Cellular Oxidative Stress Response Controls the Antiviral and Apoptotic Programs in Dengue Virus-Infected Dendritic Cells David Olagnier1¤*, Suraj Peri2, Courtney Steel1, Nadine van Montfoort1, Cindy Chiang1, Vladimir Beljanski1, Michael Slifker2, Zhong He1, Carmen N. Nichols1, Rongtuan Lin3, Siddharth Balachandran2, John Hiscott1* 1Vaccine&GeneTherapyInstituteofFlorida,PortSt.Lucie,Florida,UnitedStatesofAmerica,2FoxChaseCancerCenter,Philadelphia,Pennsylvania,UnitedStatesof America,3LadyDavisInstitute,JewishGeneralHospital,McGillUniversity,Montreal,Quebec,Canada Abstract Dengue virus (DENV) is a re-emerging arthropod borne flavivirus that infects more than 300 million people worldwide, leadingto50,000deathsannually.Becausedendriticcells(DC)intheskinandbloodarethefirsttargetcellsforDENV,we sought toinvestigatetheearly moleculareventsinvolvedinthehost responsetothe virusinprimary humanmonocyte- deriveddendriticcells(Mo-DC).Usingagenome-widetranscriptomeanalysisofDENV2-infectedhumanMo-DC,threemajor responses were identified within hours of infection - the activation of IRF3/7/STAT1 and NF-kB-driven antiviral and inflammatorynetworks,aswellasthestimulationofanoxidativestressresponsethatincludedthestimulationofanNrf2- dependentantioxidantgenetranscriptionalprogram.DENV2infectionresultedintheintracellularaccumulationofreactive oxygenspecies(ROS)thatwasdependentonNADPH-oxidase(NOX).AdecreaseinROSlevelsthroughchemicalorgenetic inhibitionoftheNOX-complexdampenedtheinnateimmuneresponsestoDENVinfectionandfacilitatedDENVreplication; ROS were also essential in driving mitochondrial apoptosis in infected Mo-DC. In addition to stimulating innate immune responses to DENV, increased ROS led to the activation of bystander Mo-DC which up-regulated maturation/activation markers andwere lesssusceptibleto viral replication.Wehave identifieda criticalroleforthe transcription factorNrf2 in limitingbothantiviralandcelldeathresponsestothevirusbyfeedbackmodulationofoxidativestress.SilencingofNrf2by RNA interference increased DENV-associated immune and apoptotic responses. Taken together, these data demonstrate thatthelevelofoxidativestressiscriticaltothecontrolofbothantiviralandapoptoticprogramsinDENV-infectedhuman Mo-DCand highlight the importanceofredox homeostasis inthe outcomeofDENVinfection. Citation:OlagnierD,PeriS,SteelC,vanMontfoortN,ChiangC,etal.(2014)CellularOxidativeStressResponseControlstheAntiviralandApoptoticProgramsin DengueVirus-InfectedDendriticCells.PLoSPathog10(12):e1004566.doi:10.1371/journal.ppat.1004566 Editor:GlennRandall,TheUniversityofChicago,UnitedStatesofAmerica ReceivedJuly14,2014;AcceptedNovember10,2014;PublishedDecember18,2014 Copyright:(cid:2)2014Olagnieretal.Thisisanopen-accessarticledistributedunderthetermsoftheCreativeCommonsAttributionLicense,whichpermits unrestricteduse,distribution,andreproductioninanymedium,providedtheoriginalauthorandsourcearecredited. DataAvailability:Theauthorsconfirmthatalldataunderlyingthefindingsarefullyavailablewithoutrestriction.Allrelevantdataarewithinthepaperandits SupportingInformationfilesexceptforthearraydatawhichisavailableintheNCBIGeneExpressionOmnibus(GEOSeriesaccessionnumberGSE58278). Funding:ThisprojectwassupportedbyfundingfromVGTIFloridaandtheCanadianInstituteofHealthandResearch.Thefundershadnoroleinstudydesign, datacollectionandanalysis,decisiontopublish,orpreparationofthemanuscript. CompetingInterests:Theauthorshavedeclaredthatnocompetinginterestsexist. *Email:[email protected](DO);[email protected](JH) ¤Currentaddress:LadyDavisInstitute,JewishGeneralHospital,McGillUniversity,Montreal,Quebec,Canada Introduction infection manifests as self-limiting dengue fever or progresses to life-threatening illnessremains unanswered. Dengue virus (DENV) is the leading arthropod-borne viral Dengue is an RNA virus of the Flaviviridae family with 4 infectionintheworld,andrepresentsamajorglobalhumanhealth closely related serotypes that exhibit inter- and intra-serotypic concern.DENVisendemicinmorethan100countrieswithupto genetic diversity [5–9]. Innate recognition of DENV involves a 3billionpeopleintropicalregionsoftheworldatriskofinfection spectrum of pattern recognition receptors (PRR) that sense [1–3].Recently,DENVhasexpandeditsglobalrange,withlong- conserved molecular components termed pathogen associated term outbreaks in South America and reintroduction into North molecular patterns (PAMP), and together orchestrate antiviral AmericathroughFloridaandTexas,witheachoftheseoutbreaks responses to the viral infection. The cytoplasmic helicases RIG-I accompanied by increased disease severity. Of the estimated 50– andMDA-5haveacentralroleinthehostresponsetoDENVby 100 million annual cases, the majority of infected individuals contributing to DENV protection in hepatocytes [10]. Addition- develop a self-limiting febrile illness, but approximately 500,000 ally,TLR3andTLR7recognizeDENVRNAandmountarapid clinicalcasesresultinmoreseveremanifestations,suchasDENV- protective immune response in human monocytic cells and induced hemorrhagic fever and shock syndrome [1], leading to plasmacytoid dendritic cells, respectively [11,12]. Signaling 25–50,000 deaths per year [4]. The pathogenesis of dengue is through these different cellular sensors