Goldentoutoulu.com
toutoulu.com 时间:2021-03-18 阅读:()
RESEARCHOpenAccessNoevidenceofprenataldiversifyingselectionatlocusorsupertypelevelsinthedogMHCclassIIlociAlinaK.
Niskanen1,2*,LornaJ.
Kennedy3,HannesLohi4,JouniAspi1andTanjaPyhjrvi1AbstractBackground:Despitedecadesofstudying,themechanismsmaintaininghighdiversityinthegenesoftheMajorHistocompatibilityComplex(MHC)arestillpuzzlingscientists.
Inadditiontopathogenrecognitionandotherfunctions,MHCmoleculesmayactprenatallyinmatechoiceandinmaternal-foetalinteractions.
TheseinteractionsarepotentialselectivemechanismsthatincreasegeneticdiversityintheMHC.
Duringpregnancy,immuneresponsehasadualrole:thefoetusrepresentsforeigntissuecomparedtomother,buthisto-incompatibilityisrequiredforsuccessfulpregnancy.
WehavestudiedtheprenatalselectioninMHCclassIIloci(DLA-DQA1,DLA-DQB1andDLA-DRB1)indomesticdogsbycomparingtheobservedandexpectedoffspringgenotypeproportionsin110dogfamilies.
Severalpotentialselectiontargetswereaddressed,includingthepeptide-bindingsite,theMHClocus,three-locushaplotypeandsupertypelevels.
Forthesupertypeanalysis,thefirstcaninesupertypeclassificationwascreatedbasedoninsilicoanalysisofpeptide-bindingamino-acidpolymorphism.
Results:Inmostlociandlevels,nodeviationfromtheexpectedgenotypefrequencieswasobserved.
However,onepeptide-bindingsiteinDLA-DRB1hadanexcessofheterozygotesamongtheoffspring.
Inaddition,ifthefathersharedaDLA-DRB1allelewiththemother,thatallelewasinheritedbytheoffspringmorefrequentlythanexpected,suggestingtheselectiveadvantageofahisto-compatiblefoetus,incontrasttoourexpectations.
Conclusions:Weconcludethatthereissomeevidenceofpost-copulatoryselectionatnucleotidesitelevelintheMHClociofpetdogs.
Butduetonoindicationofselectionatlocus,three-locus,orsupertypelevels,weestimatedthattheprenatalselectioncoefficientislessthan0.
3indomesticdogsandverylikelyotherfactorsaremoreimportantinmaintainingthegeneticdiversityinMHCloci.
Keywords:Canisfamiliaris,MHC,Prenatalselection,SupertypeclassificationPlainEnglishsummaryMajorHistocompatibilityComplex(MHC)isalargegeneticregioncodingforimmunedefensemoleculesanditisfoundinallvertebrates.
MHCmoleculestriggerimmuneresponsebybindingmoleculesderivedfrompathogensandrecogniseself-tissuesfromnon-self-tissues.
TheMHCmoleculesareoftenverydiverse.
Thisdiversityismainlythoughttobeaconsequenceofbalancingselectioncausedbypathogenpressure.
However,alsoselectionbeforebirththatmaytakeplaceduringmatechoiceorduringpregnancyhavebeendemonstratedinsomespecies.
Sincedomesticdogsareoftenunderintenseveterinarycare,pathogenpressuremightbelessstrongcomparedtowildanimals.
Yet,selectivepatternsofMHCdiversitybeforebirthhavenotbeenstudiedbeforeindogs.
WestudiedtheselectionbeforebirthatthreeMHCclassIIgenesin10dogbreedsandfourmixed-breedgroups.
Wecomparede.
g.
theexpectedandobservedratiosofheterozygousandhomozygouspuppiesin110dogfamilies.
Weconductedselectionanalysesinmanydifferentlevels,including:I)sitesthatbindmoleculesderivedfrompathogens(=peptide-bindingsite),II)the*Correspondence:alina.
niskanen@gmail.
com1DepartmentofGeneticsandPhysiology,UniversityofOulu,POBox3000,OuluFIN-90014,Finland2Presentaddress:CentreforBiodiversityDynamics,DepartmentofBiology,NorwegianUniversityofScienceandTechnology,NO-7491,Trondheim,NorwayFulllistofauthorinformationisavailableattheendofthearticleTheAuthor(s).
2016OpenAccessThisarticleisdistributedunderthetermsoftheCreativeCommonsAttribution4.
0InternationalLicense(http://creativecommons.
org/licenses/by/4.
0/),whichpermitsunrestricteduse,distribution,andreproductioninanymedium,providedyougiveappropriatecredittotheoriginalauthor(s)andthesource,providealinktotheCreativeCommonslicense,andindicateifchangesweremade.
TheCreativeCommonsPublicDomainDedicationwaiver(http://creativecommons.
org/publicdomain/zero/1.
0/)appliestothedatamadeavailableinthisarticle,unlessotherwisestated.
Niskanenetal.
CanineGeneticsandEpidemiology(2016)3:9DOI10.
1186/s40575-016-0038-9MHCgenes,III)three-genecombinationsandIV)super-types.
WecreatedthefirstcaninesupertypeclassificationthatjoinsdifferentMHCallelesintolargersupertypesbasedontheirchemicalsimilarities.
Wefoundthatpuppiesweremoreheterozygousthanexpectedatonepeptide-bindingsiteofoneMHCgene(DLA-DRB1).
WealsofoundthatifthefathersharedaDLA-DRB1allelewiththemother,thatallelewasinhe-ritedbytheoffspringmorefrequentlythanexpected.
Otherlevelsofinvestigationdidnotshowdeviationsfromexpectedratiosofhomozygousandheterozygousoffspring.
Asaconclusion,wefoundsomeevidenceofselectionbeforebirthindogs.
However,otherfactorsarelikelycontributingmoretotheMHCdiversityindogs.
BackgroundNaturalselectioninthegenesoftheMajorHistocom-patibilityComplex(MHC)hasbeenstudiedextensivelyinwildanimalpopulations[1].
Acrossmanyspecies,theMHCgenesarethemostpolymorphicinthegenome.
ThemainfunctionofclassicalMHCmoleculesistobindantigenicpeptidesandpresentthemtoT-cells[2].
TheMHCgenesaredividedintotwoclasses;classImoleculesareexpressedonthesurfaceofalmostallnucleatedcells,while,inman,classIImoleculesareexpressedonspecificantigen-presentingcells(e.
g.
macrophagesandlymphocytes).
Inman,theextendedMHCregionalsoincludesnon-classicalgenesthathavediverseimmuneandnon-immunefunctions,includingthecomplementsystem,andolfactoryreceptor(OR)genes[3,4].
SincethefunctionofMHCmoleculesistriggeringimmuneresponse,parasitesanddiseaseshavebeenthoughttoplaythemajorroleinthemaintenanceofMHCdiversity[5,6].
However,indomesticatedanimalspathogenpressurehasprobablyhadasmallercontri-butiontotherecentselectiontargetingMHC,comparedtowildspecies.
Instead,thereareotherpotentialforcesdrivingselection,suchasmatechoice,spermselection,maternal-foetalinteractionsandautoimmunediseases[7–9].
ManyMHCassociationswithautoimmunediseaseshavebeenidentifiedinman[10].
SimilarMHCasso-ciationshavealsobeendemonstratedincanineauto-immunedisorders[11].
TheseassociationsmaybeoneoftheselectiveforcesinfluencingMHCdiversityamongdogs(Canisfamiliaris).
Thecloserelationshipbetweentheownerandtheirdoghasbeenakeydriverinacquiringextensiveknow-ledgeofcaninediseasesandresultedinahighlevelofveterinarycare,especiallyinWesternsocieties.
Theformationofdogbreedsandartificialselectionbyhumanshasbeenespeciallystrongduringthelast100–150years.
Asaconsequence,dogbreedshavebecomesmallandgeneticallydistinctgroups.
MHCdiversitydiffersfrombreedtobreedandisoftenlow[12],probablyduetodrifteffectsinsmallpopulations.
However,evenwithoutfreematechoice,MHCdiversityinadogbreedcanbeclosetothediversityofawildwolfpopulation[13,14].
HumanshavehadmajorimpactinsomeoftheaspectsthatmayinfluencethedogMHCdiversity,forinstance,throughveterinarycareandmatechoice.
Therefore,ifMHCin-compatibilitycausesprenatalselection,itmaybepossibletodetectsuchselectioninthedomesticdog,whereotherfactorsofnaturalselectionmaybelessstrong.
PrenatalselectionAlthoughtheoverallMHCdiversityofdogshasbeenstudiedwidely[12,15],theroleofprenatalselectioninthemaintenanceoftheMHCdiversityhasnotprevi-ouslybeenexamined.
Matechoiceindogsismoreoftendecidedbytheownerratherthanthedog,especiallyinthepurebreddogpopulation.
Hence,naturalselectionmayhaveitsstrongestimpactontheMHCdiversityatthepost-copulatorystagesfromeggfertilisationuntilbirth.
AssociationbetweenMHCvariationandfertili-sationsuccesshasbeendemonstratedinotherspecies,e.
g.
mice[16].
Prenatalselectioncanbedetectedfromnon-MendelianproportionsofMHCgenotypesintheoffspring.
Forinstance,theexcessofheterozygousoff-spring[17]andincreasedfoetallossamongMHC-matchingcouples[8]hasbeenfoundinhumansandinpigtailedmacaques[18].
Fromthemother'sperspective,afoetuscorrespondstosemiallograftwithhalfofitsgenomebeingforeign.
Tomaintaingestation,theimmuneresponseofthemotherisdown-regulated.
However,asuccessfulpreg-nancyrequiressomedegreeofmaternalimmunestimu-lation,whichcouldbeinducedbyhisto-incompatibility(immunologicaldissimilarity)betweenthemotherandthefoetus[19,20].
MHCmoleculeshavecrucialroleinsuccessfulpregnancy.
Inhumans,mostclassicalMHCmoleculesaredownregulatedandonlyoneclassicalMHCclassIgene(HLA-C)andtwononclassicalMHCclassIgenes(HLA-Eand-G)areexpressedinthefoetalsideoftheplacenta,intheextravilloustrophoblast(EVT)cells[21–23].
TheEVTcellsmigrateintothemother'suterusandtransformsmallarteriesintolargebloodvesselstoensuresufficientnourishmentofthefoetus[24].
Theprocessoftransformingthearteriesiscontrolledbythematernalnaturalkiller(NK)cells.
MostlikelyallthreeMHCmolecules,HLA-C,Eand-G,interactwithmaternalNKcellsinregulatingplacentalvascularisation[25].
HLA-CmoleculeistheonlypolymorphicMHCmoleculeexpressedintheEVTcells,sothesuccessofpregnancyisinfluencedbythecompatibilityofthefoetalHLA-CmoleculesandmaternalNKcellreceptors.
Incon-gruityofthefoetalMHCmoleculesandmaternalNKcellNiskanenetal.
CanineGeneticsandEpidemiology(2016)3:9Page2of12receptorsmightleadtoinsufficientbloodsupplytotheplacentawhichcancausepre-eclampsia,miscarriageandotherpregnancydisorders[26].
TheexpressionofMHCgenesduringpregnancyandmechanismsofmaternal-foetalinteractionsindogsarelargelyunknown.
IthasbeenshownthatsomeclassicalclassI(DLA-88)andII(DLA-DRA)MHCmoleculesarenotexpressedinearlycanineembryos[27]butthereisanincreaseofMHCclassIIexpressingcellsintheuterusofpregnantfemales[28].
MHCsupertypesThepeptiderecognitionrepertoiresofMHCmoleculesproducedbydifferentallelesatthesameMHClocusoverlapwitheachothertosomeextent.
Aspartoftheefforttodevelopepitope-basedvaccineswiththeabilitytoimmunisehumanpopulationswithdiverseMHCgenotypes[29],MHCalleleshavebeengroupedintosupertypes,basedontheirstructuralfeaturesorpeptide-bindingspecificity.
IftheactualdifferencesinpathogenrecognitionandpeptidebindingareattheMHCsuper-typelevel,theymightrepresenttheunitofnaturalselec-tionandshouldbetakenintoaccountwhenstudyingselectionintheMHCgenes.
Untilnow,supertypeclassi-ficationshavemainlybeenproducedforprimatesandbirds[30–32],butnonehavebeenattemptedforcanines.
AimsofthestudyTheaimofthisstudywastoinvestigatepossibleprenatalselectioninthreecanineMHCclassIIloci,DLA-DRB1,DLA-DQA1,andDLA-DQB1,byexaminingtheMHCgenotypeproportionsindogfamilies.
WealsoanalysedtheMHCallelecompositionandheterozygosityoftheoffspringrelativetotheirmothers,tostudypotentialmaternal-foetalinteractions.
Sincesuccessfulpregnancyrequireshisto-incompatibilitybetweenthemotherandthefoetus,wewouldexpecttofindanexcessofhete-rozygousoffspringundertheinfluenceofprenatalselec-tion.
Similarly,wewouldexpectthat,whenparentssharethesameallele,theoffspringwillinheritthepater-nalnon-sharedallelemoreoftenthansharedallelefromtheirfather.
Finally,wecomparedthefunctionalsimila-ritiesoftheidentifiedcaninealleles,andgeneratedthefirstsupertypeclassificationforDLA-DRB1andDLA-DQB1alleles.
MethodsSamplesandDNAtechniquesThedatacomprise430puppiesand220parentsin110familiesthatbelongtotenpurebredandfourdiffe-rentmixed-breedgroups(Table1andAdditionalfile1:TableS1).
Eachfamilyincludesparentsand1–16puppieswiththemeanof3.
9puppies.
Apartfromafewlittersthatwerefullysampled,arandomsampleoftheoffspringwasavailableformostfamilies.
Multiplelittersfromthesameparentswerecombinedasonelitter.
ThedogsweresampledinFinland,UKandUSA.
Thirty-sixindividualswereinthedatasetbothasaparentandasanoffspring.
Thirtyparentshadmorethanonelitterwithdifferentmates(24parentshadtwo,fourhadthree,andtwohadfourlitters).
ThehighlypolymorphicsecondexonthatencodesthepeptidebindingpartoftheMHCmoleculewassequencedforDLA-DRB1(270bp),DLA-DQA1(246bp)andDLA-DQB1(267bp)genes.
Thegenesarelocatedclosetogetheroncaninechromosome12intheabove-mentionedorder[33].
ThedistancebetweenDLA-DRB1andDLA-DQA1is56.
6kb,whilethedistancebetweenDLA-DQA1andDLA-DQB1is77.
7kb.
Apre-viouslyestablishedprotocol[13]wasusedtosequencethelociinacohortof253dogs.
Inaddition,datafrom41familieswith192puppiesfromanearlierstudybyKennedyetal.
[34]plusafurther12familieswith64puppiesfromRowan[35]wereincluded.
MicrosatellitegenotypingmethodforMHCgenotypingwasusedforonemixed-breedgroupofsixfamiliesincluding37offspring[35].
Inthismethod,dogsweregenotypedusingmicrosatellitesCFA12-3(dinucleotide),CFA12-4(tetranucleotide),andCFA12-9(dinucleotide),whicharelinkedwiththestudiedDLAloci[36].
