201303+CALHM1+ion+channel+mediates+purinergic+neurotransmission+of+sweet,+bitter+and+umami+tastes.

中科院 神经科学

LETTER

doi:10.1038/nature11906

CALHM1ionchannelmediatespurinergic

neurotransmissionofsweet,bitterandumamitastes

´rieVingtdeux2*,MakotoOhmoto3,ZhongmingMa1,GennadyDvoryanchikov4,AngLi1{,LeslieAdrien2,AkiyukiTaruno1*,Vale

HaitianZhao2,SzeLeung5,MariaAbernethy5,JeremyKoppel2,PeterDavies2,6,MortimerM.Civan1,7,NirupaChaudhari4,8,

¨ranHellekant5,MichaelG.Tordoff3,PhilippeMarambaud2&J.KevinFoskett1,9IchiroMatsumoto3,Go

Recognitionofsweet,bitterandumamitastesrequiresthenon-vesicularreleasefromtastebudcellsofATP,whichactsasaneuro-transmittertoactivateafferentneuralgustatorypathways1.

However,howATPisreleasedtofulfilthisfunctionisnotfullyunderstood.Hereweshowthatcalciumhomeostasismodulator1(CALHM1),avoltage-gatedionchannel2,3,isindispensablefortaste-stimuli-evokedATPreleasefromsweet-,bitter-andumami-sensingtastebudcells.Calhm1knockoutmicehaveseverelyimpairedperceptionsofsweet,bitterandumamicompounds,whereastheirrecognitionofsourandsaltytastesremainsmostlynormal.Calhm1deficiencyaffectstasteperceptionwithoutinterfer-ingwithtastecelldevelopmentorintegrity.CALHM1isexpressedspecificallyinsweet/bitter/umami-sensingtypeIItastebudcells.ItsheterologousexpressioninducesanovelATPpermeabilitythatreleasesATPfromcellsinresponsetomanipulationsthatactivatetheCALHM1ionchannel.KnockoutofCalhm1stronglyreducesvoltage-gatedcurrentsintypeIIcellsandtaste-evokedATPreleasefromtastebudswithoutaffectingtheexcitabilityoftastecellsbytastestimuli.Thus,CALHM1isavoltage-gatedATP-releasechan-nelrequiredforsweet,bitterandumamitasteperception.

Tastesaresensedbydedicatedreceptorcellsintastebuds,whicharecomposedofthreeanatomicallydistincttypesofcells:typesI,IIandIII.OnlytypeIIIcells,whichsensesourtastes,haveneuron-likefea-tures,includingexpressionofneurotransmitterbiosynthesisenzymesandsynapticvesiclesatclassicalsynapseswithsensorynervefibres1.TypeIIcells,whichsensesweet,bitterandumamitastes,shareacom-monsignal-transductionpathway.However,theylackclassicalsyn-apticstructures,yettransmittasteinformationtogustatoryneuronsbyreleasingATPasaneurotransmitter4,5.Ourwork,describedbelow,implicatesCALHM1asacriticalcomponentresponsibleforthisATPrelease.

Calhm1(ref.6)encodesthepore-formingsubunitofavoltage-gated,non-selective,plasmamembraneionchannelinvolvedinneuronalexcitability2and,potentially,thepathogenesisofAlzheimer’sdisease6–8.Calhm1expressionwasidentifiedinprimatetastebuds9,suggestingthatCALHM1mayhavephysiologicalfunctionsoutsidethebrain.WeconfirmedthatCalhm1wasexpressedinmousetastebudsbutnotinsurroundingepithelium(Fig.1a,b,e–g).ToexamineCALHM1function,wegeneratedaconstitutiveCalhm12/2mouseandverifiedlossofCalhm1expressionintastebuds(Fig.1a,c).Calhm12/2micewereviableandfertile,withnoovertmorphologicalabnormalitiesintheirtastebudsnoranyalteredexpressionoftaste-relatedmarkergenes(Fig.1aandSupplementaryFig.1).LossofCalhm1signalintastebudsofSkn-1a(alsoknownasPou2f3)knockoutmice,fromwhichtypeIIcellsaredevelopmentallyabsent10(Fig.1dand

1

SupplementaryFig.2),andtheco-expressionofCalhm1andTrpm5(Fig.1h),demonstratedthatCalhm1expressionisconfinedtotypeIIcells.Expressionprofilingbyreverse-transcriptionPCRofpoolsofisolatedtypeIIandtypeIIIcellsandindividualtastecellsalsosupportedtheconfinedexpressionofCalhm1totypeIIcells(Sup-plementaryFig.3).Calhm1expressionwasobservednotonlyinTas1r3-expressingsweetandumamitastecells,butalsoinothertypeIIcells,indicatingthatCalhm1isexpressedinsweet,bitterandumamitastecells(Fig.1i).

a

Calhm1Entpd2Tas1r2Trpm5Snap25Pkd2l1Actb

No RT RT

TBLETBLETBLETBLE

Wild typeCalhm1–/–Skn-1a–/–

b

cd

CVPFungiformPalate

efg

Calhm1Trpm5Merge

h

Calhm1Tas1r3Merge

i

Figure1|CALHM1isselectivelyexpressedintypeIItastebudcells.

