outputkav

NeuronNeuroViewTheBrainActivityMapProjectandtheChallengeofFunctionalConnectomicsA.
PaulAlivisatos,1MiyoungChun,2GeorgeM.
Church,3RalphJ.
Greenspan,4MichaelL.
Roukes,5andRafaelYuste6,*1MaterialsScienceDivision,LawrenceBerkeleyNationalLabandDepartmentofChemistry,UniversityofCalifornia,Berkeley,Berkeley,CA94720,USA2TheKavliFoundation,Oxnard,CA93030,USA3DepartmentofGeneticsandWyssInstitute,HarvardMedicalSchool,Boston,MA02115,USA4KavliInstituteforBrainandMind,UCSD,LaJolla,CA92093,USA5KavliNanoscienceInstituteandDepartmentsofPhysics,AppliedPhysics,andBioengineering,CaliforniaInstituteofTechnology,Pasadena,CA91125,USA6HHMI,DepartmentBiologicalSciences,KavliInstituteforBrainScience,ColumbiaUniversityNewYork,NY10027,USA*Correspondence:rafaelyuste@columbia.
eduDOI10.
1016/j.
neuron.
2012.
06.
006Thefunctionofneuralcircuitsisanemergentpropertythatarisesfromthecoordinatedactivityoflargenumbersofneurons.
Tocapturethis,weproposelaunchingalarge-scale,internationalpubliceffort,theBrainActivityMapProject,aimedatreconstructingthefullrecordofneuralactivityacrosscompleteneuralcircuits.
Thistechnologicalchallengecouldprovetobeaninvaluablesteptowardunderstandingfundamentalandpathologicalbrainprocesses.
''Thebehavioroflargeandcom-plexaggregatesofelementaryparticles,itturnsout,isnottobeunderstoodintermsofasimpleextrapolationofthepropertiesofafewparticles.
Instead,ateachlevelofcomplexityentirelynewproper-tiesappear.
''–MoreIsDifferent,P.
W.
Anderson''Newdirectionsinsciencearelaunchedbynewtoolsmuchmoreoftenthanbynewconcepts.
Theeffectofaconcept-drivenrevolu-tionistoexplainoldthingsinnewways.
Theeffectofatool-drivenrevolutionistodiscovernewthingsthathavetobeexplained.
''–ImaginedWorlds,FreemanDysonEmergentPropertiesofBrainCircuitsUnderstandinghowthebrainworksisarguablyoneofthegreatestscienticchallengesofourtime.
Althoughtherehavebeenpiecemealeffortstoexplainhowdifferentbrainregionsoperate,nogeneraltheoryofbrainfunctionisuniver-sallyaccepted.
Afundamentalunderlyinglimitationisourignoranceofthebrain'smicrocircuitry,thesynapticconnectionscontainedwithinanygivenbrainarea,whichCajalreferredtoas''impenetrablejungleswheremanyinvestigatorshavelostthemselves''(RamonyCajal,1923).
Toexplorethesejungles,neuroscientistshavetraditionallyreliedonelectrodesthatsamplebrainactivityonlyverysparsely—fromonetoafewneuronswithinagivenregion.
However,neuralcircuitscaninvolvemillionsofneurons,soitisprobablethatneuronalensemblesoperateatamultineuronalleveloforgani-zation,onethatwillbeinvisiblefromsingleneuronrecordings,justasitwouldbepointlesstoviewanHDTVprogrambylookingjustatoneorafewpixelsonascreen.
Neuralcircuitfunctionisthereforelikelytobeemergent—thatis,itcouldarisefromcomplexinteractionsamongconstituents.
Thishypothesisissupportedbythewell-documentedrecurrentanddistributedarchitectureofconnectionsintheCNS.
Indeed,individualneuronsgenerallyformsynapticcontactswiththousandsofotherneurons.
Indistributedcircuits,thelargertheconnectivitymatrix,thegreatertheredundancywithinthenetworkandthelessimportanteachneuronis.
