JOURNALOFVIROLOGY,May1999,p.4156–4170 Vol.73,No.5 0022-538X/99/$04.0010 Copyright©1999,AmericanSocietyforMicrobiology.AllRightsReserved. High-Level Variability in the ORF-K1 Membrane Protein Gene at the Left End of the Kaposi’s Sarcoma-Associated Herpesvirus Genome Defines Four Major Virus Subtypes and Multiple Variants or Clades in Different Human Populations JIAN-CHAOZONG,1DOLORESM.CIUFO,1DONALDJ.ALCENDOR,2XIAOYUWAN,1 JOHNNICHOLAS,1PHILIPJ.BROWNING,3PETERL.RADY,4STEPHENK.TYRING,5 JANM.ORENSTEIN,6CHARLESS.RABKIN,7IH-JENSU,8KEVINF.POWELL,9 MARGARETCROXSON,9KIMBERLYE.FOREMAN,10BRIANJ.NICKOLOFF,10 SERHANALKAN,10ANDGARYS.HAYWARD1,2* D DepartmentofOncology,TheJohnsHopkinsSchoolofMedicine,Baltimore,Maryland212311;Departmentof o PharmacologyandMolecularSciences,TheJohnsHopkinsSchoolofMedicine,Baltimore,Maryland212052; w VanderbiltUniversity,Nashville,Tennessee37232-68383;DepartmentofPediatrics4andDepartmentof n Microbiology,5UniversityofTexasMedicalBranch,Galveston,Texas77555;DepartmentofPathology, lo a GeorgeWashingtonUniversityMedicalCenter,Washington,D.C.200376;ViralEpidemiologyBranch, d NationalCancerInstitute,Rockville,Maryland208927;DepartmentofPathology,NationalCheng e d KungUniversityHospital,Tainan704,Taiwan,RepublicofChina8;VirusDiagnosticInfectious f DiseaseLaboratory,AucklandHospital,Auckland,NewZealand9;andOncologyInstitute, ro LoyolaUniversityMedicalCenter,Maywood,Illinois60153-538510 m h Received18December1998/Accepted12February1999 t t p : Infection with Kaposi’s sarcoma (KS)-associated herpesvirus (KSHV) or human herpesvirus 8 (HHV8) is // commonincertainpartsofAfrica,theMiddleEast,andtheMediterranean,butisrareelsewhere,exceptin jv i. AIDSpatients.Nevertheless,HHV8DNAisfoundconsistentlyinnearlyallclassical,endemic,transplantand a s AIDS-associated KS lesions as well as in some rare AIDS-associated lymphomas. The concept that HHV8 m genomesfallintoseveraldistinctsubgroupshasbeenconfirmedandrefinedbyPCRDNAsequenceanalysis . o oftheORF-K1geneencodingahighlyvariableglycoproteinrelatedtotheimmunoglobulinreceptorfamilythat r g mapsattheextremeleft-handendoftheHHV-8genome.Amongmorethan60differenttumorsamplesfrom / theUnitedStates,centralAfrica,SaudiArabia,Taiwan,andNewZealand,aminoacidsubstitutionswerefound o n atatotalof62%ofthe289aminoacidpositions.Thesevariationsdefinedfourmajorsubtypesand13distinct J variantsorcladessimilartothosefoundfortheHIVENVprotein.TheBandDsubtypeORF-K1proteinsdiffer a fromtheAandCsubtypesby30and24%,respectively,whereasAandCdifferfromeachotherby15%.Inall n u cases tested, multiple samples from the same patient were identical. Examples of the B subtype were found a almostexclusivelyinKSpatientsfromAfricaorofAfricanheritage,whereastherareDsubtypeswerefound r y onlyinKSpatientsofPacificIslandheritage.Incontrast,CsubtypeswerefoundpredominantlyinclassicKS 3 andiniatrogenicandAIDSKSintheMiddleEastandAsia,whereasU.S.AIDSKSsampleswereprimarily , 2 A1,A4,andC3variants.Weconcludethatthisunusuallyhighdiversity,inwhich85%ofthenucleotidechanges 0 leadtoaminoacidchanges,reflectssomeunknownpowerfulbiologicalselectionprocessthathasbeenacting 1 9 preferentiallyonthisearlylyticcyclemembranesignallingprotein.TwodistinctlevelsofORF-K1variability b are recognizable. Subtype-specific variability indicative of long-term evolutionary divergence is both spread y throughouttheproteinaswellasconcentratedwithintwo40-amino-acidextracellulardomainvariableregions g (VR1 and VR2), whereas intratypic variability localizes predominantly within a single 25-amino-acid hyper- u e variableCysbridgeloopandapparentlyrepresentsmuchmorerecentchangesthathaveoccurredevenwithin s specificclades.Incontrast,numerousextracellulardomainglycosylationsitesandCysbridgeresiduesaswell t astheITAMmotifinthecytoplasmicdomainarefullyconserved.Overall,wesuggestthatratherthanbeing a newly acquired human pathogen, HHV8 is an ancient human virus that is preferentially transmitted in a familial fashion and is difficult to transmit horizontally in the absence of immunosuppression. The division intothefourmajorHHV8subgroupsisprobablytheresultofisolationandfoundereffectsassociatedwiththe historyofmigrationofmodernhumanpopulationsoutofAfricaoverthepast35,000to60,000years. The recently discovered Kaposi’s sarcoma (KS)-associated and AIDS-associated forms of KS (5, 15) as well as being herpesvirus (KSHV), or human herpesvirus 8 (HHV8), is involvedinAIDS-associatedprimaryeffusionlymphoma(PEL thought to be essential for the development of both classical orbodycavity-basedlymphoma[BCBL])(12)andmulticentric Castleman’s Disease (62). HHV8 DNA is present in virtually allKStumorsamplesandintheperipheralbloodmononuclear cellsinupto50%ofhomosexualAIDSpatientswithKS(16, *Correspondingauthor.Mailingaddress:DepartmentofPharma- cology&MolecularSciences,JohnsHopkinsSchoolofMedicine,725 19,22,46,67).SerologicalevidenceobtainedbyLANAimmu- N. Wolfe St., WBSB 317, Baltimore, MD 21205. Phone: (410) 955- nofluorescent antibody assay indicates that infection is also 8684.Fax:(410)955-8685.E-mail:[email protected]. widespreadinthosepartsofsouthernItaly(5to20%seropos- 4156 VOL.73,1999 HHV8 ORF-K1 SUBTYPE VARIATION AND EVOLUTION 4157 itivity) and central and southern Africa (40 to 60% seroposi- ORF-K1codingregionsofmorethan60differentHHV8DNA tivity) where endemic and classical KS have the highest inci- samplesfromawidevarietyofclassicandAIDS-associatedKS dence rates, reaching up to 1.0 and 10 per 100,000 person lesions and PEL tumors or cell lines. The results revealed years,respectively,withahighpreferenceformales(4,25,26, considerableaminoacidpolymorphism,particularlyintheex- 34, 68). Similarly, KS patients and male homosexual AIDS tracellular domain, giving ORF-K1 protein patterns that fall patients, but not human immunodeficiency virus (HIV)-posi- into four major subtypes, together with further subdivisions tive intravenous drug users and hemophiliacs, have extremely into13distinctivevariantsorclades.Moreover,individualclus- high seropositivity rates of 85 and 50%, respectively (25, 26). ters of related proteins closely correlate with the different However, the seroprevalence in blood donors in the United geographicalandethnicbackgroundsofthepatients. KingdomandUnitedStatesmaybenogreaterthan1%(68), whichcorrelateswithestimatedclassicKSincidenceratesofas