a,Reverse-transcriptionPCRofmessengerRNAofCalhm1,Actb(b-actin)andtastecellmarkergenesfromlaser-microdissectedcircumvallatepapillae(CVP)tastebuds(TB)andlingualepithelium(LE)inwild-type(1/1)andCalhm1knockout(2/2)mousetongues.RT,reversetranscriptase.b–d,Insitu

hybridizationofCalhm1inCVPtastebudsofwild-type(b),Calhm12/2(c)andSkn-1a2/2(d)mice.Scalebar,50mm.e–g,Calhm1isexpressedinsubsetsofCVP(e),fungiform(f)andpalate(g)tastebudcells.h,Double-labelinsituhybridizationdirectlyillustratescellularco-expressionofCalhm1andTrpm5inCVPtastebuds.MostcellsexpressingTrpm5alsoexpressCalhm1,withCalhm1expressionabsentfromTrpm5-negativecells.i,CVPtastebuds

illustratingthatTas1r3isexpressedinasubsetofCalhm1-positivecells.Scalebarsfore–i,20mm.

DepartmentofPhysiology,UniversityofPennsylvania,Philadelphia,Pennsylvania19104,USA.2Litwin-ZuckerResearchCenterfortheStudyofAlzheimer’sDisease,TheFeinsteinInstituteforMedicalResearch,Manhasset,NewYork11030,USA.3MonellChemicalSensesCenter,Philadelphia,Pennsylvania19104,USA.4DepartmentofPhysiologyandBiophysics,MillerSchoolofMedicine,UniversityofMiami,Miami,Florida33136,USA.5DepartmentofBiomedicalSciences,MedicalSchool,UniversityofMinnesotaDuluth,Duluth,Minnesota55812,USA.6DepartmentofPathology,AlbertEinsteinCollegeofMedicine,Bronx,NewYork10461,USA.7DepartmentofMedicine,UniversityofPennsylvania,Philadelphia,Pennsylvania19104,USA.8PrograminNeurosciences,MillerSchoolofMedicine,UniversityofMiami,Miami,Florida33136,USA.9DepartmentofCellandDevelopmentalBiology,UniversityofPennsylvania,Philadelphia,Pennsylvania19104,USA.{Presentaddress:DepartmentofAnatomy,LiKaShingFacultyofMedicine,TheUniversityofHongKong,HongKong.*Theseauthorscontributedequallytothiswork.

14MARCH2013|VOL495|NATURE|223

中科院 神经科学

RESEARCHLETTER

WeusedtwobehaviouraltestsandnerverecordingstoevaluatetheeffectofthelossofCALHM1expressionontasteperception.Two-bottlepreferencetestsrevealedthatCalhm12/2micehadanearlycompletelossbothofpreferenceforsweetandumamicom-poundsandofavoidanceofbittercompounds(Fig.2a,SupplementaryFig.4a,c,dandSupplementaryTables1and2).Incontrast,therewerelittleornodifferencesbetweenwild-typeandCalhm12/2miceintheirresponsestosaltyandsourcompounds(Fig.2a,SupplementaryFig.4aandSupplementaryTables1and2).Brief-accesstastetestsreproducedthesephenotypes(Fig.2b,SupplementaryFig.4bandSupplementaryTable3),demonstratingthatthetastephenotypeinCalhm12/2miceisasensorydefect.Recordingsofthewholechordatympaninerverevealedstronglyreducedresponsestosweet,bitterandumamicom-poundsinCalhm12/2mice,whereasresponsestoNaClandtheacidcompoundswerenotdifferentfromwildtype(Fig.2candSup-plementaryFig.5),indicatingthatthesensorydefectinCalhm12/2micewasduetotheknockoutofCALHM1functionintheperipheraltastecellsystem.Together,theseresultsindicatethatCALHM1hasacrucialroleintaste-sensingtypeIIcells1,11(seeSupplementaryInformationforfurtherdiscussion).

TypeIIcellssignaltothenervoussystembynon-vesicularATPreleasetonearbyafferentgustatorynerves5,12–14.Althoughconnexinhemichannelsandpannexin1havebeenproposed,themolecularidentityoftheATP-releasemechanismremainsuncertain1,12,13,15.

(ref.3),similartothatestimatedTheCALHM1porediameteris,14A

forconnexins.WethereforeexaminedthepossibilitythatCALHM1

a

Solution preference (%)

00 m

aATP (pmol cm–2)

bATP (pmol cm–2) at 20 min

c

ATP (pmol cm–2

)

Time (min)

[Ca2+]o (mM)

[Ca2+]Time (min)

e

ATP (pmol cm–2)

dInhibition (%)

f

ATP (pmol cm–2)

Time (min)

b

Licks change from 0 score

6040200–5–4

–3–2–10log ([Ca2+]o) (mM)

DCPIA43807HECBRuBF

1

g

ATP (pmol cm–2)

10

00cc mDehaMrinna) (3to

2nium mMQ (10)ui mniCMneitr

(1)ic a mciMd

)(320NaC mMl (17)8 mM)

ATP (pmol cm–2)at 20 min

M)um0 mQM uini(10)ne m (0M).HC3 ml (MN100)aC ml MM(100)SG m (1M00) mM)

h

5

ccha

ro

uc

to

os

(3

(3

rin

se

ni

na

cr

Sa

e

(1

,0

De

0Time (min)

S

Sa

Su

c

i

ATP (pmol cm–2)

Inhibition (%)

Relative intensity

j

K+K+K+K+

Time (min)