Despitetheseanatomicalfacts,neurophysio-logicalstudieshavegravitatedtowarddetaileddescriptionsofthestablefeatureselectivityofindividualneurons,anaturalconsequenceofsingle-electroderecord-ings.
However,giventheirdistributedconnectionsandtheirplasticity,neuronsarelikelytobesubjecttocontinuous,dynamicrearrangements,participatingatdifferenttimesindifferentactiveensem-bles.
Becauseofthis,measuringemer-gentfunctionalstates,suchasdynamicalattractors,couldbemoreusefulforchar-acterizingthefunctionalpropertiesofacircuitthanrecordingreceptiveeldresponsesfromindividualcells.
Indeed,insomeinstanceswherelarge-scalepopulationmonitoringofneuronalensem-bleshasbeenpossible,emergentcircuitstateshavenotbeenpredictablefromresponsesfromindividualcells.
Emergent-levelproblemsarenotuniquetoneuroscience.
Breakthroughsinunderstandingcomplexsystemsinothereldshavecomefromshiftingthefocustotheemergentlevel.
Examplesincludestatisticalmechanics,nonequi-libriumthermodynamics,andmany-bodyandquantumphysics.
Emergent-levelanalysishasledtorichbranchesofsciencedescribingnovelstatesofmatterinvolvingcorrelatedparticles,suchasmagnetism,superconductivity,superu-idity,quantumHalleffects,andmacro-scopicquantumcoherence.
Inbiologicalsciences,thesequencingofgenomesandtheabilitytosimultaneouslymeasuregenome-wideexpressionpatternshaveenabledemergentmodelsofgeneregula-tion,developmentalcontrol,anddiseasestateswithenhancedpredictiveaccuracy.
Webelievesimilaremergent-levelrich-nessisinstoreforcircuitneuroscience.
Anemergentlevelofanalysisappearstouscrucialforunderstandingbraincircuits.
970Neuron74,June21,20122012ElsevierInc.
Likewise,thepathophysiologyofmentalillnesseslikeschizophreniaandautism,whichhavebeenresistanttotraditional,single-celllevelanalyses,couldpoten-tiallybetransformedbytheirconsider-ationasemergent-levelpathologies.
TheBrainActivityMapastheFunctionalConnectomeToelucidateemergentlevelsofneuralcircuitfunction,weproposetorecordeveryactionpotentialfromeveryneuronwithinacircuit—ataskwebelieveisfeasible.
Thesecomprehensivemeasure-mentsmustbecarriedoutovertime-scalesonwhichbehavioraloutput,ormentalstates,occur.
Suchrecordingscouldrepresentacompletefunctionaldescriptionofaneuralcircuit:aBrainActivityMap(BAM).
Thismappingwilltranscendthe''structuralconnectome,''thestaticanatomicalmapofacircuit.
Instead,weproposethedynamicalmappingofthe''functionalconnectome,''thepatternsandsequencesofneuronalringbyallneurons.
Correlatingthisringactivitywithboththeconnectivityofthecircuitanditsfunctionalorbehav-ioraloutputcouldenabletheunder-standingofneuronalcodesandtheirregulationofbehaviorandmentalstates.
Thisemergentlevelofunderstandingcouldalsoenableaccuratediagnosisandrestorationofnormalpatternsofactivitytoinjuredordiseasedbrains,fosterthedevelopmentofbroaderbio-medicalandenvironmentalapplications,andevenpotentiallygenerateahostofassociatedeconomicbenets.
ImagingEverySpikefromEveryNeuronToachievethisvision,oneclearlyneedstodevelopnoveltechnologies.
Todate,ithasnotbeenpossibletoreconstructthefullactivitypatternsofevenasingleregionofthebrain.
Whileimagingtech-nologieslikefMRIorMEGcancapturewhole-brainactivitypatterns,thesetech-niqueslacksingle-cellspecicityandtherequisitetemporalresolutiontopermitdetectionofneuronalringpatterns.