MATERIALSANDMETHODS lowas0.014and0.165per100,000personyearsintheUnited KStumorDNAsamples.FiveKSDNAsamplesfromEastCoastU.S.AIDS KingdomandUnitedStates,respectively(4,6). patients were described previously (49). C282 and ASM70-80 were autopsy HHV8 is a gamma-2 class herpesvirus that is distantly re- samplesfrompatientswithadvanceddisseminatedKSfromNewYork’sSloan- lated to herpesvirus saimiri (HVS) and Epstein-Barr virus Kettering/Memorial Hospital in 1984, whereas AKS1, AKS2 and AKS4 were (EBV).Severalreportshavedescribedthreenovel10-to13-kb biopsysamplesobtainedfromAIDSpatientsinBaltimore,Md.,andNewYork D segments of the HHV8 genome that encode divergent viral in 1995. The samples referred to as BKS8, BKS10, BKS11, BKS13, BKS14, o BKS15,andBKS16wereallKSbiopsiescollectedfromunrelatedAIDSpatients w homologues of exogenously acquired cellular genes encoding inGalveston,Tex.EKS1,EKS2,andBKS12werebiopsysamplesfromclassical n interleukin-6(IL6),dihydrofolatereductase(DHFR),MIP-IB, KS patients in Baltimore and Galveston, respectively. BKS3 (CVU-14) and lo TS, MIP-IA, and BCL-2; several IRF-like genes; and FLIP, BKS4(CVU-1)weredirectbiopsyandculturedcellsupernatantsamplesob- a tainedondifferentoccasionsfromdifferentlesionsfromaclassicalKSpatientin d CYC-D,OX-2,andGCR,mostofwhichhavenotbeenfound Nashville,Tenn.431KAPand431NSCrepresentKSandadjacentnormalskin e previouslyinotherHHVs(14,45,47–50,59).Thenearlycom- biopsy DNA samples obtained from a male non-AIDS-associated KS patient d plete primary nucleotide sequences of the 190-kb double- from Zaire in 1984, whereas ST1 and ST2 represent KS biopsies from HIV- fr strandedDNAmoleculesoftwoHHV8genomes,onederived positivefemalesinUganda(49,70).OKS3andOKS4werederivedfrompar- o affin-blocksectionsofKSbiopsiesfrommaleAIDSpatientsinTanzania(kindly m from an AIDS BCBL cell line (59), and the other from a KS providedbyAndrewBlauvelt,DermatologyBranch,NationalCancerInstitute, h lesion(48),havebeendeterminedandwerefoundtodifferby Bethesda,Md.).OKS7,OKS8,andOKS9werefromparaffinblocksectionsof t only0.4%fromeachother. KSbiopsiesfromAIDSpatientsofmaleblackHaitian,femaleHispanicMexi- tp Because of the many questions that arise about the origin, can,andfemaleHispanicNicaraguanorigin,respectively,seenattheUniversity :/ distribution, transmissibility, and disease associations of HHV8, ofMiamiMedicalCenter,Miami,Fla.,andprovidedbytheAIDSMalignancy /jv we have been interested in examining the levels of genetic Bfraonmk.ARIKDSS1p,aRtiKenSt2s,fRroKmS3L,uRsKakSa4,,ZanadmRbiKa.ST5hweeTreKaSllsfarmozpelnesKwSebreioaplslycsoallmecptleeds i.a variationandpolymorphisminthisvirus.Ininitialstudies,we attheNationalChengKungUniversityHospital,Taiwan,RepublicofChina s m carried out PCR sequencing on three small segments of the (64). TKS1, TKS3, and TKS9 came from HIV-positive AIDS-associated KS openreadingframe26(ORF26)andORF75genesfrom12KS patients,whereasTKS2,TKS5,TKS6,TKS7,andTKS10wereallfromclassic .o HIV-negative non-AIDS associated KS patients, and TKS11 represented an r specimens from several different patients and from multiple iatrogenic renal transplant KS patient. TKS10 differed from all of the other g / distinctKSlesionsfromasinglepatient(70).Allsevenlesions Taiwan samples by being from a Hwalien patient of indigenous ethnic back- o testedfromanAIDSpatientwithaggressivedisseminatedKS groundandprobableSouthPacificancestry.SKS1,SKS2,SKS3,andSKS6to n proved to be identical to one another, but the equivalent se- SKS9wereiatrogenicKSsamplesfromrenaltransplantpatientsfromtheKing J FaisalHospital,Riyadh,SaudiArabia(23).WKS1camefromanAIDSbiopsy a quencesfromdifferentpatientsvariedsignificantlyandfellinto specimen from the University of North Carolina. Samples of KS specimens n three distinct DNA patterns, referred to as A, B, and C sub- collectedattheUniversityofAucklandHospital,Auckland,NewZealand,in- u a types. Overall nucleotide variation among the three groups cludedZKS1andZKS5fromHIV-positivemaleCaucasians,ZKS7fromarenal r transplantrecipient,ZKS3andZKS4fromtwoelderlyPolynesianmaleswith y reached between 1.0 and 1.5%, but less than 0.1% occurred classic KS, and ZKS6 from an HIV-positive immigrant of Bushman ancestry 3 withineachgroup.Incomparison,theaveragenucleotidedif- fromsouthernAfrica.JKS15andJKS20representarchivalKSparaffinblock , ferencelevelsbetweenunrelatedhumanbeingsis0.3to0.4%. samplesfromanAfrican-AmericanAIDSpatientandanHIV-negativeAfrican 2 0 Compilation of existing data from the literature at that time immigrantwithendemicKSthatwereselectedastheonlytwoBsubtypege- 1 over a very small 233-bp region of ORF26 (9, 17, 31, 46) nomes detected within a collection of 15 specimens from the Johns Hopkins 9 Department of Dermatology and AIDS Malignancy Clinic (11). Tissue DNA implied that among U.S. AIDS-associated HHV8 samples, a fromparaffinblocksections(OKSandJKSseries)orOCTfrozensections(RKS by largemajoritywereverycloselyrelatedoridenticalAsubtype series)wereextractedbytreatmentwithxylene,proteinaseK,andTween20or g genomes,whereasaminorsubpopulationwereCsubtypevari- withproteinaseK,sodiumdodecylsulfate,andphenol-chloroform-isoamylal- u ants,andseveralsamplesfromAfricaappearedtorepresenta coholrespectively,followedbyethanolprecipitationandresuspensioninasmall e volumeofdistilledwater. s distinctiveBsubtype. PELcelllinesandBCBLtumorDNAsamples.AtotalofsixdifferentPELcell t Inthepresentstudy,wesoughttotakethismolecularanal- linesweregrowninRPMImediumplus20%fetalcalfserum.Total-cellDNA ysis further by examining a much larger group of samples wasextractedfromwashedpelletedcellsbydetergentlysis,proteinaseKand RNasetreatment,phenolextraction,anddialysis.HBL6cells(24)wereobtained withinahighlyvariableregionoftheHHV8genome.Weand asagiftfromPatrickMoore(ColumbiaUniversity,NewYork,N.Y.).BCBL1 othershavepreviouslynotedunexpectedlyhighlevelsofdiver- cellswerereceivedfromtheAIDSReagentRepository(30,56),andtheinitial sityinthegeneencodingtheORF-K1proteinattheextreme BC2(13),BC3(2),andBCP1(8)cellcultureswereobtainedfromtheAmerican left-handside(LHS)ofthegenome(35,38,51).TheORF-K1 TypeCultureCollection.TheJSC1celllinewasestablishedfrompleuraleffusion cellsofanAIDS-associatedBCBLpatientattheJohnsHopkinsHospitalLym- protein