Virology484(2015)313–322 ContentslistsavailableatScienceDirect Virology journal homepage: www.elsevier.com/locate/yviro Brief Communication A G-quadruplex-binding macrodomain within the “SARS-unique domain” is essential for the activity of the SARS-coronavirus replication–transcription complex Yuri Kusova,b, Jinzhi Tana, Enrique Alvarez1,c, Luis Enjuanesc, Rolf Hilgenfelda,b,n aInstituteofBiochemistry,CenterforStructuralandCellBiologyinMedicine,UniversityofLübeck,Lübeck,Germany bGermanCenterforInfectionResearch(DZIF),Hamburg–Lübeck–BorstelSite,UniversityofLübeck,Germany cDepartmentofMolecularandCellBiology,CentroNacionaldeBiotecnología(CNB-CSIC),CampusUniversidadAutónoma,Madrid,Spain a r t i c l e i n f o a b s t r a c t Articlehistory: Themulti-domainnon-structuralprotein3ofSARS-coronavirusisacomponentoftheviralreplication/ Received1March2015 transcription complex (RTC). Among other domains, it contains three sequentially arranged macro- Returnedtoauthorforrevisions domains:theXdomainandsubdomainsSUD-NaswellasSUD-Mwithinthe“SARS-uniquedomain”. 19March2015 TheXdomainwasproposedtobeanADP-ribose-1”-phosphataseorapoly(ADP-ribose)-bindingprotein, Accepted12June2015 whereasSUD-NMbindsoligo(G)-nucleotidescapableofformingG-quadruplexes.Here,wedescribethe application of a reverse genetic approach to assess the importance of these macrodomains for the Keywords: activity of the SARS-CoV RTC. To this end, Renilla luciferase-encoding SARS-CoV replicons with SARS-CoVreplicon selectivelydeletedmacrodomainswereconstructedandtheirabilitytomodulatetheRTCactivitywas SARS-uniquedomain examined.WhiletheSUD-NandtheXdomainswerefoundtobedispensable,theSUD-Mdomainwas Macrodomain crucialforviralgenomereplication/transcription.Moreover,alaninereplacementofchargedamino-acid X-domain residuesoftheSUD-Mdomain,whicharelikelyinvolvedinG-quadruplex-binding,causedabrogationof Reversegenetics G-quadruplex RTCactivity. MERS-CoV &2015ElsevierInc.Allrightsreserved. Introduction by the finding that SARS-CoV replicons are able to replicate autonomously in transfected host cells (Almazán et al., 2006; Ge TheSevereAcuteRespiratorySyndromecoronavirus(SARS-CoV) etal.,2007;Erikssonetal.,2008;Panetal.,2008;Wangetal.,2008; belongstolineagebofthegenusBetacoronavirus.DuringtheSARS reviewed in Almazán et al., 2014). Replicons, in particular those outbreak of 2003 (see Hilgenfeld and Peiris (2013), for a recent encoding a reporter gene, considerably facilitate the functional review), the genome of SARS-CoV was sequenced within three analysis of molecular determinants that control the replication weeksofthediscoveryofthevirus(Marraetal.,2003;Rotaetal., and transcription of the SARS-CoV genome. SARS-CoV replicons 2003) and subjected to detailed annotation shortly thereafter consistofthegenomic50 and30 untranslatedregionsandanopen (Snijder et al., 2003). At the same time and in subsequent years, readingframe(ORF)encodingthetwopolyproteins(pp1aand,viaa leadsforantiviraltherapyweredescribed(Anandetal.,2003;Yang (-1) frame-shift, pp1ab) that are processed into 16 non-structural et al., 2005; see Hilgenfeld (2014), for a recent review). However, proteins (Nsp1 to Nsp16) to form the replication/transcription thedetailedmolecularcharacterizationofviralgenomereplication complex (RTC, Fig.1A). Among the Nsps, Nsp3 is a multi-domain and the assembly of viral particles was initially restricted to polypeptide comprising the following structurally organized specialized high-safety laboratories. This limitation was overcome domains(Serranoetal.,2009;Johnsonetal.,2010):theubiquitin- likedomain1(UB1),thepartiallydisorderedacidicdomain(Ac),the Xdomain,andtheSARS-uniquedomain(SUD).Thisisfollowedby the second ubiquitin-like domain (UB2), which is usually consid- nCorrespondingauthor. E-mailaddresses:[email protected](Y.Kusov), ered part of the next domain, the papain-like protease (PLpro). C- [email protected](J.Tan),[email protected](E.Alvarez), terminaltothese,there isthe nucleic-acid-bindingdomain(NAB), [email protected](L.Enjuanes), followed by a longer stretch of amino-acid residues apparently [email protected](R.Hilgenfeld). lackingsecondarystructure(occasionallycalled“coronavirusgroup iord1ePIrnevseensttigaadcdiorensess:CCeiennttríoficdaes.BUionloivgeíarsMidoadlecAuulatórnSoemvearodeOcMhoaad.riCdon(CseSjIoC-SUuApMer)-. 2 marker (G2M)”; Neuman et al., 2014) and the first transmem- NicolásCabrera1,28049Madrid,Spain. braneregion(TM1).Thezinc-finger(ZF)istheonlydomainonthe http://dx.doi.org/10.1016/j.virol.2015.06.016 0042-6822/&2015ElsevierInc.Allrightsreserved. 314 Y.Kusovetal./Virology484(2015)313–322 introducedinsubdomainspresumablyinvolvedintheinteraction withG-quadruplexes(Tanetal.,2009). Results IntroductionofdeletionsandpointmutationsintotheSARS-CoV repliconencodingRenillaluciferase Comparing genome transcription with genome replication, it was previously shown that the replicon pBAC-REP, which was lackingareportergene,wasabletoreplicateinmammaliancells (Almazán et al., 2006). These data encouraged us to introduce a luciferasereportergeneintotherepliconpBAC-REPtoenablethe quantitative detection of viral genome replication simply by measuringtheluciferaseactivity.TheRenillaluciferasewaschosen asareporterproteinbecauseofitslongerhalf-lifeascomparedto thatofthefireflyluciferase(seeMaterialsandmethodsandTanaka et al., 2012). However, the constructed full-length SARS-CoV Fig.1. Schematicpresentations(nottoscale)ofthegeneticallyengineeredSARS- CoVrepliconencodingRenillaluciferase(A),ofthedomainorganizationofNsp3 replicon DNA, pBAC-REP-RLuc, was too large (approximately (B),andofconstructswithdeletedfragments(ΔX,SUD-ΔN,SUD-ΔM,SUD-ΔNM, 33kb) to ensure the correctness of desired deletions and/or andSUD-ΔC)withindomainsX–SUDofNsp3(C).Nsp–nonstructuralprotein1–16, mutations. Therefore, the engineered mutants lacking different Mpro – main (or 3C-like, 3CL) protease, prim/pol – non-canonical polymerase genomefragments(completeSUD,orsubdomainsSUD-N,SUD-M, activity(Xiaoetal.,2012;teVelthuisetal.,2012),ssRBP–single-strandedRNA- binding protein, RdRp – RNA-dependent RNA polymerase, Hel – superfamily-1 SUD-NM,and,finally,SUD-C,Fig.1C)werefirstintroducedbysite- helicase, ExoN/(7)N-GMtase – 30-to-50 exonuclease/7N-guanylmethyltransferase, directed mutagenesis into the shorter plasmid pBAC-SfoI-MluI NendoU–U-specificendoribonuclease,20O-Mtase–20O-methyltransferase,TRS- (approx.15kb, Supplementary Fig.1B, Almazán et al., 2006). The M–transcription-regulatorysequenceoftheMprotein,RLuc–Renillaluciferase,N fragmentscontainingthedeletionswerethenexcisedfromthese – nucleocapsid protein, pA – a synthetic poly(A) tail, Rz – hepatitis deltavirus ribozyme, BGH – bovine growth hormone termination and polyadenylation pBAC-SfoI-MluI-derived plasmids and transferred back into plas- sequence,UB1–ubiquitin-likedomain1,Ac–acidicdomain,X–Xdomain,SUD midpBAC-REP-RLuc.Torecoverthereplication/transcriptionactiv- –SARS-uniquedomain,UB2–ubiquitin-likedomain2precedingthepapain-like ity, the correctly oriented MluI-MluI fragment was inserted into cysteineprotease(PLpro),NAB–nucleicacid-bindingdomain,G2M–coronavirus thefinalplasmids.SupplementaryFig.1depictsourmainstrategy group2 marker, TM1 and TM2 – transmembrane regions, ZF – zinc-finger, Y – exemplified by the construction of the SUD-deleted SARS-CoV uncharacterizeddomain.AsshownschematicallyinpanelC,Δindicatesadeletion replicon encoding Renilla luciferase as reporter protein (pBAC- oftheXdomainorofdomainsN,M,C,NM,orthecompleteSUD-codingsequences. Δ SUD-REP-RLuc). To examine whether SUD function within the RTC might be luminalsideof themembrane;thesecondtransmembraneregion connected toits binding to oligo(G)-stretches capable of forming (TM2)and,finally,theuncharacterizedYdomainarelocalizednear G-quadruplexes(Tanetal.,2009),aSARS-CoVrepliconcontaining theC-terminusofNsp3(Fig.1B,seeNeumanetal.