leads to the activation of incompletely understood and the factors that determine whether the interferon pathway that restricts viral proliferation and PLOSPathogens | www.plospathogens.org 1 December2014 | Volume 10 | Issue 12 | e1004566 RedoxRegulationoftheHostResponsetoDENV disease severity in DENV-infected patients, suggesting a possible Author Summary role for oxidative stress in DENV-induced pathogenesis [41–44]. Dengue virus (DENV), the leading arthropod-borne viral Interestingly, circulating monocytes from glucose-6-phosphate infection in the world, represents a major human health dehydrogenase (G6PD)-deficient patients, displayed an increased concern with a global at risk population of over 3 billion susceptibilitytoDENVinfectionandreplication[45].TheG6PD people. Currently, there are no antivirals or vaccines deletion affects ROS production, thus linking cellular oxidative available to treat patients with dengue fever, nor is it state and susceptibility to DENV infection. Altogether, these possible to predict which patients will progress to life- observationsunderlinetheimportanceoftheredoxhomeostasisin threatening severedengue fever.Markers associatedwith DENV infection and suggest an important interplay between the oxidative stressresponses have been reportedin patients generationofoxidativestressandtheimmunopathologyofdengue with severe DENV infection, suggesting a relationship disease. between oxidative stress and viral pathogenesis. In order Initial contact between DENV and innate immune cells plays to uncover biological processes that determine the an essential role in the outcome of the infection. Indeed, DENV outcome of disease in patients, we utilized human infection pushes monocytes towards a CD16+ inflammatory dendritic cells, the primary target of DENV infection, in an in vitro model. Transcriptional analysis of pathways phenotypethatfacilitatesplasmablastdifferentiationandinduction activated upon de novo DENV infection revealed a major of anti-DENV antibody responses [46]. Given the importance of roleforcellularoxidativestressintheinductionofantiviral, DC in bridging the innate and adaptive immune response, and inflammatory, and cell death responses. We also demon- sinceDCintheskinandperipheralbloodarethefirsttargetcells strated that antioxidant mechanisms play a critical role in for DENV after transmission via a mosquito bite [47–49], controlling antiviral and cell death responses to the virus, evaluation of the early molecular events in DC is crucial to the acting as feedback regulators of the oxidative stress understanding of DENV pathogenesis. In the present study, we response. This report highlights the importance of oxida- generatedin-depthtranscriptomeanalysis,coupledwithbiochem- tive stress responses in the outcome of DENV infection, ical and functional analyses of the early host response to DENV andidentifiesthispathwayasapotentialnewentry-point infection in primary Mo-DC. DENV infection triggered an for treatingdengue-associated diseases. NADPH-oxidase (NOX)-dependent oxidativestress responsethat was required for the activation of IRF3/7/STAT1 and NF-kB- contributestotheestablishmentofadaptiveimmuneresponsesvia mediated antiviral responses and for mitochondrial-dependent NF-kB-mediated cytokine and chemokine release [13–16]. Inter- apoptosis. Furthermore, we have identified a critical role for the estingly, the host immune response, activated in response to transcription factor Nrf2 in regulating both antiviral and DENVinfection,notonlymediatesprotectionagainstdisease,but inflammatory gene response to the virus by feedback modulation alsocontributestodiseaseseverity[1].Forexample,highlevelsof ofoxidativestress.Overall,thesestudieshighlighttheimportance circulatingpro-inflammatorycytokinessuchasIL-1borTNF-ain of redoxhomeostasis inthe outcomeof DENV infection. DENV-infected patients correlates with severe dengue fever, compared topatients suffering withmild denguefever [17]. Results Reactive oxygen species (ROS) production, generated as a consequenceofmicrobialinvasion,haslongbeenknowntoexert DENV2 highly infects Mo-DC and generates a broad an antimicrobial effect in phagocytes [18]. The activation of the antiviral response antiviral and inflammatory signaling pathways has also been An in vitro model of de novo DENV infection was established linkedwiththeproductionofROS[19–23],whichincludeoxygen using primary human monocytes differentiated in vitro with Mo- ions and peroxides that are produced as byproducts of aerobic DC-differentiation medium containing GM-CSF and IL-4. metabolism.BecauseofthehighchemicalreactivityofROS,cells Primary CD14+ CD1a2 monocytes were less permissive to possess scavenger antioxidant mechanisms that maintain redox DENV2 infection, whereas infectivity increased progressively as homeostasis [24–26]. Signaling pathways downstream of ROS the cells differentiated toward the Mo-DC (CD142 CD1a+) detectionactivatethetranscriptionfactornuclearfactor-erythroid phenotype(4.6660.45%ofDENV+cellsinmonocytesatday0vs 2-related factor 2 (Nrf2) [24–26], which binds antioxidant 79.660.47% in Mo-DC at day 7) (Fig.1A). A strong statistical responseelements (ARE)withinthepromotersofgenesencoding correlation between a