Figure2|CALHM1isessentialforsweet,bitterandumamitaste

perception.a,b,Meanpreferencepercentage(tastecompoundversuswater)from48-h,two-bottlepreferencetests(a)andbrief-accesslickscores(b)forindicatedcompoundsinCalhm12/2miceandwild-typelittermates.Errorbars,s.e.(8–12micepergroup,4–6monthsold);*P,0.01(posthocBonferroni’stest(a)andStudent’st-test(b)).c,Summaryofresponsesfromwhole-chorda-tympaninerverecordingsstimulatedwithindicatedcompoundsandnormalizedtoresponsetoNH4Clfromwild-type(n57)andCalhm12/2(n58)mice.MSG,monosodiumglutamate;IMP,inosine59-monophosphate;MCG,monocalciumdi-L-glutamate.Errorbars,s.e.;*P,0.05(Student’st-test).

224|NATURE|VOL495|14MARCH2013

Figure3|CALHM1mediatesATPrelease.a,TimecoursesofextracellularATPlevelsduetoreleasefrommock-andhCALHM1-transfectedHeLacellsexposedtonormal(1.9mM)oressentiallyzero(17nM)[Ca21]o.b,SummaryofextracellularATPlevelsat20minina.c,Low-[Ca21]o-inducedATPreleaseinCALHM1-expressingcellsisabolishedbyrutheniumred(RuR).d,Effectsonlow-[Ca21]o-stimulatedATPreleasefromhCALHM1cellsof0.5mgml21brefeldinA(BFA),10mMDCPIB,3mMA438079,1mM1-heptanol(HEP),30mMcarbenoxolone(CBX)and20mMRuR.e,TimecoursesofATPreleasefromhCALHM1cellsinducedbyvarious[Ca21]o.f,ATPlevelsat20minineplottedagainst[Ca21]oandfittedtoaHillequation.g,Depolarizationbyhigh[K1]o(117.5mMK1)inducesATPreleasespecificallyfromhCALHM1cells.h,ATPlevelsat20mining.i,

Depolarization-inducedATPreleasefromCALHM1-expressingcellsisabolishedbyRuR.j,Pharmacologicalsensitivitiesofdepolarization-inducedATPreleasefromhCALHM1cells:1mMHEP,1mMprobenecid(PROB),20mMRuR.Numberofwellsshowninparenthesesthroughout.Errorbars,s.e.(wherevisible);*P,0.01(Student’st-test).

rosAcSue (1escro00sulphe ( mMam300 )Fre mKucMto (25)Gse mluco(

30M)Sase0 mMcc (3ha00) mrinSM DeC45(30) na64mto7 M)ni(8um mQ (5M)uiSpnine mM (1Asar)0 coteinrbe mMic(5) Caci0 mitrd Mic a(20 )Ncid mMH(20)4Cl ( m10MNaC0 m)l M)M(30SG0 m (1M)0IM0 mP MM(C3

0 )G (1mM00) mM)

Suc

PROHERu

中科院 神经科学

LETTERRESEARCH

mediatesATPrelease.TheCALHM1ionchannelcanbeactivatedbymembranedepolarizationorreductionofextracellularCa21concen-tration2([Ca21]o).WefirstexploitedthelattermechanismtoactivateCALHM1inheterologousexpressionsystemstodeterminewhetherCALHM1canformanATP-releasechannel.Decreasing[Ca21]ofrom1.9mMtoessentiallyzero(17nM)activatedATPreleasefromhumanCALHM1(hCALHM1)-expressingHeLacells,whereaslittleATPeffluxwasinducedinmock-transfectedcells(Fig.3a,b).Simi-larlow-[Ca21]o-inducedATPreleasewasobservedinhCALHM1-expressingCOS-1cellsandXenopusoocytes(SupplementaryFig.6).NeitherCALHM1expressionnorlow-[Ca21]oexposurecausedcelldamage(SupplementaryFig.7).Involvementofotherpossiblemechanisms16wasruledoutbecauseATPreleasewasunaffectedbybrefeldinA(vesicularrelease),DCPIB(volume-sensitiveCl2chan-nels),A438079(P2X7receptors),heptanol(connexinhemichannels)orcarbenoxolone(pannexinsandconnexinsat30mM)(Fig.3d).Incontrast,rutheniumred(RuR),whichinhibitsCALHM1ioncurrents2,abolishedlow-[Ca21]o-evokedATPreleasefromhCALHM1-expressingcells(Fig.3c,d).Thus,CALHM1expressioninducesanovelATPpermeability.

CALHM1ionchannelgatingisregulatedby[Ca21]owithanappar-enthalf-maximuminhibitory[Ca21]ofIC50<220mMandaHillcoefficientof2.1(ref.2).ExtracellularCa21inhibitedATPreleasewithIC505495mMandaHillcoefficientof1.9(Fig.3e,f).CALHM1ioncurrentsarealsoactivatedbymembranedepolarization2.Atnormallevelsof[Ca21]o,hCALHM1-expressing,butnotmock-transfected,cellsreleasedATPinresponsetohigh-[K1]o-induceddepolarization(Fig.3g,h).Depolarization-inducedATPreleasewasinhibitedbyRuR

a

Wild type

butnotbyheptanolorprobenecid(Fig.3i,j),whicharerespectivelyconnexinandpannexin1blockers17.ExpressionofmouseCALHM1conferredATPreleasewithsimilarproperties(SupplementaryFig.8).RegulationofATPreleaseinhCALHM1-expressingcellsisthere-foresimilartothatofCALHM1channelgating,indicatingthattheCALHM1channelisaconduitforATPrelease.