(2014)). the mutations K565A, K568A, and E571A within the SUD-M The X domain was shown to have a macrodomain fold and domain(referredtoasthemut4setofmutations)wasconstructed proposedtobeanADP-ribose-1″-phosphatephosphatase(ADRP)or followingasimilarapproach.Wehadshownpreviouslythatthese apoly(ADP-ribose)-bindingmodule(Saikatenduetal.,2005;Egloff mutations abrogate G-quadruplex binding to SUD-M (Tan et al. etal.,2006).WehaveshownpreviouslythattheSUDcontainstwo 2009). The replicon containing the mutations K476A and K477A consecutive macrodomains, called SUD-N and SUD-M (Tan et al., (the mut2 set of mutations) located near the C-terminus of the 2009). Although their functional role in the viral replication cycle SUD-N domain was also prepared for comparison, since these remainsunknown,wehaveshownthatthesemacrodomainsbind mutationshadaweakereffectonSUD-oligo(G)bindingaccording nucleicacidsthatcontainlongguaninestretchescapableofforming to zone-interference gel electrophoresis (Tan et al., 2009). To G-quadruplexes(Tanetal.,2007,2009). further test our hypothesis, the mutation sets mut2 and mut4 Inthepresentstudy,weinvestigatedtheX,SUD-N,andSUD-M were also introduced into the replication-competent replicon Δ macrodomainstodeterminewhethertheyplayimportantrolesin pBAC- X-REP-RLuc lackingthe Xdomain (seebelowand Fig.4A viralgenomereplication/transcriptionandwhethertheyactincis andB). orintrans.Tothisend,wehaveintroducedin-framedeletionsinto the X and SUD regions of the SARS-CoV replicon, into which a Replication-competenceofthereportergene-containingSARS-CoV reportergene(Renillaluciferase,RLuc)wasintroducedunderthe repliconpBAC-REP-RLuc control of the transcription regulatory sequence (TRS) for SARS- CoVstructuralproteinM(Fig.1A).Inaddition,thefunctionalrole Toensurethatthereportergeneexpressiondirectlycorrelated oftheSUDC-terminalregion(SUD-C,afrataxin-likedomainwith withviralgenomesynthesis,thelevelofviralRNAsynthesiswas asyetunknownfunction(Johnsonetal.,2010)),wasassessedina measured in parallel to Renilla luciferase activity. This was similar way. The reporter gene-containing replicons were tested achievedbyquantitativeRT-PCR(qRT-PCR)usingprimerscovering fortheirabilitytosupporttheactivityoftheRTCinthesynthesis the non-structural protein 1 (Nsp1) region (see Materials and of subgenomic replicon RNA and to assess whether the deleted methods). There was a direct correlation of Renilla luciferase functionscouldberescuedintrans.Thedatapresentedindicatea expressionwiththeamountofviralRNAsynthesisinmammalian crucialrolefortheSUD-MmacrodomainforviralRTCactivity,thus cells transfected with the SARS-CoV replicon pBAC-REP-RLuc lending support to the significance of the previously observed (Fig. 2A and B, columns “REP”). Due to the specificity of the binding of SUD to oligo(G)-containing nucleic acids (Tan et al., forward and reverse primers, the qRT-PCR system gave a non- 2007, 2009). This observation was additionally reinforced by the significant background (Fig. 2B, column “mock”). A replication- analysis of replicons in which amino-acid mutations have been defective construct, pBAC-REP(NR)-RLuc, with a reverse Y.Kusovetal./Virology484(2015)313–322 315 Fig.2. ComparisonofthelevelofRenillaluciferaseexpression(A)andviralRNAsynthesis(B)byreplication-competent(REP)andreplication-deficient(NR)SARS-CoV replicons;time-courseofRLucexpressionbytheSARS-CoVreplicon(C);theeffectofSUDdeletion(twoclones,clon1andclon2)onRenillaluciferaseexpression(D),effectof deletionoftheXdomain(ΔX),SUDmacrodomains(ΔNandΔM),andoftheSUD-Csubdomain(ΔC)ontheexpressionofRenillaluciferase(E)andonviralRNAsynthesis(F). SARS-CoVrepliconswithdeleteddomainsweretransfectedinVeroE6cellsandafter24hofincubation,RenillaluciferaseactivityandtheamountofviralRNAwere measured(seeMaterialsandmethods).RenillaluciferaseactivityandviralRNAsynthesiswerealsomeasuredaftertransfectionofareplication-deficientSARS-CoVreplicon (NR)andinmock-transfectedcells(Cells).Datashownarefromquadruplicateexperiments,expressedasthemeanvalueofRNAcopies7standarddeviation(SD).The differenceinexpressionofthefull-lengthSARS-CoVrepliconanditsvariousmutantswasfoundtobesignificant,implyingthatvaluesamongthemaregreaterthanwould beexpectedbychance(0.014p40.001).Notethatinthereplication-deficientSARS-CoVreplicon(NR),thesequenceoftheMluI-MluIfragmentisreversed,thuspreventing theformationofthereplicasecomplex.RLU–relativelightunits. orientationoftheMluI-MluIfragmentusedasanegativecontrol, correlated with viralgenomesynthesis, Renilla luciferase activity neither expressed Renilla luciferase nor was able to synthesize expressed by constructs lacking individual domains (in this case viral RNA (Fig. 2, A and B, columns “NR”). These results clearly theXdomain)wasdirectlycomparedwithRNAsynthesis.Ahigh indicate that Renilla luciferase activity is a good reporter for level of viral RNA synthesis was observed for the X-domain- analyzingviralgenomereplication. deleted construct (Fig. 2F, column “ΔX”). Thus, not only is the Tofindouttheoptimumtimeforthereporter-geneexpression ADRP activity of the X domain not required for coronavirus by the engineered SARS-CoV replicon, the Renilla luciferase replication in tissue culture, as shown for HCoV 229E (Putics activity was analyzed at different time-points after transfection et al., 2005), but the entire X domain is dispensable in case of of pBAC-REP-RLuc into Vero E6 cells. As shown in Fig. 2C, the SARS-CoV(ourdata). reportergeneexpressionshowedalinearincreasebetween18and 53h posttransfection (hpt), with the highest value of Renilla SUDisindispensableforSARS-CoVgenometranscription/replication luciferase activity detected around 53hpt. Therefore, the expres- sionbetween24and48hptwasrecordedinourstudy. IncaseoftheSUD,wehaveshownthatitpreferentiallybinds oligo(G) stretches (G-quadruplexes) (Tan et al., 2007, 2009). To TheXmacrodomainofSARS-CoVisdispensableforRTCactivity assesstheimportanceofSUDfortheactivityoftheRTC,theentire SUD-encodingsequencewasdeletedfromtheSARS-CoVreplicon, Afewregionsnearthe50endofthecoronavirusgenome(nsp1 as described above. The ability of the deleted construct pBAC- Δ of murine hepatitis virus, MHV (Denison et al., 2004; Brockway SUD-REP-RLuctoreplicateitsgenomewascomparedtothatof and Denison, 2005, Tanaka et al., 2012), and nsp2 of MHV and the parental replicon pBAC-REP-RLuc. In contrast to the latter SARS-CoV (Graham et al., 2005)) have been showntobe dispen- (Fig. 2D, column “REP”), two independently prepared clones of Δ sable for virus replication (reviewed in Neuman et al., 2014). In pBAC- SUD-REP-RLuc,transfectedinVeroE6cells,expressedthe addition, the ADRP of Nsp3b, i.e. the X domain of human Renilla luciferase gene only at a level similar to background coronavirus 229E (HCoV 229E) can be inactivated by mutation activity, suggesting that the SUD is indispensable for SARS-CoV without significantly affecting viral replication in cell culture RTCactivity(Fig.2D,cf.columns“ΔSUD“withcolumn“mock”). (Putics et al., 2005). To find out whether the X domain may display another, as yet uncharacterized activity involved in viral Lackoftrans-complementationofthereplication-deficientSUD- replicationortranscription,weremovedthecodingsequencefor lackingSARS-CoVrepliconbythefull-lengthSUDandSUD-NM thecompletedomainfromtheSARS-CoVrepliconandfoundthat Δ the resulting replicon, pBAC- X-REP-RLuc, was replication- To answer the questionwhether the SUD function is required competent. Approximately 70–75% of the parental replicon exclusively in cis or can be provided in trans, the full-length (pBAC-REP-RLuc) activity (Fig. 2E, column “REP”) was observed SUD or its more stable SUD-NM fragment was co-transfected Δ for the expression of the X domain-deleted replicon (Fig. 2E, togetherwiththeSUD-deletedrepliconpBAC- SUD-REP-RLucin