were analyzed for luciferase activity using a dual luciferase assay (Promega) and for protein expression by Western blotting. The reporter gene assay was performed in triplicate.

Combined, the data obtained in this study address a central question related to the mechanism by which VP24 inhibits IFN signaling. We identified two functional regions of VP24 required for its IFN-signaling antagonist function and demonstrate that the extent to which VP24 mutants lose the ability to inhibit IFN-induced gene expression correlates with their ability to bind KPNα. We thus provide evidence that VP24 blocks the IFN-dependent cellular antiviral response by binding to KPNα proteins. This study also describes the first VP24 mutants with altered KPNα binding and IFN antagonist function. These data thus provide a starting point upon which mutant viruses might be generated by reverse genetics (28). Ebolavirus encodes a second IFN antagonist, VP35, which suppresses IFN-α/β production (1, 2). Recent studies have shown that the VP35 IFN antagonist function is critical for efficient ebolavirus replication in a nonlethal mouse model (9, 10). Generation of recombinant viruses impaired in both VP35 and VP24 IFN antagonist activity should shed light upon how these two factors cooperate to influence replication and pathogenesis.