Thethreetypesoftastebudcellcanbeclassifiedonthebasisofwhole-cellelectrophysiologicalfingerprints13,18.Weverifiedthesefinger-printsbyrecordingwhole-cellcurrentsinsingleisolatedtastebudcellsfromTRPM5–greenfluorescentprotein(GFP)micewithGFPexpressedspecificallyintypeIIcells(SupplementaryFig.9).WithtetraethylammoniumintheextracellularsolutionandCs1asthemajorcationinthepipettesolution2(toblockK1currents),theelectro-physiologicalsubtypeswereidentifiedbythecombinationofvoltage-gatedNa1currents(INa)andnon-selective,slowlyactivatingoutwardcurrents(Islow)withinwardtailcurrentsat270mV(Itail)(Fig.4a–candSupplementaryFig.9).ToidentifyCALHM1ioncurrents,werecordedwhole-cellcurrentsinisolatedtastebudcellsfromwild-typeandCalhm12/2mice(Fig4a–c).LossofCALHM1substantiallyreducedtheamplitudesofIslowandItailwithoutaffectingtheampli-tudeofINaintypeIIcells(Fig.4d–f),whereasnodifferenceswereobservedintypeIandtypeIIIcells(Fig.4b,c,g).TheslowlyactivatingoutwardcurrentintypeIIcellswasinhibitedbythenon-specificCALHM1blockerGd31butnotbyprobenecidorheptanol,rulingoutcontributionsofpannexin1orconnexins(Fig.4h).ThesedatademonstratethatCALHM1channelconductanceispresentintypeIIcellsandcontributesthemajorcomponentoftheslowlyactivat-ingoutwardcurrent.Notably,theamountofdepolarization-induced

f

pA pF–1)

Type II

b

Type III

d

(pA pF–1)

100 pA pF–1

0.5 s

Calhm1–/–

e

(pA pF–1)

Vm (mV)

g

pA pF–1)

c

Type I

Vm (mV)

h

Normalized current

i

Wild typeCalhm1–/–j

Basal [Ca2+]i (nM)

k

[Ca2+]i (nM)

Time (s)

l

ATP (fmol mm–2)

Figure4|CALHM1isrequiredfortaste-evokedATPreleasefromtastecells.

a–c,ElectrophysiologicalphenotypesoftypeI,typeIIandtypeIIIcellsidentifiedinwild-type(red)andCalhm12/2(blue)tastecells.Cellsheldat270mVandpulsedfrom280to180mVin20-mVincrementswith1-sduration.TypeIcurrentincrecordedfrom16wild-typeand9Calhm12/2cells.d–g,INa(circle;f),Islowatendofpulses(square;d)andItail(triangle;e)measuredfortypeIIcells(n59wildtype,n510Calhm12/2),andINameasuredfortypeIIIcells(n59wildtype,n56Calhm12/2)(g).Vm,membranevoltage.h,SensitivitiesofIslowinGFP-positivecellsfromTRPM5–GFPmicetoGd31(100mM),probenecid(1mM),1-heptanol(1mM)(n54).i,[Ca21]iintypeIIcellsfromwild-type(left,9cells)andCalhm12/2(right,12cells)mice.TypeIIcellsidentifiedbyrobust[Ca21]iresponsetoamixofsweetandbittersubstances(greybar).j,k,Basal(j)andtaste-evoked(k)responsesarecomparableinwild-typeandCalhm12/2cells.KO,knockout;WT,wildtype.l,Taste-evokedATPreleasefromgustatoryCVPtissueandnon-gustatorylingualepithelium.BittermixelicitsconsiderableATPreleasefromCVPversuslingualepitheliuminwild-typemice;thisisabolishedinCalhm12/2miceandby1mMtetrodotoxin(TTX).Errorbars,s.e.;*P,0.05,**P,0.01(Student’st-test).m,Schematicillustrationofsignal-transductioncascadeintypeIItastereceptorcellswithCALHM1astheATP-releasepathway.

Δ[Ca2+]i (nM)

14MARCH2013|VOL495|NATURE|225

中科院 神经科学

RESEARCHLETTER

ATPreleasefromtypeIIcellsiscorrelatedwiththemagnitudeofIslow(ref.13).

TypeIIcellsdetectsweet,bitterandumamicompoundsviaG-protein-coupledtastereceptors4thatstimulateacommonsignal-transductioncascadeinvolvingactivationofPLCB2,inositol1,4,5-trisphosphate-mediatedCa21releaseandCa21-dependentactivationofTRPM5channels11thatdepolarizestheplasmamembranetogene-rateactionpotentialsandsubsequentnon-vesicularreleaseofATP.Importantly,nodifferencesinbasal[Ca21]iandtaste-evoked[Ca21]iresponseswereobservedbetweenwild-typeandCalhm12/2typeIIcells(Fig.4i–k).Furthermore,Calhm1deficiencyhadnoeffectonINa(Fig.4f),TRPM5expressionintypeIIcells(Fig.1aandSupplemen-taryFigs1iand2)ortheATPcontentoftastebuds(SupplementaryFig.10).Strikingly,however,taste-evokedreleaseofATPfromcir-cumvallatepapillaewasabolishedinCalhm12/2mice(Fig.4landSupplementaryFig.11).ThisstronglysuggeststhatCALHM1contri-butestothemajorATP-releasemechanisminsweet/bitter/umami-sensingtastebudcells(Fig.4l,m).