Inaddition,paren-taldogsweregenotypedbysequencingtoconfirmtheMHCgenotypes.
Table1Thedogbreeds,thenumberofoffspring(NO),thenumberoffamilies(NF),thecountryofsamplingandtheMHCgenotypingmethodBreedNONFCountryaMHCgenotypingmethodKooikerhondjeb1710FinlandSequencingSalukib5116FinlandSequencingLowchenb94FinlandSequencingIcelandicSheepdogb73FinlandSequencingKromfohrlanderb9024FinlandSequencingAlaskanHuskyc276USASequencingAlaskanHuskyd41USASequencingEnglishCockerSpanield61USASequencingGoldenRetrieverd73USASequencingStaffordshireBullTerrierd73USASequencingNewfoundlandd4212UKSequencingRhodesianRidgebackd425UKSequencingMixed-breedd8416USASequencingMixed-breedc376USASequencingormicrosatellitebasedTotal430110aBirthcountrymaydifferfromthecountryofsampling,bDatafromthisstudy,cRowan[35],dKennedyetal.
[34]Niskanenetal.
CanineGeneticsandEpidemiology(2016)3:9Page3of12MHCalleleandhaplotypeassignmentMHCallelesweredeterminedfromthesequencedatausingMatchToolsandMatchToolsNavigator(AppliedBiosystems)programs.
MatchToolsincludesalibraryofcanineMHCallelesandcomparesahomozygousorheterozygoussequencepairtotheallelesequencesinthelibrary.
TwopuppiesfromthedatabyKennedyetal.
[34]hadarecombinantalleleinonelocusandthatlocuswasexcludedfromtheanalysesinthoseindividuals.
Usually,theMHCallelesfromdifferentlociareinhe-ritedashaplotypesduetostronglinkagedisequilibrium(LD,non-randomassociationofallelesbetweendifferentloci).
Three-locushaplotypesconsistingoftheDLA-DRB1,DLA-DQA1andDLA-DQB1allelesweredefinedinthreesteps:1)haplotypesweredeterminedforhomo-zygousindividuals,2)haplotypesforheterozygousindi-vidualswereinferredbasedonthehaplotypesobservedinthehomozygousindividualsand3)rarehaplotypes,detectedonlyintheheterozygousindividuals,wereiden-tifiedbysubtractionofhaplotypesdeterminedintheprevioussteps[14].
In35families,missingparentalhaplotypeswereinferredbasedonthehaplotypesoftheoffspringandtheotherparent.
TheDLA-DQA1allelesofthreeparentsandtheDLA-DQB1alleleofoneparentwereinferredbasedonthefamilymembers'alleles.
Inthemicrosatellite-basedgenotypingoftheDLAal-leles,themicrosatellitedatawereanalysedwithGenoty-per3.
7(AppliedBiosystems).
ThemicrosatellitealleleswerematchedwiththeDLAhaplotypesinthedogsthathadbeengenotypedbybothmethods.
AfteridentifyingthecorrespondingmicrosatelliteallelesandDLAhaplo-types,theDLAhaplotypeswereassignedtoindividualsthathadbeengenotypedonlywithmicrosatellites.
MHCsupertypeclassificationMHCsupertypeclassificationwasdevelopedtofindoutifprenatalselectionactsatthatlevel.
MHCallelesintheβloci,DLA-DRB1andDLA-DQB1,wereclassifiedintoMHCsupertypesbasedonthechemicalpropertiesoftheaminoacidsequencesintheputativepeptide-bindingregion(PBR).
TheclassificationwasdoneforofficiallynamedcanineMHCallelesfromtheMatch-Toolslibrarycomprisingallelesfromdifferentcaninespecies,mostlyfromdogandgreywolf(Canislupus).
Altogether205DLA-DRB1allelesand100DLA-DQB1alleleswereincluded.
Sincethesupertypeclassificationofallelesisbasedonthesharedpropertiesinpeptidebinding,theputativePBRsiteswereselectedaccordingtohumanPBRsitesproposedbyDoytchinovaandFlower[37].
Onlypoly-morphicPBRsiteswereincludedintheanalyses;19sitesinDLA-DRB1(codonpositions:9,11,13,26,28,30,38,47,56,57,60,61,67,70,71,74,78,86and90)and15sitesinDLA-DQB1(codonpositions:9,13,26,28,30,37,38,47,57,67,70,71,74,77and85).
Also37canineDLA-DQA1alleleswerecheckedforpolymorphismsinputativePBRsitesbutall10sitesweremonomorphicandthusthelocuswasexcludedfromthesupertypeclassification.
ChemicalpropertiesoftheaminoacidsinPBRsitesweredescribedusingfivez-descriptors:z1(hydrophobicity),z2(stericbulk),z3(polarity),z4andz5(electroniceffects)[38].
Z-descriptorvalueswerestandardisedsothattheirdistributionshadameanofzeroandastandarddeviationofone.
Allz-descriptor*polymorphicsitecombinationswereusedforestimatingEuclidiandistancesbetweenthedifferentMHCalleles.
BasedontheEuclidiandistances,thealleleswereclusteredusinghierarchicalclusteringwithWard'smethod[39]inRstatisticalenvironmentv.
3.
1.
2[40].
ThenumberofclusterswasdecidedbasedonthecostofmergingclustersinWard'smethod.
Thesmallestnumberofclusters,beforeasteepincreaseinthecostofmergingclusters(variancewithinclusters),waschosen.
TostudywhichPBRsiteshadmostinfluenceintheclusteringandtovisualisetherelationsofthesupertypeclusters,aprincipalcomponentanalysis(PCA)wasconductedusingtheprcompfunctioninR.
SegregationanalysesTostudywhethertheDLA-DRB1,DLA-DQA1andDLA-DQB1allelesfollowMendelianinheritanceindogfamilies,heterozygosityanalyseswereconductedinRstatisticalenvironmentv.
3.
1.
2[40].
Therearefourpossiblematingcrosstypeswheredifferencesinexpectedandobservednumbersofheterozygotescanbeobserved:A1A1*A1A2,A1A2*A1A1,A1A2*A1A2andA1A2*A1A3(where:mother*father,andtheallelesA1≠A2≠A3arecompositesofdifferentobservedalleles).
Inourdataset,altogether74(DLA-DRB1andDLA-DQB1)and83(DLA-DQA1)familiesincluding293(DLA-DRB1),313(DLA-DQA1)and295(DLA-DQB1)puppiesmetthiscriterion.
Thenumbersofheterozygousandhomozygousoffspringwerecountedwithineachmatingcrosstype.
TheexpectednumbersofheterozygousandhomozygousoffspringwerecalculatedbasedontheMendelianinheri-tancemodelandcomparedtoobservednumbersofoffspringusingPearson'schi-squaredtestwithMonteCarlosimulatedp-value(2000repeats).
Chi-squaredteststatisticswerealsocombinedoverthematingcrosstypesineachlocustoexaminetheoveralldeviationfromtheexpectednumberoftheheterozygotes.
Theanalyseswereconductedatlocusandthree-locushaplotypelevelsfortheentiredata,andseparatelyfordifferentbreeds/groupsandsexestoidentifythepoten-tialbreedorsexrelateddifferencesintheinheritancepatterns.
Informationaboutsexwasavailablefor381offspring.
Inaddition,purebredandmixed-breeddogswereanalysedseparately.
SegregationanalyseswerealsoNiskanenetal.
CanineGeneticsandEpidemiology(2016)3:9Page4of12doneforthesupertypeclassesandseparatelyforeachpolymorphicPBRsiteintheDLA-DRB1andDLA-DQB1loci.
Tocorrectformultipletesting,afalsediscoveryrate(q-value[41])wasestimatedforp-valuesofthecombinedchi-squaredtestsofthedifferentmatingcrosstypesoverthePBRsitesinonelocus.
Twohaplotypesthatwerefoundinfourbreeds/groupscarriedaduplicatedDLA-DQB1locus(DLA-DQB1*01303andDLA-DQB1*01701[34]).
Sinceitisunclearwhetherbothcopiesareexpressedornot,analyseswereconductedincludingallindividualsandseparatelyforthefamilieswith(numberofpuppies=51(DRB1)/59(DQA1)/53(DQB1))andwithout(numberofpuppies=242(DRB1&DQB1)/254(DQA1))theduplicatedlocus.
Whenincludedintheanalyses,theduplicatedlocuswashandledasoneallele(DLA-DQB1*013017).
IntheanalysesfortheDLA-DQB1locuswithsupertypeclassificationandforeachPBRsite,allfamilieswithduplicatedlocuswereexcluded.
MaternaleffectWestudiedthematernaleffectontheMHCclassIIalleletransmissioninthefamilieswheretheoffspringmayinheritthematernalsharedorpaternalnon-sharedallelefromtheirfather:A1A2*A1A3andA1A1*A1A2.
Wetestedthedistributionbetweentheexpectedandobservednumbersofthesharedandnon-sharedallelesinheritedfromfatherwithPearson'schi-squaredtestwithMonteCarlosimulatedp-value(2000repeats).
Thereissomeevidenceinman,thatthemother'sheterozygositymayinfluencetheoffspringheterozygosity–inAmerindianfamiliesonlytheheterozygousmothers'hadincreasedproportionofheterozygousoffspring[17].
So,todiscoverwhetherthemother'sheterozy-gosityhasaneffectontheheterozygosityoftheoff-springindogs,familieswithheterozygousmotherswerejoinedforanalyses.
Thematernaleffectwasstudiedinthetotaldatasetandinthesamesubsetsasdescribedinsection2.
4.
ResultsMHCallelesandhaplotypesAltogether28DLA-DRB1,11DLA-DQA1and23DLA-DQB1allelesorganisedin45three-locushaplotypeswerefoundinthedogsample(Table2).
Thereweresomecommonhaplotypesthatwerefoundinabouthalfofthebreeds/groupsstudiedbutrareuniquehaplotypeswerealsodetectedinmanybreeds.
ThreenewDLA-DRB1al-leleswerefoundamongthe253dogssequencedforthisstudy.
TheallelesequenceshavebeensubmittedtoNCBIGenBankwiththeaccessionnumbers:JQ904819.
1,FM246834,andFM246838.
SupertypeclassificationTheDLA-DRB1allelesweredividedintofivesupertypeclassesandtheDLA-DQB1allelesweredividedintoninesupertypeclassesusinghierarchicalclusteringmethod.
Thenumberofclasseswasbasedonthecostofmergingclusters(Additionalfile2:FiguresS1–S2).
SupertypeclusteringtreesforthecanineDLA-DRB1andDLA-DQB1allelesareshowninFigs.
1and2andthesupertypeclusteringtreeswithallelenamesareshowninAdditionalfile3:FiguresS3–S4.
CompletelistsoftheallelesineachsupertypeclassareinAdditionalfile4:TableS2.
Inhumans,eachMHCsupertypecouldbedescribedbycommonfingerprintaminoacidsthatweresharedwithineachsupertypeclass[37].
Incanines,suchsuper-typefingerprintscouldnotbedetected,sinceaminoacidswereshared,tosomeextent,alsobetweendifferentsupertypes.
BasedonthethreefirstPCsthePBRsites1,5,6,and9inDLA-DRB1,and4,9,10,12,and15inDLA-DQB1hadthelargestimpactonthesupertypeclustering(PC1vs.
PC2inAdditionalfile5:FiguresS5–S6).
Atthesesites,therewerefromtwotoseven(DLA-DRB1)andfromthreetofive(DLA-DQB1)differentaminoacids–asimilarlevelofdiversitycomparedtootherPBRsites.
MHCgenotypeproportionsintheoffspringAtthelocuslevel,deviationfromtheMendelianexpec-tationswasnotobservedinthegenotypeproportionsofthedogfamilies.
Whenexaminedforallofthefamiliesanddifferentmatingcrosstypes,χ2variedbetween0.
02and2.
81(df=1)andp-valuevariedbetween1and0.
13inallMHCloci(Fig.
3andAdditionalfile6:TableS3).
Accordingly,thecombinedchi-squaredanalysesovermatingcrosstypeswithineachlocusdidnotshowanydeviationfromtheexpectedheterozygosityandthecombinedχ2variedbetween2.
28and3.
09(df=1)andp-valuevariedbetween0.
69and0.
54.
AlloftheMHClocifollowedtheMendelianinheritancealsoindifferentbreeds,inpurebreds,inmixed-breeddogs,inbothsexesandinfamilieswithorwithouttheduplicatedDLA-DQB1locus,whentheywereanalysedseparately.
Nodeviationfromtheexpectedheterozygositywasdetectedeitheratthethree-locushaplotypelevel(Additionalfile6:TableS3)orusingsupertypeclassification(Fig.
4).
ThegenotypeproportionsintheoffspringwerealsostudiedineachpolymorphicPBRsite.
Figure5showsbothp-andq-valuesforthecombinedchi-squaredstatisticsoverthedifferentmatingcrosstypesinsite-wisesegregationanalyses.
Aftercorrectingformultipletesting,thePBRsites10–12inDLA-DRB1showedaslightdeviationfromtheexpectedheterozygosity,withsite12havingthesmallestq-value(q=0.
15).
ThePBRNiskanenetal.
CanineGeneticsandEpidemiology(2016)3:9Page5of12Table2Three-locusMHChaplotypesandtheirfrequenciesinthestudieddogbreedsandmixed-breedgroupsDRB1DQA1DQB1AH1AH2CB1CB2ECS2GR2ISDKKHKLLWNF2RRB2SASBT21030100101008020.
280010100101002010.
150.
240.
220.
080.
300.
020.
350010100101036010.
120.
020010100101036030.
120010300101002010.
060020100901001010.
250.
1700601005011007010.
170.
250.
011.
000.
080.
180.
0500601005011020010.
120.
3300901001010080110.
320.
420.
030.
470.
0201001100201015010.
060110100201013030.
040110200201013030.
080120100401013017a0.
330.
510.
190.
010120100401013030.
420130100101002010.
190.
040.
250.
460130100601020020.
070150100901001010.
190150100601023010.
240.
310.
190.
230.
030.
040150100601057010.
030150100601003010.
120150100601022010.
300150100601020020.
070150100401013017a0.
0301501012011035010.
010150100601011010.
180.
090150200601023010.
220.
030150300601003010.
080150300601023040.
070180100101008020.
030.
240.
140.
070200100401013030.
270.
090.
040.
010.
420230100301005010.
230.
180300100601003010.
030400101001019010.
170700101801050010.
0607401005011007010.
160.
020740100101002010.
010760100601023010.
250780100401013030.
080800100402013030.
0109401001010080110.
180950100301054010.
05Niskanenetal.
CanineGeneticsandEpidemiology(2016)3:9Page6of12sites10,11and12correspondtocodonpositions57,60and67inthedata.
MoredetailedexaminationofthePBRsite12revealedexcessofheterozygousoff-springfromthematingcrosstypeA1A2*A1A1(χ2=9.
48,df=1,p=0.
004).
Thesitewaspolymorphicforthreeaminoacids:leucine,isoleucineandphenyla-lanine.
Allthreeaminoacidsarehydrophobic,butphenylalaninediffersfromtheothertwobyhavinganaromaticsidechain.
AlsothePBRsite10showedexcessofheterozygosity(A1A2*A1A1,χ2=4.
46,df=1,p=0.
04)butthePBRsite11showedexcessofhomozygosity(A1A2*A1A2,χ2=4.
95,df=1,p=0.
04).
InDLA–DQB1,noPBRsitesshoweddeviationfromtheexpectedhete-rozygosity(Fig.
5).
MaternaleffectTostudytheputativeprenatalmaternaleffect,alleletransmissionwasstudiedinthefamiliesinwhichparentssharedonealleleandtheoffspringcouldinheritthematernalsharedorpaternalnon-sharedallelefromtheirfather.
Thepuppiestendedtoinheritthematernalsharedallele(A1)fromtheirfatherforDLA-DRB1inmatingcrosstypeA1A2*A1A3moreoftenthanexpected(χ2=3.
02,df=1,p=0.
09;Fig.
6andAdditionalfile6:TableS3).
Thistypeofmatingincludedthemajorityofthefamiliesinthisanalysis.
However,nostrongdeviationsfromtheexpectedalleleproportionswerefoundeitherwhenusingthealleleorsupertypeclassi-ficationintheanalyses.
Thematernalsharedandnon-sharedallelewereequallyofteninheritedfromthefatheralsoindifferentbreeds,inbothsexesandinfamilieswithorwithouttheduplicatedDLA-DQB1locus.
Whentheheterozygousmotherswerecombinedforsegregationanalyses,mother'sheterozygositydidnotaffecttheoffspringheterozygosityneitherattheallelenoratthesupertypelevel(Additionalfile6:TableS3).
DiscussionWehavestudiedprenatalselectionindogsbyanalysingthesegregationoftheMHCclassIIallelesinpurebredandmixed-breedfamilies.
NaturalselectiontargetingtheMHCmaybelimitedinthepetdogsduetotheirprotectedenvironment.
Onepotentialphaseofnaturalselectionisduringfoetaldevelopmentwhichislessstronglyaffectedbyhumanactionorbreedingchoices.
However,wedidnotdetectdeviationsfromtheMendelianTable2Three-locusMHChaplotypesandtheirfrequenciesinthestudieddogbreedsandmixed-breedgroups(Continued)09501001010080110.
020950100901001010.
020970100601020020.
040980100402023010.
01NA0.
020.
080.
120.
01n3964911681313371381766528313AH1AlaskanHusky,AH2AlaskanHusky,CB1Mixed-breed,CB2Mixed-breed,ECS2EnglishCockerSpaniel,GR2GoldenRetriever,ISDIcelandicSheepdog,KKHKooikerhondje,KLKromfohrlander,LWLowchen,NF2Newfoundland,RRB2RhodesianRidgeback,SASaluki,SBT2StaffordshireBullTerriern=numberofindividuals(alllitter-parentscombinations)aDuplicatedDLA-DQB1locus,allelesDLA-DQB1*01303andDLA-DQB1*017011Rowan[35],2Kennedyetal.
[34]Fig.
1SupertypeclusteringtreeforthecanineDLA-DRB1alleles.
DashedlineindicatesdivisionintothefivesupertypeclassesFig.
2SupertypeclusteringtreeforthecanineDLA-DQB1alleles.
DashedlineindicatesdivisionintotheninesupertypeclassesNiskanenetal.
CanineGeneticsandEpidemiology(2016)3:9Page7of12segregationpatternatallele,three-locushaplotype,orsupertypelevels.
Atleastaround200DLA-DRB1,100DLA-DQB1and40DLA-DQA1alleleshavebeenfoundincanines.
Obviouslythereissomefunctionaloverlapinsuchalargenumberofalleles.
Inhumans,ithasbeenshownthattheMHCclassIIalleleshavemoresharedpeptidebindingrepertoirethantheMHCclassIalleles[42].
ClusteringtheMHCallelesintosupertypesisonewaytoreachameaningfulleveloffunctionalityandvariationforstudyingthenaturalselectionintheMHC.
WehavegeneratedthefirstcaninesupertypeclassificationfortheDLA-DRB1andDLA-DQB1alleles.
TheDLA-DQA1alleleswerenotincluded,sincetheywerenotpolymorphicontheputativePBRsitesproposedbyDoytchinovaandFlower[37].
Thismighthaveonlyminoreffecttoourresults,sinceinhumansithasbeenshownthattheHLA-DQA1allelesdidnotcontributetothesupertypeclusteringwhentheywereanalysedtogetherwiththeHLA-DQB1allelestocomposeasinglefunc-tionalunit[37].
Inthisstudy,thenumberofthesupertypeclusters(fiveandnine)wasrelativelylow,andbasedoninsilicoanalysisofamino-acids.
Functionalstudies,wherethepeptidebindingrepertoiresofthedifferentMHCallelescouldbetested,wouldgreatlybenefittheextractionofabiologicallymeaningfulnumberofsuper-typeclusters[31,42].
Thesupertype-clusteredMHCallelescanbeusedinselectionanalysesinsteadofthesequence-basedalleles.
However,wedidnotfindpatternssuggestingprenatalFig.
3Theobserved(Ho)andexpected(He)numbersoftheheterozygousoffspringfromdifferentmatingcrosstypesfortheDLA-DRB1,DLA-DQA1andDLA-DQB1loci(n=totalnumberofoffspringwithinthecrosstype)Fig.
4Theobserved(Ho)andexpected(He)numbersoftheheterozygousoffspringfromdifferentmatingcrosstypesfortheDLA-DRB1andDLA-DQB1supertypes(n=totalnumberofoffspringwithinthecrosstype)Niskanenetal.
CanineGeneticsandEpidemiology(2016)3:9Page8of12selectionatsupertypelevelinthestudieddogfamilies.
Previously,naturalselectionhasbeendetectedatthesupertypelevele.
g.
inthehighermaleheterozygosityofnewbornhumans[43]andinthenon-randomasso-ciationofthealleleswithinDRB-haplotypesofbaboons[31].
Parasiteanddiseaseresistancehasalsobeenasso-ciatedwithMHCsupertypelevelclassesinvariousspecies[30,32].
Thesefindingssuggestthatsupertypeclassificationreachesaleveloffunctionalitythatmaybemissedusingindividualallelesintheselectionanalyses.
Inhumans,somesmallisolatedpopulationshaveshownnon-MendelianproportionsofMHCgenotypesinoffspring[17]orMHCassociatedfoetalloss[8].
However,ofteninstudiesoflargerhumanpopulationsnodeviationfromtheexpectedproportionshasbeenfoundintheMHCloci[1,44].
Thelackofselectioninthemodernlargehumanpopulationsmightbeaconse-quenceofimprovedhealthcare[45]whichalsoappliestodogsbutislikelytomostlyaffectselectivefactorsafterbirthandtolessaffectmaternal-foetalinteractions.
Inthisstudy,thepuppiesweresampledinvaryingagesandtheageofthedogwasnotknowninallcases.
Thus,itispossiblethatasmalldegreeofpostnatalselec-tionhasoccurred.
Wefoundthatthedogpuppiesweremorehetero-zygousthanexpectedinrespecttothePBR-site12ofDLA-DRB1forthematingcrosstypeA1A2*A1A1.
Whenhalfoftheoffspringareexpectedtobehetero-zygousandhalfhomozygous,theselectioncoefficient(s)canbeestimatedwith:s1Nii=Nij1whereNiiistheobservednumberofthehomozygousoffspringandNijistheobservednumberoftheheterozygousoffspring[46].
TheobservedexcessoftheheterozygousindividualsatthePBR-site12ofDLA-DRB1wouldrequirestrongselectionagainsthomozygotes:s=1–38/(70+1)=0.
465.
Selectionofsimilarstrength,s=0.
462overtwoMHCloci,hasbeenestimatedtofavorheterozygotesinhumanpopulations[17].
However,becausetheMHCexpressionincaninefertilisationandpregnancyisnotwellknown,furtherstudiesarerequiredtoinvestigatetheexactcauseoftheobserveddeviation,Fig.
5Thep-andq-valuesforthecombinedchi-squaredstatisticstestingtheratiooftheobservedandexpectedheterozygousoffspringforeachPBRsiteintheDLA-DRB1andDLA-DQB1lociFig.
6Theobserved(Obs.
)andexpected(Exp.
)numbersofoffspringthatinheritedthematernalsharedallelefromtheirfatherforDLA-DRB1,DLA-DQA1andDLA-DQB1lociNiskanenetal.
CanineGeneticsandEpidemiology(2016)3:9Page9of12beforereachingfurtherconclusionsabouttheselectiveadvantageofheterozygosityonthePBR-site12.
TheaminoacidsatthePBR-site12weresharedbydifferentallelesandsupertypes,whichexplainswhywedidnotdetectthesignalofthisdeviationatthehigherlevelsofanalyses.
Ourdataconsistedoftendogbreeds,thathavebeenbredfore.
g.
hunting,herding,andcompanionpurposes,andmixed-breedgroups.
Thesamplingofnumerousbreedshadtheadvantageofrepresentingthespeciesmorewidely,butasadownside,italsocausedindividualallelestohavelowfrequenciesinthehighlypolymorphicMHCloci.
Togainstatisticalpower,weclassifiedthematingcrosstypesbytheallelestheparentssharedwithoutkeepingtheinformationofalleleidentity.
Simul-taneously,wecompromisedtheputativeeffectsofindi-vidualalleles.
Weestimatedthestatisticalpoweroftheχ2testtodistinguishbetweentheexpectedandobservednumbersofpuppiesatthep=0.
05level.
UsingthesamplesizesshowninFig.
3foreachmatingcrosstypeandlocus,sshouldapproximatelybeabove0.
3–0.
6tobedetected.
Valuesthishighindicateverystrongselec-tionandarerarelyobservedinnaturalpopulations.
Itispossiblethattherearebreed-specificdifferences,whichwewerenotabletorecogniseduetothesmallbreed-wisesamplesize,particularlywhenthesamplewasfurtherdividedintomatingcrosstypes.
Asshowninearlierstudies[17],smallsamplesizecanhinderusfromdrawingfinalconclusionsaboutthedifferencesinthepost-copulatoryselectionindogbreeds,especiallyifselectionisnotverystrong.
DuetothehighLDwithintheMHCregion,MHCclassIIallelescouldshowtheeffectofprenatalselectioneveniftheclassIIallelesarenotexpressedintheplacenta.
OtherprenatallyselectedlocithatareinLDwiththeclassicalMHClocimaycontributetothemaintenanceofthehighpolymorphismintheextendedMHCregion[47].
Simultaneously,thismayleadtoapparentsegre-gationdistortionalsointheclassicalMHCgenes.
Forexample,Laurentetal.
[48]detectedasignatureofdisas-sortativematingusingsinglenucleotidepolymorphisms(SNPs)intheextendedhumanMHCregion(3.
6Mb),andconcludedthatitwascausedbymultipleMHCrelatedgenesratherthantheclassicalMHCgenes.
Matechoiceisseldompossibleindogs,butpost-copulatoryselectionmay,inasimilarway,becausedbyseveralMHC-linkedgenes.
TheORgenesmayinfluencebothmateselectionandthesuccessofspermatozoainthefertilisation[47].
Previouslyobservednon-MendeliansegregationofMHCallelesinprimatesmaybecausedbyselectiontargetedtoMHC-linkedORgenes.
Ifthatwasthecase,themissingselectionintheMHCclassIIlociofourdogfamiliesiseasilyexplainedbythefactthatnoORgeneshavethisfarbeenlocalisedinthemajor3MbMHCregiononthedogchromosome12orelsewhereonchromosome12[33,49].
DogspermatozoadoexpressORgenes[50],butputativelyMHC-linkedORgenesareonlyfoundintheminor600kbMHCclassIregionlocatedondogchromosome35[33].
ConclusionsInsummary,wefoundsomeevidenceofpost-copulatoryselectioninpetdogsatnucleotidesitelevel,butnotatlocus,three-locus,orsupertypelevelsinMHCloci.
Itispossiblethatthereisweakerselectionaffectingothersitesandlevels,butoursample,whichonlyspannedonegeneration,wasnotlargeenoughtopickupthesignal[1].
SinceselectionhasmodifiedtheMHCdiversityoveralongtimeperiod,evenweakselectionmayhavebeensufficienttopreservethediversity.
Toourknowledge,thiswasthefirststudyofprenatalselectionindogMHClociandwehavesomesugges-tionshowtoproceedwithfuturestudies.
InfurtherexaminationsofMHCrelatedprenatalselectionincanines,acomprehensivestudyincludingmatechoice,segregationofallelesandlittersizewouldbeinteresting.
Futurestudieswouldalsobenefitfromusingdogsfrommoreprimitivecircumstancesand,forcomparison,theirwildwolfrelatives.
AdditionalfilesAdditionalfile1:TableS1.
MHCgenotypedataandsampleinformationforthedogfamilies.
(XLSX53kb)Additionalfile2:FiguresS1andS2.
MergingcostshowingtheamountofvariancewithinclustersinWard'smethod.
ThisinformationwasusedindecidingthenumberofsupertypeclustersinDLA-DRB1(FigureS1)andDLA-DQB1(FigureS2)loci.
(ZIP3kb)Additionalfile3:FiguresS3andS4.
SupertypeclusteringtreesforthecanineDLA-DRB1(FigureS3)andDLA-DQB1(FigureS4)lociwithallelenamesincluded.
Thesupertypeclustersarecoloureddifferently.
(ZIP12kb)Additionalfile4:TableS2.
Supertypeclassificationalleledataandallallelesinthedogfamilydata.
(XLSX25kb)Additionalfile5:FiguresS5andS6.
BiplotsofthefirsttwoPCsinsupertypeclustersofDLA-DRB1(FigureS5)andDLA-DQB1loci(FigureS6).
ThebiplotsweredoneusingtheggbiplotfunctioninR.
EllipsesshowNormalcontourlineswith68%probabilityforeachcluster.
(ZIP20kb)Additionalfile6:TableS3.
ResulttableforsegregationofheterozygousandhomozygousoffspringindifferentmatingcrosstypesfortheMHCloci,thethree-locushaplotypeandtheMHCsupertypelevels,heterozygousmotherscombinedandthenumbersofoffspringthatinheritedmaternalsharedorpaternalnon-sharedallelefromtheirfather.
(XLSX12kb)Abbreviationsdf:degreeoffreedom;DLA:Dogleucocyteantigen;DNA:Deoxyribonucleicacid;EVT:Extravilloustrophoblastcells;FIS:Inbreedingcoefficient;He:Expectednumberofheterozygotes;HLA:Humanleucocyteantigen;Ho:Observednumberofheterozygotes;LD:Linkagedisequilibrium;Niskanenetal.
CanineGeneticsandEpidemiology(2016)3:9Page10of12MHC:Majorhistocompatibilitycomplex;NF:Numberoffamilies;NK:Naturalkillercells;NO:Numberofoffspring;OR:Olfactoryreceptor;PBR:Peptide-bindingregion;PC(A):Principalcomponent(analysis);s:selectioncoefficientAcknowledgementsWearegratefultoIngeridHagenandtwoanonymousreviewersforprovidingvaluablecommentsontheearlierversionofthismanuscript.
WethankRanjaEklundforcollectingsalukidata,breedclubsforcollaborationincollectingdogsamplesinFinland,andGeorgeHappfromtheUniversityofAlaskaatFairbanks,Alaska,forhuskysamples.
WethankFrancesJuryfortypingandpartialanalysesoftheMHCmicrosatelliteresultsforthehuskyfamiliesinManchester.
AvailabilityofdataandmaterialThedatasetssupportingtheconclusionsofthisarticleareincludedwithinthearticleanditsadditionalfiles.
ThenewDLA-DRB1allelesequenceshavebeensubmittedtoNCBIGenBankwithaccessionnumbers:JQ904819.
1,FM246834,andFM246838.
Authors'contributionsAKNconductedpartofthelaboratorywork.
AKN,LJKandHLcollectedthedata.
AKN,LJKandTPimplementedthedataanalyses.
AKN,TPandJAwrotethemanuscriptwitheditorialinputfromallauthors.
Allauthorsapprovedthefinalmanuscript.
CompetinginterestsLornaKennedyisManagingEditoronthisjournal,andwasexcludedfromtheEditorialprocess.
Otherauthorsdeclarenocompetinginterests.
FundingThisstudywaspartiallysupportedbytheFinnishPopulationGeneticsGraduateProgramme(AKN),KoneFoundation(AKN),theAcademyofFinland(HL:127065),theSigridJuseliusFoundation(HL)andtheJaneandAatosErkkoFoundation(HL).
ConsentforpublicationNotapplicable.
EthicsapprovalNodogDNAsampleswerecollectedspecificallyforthisstudy.
Authordetails1DepartmentofGeneticsandPhysiology,UniversityofOulu,POBox3000,OuluFIN-90014,Finland.
2Presentaddress:CentreforBiodiversityDynamics,DepartmentofBiology,NorwegianUniversityofScienceandTechnology,NO-7491,Trondheim,Norway.
3CentreforIntegratedGenomicMedicalResearch,UniversityofManchester,StopfordBuilding,OxfordRoad,ManchesterM139PT,UK.
4DepartmentofVeterinaryBiosciences,ResearchProgramsUnit,MolecularNeurology,UniversityofHelsinkiandFolkhlsanInstituteofGenetics,BiomedicumHelsinki,POBox63,FIN-00014Helsinki,Finland.
Received:28January2016Accepted:2October2016References1.
GarriganD,HedrickPW.
Perspective:DetectingAdaptiveMolecularPolymorphism:LessonsFromtheMHC.
Evolution(NY).
2003;57:1707–22.
2.
BenacerrafB.
RoleofMHCgeneproductsinimmuneregulation.
Science(80-).
1981;212:1229–38.
3.
EhlersA,BeckS,ForbesSA,TrowsdaleJ,VolzA,YoungerR,ZieglerA.
MHC-LinkedOlfactoryReceptorLociExhibitPolymorphismandContributetoExtendedHLA/OR-Haplotypes.
GenomeRes.
2000;10:1968–78.
4.
XieT,RowenL,AguadoB,AhearnME,MadanA,QinS,CampbellRD,HoodL.
AnalysisoftheGene-DenseMajorHistocompatibilityComplexClassIIIRegionandItsComparisontoMouse.
GenomeRes.
2003;13:2621–36.
5.
ClarkeB,KirbyDRS.
MaintenanceofHistocompatibilityPolymorphisms.
Nature.
1966;211:999–1000.
6.
DohertyPC,ZinkernagelRM.
EnhancedimmunologicalsurveillanceinmiceheterozygousattheH-2genecomplex.
Nature.
1975;256:50–2.
7.
ApaniusV,PennD,SlevPR,RuffLR,PottsW.
Thenatureofselectiononthemajorhistocompatibilitycomplex.
CritRevImmunol.
1997;17:179–224.
8.
OberC.
StudiesofHLA,fertilityandmatechoiceinahumanisolate.
HumReprodUpdate.
1999;5:103–7.
9.
PennDJ.
Thescentofgeneticcompatibility:sexualselectionandtheMajorHistocompatibilityComplex.
Ethology.
2002;108:1–21.
10.
FernandoMMA,StevensCR,WalshEC,DeJagerPL,GoyetteP,PlengeRM,VyseTJ,RiouxJD.
DefiningtheRoleoftheMHCinAutoimmunity:AReviewandPooledAnalysis.
PLoSGenet.
2008;4:e1000024.
11.
JokinenP.
Identifyinggeneticriskfactorsincanineautoimmunedisorders.
Helsinki:UniversityofHelsinki;2011.
12.
KennedyLJ,BarnesA,ShortA,BrownJJ,LesterS,SeddonJ,FleemanL,FrancinoO,BrkljacicM,KnyazevS,HappGM,OllierWER.
CanineDLAdiversity:1.
Newallelesandhaplotypes.
TissueAntigens.
2007;69:272–88.
13.
NiskanenAK,KennedyLJ,RuokonenM,KojolaI,LohiH,IsomursuM,JanssonE,PyhjrviT,AspiJ.
BalancingselectionandheterozygoteadvantageinmajorhistocompatibilitycomplexlociofthebottleneckedFinnishwolfpopulation.
MolEcol.
2014;23:875–89.
14.
KennedyLJ,AnglesJM,BarnesA,CarmichaelLE,RadfordAD,OllierWER,HappGM.
DLA-DRB1,DQA1,andDQB1allelesandhaplotypesinNorthAmericangraywolves.
JHered.
2007;98:491–9.
15.
NiskanenAK,HagstrmE,LohiH,RuokonenM,Esparza-SalasR,AspiJ,SavolainenP.
MHCvariabilitysupportsdogdomesticationfromalargenumberofwolves:highdiversityinAsia.
Heredity(Edinb).
2013;110:80–85.
16.
WedekindC,ChapuisatM,MacasE,RülickeT.
Non-randomfertilizationinmicecorrelateswiththeMHCandsomethingelse.
Heredity(Edinb).
1996;77:400–9.
17.
BlackF,HedrickP.
StrongbalancingselectionatHLAloci:evidencefromsegregationinSouthAmerindianfamilies.
ProcNatlAcadSciUSA.
1997;94(November):12452–6.
18.
KnappLA,HaJC,SackettGP.
ParentalMHCantigensharingandpregnancywastageincaptivepigtailedmacaques.
JReprodImmunol.
1996;32:73–88.
19.
BillingtonWD.
InfluenceofImmunologicalDissimilarityofMotherandFoetusonSizeofPlacentainMice.
Nature.
1964;202:317–8.
20.
WarningJC,McCrackenSA,MorrisJM.
Abalancingact:mechanismsbywhichthefetusavoidsrejectionbythematernalimmunesystem.
Reproduction.
2011;141:715–24.
21.
KovatsS,MainEK,LibrachC,StubblebineM,FisherSJ,DeMarsR.
AclassIantigen,HLA-G,expressedinhumantrophoblast.
Science(80-).
1990;248:220–3.
22.
KingA,BoocockC,SharkeyAM,GardnerL,BerettaA,SiccardiAG,LokeYW.
EvidencefortheexpressionofHLA-CclassImRNAandproteinbyhumanfirsttrimestertrophoblast.
JImmunol.
1996;156:2068–76.
23.
TrowsdaleJ,BetzAG.
Mother'slittlehelpers:mechanismsofmaternal-fetaltolerance.
NatImmunol.
2006;7:241–6.
24.
ParhamP,MoffettA.
VariableNKcellreceptorsandtheirMHCclassIligandsinimmunity,reproductionandhumanevolution.
NatRevImmunol.
2013;13:133–44.
25.
ChazaraO,XiongS,MoffettA.
MaternalKIRandfetalHLA-C:afinebalance.
JLeukocBiol.
2011;90:703–16.
26.
MoffettA,LokeC.
Immunologyofplacentationineutherianmammals.
NatRevImmunol.
2006;6:584–94.
27.
Schfer-SomiS,BeceriklisoyHB,BudikS,KancaH,AksoyOA,PolatB,CetinY,AySS,AslanS.
Expressionofgenesinthecaninepre-implantationuterusandembryo:Implicationsforanactiveroleoftheembryobeforeandduringinvasion.
ReprodDomestAnim.
2008;43:656–63.
28.
Schfer-SomiS,BeceriklisoyH,WalterI,SabitzerS,KleinD,KancaH,KayaD,AgaogluA,zgürH,AslanS.
ExpressionofMHC-Iand-IIinUterineTissuefromEarlyPregnantBitches.
ReprodDomestAnim.
2009;44:103–8.
29.
SetteA,NewmanM,LivingstonB,McKinneyD,SidneyJ,IshiokaG,TangriS,AlexanderJ,FikesJ,ChesnutR.
Optimizingvaccinedesignforcellularprocessing,MHCbindingandTCRrecognition.
TissueAntigens.
2002;59:443–51.
30.
SchwensowN,FietzJ,DausmannKH,SommerS.
Neutralversusadaptivegeneticvariationinparasiteresistance:importanceofmajorhistocompatibilitycomplexsupertypesinafree-rangingprimate.
Heredity(Edinb).
2007;99:265–77.
31.
HuchardE,WeillM,CowlishawG,RaymondM,KnappLA.
Polymorphism,haplotypecomposition,andselectionintheMhc-DRBofwildbaboons.
Immunogenetics.
2008;60:585–98.
Niskanenetal.
CanineGeneticsandEpidemiology(2016)3:9Page11of1232.
SepilI,LachishS,HinksAE,SheldonBC.
Mhcsupertypesconferbothqualitativeandquantitativeresistancetoavianmalariainfectionsinawildbirdpopulation.
ProcRSocB.
2013;280:20130134.
33.
YuhkiN,BeckT,StephensR,NeelamB,O'BrienSJ.
Comparativegenomicstructureofhuman,dog,andcatMHC:HLA,DLA,andFLA.
JHered.
2007;98:390–9.
34.
KennedyLJ,BarnesA,ShortA,BrownJJ,LesterS,SeddonJ,HappGM,OllierWER.
CanineDLAdiversity:2.
Familystudies.
TissueAntigens.
2007;69Suppl1:289–91.
35.
RowanD.
AtimeandcosteffectivemethodforDogLeukocyteAntigen(DLA)typingusingcaninemicrosatellitemarkers.
Manchester:UniversityofManchester;2007.
36.
DebenhamSL,HartEA,AshurstJL,HoweKL,QuailMA,OllierWER,BinnsMM.
GenomicsequenceoftheclassIIregionofthecanineMHC:comparisonwiththeMHCofothermammalianspecies.
Genomics.
2005;85:48–59.
37.
DoytchinovaIA,FlowerDR.
InsilicoidentificationofsupertypesforclassIIMHCs.
JImmunol.
2005;174:7085–95.
38.
SandbergM,ErikssonL,JonssonJ,SjstrmM,WoldS.
Newchemicaldescriptorsrelevantforthedesignofbiologicallyactivepeptides.
Amultivariatecharacterizationof87aminoacids.
JMedChem.
1998;41:2481–91.
39.
WardJH.
Hierarchicalgroupingtooptimizeanobjectivefunction.
JAmStatAssoc.
1963;58:236–44.
40.
Rcoreteam.
R:Alanguageandenvironmentforstatisticalcomputing.
2014.
41.
StoreyJD,TibshiraniR.
Statisticalsignificanceforgenomewidestudies.
ProcNatlAcadSciUSA.
2003;100:9440–5.
42.
GreenbaumJ,SidneyJ,ChungJ,BranderC,PetersB,SetteA.
FunctionalclassificationofclassIIhumanleukocyteantigen(HLA)moleculesrevealssevendifferentsupertypesandasurprisingdegreeofrepertoiresharingacrosssupertypes.
Immunogenetics.
2011;63:325–35.
43.
DorakMT,LawsonT,MachullaHK,MillsKI,BurnettAK.
IncreasedheterozygosityforMHCclassIIlineagesinnewbornmales.
GenesImmun.
2002;3:263–9.
44.
KlitzW,LoSK,NeugebauerM,BaurMP,AlbertED,ThomsonG.
AcomprehensivesearchforsegregationdistortioninHLA.
HumImmunol.
1987;18:163–80.
45.
MarkowT,HedrickPW,ZuerleinK,DanilovsJ,MartinJ,VyvialT,ArmstrongC.
HLApolymorphismintheHavasupai:evidenceforbalancingselection.
AmJHumGenet.
1993;53:943–52.
46.
HaldaneJBS.
Theestimationofviabilities.
JGenet.
1956;54:294–6.
47.
ZieglerA,SantosPSC,KellermannT,Uchanska-ZieglerB.
Self/nonselfperception,reproductionandtheextendedMHC.
SelfNonself.
2010;1:176–91.
48.
LaurentR,ToupanceB,ChaixR.
Non-randommatechoiceinhumans:insightsfromagenomescan.
MolEcol.
2012;21:587–96.
49.
QuignonP,GiraudM,RimbaultM,LavigneP,TacherS,MorinE,RetoutE,ValinA-S,Lindblad-TohK,NicolasJ,GalibertF.
Thedogandratolfactoryreceptorrepertoires.
GenomeBiol.
2005;6:R83.
50.
VanderhaeghenP,SchurmansS,VassartG,ParmentierM.
Specificrepertoireofolfactoryreceptorgenesinthemalegermcellsofseveralmammalianspecies.
Genomics.
1997;39:239–46.
Weacceptpre-submissioninquiriesOurselectortoolhelpsyoutondthemostrelevantjournalWeprovideroundtheclockcustomersupportConvenientonlinesubmissionThoroughpeerreviewInclusioninPubMedandallmajorindexingservicesMaximumvisibilityforyourresearchSubmityourmanuscriptatwww.
biomedcentral.
com/submitSubmityournextmanuscripttoBioMedCentralandwewillhelpyouateverystep:Niskanenetal.
CanineGeneticsandEpidemiology(2016)3:9Page12of12
Niskanen1,2*,LornaJ.
Kennedy3,HannesLohi4,JouniAspi1andTanjaPyhjrvi1AbstractBackground:Despitedecadesofstudying,themechanismsmaintaininghighdiversityinthegenesoftheMajorHistocompatibilityComplex(MHC)arestillpuzzlingscientists.
Inadditiontopathogenrecognitionandotherfunctions,MHCmoleculesmayactprenatallyinmatechoiceandinmaternal-foetalinteractions.
TheseinteractionsarepotentialselectivemechanismsthatincreasegeneticdiversityintheMHC.
Duringpregnancy,immuneresponsehasadualrole:thefoetusrepresentsforeigntissuecomparedtomother,buthisto-incompatibilityisrequiredforsuccessfulpregnancy.
WehavestudiedtheprenatalselectioninMHCclassIIloci(DLA-DQA1,DLA-DQB1andDLA-DRB1)indomesticdogsbycomparingtheobservedandexpectedoffspringgenotypeproportionsin110dogfamilies.
Severalpotentialselectiontargetswereaddressed,includingthepeptide-bindingsite,theMHClocus,three-locushaplotypeandsupertypelevels.
Forthesupertypeanalysis,thefirstcaninesupertypeclassificationwascreatedbasedoninsilicoanalysisofpeptide-bindingamino-acidpolymorphism.
Results:Inmostlociandlevels,nodeviationfromtheexpectedgenotypefrequencieswasobserved.
However,onepeptide-bindingsiteinDLA-DRB1hadanexcessofheterozygotesamongtheoffspring.
Inaddition,ifthefathersharedaDLA-DRB1allelewiththemother,thatallelewasinheritedbytheoffspringmorefrequentlythanexpected,suggestingtheselectiveadvantageofahisto-compatiblefoetus,incontrasttoourexpectations.
Conclusions:Weconcludethatthereissomeevidenceofpost-copulatoryselectionatnucleotidesitelevelintheMHClociofpetdogs.
Butduetonoindicationofselectionatlocus,three-locus,orsupertypelevels,weestimatedthattheprenatalselectioncoefficientislessthan0.
3indomesticdogsandverylikelyotherfactorsaremoreimportantinmaintainingthegeneticdiversityinMHCloci.
Keywords:Canisfamiliaris,MHC,Prenatalselection,SupertypeclassificationPlainEnglishsummaryMajorHistocompatibilityComplex(MHC)isalargegeneticregioncodingforimmunedefensemoleculesanditisfoundinallvertebrates.
MHCmoleculestriggerimmuneresponsebybindingmoleculesderivedfrompathogensandrecogniseself-tissuesfromnon-self-tissues.
TheMHCmoleculesareoftenverydiverse.
Thisdiversityismainlythoughttobeaconsequenceofbalancingselectioncausedbypathogenpressure.
However,alsoselectionbeforebirththatmaytakeplaceduringmatechoiceorduringpregnancyhavebeendemonstratedinsomespecies.
Sincedomesticdogsareoftenunderintenseveterinarycare,pathogenpressuremightbelessstrongcomparedtowildanimals.
Yet,selectivepatternsofMHCdiversitybeforebirthhavenotbeenstudiedbeforeindogs.
WestudiedtheselectionbeforebirthatthreeMHCclassIIgenesin10dogbreedsandfourmixed-breedgroups.
Wecomparede.
g.
theexpectedandobservedratiosofheterozygousandhomozygouspuppiesin110dogfamilies.
Weconductedselectionanalysesinmanydifferentlevels,including:I)sitesthatbindmoleculesderivedfrompathogens(=peptide-bindingsite),II)the*Correspondence:alina.
niskanen@gmail.
com1DepartmentofGeneticsandPhysiology,UniversityofOulu,POBox3000,OuluFIN-90014,Finland2Presentaddress:CentreforBiodiversityDynamics,DepartmentofBiology,NorwegianUniversityofScienceandTechnology,NO-7491,Trondheim,NorwayFulllistofauthorinformationisavailableattheendofthearticleTheAuthor(s).
2016OpenAccessThisarticleisdistributedunderthetermsoftheCreativeCommonsAttribution4.
0InternationalLicense(http://creativecommons.
org/licenses/by/4.
0/),whichpermitsunrestricteduse,distribution,andreproductioninanymedium,providedyougiveappropriatecredittotheoriginalauthor(s)andthesource,providealinktotheCreativeCommonslicense,andindicateifchangesweremade.
TheCreativeCommonsPublicDomainDedicationwaiver(http://creativecommons.
org/publicdomain/zero/1.
0/)appliestothedatamadeavailableinthisarticle,unlessotherwisestated.
Niskanenetal.
CanineGeneticsandEpidemiology(2016)3:9DOI10.
1186/s40575-016-0038-9MHCgenes,III)three-genecombinationsandIV)super-types.
WecreatedthefirstcaninesupertypeclassificationthatjoinsdifferentMHCallelesintolargersupertypesbasedontheirchemicalsimilarities.
Wefoundthatpuppiesweremoreheterozygousthanexpectedatonepeptide-bindingsiteofoneMHCgene(DLA-DRB1).
WealsofoundthatifthefathersharedaDLA-DRB1allelewiththemother,thatallelewasinhe-ritedbytheoffspringmorefrequentlythanexpected.
Otherlevelsofinvestigationdidnotshowdeviationsfromexpectedratiosofhomozygousandheterozygousoffspring.
Asaconclusion,wefoundsomeevidenceofselectionbeforebirthindogs.
However,otherfactorsarelikelycontributingmoretotheMHCdiversityindogs.
BackgroundNaturalselectioninthegenesoftheMajorHistocom-patibilityComplex(MHC)hasbeenstudiedextensivelyinwildanimalpopulations[1].
Acrossmanyspecies,theMHCgenesarethemostpolymorphicinthegenome.
ThemainfunctionofclassicalMHCmoleculesistobindantigenicpeptidesandpresentthemtoT-cells[2].
TheMHCgenesaredividedintotwoclasses;classImoleculesareexpressedonthesurfaceofalmostallnucleatedcells,while,inman,classIImoleculesareexpressedonspecificantigen-presentingcells(e.
g.
macrophagesandlymphocytes).
Inman,theextendedMHCregionalsoincludesnon-classicalgenesthathavediverseimmuneandnon-immunefunctions,includingthecomplementsystem,andolfactoryreceptor(OR)genes[3,4].
SincethefunctionofMHCmoleculesistriggeringimmuneresponse,parasitesanddiseaseshavebeenthoughttoplaythemajorroleinthemaintenanceofMHCdiversity[5,6].
However,indomesticatedanimalspathogenpressurehasprobablyhadasmallercontri-butiontotherecentselectiontargetingMHC,comparedtowildspecies.
Instead,thereareotherpotentialforcesdrivingselection,suchasmatechoice,spermselection,maternal-foetalinteractionsandautoimmunediseases[7–9].
ManyMHCassociationswithautoimmunediseaseshavebeenidentifiedinman[10].
SimilarMHCasso-ciationshavealsobeendemonstratedincanineauto-immunedisorders[11].
TheseassociationsmaybeoneoftheselectiveforcesinfluencingMHCdiversityamongdogs(Canisfamiliaris).
Thecloserelationshipbetweentheownerandtheirdoghasbeenakeydriverinacquiringextensiveknow-ledgeofcaninediseasesandresultedinahighlevelofveterinarycare,especiallyinWesternsocieties.
Theformationofdogbreedsandartificialselectionbyhumanshasbeenespeciallystrongduringthelast100–150years.
Asaconsequence,dogbreedshavebecomesmallandgeneticallydistinctgroups.
MHCdiversitydiffersfrombreedtobreedandisoftenlow[12],probablyduetodrifteffectsinsmallpopulations.
However,evenwithoutfreematechoice,MHCdiversityinadogbreedcanbeclosetothediversityofawildwolfpopulation[13,14].
HumanshavehadmajorimpactinsomeoftheaspectsthatmayinfluencethedogMHCdiversity,forinstance,throughveterinarycareandmatechoice.
Therefore,ifMHCin-compatibilitycausesprenatalselection,itmaybepossibletodetectsuchselectioninthedomesticdog,whereotherfactorsofnaturalselectionmaybelessstrong.
PrenatalselectionAlthoughtheoverallMHCdiversityofdogshasbeenstudiedwidely[12,15],theroleofprenatalselectioninthemaintenanceoftheMHCdiversityhasnotprevi-ouslybeenexamined.
Matechoiceindogsismoreoftendecidedbytheownerratherthanthedog,especiallyinthepurebreddogpopulation.
Hence,naturalselectionmayhaveitsstrongestimpactontheMHCdiversityatthepost-copulatorystagesfromeggfertilisationuntilbirth.
AssociationbetweenMHCvariationandfertili-sationsuccesshasbeendemonstratedinotherspecies,e.
g.
mice[16].
Prenatalselectioncanbedetectedfromnon-MendelianproportionsofMHCgenotypesintheoffspring.
Forinstance,theexcessofheterozygousoff-spring[17]andincreasedfoetallossamongMHC-matchingcouples[8]hasbeenfoundinhumansandinpigtailedmacaques[18].
Fromthemother'sperspective,afoetuscorrespondstosemiallograftwithhalfofitsgenomebeingforeign.
Tomaintaingestation,theimmuneresponseofthemotherisdown-regulated.
However,asuccessfulpreg-nancyrequiressomedegreeofmaternalimmunestimu-lation,whichcouldbeinducedbyhisto-incompatibility(immunologicaldissimilarity)betweenthemotherandthefoetus[19,20].
MHCmoleculeshavecrucialroleinsuccessfulpregnancy.
Inhumans,mostclassicalMHCmoleculesaredownregulatedandonlyoneclassicalMHCclassIgene(HLA-C)andtwononclassicalMHCclassIgenes(HLA-Eand-G)areexpressedinthefoetalsideoftheplacenta,intheextravilloustrophoblast(EVT)cells[21–23].
TheEVTcellsmigrateintothemother'suterusandtransformsmallarteriesintolargebloodvesselstoensuresufficientnourishmentofthefoetus[24].
Theprocessoftransformingthearteriesiscontrolledbythematernalnaturalkiller(NK)cells.
MostlikelyallthreeMHCmolecules,HLA-C,Eand-G,interactwithmaternalNKcellsinregulatingplacentalvascularisation[25].
HLA-CmoleculeistheonlypolymorphicMHCmoleculeexpressedintheEVTcells,sothesuccessofpregnancyisinfluencedbythecompatibilityofthefoetalHLA-CmoleculesandmaternalNKcellreceptors.
Incon-gruityofthefoetalMHCmoleculesandmaternalNKcellNiskanenetal.
CanineGeneticsandEpidemiology(2016)3:9Page2of12receptorsmightleadtoinsufficientbloodsupplytotheplacentawhichcancausepre-eclampsia,miscarriageandotherpregnancydisorders[26].
TheexpressionofMHCgenesduringpregnancyandmechanismsofmaternal-foetalinteractionsindogsarelargelyunknown.
IthasbeenshownthatsomeclassicalclassI(DLA-88)andII(DLA-DRA)MHCmoleculesarenotexpressedinearlycanineembryos[27]butthereisanincreaseofMHCclassIIexpressingcellsintheuterusofpregnantfemales[28].
MHCsupertypesThepeptiderecognitionrepertoiresofMHCmoleculesproducedbydifferentallelesatthesameMHClocusoverlapwitheachothertosomeextent.
Aspartoftheefforttodevelopepitope-basedvaccineswiththeabilitytoimmunisehumanpopulationswithdiverseMHCgenotypes[29],MHCalleleshavebeengroupedintosupertypes,basedontheirstructuralfeaturesorpeptide-bindingspecificity.
IftheactualdifferencesinpathogenrecognitionandpeptidebindingareattheMHCsuper-typelevel,theymightrepresenttheunitofnaturalselec-tionandshouldbetakenintoaccountwhenstudyingselectionintheMHCgenes.
Untilnow,supertypeclassi-ficationshavemainlybeenproducedforprimatesandbirds[30–32],butnonehavebeenattemptedforcanines.
AimsofthestudyTheaimofthisstudywastoinvestigatepossibleprenatalselectioninthreecanineMHCclassIIloci,DLA-DRB1,DLA-DQA1,andDLA-DQB1,byexaminingtheMHCgenotypeproportionsindogfamilies.
WealsoanalysedtheMHCallelecompositionandheterozygosityoftheoffspringrelativetotheirmothers,tostudypotentialmaternal-foetalinteractions.
Sincesuccessfulpregnancyrequireshisto-incompatibilitybetweenthemotherandthefoetus,wewouldexpecttofindanexcessofhete-rozygousoffspringundertheinfluenceofprenatalselec-tion.
Similarly,wewouldexpectthat,whenparentssharethesameallele,theoffspringwillinheritthepater-nalnon-sharedallelemoreoftenthansharedallelefromtheirfather.
Finally,wecomparedthefunctionalsimila-ritiesoftheidentifiedcaninealleles,andgeneratedthefirstsupertypeclassificationforDLA-DRB1andDLA-DQB1alleles.
MethodsSamplesandDNAtechniquesThedatacomprise430puppiesand220parentsin110familiesthatbelongtotenpurebredandfourdiffe-rentmixed-breedgroups(Table1andAdditionalfile1:TableS1).
Eachfamilyincludesparentsand1–16puppieswiththemeanof3.
9puppies.
Apartfromafewlittersthatwerefullysampled,arandomsampleoftheoffspringwasavailableformostfamilies.
Multiplelittersfromthesameparentswerecombinedasonelitter.
ThedogsweresampledinFinland,UKandUSA.
Thirty-sixindividualswereinthedatasetbothasaparentandasanoffspring.
Thirtyparentshadmorethanonelitterwithdifferentmates(24parentshadtwo,fourhadthree,andtwohadfourlitters).
ThehighlypolymorphicsecondexonthatencodesthepeptidebindingpartoftheMHCmoleculewassequencedforDLA-DRB1(270bp),DLA-DQA1(246bp)andDLA-DQB1(267bp)genes.
Thegenesarelocatedclosetogetheroncaninechromosome12intheabove-mentionedorder[33].
ThedistancebetweenDLA-DRB1andDLA-DQA1is56.
6kb,whilethedistancebetweenDLA-DQA1andDLA-DQB1is77.
7kb.
Apre-viouslyestablishedprotocol[13]wasusedtosequencethelociinacohortof253dogs.
Inaddition,datafrom41familieswith192puppiesfromanearlierstudybyKennedyetal.
[34]plusafurther12familieswith64puppiesfromRowan[35]wereincluded.
MicrosatellitegenotypingmethodforMHCgenotypingwasusedforonemixed-breedgroupofsixfamiliesincluding37offspring[35].
Inthismethod,dogsweregenotypedusingmicrosatellitesCFA12-3(dinucleotide),CFA12-4(tetranucleotide),andCFA12-9(dinucleotide),whicharelinkedwiththestudiedDLAloci[36].
Inaddition,paren-taldogsweregenotypedbysequencingtoconfirmtheMHCgenotypes.
Table1Thedogbreeds,thenumberofoffspring(NO),thenumberoffamilies(NF),thecountryofsamplingandtheMHCgenotypingmethodBreedNONFCountryaMHCgenotypingmethodKooikerhondjeb1710FinlandSequencingSalukib5116FinlandSequencingLowchenb94FinlandSequencingIcelandicSheepdogb73FinlandSequencingKromfohrlanderb9024FinlandSequencingAlaskanHuskyc276USASequencingAlaskanHuskyd41USASequencingEnglishCockerSpanield61USASequencingGoldenRetrieverd73USASequencingStaffordshireBullTerrierd73USASequencingNewfoundlandd4212UKSequencingRhodesianRidgebackd425UKSequencingMixed-breedd8416USASequencingMixed-breedc376USASequencingormicrosatellitebasedTotal430110aBirthcountrymaydifferfromthecountryofsampling,bDatafromthisstudy,cRowan[35],dKennedyetal.
[34]Niskanenetal.
CanineGeneticsandEpidemiology(2016)3:9Page3of12MHCalleleandhaplotypeassignmentMHCallelesweredeterminedfromthesequencedatausingMatchToolsandMatchToolsNavigator(AppliedBiosystems)programs.
MatchToolsincludesalibraryofcanineMHCallelesandcomparesahomozygousorheterozygoussequencepairtotheallelesequencesinthelibrary.
TwopuppiesfromthedatabyKennedyetal.
[34]hadarecombinantalleleinonelocusandthatlocuswasexcludedfromtheanalysesinthoseindividuals.
Usually,theMHCallelesfromdifferentlociareinhe-ritedashaplotypesduetostronglinkagedisequilibrium(LD,non-randomassociationofallelesbetweendifferentloci).
Three-locushaplotypesconsistingoftheDLA-DRB1,DLA-DQA1andDLA-DQB1allelesweredefinedinthreesteps:1)haplotypesweredeterminedforhomo-zygousindividuals,2)haplotypesforheterozygousindi-vidualswereinferredbasedonthehaplotypesobservedinthehomozygousindividualsand3)rarehaplotypes,detectedonlyintheheterozygousindividuals,wereiden-tifiedbysubtractionofhaplotypesdeterminedintheprevioussteps[14].
In35families,missingparentalhaplotypeswereinferredbasedonthehaplotypesoftheoffspringandtheotherparent.
TheDLA-DQA1allelesofthreeparentsandtheDLA-DQB1alleleofoneparentwereinferredbasedonthefamilymembers'alleles.
Inthemicrosatellite-basedgenotypingoftheDLAal-leles,themicrosatellitedatawereanalysedwithGenoty-per3.
7(AppliedBiosystems).
ThemicrosatellitealleleswerematchedwiththeDLAhaplotypesinthedogsthathadbeengenotypedbybothmethods.
AfteridentifyingthecorrespondingmicrosatelliteallelesandDLAhaplo-types,theDLAhaplotypeswereassignedtoindividualsthathadbeengenotypedonlywithmicrosatellites.
MHCsupertypeclassificationMHCsupertypeclassificationwasdevelopedtofindoutifprenatalselectionactsatthatlevel.
MHCallelesintheβloci,DLA-DRB1andDLA-DQB1,wereclassifiedintoMHCsupertypesbasedonthechemicalpropertiesoftheaminoacidsequencesintheputativepeptide-bindingregion(PBR).
TheclassificationwasdoneforofficiallynamedcanineMHCallelesfromtheMatch-Toolslibrarycomprisingallelesfromdifferentcaninespecies,mostlyfromdogandgreywolf(Canislupus).
Altogether205DLA-DRB1allelesand100DLA-DQB1alleleswereincluded.
Sincethesupertypeclassificationofallelesisbasedonthesharedpropertiesinpeptidebinding,theputativePBRsiteswereselectedaccordingtohumanPBRsitesproposedbyDoytchinovaandFlower[37].
Onlypoly-morphicPBRsiteswereincludedintheanalyses;19sitesinDLA-DRB1(codonpositions:9,11,13,26,28,30,38,47,56,57,60,61,67,70,71,74,78,86and90)and15sitesinDLA-DQB1(codonpositions:9,13,26,28,30,37,38,47,57,67,70,71,74,77and85).
Also37canineDLA-DQA1alleleswerecheckedforpolymorphismsinputativePBRsitesbutall10sitesweremonomorphicandthusthelocuswasexcludedfromthesupertypeclassification.
ChemicalpropertiesoftheaminoacidsinPBRsitesweredescribedusingfivez-descriptors:z1(hydrophobicity),z2(stericbulk),z3(polarity),z4andz5(electroniceffects)[38].
Z-descriptorvalueswerestandardisedsothattheirdistributionshadameanofzeroandastandarddeviationofone.
Allz-descriptor*polymorphicsitecombinationswereusedforestimatingEuclidiandistancesbetweenthedifferentMHCalleles.
BasedontheEuclidiandistances,thealleleswereclusteredusinghierarchicalclusteringwithWard'smethod[39]inRstatisticalenvironmentv.
3.
1.
2[40].
ThenumberofclusterswasdecidedbasedonthecostofmergingclustersinWard'smethod.
Thesmallestnumberofclusters,beforeasteepincreaseinthecostofmergingclusters(variancewithinclusters),waschosen.
TostudywhichPBRsiteshadmostinfluenceintheclusteringandtovisualisetherelationsofthesupertypeclusters,aprincipalcomponentanalysis(PCA)wasconductedusingtheprcompfunctioninR.
SegregationanalysesTostudywhethertheDLA-DRB1,DLA-DQA1andDLA-DQB1allelesfollowMendelianinheritanceindogfamilies,heterozygosityanalyseswereconductedinRstatisticalenvironmentv.
3.
1.
2[40].
Therearefourpossiblematingcrosstypeswheredifferencesinexpectedandobservednumbersofheterozygotescanbeobserved:A1A1*A1A2,A1A2*A1A1,A1A2*A1A2andA1A2*A1A3(where:mother*father,andtheallelesA1≠A2≠A3arecompositesofdifferentobservedalleles).
Inourdataset,altogether74(DLA-DRB1andDLA-DQB1)and83(DLA-DQA1)familiesincluding293(DLA-DRB1),313(DLA-DQA1)and295(DLA-DQB1)puppiesmetthiscriterion.
Thenumbersofheterozygousandhomozygousoffspringwerecountedwithineachmatingcrosstype.
TheexpectednumbersofheterozygousandhomozygousoffspringwerecalculatedbasedontheMendelianinheri-tancemodelandcomparedtoobservednumbersofoffspringusingPearson'schi-squaredtestwithMonteCarlosimulatedp-value(2000repeats).
Chi-squaredteststatisticswerealsocombinedoverthematingcrosstypesineachlocustoexaminetheoveralldeviationfromtheexpectednumberoftheheterozygotes.
Theanalyseswereconductedatlocusandthree-locushaplotypelevelsfortheentiredata,andseparatelyfordifferentbreeds/groupsandsexestoidentifythepoten-tialbreedorsexrelateddifferencesintheinheritancepatterns.
Informationaboutsexwasavailablefor381offspring.
Inaddition,purebredandmixed-breeddogswereanalysedseparately.
SegregationanalyseswerealsoNiskanenetal.
CanineGeneticsandEpidemiology(2016)3:9Page4of12doneforthesupertypeclassesandseparatelyforeachpolymorphicPBRsiteintheDLA-DRB1andDLA-DQB1loci.
Tocorrectformultipletesting,afalsediscoveryrate(q-value[41])wasestimatedforp-valuesofthecombinedchi-squaredtestsofthedifferentmatingcrosstypesoverthePBRsitesinonelocus.
Twohaplotypesthatwerefoundinfourbreeds/groupscarriedaduplicatedDLA-DQB1locus(DLA-DQB1*01303andDLA-DQB1*01701[34]).
Sinceitisunclearwhetherbothcopiesareexpressedornot,analyseswereconductedincludingallindividualsandseparatelyforthefamilieswith(numberofpuppies=51(DRB1)/59(DQA1)/53(DQB1))andwithout(numberofpuppies=242(DRB1&DQB1)/254(DQA1))theduplicatedlocus.
Whenincludedintheanalyses,theduplicatedlocuswashandledasoneallele(DLA-DQB1*013017).
IntheanalysesfortheDLA-DQB1locuswithsupertypeclassificationandforeachPBRsite,allfamilieswithduplicatedlocuswereexcluded.
MaternaleffectWestudiedthematernaleffectontheMHCclassIIalleletransmissioninthefamilieswheretheoffspringmayinheritthematernalsharedorpaternalnon-sharedallelefromtheirfather:A1A2*A1A3andA1A1*A1A2.
Wetestedthedistributionbetweentheexpectedandobservednumbersofthesharedandnon-sharedallelesinheritedfromfatherwithPearson'schi-squaredtestwithMonteCarlosimulatedp-value(2000repeats).
Thereissomeevidenceinman,thatthemother'sheterozygositymayinfluencetheoffspringheterozygosity–inAmerindianfamiliesonlytheheterozygousmothers'hadincreasedproportionofheterozygousoffspring[17].
So,todiscoverwhetherthemother'sheterozy-gosityhasaneffectontheheterozygosityoftheoff-springindogs,familieswithheterozygousmotherswerejoinedforanalyses.
Thematernaleffectwasstudiedinthetotaldatasetandinthesamesubsetsasdescribedinsection2.
4.
ResultsMHCallelesandhaplotypesAltogether28DLA-DRB1,11DLA-DQA1and23DLA-DQB1allelesorganisedin45three-locushaplotypeswerefoundinthedogsample(Table2).
Thereweresomecommonhaplotypesthatwerefoundinabouthalfofthebreeds/groupsstudiedbutrareuniquehaplotypeswerealsodetectedinmanybreeds.
ThreenewDLA-DRB1al-leleswerefoundamongthe253dogssequencedforthisstudy.
TheallelesequenceshavebeensubmittedtoNCBIGenBankwiththeaccessionnumbers:JQ904819.
1,FM246834,andFM246838.
SupertypeclassificationTheDLA-DRB1allelesweredividedintofivesupertypeclassesandtheDLA-DQB1allelesweredividedintoninesupertypeclassesusinghierarchicalclusteringmethod.
Thenumberofclasseswasbasedonthecostofmergingclusters(Additionalfile2:FiguresS1–S2).
SupertypeclusteringtreesforthecanineDLA-DRB1andDLA-DQB1allelesareshowninFigs.
1and2andthesupertypeclusteringtreeswithallelenamesareshowninAdditionalfile3:FiguresS3–S4.
CompletelistsoftheallelesineachsupertypeclassareinAdditionalfile4:TableS2.
Inhumans,eachMHCsupertypecouldbedescribedbycommonfingerprintaminoacidsthatweresharedwithineachsupertypeclass[37].
Incanines,suchsuper-typefingerprintscouldnotbedetected,sinceaminoacidswereshared,tosomeextent,alsobetweendifferentsupertypes.
BasedonthethreefirstPCsthePBRsites1,5,6,and9inDLA-DRB1,and4,9,10,12,and15inDLA-DQB1hadthelargestimpactonthesupertypeclustering(PC1vs.
PC2inAdditionalfile5:FiguresS5–S6).
Atthesesites,therewerefromtwotoseven(DLA-DRB1)andfromthreetofive(DLA-DQB1)differentaminoacids–asimilarlevelofdiversitycomparedtootherPBRsites.
MHCgenotypeproportionsintheoffspringAtthelocuslevel,deviationfromtheMendelianexpec-tationswasnotobservedinthegenotypeproportionsofthedogfamilies.
Whenexaminedforallofthefamiliesanddifferentmatingcrosstypes,χ2variedbetween0.
02and2.
81(df=1)andp-valuevariedbetween1and0.
13inallMHCloci(Fig.
3andAdditionalfile6:TableS3).
Accordingly,thecombinedchi-squaredanalysesovermatingcrosstypeswithineachlocusdidnotshowanydeviationfromtheexpectedheterozygosityandthecombinedχ2variedbetween2.
28and3.
09(df=1)andp-valuevariedbetween0.
69and0.
54.
AlloftheMHClocifollowedtheMendelianinheritancealsoindifferentbreeds,inpurebreds,inmixed-breeddogs,inbothsexesandinfamilieswithorwithouttheduplicatedDLA-DQB1locus,whentheywereanalysedseparately.
Nodeviationfromtheexpectedheterozygositywasdetectedeitheratthethree-locushaplotypelevel(Additionalfile6:TableS3)orusingsupertypeclassification(Fig.
4).
ThegenotypeproportionsintheoffspringwerealsostudiedineachpolymorphicPBRsite.
Figure5showsbothp-andq-valuesforthecombinedchi-squaredstatisticsoverthedifferentmatingcrosstypesinsite-wisesegregationanalyses.
Aftercorrectingformultipletesting,thePBRsites10–12inDLA-DRB1showedaslightdeviationfromtheexpectedheterozygosity,withsite12havingthesmallestq-value(q=0.
15).
ThePBRNiskanenetal.
CanineGeneticsandEpidemiology(2016)3:9Page5of12Table2Three-locusMHChaplotypesandtheirfrequenciesinthestudieddogbreedsandmixed-breedgroupsDRB1DQA1DQB1AH1AH2CB1CB2ECS2GR2ISDKKHKLLWNF2RRB2SASBT21030100101008020.
280010100101002010.
150.
240.
220.
080.
300.
020.
350010100101036010.
120.
020010100101036030.
120010300101002010.
060020100901001010.
250.
1700601005011007010.
170.
250.
011.
000.
080.
180.
0500601005011020010.
120.
3300901001010080110.
320.
420.
030.
470.
0201001100201015010.
060110100201013030.
040110200201013030.
080120100401013017a0.
330.
510.
190.
010120100401013030.
420130100101002010.
190.
040.
250.
460130100601020020.
070150100901001010.
190150100601023010.
240.
310.
190.
230.
030.
040150100601057010.
030150100601003010.
120150100601022010.
300150100601020020.
070150100401013017a0.
0301501012011035010.
010150100601011010.
180.
090150200601023010.
220.
030150300601003010.
080150300601023040.
070180100101008020.
030.
240.
140.
070200100401013030.
270.
090.
040.
010.
420230100301005010.
230.
180300100601003010.
030400101001019010.
170700101801050010.
0607401005011007010.
160.
020740100101002010.
010760100601023010.
250780100401013030.
080800100402013030.
0109401001010080110.
180950100301054010.
05Niskanenetal.
CanineGeneticsandEpidemiology(2016)3:9Page6of12sites10,11and12correspondtocodonpositions57,60and67inthedata.
MoredetailedexaminationofthePBRsite12revealedexcessofheterozygousoff-springfromthematingcrosstypeA1A2*A1A1(χ2=9.
48,df=1,p=0.
004).
Thesitewaspolymorphicforthreeaminoacids:leucine,isoleucineandphenyla-lanine.
Allthreeaminoacidsarehydrophobic,butphenylalaninediffersfromtheothertwobyhavinganaromaticsidechain.
AlsothePBRsite10showedexcessofheterozygosity(A1A2*A1A1,χ2=4.
46,df=1,p=0.
04)butthePBRsite11showedexcessofhomozygosity(A1A2*A1A2,χ2=4.
95,df=1,p=0.
04).
InDLA–DQB1,noPBRsitesshoweddeviationfromtheexpectedhete-rozygosity(Fig.
5).
MaternaleffectTostudytheputativeprenatalmaternaleffect,alleletransmissionwasstudiedinthefamiliesinwhichparentssharedonealleleandtheoffspringcouldinheritthematernalsharedorpaternalnon-sharedallelefromtheirfather.
Thepuppiestendedtoinheritthematernalsharedallele(A1)fromtheirfatherforDLA-DRB1inmatingcrosstypeA1A2*A1A3moreoftenthanexpected(χ2=3.
02,df=1,p=0.
09;Fig.
6andAdditionalfile6:TableS3).
Thistypeofmatingincludedthemajorityofthefamiliesinthisanalysis.
However,nostrongdeviationsfromtheexpectedalleleproportionswerefoundeitherwhenusingthealleleorsupertypeclassi-ficationintheanalyses.
Thematernalsharedandnon-sharedallelewereequallyofteninheritedfromthefatheralsoindifferentbreeds,inbothsexesandinfamilieswithorwithouttheduplicatedDLA-DQB1locus.
Whentheheterozygousmotherswerecombinedforsegregationanalyses,mother'sheterozygositydidnotaffecttheoffspringheterozygosityneitherattheallelenoratthesupertypelevel(Additionalfile6:TableS3).
DiscussionWehavestudiedprenatalselectionindogsbyanalysingthesegregationoftheMHCclassIIallelesinpurebredandmixed-breedfamilies.
NaturalselectiontargetingtheMHCmaybelimitedinthepetdogsduetotheirprotectedenvironment.
Onepotentialphaseofnaturalselectionisduringfoetaldevelopmentwhichislessstronglyaffectedbyhumanactionorbreedingchoices.
However,wedidnotdetectdeviationsfromtheMendelianTable2Three-locusMHChaplotypesandtheirfrequenciesinthestudieddogbreedsandmixed-breedgroups(Continued)09501001010080110.
020950100901001010.
020970100601020020.
040980100402023010.
01NA0.
020.
080.
120.
01n3964911681313371381766528313AH1AlaskanHusky,AH2AlaskanHusky,CB1Mixed-breed,CB2Mixed-breed,ECS2EnglishCockerSpaniel,GR2GoldenRetriever,ISDIcelandicSheepdog,KKHKooikerhondje,KLKromfohrlander,LWLowchen,NF2Newfoundland,RRB2RhodesianRidgeback,SASaluki,SBT2StaffordshireBullTerriern=numberofindividuals(alllitter-parentscombinations)aDuplicatedDLA-DQB1locus,allelesDLA-DQB1*01303andDLA-DQB1*017011Rowan[35],2Kennedyetal.
[34]Fig.
1SupertypeclusteringtreeforthecanineDLA-DRB1alleles.
DashedlineindicatesdivisionintothefivesupertypeclassesFig.
2SupertypeclusteringtreeforthecanineDLA-DQB1alleles.
DashedlineindicatesdivisionintotheninesupertypeclassesNiskanenetal.
CanineGeneticsandEpidemiology(2016)3:9Page7of12segregationpatternatallele,three-locushaplotype,orsupertypelevels.
Atleastaround200DLA-DRB1,100DLA-DQB1and40DLA-DQA1alleleshavebeenfoundincanines.
Obviouslythereissomefunctionaloverlapinsuchalargenumberofalleles.
Inhumans,ithasbeenshownthattheMHCclassIIalleleshavemoresharedpeptidebindingrepertoirethantheMHCclassIalleles[42].
ClusteringtheMHCallelesintosupertypesisonewaytoreachameaningfulleveloffunctionalityandvariationforstudyingthenaturalselectionintheMHC.
WehavegeneratedthefirstcaninesupertypeclassificationfortheDLA-DRB1andDLA-DQB1alleles.
TheDLA-DQA1alleleswerenotincluded,sincetheywerenotpolymorphicontheputativePBRsitesproposedbyDoytchinovaandFlower[37].
Thismighthaveonlyminoreffecttoourresults,sinceinhumansithasbeenshownthattheHLA-DQA1allelesdidnotcontributetothesupertypeclusteringwhentheywereanalysedtogetherwiththeHLA-DQB1allelestocomposeasinglefunc-tionalunit[37].
Inthisstudy,thenumberofthesupertypeclusters(fiveandnine)wasrelativelylow,andbasedoninsilicoanalysisofamino-acids.
Functionalstudies,wherethepeptidebindingrepertoiresofthedifferentMHCallelescouldbetested,wouldgreatlybenefittheextractionofabiologicallymeaningfulnumberofsuper-typeclusters[31,42].
Thesupertype-clusteredMHCallelescanbeusedinselectionanalysesinsteadofthesequence-basedalleles.
However,wedidnotfindpatternssuggestingprenatalFig.
3Theobserved(Ho)andexpected(He)numbersoftheheterozygousoffspringfromdifferentmatingcrosstypesfortheDLA-DRB1,DLA-DQA1andDLA-DQB1loci(n=totalnumberofoffspringwithinthecrosstype)Fig.
4Theobserved(Ho)andexpected(He)numbersoftheheterozygousoffspringfromdifferentmatingcrosstypesfortheDLA-DRB1andDLA-DQB1supertypes(n=totalnumberofoffspringwithinthecrosstype)Niskanenetal.
CanineGeneticsandEpidemiology(2016)3:9Page8of12selectionatsupertypelevelinthestudieddogfamilies.
Previously,naturalselectionhasbeendetectedatthesupertypelevele.
g.
inthehighermaleheterozygosityofnewbornhumans[43]andinthenon-randomasso-ciationofthealleleswithinDRB-haplotypesofbaboons[31].
Parasiteanddiseaseresistancehasalsobeenasso-ciatedwithMHCsupertypelevelclassesinvariousspecies[30,32].
Thesefindingssuggestthatsupertypeclassificationreachesaleveloffunctionalitythatmaybemissedusingindividualallelesintheselectionanalyses.
Inhumans,somesmallisolatedpopulationshaveshownnon-MendelianproportionsofMHCgenotypesinoffspring[17]orMHCassociatedfoetalloss[8].
However,ofteninstudiesoflargerhumanpopulationsnodeviationfromtheexpectedproportionshasbeenfoundintheMHCloci[1,44].
Thelackofselectioninthemodernlargehumanpopulationsmightbeaconse-quenceofimprovedhealthcare[45]whichalsoappliestodogsbutislikelytomostlyaffectselectivefactorsafterbirthandtolessaffectmaternal-foetalinteractions.
Inthisstudy,thepuppiesweresampledinvaryingagesandtheageofthedogwasnotknowninallcases.
Thus,itispossiblethatasmalldegreeofpostnatalselec-tionhasoccurred.
Wefoundthatthedogpuppiesweremorehetero-zygousthanexpectedinrespecttothePBR-site12ofDLA-DRB1forthematingcrosstypeA1A2*A1A1.
Whenhalfoftheoffspringareexpectedtobehetero-zygousandhalfhomozygous,theselectioncoefficient(s)canbeestimatedwith:s1Nii=Nij1whereNiiistheobservednumberofthehomozygousoffspringandNijistheobservednumberoftheheterozygousoffspring[46].
TheobservedexcessoftheheterozygousindividualsatthePBR-site12ofDLA-DRB1wouldrequirestrongselectionagainsthomozygotes:s=1–38/(70+1)=0.
465.
Selectionofsimilarstrength,s=0.
462overtwoMHCloci,hasbeenestimatedtofavorheterozygotesinhumanpopulations[17].
However,becausetheMHCexpressionincaninefertilisationandpregnancyisnotwellknown,furtherstudiesarerequiredtoinvestigatetheexactcauseoftheobserveddeviation,Fig.
5Thep-andq-valuesforthecombinedchi-squaredstatisticstestingtheratiooftheobservedandexpectedheterozygousoffspringforeachPBRsiteintheDLA-DRB1andDLA-DQB1lociFig.
6Theobserved(Obs.
)andexpected(Exp.
)numbersofoffspringthatinheritedthematernalsharedallelefromtheirfatherforDLA-DRB1,DLA-DQA1andDLA-DQB1lociNiskanenetal.
CanineGeneticsandEpidemiology(2016)3:9Page9of12beforereachingfurtherconclusionsabouttheselectiveadvantageofheterozygosityonthePBR-site12.
TheaminoacidsatthePBR-site12weresharedbydifferentallelesandsupertypes,whichexplainswhywedidnotdetectthesignalofthisdeviationatthehigherlevelsofanalyses.
Ourdataconsistedoftendogbreeds,thathavebeenbredfore.
g.
hunting,herding,andcompanionpurposes,andmixed-breedgroups.
Thesamplingofnumerousbreedshadtheadvantageofrepresentingthespeciesmorewidely,butasadownside,italsocausedindividualallelestohavelowfrequenciesinthehighlypolymorphicMHCloci.
Togainstatisticalpower,weclassifiedthematingcrosstypesbytheallelestheparentssharedwithoutkeepingtheinformationofalleleidentity.
Simul-taneously,wecompromisedtheputativeeffectsofindi-vidualalleles.
Weestimatedthestatisticalpoweroftheχ2testtodistinguishbetweentheexpectedandobservednumbersofpuppiesatthep=0.
05level.
UsingthesamplesizesshowninFig.
3foreachmatingcrosstypeandlocus,sshouldapproximatelybeabove0.
3–0.
6tobedetected.
Valuesthishighindicateverystrongselec-tionandarerarelyobservedinnaturalpopulations.
Itispossiblethattherearebreed-specificdifferences,whichwewerenotabletorecogniseduetothesmallbreed-wisesamplesize,particularlywhenthesamplewasfurtherdividedintomatingcrosstypes.
Asshowninearlierstudies[17],smallsamplesizecanhinderusfromdrawingfinalconclusionsaboutthedifferencesinthepost-copulatoryselectionindogbreeds,especiallyifselectionisnotverystrong.
DuetothehighLDwithintheMHCregion,MHCclassIIallelescouldshowtheeffectofprenatalselectioneveniftheclassIIallelesarenotexpressedintheplacenta.
OtherprenatallyselectedlocithatareinLDwiththeclassicalMHClocimaycontributetothemaintenanceofthehighpolymorphismintheextendedMHCregion[47].
Simultaneously,thismayleadtoapparentsegre-gationdistortionalsointheclassicalMHCgenes.
Forexample,Laurentetal.
[48]detectedasignatureofdisas-sortativematingusingsinglenucleotidepolymorphisms(SNPs)intheextendedhumanMHCregion(3.
6Mb),andconcludedthatitwascausedbymultipleMHCrelatedgenesratherthantheclassicalMHCgenes.
Matechoiceisseldompossibleindogs,butpost-copulatoryselectionmay,inasimilarway,becausedbyseveralMHC-linkedgenes.
TheORgenesmayinfluencebothmateselectionandthesuccessofspermatozoainthefertilisation[47].
Previouslyobservednon-MendeliansegregationofMHCallelesinprimatesmaybecausedbyselectiontargetedtoMHC-linkedORgenes.
Ifthatwasthecase,themissingselectionintheMHCclassIIlociofourdogfamiliesiseasilyexplainedbythefactthatnoORgeneshavethisfarbeenlocalisedinthemajor3MbMHCregiononthedogchromosome12orelsewhereonchromosome12[33,49].
DogspermatozoadoexpressORgenes[50],butputativelyMHC-linkedORgenesareonlyfoundintheminor600kbMHCclassIregionlocatedondogchromosome35[33].
ConclusionsInsummary,wefoundsomeevidenceofpost-copulatoryselectioninpetdogsatnucleotidesitelevel,butnotatlocus,three-locus,orsupertypelevelsinMHCloci.
Itispossiblethatthereisweakerselectionaffectingothersitesandlevels,butoursample,whichonlyspannedonegeneration,wasnotlargeenoughtopickupthesignal[1].
SinceselectionhasmodifiedtheMHCdiversityoveralongtimeperiod,evenweakselectionmayhavebeensufficienttopreservethediversity.
Toourknowledge,thiswasthefirststudyofprenatalselectionindogMHClociandwehavesomesugges-tionshowtoproceedwithfuturestudies.
InfurtherexaminationsofMHCrelatedprenatalselectionincanines,acomprehensivestudyincludingmatechoice,segregationofallelesandlittersizewouldbeinteresting.
Futurestudieswouldalsobenefitfromusingdogsfrommoreprimitivecircumstancesand,forcomparison,theirwildwolfrelatives.
AdditionalfilesAdditionalfile1:TableS1.
MHCgenotypedataandsampleinformationforthedogfamilies.
(XLSX53kb)Additionalfile2:FiguresS1andS2.
MergingcostshowingtheamountofvariancewithinclustersinWard'smethod.
ThisinformationwasusedindecidingthenumberofsupertypeclustersinDLA-DRB1(FigureS1)andDLA-DQB1(FigureS2)loci.
(ZIP3kb)Additionalfile3:FiguresS3andS4.
SupertypeclusteringtreesforthecanineDLA-DRB1(FigureS3)andDLA-DQB1(FigureS4)lociwithallelenamesincluded.
Thesupertypeclustersarecoloureddifferently.
(ZIP12kb)Additionalfile4:TableS2.
Supertypeclassificationalleledataandallallelesinthedogfamilydata.
(XLSX25kb)Additionalfile5:FiguresS5andS6.
BiplotsofthefirsttwoPCsinsupertypeclustersofDLA-DRB1(FigureS5)andDLA-DQB1loci(FigureS6).
ThebiplotsweredoneusingtheggbiplotfunctioninR.
EllipsesshowNormalcontourlineswith68%probabilityforeachcluster.
(ZIP20kb)Additionalfile6:TableS3.
ResulttableforsegregationofheterozygousandhomozygousoffspringindifferentmatingcrosstypesfortheMHCloci,thethree-locushaplotypeandtheMHCsupertypelevels,heterozygousmotherscombinedandthenumbersofoffspringthatinheritedmaternalsharedorpaternalnon-sharedallelefromtheirfather.
(XLSX12kb)Abbreviationsdf:degreeoffreedom;DLA:Dogleucocyteantigen;DNA:Deoxyribonucleicacid;EVT:Extravilloustrophoblastcells;FIS:Inbreedingcoefficient;He:Expectednumberofheterozygotes;HLA:Humanleucocyteantigen;Ho:Observednumberofheterozygotes;LD:Linkagedisequilibrium;Niskanenetal.
CanineGeneticsandEpidemiology(2016)3:9Page10of12MHC:Majorhistocompatibilitycomplex;NF:Numberoffamilies;NK:Naturalkillercells;NO:Numberofoffspring;OR:Olfactoryreceptor;PBR:Peptide-bindingregion;PC(A):Principalcomponent(analysis);s:selectioncoefficientAcknowledgementsWearegratefultoIngeridHagenandtwoanonymousreviewersforprovidingvaluablecommentsontheearlierversionofthismanuscript.
WethankRanjaEklundforcollectingsalukidata,breedclubsforcollaborationincollectingdogsamplesinFinland,andGeorgeHappfromtheUniversityofAlaskaatFairbanks,Alaska,forhuskysamples.
WethankFrancesJuryfortypingandpartialanalysesoftheMHCmicrosatelliteresultsforthehuskyfamiliesinManchester.
AvailabilityofdataandmaterialThedatasetssupportingtheconclusionsofthisarticleareincludedwithinthearticleanditsadditionalfiles.
ThenewDLA-DRB1allelesequenceshavebeensubmittedtoNCBIGenBankwithaccessionnumbers:JQ904819.
1,FM246834,andFM246838.
Authors'contributionsAKNconductedpartofthelaboratorywork.
AKN,LJKandHLcollectedthedata.
AKN,LJKandTPimplementedthedataanalyses.
AKN,TPandJAwrotethemanuscriptwitheditorialinputfromallauthors.
Allauthorsapprovedthefinalmanuscript.
CompetinginterestsLornaKennedyisManagingEditoronthisjournal,andwasexcludedfromtheEditorialprocess.
Otherauthorsdeclarenocompetinginterests.
FundingThisstudywaspartiallysupportedbytheFinnishPopulationGeneticsGraduateProgramme(AKN),KoneFoundation(AKN),theAcademyofFinland(HL:127065),theSigridJuseliusFoundation(HL)andtheJaneandAatosErkkoFoundation(HL).