CD142CD1a+ phenotype and DENV antioxidantanddetoxifyingenzymes.Nrf2-dependentantioxidant infection was confirmed by the nonparametric Spearman test genes act synergistically to reduce oxidative stress by quenching (r=0.9829;p,0.0001;n=15).DENV2viralRNAaccumulation ROS [24–26]. wasdetectedafteralagperiodof6 handincreasedexponentially IncreasedgenerationofROSandchangesinredoxhomeostasis thereafter(Fig.1B),whichcorroboratesapreviousreportdemon- have been described in the context of many viral infections strating release of infectious particles [50]. Prior to the onset of [23,27–33] and the failure to maintain an appropriate redox detectable DENV replication, an antiviral response was mounted balance contributes to viral pathogenesis through alterations of by the infected Mo-DC population, as demonstrated by the biological structures and themassive induction of celldeath [34– increase in IFN-b, IFIT1 and CCL5 gene expression (Fig.1B). 36]. In the flavivirus family, hepatitis C virus (HCV) has been DENV infected Mo-DC in a dose dependent manner to a shown to promote oxidative stress and manipulate antioxidant maximum of ,80% infectivity at a MOI of 20 (Fig.1C). As a systems,leadingtochronicdisease[31,37,38].Aswell,DENVwas consequence of early virus sensing, a broad antiviral and shown to stimulate oxidative stress in hepatocytes leading to inflammatoryresponsewasgeneratedasshownbythephosphor- production of the chemokine CCL5 and to activation of the ylation of IRF3 and STAT1 (Fig.1D) and significant release of transcriptional regulator C/EBP beta [39]. Furthermore, HepG2 IFN-a, TNF-a and IL-6 (Fig.1E) by the infected cells. Previous xenografted SCID mice presented alterations in oxidative stress studies reported cleavage of the endoplasmic reticulum adaptor status and increased inflammatory cytokines following DENV STING upon DENV infection in Mo-DC [51]. However, in our infection [40]. More recently, oxidative stress-induced damage experimentalmodelandwiththeviralstrainused,amodest20% andalterationsinredoxstatushavebeenassociatedwithincreased decreaseinSTINGexpressionwasobservedat48hafterinfection PLOSPathogens | www.plospathogens.org 2 December2014 | Volume 10 | Issue 12 | e1004566 RedoxRegulationoftheHostResponsetoDENV Fig.1.Mo-DCarehighlysusceptibletoDENVinfectionandgenerateabroadantiviralandinflammatoryresponse.(A)Thecorrelation betweenthepercentageofinfectedcellsdeterminedbyintracellularstainingofDENVEproteinexpressionandthepercentageofCD142CD1a+Mo- DC was calculated during the course of Mo-DC differentiation (n=15; Spearman test). The FACS panels on the left hand side represent the progressionofMo-DCdifferentiationatdays0,3and7.(B)DENVRNAlevelsaswellasIFN-b,IFIT1,andCCL5geneexpressionlevelsweredetermined byqPCRatvarioustimesofDENV2infection(MOI10).Pvaluesweredeterminedbasedonthecomparisonwithcellsattime3hafterinfection.Data arefromoneexperimentperformedonthreeindividualdonors.(C)Mo-DCwereinfectedwithincreasingamountsofvirus(MOI0.04-MOI20)for24h. PercentageofDENV2-infectedcellswasdeterminedbyintracellularstaining(ICS)ofDENVEproteinexpressionusingflowcytometry.Dataarethe means 6 SEM from one experiment performed on three different donors. (D) Mo-DC were challenged with DENV2 (MOI 20) and antiviral and inflammatoryresponseswereexaminedbyimmunoblotting.Dataarefromonerepresentativeexperiment.(E)Cytokinereleasewasevaluatedby cytokine bead array (CBA) 24h after DENV2 challenge (MOI 10) in the supernatants of infected cells. P values were determined based on the comparisonwithuninfectedcells.Dataarethemeans6SEMfromoneexperimentperformedonthreedifferentdonors. doi:10.1371/journal.ppat.1004566.g001 (Fig.1D).AltogetherthesedatademonstratethatDENV-infected antiviral pathways as well as the NF-kB-dependent pro-inflam- Mo-DC generate a broad host response and secrete an array of matory pathways were all highly enriched after de novo DENV2 antiviral and inflammatory cytokines inresponsetothevirus. infection (Fig.2B). We also noticed an enrichment of networks associated with the generation of a pro- and anti-oxidant stress Transcriptome kinetics of the host response to DENV2 response (Fig. 2B). Further gene analysis represents the top 50 infection DEG over time following DENV infection (Fig.2C); among the Tocharacterizesignalingpathwaysinvolvedinthehostintrinsic topup-regulatedgenes,twosubclassespredominated–interferon- responsetoDENV2infection,atranscriptomeanalysisofDENV- stimulated genes (ISGs) such as ISG15, IFIT1, IFIT2, IFIT3, infected Mo-DC was performed; Fig.2A represents a waterfall OASL, OAS2, CCL5, HES4 (presented in black) and more plotofdifferentiallyexpressedgenes(DEG;selectedbasedonfold surprisingly a large set of antioxidant genes belonging to the change .61.3, p value ,0.05) after DENV2 infection. Most metallothionein familyincluding MT1A,MT2A, MT1E,MT1X, changesingeneexpressionappearedearly,withover7000genes MT1G,MT1H,andMT1F(presentedinred)(Fig.2C).Basedon either up- or down-regulated by 6h after infection (Fig.2A). the regulation of gene networks activated or repressed after Pathway analysis identified multiple canonical networks coordi- DENV2infection,Fig.2Dillustratesawordcloudmapofpossibly natelyregulatedatalltimesafterinfection;theexpectedIFN/IRF activated(red)orinhibited(green)transcriptionfactorscontrolling PLOSPathogens | www.plospathogens.org 3 December2014 | Volume 10 | Issue 12 | e1004566 RedoxRegulationoftheHostResponsetoDENV Fig. 2. Transcriptome analysis of the host response to DENV2 infection in Mo-DC. Mo-DC were infected with DENV2 (MOI 20) for designatedperiodsoftime.SampleswereanalyzedbyIlluminageneexpressionarrayanddifferentiallyexpressedgenes(DEGs)thatsatisfiedap value (,0.05) with $1.3 fold change (up or down) were selected. (A) Waterfall plot representing the total number of up-regulated and down- regulatedgenesateachtimepoint.