OurstudyshedslightonanovelcellularATP-releasingmechanismbydemonstratingthatCALHM1isavoltage-gatedionchannelthatmediatestetrodotoxin-sensitiveATPreleaseintastebuds(Fig.4l)asanessentialmechanismofsweet,bitterandumamitasteperception.Assuch,CALHM1providesamissinglinkinthesignal-transductioncascadeintypeIIcells,connectingtastereceptoractivationandthegenerationofNa1actionpotentialstotheactivationofafferentgustatoryneuralpathways1(Fig.4m).IthasnotescapedourattentionthatotherCALHMisoforms6aswellaspannexin1andconnexinsarealsopresentintastebuds9,12,13andmightalsobeinvolvedinATPreleaseintaste,perhapsactinginparallelwithorinacomplexwithCALHM1.SignallingbyextracellularATPisawidespreadphenomenonthatregulatesmanyphysiologicalactivities19,includingneurotransmission20,21,intercellularcommunication22,23,vasculartone24andsensorytransduction5,25–27.CALHM1anditsisoformsmayparticipateinphysiologicallyimportantATPreleaseelsewhere.

6.7.8.9.10.11.12.13.14.15.16.17.18.19.20.21.22.23.24.25.26.27.

Dreses-Werringloer,U.etal.ApolymorphisminCALHM1influencesCa21

homeostasis,Ablevels,andAlzheimer’sdiseaserisk.Cell133,1149–1161(2008).Koppel,J.etal.CALHM1P86LpolymorphismmodulatesCSFAblevelsincognitivelyhealthyindividualsatriskforAlzheimer’sdisease.Mol.Med.17,974–979(2011).

Lambert,J.C.etal.TheCALHM1P86LpolymorphismisageneticmodifierofageatonsetinAlzheimer’sdisease:ameta-analysisstudy.J.AlzheimersDis.22,247–255(2010).

Moyer,B.D.etal.Expressionofgenesencodingmulti-transmembraneproteinsinspecificprimatetastecellpopulations.PLoSONE4,e7682(2009).

Matsumoto,I.,Ohmoto,M.,Narukawa,M.,Yoshihara,Y.&Abe,K.Skn-1a(Pou2f3)specifiestastereceptorcelllineage.NatureNeurosci.14,685–687(2011).

Zhang,Y.etal.Codingofsweet,bitter,andumamitastes:differentreceptorcellssharingsimilarsignalingpathways.Cell112,293–301(2003).

Huang,http://wendang.chazidian.comA104,6436–6441(2007).

Romanov,R.A.etal.Afferentneurotransmissionmediatedbyhemichannelsinmammaliantastecells.EMBOJ.26,657–667(2007).

Vandenbeuch,A.,Zorec,R.&Kinnamon,S.C.Capacitancemeasurementsofregulatedexocytosisinmousetastecells.J.Neurosci.30,14695–14701(2010).Romanov,R.A.etal.DispensableATPpermeabilityofpannexin1channelsinaheterologoussystemandinmammaliantastecells.J.CellSci.125,5514–5523(2012).

Sabirov,R.Z.&Okada,Y.ATPreleaseviaanionchannels.PurinergicSignal.1,311–328(2005).

Silverman,W.,Locovei,S.&Dahl,G.Probenecid,agoutremedy,inhibitspannexin1channels.Am.J.Physiol.CellPhysiol.295,C761–C767(2008).

Romanov,R.A.&Kolesnikov,S.S.Electrophysiologicallyidentifiedsubpopulationsoftastebudcells.Neurosci.Lett.395,249–254(2006).

Burnstock,G.&Kennedy,C.P2Xreceptorsinhealthanddisease.Adv.Pharmacol.61,333–372(2011).

Edwards,F.A.,Gibb,A.J.&Colquhoun,D.ATPreceptor-mediatedsynapticcurrentsinthecentralnervoussystem.Nature359,144–147(1992).

Evans,R.J.,Derkach,V.&Surprenant,A.ATPmediatesfastsynaptictransmissioninmammalianneurons.Nature357,503–505(1992).

Cotrina,M.L.,Lin,J.H.,Lopez-Garcia,J.C.,Naus,C.C.&Nedergaard,M.ATP-mediatedgliasignaling.J.Neurosci.20,2835–2844(2000).

Cotrina,M.L.,Lin,J.H.&Nedergaard,M.CytoskeletalassemblyandATPreleaseregulateastrocyticcalciumsignaling.J.Neurosci.18,8794–8804(1998).

Burnstock,G.Dualcontroloflocalbloodflowbypurines.Ann.NYAcad.Sci.603,31–44(1990).

Chizh,B.A.&Illes,P.P2Xreceptorsandnociception.Pharmacol.Rev.53,553–568(2001).

Cockayne,D.A.etal.Urinarybladderhyporeflexiaandreducedpain-relatedbehaviourinP2X3-deficientmice.Nature407,1011–1015(2000).

