forinnateimmunity
JamieTwycross,UweAickelin
SchoolofComputerScience,UniversityofNottingham,UK
jpt@cs.nott.ac.uk
Abstract.Innateimmunitynowoccupiesacentralroleinimmunology.However,artificialimmunesystemmodelshavelargelybeeninspiredbyadaptivenotinnateimmunity.Thispaperreviewsthebiologicalprin-ciplesandpropertiesofinnateimmunityand,adoptingaconceptualframework,askshowthesecanbeincorporatedintoartificialmodels.Theaimistooutlineameta-frameworkformodelsofinnateimmunity.
1Introduction
Immunologyhastraditionallydividedtheimmunesystemintoinnateandadap-tivecomponentswithdistinctfunctionalroles.Formanyyears,researchwasfocusedontheadaptivecomponent.However,theprevailingviewinimmunol-ogynowshowstheinnatesystemtobeofcentralimportance[1].Thefirstpartofthispaperfocusesontheinnateimmunesystemandonwaysinwhichitinteractswithandcontrolstheadaptiveimmunesystemanddiscussesresearchoverthelastdecadewhichhasuncoveredthemolecularbasisformanyofthesemechanisms,reviewedin[2].Itfirstcontraststheinnateandadaptiveimmunesystemsandbrieflyreviewsessentialbiology.Itthendiscussesspecificmecha-nismsofinteractionbetweencellsoftheinnateandadaptiveimmunesystems,andconcludesbyshowinghowthesemechanismsareexamplesofmoregeneralsystemicproperties.
Whiletheintegralroleoftheinnateimmunesystemhasbeenestablishedinimmunology,artificialimmunesystemmodels,surveyedin[3,4],havelargelytakentheirinspirationfromadaptiveimmunity.ThesecondpartofthispaperadoptstheconceptualframeworkofStepneyetal.[5]andaddresseshowideasfrominnateimmunitymightbemodelledinartificialimmunesystems.Thecon-ceptualframeworkisfirstbrieflysummarisedandthenageneralmeta-frameworkformodelsincorporatinginnateimmunityispresentedandrefinedthroughthediscussionofspecificmodelsproperties.
2Innateimmunity
Thissectionbeginswithanoverviewofwell-establishedconceptionsofinnateimmunity.Researchwhichoverthelastdecadehasservedtohighlightthecentral
roleoftheinnateimmunesystemisthendiscussed.Lastly,generalpropertiesoftheinnateimmunesystemwhichhavebeendrawnoutbythisresearcharepresented.Reviewpapersaswellastheoriginalarticlesarecited,andorigi-nalfiguresarereproducedtoenhancethenecessarilybriefsummariesofthemechanisms.2.1
Contrastinginnateandadaptiveimmunity
Differencesbetweentheinnateandadaptiveimmunesystemscanbeseenonanumberoflevels(Table1).Theadaptiveimmunesystemisorganisedaroundtwoclassesofcells:TcellsandBcells,whilethecellsoftheinnateimmunesystemaremuchmorenumerous,includingnaturalkiller(NK)cells,dendriticcells(DCs),andmacrophages.Thereceptorsofinnatesystemcellsareentirelygermline-encoded,inotherwordstheirstructureisdeterminedbythegenomeofthecellandhasafixed,genetically-determinedspecificity.Adaptiveimmunesystemcellspossesssomaticallygeneratedvariable-regionreceptorssuchastheTCRandBCR(TandBcellreceptors)withvaryingspecificities,createdbyacomplexprocessofgenesegmentrearrangementwithinthecell.Onapopulationlevel,thisleadstoanon-clonaldistributionofreceptorsoninnateimmunesystemcells,meaningthatallcellsofthesametypehavereceptorswithidenticalspecificities.Receptorsonadaptiveimmunesystemcellshowever,aredistributedclonallyinthattherearesubpopulationsofaspecificcelltype(clones)whichallpossessreceptorswithidenticalspecificities,butthatgenerally,cellsofthesametypehavereceptorswithdifferentspecificties[1,6,7].
property
DC,NK,macrophage.germline-encoded.
rearrangementnotnecessary.non-clonaldistribution.
conservedmolecularpatterns.selectedoverevolutionarytime.cytokines,chemokines.
immediateeffectoractivation.vertebratesandinvertebrates.
