fugedios11

MUCOSALIMMUNOLOGYIntestinalepithelialexosomescarryMHCclassII/peptidesabletoinformtheimmunesysteminmiceGVanNiel,JMallegol,CBevilacqua,CCandalh,SBrugie`re,ETomaskovic-Crook,JKHeath,NCerf-Bensussan,MHeymanSeeendofarticleforauthors'affiliationsCorrespondenceto:DrMHeyman,INSERMEMI0212,FaculteNecker-EnfantsMalades,156ruedeVaugirard,75730ParisCedex15,France;heyman@necker.
frAcceptedforpublication23July2003Gut2003;52:1690–1697Background:Intestinalepithelialcellssecreteexosome-likevesicles.
Theaimofthisstudywastocharacterisemurineintestinalepithelialexosomesandtoanalysetheircapacitytoinformtheimmunesysteminvivoinmice.
Methods:EpithelialexosomeswereobtainedfromthemurineepithelialcelllineMODEKincubatedinthepresenceorabsenceofinterferonc(IFN-c)togetherwithpepsin/trypsinovalbuminhydrolysate(hOVA)tomimicluminaldigestion.
ExosomesisolatedfromMODEKconditionedmedia(EXO-hOVAandEXO-hOVA-IFN)werecharacterisedbywesternblot,peptidemapping,andmassspectrometry.
TheywereinjectedintraperitoneallytoC3H/HeNmicetotesttheirimmunocompetence.
Results:MODEKepithelialexosomesdisplayedmajorhistocompatibilitycomplex(MHC)classIandclassII(upregulatedbyIFN-c)moleculesandtetraspanproteins(CD9,CD81,CD82)potentiallyinvolvedinthebindingtotargetcells.
A33antigen,anIg-likemoleculehighlyspecificforintestinalepithelialcells,wasenrichedinexosomesandwasalsofoundinmicemesentericlymphnodes,suggestingexosomemigrationtowardsthegutassociatedlymphoidtissues.
IntraperitonealinjectionofEXO-hOVAorEXO-hOVA-IFNdidnotinducehumoralorcellulartolerancetoOVAinmice.
Incontrast,exosomesobtainedafterincubationwithIFN-c(EXO-hOVA-IFN),bearingabundantMHCclassII/OVAcomplexes,inducedaspecifichumoralimmuneresponse.
Conclusions:EpithelialexosomesareantigenpresentingvesiclesbearingMHCclassII/peptidecomplexesthatprimeforanimmunogenicratherthantolerogenicresponseinthecontextofasystemicchallenge.
Intheintestine,boththemucosalmicroenvironmentandlocaleffectorcellsareprobablykeyplayersindeterminingtheoutcomeoftheimmuneresponsetoexosomederivedepitopes.
Thehighconcentrationofforeignantigenspresentinthegutlumenisseparatedfromthelocalintestinalimmunesystembyasinglelayerofpolarisedepithelialcells.
Intestinalepithelialcells(IEC)canendocytosedietaryproteins,processthemintopeptides,1andinvitrocanpresentthesepeptidestoCD4+Tlymphocytes,particularlywhenmajorhistocompatibilitycomplex(MHC)classIImoleculesareupregulated,2–4andinthepresenceofaccessorymole-cules.
56However,invivo,directinteractionsbetweenIECandCD4+Tcellspredominatinginthelaminapropriaarelimitedduetothepresenceofthebasementmembrane.
IEChavebeenshowntoreleaseexosome-likevesicles7morpho-logicallysimilartotheexosomessecretedbyprofessionalantigenpresentingcells(APC)suchasBcells8ordendriticcells.
9EpithelialexosomesreleasedfrombasolateralsidesofIECbearhighamountsofMHCclassIandclassIImolecules,suggestingthattheymayinteractwiththeunderlyingimmunesystem.
TheyalsocontainA33antigen,7amoleculewhoseexpressionisrestrictedessentiallytotheintestinalepithelium.
1011ExosomesderivedfromprofessionalAPCmoderatelyactivateCD48orCD812Tlymphocytes,anditwasrecentlyshownthattheirantigenpresentingcapacityhighlydependedupondendriticcells,14suggestingthatexosomeshavetobetakenupbyprofessionalAPCtoinduceefficientTcellactivation.
TheymaythereforebeinvolvedintherapiddisseminationofimmuneinformationviaAPC.
Inthepresentstudy,wetestedthehypothesisthatintestinalepithelialexosomesbearingMHCclassII/ovalbu-min(OVA)peptidecomplexes,generatedinthepresenceofinterferonc(IFN-c)andOVAthathadbeenpredigestedwithgastricandpancreaticenzymes,cantransmitimmuneinformationandtherebyactassensorsoftheantigenicinformationpresentintheintestinallumen.
MurineIECderivedexosomeswerecharacterisedandtheirinteractionwiththeimmunesystemwastestedinvivoinmice.
WetestedthepossibilitythatintraperitonealinjectionofOVAloadedepithelialexosomescouldeither:(1)mimicoralingestionofOVAandinhibitspecifichumoraland/orcellularimmuneresponses(oraltolerance),aspreviouslysuggested15or(2)directlyprimethehumoralimmuneresponsetoOVA(sensitisation).
METHODSAntigensWeincubatedepithelialcellsinthepresenceofOVAthathadbeenpredigestedwithpepsinandtrypsintomimicthefateofthisproteinintheintestinallumen.
OVA10mg/mlwastreatedforfourhours(37C,pH1.
8)withpepsin(20mg/gprotein;Sigma,StLouis,Missouri,USA)andtrypsin(20mg/gprotein)atpH7.
8.
Enzymeswereinactivatedfor45minutesat85Candeliminatedbyultrafiltrationonmembranesretainingmoleculesgreaterthan10000DaAbbreviations:IFN-c,interferonc;APC,antigenpresentingcell;IEC,intestinalepithelialcells;OVA,ovalbumin;hOVA,ovalbuminhydrolysate;MHC,majorhistocompatibilitycomplex;MALDI-TOF-MS,matrixassistedlaserdesorptionionisation-timeofflight-massspectrometry;DMEM,Dulbecco'smodifiedEagle'smedium;MLN,mesentericlymphnodes;SDS-PAGE,sodiumdodecylsulphate-polyacrylamidegelelectrophoresis;CFA,completeFreund'sadjuvant;DTH,delayedtypehypersensitivity;PBS,phosphatebufferedsaline1690www.
gutjnl.
comonJanuary7,2021byguest.
Protectedbycopyright.
http://gut.
bmj.
com/Gut:firstpublishedas10.
1136/gut.
52.
12.
1690on21November2003.
Downloadedfrom(centriplus,YM10;Millipore,StQuentinenYuelines,France).
