Secreted Vago restricts West Nile virus infectionin Culex mosquito cells by activating theJak-STAT pathwayPrasad N. Paradkara, Lee Trinidada, Rhonda Voyseya, Jean-Bernard Ducheminb, and Peter J. Walkera,1

aCSIRO Animal, Food and Health Sciences, and bCSIRO Ecosystem Sciences, Australian Animal Health Laboratory, Geelong, VIC 3220 Australia

Edited by Charles M. Rice, The Rockefeller University, New York, NY, and approved September 12, 2012 (received for review March 28, 2012)

Although West Nile virus (WNV) and other arthropod-borne virusesare a major public health problem, the mechanisms of antiviralimmunity inmosquitoes are poorly understood. Dicer-2, responsiblefor the RNAi-mediated response through the C-terminal RNase-IIIdomain, also contains an N-terminal DExD/H-box helicase domainsimilar to mammalian RIG-I/MDA5 which, in Drosophila, was foundto be required for activation of an antiviral gene, Vago. Here weshow that the Culex orthologue of Vago (CxVago) is up-regulatedin response to WNV infection in a Dicer-2–dependent manner. Fur-ther, our data show that CxVago is a secreted peptide that restrictsWNV infection by activation of the Jak-STAT pathway. Thus, Vagoappears to function as an IFN-like antiviral cytokine in mosquitoes.

innate immunity | insect cytokine

Flaviviruses such as West Nile virus (WNV), transmitted by mos-quitoes, are a major risk to human and animal health worldwide.

More than one third of the world’s population is at risk for flavivirusinfection, and their geographical distribution is expanding as aresult of increased travel and trade and global climate change (1).WNV, first isolated in Uganda, is now endemic to parts of Africa,Europe, the Middle East, Asia, and the Americas (2). Currently,no specific therapy or vaccine has been approved for use againstWNV infection in humans. Although most research to date hasfocused on the human response to WNV infection, some recentstudies have identified insect immune pathways thatmay play a rolein restricting viral replication in mosquitoes (3). Insects and otherinvertebrates are believed to lack key components of vertebrateadaptive and innate immune systems such as antibodies, MHCreceptors, and various cytokines, including interferons (IFNs) (3).However, the evolutionarily conserved Jak-STAT pathway, whichsignals the mammalian IFN response, is induced by viral infectionin insects, and depletion of components of the pathway results inincreased susceptibility to infection (4, 5). InDrosophila Schneider2 cells, expression of virus-inducible gene vir-1 in response toDrosophilaC virus (DCV) infection has been shown to be under thecontrol of a STAT-binding site in the promoter element, and vir-1induction has been shown to be substantially reduced in hopscotch-deficient flies lacking the only known Janus kinase (4). However,overexpression or knockdown of vir-1 had no effect on resistance toDCV infection, and its function is presently unknown (4).Another important aspect of the insect antiviral response is the

RNAi pathway through which viral dsRNA is processed by thecarboxyl-terminal RNase-III domain of Dicer-2 (6-8). A recentstudy inDrosophilahas shown that the amino terminalDExD/H-boxhelicase domain of Dicer-2, which is similar to the mammaliandsRNA-sensing RIG-I/MDA5 helicase, is responsible for RNAi-independent activation of a novel antiviral gene, vago (9). Identifiedin the microarray screen, vago mutant flies showed increased viralload in fat bodies. Drosophila vago encodes an 18-kDa polypeptide(DmVago) containing a single vonWillebrand factor typeC (VWC)domain, featuring eight conserved cysteine residues (9). Found onlyin arthropods, proteins containing a singleVWCdomain respond toenvironmental changes such as bacterial infections and nutritionalstatus (10, 11). However, the mode of antiviral action of Vago is not

known, and the significance of this Dicer-2–dependent pathway ina medically relevant infection model has not been determined toour knowledge.WNV is generally transmitted in a bird–mosquito cycle, with oc-

casional infections of humans and horses, which are considereddead-end hosts. Many species of mosquitoes have been shown tohost WNV infection, although most isolations have been fromculicine mosquitoes (12). Here we use a WNV-Culex cell infectionmodel to study the function of vago. We show that the Culexorthologue of DmVago is a secreted peptide that is up-regulated inresponse to WNV infection in a Dicer-2–dependent manner. Wealsodemonstrate that, likemammalian IFNs, secretedVago restrictsWNV infection in Culex cells by activating the Jak-STAT pathwayand up-regulates expression of the STAT-dependent virus-induciblegene vir-1. Vago, therefore, appears to function as a cytokine thatacts in a manner that is similar to mammalian interferons.