adaptiveimmunesystem
organismbyprocessessuchasclonalexpansion,deletionoranergyandareunderadaptivenotevolutionarypressure.Conversely,innateimmunesystemreceptorsrecogniseagenetically-determinedsetofligandsunderevolutionarypressure.Onekeygroupofinnatereceptorsisthepatternrecognitionreceptor(PRR)superfamilywhichrecognisesevolutionary-conservedpathogen-associatemolec-ularpatterns(PAMPs).PRRsdonotrecogniseaspecificfeatureofaspecificpathogenasvariable-regionreceptorsdo,butinsteadrecognisecommonfeaturesorproductsofanentireclassofpathogens.Theimmunesystemutilisesadap-tationofvariable-regionreceptorstokeeppacewithevolutionarymorerapidpathogens[1,6].
Theenvironmentofacellinvivoisthetissueinwhichitislocated.Tissueisformedbyspecialisedgroupsofdifferentiatedcells,anditselfformsmajorcomponentsoforgans.Asubstantialpartoftissuevolumeisextracellularspaceandfilledbyastructurednetworkofmacromoleculescalledtheextracellularmatrix.Manyofthemoleculesfoundintheextracellularmatrixareactivelyproducedbycellsandinvolvedinintercellularsignalling[8,9].Cytokinesaresecretedmoleculeswhichmediateandregulatecellbehaviour,twoimportantsubsetsofwhicharetissuefactors,inflammation-associatedmoleculesexpressedbytissuecellsinresponsetopathogeninvasion,andchemokines,cytokineswhichstimulatecellmovementandactivation.Cytokinesbindtogermline-encodedcytokinereceptorspresentonallcellsandarewidelyproducedandconsumedbybothinnateandadaptiveimmunesystemcellsduringanimmuneresponse.Recognitionbytheinnateimmunesystemleadstotheimmediateinitiationofcomplexnetworksofcytokinesignallingwhichorchestratetheensuingimmuneresponse.Adaptiveresponsesadditionallyinvolveprocessesofcellselectionsuchasclonalexpansion,deletionandanergy,whichtakeseveraldays[1,6].2.2
Recentdevelopments
Thissectionreviewskeydevelopmentsoverthelastdecadeinourunderstandingoftheinnateimmunesystem.Overthisperiod,intenseresearchhashighlightedthecentralroleoftheinnatesysteminhostdefensethroughitsinteractionwiththeadaptiveimmunesystemandwithtissue,anduncoveredthemolecularbasisfortheseinteractions.Thesedevelopmentshaveleadimmunologiststoreevaluatetherolesofboththeinnateandadaptiveimmunesystemsinthegenerationofimmunity,installinginnateimmunityasavitalcomponentintheinitiationandmodulationoftheadaptiveimmuneresponse[2].
NKcellsoftheinnateimmunesystemrespondtothedisruptionofnor-malcellphysiologyinwhathasbeentermedthe“missingself”modelofNKcellactivation[10].MostnormaltissuecellsconstitutivelyexpressMHCclassImolecules,whichpresentintracellularhost-derivedpeptidesonthecellsurface.Presentationofvirus-derivedpeptidesleadstoactivationofCTL(cytotoxicTlymphocyte)cellsandapoptosisintheinfectedcellthroughligationwiththeTCRoftheCTL[11].However,virusesandotherinfectiousagentshaveevolvedtointerferwithMHCclassIantigenpresentation[12]andsoevadeaCTLre-sponse.Inthe“missingself”model(Figure1),NKcellsareactivatedeither
Fig.1:NKreceptorsandNKrecognition,from[2]
byreducedsignallingthroughreceptorsoftheKIRfamily,inhibitoryreceptorsspecificforhostMHCclassI,leadingtoapoptosisofthecognatecell[13].Thiscreatesano-winsituationforthevirus:ifMHCclassIexpressionisunaffected,itwillbeopentodetectionandremovalthroughaCTL-basedadaptiveimmuneresponse,butifitaffectsMHCclassIexpression,itwillbeopentodetectionandremovalthroughanNK-basedinnateimmuneresponse.