Topreservesingle-cellinformationwhilerecordingtheactivityofcompletecircuits,vigorouseffortsmustbelaunchedtomassivelyupscalethecapabilitiesofbothimagingandnanoprobesensing.
Overthelasttwodecades,neuro-scientistshavemadetransformationaladvancesintechniquestomonitortheactivityofneuronalensembles.
Opticaltechniquesareminimallyinvasiveandcanprovidegreatspatialandtemporalexibility,havesingle-cellresolution,andcanbeappliedtolivingpreparations,evenawakebehavingones(Helmchenetal.
,2011).
Calciumimagingcanmeasurethemultineuronalactivityofacircuit(YusteandKatz,1991)(Figure1),anddespitealimitedtimeresolution,thistechniquecanpartiallyreconstructringpatternsoflarge(>1,000)populationsofneuronsinvitroorinvivo(GrienbergerandKonnerth,2012).
Calciumimaging,whileuseful,canonlyapproximatetherealfunctionalsignalsofneurons,anditispreferabletocapturethecompleteactivityofacircuitbyvoltageimaging(Peterkaetal.
,2011).
Currentmethodsforvoltageimaginginvertebratecircuits,however,cannotcaptureactionpotentialsatalargescalewithsingle-cellresolution.
Novelvoltagesensorswithbettersignal-to-noise,lessphotodam-age,andfastertemporalresolutionareneeded.
Continuedimprovementsarebeingmadeinvoltageindicators,andparticularlypromisingarenanoparticles,smallinorganiccompoundsthathavelargeabsorptionandhighlyefcientemis-sion.
Thesearerobustduringextendedilluminationandcanbeverysensitivetotheexternalelectriceld.
Zero-dimen-sionalnanoparticles,i.
e.
,quantumdots,couldbedirectlyusedtomeasurevoltageinneurons.
Othernanoparticles,suchasnanodiamonds(Mochalinetal.
,2012),mayprovideanevenhighersensitivitytomagneticandelectricelds.
Inaddition,byactingas''antennas''forlight,nano-particlescangreatlyenhanceopticalsignalsemittedbymoretraditionalvoltagereporters.
Butregardlessofthemethodchosenforimagingneuronalactivity,tocaptureallspikesfromallneurons,oneneedstoincreasethenumberofimagedneuronsandextendthedepthoftheimagedtissue.
Avarietyofrecentad-vancementsinopticalhardwareandcomputationalapproachescouldover-comethesechallenges(Yuste,2011).
Novelmethodsincludepowerfullightsourcesfortwo-photonexcitationofdeeptissue,fasterscanningstrategies,scanlessapproachesusingspatio-light-modulatorsto''bathe''thesamplewithlight,high-numericalapertureobjectivesFigure1.
Large-ScaleCalciumImagingofNeuronalActivity(A)Livingbrainslicefromprimaryvisualcortexofamousestainedwiththecalciumindicatorfura-2AM.
Morethanathousandneuronsarelabeledandcanbeimagedwithatwo-photonmicroscope.
FromYusteetal.
(2011).
(B)Thecalciumconcentrationinthesomaofaneuron(bottom)faithfullytrackstheelectricalringpatternofthecell(top).
FromSmettersetal.
(1999).
(C)Reconstructed''rasterplot''ofthespontaneousspikingactivityof754cellsfromasimilarexperiment.
FromCossartetal.
,2003.
NeuronNeuroViewNeuron74,June21,20122012ElsevierInc.
971withlargeeldsofview,engineeredpointspreadfunctionsandadaptiveopticscorrectionsofscatteringdistortions,light-eldcamerastoreconstructsignalsemanatingin3D,and,nally,advancesincomputationalopticsandsmartalgorithmsthatusepriorinformationofthesample.
Acombinationofmanyofthesenovelmethodsmayallowsimulta-neous3Dimagingofneuronslocatedinmanydifferentfocalplanesinanawakeanimal.
Inaddition,GRINbersandendo-scopesallowimagingdeeperstructures,suchasthehippocampus,albeitwithsomeinvasiveness.