column “ΔX”). To ensure that the reporter-gene expression Huh-T7 cells susceptible to SARS-CoV (Gillim-Ross et al., 2004; 316 Y.Kusovetal./Virology484(2015)313–322 Hattermannetal.,2005).However,thereportergeneactivitydid expression using vaccinia virus (VV) MVA-T7 as a helper virus. not exceed background level (Fig. 3, panel A). None of the Previously, we have successfully used MVA-T7 to efficiently constructsexpressingSUDorSUD-NM(columns2and3,respec- express hepatitis A virus genes (Kusov et al., 2002). Indeed, the tively) was able to considerably enhance the extremely low transfectionofconstructsexpressingSUDorSUD-NMfollowedby reporter gene activity of the SUD-lacking replicon co-transfected infection with the helper virus MVA-T7 (a procedure known as with vector alone (column 1). Intriguingly, the co-expression of transinfection, Kusov et al., 2002) allowed immunological detec- SUD or SUD-NM with the replication-competent SARS-CoV repli- tion of SUD and SUD-NM using either polyclonal SARS-CoV anti- con (REP-RLuc, columns 5 and 6, respectively) slightly inhibited Nsp3(Rockland;resultnotshown)ormonoclonalanti-His (Qia- 4 the Renilla luciferase activity expressed by the replicon co- gen) (Fig. 3C). Note that the level of Renilla luciferase expression transfected with the vector pIVEX used as a control (column 4). bytheSUD-deletedrepliconwasalsoincreased,probablybecause Assuming that the inability to complement the SUD-deleted of more efficient mRNA transcription from a cryptic promoter or replicon by providing SUD or SUD-NM in trans was due to low duetoahelpereffectofVVforrepliconRNAsynthesis,asweand levels of protein production, we increased the amount of others have noticed previously (Sutter et al., 1995; Kusov et al., Fig.3. ImpactofSUDandSUD-NMproteins,providedintrans,ontheactivityofthefull-lengthandSUD-deletedSARS-CoVreplicons.IndicatedDNAswereco-transfectedin Huh-T7cells(A)oradditionallyinfectedwithvacciniavirusMVA-T7asahelpervirus(seeMaterialsandmethods)(B).Theinfectionand/orco-transfectionmixtureswere replacedwithgrowthmediumandincubatedforanadditional48h.ThecelllysateswereanalyzedforRenillaluciferase(AandB,po0.01)andanti-Hisimmunological activityinmixtures2and3showninpanelB(C).Allexperimentswererunintriplicateandtheaveragewasemployedforanalysis.Errorbarsrepresentstandarddeviations ofthemeanvalues.Thedifferenceinreporterproteinactivityofthefull-lengthrepliconandtheSUD-lackingrepliconwasfoundtobestatisticallysignificant(panelA, p¼0.005;panelB,p¼0.01). Y.Kusovetal./Virology484(2015)313–322 317 α 2002).However,noneoftheproteins,SUD(Fig.3B,column2)or are located in the loop connecting the second -helix with the β SUD-NM (column 3), was able to increase the reporter-gene third -strand of the SUD-M domain, which was found to be Δ expression of the SUD-deleted replicon pBAC- SUD-REP-RLuc essentialforreplicationoftheSARS-CoVreplicon(Fig.2EandF).In (column 1), indicating that the function(s) of SUD cannot be contrast, the mutations K476A and K477A (the mut2 set of complemented in trans. To double-check these results, we have mutations) located in the dispensable SUD-N domain (Fig. 2E tried to supplement the parental SARS-CoV replicon pBAC-REP- and F) had only a minor effect on oligo(G)-binding in vitro (Tan RLuc.InlinewiththedatapresentedinFig.3A,theco-expression etal.,2009). ofSUDorSUD-NMratherpartiallyinhibitedtheRenillaluciferase Tocomparetheeffectofmutationsonviralgenomereplication/ activity of the replication-competent replicon (Fig. 3, cf. columns transcriptioninvivo,the twosets ofmutations (mut2 and mut4) 4and5inpanelsAandB),implyingthatpossiblyafinebalanceof wereintroducedseparatelyintotwoSARS-CoVrepliconscontain- Δ the proteins is crucial for the formation of the active SARS-CoV ing the full-length SUD sequence, pBAC-REP-RLuc and pBAC- X- replicasecomplex.Takentogether,thefunctionalactivityofSUDin REP-RLuc,whichwereabletoefficientlyreplicatetheirgenomeas ciscouldnotbesupplementedintrans,neitherbythefull-length judged by Renilla luciferase expression and viral RNA synthesis Δ SUDnorbyitsmorestableversionSUD-NMconsistingofthetwo (Fig. 2). A replicon lacking the SUD-C subdomain, pBAC-SUD C- macrodomains,SUD-NandSUD-M. REP-RLuc, which contains both the SUD-N and SUD-M domains, was not considered for site-directed mutagenesis because of its TheSUD-Mmacrodomainiscrucialfortheactivityofthereplication/ lowactivity. transcriptioncomplex Theimpactofthemut2andthemut4setofmutationsonthe activityoftheRTCwasexaminedbyreportergeneexpressionand ToassesstheroleofthesubdomainsofSUDfortheactivityof viralRNAsynthesisasmentionedabove.Comparedtotheoriginal theRTC,theSUD-NandSUD-MmacrodomainsandtheC-terminal constructs,thenon-mutatedrepliconpBAC-REP-RLuc(Fig.4A)and Δ subdomain (SUD-C) were deleted in separate experiments from itsXdomain-deleted derivativepBAC- X-REP-RLuc (Fig.4B),the thesequenceoftheSARS-CoVreplicon(Fig.1).Wecomparedthe corresponding replicons containing the mut2 set of mutations activities of the replicons lacking individual SUD subdomains to (K476A and K477A) expressed Renilla luciferase activity three thatoftheparentalSARS-CoVreplicon,pBAC-REP-RLuc.TheSARS- times more weakly (Fig. 4A, -SUDmut2-), or even at the same Δ CoVrepliconlackingtheSUD-Ndomainexpressed30to35%ofthe level (Fig. 4B, - X-SUDmut2-). Probably, this difference is con- Renillaluciferaseactivityoftheparentalreplicon(Fig.2E,column nectedwiththereducedactivityoftheSARS-CoVrepliconlacking “ΔN”).Accordingly,itsRNAlevelwaswelldetectableincontrastto theXdomain(seeFig.2EandF).Incontrasttotheeffectofmut2 other SUD subdomain-deleted replicons (Fig. 2F, compare the variants,theRenillaluciferaseactivityofrepliconscontainingthe column “ΔN” with those for ΔM and ΔC). These data indicate mut4 set of mutations was negligible, thereby emphasizing the thattheSUD-Nmacrodomainmayalsobeconsidereddispensable crucialroleoftheresiduesalteredinthemut4setofmutationsfor Δ for SARS-CoV replication, similar to the X domain (see above), genomereplication(Fig.4AandB,-SUDmut4-and- X-SUDmut4- althoughtheireffectsonreplicationmaybeconsidered“minor”(X ). These results were in close agreement with the levels of domain) and “moderate” (SUD-N domain), respectively. In sharp quantified viral RNA (not shown). Taken together, the effect of contrast, the Renilla luciferase activity expressed by the replicon chargedamino-acidresiduesonthein-vitrobindingofSUD-NMto with the SUD-M domain deleted (Fig. 2E, column “ΔM”) did not G-quadruplexesstrictlycorrelatedwiththeactivityofthemutated exceed the level of the activity expressed by the replication- SARS-CoVrepliconsinvivo. deficientrepliconpBAC-REP(NR)-RLuc(Fig.2E,column“NR”).This activity was similar to the background level detected in mock- transfectedcells(Fig.2E,column“Cells”).Thelackofreplicationof Discussion theSUD-Mdomain-deletedreplicon,deducedfromthenegligible levelofRenillaluciferaseactivity,wasconfirmedbyquantification The highly infectious and virulent Severe Acute Respiratory of the viral genome (Fig. 2F, column “ΔM”). The pBAC-SUDΔC- Syndromecoronavirus(SARS-CoV),classifiedasabiosafetylevel-3 REP-RLuc replicon was able to replicate only to a significantly agent (BSL3), can only be handled in specially equipped labora- lowerextentthantheconstructlackingtheSUD-Ndomain(Fig.2, tories. To overcome this limitation and to avoid the use of E and F, cf. columns “ΔC” and “ΔN”). Nevertheless, its Renilla dangerous live virus, SARS-CoV replicons have been engineered luciferaseexpressionandthesynthesisofviralRNAwerealwaysat (Almazánetal.2006,2014;Geetal.,2007;Erikssonetal.,2008; detectablelevels(Fig.2,EandF,columns“ΔC”),incontrasttothe Pan etal.,2008; Wangetal.,2008).Here,wehaveconstructeda activityoftheSUD-M-lackingreplicon(column“ΔM”)ortothatof SARS-CoV replicon containing Renilla luciferase as reporter