SomeofthemostexcitingrecentadvanceshavebeenmadeinuncoveringtheroleofTLRsindeterminingDCdifferentiationandsoamechanismbywhichtheinnateimmunesystemmediatesthequalityofanadaptiveimmunesystemresponse[2,15](Figure2).InitialligationbydifferentPAMPsandtissuefactorsofdifferentTLRsonDCs“primes”DCstodifferentiatealongdifferentpathways,resultinginmatureandimmatureDCswhichproducedifferentTh(Thelper)cellpolarisationfactors.ReleaseofthesepolarisationfactorsuponinteractionwithnaiveTcellscausesthenaivecelltodifferentiateintoTh1,Th2orTregcells,alldistincttypesofTcell[14].DCs,throughTLRs,couplethequalityoftheadaptiveimmuneeffectorresponsetothenatureofthepathogen.OtherPRRreceptorfamilieshavealsobeenimplicatedinThpolarisation[16].Interestingly,recentresearch[17]suggestsarenewedroleforvariable-regionreceptorsnotjustinthedeterminationoftheantigenspecificityofanimmuneresponse,butalsointheregulationofthisresponse.Inplaceofthepurely“instructive”DCtoTcellparadigm,therespondingTh1orTh2cellsreinforcesignalstoBcellorCTLeffectorsina“success-driven”consensualmodelofTcellpolarisation.
Fig.2:DCpolarisationofThcells,from[14]
AswellaspolarisingThcells,DCsplayakeyroleinmaintenanceofpop-ulationsofTcells.Toleranceistheabilityoftheimmunesystemtoreactinanon-biodestructivemannertostimuliandhaslongbeenassociatedwithadap-tiveimmunity.Toleranceisusuallydiscussedintermsofapoptosisoranergyofself-reactiveTandBcells,andwasinitiallyproposedtooccurcentrallyinarelativelyshortperinatalperiod,asepitomisedintheclonalselectiontheoryofBurnet[19,20].Whilerecentresearchshowsthecontinuingimportanceofcentraltolerancemechanism[21],itisnowacceptedthatperipheraltolerancemechansimswhichoperatetocensorcellsthroughoutthelifetimeofthehostareofequalimportance.DCsoftheinnateimmunesystemlieattheheartofthegenerationofperipheraltolerance.ModelsproposethatDCscontinuallyuptakeapoptoticandothermaterialfromperipheraltissuesundernormalsteady-state,nonpathogenicconditions.Periodically,DCsmigratetodraininglymphnodeswheretheydeletelymphocytesbypresentingtheprocessedmaterialwhich,asrepresentativeoftissueintheabsenceofpathogen,needstobetoleratedbythehost[22].SignalsreceivedbyDCs‘license”[18](Figure3)themtopromoteeitherTcellclonalexpansion,orTcellclonalanergyordeletion.Researchhasestablishedthemolecularbasisforsuchmodels.TheabsenceofTLRsignallingonDCs[22,23]orthepresenceofsignallingthroughreceptorsinvolvedintheup-takeofapoptoticmaterial[24,25]leadstodistinctsemimatureandmatureDCpopulationswhichinteractwithTcellstopromotetoleranceorimmunogenicityrespectively.
Fig.3:DCThtolerance,from[18]
CosignallingreceptorsandtheirligandsprovideanothermechanismbywhichDCsdeterminethequalitativeandquantitativenatureofadaptiveimmunere-sponses.CD80andCD86arecostimulatorymoleculesexpressedonDCsandbindwiththeCD28andCTLA-4cosignallingreceptorsonThcells.BindingtoCD28leadstoupregulationofThactivityandanimmunogenicresponse,whereasCTLA-4bindingtodownregulationofactivityandtolerance.CD28isconstitutivelyexpressedbyThcells,whereasthelatterinproportiontothestrengthofTCRstimulation.CD80andCD86donotbindequivalentlytoCD28andCTLA-4,andthroughselectiveexpressionbyDCsofthesemolecules,innateimmunesystemcellsinitiateandregulateThcellactivity.Akeyconceptwhichhasemergedfromthisresearchistheimportanceofsequentialandproperlytimedinteractionsinthedevelopmentofanimmuneresponse[2,26,27].