(B)Heatmapshowsstatisticallysignificantcanonicalpathways(IngenuityPathwayAnalysisSoftware)commonly regulatedat6h,12h,18hand24hwhencomparedtobaseline.Genesthathadadjustedp-value,0.05ateachtimepointandfoldchange.1.3or ,21.3andassociatedwithacanonicalpathwayinIngenuity’sKnowledgeBasewereusedforpathwayanalysis.Heatmapcolorsrepresenttheratio of regulated genes/pathway genes after dengue infection (red and blue correspond to over- and under-represented, respectively). The over- representationtestwasperformedusingFisherExactTest.Statisticalsignificanceachievedatp,0.05.Thedataarerepresentativeofoneexperiment performedonthreedifferentdonors.(C)Geneexpressionheatmapofthetop50differentiallyexpressedgenesinducedbydengueinfectioninMo- DCatvarioustimeswhencomparedtobaseline.GenesareselectedasdifferentiallyexpressedinatleastonecomparisonfollowingANOVAFtestas implementedintheLIMMApackage.Thescaleshowsthelevelofgeneexpressionwhereredandbluecorrespondtoup-anddown-regulation respectively.Apanelofantiviral(black)andantioxidant(red)DEGisrepresentedontherighthandsideoftheheatmap.(D)Wordcloudsrepresenting potentiallyactivated(red)/inhibited(green)transcriptionfactorsat6hand24hafterDENV2challenge.IPAUpstreamRegulatorAnalysiswasusedto identifymoleculesupstreamofthegenesinthedatasetthatcouldexplainthedetectedexpressionchanges.Thep-valueofoverlap,whichmeasures the enrichment of network-regulated genes in the data set, is represented by the size of the word. The activation z-score which predicts likely regulatingmoleculeswasusedtocolorthepredictedactivationstate. doi:10.1371/journal.ppat.1004566.g002 PLOSPathogens | www.plospathogens.org 4 December2014 | Volume 10 | Issue 12 | e1004566 RedoxRegulationoftheHostResponsetoDENV PLOSPathogens | www.plospathogens.org 5 December2014 | Volume 10 | Issue 12 | e1004566 RedoxRegulationoftheHostResponsetoDENV Fig.3.FunctionalcharacterizationofgenesdifferentiallyexpressedbyDENV2infection.Differentiallyexpressedgenesweresubjectedto IngenuityPathwayAnalysisforboth6h(A)and24h(B)timepointsbasedonp-values(,0.001)andfoldchange$61(log2).Redrepresentgenes inducedbyDENV,andgreenindicatedgenesdownregulatedbythevirus;theintensityofthecolorisrepresentativeofthefoldchange.Largercircles indicatetranscriptionfactors.Genesenrichedforapathwayarerepresentedasacloud. doi:10.1371/journal.ppat.1004566.g003 genenetworksat6hand24hafterDENV2challenge(Fig.2D). (Fig.4C). Although ROS are generated intracellularly, the At 6 h after infection, two subclasses of transcription factors primary sources of ROS are plasma membrane oxidases, predominated: 1) transcription factors associated with cellular particularly NADPH oxidases. ROS were detected as early as stress-responses including TP53 (p53), EPAS1, HIF1A and 3 h after infection, and ROS production was suppressed by pre- NFE2L2 (Nrf2); and 2) transcriptional regulators associated with treatment with the antioxidant diphenyleneidonium chloride the antiviral program including IRF1/3/7, STAT1/ISGF3 and (DPI), an NADPH-oxidase (NOX) inhibitor (Fig.4D). ROS NF-kB complex (Fig.2D). By 24h post-infection, the activity of production was independently confirmed in Mo-DC by the use stress-relatedtranscriptionfactorsdecreased,withtheexceptionof of pyocyanin (N-methyl-1-hydroxyphenazine), an oxidative stress TP53, while transcription factors driving the antiviral response - inducer,asdenotedbythe1.8foldincreaseinROSgenerationat predominantly IRF7andNF-kB-were highlyactive (Fig.2D). 3 h after stimulation (Fig.4D). The involvement of NOX in AFluidigmBioMarkhighthroughputqPCRassayencompass- DENV-inducedROSaccumulationwasfurtherconfirmedbythe ing a cross-section of genes identified inthe genomicanalysis(S1 increased phosphorylation of the p47 subunit of the NADPH- Table)wasusedtovalidatethetranscriptomedata;thepatternof oxidase (p=0.0404) (Fig.4E). Interference with NADPH-oxidase gene expression at various times after DENV2 infection was activityusingsiRNA-mediatedsilencingofthecatalyticgp91phox similar for three different donors (S1A Figure). Computational subunitlimitedROSaccumulationinresponsetodenovoDENV analysis identified different kinetics of IFN induction, as well as infection (p=0.0328) (Fig.4F). To examine whether cellular sustained up-regulation of chemokines, Th1 cytokines, ISGs and oxidative stress impactedtheimmediatehost responsetoDENV, antiviral transcription factors (S1B Figure). A strong statistical weevaluatedtheeffectofexogenousROSadditiononexpression correlationbetweenthelogfoldchangeforthemicroarrayvalues of DENV-induced antiviral genes. Treatment with increasing andthelogfoldchangefortheBioMarkvalueswasconfirmedbya concentrations