is a highly glycosylated cytoplasmic and membrane phomaClinicandwasagiftfromJenniferS.CannonandRichardAmbinder protein similar to the immunoglobulin receptor family that is (10).Inaddition,DNAwasextracteddirectlyfromthreeotherPELtumorcell expressedasaninducibleearly-lytic-cyclegeneproductinPEL samples,namely,BCBL-R,obtainedfromtumorcellsofanAIDSpatientin cell lines (35, 38, 52). It is not related to any other known Washington,D.C.;BKS1(CVU-30),obtainedfromearly-passageadherentcells fromapleuraleffusionsamplefromanAIDSKSpatientinNashville,Tenn.;and herpesvirusprotein,buthastheinterestingpropertiesofpro- BKS6 (CVU-27) and BKS5 (CVU-19), representing filtered supernatant and ducingfociinDNAtransfectedprimarycellsandsubstituting unculturedpleuraleffusioncells,respectively,obtainedondifferentoccasions forthesaimiritransformationprotein(STP)ofHVSinT-cell from an unusual AIDS-associated T-cell BCBL in Nashville. The latter two transformation andtumorigenesisassays,anditalsocontainsa samplesarecollectivelyreferredtohereasBCBL-B. HHV8genomicDNAlibrariesandORF-K1phageandplasmidsubclones. functional immunoglobulin-receptor tyrosine-based activation PhagelambdaclonesderivedfromtwoBCBL-RtumorDNAgenomiclibraries motif (ITAM) (36, 38). Here we have analyzed the complete inthelEMBL3andlDASHIIbackgroundsweredescribedpreviously(28,51, 4158 ZONG ET AL. J.VIROL. 70).Oneclone(lD-S1),containinga16-kbinsertfromtheLHSgenomicter- minus equivalent to HHV8 (BC1) positions 21.8 to 14.5 kb, was isolated by hybridizationwithaPCRDNAproberepresentingpartoftheORF6(SSB)gene fromtheextremeleft-handterminusoflD3-80(49).A3.5-kbBamHI-BamHI plasmid genomic subclone (pDJA61) encompassing 1,600 bp of the proximal LHSterminaltandemrepeat(TTR)sequences,plusallofORF-K1andpartof ORF4fromlD-S1,wasisolatedandsequencedbyprimerwalkingprocedures. PrimersbasedonthissequencewerethenusedtoobtainthecompleteDNA sequencefromgenomicpositions20to1085coveringthewholeoftheORF-K1 codingregionfor71KSandPELorBCBLsamples. PCRamplificationandsequencingprimers.The870-bpORF-K1codingre- gionfromnucleotidepositions105to974wasusuallyamplifieddirectlyfrom HHV8-positiveDNAsamplesasa1,066-bpPCRproduct.BRLTaqpolymerase (GIBCOBRLcatalogno.18038-042)wasincubatedwith20to100ngofDNA templateinaTechnePHC-3thermocyclersetat94°Cfor1min,50°Cfor1min, and72°Cfor2minover35cycleswiththefollowingoutsideprimerpair(genomic nucleotidepositionsindicatedinparentheses):LGH2089(59-GTTCTGCCAG GCATAGTC-39[21to38])andLGH2088(59-AATAAGTATCCGACCTCAT-39 [1085to1067]).OthercommonlyusedadditionalinternalORF-K1primersfor eithernestedordirectPCRamplificationandPCRsequencingofmostorall D subtypes were as follows: LGH2090 (59-GAGTGATTTCAACGCCTTAC-39 o [193 to 212]), LGH2091 (59-GAGTATTGTTGCAATACC-39 [270 to 253]), w LGH2505(59-CAACCTGTCTTACAAACC-39[401to418]),LGH2500(59-GT n AACATGCTGACCACAAG-39[445to427]),LGH2507(59-CGTCTCGCCTG lo TCAAATC-39[589to606]),LGH2508(59-AGATACCACACATGGTT-39[840 a to864]),andLGH2092(59-GACACTCGTAGCTCTGAT-39[802to820]).Di- de deoxynucleotide double-stranded cycle sequencing (GIBCO BRL catalog no. d 18196)withsingle32P-labelledprimerswasthencarriedoutwithisolatedagarose f gel-purifiedDNAbands.Thereactionproductswerefractionatedon7Murea– ro 6%polyacrylamidegels,andtheautoradiographswerereadmanually.Allse- m quencingwascarriedoutwithbothcomplementarystrands,andmostanalyses involvedconfirmationwithredundantoverlappingfragmentsfrommultiplein- h dependent PCR-amplified products. Several representative samples were also tt subjectedtoPCRsequencingoverthe39endoftheORF4genebyusingprimers p: LGH2094(59-GGTACTGACATTCAGG-39[966to982])andLGH2095(59- // TCACAAAAAGGAGGAGAAG-39[1540to1522]). jv Phylogeneticanalysis.Deducedaminoacidsequenceswerealignedmanually i. a withtheprogramVisEd(version1.2).Thefullalignmentcontained63ORF-K1 s sequences.Foreachvariant,therewere289characters(aminoacidsplusgaps). m Regionsofambiguousalignment(gappositions)wereexcludedfromthephylo- . geneticanalysis,and,consequently,278ofthe289aminoacidswerecompared. o r AllphylogeneticreconstructionswerecarriedoutbyusingthePHYLIPpackage, g version3.5C.Oncethesequenceswerealigned,theaminoaciddivergencewas / calculated by using PRODIST, with the Dayhoff PAM matrix option. The o n branchingpatternwasestimatedfromthedistancematrixbyneighborjoining withtheprogramNEIGHBOR.TheSKS1variantwasusedasanoutgroup,but J a a similar topography was obtained by using other variants as the outgroup. n Confidence levels for the branching pattern were estimated by a bootstrap u resamplingofthedata.Bootstrapvalueswereobtainedfromaconsensustree a basedon1,000randomlygenerateddatasetsbyusingSEQBOOT,PRODIST, ry CONDENSE,andNEIGHBOR. 3 proNtuoctylepoetiAde1(sBeqCuBeLnc-Re)a,cAce4s(sBioCnBnLu-Bm)b,eBr.(4T3h1eKADPN)A,Cs1eq(AueSnMce72d),atCa3f(oBrCt2h)e, ingFtIhGe.O1.ROF-rKga1ngizeanteio.n(Ao)ftMheaplefotf-htahnedgeenndomoficthloecHatHioVn8ogfeHnoHmVe8e(nBcComBpLa-Rss)- , 2 aGnednBDa1nk(T(KacSc1e0s)s,ioanndnoD.2A(FZ1K33S033)8sutobtyApFe1O33R04F4-K).1OgtehneerspaurbeliashvaeidlabOleRFfr-oKm1 lambda clone lD-S1 relative to other LHS clones and major features of the 01 sequencesinthisanalysisincludethoseforHHV8samplesBC1(U75678),KS-F HHV8DNAmolecule.DLandDR,duplicated1-kbORI-likeregions(leftand 9 right,respectively)atgenomiccoordinates23kband119kb.(B)Organization (U93872),andBCBL-1(U86667). b andorientationofORFsandterminalrepeatunitsequenceswithinanexpanded y area across the LHS TTR-unique region boundary encompassed within the g RESULTS n3.a5t-eksb2B1a6m0H0It-oBa1m1H88I5f)r.agNmueclnetoitnidpelapsomsiidtiosnusbcolofntehepDinJiAtia6t1or(gaenndomteicrmcoinoartdoir- ue codonsintheuniqueregionaregivenabovetheORFs(openarrows),andthose s UnusuallyhighlevelsofvariabilitybetweentheA,B,andC oftheTTRsequencesaregivenabovethedifferentrepeatunitfragments(solid t subtypes of the ORF-K1 protein. Analysis of the DNA se- arrows).