Thefiltrate,referredtoasovalbuminhydrolysate(hOVA),wassterilisedandfrozenuntiluse.
Murineintestinalepithelialcellline(MODEK)ThemurineintestinalepithelialcelllineMODEK,16kindlyprovidedbyDKaiserlian,wasused.
ThiscelllineisderivedfromC3H/HemiceandexpressesI-AkMHCclassIImolecules.
CellswereculturedinDulbecco'smodifiedEagle'smedium(DMEM;Gibco,LifeTechnologies,Cergy-Pontoise,France)containing10%heatinactivatedfetalcalfserum,4mMglutamine,and50mg/mlgentamicin,inahumidifiedatmosphereof5%CO2/95%air.
ExosomepreparationfromMODEKepithelialcelllineconditionedmediaMODEKcellswereseeded(56106cells)into75cm2FalconflasksinDMEMcontaining10%Prolifix(syntheticserum;Biomedia,France),4mMglutamine,50mg/mlgentamicin,andOVA(10mM)orOVApredigestedwithpepsinandtrypsintomimicdegradationintheintestinallumen(hOVAcorrespondingto10mMOVA).
Fourdayslater,cellsuper-natantswerecollectedtoisolateexosomesunderbasalconditions(EXO-OVAorEXO-hOVA).
ToobtainexosomesfromMHCclassIIenrichedenterocytes,MODEKcellswerefurtherincubatedfor48hourswithOVAorhOVAinthepresenceofmurineIFN-c(100U/ml)beforeexosomeswerecollected(EXO-OVA-IFNorEXO-hOVA-IFN).
ExosomesobtainedunderbasalandIFN-cconditionswereisolatedandquantifiedaspreviouslydescribed.
7Generally,10mgMODEKexosomeswerecollectedfromthe48hourconditionedmediumfroma75cm2cultureflask.
Insomeexperiments,exosomeswereisolatedfromhumanormiceserumtoanalysewhetherintestinalepithelialexosomescanaccesstheperipheralbloodstream.
SerumexosomalproteincontentwasestimatedbywesternblotanalysisusingtheA33antigen,aspecificmarkerofintestinalepithelialexosomes.
IdentificationofepithelialexosomesinmousemesentericlymphnodesWelookedforthepresenceofA33antigeninmesentericlymphnodes(MLN)totestthehypothesisthatepithelialexosomesreleasedatthebasalpoleofenterocytescanmigrate(aloneorboundtodendriticcells)towardsthesesecondarylymphoidorgans.
AsA33antigenishighlyexpressedinIEC,11acomparisonbetweenMLNandfreshlyisolatedmouseIECwasperformed.
IECorMLNwerehomogenisedin1%TritonX100in10mMHEPESbufferedsaline,pH7.
4,containingaproteaseinhibitorcocktail.
Thelysateswereincubatedfor30minutesat4Candultracen-trifugedat1000006gforonehour.
PartiallypurifiedproteinswerethenpreparedusinganionexchangehighpressureliquidchromatographybyinjectingtheresultantsupernatantontoaMono-QHR10/10columnatpH7.
8,andbyelutingtheproteinsfromthecolumnusingalinear0–1MNaClgradient.
TheA33antigen(molecularweight,50kDa)ineluantfractionswasdetectedbywesternblotanalysisusingpolyclonalrabbitanti-mA33antigenIgG(1.
2mg/ml).
ControlstoconfirmthespecificityoftheproteindetectionincludedMLNharvestedfromA33antigendeficientmice(A332/2;TebbuttNC,ErnstM,andHeathJK,unpublisheddata)andpreabsorptionoftheanti-mA33antigenantibodieswiththeimmunisingA33antigenCterminalpeptide.
11ImmunohistochemicaldetectionofthemurineA33antigenwasalsoperformedonsectionsofparaffinembeddedMLNfromnormalmiceusingthesamerabbitpolyclonalantibody.
CharacterisationofMODEKexosomesbywesternblotanalysisMODEKcellswerelysedover30minutesat4Cin50mMTris-HCl,pH8,150mMNaCl,1%EDTA,and0.
5%TritonX100,containingaproteaseinhibitorcocktailandcentri-fugedfor10minutesat700g(4C)toremovenuclei.
Thesupernatantswereultracentrifugedforonehourat100000g(4C).
Supernatantscontainingwholecellproteinswereadjustedto2mg/mlinphosphatebufferedsaline(PBS),diluted(1/2)insamplebuffer(26),andstoredat220C.
Exosomalproteinswereseparatedbysodiumdodecylsulphate-polyacrylamidegelelectrophoresis(SDS-PAGE)in10%acrylamidegels.
Gelsweretransferredontonitrocellu-losemembranes(AmershamPharmaciaBiotech,AmershamBiosciencesEurope,Saclay,France)in10mMTris,0.
2Mglycine,and30%methanol.
Non-specificbindingsiteswereblockedbyincubationinTris-HCl,pH7.
6,containing5%bovineserumalbuminand0.
1%Tween.
Blotswereincubatedforonehourwithmouseanti-I-A(mouse/rat)antibodies(cloneMRCOX-6;SerotecCergy-StChristophe,France)todetectMHCclassIImoleculesandwithperoxidasecon-jugatedsheepantimouseIg(Amersham)(1:20000).
A33antigenwasdetectedwithpolyclonalrabbitantimouseA33antigenantibodiesinnon-reducingconditions.
11TheseantibodiesweredetectedusingperoxidaseconjugatedgoatantirabbitIgG(BioradIvrysurSeine,France)(1:15000),followedbyECL(AmershamPharmaciaBiotech).
MODEKderivedexosomes:peptidemappinganalysisusingmatrixassistedlaserdesorptionionisation-timeofflight-massspectrometry(MALDI-TOF-MS)IdentificationofproteinspresentinMODEKderivedexosomeswasanalysedusingMALDI-TOF-MS.
MODEKderivedexosomes(20mg),preparedinsamplebuffer,wererunon10%SDS-PAGEandstainedwithCoomassieblue.
Themajorbandswereexcisedfromthegelandin-geldigestedwithtrypsin.
ThedigestsolutionwasanalysedbyMALDI-TOF-MSusingaBrukerBiflexmassspectrometer(Bruker-FranzenAnalytikBremen,Germany),aspreviouslydescribed.
17Monoisotopicpeptidemasseswereassignedandusedindatabasesearches.
Typicalsearchparameters,usingtheMicrosoftFITprogram,wereasfollows:maximumallowedpeptidemasserrorof100ppm;considerationofoneincompletecleavageperpeptide;norestrictionwasplacedontheisoelectricpointoftheprotein;andaproteinmassrangeof0–200kDawasallowed.
InvivostudiesinmiceWetestedthecapacityofMODEKepithelialexosomestoinducespecifictoleranceorsensitisationinmice.