2.3Summary
Asthebiologydescribedinthissectionshows,theprotectionaffordedtothehostbytheimmunesystemasawholearisesfrommechanismsoftheinnateandadaptiveimmunesystems,whichhelpformanintegratedsystemofhostprotection.Whiletherecanbenodoubtthatspecificrecognitionbytheadaptiveimmunesystemplaysanimportantroleinfunctionssuchaspathogenrecognitionandremoval,itisnowclearthatinnateimmunesystemmechanismsplayanequallyimportantrole.Themechanismsdiscussedabovearespecificexamplesofmoregeneralpropertiesofinnateandadaptiveimmunesystemfunctionandinteraction,whicharesummarisedinTable2.
property1property2property3property4property5property6
aswellasrelyingonantigenprocessedbyDCs,alsoshowshowinnateimmunesystemcellsmaintainpopulationsofcells(Property5).Theadaptiveresponseisdrivenbyinformationnotonlydirectlysensedbyadaptiveimmunesystemcells,butequallybyinformationgatheredandprocessedbyinnateimmunesystemcells,aswithDCcollection,processingandpresentationofantigentoTcells(Property6).
3Modellinginnateimmunity
Asartificialimmunesystemsdevelopintheirsophisticationandsoaremoreabletorealisethefunctionsofbiologicalimmunesystems,theywillneedtoincorporatepropertiesofinnateimmunityintotheirmodels.ThissectionfirstreviewstheconceptualframeworkforartificialimmunesystemsofStepneyetal.[5].Adoptingthisframeworkanddrawingonthebiologyoftheprevioussection,itthenproposesanumberofgeneralpropertiesofmodelsincorporatinginnateimmunity.Lookingfirstatthemechanismsoftheprevioussectionasawhole,andthenindividually,thesegeneralpropertiesarediscussedandrefined.Theaimistosuggestameta-frameworkwhichhighlightsthekeypropertiesofmodelsingeneralandhowtheymightberealisedinvariousindividualmodels.3.1
Conceptualframeworks
In[5],Stepneyetal.presentaconceptualframeworkwithinwhichbiologically-inspiredmodelsandalgorithmscanbedevelopedandanalysed.Figure4sum-marisestheirframework,inwhichprobesprovidetheexperimenterwithanin-completeandbiasedviewofacomplexbiologicalsystemwhichthenallowstheconstructionandvalidationfirstofsimplifyingabstractrepresentions,andconsequentlyofanalyticalcomputationalframeworks,whichthemselvesprovideprinciplesforthedesignandanalysisofbiologically-inspiredalgorithms.
beanalysedbyaskingquestions,calledmeta-probes,ofeachofthemodelsun-derconsideration.Theysuggestanumberofquestionsbasedaroundpropertieswhicharethoughttoaffectcomplexbehaviouringeneral.Theseareasrelatetoopenness,diversity,interaction,structureandscale(ODISS).Usingthismeta-framework,theauthorsanalysethecommonalitiesofpopulationandnetworkmodels.
Fig.5:Aconceptualframeworkforintegratingbiologically-inspiredcomputationaldomains[5].
WhileStepneyetal.usethemeta-frameworktoanalyseartificialmodelsforessentialfeaturesandcommonalities,thispaperusesittoanalysebiologicalmodels.Thelatterapproach,apartfrombeingpragmaticasveryfewartificialmodelscurrentlyexist,alsoallowsbiologytohavemuchmoreofaninfluenceonthemeta-framework.Whicheverapproachistaken,meta-frameworksandthedevelopmentofcomputationalandmathematicalmodelspresentaroutethroughwhichartificialimmunesystemresearchcanhelpbiologistsanswerresearchques-tionsintheirfield.3.2
Ameta-frameworkforinnatemodels
ThissectiontakesthegeneralpropertiesoftheinnateimmunesystempresentedinSection2andabstractsthembyadoptingtheconceptualframework.Theabstractedpropertiesformthebasisofameta-frameworkforinnatemodelsandarepresentedintermsofeachoftheODISSareasoftheconceptualframework:
openness:theinteractionbetweentheimmunesystemandthehostisoneofapoisedsystemindynamicequilibriumcoupledtoanever-changingenvironment.Therelativelyconstantpopulationsofinnateimmunesystemcellscontrastswiththefluctuatingpopulationsoftheadaptivesystem.Theinnateimmunesystemprovidesexamplesofmechanismsforcontrollingthedynamicallocationofresourcesofpopulationsofagents.
diversity:thedifferentclassesofcellsoftheinnateandadaptiveimmunesystemsleadstotheideaofdistinctgroupsoffunctionallysimilaragents.Atadifferentlevel,clonaldistributionofreceptorsisangoodexampleofdifferentwaysinwhichdiversitymanifestsitselfinbiologicalsystems.Theunderlyingpro-
cesseswhichdrivediversityofinnatereceptorsareevolutionary,whileadaptivereceptordiversityisestablishedthroughadaptation.
interaction:inthewidersenseconsideringtheinnateimmunesystemshowshowcomputationislargelycommunication,withimmunityarisingfromthecytokinenetworksofsignallinginteractionsbetweenintercommunicatingtissuecellsandtheinnateandadaptiveimmunesystems.Adaptiveandinnateimmunecellsarealsospecialisedtoaccessdifferentinformationallevels.Innatecellsfocusonclassfeatures,whileadaptivecellsonindividualfeatures.Crosstalkbetweensignallingnetworksisalsoaprevalentpropertyintheimmunesystem.Spatialityandtemporallyarekeyfeaturesofinteractionsacrossalllevels.
structure:consideringtheinnateimmunesystemnecessitatesaviewoftheimmunesystemcomposedofdistinctsubsystems.Functionalsimilaritiesaswellasdifferencescanbeseenbetweentheinnateandadaptivesubsystems.Thein-nateandadaptivearethemselvescomposedofinteractingpopulationsofagents.Celldifferentiationpathwaysprovideanevenmorefine-graineddivisionofcellsintotypes.
scale:diversepopulationsoflargenumbersofcellsisahallmarkoftheim-munesystem.Achallengeforartificialimmunesystemsistheneedtosimulatelargepopulationsofagents.Exploitingtheemergentpropertiesofdistinctpop-ulationsoflargenumbersofsimpleagentsratherthanasmallernumberofmorecomplexagents,alongwithdistributedandparallelarchitecturesforartificialimmunesystems[28]mayprovideawayforward.3.3
Refiningtheframework
Asseenintheprevioussection,theinnateimmunesystemprovidesexam-plesofgeneralpropertiesforartificialsystems.Systemsofagentsformacon-venientmeta-representationofartificialsystems,andmanyartificialsystemsarebasedonpopulationsofinteractingagents.Thissectionadoptsthismeta-representationandrefinesthegeneralpropertiesoftheprevioussectionbydis-cussinghowtheymightbeinstantiatedinmodels.
Cellsseenasautonomousagentsformsthebasisofthemeta-representationdiscussedhere.TheintercellularcommunicationinvolvedinallthemechanismsofSection2suggeststheneedforsimilarmeansofintercommunicationbetweenagents.Signalswhichallowgroupsofagentstocontrolthefunctionsandstateofothergroupsofagentsarenecessary.Afiner-grainrepresentationofintercellularsignalsintodistinctclasses,asseeninthebiologicalimmunesystem,suchascos-timulatory,primerorchemokinesignals,wouldallowartificialsystemstomorecloselyapproximatethecontrolmechanismsandsystemicpropertiesofbiologi-calsystems.Akeyroleoftheenvironmentwhichtheseagentsexistin,termedartificialtissuehere,istheprovisionofamilieuinwhichagentscaninteractviasignalling.Aswellaspassingsignalsbetweenagents,mechanismssuchasanti-genprocessingandpresentationtoThcellsbyDCssuggesttheneedforagentswiththeabilityto“consume”,processandpassoninformationtootheragents.Somegroupsofagents,akintoThorCTLcells,wouldnothavedirectaccesstoinformation,butinsteadseeitthroughthefilterprovidedbytheseinformation
processingagents.Artificialtissuewouldprovidethemechanismsforthesekindsofinteractionwiththeenvironmentandotheragents.
Therepresentationofpathogensatmultiplelevelssuggestsanother“service”whichartificialtissuehastoprovide.Aproblemmustberepresentedatmulti-plelevels.Theartificialtissueallowsagentsoftheartificialimmunesystemtoaccessdifferentlevelsofinformationaboutevents.Attheveryleast,informa-tionconcerningthestructureofeventsandsignalsrelatingtothewayelementsbehaveorinteractwiththetissueasawholeneedstobeaccessible.Classicalstaticclassificationproblemscouldperhapsbetranslatedintosuchamultilevelrepresentationbyclusteringalgorithmsorstatisticalmethodswhichgiveindi-cationsofhowindividualfeaturevectorsrelatetoawholesetofothervectors.However,theinnateimmunesystemclearlyreliesonsensingthebehaviouraswellasstructureofpathogens,andtissuemodelsbuiltentirelyfrominformationderivedfromstructuralconsiderations,suchassimilarityordifferencesbetweenfeaturevectors,failtocapturethisreliance.Dynamic,realtimeproblemssuchasintrusiondetectionofferamuchmoreamenabledomainastheynaturallyincludenotionsofbehaviour.Forexample,acomputervirusnotonlyhasapar-ticularstructure,itsprogramcode,butalsobehavesinacertainwaythroughitsinteractionswithotherprogramsandoperatingsystems,searchingforothermachines,subvertingthefunctionofexistingprograms,installingbackdoorsonsystems,andsoon.
Overitslifetimeacelldifferentiatesalongaparticularpathway,witheachdifferentiationstagealongthispathwayrepresentingaspecificcelltype.Allcellsatthesamestageofdifferentiationareofthesametypeandhavethesamephenotypicconfigurationandfunctionalcharacteristics.Whichpathwayacellfollowsistheresultoftheenvironmentalpressuresthecellexperiences.Littleofthedynamicsoftheimmunesystemcanbecapturedifagentsinartificialimmunesystemmodelsdonotpossesssimilardevelopmentalcharacteristics.Thiscouldbemodelledbyendowingagentswithasetoffunctions,subsetsofwhichtheagentperformsatanyonetimeandwhichrepresentthecurrenttypeoftheagent.Transitionsfromonetypetoanotherarearesultofinteractionsoftheagentwithitsenvironmentandcouldbepicturedasabranchingtreestructure.
Whilecellsactasindividuals,differentiatingalongtheirownindividualpath-ways,theyalsoactaspartofagroup.Atthispopulationlevel,consideringtheinnateimmunesystemhighlightstheneedforgroupsofagentswhichrespondtodifferenttypesofinformation.Certainagentsmightidentifyfixedpatternsinthisinformation,embodyingsometypeofnotionalTLR,whileotherswouldidentifyvariablepatterns,akintoTCRs.Theprocesseswhichdrivethespecificityofreceptorsmaybeadaptiveorevolutionary,withdifferentpressuresbiasingthetypeofinformationsurveyedbyagents.
Cellscontrolothercellsonanindividualcontact-dependantlevel.Theyalsocontrolcellsinalocalneighbourhoodthroughtheproductionofcytokines.Thislocalisedcontrolleadstodynamicalpatternsatthepopulationlevel.DCcontrolofThproliferationthroughcostimulatorymoleculesisagoodexampleofhowlocalinteractionscontrolthepopulationofThcellsanddeterminepopulation-
levelphenomenasuchasclonaldistribution.Effectsoftheartificialtissueononegroupofagentsshouldhaveresultingeffectsonpopulationsofotheragents.ThegenerationofperipheraltolerancebyDCssuggestsamechanismbywhichsignalspresentedbytheartificialtissuearereceivedbyonegroupofagentsandhaveadirecteffectonothergroupsofagents.Thiscontrolmightnotbeasclearcutasliveordie,butmoreadirectionofdifferentiationpathways,ofwhichpolarisationofThcellsbyTLRsonDCsisagoodexample.
Lastly,mechanismsoftrustorobligationareestablished.TheNK“missingself”modelisagoodexampleofthis.TheprovisionofsufficientquantitiesofMHCcanbeseenasamonitoringrequirement,imposedbyNKcells,ofthesystem.IftissuecellsfailtoprovideMHCtheyaredestroyed.Inrealtimemonitoringsituations,modelsofsuchasuppression-basedmechanismmightbeusedtoestablishifgroupsofdataprovidingagentsarefunctioning.3.4
Summary
Usingthebiologyoftheprevioussectionasabasis,thissectionhassketchedoutameta-frameworkformodelsofinnateimmunity,discussinggeneralpropertiesofsuchmodelsandalsohowtheymightberealisedmoreconcretely.Whilethepropertiespresentedhavetriedtocapturethecorefeaturesofinnateimmunity,duetospaceandintellectualconstraintstheyarenotexhaustiveandneedtobecombinedwithexistingframeworksofadaptivemodels[5]ifintegratedmodelsaretobebuilt.