(C)Predicteddomainstructureandkeyfeaturesofthehighlyvariable quencesencompassingtheintactORF-K1genecodingregion 289-amino-acid ORF-K1 membrane receptor-like protein encoded between genomicnucleotidecoordinates105and974.Hatchedbarsdenotesignalpeptide attheextremeLHSoftheuniquesegmentofHHV8fromeach andtransmembrane(TM)domainswithaminoacidboundariesindicated.Pre- of the three prototype strains of the proposed A, B, and C dictedN-glycosylationsites(NXS/NXT)(solidtriangles)andthe12conserved subtypesrevealedextraordinarilyhighlevelsofnucleotidevari- Cys residues (solid circles) are indicated. Cytopl, cytoplasmic. (Lower panel) ation (Fig. 1A and B). For example, a 1,066-bp fragment of LocationsofhighlyvariableVR1andVR2domainsandtheproposedhyper- variableAsubtypeVR*domainandsummarylistingofmajorsubtypeamino cloned viral genomic DNA from BCBL-R (A) showed 8 and aciddifferencevaluesbothwithinandoutsideoftheVRblocks. 16%nucleotidevariations(notcountingdeletions)relativeto the equivalent region of viral DNA from ASM72 (C) and 431KAP(B),respectively.Thiscomparestoonly0.4and1.4% nucleotide variations within an immediately adjacent 474-bp subtype. However, even within the A subgroup, the ORF-K1 PCRproductfromtheORF4generegion(notshown).Based genesfromBCBL-RandBC1display16nucleotidedifferences on these data, the published PEL cell line ORF-K1 DNA (1.5% variation), compared to only 0.2% variation in the sequencesfromtheBC1andBCBL-1celllines(35,59)aswell ORF4 gene block. Lagunoff and Ganem (35) have also re- astheKSgenomeofLeeetal.(38)allrepresentAsubtypes, ported that the ORF-K1 coding region of BCBL1 shows 2% whereas the KS genome of Neipel et al. (48) represents a C nucleotidedifferencesfromtheequivalentregionofBC1. VOL.73,1999 HHV8 ORF-K1 SUBTYPE VARIATION AND EVOLUTION 4159 The HHV8 ORF-K1 gene is predicted to encode a 289- tionships among the different subtypes and variants fully amino-acid polypeptide that has features of a membrane- supporttheseinterpretations(Fig.4). bound and largely extracellular immunoglobulin receptor-like Lack of variability among multiple samples from the same protein(Fig.1C).Surprisingly,unlikethesituationinthecon- patient.Consideringthehighdegreeofvariabilityencountered servedORF26andORF75loci,alargemajority(almost85%) among the ORF-K1 genes, it was important to establish of the nucleotide variations observed in ORF-K1 also proved whetheranydiversityoccurswithinasingleinfectedhost.We toleadtoaminoacidsubstitutions.Forexample,only17ofthe were able to address this question for six patients, but in all total of 135 nucleotides that differ between BCBL-R (A) and cases obtained identical sequence data for multiple samples 431KAP (B) within the 870-bp ORF-K1 coding region repre- derived at different times or from different lesions from the sent silent (synonymous) mutations. Overall, the prototype A same infected individual. First, the ORF-K1 gene from the andCversionsdifferbyatotalof39aminoacids(15%),and HBL6 cell line proved to have a sequence identical to that the A and B versions vary by 85 amino acids (29%). The presented for BC1 by Russo et al. (59), which represents a differences between the A and C versions are concentrated separatecelllineestablishedatadifferenttimefromthesame primarilywithintwo40-amino-acidblocks,referredtoasVR1 PEL patient (13, 24). Second, three separate KS lesions ob- andVR2,extendingfromcodons54to92and199to227,with tained at autopsy for our prototype C strain—one from skin the latter including two small in-frame deletions. These two (ASM72)andtheothertwofromlungorlymphnodelesions D highly variable blocks extend even further in the B subtype, (ASM70 and ASM75)—were identical. Third, for our proto- o w encompassingcodons1to92surroundingVR1(includingthe typeBstrainfromaUgandanendemicKSpatient(1a),botha n N-terminalsignalpeptideatpositions1to22),andcodons191 KSskinlesionbiopsysample(431KAP)andanadjacentcon- lo to228surroundingVR2,althoughtherearenodeletionsrel- trolnormalskinbiopsysample(431NSC)provedtohaveiden- a d ative to the A pattern. The putative transmembrane domain tical ORF-K1 sequences. Fourth, both a direct KS lesion bi- e lyingbetweenpositions229and261(Fig.1C)islargelyinvari- opsy(BKS3)andaseparatecellculture-grownsample(BKS4) d ant among all three subtypes, but in contrast, although the A fromanotherlesion,whichwasstillHHV8positiveatsecond fr o andCsubtypesshowfewornodifferencesinthecentralregion passage, proved to be identical. Fifth, two distinct KS biopsy m (positions 105 to 148) or in the C-terminal cytoplasmic tail specimens from different skin lesions biopsied at different h (positions262to289),theBsubtypedisplaysnearly30%(12of timesfromthesamepatientwithclassicKSatJohnsHopkins t t 38)aminoaciddifferenceswithinthecytoplasmictailregion. Hospital (EKS1 and EKS2) were identical. Sixth, two pleural p Clade analysis of multiple HHV8 genomes based on effusionsamples(BKS5andBKS6)obtainedatdifferenttimes :/ / ORF-K1proteinpatterns.ToinvestigatewhethertheORF-K1 (includingoneundergoingcellculturepassaging)provedtobe jv protein patterns found within a much larger set of KS, PEL, identicaltoanoriginalT-cell-likePELtumorsamplefromthe i.a andBCBLsamplesalsoallfellwithinthesamethreegroups, samepatient,referredtoasBCBL-B.Inaddition,manyindi- s m or instead either represented a continuum or formed addi- vidual samples were independently amplified and sequenced . tional groups, we PCR amplified and sequenced the entire onbothDNAstrandsacrossthesamesegmentoftheORF-K1 o r 1,066-bp ORF-K1 gene coding region encompassing genomic gene on multiple occasions with different overlapping primer g / nucleotide positions 20 to 1085 from a total of 71 distinct pairs without detecting any confirmed discrepancies. There- o HHV8-positivesamplesrepresenting63differentpatients.The fore, although admittedly the procedures used would detect n samples included 11 of the 12 analyzed previously in the only the most abundant molecular type if mixtures existed, J a ORF26andORF75constantregion(70),plusadditionalsam- thereisnoevidenceasyetthatanypatientswereinfectedwith n ples representing several BCBL cell lines (BC2, BC3, BCP1, multipleHHV8isolatesorthatanyofthevariabilityisgener- u a BCBL1,andJSC1),anunusualT-celllikePELtumor(BCBL- atedwithinasingleinfectedindividual. r y B),andKSsamplesderivedprimarilyfromtheEastCoastand Distinctive variants and clades within subtype A ORF-K1 3 southern United States, central and southern Africa, Saudi proteins.The22ORF-K1proteinsequenceswithAsubgroup , Arabia,Taiwan,andNewZealand. patternsfellintofivedistinctsubgroups,referredtoasA1,A2, 2 0 