ResultsMosquito Orthologues of Drosophila Vago. As Culexmosquitoes areknown vectors for transmission of WNV, a cell line derived fromCulex quinquefasciatus (Hsu) was used to determine the significanceof Vago activation in the mosquito antiviral response (13). By usingVectorBase (http://www.vectorbase.org) (14), salivary cysteine-richsecreted peptide (XP_001842264)was identified as an orthologue ofDmVago in the Cx quinquefasciatus genome (Fig. 1A). The 113-aapeptide showed ∼30% identity to DmVago at the amino acid level,with eight conserved cysteine residues forming a VWC domain.Orthologous proteins in Aedes and Anopheles mosquito genomeswere also identified. Phylogenetic analysis of Vago-related genesindicated that the selected Cx quinquefasciatus peptide and the Aeaegyptii orthologue XP_001658930 were most closely related toDmVago (SI Appendix, Fig. S1A). By using antibody preparedagainst Aedes Vago, CxVago was detected in adult Culex pipiens f.molestus mosquitoes by Western blotting, with expression levelshigher in female than male mosquitoes (SI Appendix, Fig. S1B).

CxVago Induction by WNV Is Dicer-2–Dependent. To determine ifCxVago mRNA and protein levels are up-regulated followingWNV (NY99-4132 strain) infection, Hsu cells were infected withWNV [multiplicity of infection (MOI) of 5] and total RNA orprotein lysates were collected at 0, 12, 24, and 48 h postinfection(hpi). Real-time quantitative RT-PCR (qRT-PCR) using primersspecific for CxVago showed increased Vago mRNA levels within12 hpi, with approximately three- to fourfold increase at 48 hpi

Author contributions: P.N.P., L.T., R.V., J.-B.D., and P.J.W. designed research; P.N.P., L.T., R.V.,and J.-B.D. performed research; P.N.P. and P.J.W. analyzed data; and P.N.P. and P.J.W.wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

See Commentary on page 18639.1To whom correspondence should be addressed. E-mail: peter.walker@csiro.au.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1205231109/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1205231109 PNAS | November 13, 2012 | vol. 109 | no. 46 | 18915–18920

MICRO

BIOLO

GY

SEECO

MMEN

TARY

(Fig. 1B). Similar results were observed following WNV infectionof Culex tarsalis (Chao-Ball) cells (SI Appendix, Fig. S2A).Western blot analysis also showed an increase in Vago proteinlevels in a time-dependent manner (Fig. 1C). UV-inactivatedWNV failed to induce CxVago mRNA expression, indicatinga requirement of replicating virus for activation of Vago (Fig.1B). The induction of Vago expression was also tested followingmicroinjection of female Culex pipiens f. molestus mosquitoeswith 1.0 × 103 pfu of WNV (Kunjin strain). CxVago expression,as assayed by RT-qPCR, was induced 3.5 fold by 24 hpi and thelevels returned to baseline by 8 d postinfection (Fig. 1F). WNVinfection and virus replication were confirmed by plaque assaywith a mean recovery of 1.37 × 106 pfu/mosquito at 8 d.As an earlier report (9) had shown that DmVago mRNA was

induced after infection by Dicer-2 activation, Hsu cells weretransfected with CxDicer-2 dsRNA to silence the gene. Effectivesilencing of Dicer-2 mRNA and the effect of WNV infection onCxVagomRNA expression in Dicer-2–silenced cells were assessedby using RT-qPCR. The results showed more than fivefold de-crease in Dicer-2 mRNA levels, indicating effective silencing (Fig.1D). There was significantly less increase in CxVago mRNA levelsafterWNV infection in Dicer-2–silenced cells compared with cellstransfected with GFP dsRNA (Fig. 1E), indicating that Dicer-2 isrequired for WNV-induced up-regulation of CxVago. We alsoconfirmed that the RNAi pathway was not directly involved inVago induction by demonstrating that it was not affected byR2D2 knockdown (SI Appendix, Fig. S3B and Fig. S8).

CxVago Inhibits WNV Infection in Mosquito Cells. Previous studieshad shown that ΔVago-mutant Drosophila developed higher viral

loads in fat bodies compared with WT flies (9). To determinewhether CxVago has antiviral action, Hsu cells were transfectedwith a plasmid expressing V5-tagged CxVago and infected withWNV (MOI of 5) at 24 h posttransfection. The cells and super-natant medium were harvested at 48 hpi. Western blot performedwith anti-V5 antibody showed overexpression of CxVago (Fig.2A), and RT-qPCR with WNV NS1-specific primers showeddecreased levels of WNV NS1 gene in cells at 48 and 72 hpi (Fig.2B), indicating decreased viral replication in Vago-overexpressingcells. Plaque assays (Fig. 2C) showed ∼40-fold decrease in viraltiter at 72 hpi in the supernatant medium from cells over-expressing Vago (6.4 × 102 pfu/mL) compared with those trans-fected with the empty vector (2.5 × 104 pfu/mL).To determine the effect of Vago knockdown on viral titer, Hsu

cells transfected with the CxVago dsRNA or GFP dsRNA(nonspecific control) were infected with WNV. The cells and thesupernatant medium were harvested at 48 hpi. RT-qPCR wasperformed using Vago-specific and NS1-specific primers, andWestern blot was performed with anti-Vago antibody. The resultsshowed significant decreases in Vago protein (Fig. 3A) andmRNA (Fig. 3B) levels in cells transfected with Vago dsRNA andincreased levels of NS1 RNA (Fig. 3C). Plaque assays performedon the supernatant medium collected from the cells at 48 hpi(Fig. 3D) showed more than 25-fold higher viral titers from cellssilenced for CxVago (5.5 × 104 pfu/mL) compared with controlcells transfected with GFP dsRNA (2.1 × 103 pfu/mL). Theresults of overexpression and gene silencing experiments suggestCxVago has antiviral activity in WNV-infected mosquito cellcultures. Interestingly, overexpression of Aedes albopictus Vago(AaVago) in CxVago-silenced cells showed decreased viral titers

