IRAV (FLJ11286), an Interferon-Stimulated Gene with Antiviral Activity against Dengue Virus, Interacts with MOV10

ABSTRACT Dengue virus (DENV) is a member of the genus Flavivirus and can cause severe febrile illness. Here, we show that FLJ11286, which we refer to as IRAV, is induced by DENV in an interferon-dependent manner, displays antiviral activity against DENV, and localizes to the DENV replication complex. IRAV is an RNA binding protein and localizes to cytoplasmic processing bodies (P bodies) in uninfected cells, where it interacts with the MOV10 RISC complex RNA helicase, suggesting a role for IRAV in the processing of viral RNA. After DENV infection, IRAV, along with MOV10 and Xrn1, localizes to the DENV replication complex and associates with DENV proteins. Depletion of IRAV or MOV10 results in an increase in viral RNA. These data serve to characterize an interferon-stimulated gene with antiviral activity against DENV, as well as to propose a mechanism of activity involving the processing of viral RNA. IMPORTANCE Dengue virus, a member of the family Flaviviridae, can result in a life-threatening illness and has a significant impact on global health. Dengue virus has been shown to be particularly sensitive to the effects of type I interferon; however, little is known about the mechanisms by which interferon-stimulated genes function to inhibit viral replication. A better understanding of the interferon-mediated antiviral response to dengue virus may aid in the development of novel therapeutics. Here, we examine the influence of the interferon-stimulated gene IRAV (FLJ11286) on dengue virus replication. We show that IRAV associates with P bodies in uninfected cells and with the dengue virus replication complex after infection. IRAV also interacts with MOV10, depletion of which is associated with increased viral replication. Our results provide insight into a newly identified antiviral gene, as well as broadening our understanding of the innate immune response to dengue virus infection.

importance to public health. Dengue virus infects an estimated 390 million people each year, with more than 2.5 billion people living in regions where they are at risk of dengue virus transmission (1)(2)(3)(4).
Previously, we identified FLJ11286 as one of a number of genes that were upregulated in Daudi cells in response to treatment with IFN (20) and showed that they displayed antiviral activity against encephalomyocarditis virus (EMCV) (21). FLJ11286 has also been shown by our laboratory and others to have antiviral activity against DENV (22,23). FLJ11286, which we refer to here as IRAV (interferon-regulated antiviral gene) (also annotated as C19orf66, UPF0515, or RyDEN), encodes a protein 291 amino acids (aa) in length with a calculated molecular mass of 33.1 kDa. Analysis of published microarray data suggests that IRAV (FLJ11286) is upregulated in response to type I and type II IFNs (6,20,(24)(25)(26). IRAV (FLJ11286) has been shown in microarray screens to be upregulated in response to the yellow fever virus vaccine (27), as well as after infection with a number of different pathogens, including adenovirus (28), influenza virus (29), Lassa virus (30), and ebola and Marburg viruses (31), as well as after infection with human herpesvirus 8 and human herpesvirus 1 (32,33). In addition, proteomic analysis of the hepatitis C virus (HCV) interactome identified IRAV (FLJ11286) as one of 214 human proteins interacting with the HCV NS3 protein (34), suggesting a role for IRAV in the host pathogen response.
Here, we show that IRAV is upregulated in response to DENV infection in an IFN-dependent manner. Upregulation of IRAV in response to IFN-␤ treatment can be blocked by disrupting the canonical ISGF3 pathway. CRISPR/Cas9-mediated knockout of IRAV resulted in increased titers of DENV, as well as of EMCV. We also demonstrate that IRAV associates with DENV proteins and localizes to the viral replication complex. IRAV is an RNA binding protein and localizes to P bodies in uninfected cells. IRAV also associates with the host RNA binding proteins UPF1 and HuR (ELAV1) and interacts with MOV10 (a RISC complex RNA helicase), suggesting a role for IRAV in processing or stability of RNA. Furthermore, we propose a mechanism of action for IRAV that utilizes intrinsic RNA decay pathways. These pathways have been shown to be of increasing importance to the life cycles of multiple viruses, as well as in an array of cellular processes.

IRAV is upregulated after dengue virus infection in an interferon-dependent manner.