Oraltolerance(humoralresponse)FiftytwoC3H/HeNmice(threeweekoldfemales)weredividedintosixexperimentalgroupscomprising5–10miceeach.
StandardoraltoleranceprotocolThefirsttwogroupscomprisedtolerisedandsensitisedmice,obtainedusingstandardprotocolsoforaltolerance.
OVAtolerisedmice(TOL)received20mgOVAin200ml0.
2Msodiumbicarbonatebyintragastricgavageusingastainlesssteelbluntfeedingneedle.
OVAsensitisedmice(SENS)receivedbicarbonatealonebygastricgavage.
Oneweeklater(day8),allmicewereimmunised:theyreceivedasub-cutaneousinjection(100ml),tothebaseofthetail,of25mgOVA/completeFreund'sadjuvant(CFA,v/v)andasecondinjection(100ml)of10mgOVAinPBStwoweekslater.
Oneweekafterthesecondinjection,bloodwascollectedforanti-OVAIgGandIgEmeasurements.
Functionofintestinalepithelialexosomes1691www.
gutjnl.
comonJanuary7,2021byguest.
Protectedbycopyright.
http://gut.
bmj.
com/Gut:firstpublishedas10.
1136/gut.
52.
12.
1690on21November2003.
DownloadedfromInductionoftolerancebyepithelialexosomesTotestforthetolerogeniccapacityofexosomes,fouradditionalgroupsweretestedinparallelwiththefirsttwogroups.
Inthesegroups,OVAloadedexosomeswereadministeredintraperitoneallywiththeaimofmimickingoraladministrationofOVA.
EXO-OVAmicereceived,insteadofOVAgavage,anintraperitonealinjectionof10mgexosomes(100ml)derivedfromtheMODEKcelllineculturedinthepresenceofOVAunderbasalconditions.
EXO-OVA-IFNmicereceivedanintraperitonealinjectionof10mgOVA-exosomesderivedfromtheMODEKcelllineculturedinthepresenceofOVAand100U/mlIFN-c.
Insomeexperiments,hOVAwasusedinsteadofintactOVAtoproduceexosomesfromMODEKcellsincubatedinthepresenceorabsenceofIFN-c.
Miceinthesegroupsreceivedanintraperitonealinjectionof10mgexosomesandarereferredtoasEXO-hOVAandEXO-hOVA-IFN,respectively.
Afteroneweek,allmicewereimmunisedasstatedaboveandbloodwascollectedforanti-OVAIgGandIgEmeasurements.
Oraltolerance(DTHresponse)Tolerised(TOL)andsensitised(SENS)miceweregivenOVAorbicarbonatebuffer,respectively,bygastricgavage,asdescribedabove(seehumoralresponsesectionabove).
TheothermicereceivedintraperitonealOVAexosomesinsteadofOVAbygavage:EXO-hOVAmicereceivedanintraperitonealinjectionof10mgexosomesderivedfromMODEKcellsfedhOVAunderbasalconditionsandEXO-hOVA-IFNmicereceivedanintraperitonealinjectionof10mgexosomesderivedfromMODEKcellsfedhOVAinthepresenceofIFN-c.
Afteroneweek,allmicewereinjectedsubcutaneously(baseofthetail)with100mgOVA/CFA.
Twoweekslater,micereceivedasubcutaneousinjectionof30mlOVA(1mg/ml)intheearandearswellingwasmeasuredasthedifferenceinearthicknessbeforeand24hoursafterOVAinjectionusingamicrometrecalliper(Oditest,Kro¨plin,Germany)toquantifythedelayedtypehypersensitivity(DTH)response.
Sensitisationprotocol(humoralresponse)Toanalysewhetherintestinalepithelialexosomescoulddirectlyprimetheimmunesystem,micereceivedanintraperitonealinjectionofexosomes.
EXO-hOVAmicereceivedaninjectionof10mgexosomes(100ml)derivedfromMODEKcellsfedhOVAintheabsenceofIFN-cwhereasEXO-hOVA-IFNmicereceivedaninjectionof10mgexosomesderivedfromMODEKcellsfedhOVAinthepresenceofIFN-c.
Acontrolgroupreceivedanintraperito-nealinjectionof100mlPBSinsteadofexosomes.
Afteroneweek,allmicegroupswereinjectedsubcutaneouslywith25mgOVA/CFAin100ml.
Oneweeklatermicewerebledandanti-OVAIgGandIgEantibodyresponsesweremeasured.
Inanotherexperiment,thecapacityofOVAloadedepithelialexosomesandfreeOVApeptidestostimulatetheimmunesystemwerecompared:0.
2mgofpepsin/trypsinovalbuminpeptides(hOVA),calculatedasbeingequivalenttotheamountloadedon10mgepithelialexosomesinIFN-cconditions,wereinjectedintraperitoneallyinordertoprimethemicebeforethesubcutaneousOVAboost.
MeasurementofOVAspecificIgEandIgGantibodies(ELISA)IgGandIgEtitresinplasmaweremeasuredusinganenzymelinkedimmunosorbentassayasfollows:96wellplates(ImmulonII,Labsystems,France)werecoatedovernightat4Cwith100mlof25mg/mlOVA(Sigma)inPBS.
AfterwasheswithPBS-0.
05%Tween20(PBS-Tween),non-specificbindingsiteswereblockedforonehourat37Cwith100ml/wellofPBS-1%bovineserumalbumin.
Theplateswerethenincubatedfortwohoursat37Cwith100mlplasmasamplesseriallydilutedinPBS-Tween.
PeroxidaseconjugatedratantimouseIgG(Serotec)diluted1:1000inPBS-Tweenwereadded(100ml/well)andincubatedfortwohoursat37C.
Finally,100mlofcitratebuffer,pH4.
0,containing10mgortho-phenylenediamine(Sigma)andH2O2(Sigma)wereaddedtoeachwell.
Thereactionwasstoppedwith50ml/well6NH2SO4andplateswerereadat450nmusinganautomaticMultiskanmicroplatereader(Labsystems,Cergy-Pontoise,France).
StatisticalanalysisStatisticalanalysiswasperformedusingSASsoftware.
Resultsareexpressedasmean(SD).
Comparisonofmeanswasperformedusinganalysisofvarianceandnon-para-metricWilcoxontests.
Thegenerallinearmodelprocedurewasalsousedformultiplegrouptogroupcomparisons.
Differenceswereconsideredsignificantwhenp,0.
05.
RESULTSCharacterisationofMODEKintestinalepithelialexosomesTwomoleculesfoundinhumanepithelialexosomes,7namelyMHCclassIImoleculesandA33antigen,wereanalysedbywesternblot.
Asshowninfig1A,MHCclassIImoleculeswerenotdetectableinMODEKcelllysates,eveninthepresenceofIFN-c.