Fig. 1. CxVago is induced by WNV infection in a Dicer-2–dependent manner. (A) Clustal alignment of the amino acid sequences of Vago from A. albopictus(Aa_Vago), Cx quinquefasciatus (Cq_Vago), Anopheles gambiae (Ag_Vago) and Drosophila melanogaster (Dm_Vago). (B) RT-qPCR performed by usingCxVago primers on Hsu cells infected with WNV (NY99-4132 strain) at various times after infection. Hsu cells treated with UV-inactivated WNV were alsotested at 48 h posttreatment. CxRpL32 mRNA (ribosomal protein L32) was used as an internal control. (C) Western blot performed on lysates from cells treatedas described earlier with the use of anti-Vago and anti–β-actin antibodies. (D) RT-qPCR using CxDicer-2 primers performed on cells transfected with Dicer-2dsRNA. (E) RT-qPCR performed by using CxVago primers on cells transfected with Dicer-2 dsRNA. (F) RT-qPCR performed at various time postinfection by usingCxVago primers on adult female Culex mosquitoes microinjected with WNV (Kunjin strain). For RT-qPCR assays, ΔΔCt was calculated and fold increase vs.controls was plotted. Error bars represents SE from three separate experiments with assays performed in triplicate (*P < 0.05, Student t test).

18916 | www.pnas.org/cgi/doi/10.1073/pnas.1205231109 Paradkar et al.

(7.5 × 102 pfu/mL), demonstrating conserved functionality ofVago in Culex and Aedes mosquitoes (Fig. 3D).

CxVago Is Secreted. SignalP 4.0 (15) analysis of the CxVago aminoacid sequence revealed a predicted signal peptide cleavage sitebetween amino acids 20 and 21, suggesting that CxVago may besecreted (SI Appendix, Fig. S5A). Competition ELISA performedon the supernatant medium of WNV-infected Hsu cells usinganti-Vago antibody indicated WNV infection induced a morethan fourfold increase in secreted Vago at 48 hpi (Fig. 4A), andthere was a significant decrease in Vago secretion from WNV-infected cells following Dicer-2 knockdown compared with con-trol cells treated with GFP dsRNA (Fig. 4A). This indicates thatDicer-2–mediated induction of Vago leads to its secretion fromCulex cells.

Secreted CxVago Has Antiviral Activity. Experiments were con-ducted to determine the antiviral action of secreted Vago. Su-pernatant medium was collected from Hsu cells overexpressingCxVago (without V5-tag) at 48 h posttransfection and was addedto fresh cultures of Hsu cells infected simultaneously with WNV(MOI of 5). The supernatant medium was collected at 48 hpi,and viral titers were measured by plaque assay. The results (Fig.4B) showed a 30-fold reduction in viral titers (1.1 × 102 pfu/mL) insupernatants from cells treated with CxVago compared with themedium from untreated cells (3.5 × 103 pfu/mL). Treatment withmedium from GFP-transfected cells showed viral titers similar tocontrol cells (3.2 × 103 pfu/mL). The results indicate that the se-creted form of Vago restricts WNV replication in Culex cells. Tofurther confirm the antiviral action of secreted Vago, Hsu cellswere transfected with plasmid containing CxVago, in which thepredicted signal peptide cleavage site was mutated (Ala20 > Thr),abolishing Vago secretion (SI Appendix, Fig. S5C). The superna-tant medium was collected at 48 h posttransfection and added tofresh cultures of Hsu cells infected simultaneously with WNV.Plaque assays indicated significantly higher titers in the supernatantmedium from these cells compared with the medium from cells

transfected with plasmid expressing WT CxVago (Fig. 4B), con-firming the antiviral action of secreted Vago.Similar experiments were conducted to compare the effect of

preinfection and postinfection treatment of Hsu cells withCxVago.Cells treated with supernatant medium containing secretedCxVago2 h before infection or simultaneously with WNV were protectedto the same extent, showing 30- and 25-fold lower viral titers,respectively, by plaque assays at 48 hpi compared with cellstreated the medium control. However, cells treated with CxVagoat 2 hpi showed no significant difference in viral titer comparedwith control cells (Fig. 4C), indicating that CxVago is effectiveonly in the initial stages of WNV infection. RT-qPCR assaysusing vir-1 and NS1 gene-specific primers confirmed that viralreplication in Hsu cells was significantly reduced relative tocontrols by CxVago 2 h pretreatment but not by treatment 2 hpi.Furthermore, although vir-1 was induced by both CxVago pre-treatment and postinfection treatment, the level of induction wassignificantly lower in cells treated postinfection treatment andsimilar to the level of induction in cells infected with WNV in theabsence of CxVago (Fig. 4D). This suggests that the Jak-STATpathway was activated by CxVago postinfection treatment but itfailed to restrict WNV infection. African green monkey (Vero)cells were also treated with supernatant medium from CxVago-overexpressing cells simultaneously with WNV infection. Plaqueassays showed no change in the viral titers in these cells comparedwith control medium-treated cells (SI Appendix, Fig. S5E), in-dicating that CxVago does not affect WNV infectivity per seand that the target of CxVago is insect cell-specific.

Fig. 2. Antiviral action of overexpressed CxVago. (A) Hsu cells transfectedwith plasmids containing GFP (control) or V5-tagged CxVago (Vago) 24 hbefore WNV (NY99-4132 strain) infection and probed at 48 hpi with anti-V5,anti-Vago, and anti–β-actin antibodies. (B) RT-qPCR using WNV NS1 primerson nucleic acid extracts from these cells at 48 hpi and 72 hpi. (*Significantdifference between 48 h and 72 h.) (C) Viral titer estimation by plaque assaysconducted on supernatant media from these cells at 48 hpi. Error bars rep-resent SE from three separate experiments with assays performed in tripli-cate (*P < 0.05, Student t test).

Fig. 3. Increased viral load in CxVago silenced cells. (A) Hsu cells transfectedwith GFP dsRNA (control) or CxVago dsRNA (Vago) 24 h before WNV (NY99-4132 strain) infection and probed at 48 hpi using anti-Vago and anti–β-actinantibodies to confirm gene silencing. (B) RT-qPCR was performed by usingVago-specific primers to determine effective silencing. ΔΔCt was calculatedbased on the internal control (RpL32) and fold change vs. control wasplotted. Error bars represent three separate experiments performed intriplicate. (C) RT-qPCR using WNV NS1 primers on nucleic acid extracts fromthese cells at 48 hpi. (D) Viral titer estimation by plaque assays conducted onsupernatant media from these cells at 48 hpi. Plaque assays were also con-ducted at 48 hpi on cells silenced for CxVago and transfected with AaVagoto assess complementation (dsRNA CxVago + AaVago + WNV). Error barsrepresent SE from three separate experiments with assays performed intriplicate (*P < 0.05, Student t test).

Paradkar et al. PNAS | November 13, 2012 | vol. 109 | no. 46 | 18917

MICRO

BIOLO

GY

SEECO

MMEN

TARY

Secreted CxVago Activates the Jak-STAT Pathway to Restrict WNVInfection. Experiments were conducted to determine if the antivi-ral activity of secreted CxVago occurs through receptor-mediatedactivation of a cell signaling pathway. RT-qPCR assays were per-formed on Hsu cells treated with secreted CxVago using markersfor Jak-STAT (vir-1) activation (4) and NF-kB (defensin-A) acti-vation (16). Hsu cells treated for 24 h with supernatant mediumcontaining secreted CxVago showed two- to threefold up-regula-tion of vir-1 expression but no significant change in defensin-Aexpression (Fig. 5A). To determine whether activation of the Jak-STAT pathway is responsible for CxVago antiviral activity, Hsucells transfected with CxSTAT2 dsRNA were infected with WNVand treated simultaneously with the supernatant medium con-taining secreted CxVago. RT-qPCR assays confirmed effectivesilencing of CxSTAT2 and no induction of the STAT marker gene(vir-1) after STAT silencing (Fig. 5C). Plaque assays conducted at48 hpi showed 25- to 30-fold higher viral titer in the supernatantfrom cells transfected with CxSTAT2 dsRNA (7.8 × 103 pfu/mL)compared with control dsRNA (i.e., GFP; 2.8 × 102 pfu/mL; Fig.5B), indicating that the antiviral activity ofCxVago is dependent onCxSTAT2. Silencing of CxJak (DmHopscotch orthologue) alsoprevented activation of vir-1 and abolished antiviral activity of se-creted CxVago treatment (SI Appendix, Fig. S6). These data in-dicate that CxVago activates the Jak-STAT pathway, stimulating

expression of downstream genes that restrict WNV replication inmosquito cells.In Drosophila, Jak-STAT activation occurs when the ligand

“unpaired” (UPD) binds to the receptor Domeless on the cellsurface (17). To determine whether this receptor is involved inVago-induced signaling in mosquitoes, the Culex orthologue ofDrosophila Domeless (CxDome) was silenced by transfection ofHsu cells with CxDome dsRNA. After 24 h, the cells wereinfected with WNV and treated simultaneously with supernatantmedium containing secreted CxVago. RT-qPCR assays con-ducted to detect expression of Dome, vir-1, and WNV NS1 asa marker of virus replication, confirmed that there was effectivesilencing of CxDome but that silencing had no significant effecton CxVago-induced vir-1 expression or on CxVago-induced sup-pression of WNV replication (Fig. 5D). Although some leakage ofCxDome-mediated signaling following knockdown cannot be ex-cluded, the data suggest that CxVago-stimulated vir-1 expression isindependent of Dome and that the antiviral response occurs via analternative Jak receptor.

DiscussionDespite the importance of arthropods as vectors of disease, the im-mune response of arthropods to viral infection is poorly understood.Previous studies in Drosophila and mosquitoes have established theimportance of RNAi (6–8) and demonstrated that the Toll, Imd, andJak-STAT signaling pathways have roles in antiviral immunity (4, 5,18, 19). Evidence in Drosophila that dsRNA sensing by the Dicer-2DExD/H-box helicase domain stimulates an antiviral response me-diated by the cysteine-rich peptide Vago has suggested the existenceof a pathway that may function similarly to the RIG-I/MDA5-acti-vated IFN response in vertebrates. However, there has been noprevious evidence in insects or other invertebrates of cytokine activitythrough which the innate antiviral response is communicated locallyor systemically (3). Here, we have identified the Culex orthologue ofDmVago and determined its mode of action during WNV infectionof mosquito cells. Our data demonstrate that CxVago is a stable,secreted cytokine that stimulates an insect cell-specific antiviral re-sponse by activating the Jak-STAT pathway. We demonstrate thatVago induction in mosquitoes occurs by a Dicer-2–dependentmechanism, elaborating the significance of the previous work inDrosophila (9), and that Vago activates expression of the Culexorthologueof the virus-inducibleDrosophila gene vir-1.Wealso showthat this antiviral response requires viable replicating WNV. Thus,although Vago displays no structural homology with mammalianIFNs, it appears to serve a similar function in mosquitoes.In contrast to our results, Deddouche et al. (9) were unable to

establish a link between DCV-stimulated Vago up-regulation andthe induction of vir-1 in Drosophila. They observed that vir-1remained fully inducible in ΔVago-mutant flies and that suppres-sion of RNAi by the dsRNA-binding Flock House virus B2 proteindecreased DCV-induced up-regulation of Vago but did not sup-press vir-1 induction. Furthermore, it was argued that the fat bodytissue specificity of Vago gene expression, sequence divergenceamong Drosophila species, and lack of conservation of the gene inother insects reinforce the view that it does not function as a cir-culating cytokine. However, direct functional analysis was notpossible in previous studies as a result of the poor stability ofexpressed DmVago, and it could be argued that induction of vir-1in the absence of a Vago response may occur by an alternativemechanism. In mammalian cells, the Jak-STAT pathway can beactivated by a number of different cytokines including IFNs, ILs,and various growth factors (20, 21).Whereas IFNs are activated bydsRNA-sensing via RIG-I/MDA5 or Toll-like receptors (e.g.,TLR3) (22), ILs are activated by ssRNA via TLR7/8 (23). Argu-ably, the Jak-STAT pathway in insects may also be activated bymultiple cytokines induced by various virus-specific molecularpatterns. If so, vir-1 could be induced by Vago in response to viral

Fig. 4. CxVago is a secreted antiviral peptide. (A) Competition ELISA fordetection of CxVago performed on supernatant media collected at 48 hpifrom mock-infected Hsu cells (control), WNV (NY99-4132 strain)-infected Hsucells, and Hsu cells transfected with Dicer-2 dsRNA 24 h before WNV in-fection (DCR2-dsRNA + WNV). Fold increase was calculated and plotted overcontrol (baseline). (B) Viral titers determined by plaque assays on supernatantmedia collected from WNV-infected Hsu cells at 48 hpi. Cells were treated atthe time of infection with WNV alone or supernatant media collected fromHsu cells overexpressing WT CxVago (CxVago media + WNV) or CxVago witha mutation (A-T) in the predicted signal peptide cleavage site [CxVago (A-T)media + WNV]. (C) Viral titers determined by plaque assays on supernatantmedia collected from WNV-infected Hsu cells at 48 hpi. Cells were treated atthe virus alone (i.e., WNV) or WNV and supernatant media collected from Hsucells overexpressing WT CxVago at the time of infection (CxVago +WNV), 2 hpreinfection (Pre-CxVago +WNV), or 2 hpi (Post-CxVago +WNV). (D) RT-qPCRperformed by using primers specific for vir-1 and WNV NS1 at 48 hpi on cellstreated as in C. Fold increase was calculated and plotted vs. control (baseline).Error bars represent SE from three separate experiments with assays per-formed in triplicate (*P < 0.05, Student t test).