ConsentforpublicationNotapplicable.
EthicsapprovalNodogDNAsampleswerecollectedspecificallyforthisstudy.
Authordetails1DepartmentofGeneticsandPhysiology,UniversityofOulu,POBox3000,OuluFIN-90014,Finland.
2Presentaddress:CentreforBiodiversityDynamics,DepartmentofBiology,NorwegianUniversityofScienceandTechnology,NO-7491,Trondheim,Norway.
3CentreforIntegratedGenomicMedicalResearch,UniversityofManchester,StopfordBuilding,OxfordRoad,ManchesterM139PT,UK.
4DepartmentofVeterinaryBiosciences,ResearchProgramsUnit,MolecularNeurology,UniversityofHelsinkiandFolkhlsanInstituteofGenetics,BiomedicumHelsinki,POBox63,FIN-00014Helsinki,Finland.
Received:28January2016Accepted:2October2016References1.
GarriganD,HedrickPW.
Perspective:DetectingAdaptiveMolecularPolymorphism:LessonsFromtheMHC.
Evolution(NY).
2003;57:1707–22.
2.
BenacerrafB.
RoleofMHCgeneproductsinimmuneregulation.
Science(80-).
1981;212:1229–38.
3.
EhlersA,BeckS,ForbesSA,TrowsdaleJ,VolzA,YoungerR,ZieglerA.
MHC-LinkedOlfactoryReceptorLociExhibitPolymorphismandContributetoExtendedHLA/OR-Haplotypes.
GenomeRes.
2000;10:1968–78.
4.
XieT,RowenL,AguadoB,AhearnME,MadanA,QinS,CampbellRD,HoodL.
AnalysisoftheGene-DenseMajorHistocompatibilityComplexClassIIIRegionandItsComparisontoMouse.
GenomeRes.
2003;13:2621–36.
5.
ClarkeB,KirbyDRS.
MaintenanceofHistocompatibilityPolymorphisms.
Nature.
1966;211:999–1000.
6.
DohertyPC,ZinkernagelRM.
EnhancedimmunologicalsurveillanceinmiceheterozygousattheH-2genecomplex.
Nature.
1975;256:50–2.
7.
ApaniusV,PennD,SlevPR,RuffLR,PottsW.
Thenatureofselectiononthemajorhistocompatibilitycomplex.
CritRevImmunol.
1997;17:179–224.
8.
OberC.
StudiesofHLA,fertilityandmatechoiceinahumanisolate.
HumReprodUpdate.
1999;5:103–7.
9.
PennDJ.
Thescentofgeneticcompatibility:sexualselectionandtheMajorHistocompatibilityComplex.
Ethology.
2002;108:1–21.
10.
FernandoMMA,StevensCR,WalshEC,DeJagerPL,GoyetteP,PlengeRM,VyseTJ,RiouxJD.
DefiningtheRoleoftheMHCinAutoimmunity:AReviewandPooledAnalysis.
PLoSGenet.
2008;4:e1000024.
11.
JokinenP.
Identifyinggeneticriskfactorsincanineautoimmunedisorders.
Helsinki:UniversityofHelsinki;2011.
12.
KennedyLJ,BarnesA,ShortA,BrownJJ,LesterS,SeddonJ,FleemanL,FrancinoO,BrkljacicM,KnyazevS,HappGM,OllierWER.
CanineDLAdiversity:1.
Newallelesandhaplotypes.
TissueAntigens.
2007;69:272–88.
13.
NiskanenAK,KennedyLJ,RuokonenM,KojolaI,LohiH,IsomursuM,JanssonE,PyhjrviT,AspiJ.
BalancingselectionandheterozygoteadvantageinmajorhistocompatibilitycomplexlociofthebottleneckedFinnishwolfpopulation.
MolEcol.
2014;23:875–89.
14.
KennedyLJ,AnglesJM,BarnesA,CarmichaelLE,RadfordAD,OllierWER,HappGM.
DLA-DRB1,DQA1,andDQB1allelesandhaplotypesinNorthAmericangraywolves.
JHered.
2007;98:491–9.
15.
NiskanenAK,HagstrmE,LohiH,RuokonenM,Esparza-SalasR,AspiJ,SavolainenP.
MHCvariabilitysupportsdogdomesticationfromalargenumberofwolves:highdiversityinAsia.
Heredity(Edinb).
2013;110:80–85.
16.
WedekindC,ChapuisatM,MacasE,RülickeT.
Non-randomfertilizationinmicecorrelateswiththeMHCandsomethingelse.
Heredity(Edinb).
1996;77:400–9.
17.
BlackF,HedrickP.
StrongbalancingselectionatHLAloci:evidencefromsegregationinSouthAmerindianfamilies.
ProcNatlAcadSciUSA.
1997;94(November):12452–6.
18.
KnappLA,HaJC,SackettGP.
ParentalMHCantigensharingandpregnancywastageincaptivepigtailedmacaques.
JReprodImmunol.
1996;32:73–88.
19.
BillingtonWD.
InfluenceofImmunologicalDissimilarityofMotherandFoetusonSizeofPlacentainMice.
Nature.
1964;202:317–8.
20.
WarningJC,McCrackenSA,MorrisJM.
Abalancingact:mechanismsbywhichthefetusavoidsrejectionbythematernalimmunesystem.
Reproduction.
2011;141:715–24.
21.
KovatsS,MainEK,LibrachC,StubblebineM,FisherSJ,DeMarsR.
AclassIantigen,HLA-G,expressedinhumantrophoblast.
Science(80-).
1990;248:220–3.
22.
KingA,BoocockC,SharkeyAM,GardnerL,BerettaA,SiccardiAG,LokeYW.
EvidencefortheexpressionofHLA-CclassImRNAandproteinbyhumanfirsttrimestertrophoblast.
JImmunol.
1996;156:2068–76.
23.
TrowsdaleJ,BetzAG.
Mother'slittlehelpers:mechanismsofmaternal-fetaltolerance.
NatImmunol.
2006;7:241–6.
24.
ParhamP,MoffettA.
VariableNKcellreceptorsandtheirMHCclassIligandsinimmunity,reproductionandhumanevolution.
NatRevImmunol.
2013;13:133–44.
25.
ChazaraO,XiongS,MoffettA.
MaternalKIRandfetalHLA-C:afinebalance.
JLeukocBiol.
2011;90:703–16.
26.
MoffettA,LokeC.
Immunologyofplacentationineutherianmammals.
NatRevImmunol.
2006;6:584–94.
27.
Schfer-SomiS,BeceriklisoyHB,BudikS,KancaH,AksoyOA,PolatB,CetinY,AySS,AslanS.
Expressionofgenesinthecaninepre-implantationuterusandembryo:Implicationsforanactiveroleoftheembryobeforeandduringinvasion.
ReprodDomestAnim.
2008;43:656–63.
28.
Schfer-SomiS,BeceriklisoyH,WalterI,SabitzerS,KleinD,KancaH,KayaD,AgaogluA,zgürH,AslanS.
ExpressionofMHC-Iand-IIinUterineTissuefromEarlyPregnantBitches.
ReprodDomestAnim.
2009;44:103–8.
29.
SetteA,NewmanM,LivingstonB,McKinneyD,SidneyJ,IshiokaG,TangriS,AlexanderJ,FikesJ,ChesnutR.
Optimizingvaccinedesignforcellularprocessing,MHCbindingandTCRrecognition.
TissueAntigens.
2002;59:443–51.
30.
SchwensowN,FietzJ,DausmannKH,SommerS.
Neutralversusadaptivegeneticvariationinparasiteresistance:importanceofmajorhistocompatibilitycomplexsupertypesinafree-rangingprimate.
Heredity(Edinb).
2007;99:265–77.
31.
HuchardE,WeillM,CowlishawG,RaymondM,KnappLA.
Polymorphism,haplotypecomposition,andselectionintheMhc-DRBofwildbaboons.
Immunogenetics.
2008;60:585–98.
Niskanenetal.
CanineGeneticsandEpidemiology(2016)3:9Page11of1232.
SepilI,LachishS,HinksAE,SheldonBC.
Mhcsupertypesconferbothqualitativeandquantitativeresistancetoavianmalariainfectionsinawildbirdpopulation.
ProcRSocB.
2013;280:20130134.
33.
YuhkiN,BeckT,StephensR,NeelamB,O'BrienSJ.
Comparativegenomicstructureofhuman,dog,andcatMHC:HLA,DLA,andFLA.
JHered.
2007;98:390–9.
34.
KennedyLJ,BarnesA,ShortA,BrownJJ,LesterS,SeddonJ,HappGM,OllierWER.
CanineDLAdiversity:2.
Familystudies.
TissueAntigens.
2007;69Suppl1:289–91.
35.
RowanD.
AtimeandcosteffectivemethodforDogLeukocyteAntigen(DLA)typingusingcaninemicrosatellitemarkers.
Manchester:UniversityofManchester;2007.
36.
DebenhamSL,HartEA,AshurstJL,HoweKL,QuailMA,OllierWER,BinnsMM.
GenomicsequenceoftheclassIIregionofthecanineMHC:comparisonwiththeMHCofothermammalianspecies.
Genomics.
2005;85:48–59.
37.
DoytchinovaIA,FlowerDR.
InsilicoidentificationofsupertypesforclassIIMHCs.
JImmunol.
2005;174:7085–95.
38.
SandbergM,ErikssonL,JonssonJ,SjstrmM,WoldS.
Newchemicaldescriptorsrelevantforthedesignofbiologicallyactivepeptides.
Amultivariatecharacterizationof87aminoacids.
JMedChem.
1998;41:2481–91.
39.
WardJH.
Hierarchicalgroupingtooptimizeanobjectivefunction.
JAmStatAssoc.
1963;58:236–44.
40.
Rcoreteam.
R:Alanguageandenvironmentforstatisticalcomputing.
2014.
41.
StoreyJD,TibshiraniR.
Statisticalsignificanceforgenomewidestudies.
ProcNatlAcadSciUSA.
2003;100:9440–5.
42.
GreenbaumJ,SidneyJ,ChungJ,BranderC,PetersB,SetteA.
FunctionalclassificationofclassIIhumanleukocyteantigen(HLA)moleculesrevealssevendifferentsupertypesandasurprisingdegreeofrepertoiresharingacrosssupertypes.
Immunogenetics.
2011;63:325–35.
43.
DorakMT,LawsonT,MachullaHK,MillsKI,BurnettAK.
IncreasedheterozygosityforMHCclassIIlineagesinnewbornmales.
GenesImmun.
2002;3:263–9.
44.
KlitzW,LoSK,NeugebauerM,BaurMP,AlbertED,ThomsonG.
AcomprehensivesearchforsegregationdistortioninHLA.
HumImmunol.
1987;18:163–80.
45.
MarkowT,HedrickPW,ZuerleinK,DanilovsJ,MartinJ,VyvialT,ArmstrongC.
HLApolymorphismintheHavasupai:evidenceforbalancingselection.
AmJHumGenet.
1993;53:943–52.
46.
HaldaneJBS.
Theestimationofviabilities.
JGenet.
1956;54:294–6.
47.
ZieglerA,SantosPSC,KellermannT,Uchanska-ZieglerB.
Self/nonselfperception,reproductionandtheextendedMHC.
SelfNonself.
2010;1:176–91.
48.
LaurentR,ToupanceB,ChaixR.
Non-randommatechoiceinhumans:insightsfromagenomescan.
MolEcol.
2012;21:587–96.
49.
QuignonP,GiraudM,RimbaultM,LavigneP,TacherS,MorinE,RetoutE,ValinA-S,Lindblad-TohK,NicolasJ,GalibertF.
Thedogandratolfactoryreceptorrepertoires.
GenomeBiol.
2005;6:R83.
50.
VanderhaeghenP,SchurmansS,VassartG,ParmentierM.
Specificrepertoireofolfactoryreceptorgenesinthemalegermcellsofseveralmammalianspecies.
Genomics.
1997;39:239–46.
Weacceptpre-submissioninquiriesOurselectortoolhelpsyoutondthemostrelevantjournalWeprovideroundtheclockcustomersupportConvenientonlinesubmissionThoroughpeerreviewInclusioninPubMedandallmajorindexingservicesMaximumvisibilityforyourresearchSubmityourmanuscriptatwww.
biomedcentral.
com/submitSubmityournextmanuscripttoBioMedCentralandwewillhelpyouateverystep:Niskanenetal.
CanineGeneticsandEpidemiology(2016)3:9Page12of12
- Goldentoutoulu.com相关文档
- appliqutoutoulu.com
- accounttoutoulu.com
- globaltoutoulu.com
- retourstoutoulu.com
- statisticaltoutoulu.com
- Lestoutoulu.com
A400：36元/季，16.8/月kvm架构，线路优质，延迟低
A400互联是一家成立于2020年的商家，主要推行洛杉矶服务器采用kvm架构，线路优质，延迟低，稳定性高！全场产品对标腾讯云轻量，服务器线路有有美国洛杉矶cn2_gia、香港cn2+cmi，目前推行的vps服务器均为精心挑选的优质线路机房，A400互联推出了夏季优惠洛杉矶5折、香港7折促销活动，质量可靠，价格实惠！二：优惠码洛杉矶五折优惠码：20210620香港cn2七折优惠码：0710三、优惠方...
创梦网络-江苏宿迁BGP云服务器100G高防资源,全程ceph集群存储,安全可靠,数据有保证,防护真实,现在购买7折促销,续费同价！
官方网站：点击访问创梦网络宿迁BGP高防活动方案：机房CPU内存硬盘带宽IP防护流量原价活动价开通方式宿迁BGP4vCPU4G40G+50G20Mbps1个100G不限流量299元/月 209.3元/月点击自助购买成都电信优化线路8vCPU8G40G+50G20Mbps1个100G不限流量399元/月 279.3元/月点击自助购买成都电信优化线路8vCPU16G40G+50G2...
美国G口/香港CTG/美国T级超防云/湖北高防云服务器物理机促销活动 六一云
六一云 成立于2018年，归属于西安六一网络科技有限公司，是一家国内正规持有IDC ISP CDN IRCS电信经营许可证书的老牌商家。大陆持证公司受大陆各部门监管不好用支持退款退现，再也不怕被割韭菜了！主要业务有：国内高防云，美国高防云，美国cera大带宽，香港CTG,香港沙田CN2,海外站群服务，物理机，宿母鸡等,另外也诚招代理欢迎咨询。官网www.61cloud.net最新直销劲爆...
toutoulu.com为你推荐
-
公司网络被攻击最近公司频繁的受到网络攻击，导致网络瘫痪，又碰到arp攻击，有病毒的，有点崩溃。。。百度商城百度商城里抽奖全是假的商标注册流程及费用商标注册流程及费用？商标注册流程及费用注册商标的程序及费用？西部妈妈网我爸妈在云南做非法集资了，钱肯定交了很多，我不恨她们。他们叫我明天去看，让我用心的看，，说是什么...关键字关键词编故事刘祚天DJ这个职业怎么样？嘀动网动网和爱动网各自的优势是什么？seo优化工具SEO优化神器有什么比较好的？8090lu.com8090lu.com怎么样了?工程有进展吗？