Wynn,G.,Rong,W.,Xiang,Z.&Burnstock,G.Purinergicmechanismscontributetomechanosensorytransductionintheratcolorectum.Gastroenterology125,1398–1409(2003).

METHODSSUMMARY

AllanimalexperimentswereconductedaccordingtoprotocolsapprovedbytheInstitutionalAnimalCareandUseCommitteeoftheUniversityofPennsylvania,theFeinsteinInstituteforMedicalResearch,theMonellChemicalSensesCenterandtheUniversityofMinnesotaDuluth.

FullMethodsandanyassociatedreferencesareavailableintheonlineversionofthepaper.

Received6September2012;accepted15January2013.Publishedonline6March2013.1.2.3.4.5.

Chaudhari,N.&Roper,S.D.Thecellbiologyoftaste.J.CellBiol.190,285–296(2010).

Ma,Z.etal.Calciumhomeostasismodulator1(CALHM1)http://wendang.chazidian.comA109,E1963–E1971(2012).

Siebert,A.P.etal.Structuralandfunctionalsimilaritiesofcalciumhomeostasismodulator1(CALHM1)ionchannelwithconnexins,pannexinsandinnexins.J.Biol.Chem.288,6140–6153(2013).

Chandrashekar,J.,Hoon,M.A.,Ryba,N.J.&Zuker,C.S.Thereceptorsandcellsformammaliantaste.Nature444,288–294(2006).

Finger,T.E.etal.ATPsignalingiscrucialforcommunicationfromtastebudstogustatorynerves.Science310,1495–1499(2005).

SupplementaryInformationisavailableintheonlineversionofthepaper.

AcknowledgementsThisworkwassupportedbyaKeySpanawardtoP.M.,severalUSNIHgrants(GM56328,MH059937,NS072775toJ.K.F.;DC10393toM.G.T.;EY13624toM.M.C.;R03DC011143toI.M.;CoreGrantP30EY001583totheUniversityof

Pennsylvania;CoreGrantP30DC011735totheMonellChemicalSensesCenter)andtheUniversityofMinnesota’sUndergraduateResearchOpportunitiesProgramtoS.L.andM.A.A.T.andM.O.areJSPSFellows.WethankR.F.MargolskeefortheTRPM5–GFPmiceandY.Ninomiyaforcommentsonthemanuscript.

AuthorContributionsA.T.,V.V.,A.L.,Z.M.,M.O.,I.M.,H.Z.,L.A.,S.L.,M.A.,G.H.,G.D.andN.C.designedandperformedexperiments.P.M.andV.V.generatedtheCalhm1knockoutmice.J.K.andP.D.designedexperiments.M.G.T.,M.M.C.,P.M.andJ.K.F.designedexperimentsandhelpedwithdatainterpretation.A.T.,J.K.F.andP.M.wrotethemanuscript.

AuthorInformationReprintsandpermissionsinformationisavailableat

http://wendang.chazidian.com/reprints.Theauthorsdeclarenocompetingfinancialinterests.

Readersarewelcometocommentontheonlineversionofthepaper.CorrespondenceandrequestsformaterialsshouldbeaddressedtoP.M.( )andJ.K.F.( ).

226|NATURE|VOL495|14MARCH2013

中科院 神经科学

LETTERRESEARCH

METHODS

nucleotides525–2,725ofTas1r3(accessionnumber,AF337039),andnucleotides1–1,407and2,148–2,369ofCalhm1cDNAfragment,whichcontains1,047basepairsoftheentirecodingsequenceand1,322basepairsof39-non-codingregion.Immunofluorescencestaining.Forfixed-tissuepreparation,mice12weeksoldorolderwereanaesthetizedwithanoverdoseofsodiumpentobarbital(AbbottLaboratories)andtranscardiallyperfusedwithice-coldPBSfollowedby4%para-formaldehydeinice-coldPBS.TongueepitheliacontainingCVPweredissectedandpostfixedwiththesamefixativeat4uCovernight.Tissuesampleswerethencryoprotectedin30%sucroseinPBS,embeddedinfrozenOCTcompound(SakuraFinetech)andsectionedusingacryostat(CM1900,LeicaMicrosystems)at8mm.Tissuesectionsweremountedontissue-adhesive-coatedglassslides(FisherScientific)andstoredat280uCuntilanalysed.Forimmunofluorescencestaining,slideswererinsedwithPBS,incubatedinapreheatedtargetretrievalsolution(S1700,Dako)at80uCfor20min,allowedtocooltoroomtemperature(20–25uC)for20min,stillinthetargetretrievalsolution,andwashedinPBSfor3310min.Afterblockingby1-hincubationinPBScontaining5%normaldonkeyserum(PBS-5%NDS)atroomtemperature,sectionswereincubatedovernightat4uCwithprimaryantibodiesdilutedinPBS-5%NDS:1:3,000forrabbitanti-TRPM5(AlomoneLabs);1:500forrabbitanti-PLCB2(SantaCruzBiotechnology);1:500forrabbitanti-AADC(GeneTex);1:500forgoatanti-KCNQ1(SantaCruz).Thenextday,theslideswerewashedfor3310mininPBSatroomtemperatureandincubatedfor1hwithAlexa-Fluor-conjugatedantibodiesdilutedinPBS-5%NDS:1:500forAlexaFluor488donkeyanti-rabbitIgG;1:500forAlexaFluor568anti-goatIgG(Invitrogen).Finally,slideswerewashedfor3310mininPBSatroomtemperatureandmountedinVectaShieldwith49,6-diamidino-2-phenylindole(H-1500,VectorLaboratories).Stainedsec-tionswereimagedwithaconfocalscanningmicroscope(LSM710,CarlZeiss).Single-tastecellanalyses.Circumvallatetasteepitheliumwasenzymaticallydela-minated,tastebudswerecollectedfrompeeledepitheliumanddissociatedsingle-tastecellswerecollected,allasdetailedpreviously29.Forpatchclampandfura-2Ca21imagingexperiments,isolatedcellswereplacedonpoly-L-lysine-coatedcoverslipsandallowedtosettlefor30–60min.

TotalRNAsisolatedfromtastebudsandfromadjacentnon-tasteepitheliumwereusedaspositiveandnegativecontrols,respectively.IndividualdissociatedtastecellsfromPLCB2–GFP(typeII)andGAD1–GFP(typeIII)transgenicmicewereselectedforcollectiononthebasisoftheirexpressionofGFP,aspreviouslydescribed30.TypeItastecellswereselectedfromPLCB2–GFP3GAD1–GFPdouble-transgenicmicebytheabsenceofGFPexpression30.TotalRNAisolatedfromtastecellswaseithersubjectedtoT7RNApolymerase-basedlinearamp-lification(aRNA,MessageBOOSTERkit,Epicentre)orwasdirectlyconvertedtocDNA,bothaspreviouslydescribed29.ForaRNA-basedanalyses,1%oftheresultingcDNAwasusedastemplateforPCR.FordirectRT–PCR,15%ofcDNAfromeachcellwasusedtotestfordiagnosticmessengerRNAsand30%wasusedforCalhm1.PCRconditionswithHotStarTaqPlus(QIAGEN)were95uCfor5minfollowedby4590-scycles,eachofwhichcomprisedthree30-scomponents,oneat94uC,oneattheannealingtemperatureandoneat72uC.PCRprimers,annealingtemperaturesandproductsizeswereasinSupplementaryTable4.TwodifferentprimersetswereusedforCalhm1(SupplementaryTable4).Primers1and2wereusedforRT–PCRonbulktissueRNA,aswellasforaRNAfromindividualtastecells(Fig.1aandSupplementaryFig.3a,b,d).Primers3and4spananintron,andwereusedforthesingle-celldirectRT–PCRtoavoidamp-lifyingfromgenomicDNA(SupplementaryFig.3c).ThecelltypeofallsinglecellsandpoolsofcellswasconfirmedbyRT–PCRforthreediagnosticcell-typemarkers:Entpd2(typeI),Plcb2(typeII)andSnap25(typeIII).

Two-bottlepreferencetests.Calhm12/2andwild-typecontrolmicewerepre-sentedfor48hwithtwodrinkingbottles,onecontainingwaterandtheotherasolutionsupplementedwithascendingconcentrationsofthespecifictastecom-poundstobetested.Allsolutionswerepreparedwithtapwaterandservedatroomtemperature.Micehadadlibitumaccesstoastandardchowdiet(LaboratoryRodentDiet5001,LabDiet,PMINutritionInternational)anddrinkingsolutions.Twoindependentgroupsofmiceweretestedforthefollowingseriesoftastesolutions.Theorderoftestingwasasfollows:sucrose,saccharin,NaCl,denato-nium(group1);MSG,quinine,HCl(group2).Sucrose,saccharin,denatoniumbenzoate,quininesulphateandMSGwerefromSigma-Aldrich.HClandNaClwerefromThermoFisherScientificInc.Theorderofstimuluspresentationwasidenticalforwild-typeandknockoutmice.Themicehadthreedayswithasinglebottleofdrinkingwaterbetweeneachtestseries.Thetwobottleswereswitchedafter24htocontrolforapossibleeffectofthebottlepositions.Thechangeinliquidlevelswasconsideredtobethemouse’sliquidintake31.Fluidintakesweremea-suredvolumetricallytothenearest0.1ml.Thevolumesofwaterandtastecom-poundsolutionsconsumedwererecorded.Solutionpreferencewascalculatedastheintakeoftastesolutiondividedbytotalliquidintake,andthisratiowasexpressedasapercentage.Thetastecompoundswerechosenasexemplarsofthe

Calhm12/2mice.Calhm11/2foundermiceweregeneratedatgenOway(Lyon,France)asdescribedpreviously2.Briefly,Calhm1exon1deletionwasperformedbyhomologousrecombinationin129Svembryonicstemcells.PositivecloneswerescreenedbyPCRandconfirmedbySouthernblotanalysis.TheresultantembryonicstemcellswereinjectedintoblastocystsderivedfromC57BL/6JmicetoobtainchimaericmicethatpossessedgermlinetransmissionofthetargetedCalhm1locus.Miceonthemixed129Sv3C57BL/6Jgeneticbackgroundwereusedinthisstudy.Wild-type(1/1)andCalhm1knockout(2/2)miceofF1obtainedfromlittermatemicewereusedfortwo-bottlepreferencetests.Allotherexperimentswereperformedwithwild-typeandknockoutlittermates.Insomemice,theneomycincassettewasremovedandtheresultingstrainwasbackcrossedwithC57BL/6Jmice.Briefly,Calhm11/2micewerematedwithmicebearinganEIIa–cretransgene(B6.FVB-Tg(EIIa-cre)C5379Lmgd/J,TheJacksonLaboratory)ontheC57BL/6Jbackgroundtoremovetheneomycinresistancecassettebycre–loxP-mediatedexcision.TheresultingCalhm11/2pupswerefurtherbackcrossedtoC57BL/6J(TheJacksonLaboratory)foratleastfivegenerationsbeforebeingmadehomozygous.TheremovaloftheneomycincassettewasconfirmedbyPCR.Backcrossedmicewereusedfortwo-bottlepreferencetests.LossofCALHM1expressioninCalhm12/2tastebudswasverifiedbyreverse-transcription(RT)–PCRandinsituhybridization.