4Conclusion
Thispaperhaspresentedasummaryofcurrentbiologicalunderstandingoftheinnateimmunesystem,contrastingitwiththeadaptiveimmunesystem.Adopt-ingaconceptualframeworkitthenproposedandrefinedameta-frameworkforartificialsystemsincorporatingideasfrominnateimmunity.Whileemphasisingtheroleofinnateimmunity,inreality,theinnateandadaptivesystemsarein-timatelycoupledandworktogethertoprotectthehost.Asalreadysuggested,combiningthepropertiessuggestedherewiththoseoftraditionalpopulationandnetworkmodelswouldenableartificialsystemstomorecloselyreflecttheirbiologicalcounterparts.
Otherpossibilitiesforfutureworkincludeareviewwithintheproposedframeworkofartificialimmunesystemmodelssuchas[29,30]whichalreadyincludeinnateimmunity.Thiswouldhelpevaluateandcomparethesemodels,discerningcommonalitiesandprovidingdirectionforfutureresearch.Developingmoredetailedmathematicalandcomputationalmodelswouldbeanimportantnextstepinamoredetailedunderstandingofthepropertiesofinnateimmunity.Thesemodelscouldthenbeusedtoinstantiatearangeofsystemsindifferentapplicationdomains.Morerealisticandprincipledmodelscouldalsoextendun-derstandingonthedynamicsofcompetingimmunologicalmodelssuchasthose
ofinstructiveorconsensualregulationofTh1/2responses,ormodulationofcostimulatorysignals.
Couchingideasofinnateimmunitywithinanacceptedconceptualframeworkprovidesastepindevelopingmoreintegratedartificialimmunesystemmodelswhichtakeintoaccountthekeyroletheinnateimmunesystemplaysinhostpro-tection.Asalways,thebeautyandsubtletyoftheimmunesystemwillcontinuetoprovidearichsourceofinspirationfordesignersofartificialsystems.
5Acknowledgements
ManythankstoAdrianRobinsfordiscussionsontheimmunology,andtomycoworkersUweAickelin,JulieGreensmith,JungwonKim,JulieMcLeod,SteveCayzer,RachelHarry,CharlotteWilliams,GianniTedescoandPeterBentley,withoutwhomthisworkwouldnothavebeenpossible.ThisresearchissupportedbytheEPSRC(GR/S47809/01).
References
1.Medzhitov,R.,Janeway,C.A.:Innateimmunity.TheNewEnglandJournalofMedicine343(2000)338–344
2.Germain,R.N.:Aninnatelyinterestingdecadeofresearchinimmunology.NatureMedicine10(2004)1307–1320
3.Aickelin,U.,Greensmith,J.,Twycross,J.:Immunesystemapproachestointrusiondetection-areview.InGNicosiaetal.,ed.:Proc.oftheThirdInternationalCon-ferenceonArtificialImmuneSystems.Number3239inLectureNotesinComputerScience,Springer(2004)316–329
4.deCastro,L.N.,Timmis,J.:ArtificialImmuneSystems:ANewComputationIntelligenceApproach.Springer(2002)
5.Stepney,S.,Smith,R.,Timmis,J.,Tyrrell,A.,Neal,M.,Hone,A.:Conceptualframeworksforartificialimmunesystems.InternationalJournalofUnconventionalComputing1(2005)
6.Janeway,C.A.,Medzhitov,R.:Innateimmunerecognition.AnnualReviewofImmunology20(2002)197–216
7.Janeway,C.A.,Travers,P.,Walport,M.,Shlomchik,M.:Immunobiology.5theditionedn.GarlandScience,availableonlineathttp://www.ncbi.nlm.nih.gov/books/(2001)
8.Alberts,B.,Johnson,A.,Lewis,J.,Raff,M.,Roberts,K.,Walter,P.:MolecularBiologyoftheCell.4thedn.GarlandScience,availableonlineathttp://www.ncbi.nlm.nih.gov/books/(2002)
9.Lodish,H.,Berk,A.,Matsudaira,P.,Kaiser,C.A.,Krieger,M.,Scott,M.P.,Zipursky,L.,Darnell,J.:MolecularCellBiology.4ndedn.W.H.FreemanandCo.,availableonlineathttp://www.ncbi.nlm.nih.gov/books/(1999)
10.Ljunggren,H.G.,Karre,K.:Insearchofthe‘missingself’:MHCmoleculesand
NKcellrecognition.ImmunologyToday11(1990)237–244
11.Germain,R.N.:MHC-dependentantigenprocessingandpeptidepresentation:
ProvidingligandsforTlymphocyteactivation.Cell76(1994)287–299
12.Lorenzo,M.E.,Ploegh,H.L.,Tirrabassi,R.S.:Viralimmuneevasionstrategiesand
theunderlyingcellbiology.SeminarsinImmunology13(2001)1–9
13.Lanier,L.L.:Naturalkillercellreceptorsignaling.CurrentOpinioninImmunology
15(2003)308–314
14.Kapsenberg,M.L.:Dendritic-cellcontrolofpathogen-drivenT-cellpolarization.