of hydrogen peroxide (H O ) did not stimulate 2 2 Spearman correlation test (S1C Figure) (r=0.8399194; immune responses in Mo-DC; however addition of H O 2 2 p=4.576e-14; n=49). In order to gain systems-wide insight into moderately potentiated the elevation of DENV-induced antiviral DENV-modulated transcriptome, a functional clustering (node geneexpression (Fig.4G). analysis) (Fig.3), as well as gene-pathway checkerboard analysis Next, the role of ROS in triggering the early host response to (S2 Figure) of DENV-induced DEGs was performed. This DENV2 was evaluated by treating infected Mo-DC with functional clustering identified at 6 h (Fig.3A and S2A Figure) increasing concentrations of DPI, an NADPH-oxidase inhibitor. and24h(Fig.3BandS2BFigure)avarietyoftranscriptionalsub- Strikingly,phosphorylationofIRF3,STAT1andIkBa,aswellas networksandbiologicalprocessesregulatedbyDENV.TheNrf2- the induction of ISGs such as RIG-I and IFIT1 – all markers of mediated oxidative stress response pathway, the top differentially the antiviral response – were inhibited in a dose-dependent regulatedpathwayinDENV-infectedMo-DCat6 h(S2AFigure), manner by DPI (Fig.5A). The observation that NOX-inhibitor wastriggeredpriortotheonsetofviralreplicationandintersected blocked DENV-induced immune response was further confirmed with other pathways such as NF-kB, IRF and STAT signaling by quantitative intracellular measurement of STAT1 phosphory- (Fig.3A and S2A Figure). At the same time, hypoxia pathway lation. Indeed, DPI prevented the increase in STAT1 phosphor- controlled by the transcription factor HIF1-a was predominantly ylation detected by PhosFlow following DENV infection. Impor- downregulated(Fig.3AandS2AFigure).By24htheactivityof tantly,IFNb-inducedSTAT1phosphorylationwasnotaffectedby the Nrf2-driven pathway decreased, whereas the expansion and the DPI treatment (Fig.5B). Using a customized BioMark chip, increasedinteractionamongtheantiviral,inflammatoryanddeath antiviral and inflammatory genes such as type I IFNs (IFNA2, response networks predominated (Fig.3B and S2B Figure). IFNB1), pro-inflammatory cytokines and chemokines (IL1b, Concomitantly, genes related to mitochondrial function were all CCL5) and ISGs (MX1, IFITM1/2/3, OASL, IDO1, OAS3, significantly down regulated and presumably associated with an DDX58) were inhibited by DPI in a dose dependent manner in increase intheapoptotic response (Fig.3BandS2B Figure). DENV-infected cells (Fig.5C, upper right box). Cytokine release (IFN-a,TNF-aandIL-6)wasalsoimpairedinthepresenceofthe DENV-induced NOX-derived ROS accumulation is antioxidant molecule (Fig.5D). The use of antioxidant molecules essential for induction of innate immune responses withdifferentmodesofaction(S3AFigure)recapitulatedtheeffect The role of reactive oxygen species (ROS) as specific second observed with DPI and impaired the induction of antiviral and messengers in signaling cascades involved in cell proliferation, inflammatory gene expression (Fig.5E). Importantly, all antiox- differentiation andimmune activation has been well documented idant moleculestested in this paneldid not affect cell survival, as [52].InlightofthearraydataandtoevaluateifROSareinvolved shown in S3B Figure and S3C Figure. Inhibition of NADPH- in the recognition of DENV, Mo-DC were infected and ROS oxidase activity using transient knock-down of the catalytic formation was monitored by flow cytometry using the oxidant- gp91phox subunit also decreased IFIT1 protein expression sensitive fluorescent detection probe CM-H2DCFDA. ROS following de novo DENV infection (Fig.5F). No increase in production was induced in DENV-infected Mo-DC, as reflected DENV RNA accumulation was detected in the presence of the in the 2-fold increase in DCF fluorescence detected by FACS at NOX-inhibitor (3mM) after 24h of infection (S4 Figure). 18h after infection (p=0.0405) (Fig.4A). Also, DENV infection However, pre-treatment of cells with a higher concentration of increased intracellular ROS accumulation in a dose dependent DPI(30 mM)ledtoanincreaseinDENVviralRNAaccumulation manner (Fig.4B). A strong statistical correlation between DENV in the same conditions (S4 Figure). Importantly, DPI treatment infection and the accumulation of ROS was confirmed by the resulted in increased DENV infectivity and replication at 48h nonparametric Spearman test (r=0.7635; p,0.0001; n=15) post-infection, as demonstrated by the increased number of PLOSPathogens | www.plospathogens.org 6 December2014 | Volume 10 | Issue 12 | e1004566 RedoxRegulationoftheHostResponsetoDENV Fig.4.DENV-infectedcellsaccumulateintracellularNOX-derivedROSwhichpotentiatetheimmuneresponse.(A)ROSgenerationin DENV-infected Mo-DC (MOI 20) was monitored by FACS using CM-H2DCFDA (1mM) 18h after infection. Data represent the mean 6 SEM from experimentsperformedonfourdifferentdonors.(B)AccumulationofROSwasmeasuredbyflowcytometryusingCM-H2DCFDA(1mM)24hafter infectionwithincreasingamountofvirus.Datarepresentthemean6SEMfromoneexperimentsperformedonthreedifferentdonors.Pvalueswere determinedbasedonthecomparisonwithuninfectedcells(C)Thecorrelationbetweenthepercentageofinfectedcellsdeterminedbyintracellular stainingofDENVEproteinexpressionandtheintracellularaccumulationofROSmeasuredbyDCFDAstainingwascalculated(n=15;Spearmantest). (D)OxidativestressgenerationwasdetectedbyflowcytometryusingtheCM-H2DCFDA(1mM)fluorescentprobeonMo-DCpretreatedwithDPI (3mM) and infected with DENV2 (MOI 20) for 3h or treated with the stress-inducer Pyocyanin (100mM). Data represent the geometric mean fluorescence6SEMfromanexperimentperformedontwodifferentdonors.