18918 | www.pnas.org/cgi/doi/10.1073/pnas.1205231109 Paradkar et al.

infection but still remain fully inducible by other, yet unidentified,virus-activated cytokines in the absence of Vago expression.Our data also demonstrate that the antiviral function of Vago is

preserved in Drosophila and mosquitoes despite the absence ofextensive sequence homology, and we have shown that AedesVagois functionally active in Culex cells. This suggests the critical struc-ture of the receptor-binding region may be defined by a secondarystructural fold stabilized by the eight conserved cysteine residuesand antiparallel β-sheets of the of the VWC domain (10). Thesingle VWC domain (SVC) protein family appears to occur only inarthropods, including mosquitoes, flies, ticks, midges, moths, bee-tles, bees, scorpions, and shrimp (10, 24). Several of the 13 SVCfamily members identified in Drosophila are expressed only inspecific tissues, including the fat body, which regulates immune andnutritional processes, and the salivary gland, in which virus repli-cation efficiency is a critical factor in arbovirus transmission (10).SVC genes have also been shown to be under nutritional control,and it has been proposed that they may be induced in response toenvironmental stress (10, 11). However, it is not clear at this time ifantiviral activity is common function among SVC proteins andwhether, like mammalian IFNs, the various isoforms may differ-entially regulate various aspects of the antiviral response.Analysis of the kinetics of the antiviral activity of secreted

recombinant CxVago indicated that it was effective only whendelivered simultaneously or up to 2 h before WNV infection.Vago treatment at 2 hpi resulted in lower levels of vir-1 activa-tion and failed to inhibit WNV infection. A similar effect hasbeen observed during flavivirus infection of mammalian cells inwhich in IFN treatment was also far less effective in blockingviral replication and other antiviral functions when delivered 4hpi (25). In flaviviruses and other RNA viruses, the failure ofIFN treatment postinfection has been attributed to virus-inducedshut-down of host cell protein synthesis and the action of virus-encoded antagonists that block Jak-STAT–mediated expressionof IFN-inducible genes (26, 27).Further studies are required to identify the Vago receptor.

Previous experiments have indicated that DCV-induced vir-1 ex-pression is regulated byDomeless (4), encoding the only Jak-STATreceptor identified in Drosophila to date and an orthologue ofmammalian class I cytokine receptors. UPD-like secreted proteins

have been implicated as Domeless ligands (17, 28, 29), but theyshare no detectable structural homology with Vago and in silicosearches have failed to detect obvious UPD orthologues in mos-quitoes or other invertebrates (30). Furthermore, our data in-dicated that Domeless knockdown had no significant effect onCxVago-induced vir-1 expression or CxVago-induced suppressionof WNV replication. Taken together, these observations suggestthat Vago may stimulate Jak-STAT signaling and the antiviralresponse via an alternative, yet unidentified, receptor. The down-stream STAT-activated antiviral effector genes are also unknown.Overexpression or knockdown of vir-1 has been reported to haveno effect on DCV resistance in Drosophila, and its functionremains unclear (4). Our studies have also indicated that over-expression of CxVir-1 in Hsu cells does not lead to reduced viraltiters (SI Appendix, Fig. S7). A large number of other genes havebeen reported to be up-regulated in arthropods in response to viralinfection (4, 31, 32). A recent report has shown that a cecropin-likepeptide induced in mosquito salivary glands in response for Den-gue virus infection has broad-spectrum antiviral and antibacterialactivity, but the mechanisms of its induction and antiviral actionare unknown (33).The existence in insects of an immune pathway that appears

similar to the vertebrate IFN system, permitting cell-to-cell com-munication of the antiviral response, provides a complementarydefensive mechanism to the systemic RNAi response, which hasbeen shown to inhibit replication and limit virus dissemination (34,35). Further characterization of this antiviral pathway may assist inunderstanding vector competence and the dynamics of emergenceof flaviviruses and other important vector-borne pathogens, andmay present opportunities for the development of novel inter-ventions for disease control.

Materials and MethodsCell Culture and Virus Propagation. Hsu, Vero, BHK, and C6/36 cells weremaintained and viruses were propagated as detailed in SI Appendix.

Mosquito Maintenance and Viral Infections. A colony of Culex pipiens f.molestus mosquitoes was maintained at 25 °C and 65% humidity. Five-day-old mosquitoes were infected and further analyzed as detailed inSI Appendix.

Fig. 5. CxVago restricts WNV by activating the Jak-STAT pathway. (A) RT-qPCR using primers specific for vir-1 (dark gray bar) and defensin-A (light gray bar)mRNA performed on nucleic acid extracts from untreated Hsu cells (control) and cells treated for 24 h (CxVago-24h) and 48 h (CxVago-48h) with supernatantmedium from Hsu cells overexpressing CxVago. (B) Plaque assay to determine viral titers at 48 hpi in supernatant media of Hsu cells untreated (control),infected with WNV (NY99-4132 strain) and treated simultaneously with supernatant medium from Hsu cells overexpressing CxVago (CxVago + WNV), orWNV-infected and treated with CxVago supernatant medium and transfected 24 h before infection with CxSTAT2 dsRNA (CxVago + WNV + dsRNASTAT) orcontrol GFP dsRNA (CxVago + WNV + dsRNAGFP). (C) RT-qPCR using primers specific for STAT2 (dark gray bar) and vir-1 (light gray bar) performed on nucleicacid extracts of cells treated as in B. (D) Hsu cells silenced with dsRNA against CxDomeless receptor (dsRNA dome) were treated with supernatant mediumcollected from Hsu cells overexpressing WT CxVago at the time of infection (WNV + CxVago) and, at 48 hpi, RT-qPCR was performed by using primers specificfor vir-1, WNV NS1 or CxDomeless. Fold increase was calculated and plotted vs. control (baseline). Error bars in RT-qPCR assays represent SE from threeseparate experiments with assays performed in triplicate (*P < 0.05, Student t test).

Paradkar et al. PNAS | November 13, 2012 | vol. 109 | no. 46 | 18919

MICRO

BIOLO

GY

SEECO

MMEN

TARY

Plaque Assays. Viral titers were determined by plaque assay on Vero cells asdetailed in SI Appendix. Plaques were counted and titers were estimated bydilution factor.

Western Blotting. Anti-Vago antibody (GenScript) was prepared in rabbits byusing the C-terminal peptide CEKIKQDLTKDYPE. Anti-V5 antibody was pur-chased from Invitrogen. Anti–β-actin antibody was purchased from Abcam.Western blot was performed as detailed in SI Appendix.

Viral dsRNA Preparation and Transfection. Total RNAwas extracted fromWNV-infected Hsu cells or bluetongue virus-infected BHK cells at 48 hpi by usingTRIzol (Invitrogen) according to the manufacturer’s instructions. ssRNA wasremoved by precipitation with 2 M LiCl and centrifugation at 16,500 × g for10 min. dsRNA in the supernatant fraction was precipitated with isopropanoland purified by using an ethanol wash. Hsu cells were transfected with thepurified dsRNA or poly I:C (GE Healthcare) in Cellfectin (Invitrogen) accordingto the manufacturer’s instructions.

Competitive ELISA for Assay of Vago. ELISA was performed for assay of se-creted Vago in supernatant media. In brief, 96-well plates were coated with 5μg/mL CxVago (C-terminal) peptide in 50 mM carbonate buffer, pH 9.6. Afterwashing, plates were blocked for 2 h at room temperature by using BSA (1%in PBS solution). Tenfold dilutions of peptide or supernatant medium wereincubated with anti-Vago antibody (1:50,000) for 2 h at 37 °C. The mixturewas then added to wells and incubated for 2 h at 37 °C. Secondary antibody

(HRP-labeled goat anti-rabbit) was added and, after similar incubation,3,3′,5,5′-TMB liquid substrate (Sigma) was used to quantify Vago by de-termining absorbance at 560 nm.

Real-Time RT-qPCR. Total RNA was collected from Hsu cells and homogenizedmosquitoes using the RNeasy kit (Qiagen) and RT-qPCR was performed asdetailed in SI Appendix.

Overexpression and Gene Silencing. Cx quinquefasciatus Vago was cloned intopIZ-V5/His vector (Invitrogen) for overexpression in Hsu cells. For mutations inthe signal peptide, a Site-Directed Mutagenesis Kit (Ambion) was usedaccording to the manufacturer’s instructions. Cloning and mutations wereconfirmed by direct sequencing. For gene silencing, dsRNA was preparedusing a MEGAscript RNAi kit (Ambion) according to the manufacturer’sinstructions, with primers as shown in SI Appendix, Table S1. Transfection ofplasmids and dsRNA was performed by using Cellfectin II (Invitrogen).

Statistical Analysis. SEM was calculated and data analyzed by using thenonpaired Student t test for single mean comparisons.

ACKNOWLEDGMENTS. The authors thank Dr. D. Boyle for valuable adviceduring method development and Drs. K. Blasdell, A. Joubert, and M. Adamsfor useful critiques and suggestions for the experiments. This work wassupported in part by an Australian Research Council Discovery Early CareerResearcher Award Grant (to P.N.P.).

1. Gubler DJ, Meltzer M (1999) Impact of dengue/dengue hemorrhagic fever on thedeveloping world. Adv Virus Res 53:35–70.

2. Randolph SE, Rogers DJ (2010) The arrival, establishment and spread of exotic dis-eases: Patterns and predictions. Nat Rev Microbiol 8:361–371.

3. Fragkoudis R, Attarzadeh-Yazdi G, Nash AA, Fazakerley JK, Kohl A (2009) Advances indissecting mosquito innate immune responses to arbovirus infection. J Gen Virol 90:2061–2072.

4. Dostert C, et al. (2005) The Jak-STAT signaling pathway is required but not sufficientfor the antiviral response of Drosophila. Nat Immunol 6:946–953.

5. Souza-Neto JA, Sim S, Dimopoulos G (2009) An evolutionary conserved function of theJAK-STAT pathway in anti-dengue defense. Proc Natl Acad Sci USA 106:17841–17846.

6. Sánchez-Vargas I, et al. (2009) Dengue virus type 2 infections of Aedes aegypti aremodulated by the mosquito’s RNA interference pathway. PLoS Pathog 5:e1000299.

7. Galiana-Arnoux D, Dostert C, Schneemann A, Hoffmann JA, Imler JL (2006) Essentialfunction in vivo for Dicer-2 in host defense against RNA viruses in Drosophila. NatImmunol 7:590–597.

8. Wang XH, et al. (2006) RNA interference directs innate immunity against viruses inadult Drosophila. Science 312:452–454.

9. Deddouche S, et al. (2008) The DExD/H-box helicase Dicer-2 mediates the induction ofantiviral activity in Drosophila. Nat Immunol 9:1425–1432.