MHCclassIImoleculesweredetectableinMODEKderivedexosomesharvestedunderbasalconditionsandtheirexpressionwasmarkedlyupregulatedinthepresenceofIFN-c,indicatingsubstantialenrichmentinepithelialexosomes.
ThediscrepancybetweenexpressionofMHCclassIImoleculesinentirecellsandexosomesmaybeduetotheantibodiesdifferentiallyrecognisingthematurepeptideboundclassIImoleculespossiblypresentonexosomescomparedwithnascentclassIIformsmainlypresentinepithelialcells.
A33antigenwasdetectedincelllysatesandwashighlyenrichedinexosomesalthoughitsexpressioninexosomeswasnotupregulatedbyIFN-c(datanotshown).
ModeK/IFNcelllysate5gModeKexosomes5gModeK/IFNexosomes5gModeKcelllysate5gModeKexosomes5gMHCII(40kDa)A33antigen(43kDa)Figure1Expressionofmajorhistocompatibilitycomplex(MHC)classIImoleculesandA33antigeninMODEKcellsincubatedinthepresenceorabsenceofinterferonc(IFN-c),andinMODEKcellderivedexosomes.
AlthoughMODEKcelllysatesdidnotexpressdetectablelevelsofMHCclassIImolecules,eveninthepresenceofIFN-c,MODEKderivedexosomesexpresseddetectablelevelsunderbasalconditions,whichweremarkedlyupregulatedinthepresenceofIFN-c.
A33antigen,amoleculeexpressedspecificallybyintestinalepithelialcells,wasenrichedinexosomepreparationscomparedwithwholecelllysates.
1692VanNiel,Mallegol,Bevilacqua,etalwww.
gutjnl.
comonJanuary7,2021byguest.
Protectedbycopyright.
http://gut.
bmj.
com/Gut:firstpublishedas10.
1136/gut.
52.
12.
1690on21November2003.
DownloadedfromPeptidemappingbyMALDI-TOF-MSanalysisTheresultsofpeptidemappingofMODEKderivedexosomesaresummarisedintable1.
Severalproteinfamilieswereobserved:(a)proteinsplayingaroleinthebiogenesisandreleaseofexosomes,suchasintegrinprecursorsandcytoskeletonproteinsunderbasalandinflammatory(+IFN-c)conditionsandRasrelatedprotein(Rab-2,Rab-7);(b)proteinsresponsiblefordirectingexosomestotheirtargetcells,suchasMFG-E8,aproteinpreviouslyfoundindendriticcellderivedexosomes,18andheatshockproteinshsp71andhsp90thatmayplayaroleinthetransferofantigenicpeptidesandbindingtodendriticcells,andintheirmaturation,asrecentlydescribedinmastcellderivedexosomes19;(c)proteinspotentiallyinvolvedinantigenpresentation(MHCclassImolecules)orinexosomefunction(tetraspanmoleculesCD81,CD82,andundescribedtrans-membranefoursuperfamilyprotein);and(d)cytoplasmicproteinsprobablynotrelatedtoexosomefunctionwhichmayhavebeennon-specificallycapturedintotheexosomelumenduringtheirformation(histones,enzymes).
EpithelialexosomesdonotinducetoleranceinmiceHumoraltoleranceAnearlierstudysuggestedthatepitheliumderivedexosomesmightinduceoraltoleranceinrats.
15Incontrast,inourmurinemodel,MODEKderivedexosomesbearingMHCclassII/OVAmoleculesandinjectedintraperitoneallydidnotsuppressIgEandIgGantibodyresponses(fig2A,B).
Asexpected,anti-OVAantibodiesweresignificantlylowerintolerised(TOL)mice(60and3950forIgEandIgGtitres,respectively)thaninsensitised(SENS)(480and192000,forIgEandIgGtitres;p,0.
001)mice,butIgtitresobservedinEXO-OVAorEXO-OVA-IFNmicewerenotsignificantlydifferentfromthoseobservedinSENSmice,indicatingthatepithelialexosomesdidnottolerisemicetoOVA.
Furthermore,anti-OVAantibodytitresweresignificantlyhigherinEXO-hOVA-IFNmicethaninSENSmice(IgE6.
156103v0.
486103(p,0.
0001);andIgG9006103v1926103(p,0.
0001)),suggestingthatexosomesmaysensi-tiseratherthantolerisemice,atleastinourexperimentalconditions.
Table1PeptidemappingofMODEKderivedepithelialexosomesusingmatrixassistedlaserdesorptionionisation-timeofflight-massspectrometry(MALDI-TOF-MS)analysisBandNoProteinsidentifiedfromMODEKderivedexosomesMolecularmass(kDa)ProteinsidentifiedfromMODEK+IFN-cderivedexosomesMolecularmass(kDa)BandM-1A-Xactin,Q6127641.
7ClathrinheavychainP11442193.
2SimilartoubiquitinCP11276Q922Z865.
5kD273.
3FibronectinprecursorP11276273.
3HeatshockproteinHSP90-alphaP0790185.
5HeatshockproteinHSP71kDP0810971HeatshockproteinHSP71kDP0810971HeatshockproteinHSP90-betaproteinP34058HeatshockproteinHSP90-betaproteinP3405885.
5BandM-2SimilartoubiquitinC,Q922Z865.
5BandM-3Integrinbeta-1precursor,P0905591.
7Integrinbeta-1precursor,P0905591.
7Integrinalpha-3precursor,Q62470117.
8BandM-4ProstaglandinF2-alphareceptorregulatoryprotein,Q9WV9199.
6ProstaglandinF2-alphareceptorregulatoryprotein,Q9WV9199.
6BandM-5SerineproteaseHTRA1precursor,Q9R11851.
2Moesin,P2604167.
6BandM-6Moesin,P2604167.
6BandM-7EH-domaincontainingprotein1,Q9WVK460.
6BandM-8Pyruvatekinase,M2isozyme,P5248057.
7BandM-922kDaNeuronaltissue-enrichedacidicprotein.
,Q91XV322.
222kDaNeuronaltissueenrichedacidicprotein,Q91XV322.
2Pyruvatekinase,P5248057.
7CD82antigen,P4023729.
6CD82antigen,P4023729.
6BandM-10MilkfatglobuleglycoproteinMFG-E8S48.
8AlphaEnolase,P1718247.
3CalpactinIlightchain,P0820711.
1AnnexinII,P0735638.
8BandM-11AnnexinIP1010738.
8H-2classIhistocompatibilityantigen,P0422341.
6KeratineP0253557.
8BandM-12AnnexinA4(AnnexinIV),P9742936.
1AnnexinI(LipocortinI)P1010738.
8BandM-13Transmembrane4superfamily,Q6430223.