(E)ExpressionlevelofthephosphorylatedNADPH-oxidasep47subunit wasassessedbyimmunoblotting3hafterDENV2infection.Histogramsrepresentthefoldchangeratioofphosphorylated-p47overtotalp47.Data arethemeans6SEMofindependentexperimentsperformedonthreedifferentdonors.Pvaluewasdeterminedbasedonthecomparisonwith PLOSPathogens | www.plospathogens.org 7 December2014 | Volume 10 | Issue 12 | e1004566 RedoxRegulationoftheHostResponsetoDENV uninfectedcells.(F) Mo-DCweretransfected withcontrol or gp91phox siRNAand 48h later were infected withDENV2 (MOI20) for 18h. ROS accumulationwasmeasuredbyFACSusingtheCM-H2DCFDAfluorescentprobe.Dataarethemeans6SEMofthreeindependentexperiments performedondifferentdonors.PvaluewasdeterminedbasedonthecomparisonwithcellsinfectedwithDENVandtransfectedwiththecontrol siRNAsequences(G)Mo-DCweretreatedwithincreasingconcentrationsofhydrogenperoxide(10–50mM)inthepresenceorabsenceofDENV2 (MOI20).CCL5andCXCL10mRNAexpressionlevelsweremonitoredbyqPCR24hfollowingtreatmentandinfection.Datarepresentthemeans6 SEMfromoneexperimentperformedonthreeindividualdonors.Experimenthasbeenrepeatedtwice.Pvaluesweredeterminedbasedonthe comparisonwithDENV2-infectedcells. doi:10.1371/journal.ppat.1004566.g004 DENV-infectedcells(i–ii)andviraltiters(iii)(Fig.5G).TheROS- While infected cells displayed apoptotic markers as described mediated induction of antiviral and inflammatory genes required above, uninfected bystander Mo-DC cells did not undergo live and replicating virus, since formalin-inactivation and UV- apoptosis, but rather increased expression of the differentiation inactivationofDENV2completelysuppressedtheinductionofthe and activation markers CD83 and CD86 (Fig.6G and S6A–C immune response (S5A Figure and S5B Figure). Also, DPI Figure),CD40,CD80,CD86andPD-L1(S6DFigure).Whencells inhibited antiviral and inflammatory responses induced by were pre-treated with DPI prior to DENV infection, the number DENV2strain16681(S5CFigure),indicatingthatROS-mediated ofCD83-positivebystandercellsdecreasedby2.2fold,compared antiviral induction is a common feature of the DENV2 serotype to non-treated cells (Fig. 6H). To determine if ROS production and is not restricted to a specific strain. Collectively, these data altered the antiviral response in uninfected bystander cells via demonstrate that DENV infection of Mo-DC triggers an cytokinerelease,conditionedmediafromDENV-infectedDCpre- intracellular accumulation of NOX-derived ROS, which are treated or not with DPI was transferred to uninfected Mo-DC essential for the induction of the antiviral and inflammatory (Fig.6I).Pre-treatmentwithconditionedmediafromDPI-treated immune responsesandthecontrol of DENV infection. DCalteredthesusceptibilityofna¨ıvecellstoDENVinfection,as shown by the ,2-fold increase in DENV E protein expression. DENV-inducedROSdrivemitochondrialapoptosisinMo- Altogether, ROS contributes to mitochondria-dependent apopto- DC and contribute to bystander cell maturation sis,andalsocontributestothematurationofuninfectedbystander DC. DENV-infectedDCwereclearlyapoptotic,basedonAnnexin- Vstaining:2765.15%(infected)vs6.6961%(control)at24hand Nrf2 transcription factor controls DENV infection and its 73.6564.2% (infected) vs 22.9663.88% (control) at 48h (Fig.6A). Upregulation of mRNA levels for pro-apoptotic genes associated immune and apoptotic response suchasBCLX,BIM,andCASP4uponDENVinfection(Fig.6B) Defense against sustained antioxidant production and the was consistent with the transcriptome analysis that identified the inhibitionof ROSareimportantprotective mechanismsthat are induction of apoptosis-associated pathways 24h after DENV regulated by the activation of Nrf2-transcription factor and infection (Fig.2BandFig.3B). downstreamNrf2-targetgenes.Basedonthearraydata(Fig. 3A), To assess therelease of mitochondrial ROS, cells were stained Nrf2targetgenessuchasHMOX-1,SOD2,NQO1,aswellasthe with mitoSOX, a probe specific for mitochondria-derived ROS; metallothionein and ferritin families, were all rapidly stimulated the number of mitoSOX-positive cells increased from bydenovoDENV2infection(Fig. 7A)andtransientinductionof 31.9612.3% (uninfected) to 56.867.9% (infected) at 48h after these genes was confirmed by qPCR (Fig. 7B). Levels of heme- infection. DiOC was also used to determine the loss of oxygenase-1 (HMOX-1) and superoxide dismutase-2 (SOD-2) 6 mitochondrial potential upon DENV infection: only 7.960.2% mRNA were sensitive to the ROS scavenger DPI which uninfected cells were positive, whereas 47.2610.2% of infected abrogatedtheincreaseinHMOX-1andSOD-2(Fig. 7C).When cells were positive for mitochondrial depolarization. Consistent Nrf2 expression was silenced using Nrf2-specific siRNA (both at with the release of mitochondrial ROS and mitochondrial the mRNA (Fig. 7D) and at the protein level (S7A Figure), depolarization, intracellular levels of