gene, therepliconwiththeSUD-NMdomainsdeleted(notshown).The thusallowingnotonlytheeasyscreeningofchemicallibrariesfor RTCactivityofthelatterrepliconswaseitheratbackgroundlevel antivirals interfering with replication of viral RNA and the char- (columns “Cells”) or at the level of the replication-deficient acterization of antiviral lead compounds, but also studies of the construct (columns “NR”). In summary, among all tested SARS- function of various viral proteins and regulatory sequence ele- CoVrepliconswiththeabove-mentioneddeletions,thereplication mentsbyreversegenetics. ability was mostly affected by the deletion of the SUD-M Our aim was to elucidate the functional role of the three macrodomain. sequential macrodomains within Nsp3 for the activity of the SARS-CoV RTC. To this end, we have created various SARS-CoV Chargedamino-acidresiduesoftheSUD-Mmacrodomain repliconswithdeletedand/ormutatedmacrodomains.First,theX presumablyinvolvedinG-quadruplexbindingareessentialforSARS- domain-encoding sequence was deleted from the Renilla CoVRTCfunction luciferase-containingSARS-CoVrepliconandtheresultingreplicon was tested for its ability to express the reporter gene. The X Among all previously tested sets of amino-acid replacements domainhasbeenshowntoexhibit aweakADRPactivity inmost thatwereabletoaffectthebindingofSUDtoG-quadruplexes,the coronavirusesexamined, butwehave previouslyshownthatthis mut4 setof mutations, comprising alanine substitutions of K565, activityisnotcompletelyconservedacrossthefamily;thus,theX K568, and E571 of Nsp3, was most efficient in preventing the domainoftheBeaudettestrainofInfectiousBronchitisVirus(IBV) binding of oligo(G) (Tan et al., 2009). These amino-acid residues isunabletobindADP-riboseduetoreplacementofaGly-Gly-Gly 318 Y.Kusovetal./Virology484(2015)313–322 Fig.4. Effectofamino-acidreplacementsintheSUDmacrodomainsonviralgenometranscription/replication.IndicatedmutationsinSUD-NandSUD-M(designated– SUDmut2-and-SUDmut4-,respectively)wereintroducedintotheSARS-CoVrepliconpBAC-REP-RLuc(A)anditsderivativelackingtheXdomain,pBAC-ΔX-REP-RLuc(B) (seediagramsattherightsideoftheconstructs).TheabilityoftheconstructstoexhibitRenillaluciferaseactivityispresentedasmeanvaluesoftriplicateexperiments7SD (po0.05).AputativemodelofG-quadruplexbindingtoSUD-NM,obtainedbyautomateddockingintothecrystalstructure(C,modifiedfromTanetal.,2009).TheSUD-N andSUD-Mmacrodomainsareinvioletandcyan,respectively.TheG-quadruplexasfoundintheBCL2promoterregion(PDBcode:2F8U,Daietal.,2006)isinorange.The mut2setofmutations(K476AþK477A),locatedattheC-terminusoftheSUD-Ndomain,isindicatedbyyellowsticks.Themut4setofmutations(K565AþK568AþE571A),of SUD-M,isindicatedbygreensticks.Theseresiduesbelongtoaclusterofchargedamino-acidresidueslocatedinthesecondα-helixofSUD-Mandintheloopconnectingit withthethirdβ-strand(R562,K563,K565,K568,andE571,green,red,blue,magenta,andorangelabel,respectively)(D). tripletinthebindingsitebyGly-Ser-Gly(Piotrowskietal.,2009). positioned macrodomains, i.e. X, SUD-N, and SUD-M, may have Furthermore,Puticsetal.(2005)reportedthattheADRPactivityof (an) as yet unidentified function(s) enhancing SARS-CoV genome HCoV229Eisnotessentialforgenomereplicationofthisvirusin replication.Sincefull-lengthSUDwasfoundtobecrucialforSARS- cellculture.Inthisstudy,wefindthattherepliconpBAC-ΔX-REP- CoV genome replication (Fig. 2D) and not able tocomplement in RLuc expressed a high level of Renilla luciferase and was able to trans (Fig. 3), we decided to gain further insight into the role of synthesizeviralRNAaftertransfectionintomammaliancells(Vero each SUD macrodomain for RTC activity. In addition, the C- E6 or Huh-T7, a derivative of SARS-CoV-susceptible Huh-7 cells terminal SUD subdomain (SUD-C) was also investigated by (Gillim-Rossetal.,2004;Hattermannetal.,2005)).Thisindicates removalof theSUD-C-coding sequencefromtheSARS-CoVrepli- that the X domain does apparently not carry another, as yet con.ThedeletionoftheSUD-Mdomaincompletelyabolishedboth unidentified, activity that would be essential for SARS-CoV repli- replicon activities, thus indicating that the SUD-M domain is cation/transcription.InsharpcontrasttoresultswiththedeletedX indispensable for SARS-CoV genome replication. In contrast, the Δ domain, the SUD-lacking replicon, pBAC- SUD-REP-RLuc, abro- SUD-N domainwas revealed to be non-essential for RTC activity gated reporter gene expression and the synthesis of viral RNA since its removal abolished neither Renilla luciferase expression (Fig.2Dandnotshown,respectively).Toourknowledge,thisisthe norviralRNAsynthesis(Fig.2,EandF,columns“ΔN”).Thesame first description of the indispensability of SUD for SARS-CoV istruefortheSUD-Csubdomain,althoughtheRTCactivityofthe genome replication. The most plausible explanation for this SUD-C-lackingrepliconwasalwaysatalowlevel(Fig.2,EandF, observation is an essential role of the SUD-M macrodomain for columns “ΔC”). Intriguingly, three amino-acid residues (K565, α viralRNAsynthesis(seebelow). K568, and E571) located in the loop following the second - Interestingly,allourattemptstocomplementintranstheSUD helix of SUD-M and responsible for the interaction of SUD with function by co-expression of the full-length SUD or SUD-NM oligo(G)-nucleotides have previously been identified (Tan et al., failed, albeit the expression was evidenced by immunological 2009). We propose that the SUD-oligo(G) interaction is required detection of the proteins (see Fig. 3). Also, the replication of the forSARS-CoVgenomereplication.TheSARS-CoVgenomecontains completeSARS-CoVrepliconwasnotenhancedwhenafull-length a number of conserved G stretches (Johnson et al., 2010) that 4 SUD or SUD-NM were provided in trans (Fig. 3). The inability to couldbebinding-partnersforSUD.Oligo(G)sequencesarecapable supply SUD activity in trans rules out a hypothetical enzymatic of forming G-quadruplexes, even if they comprise a few non- activity of SUD. Taken together, these results indicate that the guanines;thus,theNMRstructureoftheprototypeG-quadruplex functionally active SUD is required only in cis or, alternatively, a thatweusedfordockingintotheSUD-NMcavitycontains8non- fine balance of the proteins is essential for the formation of the guaninesamong23nucleotides(Daietal.,2006).Therefore,oligo activeSARS-CoV replicasecomplex. These observations maysub- (G) regions such as, e.g., 50-GGGAGGUAGG-30, which is found stantiatethepreviouslypublisheddataonenhancementofSARS- conservedintheNsp2-andNsp12-codingregionsinthegenomes CoV reporter activity by the co-expression of Nsp3.1, which ofdifferentSARS-CoVstrains(Johnsonetal.,2010),arecandidates comprises domains X and SUD (Pan et al., 2008). Now we can forinteractionwithSUD. assumethatthisenhancementwasmainlyduetothepresenceof WhiletheseobservationssupportanessentialroleoftheSUD- the X domainwithin Nsp3.1. Alternatively, the three sequentially M domain in replication/transcription of the SARS-CoV genome, Y.Kusovetal./Virology484(2015)313–322 319 probably via interaction with G-rich stretches forming G-quad- butnegativelycharged(Fig.4D,orangelabel,andSupplementary ruplexes, we have to consider a potential negative effect of SUD Fig. 3) needs to be further investigated. In summary, we have domain deletions on the activity of the papain-like protease identified amino-acid residues essential for SARS-CoV genome (PLpro),locatedimmediatelydownstreamofSUD.Suchamodula- replication,whichrequiresSUD-oligo(G)interaction.Interestingly, tion of PLpro activity by flanking regions has been reported for thisclusterofchargedamino-acidresidueslocatedinthesecond SARS-CoV PLpro (Harcourt et al., 2004; Han et al., 2005) and α-helixoftheSUD-Mdomain,orintheloopfollowingit(Fig.4D; alphacoronavirus PL2pro (Ziebuhr et al., 2001, 2007). Therefore, R562, K565, K568, and E571, green, blue, magenta, and orange to demonstrate a direct correlation between the G-quadruplex labels,respectively),isconservedamongestablishedhumanSARS- interactionandthereplicationabilityofSARS-CoVrepliconsandto CoV strains (Urbani, Frankfurt, Tor2, GZ02, and BJ01) and SARS- exclude a possible modulating effect of domains preceding the CoV-relatedbatandcivetisolates(RsSHC014,RS3367,Rs672/2006, PLpro,wereasonedtoleavetheSUDintactbutmutatetheamino- HKU3-1, Rf1, Rm1, Rp3, SZ3, SZ16, Bat273, Bat 279, and BM48) acid residues putatively involved in SUD-binding to G- (SupplementaryFig.3). quadruplexes. The replacement of charged amino-acid residues Despite strong evidence for SUD-M domain - G-quadruplex byalanine(K565A,K568A,andE571A,mut4setofmutations)on interactiondemonstratedinvitro(Tanetal.,2007,2009)aswellas the surface of the SUD-M domain that is oriented towards the invivo(thiswork),wedonotruleoutaroleinreplicationofother SUD-N domain and remote from the PLpro (Tan et al., 2009), possibleSUDfunctions,suchastheSUD-SUD,SUD-UB1,andSUD- completely abrogated the Renilla luciferase expression (Fig. 4) X domain-domain interactions described previously (Neuman et andthesynthesisofviralRNA(notshown),bothinthecontextof al., 2008). Nevertheless, supporting a crucial role for oligo(G)- the SARS-CoV repliconand its X-lacking version. In contrast, and binding in viral genome replication, a similar, but not identical, consistent with the previously described marginal effect of the stretch of charged amino-acid residues was also found to be lysineresiduesatpositions476and477oftheSUD-Ndomainon conservedin thegenomeof the newlyemerging human Middle- G-quadruplex-binding invitro(Tan et al., 2009), the introduction East Respiratory Syndrome coronavirus (MERS-CoV) and in the of themut2setofmutations(K476AandK477A)intotheabove- genomesofthecloselyrelatedbatCoVs,HKU4andHKU5,aswell mentioned replicons had only a minor effect on expression of as in the genome of CoVs isolated from dromedars, which are Renillaluciferase(Fig.4)andviralRNAsynthesis(notshown).In supposedtobeaprimaryanimalreservoirofMERS-CoV(Memish agreementwiththis,therepliconwiththeSUD-Ndomaindeleted etal.,2014)(seeSupplementaryFig.3).Intriguingly,asseenforthe wasabletoreplicate(Fig.3EandF,columns“ΔN”). SUD-Mmacrodomain,wefoundtheputativeMdomainofMERS- These results were consistent with our hypothesis that SARS- CoVtobindexclusivelyoligo(G)(andnotoligo(A),oligo(C),oroligo CoVgenomereplicationrequirestheinteractionofSUDwitholigo (U)) nucleotides (Lei et al., personal communication). It is highly (G)-containing nucleic acids (Tan et al., 2007, 2009). Amino-acid probable that this property is attributable to the stretch of replacements(mut4),whichinin-vitroexperimentsabrogatedthe conserved charged residues on the surface of the M domain of interaction of mutated SUD-NM with oligo(G) (Tan et al., 2009), MERS-CoV (see Supplementary Fig. 3). Further studies will be resultedinareplication-defectiveconstruct.Incontrast,themut2 required to elucidate the impact of the deletions and mutations setofmutationswithnegligibleeffectonoligo(G)-bindingtoSUD- described here at the level of the full-length SARS-CoV genome NMinvitro(Tanetal.,2009)resultedinaviablereplicon. and,inparticular,toanswerthequestionwhetheranysecond-site Taken together, these data suggest SUD amino-acid residues mutation(s)canrescuethemutatedvirus. thatarestrictlyrequiredforSARS-CoVgenomereplication.These residues(lysineresidues565,568,andglutamate571)arelocated α β intheloopconnectingthesecond -helixwiththethird -strand Conclusion oftheSUD-Mmacrodomain;interestingly,thetwolysineresidues belongtoaregionshowntobeinvolvedinG -bindingbyNMR- In this contribution, we demonstrate, for the first time, the 10 shiftperturbationanalysis(Johnsonetal.,2010)andareinvolved functional role of the SUD subdomains within the replication– in interactions with a G-quadruplex according to our docking transcription activity of the Severe Acute Respiratory Syndrome model(Fig.4CandD).Consideringtheirelectrostaticpotential,the Coronavirus (SARS-CoV). Contrary to the dispensable SUD-C and replacementoftheseamino-acidresiduesbyalaninewillaffectthe SUD-N subdomains aswellas totheX domainpreceding SUD in chargedistributionwithinthebindingsiteforG-quadruplexesor thegenome,theSUD-Mmacrodomainwasfoundtobecrucialfor othernucleicacids(SupplementaryFig.2andSupplementarytext the activity. The indispensability of this subdomain might be B).Suchamodificationofthebindingsite'selectrostaticpotential connected with its ability to bind oligo(G) stretches/G-quadru- results in abrogation of the SUD – nucleic-acid interaction (Tan plexesasconcludedfromtheresultsofsite-directedmutagenesis etal.,2009)thatisrequiredforSARS-CoVgenomereplicationina ofchargedamino-acidresiduesintheloopconnectingthesecond α β cell-based assay (Supplementary Fig. 2). On the other hand, the -helix of SUD-M with the third -strand. Intriguingly, a similar, replacementoflysinebyalanineatpositions476and477didnot butnotidentical,clusterofresiduesisobservedinthegenomesof significantlyaffecteithertheelectrostaticpotentialofthebinding the newly emerging human Middle-East Respiratory Syndrome site (Supplementary Fig. 2) or the direct interaction with the coronavirus (MERS-CoV), MERS-related dromedary camel CoV, G-quadruplex(Fig.4C).Althoughwedidnotconstructthereplicon andbatCoVsHKU4andHKU5. containing arginine to alanine replacement at position 562, we speculate that this highly conserved arginine residue is also involved in G-quadruplex binding, since it is located in close Materialsandmethods proximity to the binding cavity (Supplementary Fig. 3 and Fig. 4D; R562, green label) and has also been identified as part Cellsandviruses of the nucleic-acid binding site in the NMR-shift perturbation experiments reported by Johnson et al. (2010). In contrast, the Africangreenmonkeykidneycells(VeroE6)andHuh-T7cells, lysineresidueatposition563seemstobeorientedawayfromthe a derivative of human hepatocellular carcinoma Huh-7 cells binding pocket (Fig. 4D; K563, red label), thereby preventing its (Nakabayashi et al., 1984) that constitutively expresses the T7 participation in oligo(G) interaction. The participation of the RNA polymerase (Shultz et al., 1996), were grown in Dulbecco's strictlyconservedE571,whichislocatedclosetothebindingsite modifiedminimalessentialmedium(DMEM)supplementedwith 320 Y.Kusovetal./Virology484(2015)313–322 2mM glutamine, 100U/ml penicillin, 100mg/ml streptomycin plasmid (ΔSUD), which encoded the Renilla luciferase reporter sulfate, and fetal calf serum (10% v/v). Huh-T7 cells were addi- gene,wasstillreplication-deficientbecauseoftheabsenceofthe tionally supplemented with geneticin (G-418 sulfate, 400mg/ml). MluI-MluI fragment. To restore all replicase components, this Therecombinant,non-cytopathicvacciniavirus(VV)MVA-T7was fragmentwasre-introducedattheMluIrecognitionsiteaccording usedtoproduceSUDoritssubdomainsNþM(SUD-NM)inorder totheprocedureforcloninglongDNAfragments(seeabove).The to complement in trans the activity of the SARS-CoV replicon correctorientationoftheMluI-MluIfragmentwasprovenbyStuI lacking SUD. MVA-T7 was propagated in BHK-21 baby hamster digestion prior to sequencing. The plasmid with reverse orienta- kidney cells and titrated as described previously (Kusov et al., tion of the MluI-MluI fragment was used as negative, non- 2002). Other cell and culture conditions have been described in replicating(NR)control. Almazánetal.(2006)andKusovetal.(2006). A similar cloning strategy was employed to introduce point Δ mutations into the replicon pBAC- X-REP-RLuc lacking the X ConstructionoftheSARS-CoVrepliconcontainingreportergene domainandintothefull-lengthrepliconpBAC-REP-RLuc.Twosets ofmutations–K476AandK477A(mut2)intheSUD-Ndomainand TogenerateaSARS-CoVrepliconcontainingareportergene,we K565A, K568A, and E571A (mut4) in the SUD-M domain – were havetakenadvantageofthestrategypreviouslydescribedforthe introduced into both replicons. Phosphorylated asymmetric for- construction of a SARS-CoV replicon lacking a reporter gene ward and reverse primers overlapping only within a short (Almazán et al., 2006). A Renilla luciferase as a reporter gene sequence (Supplementary Table 1) were employed for site- (RLuc, Renilla reniformis, also known as sea pansy) was PCR- directed mutagenesis as described above. Mut2- and mut4- amplified using pRL-SV40 DNA as a template (Promega; acc. containing clones were identified by BtsI and BstAPI digestion, AF025645), forward and reverse primers (Supplementary respectively.TheORFofallconstructedSARS-CoVrepliconsbear- Table 1), and proof-reading DNA polymerase (AkkuPrime Pfx ingdeletionsormutationswithinNsp3,schematicallydepictedin Δ Δ Δ Δ Δ Δ SuperMix, Invitrogen). The transcription regulatory sequence for Fig.1C ( SUD, X, SUD- N, SUD- M, SUD- NM, and SUD- C Δ theSARS-CoVMprotein(TRSM)andaKozaksequenceenhancing replicons)andFig.4(-SUDmut2-and-SUDmut4-,- X-SUDmut2- expression in eukaryotic cells were included in the forward RLuc and-ΔX-SUDmut4-),wasverifiedbycompletesequencingofSfoI- primer(Almazánetal.,2004;Kozak,1989).ThePCRampliconwas MluIfragments(LGCGenomics).Detailsofthecloningprocedures, treated with AscI and BamHI and introduced between the same restriction analysis of constructed plasmids, their maps and restrictionsitesoftheSARS-CoVrepliconpBAC-REP-URB(Almazán sequencescanbeprovideduponrequest. etal.,2006)producingthereportergene(RLuc)-containingSARS- CoV replicon, referred to as pBAC-REP-RLuc (Supplementary TransfectionofSARS-CoVrepliconsinVeroE6orHuh-T7cells Fig. 1). To exactly conform to the Kozak sequence, the second amino-acidresidueofRluc(tyrosine)wasreplacedbyalanine.This Grown in twelve-well plates to 95% confluence, Vero E6 or replacement was successfully employed in the RLuc expression Huh-T7cells(1(cid:2)105cells/well)weretransfectedwithdeletedor vectorpBS-35S-Rluc-Ala(acc.numberAY189983). mutatedSARS-CoV repliconsbyusingLipofectamin 2000accord- ingtothemanufacturer'sspecifications(Invitrogen).Atindicated IntroductionofdeletionsandpointmutationsintotheRenillagene- time-points (see figure legends), the cells were lysed and the containingSARS-CoVreplicon Renillaluciferaseactivityand/orviralRNAgenomewasmeasured in cell lysates (see below). All experiments were performed in Alldesireddeletionsandpointmutationswereintroducedinto triplicate or quadruplicate and the mean values and standard the pBAC-REP-RLuc plasmid encoding polyproteins pp1a and deviations(SD)arepresented.PlasmidpRL-SV40DNAs(Promega, pp1ab as replicase and RLuc as reporter protein (see above). acc.AF025645)andthewild-typeSARS-CoVrepliconwereapplied However,tosimplifythecloningprocedure,weusedastemplate ascontrolreportersinalltransfectionexperiments. forsite-directedmutagenesistheshorterplasmidpBAC-SfoI-MluI encodingonlytheN-terminalhalfofpolyprotein1a(Nsp1–Nsp3) AssayofRenillaluciferaseactivity of SARS-CoV (Almazán et al., 2006). In brief, a Phusion Hot Start DNA polymerase (Phusion Site-Directed Mutagenesis Kit, Finn- To lyse the cells,150ml/well of Passive lysis buffer (Promega) zymes),which ensureshighfidelityfor the amplification oflarge wasaddedtothewashedcellmonolayerin12-wellplates(1ml/ plasmids, was employed to extend perfectly matched 50-phos- well,phosphate-bufferedsaline,PBS)andincubatedfor20minat phorylatedforwardandreverseprimers(SupplementaryTable1) room temperature (RT) with rigorous shaking. The cells were that border the deleted area as schematically exemplified in further lysed by a freeze ((cid:3)801C) / thaw procedure, vortexed, SupplementaryFig.1BfordeletionofthecompleteSUDsequence. and centrifuged (18,400(cid:2)g, 1min). To a 20-ml aliquot of clear The amplification mixture was treated with DpnI to destroy the supernatant, a mixture of Renilla luciferase assay and enhancer original template DNA and the amplicon was circularized with solutions (50ml each, Biotium) was added and the luminescence Quick T4 DNA ligase (Finnzymes). An aliquot of the ligation wasimmediatelymeasuredusinganAnthosLucy-3luminescence mixture was transformed by electroporation (2.5kV, 200Ω, plate reader (Anthos Labtec Instruments). Data presented in 25mF) into electrocompetent E. coli cells (DH10B, NEB 10-beta, Figs. 2, 3, and 4 are from quadruplicate experiments and are New England Biolabs) or HST02 (Takara) that are suitable for expressed as the mean value7standard deviation (SD). The transformation of long-size plasmids. Positive clones were differences in Renilla luciferase expression of the full-length initially identified by restriction analysis and then confirmed by SARS-CoV replicon and its various mutants were analyzed with sequencing. An agarose gel-purified SfoI-ΔSUD-MluI fragment Sysstat (SigmaPlot Software Inc.) and found to be statistically was transferred into dephosphorylated pBAC-REP-RLuc significant implying that values are greater than would be (Supplementary Fig. 1A) that was restricted with SfoI and MluI. expectedbychance. For this procedure, the DNA ligase 〈long〉 optimized for cloning largeDNAfragmentswasusedasrecommendedbythemanufac- ViralRNAquantification turer(Takara).Theefficiencyoftransformationwastremendously increased after removing components of the ligation buffer by To isolate the viral RNA from Vero E6 cells transfected with sodium acetate/ethanol precipitation. The resulting SUD-lacking SARS-CoV replicons containing deletions or mutations, the cells Y.Kusovetal./Virology484(2015)313–322 321 were trypsinized as usual and spun down (1000(cid:2)g, RT, 5min). virus MVA-T7 (multiplicity of infection, moi, around 5 as pre- ThecellpelletwaswashedwithPBSandcellsweresuspendedin viouslydescribed(Sutteretal.1995;Kusovetal.,2002)).Afterone 100mlPBSbeforedividingintotwoaliquots.A20-mlaliquotofcell hour, the infection mixture was replaced with growth medium suspension was used for the evaluation of Renilla luciferase and incubation was continued for 48h. To measure Renilla activityafter lysis of pelleted cells in 50ml Passive lysis buffer as luciferase activity, the cells were lysed using Passive Lysis Buffer described(seeabove).ThetotalRNAwasisolatedfrom80mlofcell (Promega) as described above. Complementation experiments suspensionusingtheRNeasyextractionkitandDNaseItreatment were run in triplicate and the mean values were employed for asrecommendedbythesupplier(Qiagen).TraceamountsofDNA analysis. To detect viral proteins, aliquots of cell lysates were wereremovedfromRNApreparations(20ml)byadditionaltreat- boiled with 1% sodium dodecyl sulfate (SDS) and proteins were ment (371C, 30min) with 1 unitofRNase-free DNaseIin DNase separated by 12% denaturing polyacrylamide gel (SDS-PAGE) buffercontainingMgCl .TheDNasewasinactivatedbyadding1ml before transfer to PVDF membranes (Immobilon-P, Millipore). 