1AnnexinI,P1010738.
8BandM-14AnnexinV,P4803635.
7CD9antigen,P4024025.
0AnnexinI,P1010738.
5CD81antigen,P3576225.
8AnnexinA4,P9742935,8HistoneH1.
3,P1586422.
2HistoneH1.
5,P4327622.
4HistoneH1.
4,P4327421.
8BandM-15AnnexinA4,P9742935.
8AnnexinI,P1010738.
7BandM-16HistoneH3,P0230114.
914-3-3proteinzeta/delta,P3521527.
7HistoneH1.
1,P1586421.
1BandM-17Ferritinlightchain1,P2939120.
6RasrelatedproteinRab-2,P5399423.
5RasrelatedproteinRab-7,P5115023.
5Myeloidassociateddifferentiationmarker,O3568232.
08BandM-18HistoneH1.
3,P4327721.
9BandM-19HistoneH2BF,P1085313.
8Ferritinheavychain,P0952821.
0MODEKderivedexosomes(20mg)wererunon10%sodiumdodecylsulphate-polyacrylamidegelelectrophoresisandstainedwithCoomassieblue.
Themajorbandswereexcisedfromthegelandin-geldigestedwithtrypsin.
ThedigestsolutionwasanalysedbyMALDI-TOF-MS.
IFN-c,interferonc.
Functionofintestinalepithelialexosomes1693www.
gutjnl.
comonJanuary7,2021byguest.
Protectedbycopyright.
http://gut.
bmj.
com/Gut:firstpublishedas10.
1136/gut.
52.
12.
1690on21November2003.
DownloadedfromDTHtoleranceAsshowninfig3,thecellularresponse(DTH)wasdecreasedinorallytolerisedmicecomparedwithsensitisedmice(28v89mm;p,0.
05)asexpected,butnotinmicethatreceived10mgEXO-hOVAorEXO-hOVA-IFNbyintraperitonealinjection(89and98mm,respectively),indicatingthatepithelialexosomesadministeredintraperitoneallydidnottolerisethecellularimmuneresponsetoOVA.
EpithelialexosomescanprimemousehumoralresponsesIECderivedexosomes,underourexperimentalconditions,wereunabletoinducetolerisationofmicetowardsOVAandevenseemedtoenhancethesensitisationprocess.
Asshowninfig4,exosomesderivedfromMODEKcellsstimulatedbyIFN-cinthepresenceofhOVA(EXO-hOVA-IFN)primedaspecificanti-OVAhumoralimmuneresponse.
ThisresponsewasnotobservedwithexosomespreparedintheabsenceofIFN-c(EXO-hOVA)thatexpressfewerornon-functionalMHCclassII/OVApeptidecomplexes.
TheseresultssuggestthatIFN-promotesthereleaseoffunctionalepithelialexosomesbearingMHCII/OVA-peptidecomplexesabletostimulatetheimmunesystem.
Inaddition,immunisationofmicewith0.
2mgofOVApeptides(calculatedasbeingequivalenttotheamountexpressedon10mgepithelialexosomesunderIFN-cconditions),withorwithoutCFA,didnotinduceasignificantIgGresponse.
Thisresulthighlightedtheincreasedefficiencyofpeptideloadedepithelialexosomesinstimulatingtheimmunesystemcomparedwithfreepeptides.
MouseepithelialexosomesarepresentinmesentericlymphnodesbutnotinserumAsIECderivedexosomesseemtoactasvehiclesoftheantigenicinformationoriginatingfromthegutlumen,wesearchedfortheirpresenceinmiceMLN.
WeusedA33antigenasaspecificmarkerofepithelialexosomestodetecttheirpresenceinthelymphnodesdrainingthesmallintestine.
Asshowninfig5A,MLNlysatescontainedA33antigen,suggestingthepresenceofIECderivedexosomesinthesesecondarylymphoidorgans.
TheamountofA33antigendetectedinMLNwasmuchlowerthanthatobservedusinglysatesfrompurifiedIEC.
TheA33antigenbandwasspecific,aspreabsorptionoftheimmuneserumwithA33antigenimmunisingpeptideledtothedisappearanceoftheband,andA332/2micedidnotdisplayanyspecificbands.
AnimmunohistochemicalstudyofthelocalisationofA33antigeninMLN(fig5B)suggestedthatA33antigenwasenrichedontheexternalmembranesofimmunecellsenteringtheMLNinthesubcapsularsinus.
However,onacautionarynote,whilethepresenceofA33antigeninMLNmightbeareliableindicatorofthepresenceofepithelialexosomes,analternativeexplanationwouldbethattheA33antigenwaspresentasaresultofasubpopulationofdendriticcellstransportingapoptoticintestinalepithelialcellsinphagocyticvacuolestoTcellareasofMLN.
20TheapparentdistributionofA33antigenontheexternalAnti-OVAlgGtitre(10_3)100060020001400*********AAnti-OVAlgEtitre(10_3)621010*B(SD3.
8)TOLEXO-OVAEXO-hOVASENSEXO-OVA-IFNEXO-hOVA-IFNFigure2Ovalbumin(OVA)loadedepithelialexosomes,injectedintraperitoneally,didnotabrogatehumoralimmuneresponses.
Anti-OVAIgGandIgEtitresobtainedinC3H/HeNmicewereorallytolerisedtoOVA(TOL),OVAsensitised(SENS),orinjectedintraperitoneallywithexosomesderivedfromMODEKcellsfedintactOVA(EXO-OVA)orpepsin-trypsinhydrolysedOVA(EXO-hOVA),orfromMODEKcellstreatedwithinterferonc(IFN-c)(EXO-OVA-IFNandEXO-hOVA-IFN).
Notonlydidexosomesnotinducespecifichumoraltolerance,butonthecontrary,EXO-OVAh-IFNandEXO-OVAincreasedsensitisationtoOVA.
*p,0.
05comparedwithtolerisedmice;p,0.
03comparedwithsensitisedmice(n=5–16micepergroup).
Earthickness(m)016014012010080604020***TOLSENSEXO-hOVAEXO-hOVA-IFNFigure3Ovalbumin(OVA)loadedepithelialexosomesinjectedintraperitoneallydidnotabrogatethecellularimmuneresponse(delayedtypehypersensitivity),asmeasuredbythedifferenceinearthicknessbeforeandafterlocalinjectionofOVAorpepsin-trypsinhydrolysedOVA(EXO-hOVA),orfromMODEKcellstreatedwithinterferonc(IFN-c)(EXO-hOVA-IFN).
*p,0.
05comparedwithtolerised(TOL)mice(n=5–8micepergroup).
SENS,sensitisedmice.
1694VanNiel,Mallegol,Bevilacqua,etalwww.
gutjnl.
comonJanuary7,2021byguest.
Protectedbycopyright.
http://gut.
bmj.
com/Gut:firstpublishedas10.
1136/gut.
52.
12.
1690on21November2003.
DownloadedfrommembranesoftheimmunecellsintheMLN(ratherthaninapoptoticbodies)wouldtendtoargueagainstthispossibility.
WealsosearchedforthepresenceofintestinalepithelialexosomesatthesystemiclevelbyanalysinghumanandmiceserumderivedexosomesfortheircontentinA33antigen.
WewereunabletodetectsignificantamountsofA33antigeninserumderivedexosomes(datanotshown)suggestingthatfreeintestinalepithelialexosomesdonotachievesignificantconcentrationsinthecirculation.
DISCUSSIONThepresentresultsshowthatmurineIECstimulatedwithIFN-cinvitroreleaseexosomesbearingMHCclassIIAnti-OVAlgGtitre(10_3)1020304050600*AIIDay8(sc)IDay0(ip)PBS10g10g0.
2gEXO-hOVA-IFNEXO-hOVAhOVAAnti-OVAlgEtitre(10_3)0.
0200.
0400.
0600.
0800.
1000.
1200BPBS10g10g0.
2gEXO-hOVA-IFNEXO-hOVAhOVAIIIDay15Bloodsampling25gOVA/CFA(SD0.
625)(SD20.
5)Figure4Epithelialexosomescanprimethehumoralimmuneresponseinmice.
Anti-ovalbumin(OVA)IgGandIgEtitresobtainedinC3H/HeNmiceinjectedintraperitoneally(ip)withphosphatebufferedsaline(PBS)orexosomesobtainedfromMODEKcellsexposedtopepsin-trypsinOVAhydrolysate(hOVA)andincubatedinthepresenceorabsenceofinterferonc(IFN-c)(EXO-hOVAandEXO-hOVA-IFN)beforesubcutaneous(sc)injectionofOVA/completeFreund'sadjuvant(CFA).
ExosomesobtainedinthepresenceofIFN-c(thusbearingmoreMHCII/OVApeptidecomplexes)werecapableofinducingspecificIgGandIgEresponses.
TheamountofhOVApeptidescorrespondingtothatloadedon10mgEXO/hOVA/IFN(thatis,0.
2mg)didnotinduceanIgGresponse,eveninthepresenceofCFA.
IgEresponsesto0.
2mgOVApeptidesdidnotparallelIgGresponses,suggestingdifferentialsusceptibilitiestolowdoseantigens(n=4–9micepergroup).
*p,0.
001comparedwiththeothergroups;comparedwithEXO-hOVA(p,0.
02),PBS(p,0.
08),andhOVA(p,0.
02).
Figure5(A)WesternblotanalysisofA33antigeninmicemesentericlymphnode(MLN)lysatesshowingsignificantexpressionofA33antigenandsuggestingthatA33antigen,amarkerofintestinalepithelialcells(IEC),alsoexpressedinexosomes,canreachthedraininglymphnodes.
Thewesternblotwasperformedunderreducingconditionsusingarabbitpolyclonalanti-mA33antigenIgG(1.
2mg/ml).
ProteinswereextractedfrommouseIEC(1mgprotein:lanes1and2),MLN(5mgprotein:lanes3and4),andMLNfromA33antigendeficientmice(A332/2)(5mgprotein:lanes5and6).
Inlanes1,3,and5,theanalysiswasconductedafterpreabsorptionoftheanti-mA33antigenIgGwith0.
75mg/ml(,506molarexcess)oftheimmunisingmA33antigenpeptide(+P).
(B)ImmunohistochemistalanalysisofMLNfromC3H/HeNmice,showingimmunecellsenteringthesubcapsularsinusanddisplayingtheA33antigenattheirsurface(seemagnifiedinset),suggestingthatepithelialexosomesmayreachMLNviamigrationofantigenpresentingcells.
ItcannotbeexcludedhoweverthattheA33antigenimmunoreactivityisduetothepresenceofintestinalepithelialapoptoticbodies,althoughenrichmentofthestainingaroundtheperipheryofthecellsarguesagainstthisexplanation.
Functionofintestinalepithelialexosomes1695www.
gutjnl.
comonJanuary7,2021byguest.
Protectedbycopyright.
http://gut.
bmj.
com/Gut:firstpublishedas10.
1136/gut.
52.
12.
1690on21November2003.
Downloadedfrommolecules/peptidecomplexesthat,whenadministeredintra-peritoneallytomice,induceaspecificimmuneresponse.
ThemurineintestinalepithelialcelllineMODEK16producedexosomesdisplayingapatternofmoleculeexpres-sionsimilartothatpreviouslydescribedinhumanintestinalepithelialexosomes,suggestingthatthesemoleculesplayanimportantroleinexosomebiogenesisandfunction.
ThepresenceofthetetraspanproteinsCD9,CD81,andCD82confirmstheirendosomaloriginandMHCclassIImolecules,upregulatedbyIFN-c,underlinetheirpotentialtocarrydietaryproteinderivedpeptides.
Theheatshockproteinshsp71andhsp90couldplayaroleinthetransferofpeptidesfromIECtoprofessionalAPC.
Indeed,hspshavebeeninvolvedinpeptideexchangewithMHCclassIImoleculesafterinternalisationofhsp-peptidecomplexesthroughtheCD91receptorpresentonAPC,andtheirroleincrosspresentationbydendriticcellsofantigenassociatedexo-somesfrommastcellshasrecentlybeendescribed.
19Hspsmayalsoinducematurationandmigrationofdendriticcellstothedraininglymphnodes.
21EpithelialexosomeswerealsoshowntobearA33antigen,amoleculethatishighlyspecificforintestinalepithelialcells,andMFG-E8,2223amoleculethatbindstodendriticcellintegrinsthatcouldplayaroleinadhesionofexosomestotheirtargetcell.
Itisnoteworthythatnocostimulatorymoleculeswerefoundonepithelialexosomes.
Inthepresentstudy,wetriedtoreproducethearrayofOVApeptidesproducedinthedigestivetractinvivobymimickinggastricandpancreatichydrolysisbypepsinandtrypsin.
TheresultingOVApeptideswereplacedincontactwiththeintestinalepithelialcellline,MODEK,allowingfurtheruptakeandprocessingofOVApeptidesbyIECpriortotheirloadingonMHCclassIImolecules,especiallyinthepresenceofIFN-c.
Itisgenerallyacceptedthatantigenpresentationattheintestinallevelistolerogenic,atleastunderbasalconditions.
1324Inthepresentstudy,OVAloadedMODEKexosomeswereusedinatolerisationprotocol,inordertocheckwhethertheirintraperitonealadministrationcouldmimicthetolerisingeffectinducedbyanoraldoseofOVA.