To determine if IRAV is upregulated in response to DENV infection, A549 cells were infected with DENV for 72 h. Quantitative real-time PCR (qRT-PCR) analysis showed upregulation of IRAV starting at 24 h postinfection, corresponding to increased expression of DENV RNA, IFN-␤, and ISGs, including IFIT3 (Fig. 1A). To determine if IRAV is regulated through the canonical IFN (ISGF3) signaling pathway, IRF9 knockouts (KO) were generated in A549 cells using CRISPR/Cas9. Knockout of IRF9 resulted in decreased expression of IRAV after DENV infection (Fig. 1B), as well as after IFN-␤ treatment (Fig. 1C), indicating that the canonical ISGF3 pathway plays a role in induction of IRAV.
To further characterize IRAV induction in response to IFN, HeLa cells were treated with various concentrations of IFN-␤ for 16 h or were treated with IFN-␤ at a concentration of 3 ng/ml and collected at the indicated time points. Samples were analyzed by Western blotting for expression of IRAV and IFIT3. Analysis showed IRAV to be detectable in unstimulated HeLa cells and to be upregulated in response to treatment with IFN-␤ in both dose-dependent (Fig. 1D) and time-dependent (Fig. 1E)  that monocytes and U937 and Daudi cells produced an additional low-molecular-mass band of approximately 26 kDa that matched the predicted molecular mass of an alternate IRAV isoform (Fig. 1F). CRISPR-mediated knockout of IRAV results in increased titers of DENV and EMCV. The CRISPR/Cas9 system was used to generate A549 IRAV KO cells. Treatment of A549 cells with IFN-␤ resulted in upregulation of IRAV, as well as of other ISGs. However, in the KO cells, IRAV was not detected at either the RNA ( Fig. 2A) or protein (Fig. 2B) level, while expression of other ISGs remained unaffected. Infection of A549 or IRAV KO cells with DENV ( Fig. 2C) resulted in increased titers of DENV (Fig. 2D), as well as significant increases in DENV RNA relative to control cells (Fig. 2E). Similar effects were observed after infection with EMCV ( Fig. 2F and G). When DENV-infected IRAV KO cells were examined for expression of other genes (Fig. 2H), we observed increased expression of the IFIT3, Mx1, IRF9, and IFN-␤ ISGs in IRAV KO cells relative to the control A549 cells. These results suggest that IRAV directly affects DENV replication rather than exerting its effects via regulation of other ISGs.
IRAV associates with the dengue virus replication complex. To determine if IRAV has a direct influence on virus replication, HEK293 cells were stably transfected with an expression vector encoding IRAV with an amino-terminal (N-terminal) green fluorescent protein (GFP) fusion or with a negative-control vector with an N-terminal GFP tag. Because IRAV is poorly expressed in HEK293 cells (Fig. 1F), they are an ideal cell line for overexpression experiments. Cells were infected with DENV (multiplicity of infection [MOI], 0.1), and supernatants were collected at time zero and at 72 h postinfection, and the titer was determined by plaque assay. While detectable in the negative-control group, no virus was detected in cells stably expressing IRAV (Fig. 3A). The IRAVexpressing HEK293 cells were also treated with a small interfering RNA (siRNA) targeting IRAV or with a negative-control siRNA. The cells were then infected with DENV at an MOI of 0.1 for 72 h. Virus titration showed a significant increase in virus titers in IRAV-expressing HEK293 cells after knockdown of IRAV (Fig. 3B), demonstrating that IRAV has a significant influence on DENV replication even in the absence of interferon treatment. Because IRAV was previously identified by mass spectrometry (MS) as an interaction partner for HCV NS3 (34), we examined whether IRAV was interacting with DENV proteins. Coimmunoprecipitation (co-IP) experiments were performed on DENVinfected cells using GFP-IRAV as bait and GFP-chloramphenicol acetyltransferase (CAT) as a negative control. Western blotting showed that the DENV protease/helicase NS3 and the replication complex-associated protein NS4A interacted with IRAV, while the DENV envelope (E) protein did not (Fig. 3C). Interactions between IRAV and NS3 were not affected by RNase A treatment (data not shown). To further substantiate our findings, colocalization experiments were performed in A549 cells. Here, A549 cells were infected with DENV serotype 2 (DENV-2) for 48 h, followed by fixation and immunostaining for IRAV and either DENV NS3 (Fig. 3D) or NS4A (Fig. 3E) protein. While IRAV was localized to small puncta in mock-infected cells, colocalization between IRAV and NS3 or NS4A was observed in perinuclear regions of DENV-infected cells (Fig. 3F). This suggests that IRAV relocalizes to the DENV replication complex after infection. In order to demonstrate association of IRAV with the DENV replication complex in primary human cells, monocyte-derived macrophages were infected with DENV-2 as described above. As shown for A549 cells, colocalization was observed between IRAV and DENV NS3 (Fig. 3G) or NS4A (Fig. 3H) in DENV-infected monocyte-derived macrophages.
IRAV associates with host RNA binding proteins. To determine if IRAV interacted with host proteins in DENV-infected HEK293 cells, co-IP experiments were performed using overexpressed IRAV as bait. The immunoprecipitated proteins were then analyzed by MS. Gene ontology (GO) analysis was performed using the PANTHER classification system (http://pantherdb.org) (35) on MS hits with a spectral count of Ͼ50. Ontology derived based on the protein class revealed a strong preference (74%) for interactions with nucleic acid binding proteins (Fig. 4A). To characterize the nucleic acid binding properties of IRAV, fluorescence polarization (FP) assays were performed using Here, dilutions of IRAV were incubated with either synthetic single-stranded (ssRNA) or double-stranded (dsRNA) RNA or single-or double-stranded DNA containing complementary nucleic acid sequences. FP showed IRAV to be a nucleic acid binding protein with a higher affinity for ssRNA and ssDNA than for dsDNA (Fig. 4B).
To further explore interactions between IRAV and proteins associated with RNA processing, MS hits were ranked based on the spectral count, and the top-scoring proteins were further characterized (see Table S1 in the supplemental material). We were able to verify interactions between IRAV and MOV10, UPF1, and HuR by co-IP of overexpressed IRAV, followed by Western blotting (Fig. 4C). Because MOV10, UPF1, and HuR are all known to be RNA binding proteins, we examined the influence of RNase A treatment on these interactions. RNase treatment resulted in either complete or partial loss of interactions with IRAV (Fig. 4D). Partial loss of interactions due to RNase A treatment was previously described for interactions between MOV10 and UPF1 (36).
IRAV colocalizes with P bodies. Because many of the RNA binding proteins shown to interact with IRAV are associated with ribonucleoprotein (RNP) granules, we next investigated whether IRAV colocalized with specific RNP granule markers. Here, we examined A549 cells for the localization of P body markers (XRN1, DCP1a, and DDX6/ RCK), as well as the stress granule marker G3BP1 (Fig. 5A). We observed colocalization between IRAV and all three P body markers in IFN-treated cells; however, no statistically significant colocalization was observed between IRAV and G3BP1, indicating that IRAV associated with P bodies in the cytoplasm of IFN-treated cells (Fig. 5B). To characterize the association of IRAV with P body components after DENV infection, A549 cells were infected with DENV, followed by staining for IRAV and XRN1 ( Fig. 6A and B). As previously described for XRN1 (37), IRAV relocalized to the DENV replication complex, colocalizing with DENV NS3.
IRAV interacts with MOV10. Due to its known role as a restriction factor for retroviruses (38)(39)(40)(41), hepatitis C virus (6), and hepatitis B virus (42), we further characterized interactions between MOV10 and IRAV. Co-IPs and reciprocal co-IPs were performed using HEK293 cells transfected with either IRAV (GFP-IRAV) or a negative control (GFP-CAT). Co-IPs were then performed using an antibody either to overexpressed IRAV or to endogenous MOV10, and Western blots were analyzed for the presence of either MOV10 or IRAV (Fig. 7A). In addition, confocal microscopy and fluorescence resonance energy transfer (FRET) by fluorescence lifetime imaging (FLIM) analyses were used to characterize interactions between endogenous IRAV and MOV10 in A549 cells. Confocal microscopy revealed colocalization between IRAV and MOV10, as determined by Pearson's linear correlation coefficient ( Fig. 7B and C). Moreover, FRET-by-FLIM analysis demonstrated interactions between IRAV and MOV10, with lifetime values of 1.8 to 2.1 compared to values of 2.5 to 2.8 for the donor control, representing 25% to 28% FRET efficiency. The endosomal marker Rab5 was used as a negative control. Rab5 did not demonstrate significant colocalization with IRAV or a shift in lifetime values ( Fig. 7D and E).
Confocal microscopy was also used to examine localization of IRAV and MOV10 in DENV-infected and mock-infected A549 cells ( Fig. 8A and B). Here, it was observed that, similar to XRN1, both IRAV and MOV10 relocalized to the viral replication complex, colocalizing with DENV NS3. To further substantiate a role for MOV10 in restriction of  (Continued on next page) DENV replication, siRNA-mediated knockdown experiments were performed in A549 and IRAV KO cells (Fig. 8C). Depletion of MOV10 resulted in an increase in virus replication for DENV, as well as for EMCV, which was significantly greater in IRAV KO cells than in IRAV-expressing A549 cells (Fig. 8D and E), indicating that both IRAV and MOV10 play roles in the restriction of virus replication.

DISCUSSION
Interferon-stimulated genes are key mediators of antiviral immunity and play a pivotal role in the immune response to DENV infection. While the IFN response is an innate and broad-spectrum response to many pathogens, ISGs have been demonstrated to have remarkable specificity both for the viruses they target and the pathways through which they inhibit viral replication (6). Hundreds of ISGs have been shown to be upregulated in response to induction of IFN signaling pathways, illustrating the  scope and complexity of the IFN response. To date, only a fraction of these genes have been characterized.
Here, we show that IRAV is an IFN-stimulated gene with antiviral activity against DENV. Knockout of the IFN signaling component IRF9 results in a significant reduction of IRAV expression after IFN treatment, suggesting regulation through the canonical ISGF3 complex; however, it should be noted that there appears to be some baseline constitutive expression of IRAV. In addition, IRAV homologues were identified in hemichordates and echinoderms, organisms that lack a "classical" IFN response, suggesting the possibility of IFN-independent functions for IRAV-like proteins.
IRAV associates with P bodies, as evidenced by interactions with P body-associated proteins (MOV10 and UPF1) and colocalization with P body markers (DCP1a, XRN1, and DDX6). P bodies are discrete cytoplasmic structures composed of RNA and proteins that are associated with various aspects of RNA turnover, including nonsense-mediated decay, adenylate-uridylate-rich element (ARE)-mediated decay, and gene silencing (43). Notably, both DENV and West Nile virus (WNV) have been shown to associate with components of P bodies. Subgenomic flavivirus RNA (sfRNA) inhibits the activity of XRN1 and TRIM25 (44)(45)(46), and DDX6 interacts with DENV untranslated regions (UTRs) and localizes to DENV replication complexes, possibly playing a role in virus replication (47). Furthermore, WNV infection of HeLa cells results in a decrease of both the size and number of P bodies, along with a relocalization of P body components (including GW182, DDX3, and XRN1) to WNV replication complexes, colocalizing with NS3 (37). This is similar to what we observed with IRAV during DENV infection, with IRAV puncta becoming more diffuse and colocalizing with NS3 in distinct perinuclear complexes. A number of other viruses, including hepatitis C virus (48), poliovirus (49), influenza virus (50), and bunyaviruses (51,52), have been shown to interact with P bodies or with P body components (53).
MOV10 and UPF1 are both members of the SF1 family of helicases and have been previously shown to interact with one another and to form complexes with APOBEC3G and Argonaute 2 (40,54), as well as with the antiviral protein ZAP (55). Interactions between MOV10 and UPF1 are partially sensitive to RNase treatment (36), as we have also demonstrated for IRAV interactions with MOV10 or UPF1. Knockdown of MOV10 results in an increase in the half-lives of mRNA targets (36). MOV10 has been shown to be involved in retrotransposition of LINE elements (56)(57)(58), as well as in microRNA pathways (54,59). MOV10 may also play a role in mediating antiviral response (60) and restricts replication of retroviruses (38)(39)(40)(41), hepatitis C virus (6), hepatitis B virus (42), and influenza virus (61). UPF1 is a key component of nonsense-mediated RNA decay pathways (62) and has been demonstrated to play a role in viral replication cycles (63,64). IRAV also associates with the RNA binding protein HuR. HuR has been shown to be involved in RNA decay pathways (65) and has been linked to viral replication and modulation of the host response (66)(67)(68)(69). In addition, IRAV has been previously shown to interact with RNA binding proteins LARP1 and PABPC1, both of which have also been shown to be involved in RNA decay (23). Taken together, IRAV's role as an RNA binding protein, its association with proteins linked to RNA decay, and its localization to P bodies (sites of RNA decay) and the marked increase in viral RNA observed after knockout of IRAV all suggest a role for IRAV in degradation of viral RNA. The role of MOV10 in IRAV-mediated antiviral activity remains unclear; however, given MOV10's role as an RNA helicase involved in RNA decay pathways, as well as its localization at the viral replication complex and the increase in viral RNA observed after its knockdown, MOV10 may function in conjunction with IRAV and other proteins to aid in destabilization of viral RNA.
In conclusion, we show that IRAV is an ISG that is regulated through the canonical type I interferon signaling pathway. IRAV displays antiviral activity against DENV and EMCV and interacts with DENV proteins. IRAV is an RNA binding protein that localizes to P bodies, sites of RNA decay. Additionally, IRAV interacts with MOV10 and UPF1, two proteins previously shown to interact with each other and to be involved in RNA decay pathways. These data serve to identify an ISG with antiviral activity against DENV, as well as to suggest a mechanism of action involving the destabilization of viral RNA.
Oligomers were dissolved in binding buffer (100 mM KCl, 20 mM HEPES, pH 7.4), heated to 85°C for 10 min, and allowed to cool slowly. The oligomers were then added to serial dilutions of protein solution in triplicate, and FP was measured on a Corning black 96-well, half-area assay plate (Corning, NY) using a Hidex Sense Microplate Reader (Turku, Finland).
Confocal microscopy and fluorescence resonance energy transfer by fluorescence lifetime imaging. A549 cells were plated on poly-D-lysine-coated 35-mm culture dishes (MatTek, Ashland, MA) and either treated with IFN-␤ (10 ng/ml) for 16 h or infected with DENV-2 (MOI, 3) for 48 h. The cells were fixed with 4% paraformaldehyde in PBS and permeabilized with 0.1% or 0.5% Triton X-100, followed by blocking with 5% bovine serum albumin (BSA) for 30 min. Primary antibodies were diluted in 10% normal goat serum and incubated at room temperature for 2 h, followed by three washes in PBS and staining with fluorescently labeled secondary antibody (1:500) and nuclear DAPI (4=,6-diamidino-2-phenylindole) stain (Life Technologies). Primary labeled antibodies were diluted in 10% normal goat serum and incubated at room temperature for 2 h, followed by three washes in PBS, and postfixed with 4% paraformaldehyde in PBS. Images were collected on a Leica SP8 inverted confocal microscope with a 63ϫ oil immersion objective (Leica Microsystems, Buffalo Grove, IL). Colocalization analysis was performed using Imaris software (Bitplane Inc., South Windsor, CT). FRET-by-FLIM analysis was performed as previously described (73).
Gene ontology and statistical analyses. GO analysis was performed using proteins identified by MS with spectral counts above 50. Selected protein accessions were analyzed using the PANTHER classification system (http://pantherdb.org) (35). All statistical analyses were performed on Prism (GraphPad Software, Inc.), using one-way analysis of variance (ANOVA) with Tukey's post hoc test and a P value of 0.05.

ACKNOWLEDGMENTS
We thank Tsan Xiao and David Garboczi for their insights into protein structure; Ming Zhao and Renee Olano from the Research Technologies Branch, NIAID, NIH, for excellent technical assistance; Yajuan Li from the School of Life Sciences, University of Science and Technology of China, for assistance with protein purification; Daniel Green for providing human monocytes; Xavier Ambroggio and Vijayaraj Nagarajan for their help with computational analysis; Emerito Amaro-Carambot for technical expertise; Stephen Whitehead for providing reagents and support; and Sonja Best and Ted Pierson for their valuable insight and critical readings of the manuscript.
This work was supported by the Intramural Research Program of the NIAID, NIH.