10. Sheldon TJ, Miguel-Aliaga I, Gould AP, Taylor WR, Conklin D (2007) A novel family ofsingle VWC-domain proteins in invertebrates. FEBS Lett 581:5268–5274.

11. Zinke I, Schütz CS, Katzenberger JD, Bauer M, Pankratz MJ (2002) Nutrient control ofgene expression in Drosophila: Microarray analysis of starvation and sugar-de-pendent response. EMBO J 21:6162–6173.

12. Kilpatrick AM (2011) Globalization, land use, and the invasion of West Nile virus.Science 334:323–327.

13. Hsu SH, Mao WH, Cross JH (1970) Establishment of a line of cells derived from ovariantissue of Clex quinquefasciatus Say. J Med Entomol 7:703–707.

14. Lawson D, et al. (2009) VectorBase: A data resource for invertebrate vector genomics.Nucleic Acids Res 37:D583–D587.

15. Petersen TN, Brunak S, von Heijne G, Nielsen H (2011) SignalP 4.0: Discriminatingsignal peptides from transmembrane regions. Nat Methods 8:785–786.

16. Meng X, Khanuja BS, Ip YT (1999) Toll receptor-mediated Drosophila immune re-sponse requires Dif, an NF-kappaB factor. Genes Dev 13:792–797.

17. Harrison DA, McCoon PE, Binari R, Gilman M, Perrimon N (1998) Drosophila unpairedencodes a secreted protein that activates the JAK signaling pathway. Genes Dev 12:3252–3263.

18. Xi Z, Ramirez JL, Dimopoulos G (2008) The Aedes aegypti toll pathway controlsdengue virus infection. PLoS Pathog 4:e1000098.

19. Costa A, Jan E, Sarnow P, Schneider D (2009) The Imd pathway is involved in antiviralimmune responses in Drosophila. PLoS ONE 4:e7436.

20. Platanias LC, Fish EN (1999) Signaling pathways activated by interferons. Exp Hematol

27:1583–1592.21. Subramaniam PS, Torres BA, Johnson HM (2001) So many ligands, so few transcription

factors: A new paradigm for signaling through the STAT transcription factors. Cyto-

kine 15:175–187.22. Alexopoulou L, Holt AC, Medzhitov R, Flavell RA (2001) Recognition of double-

stranded RNA and activation of NF-kappaB by Toll-like receptor 3. Nature 413:

732–738.23. Heil F, et al. (2004) Species-specific recognition of single-stranded RNA via toll-like

receptor 7 and 8. Science 303:1526–1529.24. Chen YH, et al. (2011) Identification and functional characterization of Dicer2 and

five single VWC domain proteins of Litopenaeus vannamei. Dev Comp Immunol 35:

661–671.25. Diamond MS, et al. (2000) Modulation of Dengue virus infection in human cells by

alpha, beta, and gamma interferons. J Virol 74:4957–4966.26. Yamazaki S, Wagner RR (1970) Action of interferon: Kinetics and differential effects

on viral functions. J Virol 6:421–429.27. Muñoz-Jordan JL, Sánchez-Burgos GG, Laurent-Rolle M, García-Sastre A (2003) In-

hibition of interferon signaling by dengue virus. Proc Natl Acad Sci USA 100:

14333–14338.28. Gilbert MM, Weaver BK, Gergen JP, Reich NC (2005) A novel functional activator of

the Drosophila JAK/STAT pathway, unpaired2, is revealed by an in vivo reporter of

pathway activation. Mech Dev 122:939–948.29. Hombría JC, Brown S, Häder S, Zeidler MP (2005) Characterisation of Upd2, a Dro-

sophila JAK/STAT pathway ligand. Dev Biol 288:420–433.30. Arbouzova NI, Zeidler MP (2006) JAK/STAT signalling in Drosophila: Insights into

conserved regulatory and cellular functions. Development 133:2605–2616.31. Veloso A, Warr GW, Browdy CL, Chapman RW (2011) The transcriptomic response to

viral infection of two strains of shrimp (Litopenaeus vannamei). Dev Comp Immunol

35:241–246.32. Colpitts TM, et al. (2011) Alterations in the Aedes aegypti transcriptome during in-

fection with West Nile, dengue and yellow fever viruses. PLoS Pathog 7:e1002189.33. Luplertlop N, et al. (2011) Induction of a peptide with activity against a broad spec-

trum of pathogens in the Aedes aegypti salivary gland, following infection with

Dengue virus. PLoS Pathog 7:e1001252.34. Saleh MC, et al. (2009) Antiviral immunity in Drosophila requires systemic RNA in-

terference spread. Nature 458:346–350.35. Attarzadeh-Yazdi G, et al. (2009) Cell-to-cell spread of the RNA interference response

suppresses Semliki Forest virus (SFV) infection of mosquito cell cultures and cannot be

antagonized by SFV. J Virol 83:5735–5748.

18920 | www.pnas.org/cgi/doi/10.1073/pnas.1205231109 Paradkar et al.