Acomparisonofthecompleteornearlycomplete289-ami- A3,A4,andA5variants,withthemajorbranchcontainingthe 1 no-acidORF-K1proteinsequencesfrom63differentpatients A1,A4,andA2variantsseparatingfromtheA3andA5branch 9 ispresentedinFig.2.WiththeexceptionofTKS10,ZKS3,and with a bootstrap confidence level of 83% (Fig. 4A). Twelve b y ZKS4,whichformedanewDsubtype,allsampleswerereadily genomes,includingBCP1andBCBL-RfromPELtumors,plus g differentiatedintothreeverydistinctivegroupscorresponding 4 KS genomes from New York and Maryland (C282, AKS1, u to our previously assigned A, B, and C subtypes. Moreover, AKS2, and AKS4), 3 KS samples from Texas and Tennessee e s because of the high variability here, we were also able to (BKS8,BKS11,andBKS16),2fromNewZealand(ZKS5and t furthersubdividetheORF-K1proteinpatternsintoatotalof ZKS7), and 1 from Saudi Arabia (SKS8) are all classified as 13 distinct variants based either on amino acid differences havingtheA1pattern.Althoughverysimilar,nonewereiden- totalling5%orgreateroronthepresenceofcommondistinc- tical to one another, and almost all nucleotide changes gave tivein-framedeletionswithintheVR2domain.Thesevariant aminoacidsubstitutions.EvenC282,AKS1,andAKS2,which subgroupshavebeenassignedadditionalnumericaldescriptors showednonucleotidedifferencesatallwithinthe2,000bpof (e.g.,A1toA5,C1toC5,andD1orD2).Forthepurposesof constantregionsequenceanalyzedpreviously(70),displayed3, this analysis, we have reserved the term “clade” for several 8, and 5 nucleotide changes, respectively, in ORF-K1. The evennarrowerclustersofcloselyrelatedgenomeswithobvious maximumvariationfoundwithintheA1subsetwas10of289 relativelyrecentcommonorigins(suchasthosedescribedbe- amino acids (3%), and 7 of the 23 overall variant amino acid lowasA19ortheC39clusterfromTaiwan).Atabulardiagram positions were concentrated over a 10-amino-acid stretch be- summarizing the geographic origins, subgroup assignments, tweencodons60and69(referredtoaspartoftheVR*domain andkeyclinicalfeaturesofeachofthe63patientsispresented [described later]). A distinctive cluster of four very closely inFig.3.TheclassificationandclusteringshowninFig.2and relatedA1genomes(BCP1,BKS8,ZKS5,andZKS7)havinga 3weregeneratedbyvisualinspection,butsubsequentcomput- common LGVN feature at amino acid positions 66 to 69 is er-generatedtreeandradialdendrogramsillustratingtheover- referredtoastheA19clade. allphylogeneticdistanceandsimilarevolutionarybranchrela- Five more samples formed an A4 variant cluster (Fig. 3). 4160 ZONG ET AL. J.VIROL. D o w n lo a d e d f r o m h t t p : / / jv i. a s m . o r g / o n J a FIG. 2. AminoacidalignmentoftheORF-K1proteinsof63distinctHHV8genomes.ThecompleteaminoacidsequenceisgivenonlyforHHV8(BCBL-R)on n thetopline,withaminoacididentitiesindicatedbydashesfortheothergenomes.DeletionsinVR2areindicatedbygapsinparentheses,andsubtypedesignations u aregiventothefarright.TheA1,A2,A3,B,andD2subtypeORF-K1proteinsconsistof289aminoacids,whereastheA4subtypehas285aminoacids,theC3and a C2subtypeshave284aminoacids,theC1subtypehas282aminoacids,andthesingleexampleofsubtypeD1has302aminoacids.TheconservedCysresidues((cid:217)) ry surroundingthepredictedVR*loopandkeyaminoacidsinthepredictedITAMs(w)arehighlighted.;,potentialN-glycosylationsite;1,BCBLorPELtumoror 3 cellline;C,classicorendemicnon-HIV-associatedKS;R,renaltransplant-associatediatrogenicKS;A,patientlivedinorwasadirectimmigrantfromAfrica;P,Pacific , Islander;F,aggressiveKSfromFlorida.BlanksindicateincompletedataforAKS4,ST1,andJKS20,forwhichinsufficientoriginalDNAwasavailable.Allsamples 2 designatedasmembersoftheA1,A4,A2,A3,C3,C2,C1,B,andDvariantsaregroupedtogether.AKS,AIDSKSfromMaryland;BKS,samplesfromTexasand 0 1 Tennessee;OKS,AIDSKSsamplesfromTanzaniaorFlorida;RKS,AIDSKSsamplesfromZambia;SKS,renaltransplantKSfromSaudiArabia;TKS,samplesfrom 9 Taiwan;ZKS,samplesfromNewZealand.NotethatthedataforHBL6(BCBLcellline)andWKS1(KS)areidenticaltothepublishedsequencefortheBC1 b cellline(59).Similarly,ourresultsfortheBCBL1celllineandBKS14KSareidenticaltoeachotherandtopublisheddataforBCBL1(35).Finally,thedata y fortheBCBL-BcelllineandfortheBKS10KSsamplesfromdifferentpatientsarealsoidentical.KS-FshowsdatareportedbyNeipeletal.(48)foranAIDSKSsample g fromGermany. u e s t AlthoughoveralltheA4patternsareveryclosetothoseofthe variants were found in both the HBL6 and BC1 cell lines, as A1 group (i.e., only six to eight amino acid differences from wellasintheWKS1KSlesion.TheA2ORF-K1patterndiffers BCBL-R), all five contain a common 12-bp deletion encom- from the prototype A1 (BCBL-R) pattern by 14 amino acids passingcodons207to210intheVR2block.Mostofthemare (5%),withjustasingleadditionalsynonymousnucleotidedif- also distinguishable from the A1 pattern at positions D-54, ference,andfromallotherAvariantsbybetween7and8%of P-71, and Q-183. Two of the A4 DNAs, one from Tennessee their amino acids (Fig. 3). Overall, 17 of the 19 A1, A19, A2, (BCBL-B)andonefromTexas(BKS10),wereidenticalatthe andA4samplesinTable1camefromAIDSpatients,andthe aminoacidlevel,andonefromNewZealand(ZKS1)differed other2werefromrenaltransplantpatients(SKS8andZKS7). byonlyoneaminoacid,whereastheothersdifferedbythreeto However,recentVR1andVR2datafromthreenewsamples five amino acids from each other and from the first three. of classic KS from Israel (not shown) reveals that they also Although,thedendrogramanalysis(Fig.4A)separatedtheA4 clustertightlyintotheA19clade,confirmingthevalidityofthis variant subset from the A1 subset with a bootstrap value of designation(9a). only 44, this does not take into account the characteristic ThesecondmajorbranchoftheAsubtypepatternscontains four-amino-aciddeletion. atleasttwovariants,A3andA5,thataredistinguishablefrom The A2 pattern separates from the A1-plus-A4 branch on theA1,A4,andA2branchwithabootstrapvalueof83.Again, the dendrogram with a bootstrap value of 98. Identical A2 theonlytwoA3variantsfound,onefromtheBCBL1cellline, VOL.73,1999 HHV8 ORF-K1 SUBTYPE VARIATION AND EVOLUTION 4161 D o w n lo a d e d f r o m h t t p : / / jv i. a s m . o r g / o n J a n u a r FIG. 2—Continued. y 3 , 2 0 and the other from the BKS14 lesion, were identical to one variants or C3 and C4 variants (bootstrap value of 87). All C 1 another,buttheydifferedfromtheA1,A4,andA2aminoacid subtype ORF-K1 genes have the primary distinguishing char- 9 patternsby6to8%.Finally,OKS3,theonlyAsubtypesample acteristic of a 15-bp in-frame deletion of amino acids 201 to b y fromAfrica,showedadistinctiveA5patternthatalsodiffered 205inVR2,althoughthetwoC1samplesalsocontainasecond g from each of the other A variants by between 6 and 8%. 