Incontrastwithresultspreviouslyreportedonepithelialtolerosomes,15intraperitonealinjectionofOVAloadedepithelialexosomesinmicedidnotinduceatolerogenichumoralorcellularimmuneresponsetoOVA.
ThefactthatEXO-OVA-IFNandEXO-hOVA-IFNgenerateddifferentresponses(hOVAhavinggreaterpotentialtosensitisemice)suggeststhatprocessingoffoodantigensbyepithelialcellsmayleadtotheformationofdistinctpeptideepitopes.
Thediscrepancybetweenourresultsandthosepreviouslyreportedonepithelialtolerosomes15isdifficulttoexplainalthoughthetwostudieswereconductedusingdifferentepithelialcelllinesanddifferentanimalmodels.
Aswedidnotobserveanytolerogeniceffectofepithelialexosomesdeliveredthroughtheintraperitonealroute,wetestedwhethertheymight,perse,induceaspecificimmuneresponsetoOVA.
WefoundthatMODEKderivedexosomesloadedwithOVApeptidesinthepresenceofIFN-cwerecapableofinducingastrongOVAspecificimmuneresponsewhenadministeredintraperitoneally.
TheabsenceofresponsewithOVAexosomesobtainedintheabsenceofIFN-csuggestedthatacertaindensityofMHCclassIImoleculesontheexosomalsurfaceismandatorytotheimmuneresponse.
Alternatively,MHCclassIImoleculespresentonexosomesintheabsenceofIFN-chavebeendescribedasnascentratherthanmatureMHCclassIIforms25;theseimmaturenascentformshavebeensuggestedtoinducetolerogenicratherthanimmunogenicresponses.
OurresultsindicatethatepithelialexosomescaninduceasignificantimmuneresponsetoOVAdespitetheirlackofcostimulatorymolecules.
ThiswouldindicatethattheywouldmostprobablyhavetointeractwithprofessionalAPCexpressingthesemoleculesinordertoinduceanimmuneresponse,aspreviouslydemonstratedwithdendriticcellderivedexosomes.
14Indeed,ourresultsareinagreementwiththoserecentlyreportedinmiceondendriticcellderivedexosomeswhich,afterinjection,cangenerateanantigenspecificMHCclassIIrestrictedCD4+Tcellproliferationandupregulationofactivationmarkers.
Thisactivationmainlydependsoncostimulatorymoleculesexpressedbydendriticcells,andnotthosepresentonexosomes.
14Inthesameway,mastcellderivedexosomesalsoneeddendriticcellstoefficientlyactivateTcellsinvitro,26suggestingthatMHCclassII-peptidecomplexesaretransferredfromexosomestodendriticcells.
TheinteractionsbetweenAPCandexosomeswasalsodemonstratedwithBcellderivedexosomeswhichwereshowntobindspecificallytofolliculardendriticcellsingerminalcentres.
27Inourstudy,intestinalepithelialexo-somesalsoseemtoplayaroleasMHC/peptidecarriers.
Weartificiallyinjectedtheseepithelialexosomesintraperitone-ally,whichisnotthesite(intestinalmucosa)wheretheyarereleasednormally.
Wehypothesise(fig6)thatepithelialexosomesmayallowtheluminalantigenstakenupandprocessedbyepithelialcellstobetransferredfromtheintestinallumen,viaepithelialexosomes,tothelocalintestinaldendriticcells''sensing''thecontentofthegutlumen,andtotheMLN.
Indeed,A33antigen,aspecificmarkerofepithelialcells11highlyenrichedonepithelialexosomes,wasdetectedinlysatesfromMLN.
ThisantigenFoodantigensIntestinallumenIEC14BasementmembranePeptideloadedexosomesLaminapropria23MaturationLocalpresentationMaturationandmigrationtoMLNImmaturedendriticcellFigure6Workinghypothesisontheroleofepithelialderivedexosomes.
Exosomesreleasedatthebasolateralsurfacesofenterocytesandloadedwithluminalantigenderivedpeptidescrossthebasementmembranetoreachthelaminapropria(1)wheretheymayinteractwithdendriticcells(DCs)(2).
AftermaturationofexosomeloadedDCs,antigenpresentationmaytakeplacelocally(3),orepithelialexosomes/DCsmaymigratetothedraininglymphnodes(4).
IEC,intestinalepithelialcells;MLN,mesentericlymphnodes.
1696VanNiel,Mallegol,Bevilacqua,etalwww.
gutjnl.
comonJanuary7,2021byguest.
Protectedbycopyright.
http://gut.
bmj.
com/Gut:firstpublishedas10.
1136/gut.
52.
12.
1690on21November2003.
Downloadedfromsamplingpathwayviaexosomescouldaddtothepreviouslydescribeddirectsampling,bydendriticcells,ofbacteriapresentintheintestinallumen,throughtheepithelialtightjunctions.
28Intheintestinalsuppressiveenvironment(trans-forminggrowthfactorb),theimmunerecognitionofexosomederivedpeptidesbymucosallymphocytepopula-tions,eitherdirectly(withoutcostimulation)orindirectlythroughprofessionalAPC,mightdriveatolerogenicresponseinthegut,the''natural''environmentforexosomes,atleastunderphysiologicalconditions.
Thuseffectorcellsthatinteractwithexosomederivedepitopes,togetherwiththemicroenvironmentinwhichthisoccurs,areprobablykeyfactorsindeterminingtheoutcomeoftheimmuneresponse,thesamedeterminantstriggeringdifferentresponseswithindifferentsites.
Ininfectiousconditions,maturationofdendriticcellsinthepresenceof''dangersignals''(lipo-polysaccharide,TNF-a),29aswellastheincreaseinMHCclassII/peptidedensityonepithelialexosomes,mayexplainanabnormalimmuneresponsetoluminalantigens.
Thishypothesisrequiresfurtherinvestigation.
Insummary,ourresultssuggestthatepithelialexosomesmayserveasmessengerssensingtheantigeniccontentofthegutlumen.
TransmissionofimmuneinformationviaepithelialexosomestoprofessionalAPCpresentintheintestinalmucosamaybeoneofthepathwaysinvolvedintheimmuneresponsetointestinallydeliveredantigens.
ACKNOWLEDGMENTSGvanNielwasarecipientofagrantfromtheFondationpourlaRechercheMedicale.
Authors'affiliationsGVanNiel,JMallegol*,CBevilacqua*,CCandalh,NCerf-Bensussan,MHeyman,INSERMEMI0212,FaculteNecker-EnfantsMalades,156ruedeVaugirard,Paris,FranceSBrugie`re,CEA,LaboratoiredeChimiedesProteines,Grenoble,FranceETomaskovic-Crook,JKHeath,LudwigInstituteforCancerResearch,PostOfficeRoyalMelbourneHospital,Parkville,Victoria3050,Australia*JMallegolandCBevilacquacontributedequallytothisworkREFERENCES1TerpendK,BoisgeraultF,BlatonMA,etal.