NatureReviewsinImmunology3(2003)984–993
15.ReiseSousa,C.:Toll-likereceptorsanddendriticcells:fromwhomthebugtolls.
SeminarsinImmunology16(2004)27–34
16.Pulendran,B.,Kumar,P.,Cutler,C.W.,Mohamadzadeh,M.,VanDyke,T.,
Banchereau,J.:Lipopolysaccharidesfromdistinctpathogensinducedifferentclassesofimmuneresponsesinvivo.TheJournalofImmunology167(2001)5067–5076
17.Kalinski,P.,Moser,M.:Consensualimmunity:success-drivendevelopmentofT-helper-1andT-helper-2responses.NatureReviewsinImmunology5(2005)251–260
18.Heath,W.R.,Villadangos,J.A.:Nodrivingwithoutalicense.NatureImmunology
6(2005)125–126
19.Baxter,A.G.,Hodgkin,P.G.:Activationrules:thetwo-signaltheoriesofimmune
activation.NatureReviewsinImmunology2(2002)439–446
20.Mondino,A.,Khoruts,A.,Jenkins,M.K.:TheanatomyofT-cellactivationand
tolerance.Proc.oftheNationalAcademyofScienceUSA93(1996)2245–225221.Kyewski,B.,Derbinski,J.:Self-representationinthethymus:anextendedview.
NatureReviewsinImmunology4(2004)688–698
22.Scheinecker,C.,McHugh,R.,Shevach,E.M.,Germain,R.N.:Constitutivepresen-tationofanaturaltissueautoantigenexclusivelybydendriticcellsinthedraininglymphnode.TheJournalofExperimentalMedicine196(2002)1079–109023.Sp¨orri,R.,ReiseSousa,C.:Inflammatorymediatorsareinsufficientforfullden-driticcellactivationandpromoteexpansionofCD4+Tcellpopulationslacking
helperfunctions.NatureImmunology6(2005)163–170
24.Lutz,M.B.,Schuler,G.:Immature,semi-matureandfullymaturedendriticcells:
whichsignalsinducetoleranceorimmunity?TrendsinImmunology23(2002)445–449
25.Mahnke,K.,Knop,J.,Enk,A.H.:InductionoftolerogenicDCs:‘youarewhatyou
eat’.TrendsinImmunology24(2003)646–651
26.Sharpe,A.H.,Freeman,G.J.:TheB7-CD28superfamily.NatureReviewsinIm-munology2(2002)116–126
27.Carreno,B.M.,Collins,M.:TheB7familyofligandsanditsreceptors:newpath-waysforcostimulationandinhibitionofimmuneresponses.AnnualReviewofImmunology20(2002)29–53
28.Watkins,A.,Timmis,J.:ExploitingparallelisminherentinAIRS,anartificial
immuneclassifier.InGNicosiaetal.,ed.:Proc.oftheThirdInternationalConfer-enceonArtificialImmuneSystems.Number3239inLectureNotesinComputerScience,Springer(2004)427–438
29.Zhang,X.,Dragffy,G.,Pipe,A.G.,Zhu,Q.M.:Artificialinnateimmunesystem:
Aninstantdefencelayerofembryonics.InGNicosiaetal.,ed.:Proc.oftheThirdInternationalConferenceonArtificialImmuneSystems.Number3239inLectureNotesinComputerScience,Springer(2004)302–315
30.Na,D.,Park,I.,Lee,K.,Lee,D.:Integrationofimmunemodelsusingpetrinets.
InGNicosiaetal.,ed.:Proc.oftheThirdInternationalConferenceonArtificialImmuneSystems.Number3239inLectureNotesinComputerScience,Springer(2004)205–216
因篇幅问题不能全部显示，请点此查看更多更全内容