6-bpin-framedeletionofaminoacids213and214.Additional u AlthoughclosesttoA3inthedendrogram(bootstrapvalueof uniquefeaturesthatarecommontonearlyallCsubtypepat- e s 47),theA5patternalsohasseveralfeaturesmoretypicalofB terns, but are not found in A or B patterns, are Q-22, T-44, t orCsubtypesratherthanApatterns(e.g.,Q-22,Q-63,V-67, T-48,G-78,L-86,G-101,A-165,andL-168. N-222, P-223, V-227, and P-228). Importantly, among an ad- The only two subtype C genomes that occurred among our ditional 11 subtype A ORF-K1 genes examined that are not original 12 samples (70) are referred to as the prototype C1 includedinFig.3(becauseonlyVR1andVR2blockdataare (ASM72)andC2(EKS1)patterns.OneadditionalU.S.sample available at present), there are several other distinctive se- (BKS13), one Taiwan sample (TKS11), and four Saudi Ara- quence variants that fit best into the A3 category, especially biansamples(SKS2,SKS3,SKS6,andSKS7)turnedouttobe among U.S. and European patients with classic KS. In addi- C1 or C2 variants, whereas six other U.S. samples, including tion,weandothershavealsonowdetectedarelativelylarge thoseinthreePELcelllines(BC2,BC3,andJSC1),plusseven subset of African samples that closely resemble the A5 ofthenineTaiwanKSsamplesandKS-FfromGermanyallfell ORF-K1 pattern in the VR1 and VR2 regions (1, 33a, 51a, into a large new C3 subgroup. The eight C1 and C2 patterns 68a). have six conserved amino acid variations at D-63, N-69, S Subdivision of the C subtype ORF-K1 proteins into two (codonTCT)-75,M-88,A-104,andF-220thatdistinguishthem major classes. All subtype C ORF-K1 patterns differ from from all other ORF-K1 proteins, plus three at P-228, S-258, subtypeApatternsbyanaverageof14to15%oftheiramino andR-269thattheyshareincommonwithBandD1subtypes acids (bootstrap value of 70), and the majority (22 of 24) fall and three more at Y-20, D-43, and P-223 that they share in intooneoftwomajorbranchesconsistingofeitherC1andC2 commononlywithA3andA5orA5alone.Ourrationalefor 4162 ZONG ET AL. J.VIROL. D o w n lo a d e d f r o m h t t p : / / jv i. a s m . o r g / o n J a n u a r y 3 , 2 0 FIG. 2—Continued. FIG. 3. Summary of ORF-K1 subtype and clade patterns compared with 1 diseasetypeandgeographicorigins.Thediagramincludesalistingoftheclinical 9 characteristicsandgeographicsourceofeachofthe63individualHHV8samples b included in Fig. 2 together with the evolutionary relationships between the y distinguishing ASM72 and BKS13 as C1 patterns relative to variousORF-K1clades.p,prototypeA,B,andCsamplescollectedin1984;1, g EKS1,TKS11,SKS2,SKS3,SKS6,andSKS7asC2patternsis AIDS-associated,HIV-positivepatients;X,patientsfromwhichmultipleinde- u based primarily on the presence or absence of the 6-bp dele- pendentsamplesweresequenced(allprovedtobeidentical).KS,lesionbiopsy, e autopsy,orarchivalparaffinblocksamples;PEL,PELtumorsamples;BCBL, s tion. The two C1 samples also have three unique positions in t establishedlymphoblastoidcelllines;ClassicandEndem,non-HIV-associated common,A-43,H-62,andG-92,anddifferfromeachotherby KSpatients;Renal,iatrogenicrenaltransplantKS.Solidcirclesdenotethose only one amino acid. The dendrogram analysis also places genomesthathaveMratherthanPallelesoftheORF-K15gene(53).$,data SKS3closertoC1thantoC2whennottakingintoaccountthe fromNeipeletal.(48). characteristic6-bpdeletion(bootstrapvalueof46).Overallthe eight C1 plus C2 variants display a total of 22 amino acid differences among them (8%) and are distinguished from the samples into the C39 clade (bootstrap value of 44), no other C3groupbyabootstrapvalueof98. convincing clustering is discernible. Overall, the prototype The 14 examples of C3 variant ORF-K1 proteins are all C1-C2pattern(ASM72)differsfromtheprototypeC3pattern different from one another, but show many common charac- (BC2)bynearly10%,fromAsubtypepatternsby11to13.5% teristics that make them a very distinctive branch (bootstrap (plusthe5-amino-acidVR2deletion),andfromtheBpattern value of 92), including the presence of Q-63, P-92, S-172, by29%,whereasC3differsfromtheAsubtypesby12to14.5% M-215, L-223, and Q-226 and the absence of the C1- and (plusthedeletion)andfromtheBpatternby31%(Table1). C2-specificchangeslistedabove.Overall,therewereatotalof Similarities and divergence among renal transplant sam- 38 variant amino acid positions within the group, with the 20 plesfromSaudiArabia.AhighincidenceofiatrogenicKShas differences between BC2 and BC3 (7%) representing the ex- beenreportedinstudiesofrenaltransplantpatientsofMiddle tremes, but except for the clear separation of seven Taiwan Eastern and Jewish origin in both Saudi Arabia (54, 55) and VOL.73,1999 HHV8 ORF-K1 SUBTYPE VARIATION AND EVOLUTION 4163 Toronto (29). Foreman et al. (23) reported that four of five HHV8 genomes from the Saudi Arabian KS patients had in commonanunusualORF26patternthatdiffersfromthoseof fivetestedU.S.renaltransplantKSsamplesthatweremostlyA subtypes. Seven of the same Saudi Arabian samples were se- quencedhereintheORF-K1region,andsixprovedtohaveC subtype patterns. Only SKS8 was different by being a typical A1 genome. In contrast, SKS2, SKS3, SKS6, and SKS7 all closelyresembleourtwootherC2patterns(EKS1andTKS11),a subtypethathasnotyetbeendetectedamongourU.S.AIDS patientsamples.Overall,13ofthetotalof20KSsamplesthat we have tested that were not AIDS associated had C-subtype ORF-K1 genes, including 4 from Taiwan and 6 from Saudi Arabia, plus 1 each from Texas, Tennessee, and Maryland. Furthermore, the C2 variant EKS1-EKS2 specimens from MarylandalsocamefromanHIV-negativeclassicKSpatient D ofMiddleEasternbackground. o w TwoSaudiArabiansamples,SKS1andSKS9,provedtobe n unusualandhavebeenassignedastheprototypesofnovelC4 lo and C5 variants. Both include the characteristic C subtype a d 15-bpdeletion,butSKS1differsfromallotherC1,C2,andC3 e genomes at between 5 and 7% of its amino acids and was d excluded from the C1-C2 variant cluster on the dendrogram fr o (withabootstrapvalueof98).SKS9representsaspecialcase m that is otherwise closer to the A subtypes than to all other C h subtypesbyhaving11to13%aminoaciddifferencesfromthe t t C1,C2,andC3prototypesand8to10%differencesfromeach p : oftheAsubgroupORF-K1proteins(Table1).Thedendro- / / gramanalyses(Fig.4)placeSKS9asanomadicoutlierofthe jv Asubgroup,butthisdoesnottakeintoaccountthecharacter- i.a istic C subtype VR2 deletion, which we consider to be an s m overriding factor leading to its classification as a C5 variant . (Fig. 3). In fact, SKS9 more closely resembles A subtype pat- o r terns in the N-terminal half of the protein and C subtype g / patterns in the C-terminal half of the protein, although in o neither case does it fit clearly into any of the current variant n subgroups (Fig. 2). Therefore, SKS9 may either be an A/C J a chimera that arose originally by intertypic recombination n within the ORF-K1 gene, or it could represent a true evolu- u a tionaryintermediate. r y AfricanKSsamplesfallprimarilyintotheORF-K1Bsub- 3 type pattern. Among the 14 different HHV8 KS samples se- , quenced that fell into the B subgroup, 9 came directly from 2 0 Africa, including 1 endemic case of infection from 1984 in 1 Zaire(431KAP),2fromHIV-positivefemalesinUganda,1of 9 2fromHIV-positivepatientsinTanzania,andall5fromHIV- b y positivepatientsinZambia.AmongtheotherfiveBsamples, g at least four probably came indirectly from Africa. Two were u fromAfricanimmigrantstoeitherNewZealand(ZKS6/AIDS) e s ortheUnitedStates(JKS20classic),whereasOKS7andOKS8 t came from AIDS KS patients of black Haitian and Hispanic Mexican backgrounds, respectively (both from Miami), and JKS15 came from an African-American AIDS patient in Baltimore. Only 1 of the 10 samples tested from Africa (OKS35A5)didnothaveaBsubtypeORF-K1gene. The B subtype ORF-K1 patterns show up to 30% overall FIG. 4. Predictedphylogeneticrelationshipsamongthe63HHV8ORF-K1 aminoaciddifferencesfromtheA,C,andDsubgrouppatterns proteinsubtypesandvariantsexaminedhere.Thediagramsweregeneratedby (Table1),withnumerouscharacteristicsubtype-specificdiffer- PHYLIP,PRODIST,andNEIGHBORprogramanalysisbyRaphaelViscidiof encesoccurringessentiallythroughoutthewholelengthofthe the Department of Pediatrics, Johns Hopkins School of Medicine, based on estimated PAM distances and variances for each pair of the intact protein protein (Fig. 2). Even within the N-terminal signal peptide sequences.In-framedeletionsandinsertionsintheVR2blockarenottakeninto region, 13 of 26 positions differ from BCBL-R (A1) and account.(AandB)Linear(A)andradial(B)unrootedphylogeneticdendro- ASM72 (C1), and across the C-terminal cytoplasmic domain, grams with the SKS1 sequence as an outgroup. The length of each branch 12 of 38 positions differ, and yet these domains are both vir- indicatesgeneticdistancewiththesizescalefor0.1(10%difference)indicated. Confidencelevels(percent)frombootstrapanalysisatmajorbranchpointsare tually invariant between and within the A and C subtypes. given.Notallsampleshavebeenlabelledintheradialdiagrambecauseofspace Therefore,thedivergenceofBsubtypesfromaprogenitorof constraints. theAandCsubtypesprobablyrepresentsamucholderevent 4164 ZONG ET AL. J.VIROL. TABLE 1. LevelsofvariabilityamongtheORF-K1proteinmajorsubtypesandvariantsa Totalaminoaciddifferences Subtypeor variant A1 A4 A2 A3 A5 C5 C3 C4 C2 C1 B D1 D2 A1 51 15 19 19 26 40 33 35 39 85 75 73 A4 1.71 20 17 23 26 40 32 33 35 84 71 75 A2 5.2 6.9 22 23 25 42 38 36 39 85 72 78 A3 6.6 5.9 7.6 17 23 39 30 30 33 83 69 73 A5 6.6 8.0 8.0 5.9 29 35 30 28 32 81 70 74 C5 9.0 9.0 8.6 8.0 10.0 36 32 32 36 86 76 72 C3 13.8 13.8 14.5 13.5 12.1 12.5 28 24 28 89 80 78 C4 11.4 11.1 13.1 10.4 10.4 11.1 9.7 15 17 82 71 70 C2 12.1 11.4 12.5 10.4 9.7 11.1 8.3 5.2 91 84 72 73 C1 13.5 12.1 13.5 11.4 11.1 12.5 9.7 5.9 3.11 84 77 75 B 29.4 29.1 29.4 28.7 28.0 29.8 30.8 28.4 29.1 29.1 77 83 D1 26.0 24.6 24.9 23.9 24.2 26.3 27.7 24.6 24.9 26.6 26.6 601 D2 25.3 26.0 27.0 25.3 25.6 24.9 27.0 24.2 25.3 26.0 28.7 20.81 D o Totalaminoacids 289 285 289 289 289 284 284 284 284 282 289 302 289 w n aSummaryshowingnumbers(tophalf)andpercentages(bottomhalf)ofpairwiseaminoaciddifferencesbetweentheprototypeexamplesofeachofthe13ORF-K1 lo subtypes and variants recognized here. Deletions and insertions are not taken into account, but their contributions to variant classification designations are a acknowledgedbysuperscriptplussignswheresignificant. d e d that the separation of the A and C subtypes. There are no 2). In addition, the TKS10 protein differed from all A and C fr o fewer than 40 totally invariant B-specific amino acid changes variantsby24to28%andfromtheBprototypeby27%atthe m relativetotheAandCpatterns,with22othersthatarenearly amino acid level (Table 1). The D1 ORF-K1 protein is more h invariant. All B subtype examples also differed significantly similar to B than to A and C in the VR1 block, but more t t fromoneanother,althoughnoclearclusteringorsubgrouping similar to A and C than to B in VR2 and at the N and C p : was apparent even in the dendrogram analysis. A total of 58 termini. There are 18 common amino acid positions within B / / residues show variations across the whole B subset, with the andD1thatdifferfromsitesconservedinAandC,aswellas jv maximum range being the 26 amino acid (9%) differences 31 common amino acid positions between A, C, and D1 that i.a between OKS7 and RKS5. RKS3 and RKS5 show significant differ from those in B subtypes (Fig. 2). On the other hand, s m differences from the other Zambian samples, whereas RKS5 TKS10 also has D-43 in common with the A3, A5, and C2 . andZKS6showsomecommonuniquefeatures,suchasF-33, patterns; P-227 in common with the A5, C2, and B patterns; o r K-34, Y-84, Q-89, E-93, and G-189. The three nonimmigrant and both P-92 and R-200 in common only with many C3 g / American B samples have two unique positions in common patterns. The overall interpretation from both manual align- o (Y-5 and R-223), but they also differ significantly elsewhere. ment(Fig.2and3)andaccordingtophylogenetictreeanalysis n Only 431KAP and JKS20 represent endemic non-AIDS-asso- (Fig.4)isthatthedivergenceoftheDbranchoccurredatan J a ciatedKSsamples,andnoPELsamplesfromAfricahavebeen intermediate level relative to the divergence of the A-plus-C n examinedasyet.Importantly,theBsubtypeORF-K1proteins branch from the B branch. Both the radial dendrogram (Fig. u a arenotonlymoredivergedamongthemselvesthanareeither 4B)andtheaminoaciddifferencetable(Table1)suggestthat r y theAorCsubgroups,buttheirbranchlengthsinthedendro- the D branch point lies somewhat closer to the A-plus-C lin- 3 gram (Fig. 4) are also noticeably longer (especially ZKS6), eagethantotheBlineage. , supporting the idea that they may represent a more ancient AdditionalDsubtypegenomesfromtheSouthPacific.After 2 0 anddiversegroupthantheothers. discoveringthattheunusualDsubtypegenome(TKS10)came 1 CommonfeaturesoftheTaiwanC3*cladeandidentification from a classic KS patient who had a Pacific Island ethnic 9 ofanovelDsubtype.Interestingly,thesevenC3subtypeTai- heritageratherthantheChineseheritageoftheotherTaiwan b y wansamples(fourclassicandthreeAIDSassociated)allrep- patients, we searched for additional archived diagnostic sam- g resentadistinctivesubsetwithfourcommonnovelaminoacid ples that might be representative of a Pacific branch of the u variationsthatarenotfoundinanyotherCsubtypegenomes virus. A total of six HHV8-positive KS lesion DNA samples e s (represented by R-165, R-200, D-222, and the absence of from New Zealand were examined and proved to represent t S-65).However,theyarealsoquitevariableamongthemselves, severaldifferentsubtypes.SamplesZKS1andZKS5frommale with TKS2 and TKS9 having six common unique nucleotide Caucasian AIDS patients had typical A4 and A1 variant variations at A-79, C-341, G-392, C-765, T-741, and C-742, ORF-K1genes,respectively,andthesamplefromaCaucasian althoughalsodifferingatthreepositions,andwithTKS5being renal transplant patient (ZKS7) was also an A1 subtype. In quite distinctive from all other HHV8 genomes tested, with contrast,sampleZKS6camefromanHIV-positiverecentim- uniquenucleotidevariationsthatincludeG-75,A-306,G-307, migrant from southern Africa and was a B subtype closely G-308,G-317,andA-734.Basedontheobviouscommonlin- relatedtoRKS5fromZambia.Finally,thegenesintwosam- eageofthissetofsampleswithintheC3subgroup,wereferto ples, referred to as ZKS3 and ZKS4, from HIV-negative el- themasadistinctC39clade(bootstrapvalueof44). derlymalePacificIslanderswithclassicKSweredramatically Among the other two samples from Taiwan, TKS11 came different from all other ORF-K1 genes examined and have from a renal transplant recipient and proved to have a C2 been provisionally designated as D2 variants within the same subtypeORF-K1gene,whereasTKS10,representingaclassic PacificDsubtypeasTKS10. non-AIDS-associatedKScaseinanaboriginalHwalianpatient The ZKS3 and ZKS4 ORF-K1 proteins differ by only one fromeasternTaiwan,provedtobetheprototypeofanovelD1 aminoacidfromeachother,andalthoughconsiderablydiffer- subtype genome. A major feature of TKS10 is a 39-bp dupli- ent from the D1 pattern, they also display several features in cationofaminoacids181to193closetotheVR2region(Fig. common only with TKS10, such as L-32, D-43, S-83, Q-89, VOL.73,1999 HHV8 ORF-K1 SUBTYPE VARIATION AND EVOLUTION 4165 D o w n lo a d e d f r o m h t t p : / / jv i. a s m . o r g / o n J a FIG. 5. HomologybetweenORF-K1andthevariableregionofimmunoglobulin(IG)lightchains.(Toppanel)Typicalvariationsseenwithinthemajorsubtypes n ofORF-K1inthevicinityoftheimmunoglobulinhomology.DashesindicatematchestotheORF-K1subtypeAsequence(topline).TheconservedCysatposition u 117inORF-K1isindicatedinablackbox.(Centerpanel)AminoacidalignmentsbetweentheNterminiofmatureandunprocessedimmunoglobulinlambdachains a (positions7to35and21to56,respectively),aBenceJonesprotein,andaBCRproteinwithpositions86to140ofORF-K1subtypeA.Dashesindicatespaces,asterisks ry denoteidentitiestotheORF-K1Aversion,andverticalbarsdenotesimilarities.(Bottompanel)Homologiesamongdifferentlambdachainvariableregionsubtypes 3 comparedtotheORF-K1(A)immunoglobulinhomologyregion.Thenumberingusedrepresentsthatofaminoacidpositionsinthematureimmunoglobulinforms , lackingthesignalpeptide.FRI,flankingregionI;CDR1,complementarity-determiningregionI.Othersymbolsarethesameasforthecenterpanelabove. 2 0 1 9 R-112,C-116,A-121,T-123,andG-133.Theyalsohaveanum- single-domain immunoglobulin family receptor with two ex- b berofotherfeaturesincommononlywithboththeBandD1 tendedloops(VR1andVR2)addedtothelargeextracellular y subtypes,suchasP-14,H-20,G-54,I-126,R-128,Q-135,P-206, domain.Intriguingly,ORF-K1containsa36-amino-acidblock g u A-211,E-216,andF-277,buttheydonothavethe13-amino- between positions 97 and 133 that has 50 to 55% amino acid e acidduplicationseeninTKS10(Fig.2).Overall,theD2ORF- identitytotheextremeNterminusofthematureformsofthe st K1proteinsdifferfromtheAandCprototypesby25to27%, variable domains of certain immunoglobulin lambda light fromBby29%,andfromD1bynearly21%attheaminoacid chains, in particular those referred to as subtypes VII, VIII, level (Table 1). Although quantitatively the differences be- andXII(Fig.5).Thishomologyisnotsignificantlyaffectedby tweentheD1andD2ORF-K1proteinsaregreaterthanthose polymorphismofdifferentORF-K1subtypesandincludescon- between the A and C subtype patterns, the dendrogram pat- servationofthefirstCysbridgeresidueoftheimmunoglobulin ternsclearlyshowthemtobeonacommondistinctivebranch variableregion(positionC-117inORF-K1).Relativetoclas- (Fig.4B),anditwasdeemedunwisetointroduceanadditional sical immunoglobulin domain family proteins, VR1 is posi- Edesignationuntilmoresamplesofthesesubtypeshavebeen tionedbetweentheN-terminalsignalpeptideandthebeginning found and evaluated. Furthermore, D1 and D2 proved to be ofthecomplementarity-determiningregionofthevariabledo- virtuallyidenticalacrosstherestoftheirgenomes(53). mainhomologyregion,whereasVR2liesadjacenttothesingle C-terminaltransmembranespanning(orTM)domain(seeFig. DISCUSSION 1C).Thereareatotalof10completelyconservedCysresidues at positions 28, 40, 52, 76, 100, 109, 117, 154, 157, and 170 in StructuralsimilaritybetweenORF-K1andimmunoglobulin thematureextracellulardomainofORF-K1,aswellasseven familyreceptors.Thesizeandstructureofthe289-amino-acid predominantlyconservedN-glycosylationmotifs. ORF-K1 protein show a distinct resemblance to those of a The 38-amino-acid cytoplasmic tail domain in all ORF-K1