Lasercapturemicrodissection,RNAamplificationandRT–PCR.Circum-vallatetastetissuefromCalhm11/1andCalhm12/2micewasembeddedincryo-moldsusingOCTfreezingmedium.Twelvemicrometre-thicktissuesectionswerecutonaLeicaCM1850cryostat(LeicaMicrosystems),collectedonRNase-freemembraneslides(Leica)andstainedwithcresylviolet.TastebudsfromtheCVPandsurroundinglingualepitheliumwereisolatedusingaLeicalasermicrodissec-tionsystemLMD7000(Leica).Tastebudsandsurroundingepitheliumwerepooledfromatotaloffourmiceofeachgenotype.TotalRNAfromtastebudandlingualepitheliumsampleswaspurifiedusingaRNAqueous-MicroKit(Ambion).TotalRNAwasamplifiedusingtwosequentialroundswithMessageAmpIIaRNAkit(Ambion),asperthemanufacturer’sinstructions.OnemicrogramofRNAwastranscribedintocomplementaryDNA(cDNA)usingInvitrogen’sSuperScriptIIIFirst-StrandSynthesisSystemforRT–PCRwithran-domhexamers,accordingtothesuppliedprotocol.A50-mlPCRreactionwasrunwiththefollowingfinalconcentrations:450ngofeachprimer(seeSupplementaryTable4forPCRprimersequences),2mMMgCl2,0.3mMdNTPs,2.5UTaqpolymerase(PromegaGoFlexDNApolymerase)and1mlof10ngml21DNA.PCRcyclingconditionsusedforCalhm1:95uCfor3min;35cyclesof95uCfor30s,65uCfor30sand72uCfor45s;72uCfor7min;4uC(hold).

Insituhybridization.Oralepitheliacontainingtastebudsweredissectedfromadultmalemicedeeplyanaesthetizedwithanoverdoseofsodiumpentobarbital(AbbottLaboratories)andembeddedinthefrozenOCTcompound(SakuraFinetechUSA).Fresh-frozensectionsof8-mmthicknesswerepreparedusingacryostat(CM3050S,LeicaMicrosystems).Theinsituhybridizationprocedurewasdescribedpreviously28.Inbrief,digoxigenin-andfluorescein-labelledantisenseRNAspreparedusingRNAlabellingmix(RocheDiagnostics)andanRNApoly-merase(Stratagene)wereusedforhybridizationafterfragmentationunderalkal-ineconditionstoalengthofabout150bases.Fresh-frozensectionswerefixedwith4%PFA,treatedwithdiethylpyrocarbonate,prehybridizedwithsalmonspermDNAfor2hat65uCandhybridizedwithantisenseriboprobesfor40hat65uC.ForsinglelabellingofTas2r108andTrpm5,however,prehybridizationandhybridizationwerecarriedoutat58uC.Afterhybridization,thesectionswerewashedin30.2SSCat58uCor65uCandblockedwith0.5%blockingreagent(RocheDiagnostics)inTris-bufferedsaline.Chromogenicsignals,exceptthosefromCalhm1,weredevelopedforonedayusingalkalinephosphatase-conjugatedanti-digoxigeninantibody(1:500,RocheDiagnostics)and4-nitrobluetetrazo-liumchloride/5-bromo-4-chloro-3-indolyl-phosphateaschromogenicsubstrates.ChromogenicCalhm1signalsweredevelopedfortwodaystoclarifythecellsexpressingCalhm1messengerRNA.NoCalhm1signalwasobservedinSkn-1orCalhm1knockouttastetissuesaftersignaldevelopmentforthreedays.StainedimageswereobtainedwithaNikonEclipse80imicroscope(NikonInstru-mentsInc.)equippedwithaDXM1200Cdigitalcamera(Nikon).Fluorescentsignalsweredevelopedusingbiotin-conjugatedanti-fluoresceinantibody(1:500,VectorLaboratories),avidin–biotincomplex(VectorLaboratories),tyramidesig-nalamplificationbiotinsystem(1:50,PerkinElmer)andAlexa488-conjugatedstreptavidin(4mgml21,Invitrogen).FluorescentimageswereobtainedwithaLeicaSP2confocalscanningmicroscope(Leica).FordoublelabellingofCalhm1withTrpm5andTas1r3,fluorescentsignalswerefirstdevelopedandobtained,thenchromogenicsignalsweredevelopedandobtainedasdescribedabove,andfinallytheimagesweresuperimposedusingPhotoshopCS3(Adobe).RNAprobesgeneratedweretonucleotides1–894ofTas2r108(GenBankaccessionnumber,AF227148),nucleotides310–3,491ofTrpm5(accessionnumber,AF228681),