2 of25mMEDTAsolutionandheatingat651Cfor10min.Theyield His-taggedproteins (full-length SUD or SUD-NM) wereimmuno- of total RNA was quantified by using a NanoDrop 1000 UV/Vis logically detected using either polyclonal SARS-CoV anti-Nsp3 spectrophotometer(ThermoFisher).Anequalamount(500ng)of (dilution 1:1000; Rockland, not shown) or monoclonal anti-His 4 RNAsamplesextractedfromcellstransfectedwithdeletedSARS- (1:5000,Qiagen)asprimaryantibody.Alkalinephosphatase(AP)- CoV replicons was directly used as template for the first-strand conjugatedanti-rabbitoranti-mouseIgGwasusedasasecondary cDNAsynthesis.Thereactionmixture(25ml)additionallycontain- antibody(1:10000,Sigma). Δ ingareverse Xprimer(200nM,seeSupplementaryTable1)and allfourstandarddNTPs(800mMeach)waspre-incubatedat651C DockingofaG-quadruplextoSUD-NM for5min,chilledonicetodestroyanysecondarystructureofviral RNAand,afteradditionofRNaseinhibitor(1ml,Ribolock,Thermo Atomic coordinates for a typical G-quadruplex were obtained Scientific)andreversetranscriptase(40units,ThermoScientific), fromtheProteinDataBank(PDB;PDBcode2F8U,Daietal.,2006). further incubated at 451C for 60min. The reverse transcriptase The G-quadruplex was docked into the SUD-NM structure deter- wasinactivatedbyincubationat701Cfor5min.Asimilarmixture minedpreviously(Tanetal.,2009)usingtheprogramAUTODOCK withoutreversetranscriptasewasusedasacontrol.A5-mlaliquot (Morrisetal.,2009). of the mixture was used for Real-Time PCR after addition of forward and reverse primers (0.3mM each, see Supplementary Table 1; note that these primers represent a sequence of Nsp1 Acknowledgments allowingtodeterminegenomereplication,butnottranscription), a fluorescence-quenching primer (6FAM-ACCATCAAGTATGGTGA- CAGCTGCTCT-BBQ, 0.2mM; TIB MolBiol), and a Maxima Probe We thank Doris Mutschall for excellent technical assistance, qPCR Master Mix (Fermentas, MBI) in a total volume of 20ml. A Liliya Grin and Luisa Stroeh for help with the preparation and analysisofseveralconstructs,andDr.JianLeifordiscussion.This calibrationcurvewaspreparedbylog dilutionofpBAC-REP-RLuc 10 workwassupportedbytheGermanCenterforInfectionResearch in nuclease-free water. The measurements were performed in (DZIF), the European Commission (through its projects SILVER triplicate or quadruplicate; mean values of genome copies and (Contract no. 260644) and EMPERIE (Contract no. 223498)), the SD are presented. Data shown are from quadruplicate experi- ments, expressed as the mean value of RNA copies7standard MinistryofScienceandInnovationofSpain(BIO2010-16705),the U.S.NationalInstitutesofHealth(Grantnos.2P01AI060699-06A1 deviation (SD). The statistical analysis of RNA copies for the full- andCRIP-HHSN266200700010C),andtheSino-GermanCenterfor length SARS-CoV replicon and mutants was performed as the Promotion of Research, Beijing, China (GZ 236 (202/9)). The describedabove. fundershadnorolein studydesign, datacollection andanalysis, decisiontopublish,orpreparationofthemanuscript. Generationofplasmidstoexpressfull-lengthSUDandSUD-NMin mammaliancells AppendixA. Supportinginformation To express in Huh-T7 cells the full-length SUD and its more stable derivative SUD-NM comprising the two macrodomains (SUD-N and SUD-M) (Tan et al., 2007), their coding sequence Supplementarydataassociatedwiththisarticlecanbefoundin theonlineversionathttp://dx.doi.org/10.1016/j.virol.2015.06.016. wasplacedundercontroloftheT7promoterinthecontextofthe pIVEXWGvector(Roche).AcompleteSUDsequencewasobtained frompET-Blue2-SUD(Tanetal.,2007)bydigestionwithNcoIand References XhoIrestrictionenzymes.Then,thepurifiedinsertwasclonedinto NcoI-XhoI-treated and dephosphorylated vectors resulting in Almazán,F.,Galan,C.,Enjuanes,L.,2004.Thenucleoproteinisrequiredforefficient plasmids named pIVEX-SUD. A SUD-NM sequence was PCR- coronavirus genome replication. J. Virol. 78, 12683–12688. http://dx.doi.org/ amplified using the proof-reading Pfu DNA polymerase, the 10.1128/JVI.78.22.12683-12688.2004. Almazán,F.,DeDiego,M.L.,Galan,C.,Escors,D.,Alvarez,E.,Ortego,J.,Sola,I.,Zuniga, template pQE30-Xa-SUD-NM, and the forward and reverse pri- S., Alonso, S., Moreno, J.L., Nogales, A., Capisco, C., Enjuanes, L., 2006. mers(SupplementaryTable1).Thepurifiedampliconwastreated ConstructionofaSevereAcuteRespiratorySyndromeCoronavirusinfectious withNcoIandSmaIandinsertedintothedephosphorylatedvector cDNAandareplicontostudycoronavirusRNAsynthesis.J.Virol.80,10900–- pIVEXWGdigestedwiththesameenzymes,resultinginplasmid 10906.http://dx.doi.org/10.1128/JVI.00385-06. Almazán, F., Sola, I., Zuniga, S., Marquez-Jurado, S., Morales, L., Becares, M., pIVEX-SUD-NM. Enjuanes,L.,2014.Coronavirusreversegeneticsystems:infectiousclonesand replicons. Virus Res. 189, 262–270. http://dx.doi.org/10.1016/j. ComplementationintransofSUDandSUD-NMandtheir virusres.2014.05.026. Anand,K.,Ziebuhr,J.,Wadhwani,P.,Mesters,J.,Hilgenfeld,R.,2003.Coronavirus immunologicaldetection mainproteinase(3CLpro)structure:basisfordesignofanti-SARSdrugs.Science 300,1763–1767.http://dx.doi.org/10.1126/science.1085658. Huh-T7 cells (1(cid:2)105 cells/well) cotransfected with 0.25mg Brockway,S.M., Denison, M.R., 2005. Mutagenesis of the murine hepatitis virus SUD-lacking or parental SARS-CoV replicon and 1mg pIVEX-SUD nRsNpA1-scyondtihnegsirse,gainodnvidireanltriefipelsicraetsioidnu.eVsiriomlopgoyrt3an40t,fo2r09p–ro2t2e3in.hpttrpoc:/e/sdsxin.dgo,i.voirrga/l or pIVEX-SUD-NM plasmids were infected with helper vaccinia 10.1016/j.virol.2005.06.035. 322 Y.Kusovetal./Virology484(2015)313–322 Dai,J.,Chen,D.,Jones,R.A.,Hurley,L.H.,Yang,D.,2006.NMRsolutionstructureof M.J.,Kuhn,P.,2008.Proteomicsanalysisunravelsthefunctionalrepertoireof themajorG-quadruplexstructureformedinthehumanBCL2promoterregion. coronavirusnonstructuralprotein3.J.Virol.82,5279–5294.http://dx.doi.org/ NucleicAcidsRes.34,5133–5144.http://dx.doi.org/10.1093/nar/gkl610. 10.1128/JVI.02631-07. Denison,M.R.,Yount,B.,Brockway,S.M.,Graham,R.L.,Sims,A.C.,Lu,X.,Baric,R.S., Neuman,B.W.,Chamberlain,P.,Bowden,F.,Joseph,J.,2014.Atlasofcoronavirus 2004. Cleavage between replicase proteins p28 and p65 of mouse hepatitis replicase structure. Virus Res. 194, 49–66. http://dx.doi.org/10.1016/j. virusisnotrequiredforvirusreplication.J.Virol.78,5957–5965.http://dx.doi. virusres.2013.12.004. org/10.1128/JVI.78.11.5957–5965.2004. Pan,J.,Peng,X.,Gao,Y.,Li, Z.,Lu,X.,Chen,Y.,Ishaq,M.,Liu,D.,DeDiego,M.L., Egloff,M.P.,Malet,H.,Putics,A.,Heinonen,M.,Dutartre,H.,Frangeul,A.,Gruez,A., Enjuanes,L.,Guo,D.,2008.Genome-wideanalysisofprotein–proteininterac- Campanacci,V.,Cambillau,C.,Ziebuhr,J.,Ahola,T.,Canard,B.,2006.Structural tionandinvolvementofviralproteinsinSARS-CoVreplication.PLoSOne3, and functional basis for ADP-ribose and poly(ADP-ribose) binding by viral e3299.http://dx.doi.org/10.1371/journal.pone.0003299. macrodomains.J.Virol.80,8493–8502.http://dx.doi.org/10.1128/JVI.00713-06. Piotrowski,Y.,Hansen,G.,Boomaars-vanderZanden,A.L.,Snijder,E.J.,Gorbalenya, Eriksson,K.K.,Makia,D.,Thiel,V.,2008.Generationofrecombinantcoronaviruses A.E.,Hilgenfeld,R.,2009.CrystalstructuresoftheX-domainsofaGroup-1and using vaccinia virus as the cloning vector and stable cell lines containing aGroup-3coronavirusrevealthatADP-ribose-bindingmaynotbeaconserved coronaviralrepliconRNAs.In:Cavanagh,D.(Ed.),MethodsMol.Biol.,Vol.454; property.ProteinSci.18,6–16.http://dx.doi.org/10.1002/pro.15. 2008,pp.237–254.http://dx.doi.org/10.1007/978-1-59745-181-9_18. Putics,A.,Fillipowicz,W.,Hall,J.,Gorbalenya,A.E.,Ziebuhr,J.,2005.ADP-ribose-1″- Ge,F.,Luo,Y.,Liew,P.X.,Hung,E.,2007.DerivationofanovelSARS-coronavirus monophosphatase: a conserved coronavirus enzyme that is dispensable for repliconcelllineanditsapplicationforanti-SARSdrugscreening.Virology360, viralreplicationintissueculture.J.Virol.79,12721–12731.http://dx.doi.org/ 150–158.http://dx.doi.org/10.1016/j.virol.2006.10.016. 10.1128/JVI.79.20.12721-12731.2005. Gillim-Ross,L.,Taylor,J.,Scholl,D.R.,Ridenour,J.,Masters,P.S.,Wentworth,D.E., Rota, P.A., Oberste, M.S., Monroe, S.S., Nix, W.A., Campagnoli, R., Icenogle, J.P., 2004.DiscoveryofnovelhumanandanimalcellsinfectedbytheSevereAcute Penaranda,S.,Bankamp,B.,Maher,K.,Chen,M.H.,Tong,S.,Tamin,A.,Lowe,L., Respiratory Syndrome Coronavirus by replication-specific multiplex Reverse Frace,M.,DeRisi,J.L.,Chen,Q.,Wang,D.,Erdman,D.D.,Peret,T.S.T.,Burns,C., Transcription-PCR.J.Clin.Microbiol.42,3196–3206.http://dx.doi.org/10.1128/ Ksiazek,T.G.,Pollin,P.E.,Sanchez,A.,Liffick,S.,Holloway,B.,Limor,J.,McCaust- JCM.42.7.3196-3206.2004. land, K.,Olsen-Rasmussen, M.,Fouchier,R.,Günther,S.,Osterhaus,A.D.M.E., Graham,R.L.,Sims,A.C.,Brockway,S.M.,Baric,R.S.,Denison,M.R.,2005.Thensp2 Drosten,C.,Pallansch,M.A.,Anderson,L.J.,Bellini,W.J.,2003.Characterization replicase proteins of murine hepatitis virus and Severe Acute Respiratory of a novel coronavirus associated with Severe Acute Respiratory Syndrome. Syndrome Coronavirus are dispensable for viral replication. J. Virol. 