ProteintransportandprocessingbyhumanHT29-19Aintestinalcells:effectofinterferongamma.
Gut1998;42:538–45.
2HershbergRM,FramsonPE,ChoDH,etal.
IntestinalepithelialcellsusetwodistinctpathwaysforHLAclassIIantigenprocessing.
JClinInvest1997;100:204–15.
3KaiserlianD,VidalK,RevillardJP.
MurineenterocytescanpresentsolubleantigentospecificclassII-restrictedCD4+Tcells.
EurJImmunol1989;19:1513–16.
4HershbergRM,ChoDH,YouakimA,etal.
HighlypolarizedHLAclassIIantigenprocessingandpresentationbyhumanintestinalepithelialcells.
JClinInvest1998;102:792–803.
5FramsonPE,ChoDH,LeeLY,etal.
PolarizedexpressionandfunctionofthecostimulatorymoleculeCD58onhumanintestinalepithelialcells.
Gastroenterology1999;116:1054–62.
6KvaleD,KrajciP,BrandtzaegP.
ExpressionandregulationofadhesionmoleculesICAM-1(CD54)andLFA-3(CD58)inhumanintestinalepithelialcelllines.
ScandJImmunol1992;35:669–76.
7vanNielG,RaposoG,CandalhC,etal.
Intestinalepithelialcellssecreteexosome-likevesicles.
Gastroenterology2001;121:337–49.
8RaposoG,NijmanHW,StoorvogelW,etal.
Blymphocytessecreteantigen-presentingvesicles.
JExpMed1996;183:1161–72.
9ZitvogelL,RegnaultA,LozierA,etal.
Eradicationofestablishedmurinetumorsusinganovelcell-freevaccine:dendriticcell-derivedexosomes.
NatMed1998;4:594–600.
10HeathJK,WhiteSJ,JohnstoneCN,etal.
ThehumanA33antigenisatransmembraneglycoproteinandanovelmemberoftheimmunoglobulinsuperfamily.
ProcNatlAcadSciUSA1997;94:469–74.
11JohnstoneCN,TebbuttNC,AbudHE,etal.
CharacterizationofmouseA33antigen,adefinitivemarkerforbasolateralsurfacesofintestinalepithelialcells.
AmJPhysiolGastrointestLiverPhysiol2000;279:G500–10.
12WolfersJ,LozierA,RaposoG,etal.
Tumor-derivedexosomesareasourceofsharedtumorrejectionantigensforCTLcross-priming.
NatMed2001;7:297–303.
13HansonDG,VazNM,MaiaLC,etal.
Inhibitionofspecificimmuneresponsesbyfeedingproteinantigens.
IntArchAllergyApplImmunol1977;55:526–32.
14TheryC,DubanL,SeguraE,etal.
IndirectactivationofnaiveCD4+Tcellsbydendriticcell-derivedexosomes.
NatImmunol2002;3:1156–62.
15KarlssonM,LundinS,DahlgrenU,etal.
''Tolerosomes''areproducedbyintestinalepithelialcells.
EurJImmunol2001;31:2892–900.
16VidalK,GrosjeanI,EvillardJP,etal.
ImmortalizationofmouseintestinalepithelialcellsbytheSV40-largeTgene.
PhenotypicandimmunecharacterizationoftheMODE-Kcellline.
JImmunolMethods1993;166:63–73.
17RabilloudT,KiefferS,ProcaccioV,etal.
Two-dimensionalelectrophoresisofhumanplacentalmitochondriaandproteinidentificationbymassspectrometry:towardahumanmitochondrialproteome.
Electrophoresis1998;19:1006–14.
18TheryC,RegnaultA,GarinJ,etal.
Molecularcharacterizationofdendriticcell-derivedexosomes.
Selectiveaccumulationoftheheatshockproteinhsc73.
JCellBiol1999;147:599–610.
19SkokosD,BotrosHG,DemeureC,etal.
Mastcell-derivedexosomesinducephenotypicandfunctionalmaturationofdendriticcellsandelicitspecificimmuneresponsesinvivo.
JImmunol2003;170:3037–45.
20HuangFP,PlattN,WykesM,etal.
AdiscretesubpopulationofdendriticcellstransportsapoptoticintestinalepithelialcellstoTcellareasofmesentericlymphnodes.
JExpMed2000;191:435–44.
21SrivastavaP.
Rolesofheat-shockproteinsininnateandadaptiveimmunity.
NatRevImmunol2002;2:185–94.
22AndersenMH,BerglundL,RasmussenJT,etal.
BovinePAS-6/7bindsalphavbeta5integrinsandanionicphospholipidsthroughtwodomains.
Biochemistry1997;36:5441–6.
23TaylorMR,CoutoJR,ScallanCD,etal.
Lactadherin(formerlyBA46),amembrane-associatedglycoproteinexpressedinhumanmilkandbreastcarcinomas,promotesArg-Gly-Asp(RGD)-dependentcelladhesion.
DNACellBiol1997;16:861–9.
24Nagler-AndersonC,ShiHN.
Peripheralnonresponsivenesstoorallyadministeredsolubleproteinantigens.
CritRevImmunol2001;21:121–31.
25VidalK,SamarutC,MagaudJP,etal.
Unexpectedlackofreactivityofallogeneicanti-IamonoclonalantibodieswithMHCclassIImoleculesexpressedbymouseintestinalepithelialcells.
JImmunol1993;151:4642–50.
26Vincent-SchneiderH,Stumptner-CuveletteP,LankarD,etal.
ExosomesbearingHLA-DR1moleculesneeddendriticcellstoefficientlystimulatespecificTcells.
IntImmunol2002;14:713–22.
27DenzerK,vanEijkM,KleijmeerMJ,etal.
FolliculardendriticcellscarryMHCclassII-expressingmicrovesiclesattheirsurface.
JImmunol2000;165:1259–65.
28RescignoM,UrbanoM,ValzasinaB,etal.
Dendriticcellsexpresstightjunctionproteinsandpenetrategutepithelialmonolayerstosamplebacteria.
NatImmunol2001;2:361–7.
29AlpanO,RudomenG,MatzingerP.
Theroleofdendriticcells,Bcells,andMcellsingut-orientedimmuneresponses.
JImmunol2001;166:4843–52.
Functionofintestinalepithelialexosomes1697www.
gutjnl.
comonJanuary7,2021byguest.
Protectedbycopyright.
http://gut.
bmj.
com/Gut:firstpublishedas10.
1136/gut.
52.
12.
1690on21November2003.
Downloadedfrom