cleaved caspase-3 increased decreases in the mRNA levels of Nrf2-dependent antioxidant from5.661.3%inuninfectedcellsto28.961.2%ininfectedcells genes were also observed (S7B Figure). Functionally, the redox (Fig.6C). Regression analysis indicated that the percentage of homeostasis was critically affected in Nrf2-deleted Mo-DC, as infected cells at 24h correlated with the percentage of apoptotic shown by the ,3 fold increase in ROS accumulation (S7C cells at 48h after infection (Fig.6D (i)). Furthermore, both Figure). Although silencing of Nrf2 only slightly increased mitochondrial ROS release and mitochondrial depolarization DENV2 RNA accumulation (Fig. 7E) and DENV infectivity were statistically associated with apoptosis induction (Fig.6D (ii) (Fig. 7F) after 24 h of infection, the impairment of Nrf2 and (iii)), thus demonstrating DENV-infected Mo-DC undergo expression drastically potentiated oxidative stress response in mitochondrial-dependent apoptosis. When DC were pre-treated DENV-infected cells (Fig. 7G). Indeed, a ,2 fold increase in withtheNOX-inhibitorDPI,astatisticallysignificantdecreasein ROS generation was observed between DENV-infected control- apoptosisofDENV-infectedcellswasobserved(,80%forDENV and siRNA-expressing, Nrf2-transfected cells (Fig. 7G) for the only infection compared to ,52% for DENV+DPI infection), samenumberofinfectedcells(Fig. 7F).Finally,themRNAlevels indicating that mitochondrial-dependent apoptosis was also of genes associatedwith the antiviral and inflammatoryresponse dependent, at least in part, on NOX-generated ROS (Fig. 6E). such as IFIT1, RSAD2, DDX58, CXCL10 and IFNb (Fig. 7H), Based on the array data, a key sensor of cellular stress, the as well as genes involved in the apoptotic response such as transcription factor p53 was strongly activated following DENV NOXA, BCLX, and RIPK1 (Fig. 7I) were all significantly infection (Fig.2D). Inhibition of p53, using the specific inhibitor increased. Altogether, these data demonstrate that the Nrf2- pifithrin-awasabletopartiallysuppressDENV-inducedapoptosis regulated antioxidant pathway is stimulated as part of the stress (Fig.6F),asdidthepan-caspaseinhibitorZ-VAD-fmkinMo-DC response after DENV infection; the Nrf2-dependent genes (Fig.6F). Altogether, these results argue that NOX-dependent regulate the levels of ROS production and thus modulate the induction of ROS stimulated p53-regulated mitochondrial and immune and apoptotic responses against DENV infection caspase-dependent apoptosis. (Fig. 8). PLOSPathogens | www.plospathogens.org 8 December2014 | Volume 10 | Issue 12 | e1004566 RedoxRegulationoftheHostResponsetoDENV Fig.5.CellularoxidativestressresponseisrequiredtomediateDENV-inducedinnateimmuneresponses.Mo-DCwerepre-treated withDPI(0.3or3mM)for1h,andsubsequentlyinfectedwithDENV2(MOI20)for24h.(A)Antiviralandinflammatoryresponsesweremonitoredby immunoblotting. Dataare representative of at leasttwoindependentexperiments onseparatedonors. (B) Intracellularlevels of phosphorylated PLOSPathogens | www.plospathogens.org 9 December2014 | Volume 10 | Issue 12 | e1004566 RedoxRegulationoftheHostResponsetoDENV STAT1weredetectedbyPhosFlowinMo-DCinfectedfor24hbyDENV2(MOI20)orstimulatedfor30minwithIFN-b(1000IU/mL)andpre-treated ornotwithDPI(1mM).Datarepresentthemeans6SEMfromtwoexperimentsperformedonseparatedonors.(C)Highthroughputanalysisofgene expression evaluated by qPCR BioMark analysis. Gene expression levels were calculated using the DDCt method and gene-wise standardized expression (z-score) were generated for each gene. The scale represents z-score values where red shows an up-regulation and blue a down- regulationingeneexpression.Dataarerepresentativeofoneexperimentperformedonthreeindividualdonors.Eachboxoftheheatmaprepresents onedonor.(D)Cytokinereleasewasevaluatedbycytokinebeadarray(CBA)onthesupernatantsofDENVinfectedcellspre-treatedornotwithDPI (0.3–3mM).Datarepresentthemeans6SEMfromthreeindividualdonors.PvaluesweredeterminedbasedonthecomparisonwithDENV2-infected cells.(E)Mo-DCwerepre-treatedwithDPI(3mM),Apocynin(3mM),TEMPOL(3mM),Ebselen(10mM),PDTC(40mM)andTrolox(5mM)for1h,and subsequentlyinfectedwithDENV2(MOI20)for24h.AntiviralandinflammatorygeneexpressionwasdeterminedbyqPCR.Datarepresentthemeans 6SEMfromoneexperimentperformedonthreeindividualdonors.PvaluesweredeterminedbasedonthecomparisonwithDENV2-infectedcells. (F) Mo-DC were transfected with control or gp91 phox siRNA and 48h later were infected with DENV2 (MOI 20). IFIT1 and gp91 phox protein expressionlevelsweremeasuredbyimmunoblotanalysis.Resultisrepresentativeofoneexperiment.