79, Science300,1394–1399.http://dx.doi.org/10.1126/science.1085952. 13399–13411.http://dx.doi.org/10.1128/JVI.79.21.13399-13411.2005. Saikatendu, K.S., Joseph, J.S., Subramanian, V., Clayton, T., Griffith, M., Moy, K., Han,Y.-S.,Chang,G.-G.,Juo,C.-G.,Lee,H.J.,Yeh,S.-H.,Hsu,J.T.-A.,Chen,X.,2005. Velasquez, J., Neuman, B.W., Buchmeier, M.J., Stevens, R.C., Kuhn, P., 2005. Papain-like protease 2 (PLP2) from severe acute respiratory syndrom StructuralbasisofSevereAcuteRespiratorySyndromeCoronavirusADP-ribose- coronavirus(SARS-CoV):expression,purification,characterization,andinhibi- 1″-phosphatedephosphorylationbyaconserveddomainofnsP3.Structure13, tion.Biochemistry44,10349–10359.http://dx.doi.org/10.1021/bi0504761. 1665–1675.http://dx.doi.org/10.1016/j.str.2005.07.02.022. Harcourt,B.H.,Jukneliene,D.,Kanjanahaluethai,A.,Bechill,J.,Severson,K.M.,Smith, Serrano,P.,Johnson,M.A.,Chatterjee,A.,Neuman,B.W.,Joseph,J.S.,Buchmeier,M.J., C.M., Rota, P.A., Baker, S.C., 2004. Identification of Severe Acute Respiratory Kuhn, P., Wüthrich, K., 2009. Nuclear magnetic resonance structure of the SyndromeCoronavirusreplicaseproductsandcharacterizationofpapain-like nucleic acid-bindingdomain ofSevereAcuteRespiratory SyndromeCorona- protease activity. J. Virol. 78, 13600–13612. http://dx.doi.org/10.1128/ virus nonstructural protein 3. J. Virol. 83, 12998–13008. http://dx.doi.org/ JVI.78.24.13600-13612.2004. 10.1128/JVI.01253-09. Hattermann,K.,Müller,M.A.,Nitsche,A.,Wendt,S.,Mantke,O.D.,Niedrig,M.,2005. Shultz,D.E.,Honda,M.,Whetter,L.E.,McKnight,K.L.,Lemon,S.M.,1996.Mutations S15u0sc,e1p0t2ib3i–li1t0y3o1f.dhitftfper:e//ndtxe.duokia.orrygo/t1ic0.c1e0l0l7li/ns0es07to05S-A0R0S4--c0o4r6o1n-a1v.irus.Arch.Virol. withinthe5‘nontranslatedRNAofcellculture-adaptedhepatitisAviruswhich enhancecap-independenttranslationinculturedAfricangreenmonkeycells. Hilgpenrofetelda,seRs.,e2n0a1b4le.FarnotmiviSraAlRdSrutogMdeEsRigSn:.cFryEsBtSalJl.og2r8a1p,h4i0c8s5tu–d40ie9s6o.nhtctpor:/o/ndaxv.diroail. J.Virol.70,1041–1049. Snijder,E.J.,Bredenbeek,P.J.,Dobbe,J.C.,Thiel,V.,Ziebuhr,J.,Poon,L.L.M.,Guan,Y., org/10.1111/febs.12936. Rosanov, M., Spaan, W.J.M., Gorbalenya, A.E., 2003. Unique and conserved Hilgenfeld,R.,Peiris,M.,2013.FromSARStoMERS:10yearsofresearchonhighly pathogenic human coronaviruses. Antiviral Res. 100, 286–295. http://dx.doi. ftehaetucorersonoafvgireunsomgreouapnd2plirnoetaegoem.eJ.oMfoSlA.RBSio-cl.o3ro31n,av9i9r1u–s,10an04e.ahrltytps:p//lditx-.odfofif.roormg/ org/10.1016/j.antiviral.2013.08.015. 10.1016/S0022-2836(03)00865-9. Johnson,M.A.,Chatterjee,A.,Neuman,B.W.,Wüthrich,K.,2010.SARSCoronavirus- Sutter, G., Ohlmann, V., Erfle, V., 1995. Non-replicating vaccinia virus vector unique domain (SUD): three-domain molecular architecture in solution and RNA binding. J. Mol. Biol. 400, 724–742. http://dx.doi.org/10.1016/j. efficientlyexpressesbacteriophageT7RNApolymerase.FEBSLett.371,9–12. Tan,J.,Kusov,Y.,Mutschall,D.,Tech,S.,Nagarajan,K.,Hilgenfeld,R.,Schmidt,C.L., jmb.2010.05.027. 2007.The “SARS-unique” domain (SUD)of SARScoronavirus isanoligo(G)- Kozak,M.,1989.Thescanningmodelfortranslation:anupdate.J.CellBiol.198, 225–239. bindingprotein.Biochem.Biophys.Res.Commun.364,877–882.http://dx.doi. org/10.1016/j.bbrc.2007.10.081. Kusov,Y.Y.,Shatirishwili,G.,Klinger,M.,Gauss-Müller,V.,2002.Avacciniavirus Tan, J., Vonrhein, C., Smart, O.S., Bricogne, G., Bollati, M., Kusov, Y., Hansen, G., MVA-T7-mediated recovery of infectious hepatitis A virus from full-size cDNAs,bothbythemselvesunabletocompletetheviruslifecycle.VirusRes. Mesters,J.R.,Schmidt,C.L.,Hilgenfeld,R.,2009.TheSARS-uniquedomain(SUD) 89,75–88. of SARS coronavirus contains two macrodomains that bind G-quadruplexes. Kusov,Y.,Kanda,T.,Palmenberg,A.,Sgro,J.-Y.,Gauss-Müller,V.,2006.Silencingof PLoSPathog.5,e1000428.http://dx.doi.org/10.1371/journal.ppat.1000428. hepatitisAvirusinfectionbysmallinterferingRNAs.J.Virol.80,5599–5610. Tanaka, T., Kamitani, W., DeDiego, M.L., Enjuanes, L., Matsuura, Y., 2012. Severe http://dx.doi.org/10.1128/JVI.01773-05. AcuteRespiratorySyndromeCoronavirusnsp1facilitatesefficientpropagation Marra,M.A.,Jones,S.J.M.,Astell,C.R.,Holt,R.A.,Brooks-Wilson,A.,Butterfield,Y.S. in cells through a specific translational shutoff of host mRNA. J. Virol. 86, N.,Khattra,J.,Asano,J.K.,Barber,S.A.,Chan,S.Y.,Cloutier,A.,Coughlin,S.M., 11128–11137.http://dx.doi.org/10.1128/jvi.01700-12. Freeman, D., Girn, N., Griffith, O.L., Leach, S.R., Mayo, M., McDonald, H., teVelthuis,A.J.W.,vandenWorm,S.H.E.,Snijder,E.J.,2012.TheSARS-coronavirus Montgomery, S.B., Pandoh, P.K., Petrescu, A.S., Robertson, A.G., Schein, J.E., nsp7þnsp8complexisauniquemultimericRNApolymerasecapableofboth Siddiqui, A., Smailus, D.E., Stott, J.M., Yang, G.S., Plummer, F., Andonov, A., denovoinitiationandprimerextension.NucleicAcidsRes.40,1737–1747.http: Artsob,H.,Bastien,N.,Bernard,K.,Booth,T.F.,Bowness,D.,Czub,M.,Drebot,M., //dx.doi.org/10.1093/nar/gkr893. Fernando,L.,Flick,R.,Garbutt,M.,Gray,M.,Grolla,A.,Jones,S.,Feldmann,H., Wang, J.-M., Wang, L.-F., Shi, Z.-L., 2008. Construction of a non-infectious SARS Meyers,A.,Kabani,A.,Li,Y.,Normand,S.,Stroher,U.,Tipples,G.A.,Tyler,S., coronavirus replicon for application in drug screening and analysis of viral Vogrig, R., Ward, D., Watson, B., Brunham, R.C., Krajden, M., Petric, M., proteinfunction.Biochem.Biophys.Res.Commun.374,138–142.http://dx.doi. Skowronski,D.M.,Upton,C.,Roper,L.R.,2003.Thegenomesequenceof the org/10.1016/j.bbrc.2008.06.129. SARS-associated coronavirus. Science 300, 1399–1404. http://dx.doi.org/ Xiao, Y., Ma, Q., Restle, T., Shang, W., Svergun, D.I., Ponnusamy, R., Sczakiel, G., 10.1126/science.1085953. Hilgenfeld,R.,2012.Non-structuralproteins7and8ofFelineCoronavirusform Memish,Z.A.,Assiri,A.,AlHakeem,R.,Yezli,S.,Almasri,M.,Zumla,A.,Al-Tawfiq,J.A., a2:1heterotrimerthatexhibitsprimer-independentRNApolymeraseactivity. Drosten,C.A.,Albarrak,A.,Petersen,E.2014.MiddleEastRespiratorySyndrome J.Virol.86,4444–4454.http://dx.doi.org/10.1128/JVI.06635-11. Coronavirus,MERS-CoV. Conclusionsfromthe2ndScientificAdvisoryBoard Yang,H.,Xie,H.,Xue,X.,Yang,K.,Ma,J.,Liang,W.,Zhao,Q.,Zhou,Z.,Pei,D.,Ziebuhr,J., Meeting of the WHO Collaborating Center for Mass Gathering Medicine, Hilgenfeld,R.,Yuen,K.Y.,Wong,L.,Gao,G.,Chen,S.,Chen,Z.,Ma,D.,Bartlam,M., Riyadh,Int.J.Infect.Dis.24:51–53.10.1016/j.ijid.2014.05.001. Rao, Z., 2005. Design of wide-spectrum inhibitors targeting coronavirus main Morris,G.M.,Huey,R.,Lindstrom,W.,Sanner,M.F.,Belew,R.K.,Goodsell,D.S.,Olson, proteases.PLoSBiol.3,1742–1752.http://dx.doi.org/10.1371/journal.pbio.0030324. A.J.,AutoDock4andAutoDockTools4:automateddockingwithselectiverecep- Ziebuhr,J.,Thiel,V.,Gorbalenya,A.E.,2001.Theautocatalyticreleaseofaputative torflexibility,2009.J.Comput.Chem.30,2785–2791.http://dx.doi.org/10.1002/ RNA virus transcription factor from its polyprotein precursor involves two jcc.21256. paralogous papain-like proteasesthatcleavethe same peptidebond. J.Biol. Nakabayashi, H., Taketa, K., Yamane, T., Miyazaki, M., Miyano, K., Sato, J., 1984. Chem.276,33220–33232.http://dx.doi.org/10.1074/jbc.M104097200. Phenotypical stability of a human hepatoma cell line, Huh-7, in long-term Ziebuhr,J.,Schelle,B.,Karl,N.,Minskaia,E.,Bayer,S.,Siddell,S.G.,Gorbalenya,A.E., culturewithchemicallydefinedmedium.Gann75,151–158. Thiel, V., 2007. Human Coronavirus 229E papain-like proteases have over- Neuman,B.W.,Joseph,J.S.,Saikatendu,K.S.,Serrano,P.,Chatterjee,A.,Johnson,M.A., lapping specificities but distinct functions in viral replication. J. Virol. 81, Liao,L.,Joseph,P.,Klaus,J.P.,YatesIII,J.R.,Wuthrich,K.,Stevens,R.C.,Buchmeier, 3922–3932.http://dx.doi.org/10.1128/JVI.02091-06.