Large-ScaleElectricalRecordingswithNanoprobesElectricalrecordingofneuronalactivityisnowbecomingpossibleonamassivelyparallelscalebyharnessingnoveldevel-opmentsinsilicon-basednanoprobes(Figure2).
Silicon-basedneuralprobeswithseveraldozenelectrodesarealreadyavailablecommercially;itisnowfeasibletorecordfromdozensofsitespersiliconneuralprobe,densely,atapitchoftensofmm(Duetal.
,2009a).
Stackingoftwo-dimensionalmultishankarraysintothree-dimensionalprobearrayswouldprovidethepotentialforhundredsofthousandsofrecordingsites.
Therearetechnicalhurdlestobesurmounted,butwhenthetechnologyisperfected,recordingfrommanythousandsofneuronsisconceivablewithadvancedspike-sortingalgorithms.
The''HolyGrail''willbetorecordfrommillionsofelec-trodes,keepingthesamebandwidth,reducingtheelectrodepitchdowntodistancesof15mm,andincreas-ingtheprobelengthtocorticaldimen-sionsofseveralcentimeters.
Thiswillrequiresignicantinnovationinsystemsengineering.
WirelessandSyntheticBiologyApproachesWealsoenvisiontechniquesforwireless,noninvasivereadoutoftheactivityofneuronalpopulations(Figure2).
Thesemightincludewirelesselectroniccircuitsbasedonsiliconverylarge-scaleintegra-tion(VLSI),syntheticbiologicalcompo-nents,ortheirhybrids.
Itiseasytounderestimatethepotentialoftoday'smicroelectronictechnology,andwethinkthatitwillultimatelybecomefeasibletodeploysmallwirelessmicrocircuits,un-tetheredinlivingbrains,fordirectmoni-toringofneuronalactivity,althoughtherearesignicanttechnologicalchallenges.
AsanalternativetosiliconVLSI,syntheticbiologymightprovideaninter-estingsetofnoveltechniquestoenablenoninvasiverecordingofactivity(Fig-ure2).
Thiscouldbeconsideredawirelessoption,albeitaradicallydifferentone.
Forexample,DNApolymerasescouldbeusedasspikesensorssincetheirerrorratesaredependentoncationconcentra-tion.
PrechosenDNAmoleculescouldbesynthesizedtorecordpatternsoferrorscorrespondingtothepatternsofspikesineachcell,encodedascalcium-inducederrors,servingasa''ticker-tape''recordoftheactivityoftheneuron.
ThecapabilityofDNAfordenseinformationstorageisquiteremarkable.
Inprinciple,a5-mm-diametersyntheticcellcouldholdatleast6billionbasepairsofDNA,whichcouldencode7daysofspikingdataat100Hzwith100-foldredundancy.
ABAMProjectRoadmapandChoiceofSpeciesForanygivencircuit,thereconstructionofactivitymightproceedinthreesteps.
First,initialmappingcouldbedoneusingcalciumimagingwithspikingreconstruc-tioncarriedoutat100Hz.
Thiscouldbeperformedwithimprovementstoexistingmethods.
Thesecondstepwouldinvolvevoltageimagingofactionpotentials(andsubthresholdelectricalactivity),ideallywithatemporalresolutionof1kHz.
Thesersttwostepscouldbecarriedoutin3Dyettheywouldbelimitedtosupercialstructures(KeyAretherecircuitattractorsWhatisthefunctionalconnectivitydiagramofacircuitWhatdetailedcomputationstakeplacelocallyWhatarethereal-time,multiple,long-rangeinteractionsthatunderliecognitivefunctionsandbehaviorHowdolocalcomputationsandlong-rangeinteractionsinuenceeachotherWhatarethepathsofinformationowDoalternativepathwaysproducesimilaroutputsWhenthebrain''organizes''itselfduringdevelopment,or''reorganizes''itselfafteraninjury,whatisactuallyhappeningtoactivitylocallyandgloballyWhenpharmacoactivedrugsalterbehavior,whatarethelocalandglobaleffectsonactivityWhenmemoriesaretransferredfromonebrainregiontoanotherovertime,howdoactivitypatternschangeWhatdesignprinciplescanbediscernedinhowthebrainfunctionsIsthereanunderlyingfunctionalarchitecturetothebrain'snetworksWhatarethetruefunctionalunderpinningsofperception,recognition,emotion,understanding,consciousness,andsubconsciousprocessesNeuron74,June21,20122012ElsevierInc.
973NeuronNeuroViewengagingdiversesetsofstakeholdersandthelaypublicearlyandthoughtfully.
OutcomesandAnticipatedBenetsTheBAMProjectwillgenerateahostofscientic,medical,technological,educa-tional,andeconomicbenetstosociety.
Indeed,thewidespreadeffectofthisresearchunderscorestheneedforittobecontrolledbythepublic.
Intermsofanticipatedscienticbenets,thegenerationofacompletefunctionaldescriptionofneuralcircuitswillbeinvaluabletoaddressmanyoutstandingquestionsinneuroscienceforwhichemergentfunctionalpropertiescouldbekey(Table1).
Together,answerstothesequestionscanopenthedoorstodecipheringtheneuralcode,aswellasunlockingthepossibilityofreverse-engi-neeringneuralcircuits.
Inadditiontopromotingbasicresearch,weanticipatethattheBAMProjectwillhavemedicalbenets,includingnovelandsensitiveassaysforbraindiseases,diagnostictools,validationofnovelbiomarkersformentaldisease,testablehypothesesforpathophysiologyofbraindiseasesinanimalmodels,anddevelop-mentofnoveldevicesandstrategiesfornecontrolbrainstimulationtorebalancediseasedcircuits.
Notleast,wemightexpectnovelunderstandingandthera-piesfordiseasessuchasschizophreniaandautism.
ManytechnologicalbreakthroughsareboundtoarisefromtheBAMProject,asitispositionedattheconvergenceofbiotechnologyandnanotechnology.
Thesenewtechnologiescouldincludeopticaltechniquestoimagein3D;sensitive,miniature,andintelligentnano-systemsforfundamentalinvestigationsinthelifesciences,medicine,engi-neering,andenvironmentalapplications;capabilitiesforstorageandmanipulationofmassivedatasets;anddevelopmentofbiologicallyinspired,computationaldevices.
AsintheHumanGenomeProject,whereeverydollarinvestedintheU.
S.
generated$141intheeconomy(Battelle,2011),technologicalandcomputinginno-vationsdevelopedinthecourseoftheBAMprojectwillprovideeconomicbenets,potentiallyleadingtotheemer-genceofentirelynewindustriesandcommercialventures.
IftheGenomeProjectwas''arguablythesinglemostinuentialinvestmenttohavebeenmadeinmodernscience''(Battelle,2011),theBAMProject,webelieve,willhavecomparableramications.
Finally,weshouldnotunderestimatetherepercussionsthatsuchaprojectcouldhaveforeducation.
Theproposedactivitiesarebroadlyinterdisciplinaryandwillleadtothetrainingofanewgenerationofscientistsandtheopeningupofnewstrategiesforevaluatingpeda-gogicaleffectiveness.
ACallforaCommunityEffortTosucceed,theBAMProjectneedstwocriticalcomponents:strongleadershipfromfundingagenciesandscienticadministrators,andtherecruitmentofalargecoalitionofinterdisciplinaryscientists.
Webelievethatneuroscienceisreadyforalarge-scalefunctionalmappingoftheentirebraincircuitry,andthatsuchmappingwilldirectlyaddresstheemergentleveloffunction,shiningmuch-neededlightintothe''impenetrablejungles''ofthebrain.
ACKNOWLEDGMENTSThiscollaborationarosefromaworkshopheldatChicheleyHall,theKavliRoyalSocietyInterna-tionalCentre,supportedbyTheKavliFoundation,theGatsbyCharitableFoundation,andtheAllenInstituteforBrainScience.
WealsothankA.
S.
Chiang,K.
Deisseroth,S.
Fraser,C.
Koch,E.
Marder,O.
Painter,H.
Park,D.
Peterka,S.
Seung,A.
Siapas,A.
Tolias,andX.
Zhuang—participantsatasmaller,subsequentKavliFuturesSymposium,whereinitialideaswerejointlyrened.
WeacknowledgesupportfromtheDOE(A.
P.
A.
),NHGRI(G.
M.
C.
),NIHandtheMathersFoundation(R.
J.
G.
),NIHandFondationpourlaRechercheetl'EnseignementSuperieur,Paris(M.
L.
R.
),andtheKeckFoundationandNEI(R.
Y.
).
Amoreextensiveversionofthispaperandaddi-tionaldocumentsabouttheBAMcanbefoundathttp://hdl.
handle.
net/10022/AC:P:13501.
REFERENCESBattelle.
(2011).
Economicimpactofthehumangenomeproject.
Inhttp://www.
battelle.
org/publications/humangenomeproject.
pdf.
Bock,D.
D.
,Lee,W.
C.
,Kerlin,A.
M.
,Andermann,M.
L.
,Hood,G.
,Wetzel,A.
W.
,Yurgenson,S.
,Soucy,E.
R.
,Kim,H.
S.
,andReid,R.
C.
(2011).
Nature471,177–182.
Chiang,A.
S.
,Lin,C.
Y.
,Chuang,C.
C.
,Chang,H.
M.
,Hsieh,C.
H.
,Yeh,C.
W.
,Shih,C.
T.
,Wu,J.
J.
,Wang,G.
T.
,Chen,Y.
C.
,etal.
(2011).
Curr.
Biol.
21,1–11.
Cossart,R.
,Aronov,D.
,andYuste,R.
(2003).
Nature423,283–288.
Du,J.
,Riedel-Kruse,I.
H.
,Nawroth,J.
C.
,Roukes,M.
L.
,Laurent,G.
,andMasmanidis,S.
C.
(2009a).
J.
Neurophysiol.
101,1671–1678.
Du,J.
,Roukes,M.
L.
,andMasmanidis,S.
C.
(2009b).
Dual-sideandthree-dimensionalmicro-electrodearraysfabricatedfromultra-thinsiliconsubstrates.
J.
Micromech.
Microeng.
19,10.
1088/0960-1317/19/7/075008.
Grienberger,C.
,andKonnerth,A.
(2012).
Neuron73,862–885.
Helmchen,F.
,Konnerth,A.
,andYuste,R.
(2011).
ImaginginNeuroscience:aLaboratoryManual(ColdSpringHarbor,NewYork:ColdSpringHarborPress).
Mochalin,V.
N.
,Shenderova,O.
,Ho,D.
,andGogotsi,Y.
(2012).
Nat.
Nanotechnol.
7,11–23.
Peterka,D.
S.
,Takahashi,H.
,andYuste,R.
(2011).
Neuron69,9–21.
RamonyCajal,S.
(1923).
Recuerdosdemivida:Historiademilaborcientca(Madrid:AlianzaEditorial).
Smetters,D.
K.
,Majewska,A.
,andYuste,R.
(1999).
Detectingactionpotentialsinneuronalpopulationswithcalciumimaging.
Methods18,215–221.
White,J.
G.
,Southgate,E.
,Thomson,J.
N.
,andBrenner,S.
(1986).
Philos.
Trans.
R.
Soc.
Lond.
BBiol.
Sci.
314,1–340.
Yuste,R.
(2011).
Imaging(ColdSpringHarbor,NewYork:ColdSpringHarborPress).
Yuste,R.
,andKatz,L.
C.
(1991).
Neuron6,333–344.
Yuste,R.
,McLean,J.
M.
,Vogelstein,J.
,andPanin-ski,L.
(2011).
ColdSpringHarb.
Protoc.
2011,985–989.
974Neuron74,June21,20122012ElsevierInc.
NeuronNeuroView