(G)Mo-DCwerepre-treatedwithDPI(0.1– 1mM)for1h,andsubsequentlyinfectedwithDENV2(MOI1)for48h.PercentageofinfectedcellswasdeterminedbyintracellularstainingofDENVE protein(i–ii).DENVtitersweredeterminedbytransferringsupernatantsfromMo-DC-infectedcellsonA549cellsandstainingforDENVEprotein(iii). DENVtiterswereexpressedasthenumberofinfectiousunits/mL.Datarepresentthemeans6SEMofexperimentsperformedonfour(i–ii)andtwo differentdonors(iii). doi:10.1371/journal.ppat.1004566.g005 Discussion syntheticdsRNAPoly(I:C)[65].Furthermore,exogenousaddition of oxidative stress potentiated the TLR3 response to dsRNA in Evaluation of the early host immune response to DENV airway epithelial cells [66]. Finally, the specific TLR7 agonist infectionisessentialforacompleteunderstandingofthecomplex imiquimod also elevated basal superoxide production through immunopathogenesis associated with the development of mild or enhancedNOX2activityinmacrophages[67].Furtherstudiesare severedenguefeverinpatients.Previousstudieshavedemonstrat- now required to determine the exact mechanism(s) involved in ed that DENV can trigger an innate immune response that DENV-inducedNOX-dependentROSproductioninhumanMo- includes the release of antiviral and inflammatory cytokines DC. [50,53,54], while other studies demonstrate the ability of DENV ROS were long considered as toxic, microbe-induced by- toantagonizetheinductionofinnateresponsesviacleavageofthe products involved inthekillingofpathogens[18];however, their endoplasmicreticulumadaptorSTING[51,55].Touncovernovel function as second messengers that regulate immune signaling regulatory pathways involved in DENV infection of Mo-DC, we suggestsamuchbroaderroleinhostdefenseagainstviruses[19– have for the first time used a transcriptome-wide expression 23]. ROS production was in fact required to trigger the antiviral analysis, coupled with biochemical dissection, to investigate the and inflammatory responses to DENV infection in DC, and was early host response to DENV infection in primary human confirmed by both chemical and genetic inhibition of the NOX dendritic cells - an important pool of cells infected early in vivo complex. Blockade of NOX activation or ROS production afterthebiteofthemosquitoAedesaegypti.Here,wedemonstrate inhibited antiviral and inflammatory responses, including the that: 1) DENV preferentially infected myeloid cells as they IRF3/STAT1 antiviral axis and the NF-kB inflammatory differentiated in vitro to mature Mo-DC; 2) DENV2 infection pathway (Fig.4). The IRF3 pathway has previously been triggered antiviral, inflammatory, and oxidative stress pathways demonstrated to be regulated by oxidative stress variations. with distinct kinetics; 3) DENV2 infection generated a NOX- Indeed,theexpressionlevelofthenon-canonicalIKK-likekinase, dependentintracellularaccumulationofROS;4)ROSproduction IKKe, is itsef NOX-regulated and participated in the immune mediated activation of the IRF3/STAT1- and NF-kB-mediated response induced by the respiratory syncytial virus (RSV) [68]. innate immune responses; 5) ROS production mediated p53 NOX-derived ROS were also shown to activate the RIG-I/ mitochondrial-dependent apoptosis and contributed to bystander MAVS/IRF3antiviralaxisinepithelialcells,andwererequiredto Mo-DCmaturation/activation;and6)Nrf2-regulatedtargetgenes maintain the constitutive level of MAVS expression [22]. In limited the oxidative stress response, and ultimately modulated contrast, statistical changes in MAVS or IKKe expression ROS-induced immune and apoptotic responses. These results following NOX inhibition in primary DENV-infected DC were highlight a requirement for the oxidative stress response in the notobservedinthisstudy(S8A–CFigure),suggestingthatDENV- generation of the host innate immune response to DENV induced ROS may regulate host response via post-translational infection. modification of proteins involved in antiviral signaling, as was Activation of the NADPH-oxidase (NOX) complex and described previously for S-glutathionylation of TRAF3 and generation of reactive oxygen species (ROS) has been described TRAF6 [19]. Other non-infectious biological processes such as for several viral infections, including hepatitis C virus (HCV), impairment of autophagy also support the idea that oxidative Rhinovirus, and HIV [56–58]. We demonstrate that DENV stress modulates the sensitivity to antiviral signaling. Indeed, infectionalsoactivatesNOX-dependentROSproductioninMo- blocking of autophagy allows for oxidative stress accumulation DC.Insomeinfectionmodels,viralproteinssuchasNefandTat through defective mitochondria and leads to the amplification of for HIV and NS3 for HCV were shown to specifically stimulate RLR signaling [69]. Altogether, these studies cumulatively the NOX complex [59–61]. NOX activity was also regulated by highlight the complexity of ROS involvement in the stimulation spleen tyrosine kinase (Syk)-mediated phosphorylation of the ofantiviralresponsesandarguesthattheinnateimmuneresponse NOX p47phox subunit [62]; Syk kinase is downstream of the integrates both viral RNA sensing and detection of homeostatic surface receptor CLEC5A, which was shown to promote perturbations tocoordinatean appropriate host response. inflammasome activation and inflammatory cytokine release in The Nrf2-mediated antioxidant response was one of the top DENVinfection[63,64].Importantly,TLR3,areceptorcritically differentially regulated pathways early after DENV infection, involvedinRNAsensing,wasrecentlyshowntostimulateNOX- resultingintheexpressionofmanycytoprotectiveenzymessuchas dependent ROS production that was required for NF-kB, IRF3 HMOX-1, SOD2, NQO1, GCLC and GCLM, that function andSTAT1activationinmurinemacrophagesinresponsetothe togethertomaintainanappropriateredoxstatus,andthusprotect PLOSPathogens | www.plospathogens.org 10 December2014 | Volume 10 | Issue 12 | e1004566
Description: