PreciseSpatiotemporalControlofVoltage-GatedExcitabilityinNeuralDendritesby

AlexisVoorheisFedorchak

Adissertationsubmittedinpartialsatisfactionofthe

requirementsforthedegreeof

JointDoctorofPhilosophywithUniversityofCalifornia,SanFrancisco

in

Bioengineering

inthe

GraduateDivision

ofthe

UniversityofCalifornia,Berkeley

Committeeincharge:

ProfessorRichardH.Kramer,ChairProfessorDanielE.Feldman

ProfessorChristophE.Schreiner

Fall2015

©2015Copyright,AlexisFedorchakAllRightsReserved

I

Abstract

PreciseSpatiotemporalControlofVoltage-GatedExcitabilityinNeuralDendritesby

AlexisVoorheisFedorchak

DoctorofPhilosophyinBioengineering

UniversityofCalifornia,Berkeley

ProfessorRichardH.Kramer(Chair) Neuraldendrites continuallyprove toharnessmorecomputational complexity thanpreviouslythought.Thevoltage-gatedionchannelsdistributedthroughoutadendritictreeare keydeterminants of dendritic excitability and computation.However, little is knownabout the specific functional impacts of voltage-gated excitability in discrete dendriticregions.Recently,anopticalrevolutioninneurosciencehasyieldedavastarrayofopticaltoolsforfunctionalinterrogationofneuronsandneuralcircuits.Onesuchtool,Quarternary-ammoniumAzobenzeneQuarternary-ammonium(QAQ),isanoptically-controllablesmall-moleculedrug thataffectsvoltage-gated ionchannels. In its trans conformation,which isphoto-induciblewithgreenlight,QAQdirectlyblocksallvoltage-gatedionchannelstested,butrapidlyun-blocksthosechannelswhenconvertedtoitscis formwithnear-ultravioletlight(Mourotetal.2012).Itdoesnotphoto-bleach,andcanberobustlyphotoswitchedbackandfourthtoeitherblockorunblockchannelsinamatterofmilliseconds.QAQisapromisingtool to control voltage-gated excitability in neural dendrites with the spatiotemporalprecisionoflight. InthisthesisweuseQAQtorapidly,reversibly,andlocallycontrolvoltage-gatedionchannelactivityinneuraldendritesusingtargetedlight.Awealthofexperimentalevidenceusingtraditionalpharmacologyisalreadyavailableaboutspecificvoltage-gatedionchannelsinCA1pyramidalcells,sowefirstapplyQAQviaapatch-pipettetoCA1pyramidalcellsandconfirmthat itworksasexpected inawhole-cell.We find that trans-QAQblockssomaticactionpotentials,blocksdendriticcalciumactivity,andenhancesEPSPsummation.TheseareallprocessesdrivenbyQAQ-sensitivevoltage-gatedionchanneltypesthateitherboost(sodiumandcalciumchannels)ordampen(potassiumchannels)intrinsicexcitability. WetheninvestigatethelevelofspatialcontrolwecanachievewithQAQusingdendriticcalciumimaging.Indeed,foruptothreesecondsafterphoto-switchingthemolecule,controlisextremelyprecise.Withthisknowledge,weuselocalblockofvoltage-gatedionchannelsand calcium imaging to confirmand extendprevious findings that voltage-gated calciumchannelactivityisrelativelyuniformthroughouttheapicaldendritictreeofCA1pyramidalcells.Finally,wespecificallyphoto-controlvoltage-gatedionchannelsintheapicaldendritesofCA1neuronstoexperimentallyprobewhetherdendrite-specificvoltage-gatedexcitabilityaffectsthedegreeofactionpotentialback-propagation.Wefindthatdendriticvoltage-gatedionchannelsdeterminewhetheraCA1pyramidalneuronwillundergostrongorweakback-

II

propagation,anotionthathasonlypreviouslybeenmodeled.

i

ContentsListofFiguresListofTables1. Introduction.............................................................................................................................1

1.1. Dendriticexcitability.....................................................................................................................11.2. Opticalmethodstomanipulateneuralactivity..................................................................41.3. OverviewofQAQ............................................................................................................................61.4. Thesispreview.................................................................................................................................7

2. Methods.....................................................................................................................................82.1. Materials............................................................................................................................................82.2. Solutionpreparation.....................................................................................................................92.3. QAQpreparation............................................................................................................................102.4. Animals..............................................................................................................................................102.5. Acutehippocampalslicepreparation...................................................................................102.6. Electrophysiologyprocedures.................................................................................................112.7. Two-photoncalciumimaging...................................................................................................122.8. Analysis….........................................................................................................................................15

3. Results......................................................................................................................................173.1. Whole-cellphoto-controlofneuronsinacuteslice........................................................17

3.1.1. Photo-controlofactionpotentialfiring....................................................................173.1.2. Long-termphotoswitchingandmembraneresistancewithQAQ...................183.1.3. VGICblockenhancesEPSPsummationinCA1pyramidalcells........................193.1.4. QAQtwo-photonsensitivity............................................................................................213.1.5. Photo-controlofCA1dendriticcalciumtransients...............................................23

3.2. TargetingVGICblocktoprecisespatialregions...............................................................263.2.1. LaserscatteringthroughCA1tissue...........................................................................263.2.2. Localdendriticcalciumblock........................................................................................27

3.3. Dendrite-specificphoto-controlofvoltage-gatedactivity...........................................303.3.1. DendriticblockofVGICsdoesnotaffectthesomaticstepresponse...............303.3.2. Localizedvoltage-gatedionchannelfunction.........................................................313.3.3. DendriticVGICsdeterminestrengthofactionpotentialback-propagation34

4. Discussion................................................................................................................................384.1. Conclusions&newhypotheses...............................................................................................384.2. ConsiderationsforusingQAQ...................................................................................................404.3. FutureDirections............................................................................................................................40

References......................................................................................................................................42

ii

ListofFigures1.1Dualrecordingsofaback-propagatingactionpotential...............................................................21.2Overviewofcompartmentalizeddendriticintegration................................................................31.3Dendriticdistributionofvoltage-gatedionchannels.....................................................................41.4SummaryofQAQ...........................................................................................................................................62.1SchematicofmodifiedMOMforcombinedCa2+imagingandphoto-control...................132.2Two-photon&488beamalignmentcalibration...........................................................................143.1Actionpotentialphoto-controlinCA1pyramidalneurons......................................................173.2Actionpotentialphoto-controlincerebellarPurkinjeneurons.............................................183.3Long-termAPphoto-controlandmembraneresistanceinwhole-cell...............................193.4VGICsdampenEPSPsummation,butdonotaffectsingleEPSPs...........................................203.5VGICsEnhanceEPSP-SpikeCouplinginCA1PyramidalNeurons.........................................213.6QAQsensitivitytotwo-photonlight...................................................................................................223.7Dark-relaxationfromcistotransduringtimeinwhole-cell....................................................233.8EffectofQAQonbAP-induceddendriticcalciumtransients....................................................253.9Laserscatteringthroughacutehippocampalcoronalslices....................................................273.10Localdendriticphoto-controlofbAPcalciumtransients.........................................................283.11Independentphoto-controlofsomaticexcitability.....................................................................303.12Localvoltage-gatedcalciumisrelativelyuniformacrossthedendriticarbor................323.13Branch-pointVGICblockhasnoeffectonEPSP-inducedspikeprobability.…...............333.14Cellsexhibitedeitherweakorstrongback-propagation..........................................................343.15DendriticVGICsdetermineback-propagationstrength............................................................363.16Dendriticmorphologydoesnotdifferbetweenstrongandweakgroups........................37

ListofTables2.1Listofchemicalsandreagents.................................................................................................................82.2ExternalSolution............................................................................................................................................92.3InternalSolution.............................................................................................................................................9

iii

Acknowledgements I’dliketothankmanypeoplefortheirsupportduringmyPhD.Firstandforemost,thankyoutomyadvisor,RichardKramer,forhisguidance,patience,andgoodhumorovertheyears.Hehasanincredibleknackforcomingupwithcreativeideasandframingscientificstories,and Ihavecertainly learneda lot fromhim. Ialsowant to thankmyother thesiscommittee members, Dan Feldman and Christoph Schreiner, and guest member HillelAdesnik for being my advocates and offering helpful and practical advice on my thesisresearch.Thanksalso toSteveConolly in thebioengineeringdepartmentwhoofferedanimportantearandadvice.Also,whethersheknowsitornot,myundergraduateprofessor,LaurelCarney,wasaphenomenalrolemodelformeasawomaninengineering.Shetookmeunderherwingandintroducedmetosystemsneuroscience,andforthatIamthankful. ThereareanumberofotherimportantpeopleIneedtothankaswell,becauseaPhDcanbealonelyendeavor,andwithoutthemIwouldhavequitgraduateschoolyearsago.Iwant to thank each and every member of the Kramer lab, past and present, for theircomraderyandunyieldingsensesofhumor—especiallywhenIneededitthemost.Specialthanks to Alexandra Polosukhina for never failing to offer enthusiasm and words ofencouragement,JeffLittforbad-movierecommendations,statisticsconsults,andoneverylongbikeride,ChrisDavenportforoversixyearsofpunsandone-linersandfindinghumorin everything, Christian Herold for dissertation motivation, Lars Holzhausen for realitychecks and coffee, Ming-Chi Tsai for engaging scientific discussions and advice, and, ofcourse,theGoldenGateBridgeforalwaysbeingthere. I alsowant to thankmy friendsoutside the lab forenrichingmy lifewith funandadventures over the years. Thanks tomy ski buddies (Katie Black, Matt Black, MichelleKonkle,JustinKonkle,DanielleDoane,MikeLee,MarkSena,etal.!)formany,manytripstoTahoe, sometimes for glorious skiing pursuits, and sometimes to go for a balmy hike inJanuary. Here’s to hoping El Niño delivers pantloads of powder this year! Thanks toStephanieGindlespergerforbeinganawesomeroommatethroughthefirstfewyearsofgradschool,andtoLowryKirkbyandMariaDadarlat,fellowmath/neurosciencenerd-ladiesandwonderfulpeople.Thanksalsotothegreatgroupofgradstudentsandpost-docswhohaveworkedwithmeattheBerkeleyScienceReview.Youmademygraduatecareeraboutmorethanjustlabwork,andintroducedmetoasciencecommunicationworldIpreviouslyknewnothingabout.BeinginvolvedwiththemagazinewasahighlightofmytimeatBerkeley.

Most of all, thank you toNate Lawrence for continually encouragingme in roughtimes,forbeingupforanyandalladventures,andforalwaysmakingmelaugh.You’vebeenaconstantsourceofloveandsupportforyears,andforthatI’meternallygrateful. Finally, of course, I want to thank my loving parents, Gaye Fedorchak and PaulFedorchak.I’vebeenincrediblyluckytohavesuchsupportiveparents,who,whentheironlychilddecided to takeoff across the country to attendgraduate school,werenothingbutunderstandingandencouraging.Myparentsarebothpsychologists,andsomeofthemostfunpeopletotalkaboutneurosciencewith.IthankthemforneverpushingmetopursueacareerIdidn’twant,orholdingmetoimpossiblestandards.Andmostofalltoday,Iwanttothankthemforgivingmeasenseofperspective,andconstantlyremindingmethatgraduateschoolistemporary.

1

1.Introduction

1.1. DendriticexcitabilityIn the central nervous system, neurons are the cells that communicate with one

another throughchemical andelectricalmeans togive rise toperceptions, thoughts, andactions.Inthelate1800s,RamonyCajalfirststainedaneurontoobserveitsstructure.Whathefoundwasthatneuronsarequitevariedinappearance.Eachneuronusuallycontainsasoma, dendrites, and an axon, but each of these structures has incredibly differentmorphologies.Cajalhypothesizedthatinformationinanervoussystementersthedendrites,flowstothesoma,andouttheaxon.Intheyearssince,hishypothesisprovedtobetrue.Forawhile,itwasgenerallyacceptedthattheadvantageofhavingdendriteswassimplytoallowa cell to fanout spatially and formmany synapseswith axonspassing through the area.Dendriteswere thought to passively relay synaptic information to the soma,which thenproducedanall-or-noneregenerativeactionpotential.

Decades later, various scientists called this view into question. Field potentialrecordingsofferedthefirstevidenceforregenerativeeventsindendrites(Andersen1960;Fujita&Sakata1962;Llinásetal.1968;Spencer&Kandel1961;Purpura&Shofer1965).Then, direct recordings from the inside of dendrites then confirmed the existence ofregenerativecurrentsinalligatorpurkinjeneurons(Llinas&Nicholson1971),andinbitsofCA1pyramidalcelldendriteonacoverslip(Wongetal.1979).Atthesametime,WilifridRallextendedanexistingcomputationalmodelingframeworkforaxonstodendrites(Rall1959),thenupdatedthisframeworkbyincorporatingdiscretecompartmentstoenabletheoreticalstudies of non-lineardendritic events (Rall 1964).Rall’s compartmentalmodel has sincebeeninfluentialinstudiesofdendriticfunction,byenablingresearcherstomakepredictionsand test experimental observations. In 1994, when whole-cell patch-clamping had beenextendedtodendrites (Stuartetal.1993),GregStuartandBertSakmannsimultaneouslypatch-clampedthedendriteandsomaofasinglelayer5corticalneurontorecordthefirstdirectevidenceofasomaticactionpotentialactivelypropagatingintothedendrites(Stuart& Sakmann 1994). This back-propagating spike was extinguished by application of thevoltage-gatedsodiumchannelblockers,QX-314andtetrodotoxin(TTX),indicatingvoltage-gatedsodiumchannelsnotonlycontrolspikingintheaxon,butalsogovernactionpotentialpropagationintothedendrites(Figure1.1).Indeed,dendritesnotonlycontainvoltage-gatedsodiumchannels,butmanyothervoltage-gatedionchannelstoo.

In fact, voltage-gated ion channels (VGICs) are main determinants of dendriticexcitability,meaninghowreadilydendriteswillundergoregenerativeeventssuchasactiveback-propagation (London & Häusser 2005; Sjöström et al. 2010). Since those earlyexperiments,VGICshavebeenbeen found toplay roles inmanydendritic computationalevents.InadditiontoL5pyramidalcells,somaticactionpotentialsactivelyback-propagateintothedendritesofCA1pyramidalcells(Sprustonetal.1995;Watersetal.2005),toanextent thatvariesevenwithinasinglecell-type(Goldingetal.2001;Bernard&Johnston2003).

2

Figure1.1:Dualrecordingsofaback-propagatingactionpotential.Inthefirstdirectevidenceofactionpotentialback-propagation,whole-cellvoltagerecordingofanactionpotentialrecordedatthesomaisoverlayedonthesameactionpotentialrecorded443µmfromthesomaofalayer5pyramidalcell.Afterthesodiumchannelblocker,QX-314diffusesintothedendrite(100slater),thedendriticpotential is greatly diminished, implicating voltage-gated conductances in dendritic back-propagation.FigureadaptedfromStuartetal1994.

Anevenmorerecentviewofdendritesisthatsingledendriticbranchesmayfunctionas independently excitableunitswithinaneuron,much like singleneurons function in acircuit (Branco & Häusser 2010; Figure 1.2). Such multiple layers of single-neuroncomplexity creates a much richer computational landscape than previously thought(Häusser&Mel2003).Studieshaveshownthatinsomecasesactivityoccurringinindividualbranches never even reaches the soma or produces an action potential (Miyakawa et al.1992;Wei2001;Ariavetal.2003;Polskyetal.2004;Gasparini&Magee2006;Losonczy&Magee2006).Theseisolateddendriticcomputationsarealsoobservedinvivo(Kamondietal.1998;Remyetal.2009;Smithetal.2013)andduringbehavior(Xuetal.2012;Smithetal.2013;Palmeretal.2014),suggestingthattheyholdapracticalrelevance. Usingtraditionalpharmacology,manipulationsofvoltage-gatedionchannelactivityinanentirecellhaverevealedavastwebofdendriticcomputationscontrolledbyspecifictypesofvoltage-gatedionchannels(Remyetal.2010;Bruneletal.2014).Whilesodiumchannelsboost back-propagating action potentials along pyramidal cell dendrites, potassiumchannelsdampenand sharpen theseback-propagating signals (Stuart&Sakmann1994).Bothsodiumandcalciumchannelshavebeenshowntoboostcoincidentexcitatorypost-synapticpotentials (EPSPs)andunderliedendritic spikes(Markrametal.1997;Stuart&Häusser 2001; Svoboda et al. 1997), whereas potassium channels have been shown tosharpen EPSPs (Takagi 2000), precisely control action potential timing (Fricker &Miles2000;Ariavetal.2003),andunderliesub-lineartemporalsummationofEPSPs(Cash&Yuste1998; Margulis & Tang 1998; Cash & Yuste 1999). Voltage-gated calcium channels alsoprovide an important source of calcium for plasticity processes in dendrites (Frick &Johnston2005;Sjöströmetal.2010;Goldberg&Yuste2005).

3

Figure1.2.Overviewofcompartmentalizeddendriticintegration.Coloredboxesontheleftdepictvariouscomputationallayersinapyramidalneuron(layer5examplemorphologyatfarleft).Signalshavebeenshowntobeintegratedinspines,branches,andsubregions liketheapicaltuft,obliquedendrites,andsoma.Within the largeright sidebox:a)examplemorphologyofaCA1pyramidalneuron.b)Modeledideathatdendritesactascomputationalsubunitswithina2-layeredmodelofintegration.c)Proposedschemeformulti-layeredsingle-neuroncomputation.FigureadaptedfromPalmer2014&HausserandMel2003.

Littleisknownabouthowtherelativedistributionofvoltage-gatedionchannelsaffectsdendriticfunction.Anatomicalimmunolabellingandelectron-microscopyhavehelpedparsegeneral locations of various voltage-gated ion channels in dendrites, which are foundthroughoutallneuraldendriteswithconsiderablevariationinrelativelevels,andchannelsubtypesfromcell-to-cell(Migliore&Shepherd2002;Trimmer&Rhodes2004;Lai&Jan2006).However,methodsforlocalizedperturbationarelimitedandtechnicallychallenging,so only a handful of studies have been done to address the functional profiles andcomputationalrelevanceofthesub-cellularvoltage-gatedionchanneldistributions(Figure1.3).

4

Figure1.3:Dendriticdistributionofvoltage-gatedionchannels.Currentlywehaveonlya low-resolutionunderstandingoftherelativedistributionsofthemanydifferenttypesofvoltage-gatedion channels in neural dendrites. These simple graphs combine information gleaned from bothexperimentalandcomputationalstudiestooutlinethecurrentknowledgeinthefield.a)Mitralcellsshowrelativelyuniformconductancesofallvoltage-gatedionchannelsacrossthedendriticarbor.b)Corticallayer5pyramidalcellsexhibitastronggradientofincreasingIhcurrentwithdistancefromthesoma,butrelativelyuniformdistributionsofallotherchannels.c)CA1pyramidalneuronshaverelatively uniform distributions of calcium and sodium channels, but greatly increasing A-typepotassiumandIhcurrentswithdistancefromthesoma.FigureadaptedfromMigliore&Shepherd2002.

Dendritic patch-clamping, though low-throughput and prohibitively difficult in thindendrites,hasrevealedthatvoltage-gatedsodiumandcalciumchannelactivityisrelativelyuniformalongCA1hippocampalpyramidalcellapicaldendrites(Magee&Johnston1995;Bittneretal.2012),butthatvoltage-gatedpotassiumchannelcurrentincreasesabout16-foldfromthesomatothedistaldendrites(Hoffmanetal.1997;Bittneretal.2012).Localperfusion of channel blockers is slow to act and has blurred spatial boundaries due todiffusion.Butithasshownthatsodiumchannelsinthemainapicaldendriteboostactionpotential back-propagation and dendritic coincidence detection of simultaneous actionpotentialsandEPSPs(Magee& Johnston1997;Stuart&Häusser2001),whilepotassiumchannelskeepthiscoincidenceactivityin-check,anddampenback-propagationinobliquedendrites(Stuart&Häusser2001;Fricketal.2003).Furtherresearchondendriticfunctionwill benefit from an ability to precisely control voltage-gated excitability in sub-cellularregionsandindividualbranches.1.2. Opticalmethodstomanipulateneuralactivity.

Much has been discovered about the nervous system by simple observation.Morphologicalimagingstudieshaverevealedextensivedendriticbranchingpatterns(Chenetal.2006;Jiangetal.2015;Markrametal.2015),aswellasthegenerallocationsofsub-cellularstructuresandproteins(Trimmer&Rhodes2004).Suchstudiesserveapurposetoconstrainourviewofhowthenervoussystemgivesrisetobehaviorandperception,buttoreally uncover the basis of neural processing one must establish cause-and-effectrelationships between functional mechanisms and neural outputs. We can do this bycarefully manipulating certain aspects of a system, rather than simply observing. Asmentioned already,methods tomanipulate neural activity can be sloppy and imprecise.Electricalstimulationnon-specificallytargetsallelectricallyactivemembranesinacertain

5

spatial vicinity with no regard to specific types of cells. Pharmacology allows us tomanipulatespecifictypesofionchannelsandproteinstoisolatetheeffecteachchannelhasingeneral,buttargetingdrugstospecificareasofadendritetodeterminethelocalizedroleachannelplaysinneuralprocessingisprohibitivelydifficult. Inthepastdecadeorso,anumberofopticaltoolstopreciselymanipulatebothneuralcircuitsandsub-cellularfunctionhaveemerged.Chemicallycagedneurotransmittershaveachemicalgroupthatcanbecleaved(uncaged)byilluminationwithspecificwavelengthlight,makingthemavailabletobindtoreceptorsonthesurfaceofcells(Trigoetal.2009).Withthese cagedcompounds, researchershavebeenable tomimic synaptic transmissionanddirectly assay its effect on dendritic processing. For example, glutamate uncaging ondendriticspineshasrevealedthatclustersofexcitatoryinputsonasingledendriticbranchwillproduceanon-lineardendriticspike,butthatinputsdistributedtomultiplebranchessummate linearly (Losonczy & Magee 2006; Yang et al. 2014). It’s also revealed thatdendrites preferentially respond to synaptic activity activated in specific temporalsequences (Branco et al. 2010). The ability to perform such experiments has greatlydeepenedourunderstandingofdendriticintegration(London&Häusser2005;Tran-Van-Minhetal.2015). In2005,amicrobialionchannelthatopenswithbluelighttodepolarizemembraneswasexpressedinneuronsandusedtocontroltheirfiringwithlight(Boydenetal.2005).Becausethepowerofthistechniquetoactivategenetically-targetedneuronswithlightwasunparalleledinspecificity,itkickedoffanentirelynewfieldcalledoptogenetics.Researchershavesincedevelopedanincrediblevarietyoftoolsalongthissamevein,fromchannelsthatwillhyperpolarizemembranesandsilenceneurons(Zhangetal.2007;Chowetal.2010),totoolsthatarefasterorhavedifferentkinetics(Yizharetal.2011),tored-shiftedvarietiesthatcanbemultiplexedtogether(Linetal.2013;Klapoetkeetal.2014).Theseopticaltoolshave already been used extensively to dissect neural circuits at a level never beforeattainable(Deisseroth2015). Jumpingoneleveldownfromthesystemstothecellularneurosciencelevel,anothertransformative method of optical perturbation is a bit a mix of both optogenetics andendogenous control of neural receptors like with optical uncaging. Optogeneticpharmacologyisaclassofresearchtoolsthataddacysteinepointmutationtoendogenousmembraneproteinssuchthatasyntheticphotoswitchableligandwillcovalentlyattachitselftotheproteins.Thesetoolsallowgenetically-targetableopticalcontrolofspecificneuralionchannelandreceptorsubtypes,likeblockofKv3.1potassiumchannels(Fortinetal.2011),activationofionotropicglutamatereceptors(Volgrafetal.2006;Szobotaetal.2007),blockofnicotinicacytylcholinereceptors(Tochitskyetal.2012),orblockofGABAareceptors(Linetal.2014;Lin&Tsaietal.2015). Lastly,anothermethodthat’sbeendevelopedisverysimple.In2008,DirkTraunerandRichardKramer’s labscollaboratedtodevelopsingle-component,non-geneticopticalmethodstocontrolactivity,termedoptopharmacology(Fortinetal.2008;Banghartetal.2009). Taking known pharmacological blockers, the labs attached a reversiblyphotoisomerizablechemical,azobenzene.Thefirstofthesechemicals,termedAAQ,hasbeenuniquelyusedtorestorelightresponsestobindretinas(Polosukhinaetal.2012),andred-shiftedversionshavealsobeendeveloped(Mourotetal.2011).

6

1.3. OverviewofQAQA recently developed addition to the optopharmacology toolbox is a smallmolecule

namedQuarternary-ammoniumAzobenzeneQuarternary-ammonium(QAQ)(Mourotetal.2012).QAQisbasedonthedrugQX-314,whichiscommonlyusedasanintracellularsodium-channelblocker,thoughithasbeenshowntoblockotherVGICsaswell(Perkins&Wong1995; Talbot & Sayer 1996). In QAQ, a light-sensitive azobenzene links two QX-314moleculestocreate,essentially,aQX-314moleculethatissensitivetolight.It’sbeenshownthat QAQ is reversibly photoisomerized to its cis-form with illumination wavelengthsbetween 360 and 390 nm, and photoisomerized back to its low energy trans-formwithwavelengthsbetween480and540nm(Fortinetal.2008;Mourotetal.2012).Azobenzeneshowsnophotobleachingeffects(Beharry&Woolley2011),andQAQhassofarbeenshowntoswitchbackandfourthbetweencisandtransformsindefinitely.

Init’strans-form,QAQ’schemicalconformationisrelativelystraight,whichallowsittoblockvoltage-gatedionchannelactivity,butinitscis-formthemoleculehasabentstructureandcannotblockvoltage-gatedionchannels(Figure1.4).PreviouslyinculturedcellsQAQwasfoundtoblockalltypesofvoltage-gatedchannelstosomedegree(Figure1.4),includinghigh-voltage activated calcium channels (L-type & N-types), sodium channels (includingTTX-resistanttypes),andaslewofpotassiumchannels,butnotchannelsfromotherfamiliesincludingKir,HCN,AMPA,andNMDAchannels(Mourotetal.2012).QAQactsonvoltage-gatedionchannelsfromtheinsideofacellandcannotcrosslipid-membranes,soit’sactionscanbeconfinedtoasingleneuron.

QAQ’sreversibility,apparentimmunitytophoto-bleaching,andsensitivitytovoltage-gated ion channels make it a great candidate for fine spatiotemporal manipulations ofdendritic excitability. Such manipulations will be critical to advance our knowledge ofdendriticfunction.

Figure1.4:SummaryofQAQ.Indarknessorafterabsorbinggreenlight,QAQresidesinthetransconfiguration, whereas near uv light rapidly photoisomerizes the molecule to the bent cisconfiguration.Trans-QAQ has been shown to block all voltage-gated ion channels tested so far(right).Photoswitchingisameasureofpercentcurrentblockintrans,calculatedas(icis–itrans)/icis.FigureadaptedfromMourotetal.2012.

7

1.4. Thesispreview

Thegoalofthisthesisistobothprovideaproof-of-conceptthatQAQwillbeusefulforlocalizedfunctionalstudiesofdendriticexcitability,andpresentnewbiologicalevidenceforpreciserolesofvoltage-gatedexcitabilityinsub-cellularregions.UsingQAQ,wereportthat ion channel block is limited to an optically-targeted area. By blocking voltage-gatedexcitability—essentiallymaking a neuron passive—we confirm previous findings in CA1pyramidal cells that voltage-gatedpotassium channels aredominant players in temporalsummation. We also provide evidence that dendritic voltage-gated calcium channelsactivated during action-potential back-propagation are relatively uniform in distributionalong the apical dendritic tree, and similarly distributed across both oblique and maindendrites. Finally,we offer the first experimental evidence that dendriticVGICs underliewhetherCA1pyramidalcellswilldisplaystrongorweakactionpotentialback-propagation,something that has only previously been simulated (Golding et al. 2001), andwe revealdifferingvoltage-gatedexcitabilityprofileswithineachcellpopulation.

8

2. Methods

We used both whole-cell patch-clamp electrophysiology, and two-photon calciumimagingtorecordneuralactivity.Herewereportthematerialsandmethodsusedtoobtainourresults.2.1. Materials

Table2.1:Listofchemicalsandreagents.

Purpose Chemical Source

Slicing&Recording

NaCl

Sigma-Aldrich

KClMgCl2NaH2PO4NaHCO3D-GlucoseCaClSodiumL-ascorbicacidPicrotoxinGluconateHEPESEGTANa-GTPMg-ATPKOH95%O2,5%CO2tank PraxairDouble-distilled(DD)H2O(>18mOhmresistance) Millipore

ImagingAlexa594

MolecularProbesFluo-4Fluo5f

OpticalVGICControl QAQ Synthesizedin-house

9

2.2. SolutionpreparationExternalSolutionforslicingandrecording(rattissue):

1) Dissolvethechemicalslistedintable2.2in~650mLDDH2O.

Table2.2:ExternalSolution.

Chemical FinalConcentration(mM)NaCl 126KCl 2.5MgCl2 1.3NaH2PO4 1.25NaHCO3 26D-Glucose 10

2) AddDDH2Otobringsolutiontojustunder750mL.3) CheckpH. If less than7.22pH,add~10µMof8molarKOHata time tobring

solutionto7.22-7.28pH.Ifover7.28,remakesolution.4) Check solution osmolarity with an osmometer. If above 296 mOsmol, bring

solutiontobetween296and290mOsmol.Iflessthan290,remakesolution.5) CalciumchlorideshouldonlybeaddedafterbubblingO2throughthesolutionfor

at least 10 minutes to saturate (this ensures that all calcium will remain insolution.CaClmayformprecipitatewithoutbubblingfirst).Forslicing,add1mMtotalCaCl.Forincubationandrecording,add2.5mMtotalCaCl.

*For calcium recording experiments, we added 100 μM picrotoxin bydissolving ~6 mg/100 mL in External Solution, and mixing for ~30minutes,oruntilcompletelydissolved.

InternalSolutionforwhole-cellcurrent-clamp:

a) Dissolvethechemicalslistedintable2.3in~40mLDDH2O.

Table2.3.InternalSolution.

Chemical FinalConcentration(mM)K-gluconate 116HEPES 20KCl 6NaCl 2EGTA .5Mg-ATP 4Na-GTP 0.3

b) AddDDH2Otobringsolutionto~45mL.

10

c) CheckpH.Iflessthan7.18pH,add~5µMof1molarKOHatatimetobringsolutionto7.18-7.22pH.Ifover7.22,remakesolution.

d) Checksolutionosmolaritywithanosmometer.Ifabove293mOsmol,bringsolutiontobetween287and293mOsmol.Iflessthan287,remakesolution.

e) Filterwith.22µmvacuumfilterandsplitinto500µLaliquots.Storeat20oCforupto2months.

f) Before eachday of recording, thawone aliquot and adddesired amounts ofQAQ,Alexa594,andcalciumindicatordye.

2.3. QAQpreparation

QAQwassynthesizedin-houseasasolidpowderwithaTFAcounter-ion.Itsmolecularweight is 722.2, and it is chemically stable in aqueous solution (we have used years-oldaqueous stock solutions with great success). In order to ensure proper and consistentconcentrationfromday-to-day,wemadea~20mMaqueousstocksolutionsofQAQ,500µLatatime,thenaliquotedout5µLportionsandstoredthemina-20oCfreezer.Beforeeachday of experiments,we added 5 µL of 20mMQAQ to 1mL internal solution for a finalconcentrationof100µMQAQinthepatchpipette.

Sometimesourin-housesynthesisyieldedleftoversiliconparticulatesintheQAQ-TFApowder.Forthisreason,afterdissolvingenoughpowderfor20mMstocksolutioninDDH2O,wealwayscentrifugedthesolutionandpipetteditintoaneweppendorftube,discardinganysiliconparticulatesthatcollectedatthebottomofthetubeaftercentrifuging.Afterremovingthis potential silicon debris,we checked the final concentration of our stock solution bymeasuring its optical densitywith a nanodrop (Thermo Scientific), and adjusted aliquotvolumesaccordingly.2.4. Animals

ExperimentalprocedureswereperformedwithapprovalfromtheUCBerkeleyAnimalCareandUseCommittee.Priortoexperiments,Sprague-Dawleyrats(CharlesRiver)werehousedinanOfficeofLaboratoryAnimalCarefacilityatUCBerkeley,keptona12h–12hlight–darkcycleandhadadlibitumaccesstofoodandwateratalltimes.

2.5. AcutehippocampalslicepreparationAcutehippocampalbrainsliceswerepreparedasfollows:

1) Priortoexperiments,wecooledexternalslicingsolution(section2.2.1)to4oC,andheatedawaterbathto34oC.

2) Rats(P14–P23)weredeeplyanesthetizedusingisoflurane.3) Whenbreathingslowedtolessthan1breath/minutewequicklydecapitatedtherat

and dissected out the brain in chilled external slicing solution bubbledwith 95%oxygen.

4) Positioneddorsal-sideuponachilledplatform,wecutoff thecerebellumandthefrontallobe(~rostralthirdofthebrain)withacleanrazorblade,carefullyseveredthemidlinetoseparatethetwohemispheres,crazy-gluedthebrainrostralsidedowninavibrotome(LeicaVT1000-S),thenfilledtheslicingchamberwithchilledexternalslicingsolutionbubbledwith95%oxygen.

5) Totargetthehippocampus,thefirstslicewasmadestartingabout2100µMdownfromthesurfaceofthetissue(caudalendup),andsubsequentsliceswere350µm

11

thick.6) Ascut,eachslicewastransferredtoaholdingchamberfilledwithexternalrecording

solution(section2.2.1),andbubbledwith95%oxygen.7) Brainsliceswereincubatedfor~20minat34°Candstoredatroomtemperaturein

anoxygenatedchamberuntilused(upto6hoursfromslicing).

2.6.ElectrophysiologicalproceduresFor experiments using only electrical recordings, neurons were visualized with

infraredDodt-contrastimagingonanuprightOlympusmicroscopeandidentifiedbasedontheirmorphology.Forcombinedimagingandelectricalrecordings,neuronswerevisualizedwithdiffuseobliqueinfraredlightfromaDodtcontrasttube(Siskiyou)positionedbelowtheslice. Recording pipettes (4–6 MΩ) were pulled from borosilicate glass, and filled withinternalsolution(section2.2.2)containingeither100µMQAQand40µMalexa594,or100µMQAQ,40µMalexa549,andeither50µMFluo-4or200µMFluo-5f(Yasudaetal.2004).SignalswereacquiredusinganAxopatchamplifier(AxonInstruments),low-passfilteredat2–3kHz,andsampledat30–50kHz. Membrane voltage was recorded under current-clamp in bridge mode. Seriesresistancesmeasuredbetween10and35MΩ,andexperimentswereterminatedifthisrangewasexceeded.Therestingmembranepotentialofallcells intheresultspresentedinthisthesis was between -70 and -55 mV. This is consistent with typical resting membranepotentials of CA1 hippocampal neurons (Staff et al. 2000). If a cell’s resting membranevoltageexceededthisrange,orweneededtoinjectmorethan50pAholdingcurrenttokeepmembrane voltage near -60 mV, experiments were terminated. For calcium imagingexperiments,restingmembranepotentialsremainedinanevennarrowerrange(-65to-58mV).Thisislikelybecausetheseexperimentsrequiredustoholdcellsforatleastanhour,andcellsrestingoutsidethisrangetypicallydidnotlastaslongasnecessary.Temporalsummation

To measure synaptic summation, a series of four EPSPs at 50 Hz was evoked bystimulatingSchaffercollateralaxonswhenrecordinginCA1pyramidalneurons.EPSPswererecordedatthesomainpresenceofpicrotoxin(100µM)toblockinhibitoryPSPs.TemporalsummationwascalculatedastheratiooftheamplitudesofthelastandfirstEPSPinatrain(EPSP4/EPSP1).EPSP-spikecoupling

ForEPSP-spikecouplingexperiments,adouble-barrelstimulatingpipettewithabouta5µmwidetipwasplacedinthestratumradiatum.WegentlymovedthepipettetodifferentpositionsuntilanEPSPwasgenerated,thenslowlyincreasedthesimulationamplitudeuntilreliableactionpotentialfiringoccurredin390nmlight(cis-QAQ),thenslowlydecreasedthestimulationamplitudebacktobaseline.Thecellwasallowedtorestfor20secondsbetweeneachstimulationpulse,andwerecordedaround100EPSPstotalduringthecourseofoneexperimenttogetanaccuratemeasureofspikeprobability.Forbranch-pointblockEPSP-spikecoupling,wefirstvisualizedthedendriteswithaflorescentreddye(Alexa594),andthenplacedthedoublebarrelpipettebetween5and20µmfromanidentifiedbranch.

12

Stepresponses For step response experiments,we injected either 120ms (Figure 3.2), or 250ms(Figures3.1,3.6,3.7,&3.11)withcurrentamplitudesnecessarytoreachthresholdvaryingbetween25pAand300pA.Forplotsofcurrentvs.numberofactionpotentials,weincreasedthestimulusby50pAabovethresholdwitheachsuccessivestep.Back-propagatingactionpotentials Toelicitback-propagatingactionpotentialsforcalciumimagingexperiments(Figures3.8,3.10,3.12,3.14,&3.15),weinjectedatrainofthree2-mslongcurrentpulsesintothesomaat40Hz.Theamplitudeofthebriefcurrentinjectionswasrelativelyhigh(~700-1000pA) in order to overcome the time needed for capacitive charging to bring a neuron tothreshold in just2ms.At thebeginningofeachexperiment,wechoseacurrent injectionamplitudethatreliablyevokedspikeswitheachpulse.Duringa40Hzpulsetrain,trans-QAQdidnotabolishsubsequentactionpotentialsinatrainasitdoesduringastep-response,butitdidgreatlyreduceboththeiramplitudesandafterhyperpolarizationscomparedwithcis-QAQ.2.7.Two-photonCalciumImagingSet-up&calibration

Two-photon calcium imagingwas performed on two separatemicroscope systems.Experiments in figure 3.8 were done on a home-built two-photon microscope runningScanimagesoftware(Pologrutoetal.2003).AChameleonTi-Sapphirelaser(Coherent,Inc.)was tuned to810nanometersand the laserbeamwasdelivered througha40xOlympuswaterimmersionobjective(.8NA)abovetheslice.TocontrolQAQ,380nmand500nmweredeliveredwithafiber-opticPrismatixLEDthroughacondenserbelowtheslice.Ashutterwas used to shield the photomultiplier tubes from potentially damaging light duringphotoswitching.Imageswereacquiredat25Hzin16×16-pixelframescovering5×5µm.ImagingwastriggeredviaaTTLpulsefromtheelectrophysiologysoftwarepClamp. Two-photonimagingexperimentsinfigures3.10,3.12,3.14,&3.15wereperformedat810nm(ChameleonTi-Sapphirelaser;Coherent,Inc.)onaMoveableObjectiveMicroscope(MOM; Sutter, Inc.) running Scanimage software (Pologruto et al. 2003). Infrared two-photon light was delivered above the slice through a 20x Olympus water immersionobjective(.95NA).Imageswereacquiredateither13.5or27Hzin32x32or16x16-pixelframes, respectively. TheMOMwasmodified to include a Lumencore light source in theepiflorescent pathway, a TTL-controlled removablemirror to switch betweenwide-fieldepiflorescentandlaserscanninglightpaths,anda488laserco-alignedwiththetwo-photonlaser (Figure 2.1). Both the home-built and MOM microscopes were equipped withelectrophysiologyheadstagesandamplifiers.

13

Figure 2.1: Schematic ofmodifiedMOM for combined calcium imaging and photo-control.AMoveable Objective Microscope (MOM) from Sutter was modified with a bright light source(Lumencore) forwide-field,whole-dendrite, and background photo-control. Itwas alsomodifiedwitharemovablemirror(RM)onamotorthatwasdrivenbyaTTLpulseforprotocolautomation.Finally,a488nmlaser(CoherentOBIS)waspipedintothemicroscopeviathescanmirrorssothatitcouldbeeasilytargetedtoaregionofinterest.Thisschematicwasmodifiedfrom(Euleretal.2009).

Duringset-up,weco-alignedthe488nmlaserbeamwiththe2plaserbeam,butminute

alignmentvariationsfromday-to-dayoffsetthetwobeamsbyuptoasmuchas10µm.Weusedatwo-photonimagetooutlineROIsfor488nmlaserscanning,soweneededtoknowthiserroreachdayinordertoadjustthetargetROIappropriatelyduringlocalVGICblockexperiments(seeFigure3.10).Tomeasuretheoffsetatthestartofeachday,weplacedaflorescentslide(Chroma)belowtheobjectiveanddefinedanROI(samerelativeshape/sizeused for local block experiments). We then scanned with high 488 laser power tophotobleachthetargetedROI,imagedthephotobleachedregionagainintwo-photonmode,andmeasuredtheoffsetbetweentheintendedROIandtheactualphotobleachedROI(Figure2.2).

14

Figure 2.2: Two-photon & 488 beam alignment calibration. Two-photon image of calibrationprocedure. Displayed is a two-photon image of a florescent slide. Yellow rectangles outline theintendedROI,anddarkerregionsdepicttheregionsactuallyscannedwiththe488laser.Here,theoffsetis10µmtotheleftand1µmdownfromthetargetchosenbasedonthetwo-photonimage.SubsequentROItargetsforthedaywereadjustedby10and1µminthexandydirection,accordingly.

ExperimentalProtocols For ideal calcium recording conditions, we added 100 µM picrotoxin to externalsolutiontoblockGABAacurrents,whichwilldampencalciumsignals(Kanemotoetal.2011).Picrotoxinisnotparticularlysoluble,soweletthesolutionmixforatleast30minutestoensurepicrotoxinpowdercompletelydissolved.Toourinternalsolution,weadded100µMQAQ,40µMAlexa594,&either50µMFluo-4(Figure3.8)or200µMFluo-5f(Figures3.10,3.12, 3.14, and 3.15). For experiments measuring somatic action potential generation,experiments were started at least 20 minutes after achieving a whole-cell patchconfigurationtoallowQAQ,Alexa594andcalciumdyesufficienttimetodiffusethroughoutthecell.Experimentsinvolvingdendriticfunctionwerestartedafteratleast30minutespostbreak-in.

TocombineQAQphoto-controlwith two-photoncalcium imaging,wereliedon twoproperties:QAQisinsensitiveto810nmtwo-photonlight(Figure3.6),andisbi-stableinthedarkwitharelaxationconstantfromcistotransof>5s.QAQ’sinsensitivityto810nmlightallowedustoimagecalciumwithoutphoto-isomerizingthemolecule.ByexploitingQAQ’sbi-stabilityandswitchingoffthephoto-isomerizationlightjustpriortoimaging,wecouldavoidhighbackgroundinthetwo-photonimageduetostimulationlightbleedingthroughtheemissionfilter.

To imageoneregionof interest(ROI)alongadendrite inexperimentsnot involvingtargeted488nmlaserlight,theentirecellwasfirstilluminatedwitheither390or540nmlightatanopticalpowerdensityof less than5mW/mm2 toavoid tissuedamageand/orphysiologicaleffectsfromlight(Yizharetal.2011).Thatlightwasthenswitchedoffand,lessthanthreeseconds later,calciumsignalsweresimultaneouslyevokedbysomaticcurrentinjectionandimaged.Thisconstitutedonetrial.Thephotoswitchingwavelengthwasthenswitched,andtheprocessrepeated.Wealternatedwavelengthsbetweentrialsandineachtrialat leastthreerecordingswereacquiredforsignalaveraging.Thecellwasallowedtorecoverfor20secondsaftereachrecording. ForlocalVGICblockexperiments,wefirstilluminatedthewholecellwith390nmlightthroughtheepiflorescentpathontheMOM.Thenweswitchedlightpathsand,fortrialswithVGICblock,scannedatargetedROIwith488nmlaserlight(<5mW/mm2).Nomorethan

15

threesecondsafterthe390nmlightwasgiven,weinjectedthree2-mscurrentpulsesatthesomatoelicitactionpotentialsat40Hz,andimagedcalciuminthedendrites.Wethenletthe cell recover for 20 seconds. This constituted one trial.With each trialwe alternatedbetweencis and trans conditions,andat least six trialswereperformed,yieldingat leastthreetrialsperconditionforeachROI.Fortheweak/strongback-propagatingexperiments,weilluminatedonlythedendriteswitheither390or540nmlightbeforeswitchingtothescannedlightpathandimagingcalcium.AllimagingandphotoilluminationwascoordinatedwithTTLpulsescontrolledbytheelectrophysiologysoftware,pClamp(AxonInstruments). During the course of an experiment,we took care to use the least amount of lightnecessaryinordertominimizeeffectsoncell-healthofdirectexposuretoUVlight,andAlexa594 photobleaching. Occasionally damaged cells exhibited strong, irreversible plateaupotentialswithexcessivelightstimulation.Weexcludedthesecellsfromouranalysis.2.8AnalysisElectrophysiology InputresistancewascalculatedastheslopeofI/Vcurveclosetorestingmembranepotential. Action potential amplitudewasmeasured from restingmembrane potential topeak. In response to a hyper polarizing current step, the sag ratiowas calculated as thesteady-statevoltagedividedbytheinitialsag.Voltagetraceswithsimilarsteady-statevalueswerechosentocompareacrossneurons.SynapticsummationwascalculatedastheratiooftheamplitudesofthelastandfirstEPSPinatrain(EPSP4/EPSP1).Imaging Average raw pixel values were extracted from semi-automatically defined ROIs.GeneralregionswerefirstoutlinedmanuallyinImageJ,thenthresholdedinMATLAB.ImagesweresortedandanalyzedwithacustomMATLABprogram.Foreachdendriticlocation,atleastthreetrialswereaveragedperexperimentalconditiontoachievebettersignal-to-noise.Calcium transients were quantified as change in calcium florescence (G; fluo-5f dye),normalizedbyredflorescence(R;alexa594),asfollows:

𝑑𝐺/𝑅(𝑡) =𝐺(𝑡) − 𝐺*𝑅(𝑡)

Wherethebaselines(Go)weretakenasanaverageofthefirst500millisecondsofeachtrace.All imagingdata shown in figures andused for statistical analysis representspeakdG/Rvalues,exceptforthedatainfigure3.8,forwhichdF/Fwasused.AmeasureofcalciumusingdG/Rismorerobusttonoisebecausetheredflorescencesignalusedfornormalizationisorders ofmagnitude stronger than the green calcium florescence. Also, dG/R inherentlynormalizesforchangesinabsolutedyeamount(differentdendriticdiameters,etc.),andisrelativelyinsensitivetobaselineflorescence(Yasudaetal.2004). Formorphologicalanalysis,wetracedneuronsusingthesimpleneuriteplugininFIJI(Longair at al. 2011), and performed a sholl analysis to examine dendritic branchingpatterns,aswellascountedthenumberofprimaryapicalbranchpointsineachcell.

16

StatisticsForparametricdata,valuesarepresentedinthetextasmeans±standarderrorofthe

mean, and plotted with error bars representing standard error. Two-tailed unpaired orpaired t testswere used to evaluate the difference between two groupmeans. For non-parametricdata,valuesaregivenasmedianswithbootstrapped95%confidenceintervals,and displayed using box plots (middle line = median; box = 25th and 75th percentiles;whiskers=mostextremedatapointsotherthanoutliers).Fornon-parametricpaireddata,weusedaWilcoxonsignedranktest,andfornon-paireddataweusedaWilcoxonrank-sumtest. To compare distributions, we used a two-sample Kolmogorov-Smirnov test. Thenumberofcellsforeachexperimentisindicatedinthefigurelegends,orinthetext.

17

3. Results

3.1. Whole-cellphoto-controlofneuronsinacuteslice. QAQ,asanoptically-controllablevoltage-gatedionchannelblocker,hasthepotentialto be a powerful tool to investigate both sub-cellular excitability and voltage-gated ionchannels’complexcontributiontodendriticcomputations.Inthisfirstsection,weintroduceQAQ intoneurons inacuteslice to confirm that itsactionsonvoltage-gated ionchannelsworkasexpectedinvarioussingle-neuroncomputations.3.1.1Photo-controlofactionpotentialfiring To confirmprevious resultsondissociatedhippocampal cell cultures (Mourotet al.2012),weloaded100µMQAQintoCA1pyramidalcellsinacutebrain.Asindissociatedcellcultures,QAQreversiblyregulatedactionpotentialfiringinCA1pyramidalneuronsinacuteslices(Fig.3.1a).

Figure 3.1: Action potential photo-control in CA1 pyramidal neurons. a) Example voltageresponsestoa300pA,250mslongcurrentinjectionatthesoma.Controlisreversible,aslightwasswitchedbackandfourthfrom390to540nm.b)Increasingcurrentinjectionamplitudesin50pAincrements yields increasinglymore action potentials in cis-QAQ conditions, but only one actionpotentialatallcurrentlevelsundertrans-QAQconditions(p<.001;n=7).

In 390 nm light, which converts QAQ to its inert cis form, we saw a normal CA1pyramidal cell step response (Staff et al. 2000). In 390 nm light, cells firedmore actionpotentialswithincreasingcurrentamplitudes(Fig.3.1b),andinter-spikeintervalsincreasedwithsubsequentactionpotentials.Whenweilluminatedcellswith540nmlighttoconvertQAQ to itsVGIC-blocking trans-form, cells still fired a first actionpotential, but repeated

18

firingwasabolished (Fig3.1a,b).Persistenceof the first actionpotential is likelydue totrans-QAQ’sfunctionasause-dependentblocker,suchthatitonlyblockschannelsoncetheyareintheopenstate(Mourotetal.2012;Hille2001).DuringthefirstactionpotentialQAQrapidlyblockstheopenchannelstothenabolishsubsequentfiring.Somephoto-controlofthefirstactionpotentialwasapparent:actionpotentialamplitudewasdecreased,andthehalf-widthwasincreased,aspreviouslynotedinculturedhippocampalneurons(Mourotetal. 2012).We also sawan increased after-depolarization (ADP) following the first actionpotential in540nm light (Fig.3.1a),whichwasnotapparent indissociatedhippocampalcells,buttheADPcanbeattributedtopartial-blockofvoltage-gatedchannels(Jensenetal.1994;Azouzetal.1996;Metzetal.2005).QAQhadasimilareffectincerebellarPurkinjeneurons(Fig.3.2), thoughcellsoften firedan initialburstof2-3spikesathighercurrentinjectionamplitudesinthesecells,ratherthanjustoneinitialspike(Fig.3.2b),whichcanagainbeattributedtoQAQ’suse-dependentnature.

Figure 3.2: Action potential photo-control in cerebellar Purkinje neurons. a) Example voltageresponsestoa150pA,120mslongcurrentinjectionatthesoma.Controlisreversible,aslightwasswitchedbackandfourthfrom380to500nm.b)Increasingcurrentinjectionamplitudesby50pAincrements yields increasinglymore action potentials in cis-QAQ conditions, but only 1-2 actionpotentialsintrans-QAQconditions(p<.001;n=5).

3.1.2Long-termphotoswitchingandmembraneresistancewithintracellularQAQ. Forcertainexperimentsitisnecessarytoholdaneuronalwhole-cellpatchforupto90minutes. For this thesis research, specifically, to ensure that QAQ has equilibratedthroughouttheextensiveCA1dendritictree,experimentsondendriticfunctiondidn’tevenstartuntil30minutesafterwhole-cellbreak-in(seeFigure3.7forapossibleequilibrationtime-course).QAQ’slong-termactioninacellhasyettobetested,soweassayedthelevelofaction potential photo-control over the course of an experiment once photo-control hadreachedsteady-state(typicallyafterabout20minutesofwash-in,thoughloworhighaccessresistances shortened or lengthened this time). Action potential photo-control was nodifferentataround90minutespostbreak-inthanitwasjustafterreachingsteadystateat

19

around20minutespost-break in (p= .92, n=11;Wilcoxon sign rank test; Figure3.3a).During a 250ms current injection about 150 pA above threshold, the number of actionpotentialsdecreasedonaverage from5 incis-QAQto1 in trans-QAQatboth timepointsassayed.Wealsomeasuredmembraneresistanceimmediatelyuponbreak-in(presumablybeforeQAQhadtimetodiffuseinatall),andattheterminationeachexperiment(excludingterminationsduetocelldeath,oralostpatch),membraneresistancewasnotsignificantlydifferentthanuponbreak-in(p=.95;n=8;Wilcoxonsignranktest;Figure3.3b).

Figure 3.3: Long-term AP photo-control and membrane resistance in whole-cell. a) Actionpotentialphoto-controlduringtimeinwhole-cell,quantifiedusinga“photoswitch(PS)index,”whichis calculated as (#APs390 -#APs540)/ (#APs390 +#APs540). b)Membrane resistancemeasured uponbreak-in, and again before the end of each experiment. Neithermeasure is significantly changedduringthecourseofalong-termrecording.

3.1.3VGICblockenhancesEPSPsummationinCA1pyramidalcells. Previousresearchhasshownthatblockingeithervoltage-gatedpotassiumchannelsorHCNchannelsenhancestemporalsummationinCA1pyramidalcells(Margulis&Tang1998;Georgeetal.2009),butthatatthesametimevoltage-gatedsodiumandcalciumchannelscanboostsub-thresholdEPSPs(J.Magee&Johnston1995;Lipowskyetal.1996;Gillessen&Alzheimer1997).To test theneteffect thatallVGICsacting togetherhaveonEPSPsandtemporalsummation,wepatchedCA1pyramidalcells,letthemfillwithQAQ(100µM),thenstimulatedaxonsinthestratumradiatumtoelicitfourEPSPsat50HzfromSchafercollateralinputs (Fig. 3.4a, b).WhenweblockedVGICs, singleEPSPswerenot affected (p = 0.08),indicatingthatVGICsactingtogetherdonotboostsingleEPSPs(Fig.3.4c).However,wedidfind that the summation ratio of the fourth to first in a 50 Hz train of four EPSPs wasenhancedwithVGICblock(Fig.3.4d),aspreviousstudieshavefoundwhenblockingvoltage-gatedpotassiumchannels(Cash&Yuste1998;Cash&Yuste1999;Ramakers&Storm2002).Thisresultindicatesthatvoltage-gatedpotassiumchannels,whichdampenactivity,arethedominantplayersinthisphenomenon.

20

Figure3.4:VGICsdampenEPSPsummation,butdonotaffectsingleEPSPs.a)Schafercollateralaxons were stimulated to produce EPSPs in the post-synaptic CA1 pyramidal neuron, whileillumination lighttoblockorunblockchannelswasappliedtothewholecell.b)Examplesomaticvoltage recording. FourEPSPswere elicitedwith an extracellular electrode at 50Hz. c)BlockingVGICswith500nmlighthadnosignificanteffectonthefirstEPSPpeak(p=.08;n=13;blackoverlayrepresentsmean±S.E.).d)ThesummationratioofthefourthtothefirstEPSPpeakwasenhancedfrom3.6(±.2)withnoVGICblock,to4.4(±.3)withVGICblock(p=.0081;n=13).Greensignifies500nm(trans-QAQ)conditions,whereaspurplesignifies380nm(cis-QAQ)conditions.

CertainvoltagegatedionchannelshavebeenshowntoincreasetheprobabilitythatanEPSPwillelicitanactionpotential,aprocesstermedEPSP-spikecoupling(Fricker&Miles2000;Jesteretal.1995;Daoudaletal.2002).Inlightoftheseresults,weexpectthatblockingallvoltage-gatedchannelsinCA1pyramidalcellswillaffectEPSP-spikecoupling.BecauseCA1cellsalwaysfireonesmalleractionpotentialundertrans-QAQconditions(Figure3.1)wewereabletophoto-controlvoltage-gatedchannelsinthewholecell,includingthesoma,andstillmeasureactionpotentialevents.Ourresultsconfirmthatthevoltage-gatedchannelsinCA1pyramidalcellsdo indeedenhanceEPSP-spikecoupling,andthatQAQaffects thisfunction,asexpected(Figure3.5).

21

Figure3.5:VGICsEnhanceEPSP-SpikeCouplinginCA1PyramidalNeurons.a)Examplevoltagerecordingincisconditions(purple)andtransconditions(green).Thestimulationartifactfromanextracellular simulating pipette in the perforant pathwaymarks the time of EPSP simulation. b)InitialEPSPslopewasmeasuredoverthefirst2msfollowingtheupwardinflectionpointoftheEPSP.Spikeprobabilityfrom3to4mV/mswassignificantlygreaterwithVGICsactive(purple),thanwithVGICblocked(green)(p=.0031),butnotforslopesfrom1to2mV/ms(p=.76;n=3cells).

3.1.4QAQtwo-photonsensitivity. Onewaytomeasuredendriticactivityiswithflorescentcalciumimaging(Yasudaetal.2004). Currently, green calcium indicator dyes are themost effective dyes available forimagingcalcium,asreddyeshaveatendencytogetsequesteredinendoplasmicreticulum.Thesegreendyesrequireone-photonexcitationwavelengthsinthesamespectralrangeasQAQ, but can also be excited with two-photon wavelengths. In order to use QAQ inconjunctionwithtwo-photoncalciumimagingforstudiesdescribedlaterinthisthesis,weneededtoknowwhetherQAQistwo-photonsensitive.Totestthis,wefirstilluminatedwith540 nm light, or 390 nm light to convert themolecule into either its trans or cis forms,respectively.Thenweimmediatelyscannedthesomawithtwo-photonlight,usingimaging-intensitiesandacquisitiontimesequaltothoseweusedforlaterexperiments(sections3.1.5,3.2.2, 3.3.2, and 3.3.3), and injected a current pulse to assay the step response in eachcondition(Figure3.6a,c). Usingaratiooftheintegratedstepresponses(totalchargetransfer)incisvstransasasensitivemeasureofactivity,wefoundnoeffectoftwo-photonlightoneitherformofQAQ.Whencompared to the typicalone-photoncharge transfer ratio, scanningcis-QAQatanytwo-photonwavelengthbetween750nmand950nmhadnoeffect(Figure3.6b).Similarly,scanningtrans-QAQwithanytwo-photonwavelengthsbetween800nmand950nmwasnotsignificantlydifferentthantheone-photonchargetransferratio(Figure3.6d),alsoindicatingthattrans-QAQisunaffectedbytwo-photonimaginglight.Theseresultsletusperformtheimagingexperimentscoveredlaterinthisthesis.

22

Figure 3.6: QAQ sensitivity to two-photon light. a) Example voltage traces used to determinewhether2plighthadanyeffectoncis-QAQ,asmeasuredinCA1pyramidalcells.Asequenceof390nmlight,then810nm2plight,then390nmwasgiven.b)Groupdatafortheratioofchargetransferincis-QAQtochargetransferntrans-QAQ,binnedin50nmincrements.Theinitialboxrepresentstheone-photonchargetransferratioforeachcelltested,whichissignificantlygreaterthanalltwo-photonwavelengthstested(p=0.032,.030,.016,.0087,respectively).c)Examplevoltagetracesusedtodeterminewhether2p lighthadanyeffecton trans-QAQ.b)Binnedgroupdata for theratioofchargetransferincis-QAQtochargetransferintrans-QAQ.Theone-photon(normal)chargetransferratioforeachcelltestedissignificantlygreaterthanalltwo-photonwavelengthstested(p= .032,.0043,.016,respectively).

Itisimportanttonotethatthestepresponsesseenwhentestingcis-QAQ’stwo-photonsensitivity (Figure3.6a)do reflect a slight voltage-gated channel block.Actionpotentialslaterinthecurrentpulseareshorterandwider,comparedtocis-QAQmeasurementsusingjustelectrophysiology,asabove(Figure3.1).Thisisduetoa5-secondshutterdelaybetweenphotoisomerizationand imaging inherent inour two-photonmicroscopeset-up for theseexperiments.Cis-QAQslowlyrelaxesbacktoit’slow-energytransformindarknesswithatimeconstantontheorderofseconds,onlyfullyrelaxingbackafter30seconds(Mourotetal. 2012). After just 5 seconds in darkness, this slow relaxation is reflected in the step

23

response.Interestingly,wefoundthatthetimecourseofdark-relaxationspeedsupwithtimeinwhole-cell(Figure3.7).ThiscouldbebecauseQAQ’sdiffusionintothedendritictreeisrelativelyslow,onlyreachingequilibriumandsteady-statedarkrelaxationkineticsatabout30 minutes post whole-cell break-in (Figure 3.7b). To control for this possibility, allsubsequent experiments on dendritic functionwere performed at least 30minutes afterbreak-intoletQAQ(100µM)fullyequilibrateinthedendrites.

Figure3.7:Dark-relaxationfromcistotransduringtimeinwhole-cell.QAQhasbeenpreviouslyshowntorelaxfromcisbacktoitslow-energytransformindarknesswithatimeconstantontheorderofseconds.a)Exampleofthesteady-staterelaxationisapparentinthestepresponseafter5seconds in darkness. b) Steady-state relaxation kinetics are not reached until >30minutes post-break-in.Photoswitch(PS)indexisameasureofthedifferenceinmeanactionpotentialamplitudesduringastepresponse:(cis-dark)/(cis+dark).Fivesecondsofdarkrelaxationyieldsabouta40%decreaseinmeanactionpotentialheightduringacurrentinjectioninsteady-state.Dataisseparatedinto20-minutebinsfordisplay.Theinitialandfinaldatapointsbeforetheendofrecordingforeachcellarestatisticallydifferent(p=.04,n=8;Wilcoxonsignranktest).Accessresistancesforeachcellremainedinarangefrom20to35mOhmduringeachrecording.

3.1.5Photo-controlofCA1dendriticcalciumtransients.

One important dendritic phenomenon observed in many neuron types, includinghippocampalCA1pyramidalneurons,isaninfluxofcalciumfollowingAPbackpropagation(Fricket al., 2003;2004; Sprustonet al., 1995; Stuart et al., 1997).BackpropagatingAPs(bAPs)canserveasacrucialsignalfortheinductionofsynapticandnon-synapticformsofplasticity, lower the threshold for the generation of dendritic regenerative potentials, ortrigger the dendritic release of transmitters/neuromodulators such as endocannabinoids(Benderetal.,2007;Watersetal.,2005).

TogaugeQAQ’simpactondendriticcalciuminflux,wemeasuredCa2+transientsindifferentlocationsalongCA1pyramidalcelldendriteswithtwo-photonmicroscopy.TherearechallengesinusinglightbothtomanipulateandtomeasurebAPcalcium.However,twokeypropertiesofQAQmadethisexperimentfeasible.Indarkness,cis-QAQisbi-stablewith

24

arelaxationtimeconstantontheorderofseconds(Mourotetal.2012),sothatwhenweswitched off photoswitching light, QAQ remained in the same photostatewhilewe thenevokedandimagedbAP-Ca2+signals(seeMethods;Figure3.8a).WealsodeterminedthatQAQisinsensitiveto810nmtwo-photonimaginglight(Fig3.6),whichallowedustoimagecalciumatthosewavelengthswithoutaffectingQAQ’sphotostate.

We found that QAQ decreased bAP-triggered Ca2+ transients. Apical dendritictransientsweresmallerintrans-QAQthanincis-QAQ(Figures3.8b-d).TheamountofCa2+blockalsoincreasedwithdistancefromthesoma(Figures3.8c-f).QAQdoesdirectlyblockvoltage-gatedcalciumchannels,whichlikelycontributestothedecreaseinCa2+florescenceundertrans-QAQ.ButgivenpreviousstudiesindicatingthatoverallcalciumchanneldensityremainshomogenousthroughouttheCA1pyramidalcelldendritictree(Mageeetal1998,MiglioreandShepherd2002),wecanattributetheprogressiveincreaseincalciumsignalblockalongthedendritetofactorsotherthancalciumchannelblockateachimagingsite.This result agreeswith previous studies that functionalVGICs are necessary for bAPs topropagatenormallythroughtheapicaldendritictreeofCA1pyramidalcells.

Onecaveatoftheexperimentsinsection3.1isthatvoltage-gatedionchannelswerecontrolledintheentirecellatonce.Wehaveconfirmedtheroleofvoltage-gatedchannelsinavarietyofdendriticfunctions,confirmedthatcontrolisquickandreversible,andconfirmedthat QAQ does, in fact, control functions previously shown to occur in the dendrites.However,we cannotbe absolutely sure that itwas the voltage-gated ion channels foundspecifically in the dendrites that accounted for these results, as previous experimenterscannotbeabsolutelysurethatvoltage-gatedchannelsinthedendritesaccountforspecificphenomenon unless they are specifically patch-clamping a dendrite, or locally perfusingchannel blockers onto just the dendrites. Particularly in the case of the calcium imagingexperiment, it isunclearwhatcontributionbothcalciumchannelsdirectlyblockedat theimagingsites,andinitialcontrolofactionpotentialgenerationhasonthetransientblockwereport.Theseconcerns,aswellasourabilitytolocalizevoltage-gatedionchannelblockwillbeaddressedinthefollowingsections.

25

Figure3.8:EffectofQAQonbAP-induceddendriticcalciumtransients.a)Experimentalprotocol.QAQwasphotoisomerizedby5 s-longpulsesof380nmor500nm light, followedby immediatestimulationofAPtrains(fourAPsat25Hz)viasomaticcurrentinjectionsandCa2+measurementsalong the apical dendrite. b) Four successive bAP-Ca2+ traces recorded under alternating lightconditions.Correspondingsomaticvoltagetracesareshownbelow.c)ExampleCA1pyramidalcell.Imagingregionsofinterest(ROIs)areindicatedbyreddotsandnumbers.d)AveragedCa2+tracesrecordedattheROIsshownin(c).PurpletracesindicateCa2+transientsrecordedfollowing380nmlightpulses,andgreentracesthoserecordedfollowing500nmlightpulses.Eachtraceisanaverageofatleastsixindividualtraces.e)AverageoftheblockoftheCa2+transientintegralbytrans-QAQasfunctionofdistancefromthesoma.f)AverageoftheblockoftheCa2+transientamplitudebytrans-QAQasfunctionofdistancefromthesoma.Reddotsindicatedatapointsfromexamplecellshownin(c)and(d).Blockiscalculatedas(data385–data505)/data385.n=6cells.

26

3.2TargetingVGICblocktoprecisespatialregions.

AkeyadvantageofusingQAQtoopticallycontrolvoltage-gatedionchannelblockisthat lightcanbe targeted tosmalloptical regionsof tissue,suchassub-cellulardendriticregions.Mainparametersthatlimitthespatialprecisiontowhichvoltage-gatedionchannelscanbecontrolledinclude:thediffractionlimitoflight,thepoint-spreadfunctionofthelight-deliverysystem,thelightscatteringimpartedbybiologicaltissueitself,andthefactthatQAQisafreely-diffusiblesmallmolecule.Theopticaldiffractionlimitrepresentsthephysicallimitforthedegreeofspatial localizationwithinwhichwecanphotoswitchQAQ.Forpracticalmicroscopy,thisfundamentallimitisapproximately½thewavelengthoflight,andwewillexploretheotherparametersinthissection.

3.2.1. LaserscatteringthroughCA1tissue.

Biologicaltissueisknowntoscatterphotons(Niemz2004).Inordertodeterminethe

limits thatourhippocampalslicepreparationplacedonspatially-targeted light,wemadeslices of rat hippocampus at varying thicknesses (89-165 µm) andmeasured the size oftargetedlaserlightspotsthroughtheseslices.

TobeabletoblockVGICsinafocalregionweneededtocombinebackgroundlightforcis-QAQphotoconversionwithtargetedlaserlightforlocaltransphotoconversion.Todoso,weusedfull-field390nmlightasinsection3.1,andequippedourtwo-photonimagingmicroscopewitha488nmlaser(seeMethods),respectively.QAQisreadilyphotoisomerizedtoitstransformwithanywavelengthsbetween480nmand550nm(Fortinetal.2008).Tomeasurethespotsizedeliveredbya488nmlaserthroughhippocampaltissue,weplacedeach acute slice on top of a florescent slide (Chroma), then focused down through thehippocampus onto the florescent slide (Figure 3.9a). With the 488 nm laser, we thenphotobleacheddiffraction-limitedspotsintheflorescentslideunderneathvariouslocationswithinboth theCA1cellbody layer (stratumpyramidale)or thedendritic layer (stratumradiatum),thenremovedthetissueandthenimagedthephotobleachedspots(Figure3.9b).

After measuring line profiles through the center of each spot (5-12 spots perthicknessandlayer;exampleprofiles:Figure3.9c),wefoundasignificantdifferenceinthehalf-width of laser light focused through the cell body layer (4.7 ±0.52), vs. through thedendriticlayer(2.0±0.22µm)attissuethicknessesover120µm(p=.002;Figure3.9d).Ingeneral,we foundasignificantly increasedhalf-width through89µmof tissue in thecellbodylayer(2.3µm,CI:[1.9,2.8])vs.notissue(1.7µm,CI:[1.6,1.8]);p=0.019;Wilxoconrank sum test), though no significant difference in the dendritic layer at 89 µm tissuethickness(1.6µmCI:[1.4,1.9]),comparedwithnotissue(p=.67;Wilcoxonranksumtest).Focalspothalf-widthinthedendriticlayerisonlybarelysignificantlyincreasedatthe165µmthickness(2.3µm,CI: [1.9,2.8];p= .05).Theseresultstellusthatthedendritic layerscatters488nmlaserlightsignificantlylessthanthecellbodylayer,andthatphoto-controlcantheoreticallyberestrictedto~2-3µmatacellbodyneartheslicesurface,1.5-2µmindendritesneartheslicesurface,andto2-3µmindeeperdendrites.

27

Figure3.9:Laserscatteringthroughacutehippocampalcoronalslices.a)Experimentalprotocol.Anacuteslicewasplacedundera20xwater-immersionobjective.Thena488laserbeamwasfocusedthrough either the CA1 cell body layer (Stratum Pyramidale), or the dendritic layer (StratumRadiatum)tophotobleachavarietyoffocalspotsonaflorescentslidebelow.Wethenremovedthesliceandimagedtheresultingphotobleachedspots.b)Exampletwo-photonimagingplanefollowingspot-photobleachingthroughaslice.StratumPyramidale(S.Pyr.)andStratumRadiatum(S.Rad.)aremarkedaccordingtowheretheywereduringlaserphotobleaching.c)Examplephotobleachedspotsthrough each tissue thickness. Top traces are line-profiles through the example spots at 138 µm(arbitrarygray-valueunits).d)Groupdatashowingsignificantlybroaderhalf-widthsthroughS.Pyr.thanthroughS.Rad.atthicknessesgreaterthan120µm(p<.05).

3.2.2. Localdendriticcalciumblock. BecauseQAQisadiffusiblesmallmolecule,wewantedtoknowtheextenttowhichitsactionscanberestrictedtoatargetedregionofdendrite.AfterconfirmingthatQAQisnottwo-photonsensitive(Figure3.6),weusedtwo-photoncalciumimagingtomeasure localdendritic activity evoked by a train of back-propagating action potentials. To only block

28

VGICsinaspecificregionofdendrite,wefirstilluminatedtheentireCA1pyramidalneuronwith390nmlighttogloballyswitchQAQtoitsinertcisform,thenscannedasmall(5-20µm)region of interest (ROI)with a 488 nm laser to photoisomerize QAQ in that region (seeMethods).Threeshort somaticcurrent injectionsat40Hzelicitedacalciumsignal in thedendrites,andwemeasuredflorescencechangesbothwithinandinthe5µmimmediatelynexttothetargetedROI(Figure3.10a,b).

Figure3.10:Localdendriticphoto-controlofbAPcalciumtransients.a)Experimentalset-upandexampleROI imagehighlightingthedifferencebetweenpeakcalciumsignals in390nmlight(cis-QAQ)vs. after scanning the targetedROI (whitebox)with488nm light (trans-QAQ).b)ExamplecalciumtransientsrecordedwithinandnexttothetargetROI.Greenindicatesrecordingsdoneaftertargeted 488 laser scanning, and purple indicates recordings done after 390 nm whole-cellillumination.Actionpotentialswereelicitedbythreebriefcurrentinjections,andbecausewealwaysilluminatedthewholecellwith390nmfirstnodifferencewasseeninvoltageatthesomabetween488nmand390nmconditions.c)TherewasasignificantdifferenceinthemedianamountofblockedcalciumsignalwithinthetargetROI,vsimmediatelyadjacenttothetargetROI(p=2.5415e-08;n=15cells,45ROIs).d)InasubsetofslightlylargerROIs,therewasnosignificantdifferenceinthelevelofcalciumblockneartheedge(inside0-3µmedge)ofatargetROIandthelevelofblockinthecenter(5-8µmfromtheedge)oftheROI(p=.602;n=8cells,16ROIs).

29

We found that calcium signal block during a back-propagating action potential isconfinedtoalocally-targetedregion(Figure3.10c).BecausetheempiricallengthconstantofCA1pyramidalcelldendriteshasbeenmeasuredtobeontheorderof200to550µm(Meyeretal.1997;Williams2004),weassumed thatanyvoltagechangecausedbyblocking ionchannelsinsuchasmalllengthofdendrite(5-20µm)isnegligible,andthatthelocalcalciumblockwemeasured reflects a direct blockof local voltage-gated calcium channel activityduringback-propagatingactionpotentials.Ofdatapooledacross13cellsand28targetedROIs,calciumisreducedinatargetedregionby17.6%(CIs:[13.2%,28.6%];p=2.5415e-08,Wilcoxonsignranktest;Figure3.10b,c),whereascalciumtransientsinthe5µmjustpastthetargetedROIarenotsignificantlyaffected(0.3%medianreduction,CIs:[-5.59,5.89];p=.784;Wilcoxonsignranktest;Figure3.10b,c).Thisisarelativelysmallamountofblock,consideringthatQAQisknowntoblockL-typecalciumchannelsbyabout78%andCav2.2channelsbyabout60%(Mourotetal.2012).However,manytypesofvoltage-gatedcalciumchannelsarepresentinCA1dendrites(Lai&Jan2006),andL-type,N-type,T-typeandR-types are all known to be active during back-propagating action potentials (Fisher et al.1990;Helmchenetal.1996).QAQ’seffectonRandT-typecalciumchannelsremainstobetested,so it ispossiblethattherelatively largelevelof leftovercalciumafter localphoto-controlcomesfromtheseionchannels. Inordertotesttheextenttowhichdiffusionintooroutofourtargetedphoto-controlareainthethree-secondgapbetweenphotoswitchingandimagingmayhaveaffectedourmeasurements, we analyzed calcium block near the edge the target ROI (Figure 3.10d).Previousstudiesfoundthatcalciumdiffusionindendritesisextremelyrestrictedtowithin3-5µmfromitssource(e.g.avoltage-gatedionchannel)(Korkotian&Segal2006;Biessetal.2011),soweusedthisasananalysisguideline.InasubsetoftargetROIs10-20µmlong,calciumblockwithin0-3µmfromtheedgeoftheROI(21.4%,CI:[10.4%,33.7%])wasnodifferentfromthelevelofblock5-8µmfromtheedge(19.3%,CI:[8.3%,31.0%];p=.602).This indicates that diffusion of QAQ (and calcium) at the boundary of a targeted ROI isnegligibleinthethreesecondsbetweenphotoswitchingandimaging,andthatphoto-controlindeedhasthepotentialtobeaspreciseasopticallypossible(Figure3.9addressesthisforacutehippocampalslices,specifically).

30

3.3Dendrite-specificphoto-controlofvoltage-gatedactivity.3.3.1.DendriticblockofVGICsdoesnotaffectthesomaticstepresponse. We next tested whether blocking only dendritic VGICs affected the somatic stepresponse.Tostart,wepatchedaCA1pyramidalcellandletQAQ(100µM)diffuseinforatleast20minutes.Fordendrite-specificphoto-control,werestrictedilluminationlighttoourfieldofview(450µmdiameter),thenmoveda20xobjectiveovertheapicaldendritessuchthatthecell’ssomawasjustoutofsight(Figure3.11a).Becausethisset-upleavesthesomainthedark,voltage-gatedionchannelsatthesomawereblockedwhenphoto-controllingthedendrites.With somatic illumination, the cell’s step response to a current injectionwassimilartowhatwesawwithwhole-cellillumination(Figure3.1).Firingreturnedtonormalwithsomatic390nmillumination,butblockingVGICsinjusttheapicaldendrites,however,hadnoeffectonthesomaticstepresponse(Figure3.11b).Thenumberofactionpotentialsfiredwithsomaticilluminationswitchedfromamedianof5.5(CI:[3.5,6.9])to1.2(CI:[1,3.1];p=.002)withVGICblock,butactionpotentialgenerationwasn’taffectedbydendriticillumination(p=.5;Figure3.11c).ForamoresensitivemeasureofwhetherthesomaticstepresponsewasaffectedbydendriticVGICactivity,welookedattotalchargetransferin390vs. 540nmdendritic illumination (Figure3.11d).With somatic illumination, total chargetransferwas7%larger(ratio=1.07;CI:[1.05,1.14];p=.002;Wilcoxonrank-sumtest)withVGICs unblocked, but with dendritic illumination there was no change in total chargetransfer(ratio=1,CI:[.99,1.002],p=.945;Wilcoxonrank-sumtest).

31

Figure 3.11: Independent photo-control of somatic excitability. a) Experimental set-up.Photoswitchinglight(graydottedcircle)waspositionedeitheroverthesomaandaxon,orovertheapicaldendrites,thena250mscurrentpulsewasinjected.b)ExamplestepresponseswhenVGICswerephoto-controlledinthesoma,orjustinthedendrites.c)Thenumberofactionpotentialselicitedwith somatic VGICs unblockedwas significantlymore thanwith somatic VGICs blocked, with orwithoutdendriticVGICsblocked(p<.01).Therewasnodifferenceinthenumberofactionpotentialsevokedwithdendriticphoto-control(p>.05).d)Therewasasignificanteffectinthechargetransferratioof390to540nmlightwithsomaticphoto-control,butnotwithdendriticphoto-control(p=.00017).

3.3.2.Localizedvoltage-gatedionchannelfunction. Dendritic patch-clamp studies have previously shown that voltage-gated calciumchannelsalongthemainapicaldendritearerelativelyuniformlydistributed,thoughvariablefrom patch-to-patch (J. C. Magee & Johnston 1995; Migliore & Shepherd 2002). QAQ-mediatedlocalblockofvoltage-gatedcalciumsignalsalongthedendritictreeconfirmedthisfinding.Asinsection3.2.2,weassumedthatVGICblockinadendriticlengthsmallerthan20µmhasanegligibleeffectonmembranepotential and that calciumblock is thusadirectmeasureof localized functional calciumchannel activity.We found that the level of local

32

calciumblockwasindependentofdendriticdistanceacrossthedendriticmeasuredrange(48µm to225µm;Figure3.12a),which confirmsprevious results fromdendriticpatch-clamping.Inadditiontothemainapicaldendrite,wewereabletoprobeobliquedendrites.Duetotheirsmallsize,obliquedendriteswerepreviouslyoff-limitstolocalinvestigation.Inthesameveinasarelativelyuniformmainapicaldendriticvoltage-gatedcalciumchanneldistribution, we found that there was also no significant difference between main andobliquelocalcalciumblock(Mainblock:15.7%,CI:[10.9%,20.7%],obliqueblock:20.4%,CI:[6.5%,29.2%];p=.943;Figure3.12b). Thoughwe found no general differences in voltage-gated calcium channel functionbetweenobliqueandmaindendrites,ithasbeenproposedthattherecouldexistspecifichotspots of excitability along the dendritic tree (Fisher et al. 1990; Wathey et al. 1992;Fitzpatricketal.2009).ThereisalsosomeevidencethatL-typecalciumchannelsclusteratbranch-points(Shitakaetal.1996),whichcouldtheoreticallycompensateforjumpsinaxialresistancethatcanpromptfailureofsignalpropagationthroughabranchpoint(Stuart&Häusser2007).Inlightoftheseideas,wecomparedthepercentsignalblockatbranch-pointstotheblockmeasuredatleast5µmawayfromanybranch-point.Wefoundnosignificantdifference between block at branch-points (15.3%block, CI: [8.2, 30.6]) and non-branchpoints(18.5%block,CI:[10.7,28.7];p=.61;n=13;Figure12c).Takentogether,theseresultsdescribeadendriticlandscapeofvoltage-gatedcalciumcurrentthatisgenerallynon-specificfordistancefromthesoma,dendritetype,orbranchpoints,thoughitisstillpossiblethatasmall fraction of dendritic regions containmore or less calcium current, and that smallfractioniswashedoutinourpopulationdata.

Figure 3.12: Local voltage-gated calcium is relatively uniform across the dendritic arbor. a)Percentageofpeakcalciumtransientblockineach5-20µmregionofdendriteplottedversesdistancefromthesoma.Blockwasvariable,butrelativelyuniformalongthedendrites from50to200µmfromawayfromthesoma(first50µmvs.last50µm,p=.78;Wilcoxonranksumtest).b)Percentagepeak calcium block is no different in oblique dendrites, vs. the main apical dendrite (p = .943;Wilcoxonranksumtest).c)Percentagepeakcalciumblockatabranchpointisnodifferentthanlocalblocknotatabranchpoint(p=.61).

33

Usingactionpotentialback-propagationasameasureofbranch-pointcalciumactivity,wefoundnodifferentialdistributionofcalciumcurrentatbranch-pointsvs.notatbranch-points(Figure3.12c).However,signalspropagatingintheforwarddirectionhaveagreaterpropensity to failsolelydueto the impedancemismatchofcurrent jumping fromasmallbranch to a larger more proximal dendrite (Stuart & Häusser 2007). Branch-pointmorphologyhasalsobeenshowntoaffectcouplingbetweenactivityinabranchandthatinthe soma (Ferrante et al. 2013), indicating that branch-points may be computationallyimportant. It ispossible thatvoltage-gatedchannelsatbranch-pointsalsohaveaspecificeffectonelectricalcouplingbetweenbranchandsoma.Totestthis,weaskedwhetherlocallyblockingVGICsatbranch-pointsaltersEPSP-spikecoupling,ameasureofsignalpropagationefficacy fromthedendrites to thesoma.Tocarryout thisexperiment,weusedadouble-barrelstimulatingpipettesothatcurrentflowoutofthepipettewaslocal,andthenplacedthebarrelopening5-20µmfromavisually identifiedbranch inorder tostimulateaxonsimpingingonaspecificbranch(Yasudaetal.2004)(Figure3.13a).WethenprovokedanEPSP,andslowlyincreasedthestimulationamplitudeuntilactionpotentialswerereliablyevoked.Inthreecells,wefoundnoevidencethatVGICblockinthe15µmregionaroundthebranch-pointofthestimulateddendritehadaneffectonEPSP-spikecoupling(Figure3.13b),indicatingthatVGICsinsuchasmallregionatabranch-pointdonotcarrymuchfunctionalrelevancetobranch-somacoupling,specifically.

Figure 3.13: Branch-point VGIC block has no effect on EPSP-induced spike probability. a)Experimentalset-up.AfteracellwasfilledwithAlexa594forvisualization,adouble-barrelpipettewasplacednearanidentifieddendrite.Background390nmlightwasusedtogloballyunblockVGICs,and a 488 nm laser was scanned to block VGICs in about a 15 µm region at a branch-pointcorrespondingtothestimulatedbranch.EPSPsofincreasingamplitudewereevoked,andionchannelblockconditionswerealternatedduring thecourseofanexperiment.b)Anempiricalcumulativedistribution function forspikeprobabilityvs. increasing initialEPSPslopes(calculatedwithinthefirst 2ms of the initial EPSP upward deflection). Upper and lower 95% confidence intervals areindicatedbythinlines.TherewasnostatisticallysignificantdifferencebetweenVGICblock(green)vs.VGICunblock(purple;p>.05;Kolmogorov-Smirnovtest).

34

3.3.3DendriticVGICsdeterminestrengthofactionpotentialback-propagation Previously,twodiscretepopulationsofCA1pyramidalcellswereidentifiedthatexhibiteitherstrongorweakdendriticback-propagation(Goldingetal.2001).Modelingsuggeststhat the relativedistributionofvoltage-gated ionchannelsalong thedendritic treecouldaccount for differences in excitability (Golding et al. 2001; Jarsky et al. 2005;Bernard&Johnston 2003). Here, using dendrite-specific illumination as in figure 3.11, we provideexperimentalevidencethatvoltage-gatedionchannelsdoindeeddeterminethedegreeofback-propagation. Becausewe found that direct block of local calcium in any one area of dendrite isindependentofdendriticdistancefromthesoma(Fig3.12a),weusedtwo-photoncalciumimagingtoexaminethedistance-dependenteffectsofactionpotentialback-propagation.Weelicitedback-propagatingactionpotentialsasinthelocalblockexperiments,andalternatedwhetherdendriticvoltage-gatedionchannelswereblockedwith540nmillumination,ornotblockedwith390nmillumination.In10cells,weimagedcalciuminatleasteightROIsalongthedendritictree,withthemostproximal(“first”)andmostdistal(“last”)ROIsseparatedbyatleast100µm,andidentifiedtwodiscretepopulationsofactionpotentialback-propagationstrength(Figure3.14a).Byexaminingcalciumtransientsalongthedendriteinun-blockedcis-QAQconditions,welabeledcellswithpropagationefficacies(first10µm/last10µm)ofmorethan90%as“strong”propagators(Figure3.14b),andcellswithpropagationefficacieslessthan60%as“weak”propagators(Figure3.14c).

Figure3.14:Cellsexhibitedeitherweakorstrongback-propagation.a)Cellswereseparatedinto“weak”or “strong”propagatorsbasedon theirpropagationefficacy inunblocked conditions (cis-QAQ),asmeasuredfromthemostproximalROItothemostdistalROI(Caprox/Cadist).Cellswithlessthan60%propagationefficacywereclassifiedasweak,andcellswithmorethan90%classificationefficacywereclassifiedasstrong.b)Exampleofastronglyback-propagatingneuron.c)Exampleofaweaklyback-propagatingneuron.GreentracesrepresentVGICblockwithdendrite-specific540nmillumination,andpurpletracesrepresentVGICunblockwith390nmdendriticillumination.Inbothb)andc),calciumtracesareequalinscale,andnumbereddotsindicatethemeasurementlocationofeachcorrespondingcalciumtransient.

35

Tocomparethetwopopulations,wepooledalldatafromstrongcellsandalldatafromweakcellsintotwogroups,thenbinnedthedatain40µmincrementsforanalysis.Theweakpropagatinggroupshowedasignificantdecayto36%(±8%;p=.0029)oftheinitialcalciumlevelby the farthestdistancemeasured,whereas thestronggroupshowednosignificantdecayfromtheinitialcalciumleveloveralldistancesmeasured(p=.2;Wilcoxonranksumtest;Figure3.15ai).Therelativecalciumbetweenthetwogroupswasalreadysignificantlydifferent40µmawayfromtheinitialbin(weak=77%±5%;strong=99%±6%;p=.011;Wilcoxonranksumtest;Figure3.15ai).Whenwethenblockeddendriticvoltage-gatedionchannels,therewasnolongerasignificantdifferenceinrelativecalciumdecaybetweenthetwogroups,exceptforabarelysignificantdifferencestillevidentatthefarthestregionfromthesoma(weak=64%±14%,strong=120%±24%;p=.046,Wilcoxonranksumtest;Figure3.15aii).

Interestingly,whenwelookedattheabsolutecalciumlevelsbetweenthetwogroupsin unblocked, 390 nm conditions, the weak propagators actually showed a significantlyhigherinitialcalciumlevel(dG/R=.055±.006)thanthestrongpropagatinggroup(dG/R=.023±002;p=.00077).However,thisleveldecayedrapidlytobecomeevenwiththestronggroupatdistancesgreaterthan150µmfromthesoma(Figure3.15b).Theabsolutecalciumlevelinthestrongpropagatinggroupremainedmoderateandconsistentalongthelengthofdendritemeasured.

When we looked at the effects of dendritic voltage-gated ion channel block onabsolutecalciumsignals,wefoundstrikinglydifferenteffectsoneachgroup.Atdistancescloserthan100µmtothesoma,theweakgroup(Figure3.15ci)displayedagreateramountofsignalblock(35.6%±2.9%)thanthestronggroup(Figure3.15cii,d;16.1%±3.6%;p=.00038).However,withincreasingdistancefromthesoma,signalblockintheweakgroupdecreased sharply to just 0.8% (± 4.2%) beyond 190 µm (p = 8.68e-07 compared withproximalsignal),butblockinthestronggroupactuallyincreasedslightlyto27.4%(±6.3%)beyond190µmfromthesoma.Thoughthiswasnotsignificantlydifferentthantheproximalblockinthestronggroup(p=.18),itwassignificantlygreaterthantheamountofblockpast190µmintheweakgroup(p=.002).Thisindicatesthatvoltage-gatedionchannelshadlittleeffect on back-propagation farther from the soma in the dendrites of weak cells, butsignificantlyboostedsignalsinfartherregionsofstrongcells.

36

Figure3.15:DendriticVGICsdetermineback-propagationstrength.Alldatafromstrongcells,andalldatafromweakcellswasbinnedin40µmincrementstocomparetwogroups.a)Relativetothefirstbin,normalpropagation(unblockedVGICs)intheweakgroupsignificantlydeviatesfromthestronggroupinallbinsafterthefirst(p=0.91,0.011,0.0032,0.00026,.000046,0.0025,withrespecttodendriticdistance).WithVGICblocked,however, there isnosignificantdifferencebetweenthetwogroups,exceptforamildeffectstillapparentatthemostdistalbin(p=0.97,0.087,0.30,0.68,0.10,0.046,respectively).b)Intheweakgroup,absolutecalciumtransientsweresignificantlylargeratproximalregionsthaninthestronggroup,butnodifferentinmoredistalregions(p=0.00077,.0000043, 0.035, 0.71, 0.32, 0.14, respectively). c) Calcium transients in the weak group weresignificantlyreducedbyblockingVGICsinonlymoreproximalregions(p=0.0020,5.39e-07,0.0014,0.041,0.38,0.55, respectively),whereascalciumtransients in thestronggroupweresignificantlyreducedwithVGICblockinbothproximalanddistalregions(p=0.0098,0.0085,0.064,0.023,0.020,0.13,respectively).d)ComparingtheeffectofVGICblockontheamountofsignalreductionacrossgroups,weseethatinproximalregionsVGICblockintheweakgroupcausedagreaterlevelofblockthaninthestronggroup,butatmoredistalregions,signalblockinthestronggroupwasgreaterthanin theweakgroup (p=0.011, 0.0024, 0.52, 0.076, 0.0062, 0.038, respectively). Forwithin groupcomparisons, p values were calculated using theWilcoxon signed rank test. For strong v. weakcomparisons,pvalueswerecalculatedwiththeWilcoxonranksumtest.

37

Modelingwork has shown that differences inmorphology can substantially affectback-propagation characteristics (Vetter et al. 2001). For this reason, we analyzed themorphologyofcellsinbothstrongandweakgroups.Ashollanalysisofdendriticcomplexity(Sholl1953)revealednodifferenceinthedendriticbranchingpatternbetweenstrongandweakpropagatingcells(Fig3.16a).TherewasalsonodifferenceinthenumberofprimarybranchesoffofthemainapicaldendritebetweenthefirstandlastimagingROIforeachcell(Fig3.16b).Theseresultsindicatethatdendriticbranchingpatternswerenotresponsibleforthedifferencesinback-propagationwesaw,andfurtherimplicateabalanceofdendriticVGICpopulationsineachcellastheunderlingfactorinback-propagationdifferences.

Figure3.16:Dendriticmorphologydoesnotdifferbetween strongandweakgroups. a) Shollanalysisofdendriticcomplexity.Thedendriticbranchingpatternbetweenweak(red)andstrong(blue) neurons is not significantly different (p = .72; Two-sample Kolmogorov-Smirnov test). b)NumbersofprimarydendriticbranchesbetweenthemostproximalanddistalROIsineachcelldidnotdifferbetweenstrongandweakgroups(p=.97;Wilxoconranksumtest).

38

4. Discussion

4.1. Conclusions&newhypotheses HereweoffernewexperimentalevidencethatdendriticVGICsunderliethedistinctionbetweenweakandstrongactionpotentialback-propagation,anotionthathaspreviouslyonly been simulated (Golding et al. 2001; Bernard& Johnston 2003).We identified twopopulations of CA1 pyramidal cells that display eitherweak or strong back-propagationcharacteristics. Weakly propagating cells exhibited a large calcium signal in the apicaldendritesproximaltothesoma,butadramaticsignalfall-offwithdendriticdistance,andstronglypropagating cells exhibitedamoderate calciumresponse initially that remainedstable along the apical dendrites. When we blocked VGICs, weak and strong cells weresuddenly indistinguishable.Usingpercent calciumblock in this situationasameasureofabsoluteexcitability,we found thatweakandstrongpopulationsdisplayed twodifferentexcitability profiles. The weak propagators contained a steep excitability gradient thatdecreasedwithdistance,whereasstrongpropagatorshadamildexcitabilitygradientthatincreasedwithdendriticdistance. Back-propagatingactionpotentialsare thought toserveasasignalof recentneuraloutput to dendrites and synapses. Back-propagation can both strengthen or weakensynapsesbasedonrecentsynapticinput(Markrametal.1997;Magee&Johnston1997),andcanincreasethelocalintrinsicexcitabilityofarecently-activateddendriticregion(Fricketal. 2004). It has alsobeendiscovered that hippocampal neurons aredifferentially-tuned,suchthatsomerespondbroadlytoavarietyofinputpatterns,whereasothersrespondonlyoccasionallytoveryspecificinputpatterns(Petersen&Crochet2013).Wehypothesizethatthestarkcontrast inexcitabilityprofilesofweakandstrongneuronsmightdifferentiallyaffecteachpopulation’sresponsepropertiesandinputcodingrules.Itispossiblethatthemildly increasing excitability gradient in the strong population, which produces a“normalized”calciumlevelalongthedendritictree,servestopotentiatenear-bysynapsestosimilardegreesduringactivity, leading to lessdendritic compartmentalization,andmoregeneralizedresponsestovariedinputpatterns—abroadly-tunedcell.Contrarytothis,thesteepexcitability gradient inweakcellsproducesa largedifference in the calcium levelsalongthedendrites,whichwouldhaveverydifferentplasticeffectsoneachactivesynapse.Wehypothesizethatthismaypromotemorecompartmentalizeddendrites,andacellthatlearnstorespondtoonlyspecificpatternsofinput—asharply-tunedneuron.Voltage-gatedion channel control of weak and strong back-propagation may alternatively be aconsequenceofthespecifictuningpropertiesofacell,orgeneticallypre-determined.Futurestudiescouldaddressthesequestions. In a separate experimentwe found that dendritic VGICs as awhole suppress EPSPsummation, which supports previous studies indicating potassium channels are thedominantplayersinEPSPsummation.HCNchannelshavealsobeenheavilyimplicatedinCA1pyramidalcelltemporalsummation,however,QAQhasbeenshowntonotaffectHCNchannels (Mourotetal.2012). It’sbeen theoreticallyproposed thatHCNchannelactivityaloneistoosimpleanexplanationtoexplainthetemporalsummationregulation(Desjardinset al. 2003), so it’s possible that both HCN and voltage-gated potassium channels haveindependentbutsimilareffectsonsummation,andindeed,thereisevidenceforthis idea(Georgeetal.2009).Additionally,oneshouldkeepinmindthatfulldose-responseprofiles

39

forallchannelsthatQAQactsonhavenotbeenmeasured,butQAQhassofarbeenshowntoaffectthesamechannelsasthedrugfromwhichitwasderived,QX-314(Mourotetal.2012),andathighconcentrations,QX-314hasbeenshowntoblockCA1pyramidalcellsagcurrentduringahyperpolarizingcurrentstep(Perkins&Wong1995),soitispossiblethatQAQdoesactually have a minor affect on HCN current in CA1 pyramidal cells, though previousexperimentsindissociatedtrigeminalneuronsdidnotfindanyeffectofQAQonHCNcurrentdirectly. WhenwespecificallytargetedilluminationlighttotheapicaldendritestoblockVGICsthere,wefoundnoeffectonthesomaticstepresponseinmeasuresofeitheractionpotentialfiringorcharge transfer.This isan important findingbecausesmall changes indendriticpropertieshavebeenshowntoaffectactionpotentialfiring(Magee&Carruth1999;Eyaletal.2014).Thisexperimentalso informsourback-propagatingactionpotentialsresults, inthatwecanbeconfidentthatthedifferenceinpropagationweseewithdendriticVGICblockisnotduetoaneffectoninitialactionpotentialfiring. Wealso found that localblockofback-propagatingcalciumactivity is confined toatargetarea,andtheamountofblockisnotdependentondistancefromthesoma.Thiswasalso an important finding to support our back-propagating results, becausewe could beconfidentthatthepropagationeffectswemeasuredwithcalciumimagingwerenotduetodistance-dependentchangesinlocalizedcalciumblockateachimagedregion.Ouranalysisoflocalcalciumblocksupportspreviousresearch(Korkotian&Segal2006;Biessetal.2011)that calcium diffusion from a source (i.e. voltage-gated calcium channels) is extremelyrestricted to within a fewmicrons, and demonstrates that diffusional blurring of QAQ’sactionboundariesisnegligibleoverthecourseofjustthreeseconds. Interestingly,wefoundnodifferenceintheamountofcalciumsignalblockedinobliquebranchescomparedwiththemainapicaltrunk.Previousresearchhasshownthatobliquedendritesmaybemoreexcitablethanthemaindendrite(Losonczy&Magee2006;Miglioreetal.2005;Fricketal.2003),butwefoundnodifferencesintheleveloflocalcalciumblockbetween these regions. Any excitability differences may then arise primarily fromdifferentialvoltage-gatedsodiumandpotassiumchanneldistributions.UsingQAQblockofactivityasameasureofvoltage-gatedexcitability,itwouldbeinterestinginfuturestudiestoexaminehow levels of branch excitability correlatewith input tuningproperties orwithEPSP-spikecoupling. Inthisstudy,wepresentthefirstexperimentalevidencethatVIGCsunderliecell-to-cell differences in action potential back-propagation. Using amethod that ismore easilytractable thandendriticpatch-clamp recordings andamenable to thindendrites,wealsoconfirm previous electrophysiological and immunolabeling evidence that the generaldistribution of voltage-gated calcium channels is relatively uniform along the apicaldendritictree.Wealsoaddthatvoltage-gatedcalciumactivityisnodifferentbetweenmainandobliquedendrites.Byblockingallvoltage-gated ionchannels toessentiallymake thedendritic tree passive, we present definitive evidence that schaffer collateral EPSPsummationissuppressedbytheVGICsinCA1hippocampalpyramidalcelldendrites.Thissuggeststhat,indeed,potassiumchannelsaredominantplayersinsub-thresholdtemporalsummation,becauseifsodiumandcalciumchannelsplayedadominantrole,thenblockingallVGICswouldyieldanoppositeeffectonsummation.

40

3.2. ConsiderationsforusingQAQQAQ can potentially address a wealth of open questions about voltage-gated

informationprocessing indendrites,but thereare,ofcourse,drawbacks tousing light tomanipulateneuralactivity.Onemaincomplicationisthattoomuchlightcandamagetissue(Niemz2004).BraintissueabsorbsUVlightparticularlywell(Yaroslavskyetal.2002),socareshouldalwaysbetakentomeasurelightoutputandavoidinundatingacellwithmorepowerthanissafe.QAQisreadilyphotoisomerizablewithshort300mspulsesof lightatpowerdensitiesbetween1-10mW/mm2,whichiswellbelowthesafeupperlimitforbraintissueof75mW/mm2withshort(<500ms)pulses(Cardinetal.2010).

ItshouldalsobenotedthatQAQisause-dependentionchannelblocker,meaningthatit’sactionwillgetstrongerwithrepeatedactivity.Ithasbeennotedthatatconcentrationshigherthan200µM,QAQwillblockopenionchannelsfasterandmayeliminateeventhefirstaction potential during a step response (Fedorchak et al, in preparation), but higherconcentrationsthan200µMdonotallowforcompleterecoveryofactionpotentialfiringincis-QAQ illumination conditions. The lack of return to a normal firing pattern at highconcentrations isattributable to the fact thatQAQphotoisomerization isonlyabout90%efficient(Fortinetal.2008).Thismeansthatevenwhenweilluminatewithnear-uvlighttoconvert QAQ to its inert cis form, there is still about 10% trans-QAQ present. At lowerconcentrations,thisproportionofresidual-transisnotenoughtoblockchannelsanddisruptthecell’snormalfiringpattern,butathigherconcentrations,theresidual-transinnear-uvlightisenoughtokeepionchannelspartiallyblocked.HigherconcentrationsofQAQmay,however, equilibrate faster in the dendrites and allow dark relaxation kinetics to reachsteady-statefaster,butthisremainstobetested. QAQisnon-specificforallvoltage-gatedionchannelstestedsofar,buttherearemanyothersthathavenotbeentested.Forthetooltobeasusefulaspossible,itwillbecriticaltobothmeasurefulldose-responsecurvesfortheionchannelsalreadytested,aswellasotherchannels. It’s possible that at certain concentrations, QAQ may be able to tease outcontributionsfromspecificionchannels.

ResearchersmaybetemptedtocombineQAQwithotheropticalmethodstocontrolneuralactivity,likeglutamateuncagingoroptogeneticactivation.Itiscertainlypossibletomultiplex optical methods, as we have multiplexed calcium imaging and voltage-gatedphoto-control in this thesis.However, careful attention should be paid to any overlap inspectralsensitivitiesforeachtool.ThevisiblelightwavelengthsusedtophotoisomerizeQAQarealsoknowntouncageneurotransmitters,andtoaffectthechannelconductingstateofpopular opsins (Bamann et al. 2008). Red-shifted versions of QAQ could mitigate thesepotential barriers in the future, but careful experimental designwill be critical for suchstudies.Toeffectivelycombinemethodswithoverlappingspectra,forinstanceinthecaseofmicrobialopsins,weproposethatonecouldtunethelightintensityusedforopsinactiviationin order to offset the change in opsin conductance after 390 or 500 nm illumination.Alternatively,ifanexperimentallows,lightmaybetargetedtodifferentspatialareastoavoidcross-talkcompletely.3.3. Futuredirections Anopticalrevolutionhasopenedupnewexperimentalavenuesinneuroscience,andwedemonstrateinthisthesisthatthespecificphotoswitch,QAQ,isauniqueandusefultoolfor previously unattainable studies on dendritic excitability.We anticipate thatwith the

41

technical knowledge presented here, one could preferentially control voltage-gatedexcitabilityinsingledendriticbranches,orevenspines,andstudy—forthefirsttime—thefunctionalimplicationsoflocalVGICpopulationsinadendritictree. WealsodemonstrateinthisthesisthatEPSP-spikecouplingisreducedwithwhole-cellVGIC block, but notwith block targeted to a specific branch-point. One open question iswhetherVGICblockinawholedendriticbranchwillaffectEPSP-spikecoupling,orwhetherblockclosertothesomamighthavealargereffect.DoesbranchexcitabilitydriveEPSP-spikecoupling? Similarly, does branch excitability determine temporal integration, or is thiscontrolledalongtheentiredendritictreeasEPSPspropagatetothesoma?It’sbeenshownthat clustered synaptic inputs on a single dendritic branch can evoke an NMDA-baseddendriticspike,butthatdistributedinputssummatelinearlyatthesoma(Yangetal.2014).Will clustered inputsdriveNMDAspikeswithoutvoltage-gated ionchannelactivity?Willpassivedendritesyieldsub-linearsummationofdistributedinputs? We found in this thesis thatQAQ’s action canbe extremely spatially restricted in adendrite. It will be interesting to examine what effect, if any, VGICs have in the tiniestdendriticregions:spines.VGICsarefoundinspines(Trimmer&Rhodes2004;Bloodgood&Sabatini 2007; Tippens et al. 2008), but their purpose there is entirely unclear. Becausespinesaresuchelectricallycompactcompartments,evenasmallnumberofvoltage-gatedion channels could powerfully affect membrane potential. By focally blocking (orunblocking) spine VGICs, one could ask whether they control bAP invasion into spines(Palmer&Stuart2009),whethertheymodulatesub-thresholdEPSPsorcalciumdynamics,orwhetherVGICsaffectchangesinspinemorphologyatall. QAQisnon-specificforavarietyofvoltage-gatedionchannels,soitwillbebestputtouseforquestionsconcerninggeneralvoltage-gatedexcitabilitygovernedbyallthechannelsinmembrane.However,ifonewantedtouseitsopticalspecificitytoteaseoutthespecificcontributionsofcertainchannels,QAQcouldbecombinedwithtraditionalpharmacologicaltools. Finally,QAQdoesnotcrosslipidmembranes,butifonewantedtouseQAQformoresystems-levelquestionsaboutbranch-excitability,previousresearchhasshownthatitcanenter cells through pore-dilating TRPV1 or P2X channels. We foresee that it would bepossibletoexogenouslyexpressTRPV1inaspecificpopulation,thenapplyQAQwithapore-dilatingagonistlikecapsasintoloaditintoawholepopulation.Dependingoncell-type,thereshouldbelittle-to-nooccurrencesofoff-targetloadingintocellswithendogenousTRPV1,asthesechannelsarerareinthecentralnervoussystem(Cavanaughetal.2011;Arenkieletal.2008),andcouldevenbeknockedoutfirstifdesired(Birderetal.2002).

Dendriticspikesintheapicaltuftofpyramidalneuronshavebeenshowntocorrelatewith behavior, though studies also confirm that activity in the tuft does not propagatereliably to the soma in-vivo. With such a TRPV1/capsasin/QAQ system, one maydifferentially“turnonandoff”voltage-gatedexcitabilityinatuft,anddirectlyaskwhateffectthismanipulationhasonbehavior.Recently,dendriticcalciumtransientshavebeenshowntoshapetheemergentplacefieldpropertiesinCA1pyramidalcells(Sheffield&Dombeck2015).Onecoulddirectlytesttheeffectthatdampeningthesecalciumtransients,aswehaveshownhere,hasonplacefieldproperties.

42

ReferencesAndersen,P.,1960.Interhippocampalimpulses.II.ApicaldendriticactivationofCAI

neurons.ActaphysiologicaScandinavica,48,pp.178–208.Arenkiel,B.R.etal.,2008.Geneticcontrolofneuronalactivityinmiceconditionally

expressingTRPV1.NatMethods,5(4),pp.299–302.Ariav,G.,Polsky,A.&Schiller,J.,2003.Submillisecondprecisionoftheinput-output

transformationfunctionmediatedbyfastsodiumdendriticspikesinbasaldendritesofCA1pyramidalneurons.TheJournalofneuroscience :theofficialjournaloftheSocietyforNeuroscience,23(21),pp.7750–7758.

Azouz,R.,Jensen,M.S.&Yaari,Y.,1996.Ionicbasisofspikeafter-depolarizationandburstgenerationinadultrathippocampalCA1pyramidalcells.TheJournalofphysiology,492(1),pp.211–23.

Bamann,C.etal.,2008.Spectralcharacteristicsofthephotocycleofchannelrhodopsin-2anditsimplicationforchannelfunction.Journalofmolecularbiology,375(3),pp.686–94.

Banghart,M.etal.,2009.PhotochromicBlockersofVoltage-GatedPotassiumChannels.AngewandteChemieInternationalEdition,48(48),pp.9097–9101.

Beharry,A.A.&Woolley,G.A.,2011.Azobenzenephotoswitchesforbiomolecules.ChemicalSocietyreviews,40(8),pp.4422–37.

Bernard,C.&Johnston,D.,2003.Distance-dependentmodifiablethresholdforactionpotentialback-propagationinhippocampaldendrites.Journalofneurophysiology,90(3),pp.1807–1816.

Biess,A.,Korkotian,E.&Holcman,D.,2011.Barrierstodiffusionindendritesandestimationofcalciumspreadfollowingsynapticinputs.PLoScomputationalbiology,7(10),p.e1002182.

Birder,L.A.etal.,2002.AlteredurinarybladderfunctioninmicelackingthevanilloidreceptorTRPV1.Natureneuroscience,5(9),pp.856–60.

Bittner,K.C.,Andrasfalvy,B.K.&Magee,J.C.,2012.IonchannelgradientsintheapicaltuftregionofCA1pyramidalneurons.PloSone,7(10),p.e46652.

Bloodgood,B.L.&Sabatini,B.L.,2007.NonlinearregulationofunitarysynapticsignalsbyCaV(2.3)voltage-sensitivecalciumchannelslocatedindendriticspines.Neuron,53(2),pp.249–60.

Boyden,E.S.etal.,2005.Millisecond-timescale,geneticallytargetedopticalcontrolofneuralactivity.Natureneuroscience,8(9),pp.1263–8.

Branco,T.,Clark,B.A.&Häusser,M.,2010.Dendriticdiscriminationoftemporalinputsequencesincorticalneurons.Science(NewYork,N.Y.),329(5999),pp.1671–5.

Branco,T.&Häusser,M.,2010.Thesingledendriticbranchasafundamentalfunctionalunitinthenervoussystem.Currentopinioninneurobiology,20(4),pp.494–502.

Brunel,N.,Hakim,V.&Richardson,M.J.E.,2014.Singleneurondynamicsandcomputation.CurrentOpinioninNeurobiology,25,pp.149–155.

Cardin,J.A.etal.,2010.Targetedoptogeneticstimulationandrecordingofneuronsinvivousingcell-type-specificexpressionofChannelrhodopsin-2.NatureProtocols,5(2),pp.247–254.

Cash,S.&Yuste,R.,1998.Inputsummationbyculturedpyramidalneuronsislinearand

43

position-independent.TheJournalofneuroscience :theofficialjournaloftheSocietyforNeuroscience,18(1),pp.10–15.

Cash,S.&Yuste,R.,1999.LinearsummationofexcitatoryinputsbyCA1pyramidalneurons.Neuron,22(2),pp.383–394.

Cavanaugh,D.J.etal.,2011.Trpv1reportermicerevealhighlyrestrictedbraindistributionandfunctionalexpressioninarteriolarsmoothmusclecells.TheJournalofneuroscience :theofficialjournaloftheSocietyforNeuroscience,31(13),pp.5067–77.

Chen,B.L.,Hall,D.H.&Chklovskii,D.B.,2006.Wiringoptimizationcanrelateneuronalstructureandfunction.ProceedingsoftheNationalAcademyofSciencesoftheUnitedStatesofAmerica,103(12),pp.4723–8.

Chow,B.Y.etal.,2010.High-performancegeneticallytargetableopticalneuralsilencingbylight-drivenprotonpumps.Nature,463(7277),pp.98–102.

Daoudal,G.,Hanada,Y.&Debanne,D.,2002.Bidirectionalplasticityofexcitatorypostsynapticpotential(EPSP)-spikecouplinginCA1hippocampalpyramidalneurons.ProceedingsoftheNationalAcademyofSciencesoftheUnitedStatesofAmerica,99(22),pp.14512–14517.

Deisseroth,K.,2015.Optogenetics:10yearsofmicrobialopsinsinneuroscience.Natureneuroscience,18(9),pp.1213–1225.

Desjardins,A.E.etal.,2003.TheinfluencesofIhontemporalsummationinhippocampalCA1pyramidalneurons:amodelingstudy.Journalofcomputationalneuroscience,15(2),pp.131–42.

Euler,T.etal.,2009.Eyecupscope--opticalrecordingsoflightstimulus-evokedfluorescencesignalsintheretina.PflügersArchiv :Europeanjournalofphysiology,457(6),pp.1393–414.

Eyal,G.etal.,2014.Dendritesimpacttheencodingcapabilitiesoftheaxon.TheJournalofneuroscience :theofficialjournaloftheSocietyforNeuroscience,34(24),pp.8063–71.

Ferrante,M.,Migliore,M.&Ascoli,G.A.,2013.Functionalimpactofdendriticbranch-pointmorphology.TheJournalofneuroscience :theofficialjournaloftheSocietyforNeuroscience,33(5),pp.2156–65.

Fisher,R.E.,Gray,R.&Johnston,D.,1990.Propertiesanddistributionofsinglevoltage-gatedcalciumchannelsinadulthippocampalneurons.Journalofneurophysiology,64(1),pp.91–104.

Fitzpatrick,J.S.etal.,2009.Inositol-1,4,5-trisphosphatereceptor-mediatedCa2+wavesinpyramidalneurondendritespropagatethroughhotspotsandcoldspots.TheJournalofphysiology,587(Pt7),pp.1439–59.

Fortin,D.L.etal.,2011.OptogeneticphotochemicalcontrolofdesignerK+channelsinmammalianneurons.Journalofneurophysiology,106(1),pp.488–496.

Fortin,D.L.etal.,2008.Photochemicalcontrolofendogenousionchannelsandcellularexcitability.Naturemethods,5(4),pp.331–338.

Frick,A.etal.,2003.NormalizationofCa2+signalsbysmallobliquedendritesofCA1pyramidalneurons.TheJournalofneuroscience :theofficialjournaloftheSocietyforNeuroscience,23(8),pp.3243–3250.

Frick,A.&Johnston,D.,2005.Plasticityofdendriticexcitability.JournalofNeurobiology,64(1),pp.100–115.

Frick,A.,Magee,J.&Johnston,D.,2004.LTPisaccompaniedbyanenhancedlocalexcitabilityofpyramidalneurondendrites.Natureneuroscience,7(2),pp.126–135.

44

Fricker,D.&Miles,R.,2000.EPSPamplificationandtheprecisionofspiketiminginhippocampalneurons.Neuron,28(2),pp.559–569.

Fujita,Y.&Sakata,H.,1962.ElectrophysiologicalpropertiesofCA1andCA2apicaldendritesofrabbithippocampus.Journalofneurophysiology,25,pp.209–22.

Gasparini,S.&Magee,J.C.,2006.State-dependentdendriticcomputationinhippocampalCA1pyramidalneurons.TheJournalofneuroscience :theofficialjournaloftheSocietyforNeuroscience,26(7),pp.2088–2100.

George,M.S.,Abbott,L.F.&Siegelbaum,S.A.,2009.HCNhyperpolarization-activatedcationchannelsinhibitEPSPsbyinteractionswithM-typeK(+)channels.Natureneuroscience,12(5),pp.577–84.

Gillessen,T.&Alzheimer,C.,1997.AmplificationofEPSPsbylowNi(2+)-andamiloride-sensitiveCa2+channelsinapicaldendritesofratCA1pyramidalneurons.Journalofneurophysiology,77(3),pp.1639–1643.

Goldberg,J.H.&Yuste,R.,2005.Spacematters:localandglobaldendriticCa2+compartmentalizationincorticalinterneurons.Trendsinneurosciences,28(3),pp.158–67.

Golding,N.L.,Kath,W.L.&Spruston,N.,2001.Dichotomyofaction-potentialbackpropagationinCA1pyramidalneurondendrites.Journalofneurophysiology,86(6),pp.2998–3010.

Häusser,M.&Mel,B.,2003.Dendrites:bugorfeature?Currentopinioninneurobiology,13(3),pp.372–83.

Helmchen,F.,Imoto,K.&Sakmann,B.,1996.Ca2+bufferingandactionpotential-evokedCa2+signalingindendritesofpyramidalneurons.Biophysicaljournal,70(2),pp.1069–81.

Hille,B.,2001.IonChannelsofExcitableMembranes.Hoffman,D.A.etal.,1997.K+channelregulationofsignalpropagationindendritesof

hippocampalpyramidalneurons.Nature,387(6636),pp.869–875.Jarsky,T.etal.,2005.Conditionaldendriticspikepropagationfollowingdistalsynaptic

activationofhippocampalCA1pyramidalneurons.Natureneuroscience,8(12),pp.1667–76.

Jensen,M.S.,Azouz,R.&Yaari,Y.,1994.Variantfiringpatternsinrathippocampalpyramidalcellsmodulatedbyextracellularpotassium.Journalofneurophysiology,71(3),pp.831–9.

Jester,J.M.,Campbell,L.W.&Sejnowski,T.J.,1995.AssociativeEPSP--spikepotentiationinducedbypairingorthodromicandantidromicstimulationinrathippocampalslices.TheJournalofphysiology,484(3),pp.689–705.

Jiang,X.etal.,2015.Principlesofconnectivityamongmorphologicallydefinedcelltypesinadultneocortex.Science,350(6264),pp.9462–9462.

Kamondi,A.,Acsády,L.&Buzsáki,G.,1998.Dendriticspikesareenhancedbycooperativenetworkactivityintheintacthippocampus.TheJournalofneuroscience :theofficialjournaloftheSocietyforNeuroscience,18(10),pp.3919–3928.

Kanemoto,Y.etal.,2011.SpatialdistributionsofGABAreceptorsandlocalinhibitionofCa2+transientsstudiedwithGABAuncaginginthedendritesofCA1pyramidalneurons.PloSone,6(7),p.e22652.

Klapoetke,N.C.etal.,2014.Independentopticalexcitationofdistinctneuralpopulations.Naturemethods,11(3),pp.338–46.

45

Korkotian,E.&Segal,M.,2006.Spatiallyconfineddiffusionofcalciumindendritesofhippocampalneuronsrevealedbyflashphotolysisofcagedcalcium.CellCalcium,40(5-6),pp.441–449.

Lai,H.C.&Jan,L.Y.,2006.Thedistributionandtargetingofneuronalvoltage-gatedionchannels.Naturereviews.Neuroscience,7(7),pp.548–62.

Lin,J.Y.etal.,2013.ReaChR:ared-shiftedvariantofchannelrhodopsinenablesdeeptranscranialoptogeneticexcitation.Natureneuroscience,16(10),pp.1499–508.

Lin,W.-C.etal.,2015.AComprehensiveOptogeneticPharmacologyToolkitforInVivoControlofGABAAReceptorsandSynapticInhibition.Neuron,88(5),pp.879–891.

Lin,W.-C.etal.,2014.Engineeringalight-regulatedGABAAreceptorforopticalcontrolofneuralinhibition.ACSchemicalbiology,9(7),pp.1414–9.

Lipowsky,R.,Gillessen,T.&Alzheimer,C.,1996.DendriticNa+channelsamplifyEPSPsinhippocampalCA1pyramidalcells.Journalofneurophysiology,76(4),pp.2181–2191.

Llinás,R.etal.,1968.DendriticspikesandtheirinhibitioninalligatorPurkinjecells.Science(NewYork,N.Y.),160(3832),pp.1132–5.

Llinas,R.&Nicholson,C.,1971.ElectrophysiologicalpropertiesofdendritesandsomatainalligatorPurkinjecells.Journalofneurophysiology,34(4),pp.532–51.

London,M.&Häusser,M.,2005.Dendriticcomputation.Annualreviewofneuroscience,28,pp.503–32.

LongairMH,BakerDA,ArmstrongJD.,2011.SimpleNeuriteTracer:opensourcesoftwareforreconstruction,visualizationandanalysisofneuronalprocesses.Bioinformatics,27,pp.2453–2454.

Losonczy,A.&Magee,J.C.,2006.IntegrativepropertiesofradialobliquedendritesinhippocampalCA1pyramidalneurons.Neuron,50(2),pp.291–307.

Magee,J.&Johnston,D.,1995.Characterizationofsinglevoltage-gatedNa+andCa2+channelsinapicaldendritesofratCA1pyramidalneurons.TheJournalofPhysiology,487(1),pp.67–90.

Magee,J.C.&Carruth,M.,1999.Dendriticvoltage-gatedionchannelsregulatetheactionpotentialfiringmodeofhippocampalCA1pyramidalneurons.Journalofneurophysiology,82(4),pp.1895–1901.

Magee,J.C.&Johnston,D.,1997.Asynapticallycontrolled,associativesignalforHebbianplasticityinhippocampalneurons.Science(NewYork,N.Y.),275(5297),pp.209–213.

Magee,J.C.&Johnston,D.,1995.Characterizationofsinglevoltage-gatedNa+andCa2+channelsinapicaldendritesofratCA1pyramidalneurons.TheJournalofphysiology,487(1),pp.67–90.

Margulis,M.&Tang,C.M.,1998.Temporalintegrationcanreadilyswitchbetweensublinearandsupralinearsummation.Journalofneurophysiology,79(5),pp.2809–2813.

Markram,H.etal.,2015.ReconstructionandSimulationofNeocorticalMicrocircuitry.Cell,163(2),pp.456–492.

Markram,H.etal.,1997.RegulationofsynapticefficacybycoincidenceofpostsynapticAPsandEPSPs.Science(NewYork,N.Y.),275(5297),pp.213–5.

Metz,A.E.etal.,2005.R-typecalciumchannelscontributetoafterdepolarizationandburstinginhippocampalCA1pyramidalneurons.TheJournalofneuroscience :theofficialjournaloftheSocietyforNeuroscience,25(24),pp.5763–73.

Meyer,E.,Müller,C.O.&Fromherz,P.,1997.Cablepropertiesofdendritesinhippocampalneuronsoftheratmappedbyavoltage-sensitivedye.TheEuropeanjournalof

46

neuroscience,9(4),pp.778–85.Migliore,M.,Ferrante,M.&Ascoli,G.A.,2005.Signalpropagationinobliquedendritesof

CA1pyramidalcells.Journalofneurophysiology,94(6),pp.4145–55.Migliore,M.&Shepherd,G.M.,2002.Emergingrulesforthedistributionsofactivedendritic

conductances.Naturereviews.Neuroscience,3(5),pp.362–370.Miyakawa,H.etal.,1992.SynapticallyactivatedincreasesinCa2+concentrationin

hippocampalCA1pyramidalcellsareprimarilyduetovoltage-gatedCa2+channels.Neuron,9(6),pp.1163–73.

Mourot,A.etal.,2012.Rapidopticalcontrolofnociceptionwithanion-channelphotoswitch.Naturemethods,9(4),pp.396–402.

Mourot,A.etal.,2011.Tuningphotochromicionchannelblockers.ACSchemicalneuroscience,2(9),pp.536–43.

Niemz,M.H.,2004.Laser-TissueInteractions.Palmer,L.M.etal.,2014.NMDAspikesenhanceactionpotentialgenerationduringsensory

input.Natureneuroscience,17(3),pp.383–90.Palmer,L.M.&Stuart,G.J.,2009.Membranepotentialchangesindendriticspinesduring

actionpotentialsandsynapticinput.TheJournalofneuroscience :theofficialjournaloftheSocietyforNeuroscience,29(21),pp.6897–903.

Perkins,K.L.&Wong,R.K.,1995.IntracellularQX-314blocksthehyperpolarization-activatedinwardcurrentIqinhippocampalCA1pyramidalcells.Journalofneurophysiology,73(2),pp.911–5.

Petersen,C.C.H.&Crochet,S.,2013.Synapticcomputationandsensoryprocessinginneocorticallayer2/3.Neuron,78(1),pp.28–48.

Pologruto,T.A.,Sabatini,B.L.&Svoboda,K.,2003.ScanImage:flexiblesoftwareforoperatinglaserscanningmicroscopes.Biomedicalengineeringonline,2,p.13.

Polosukhina,A.etal.,2012.PhotochemicalRestorationofVisualResponsesinBlindMice.Neuron,75(2),pp.271–282.

Polsky,A.,Mel,B.W.&Schiller,J.,2004.Computationalsubunitsinthindendritesofpyramidalcells.Natureneuroscience,7(6),pp.621–7.

Purpura,D.P.&Shofer,R.J.,1965.Spike-generationindendritesandsynapticinhibitioninimmaturecerebralcortex.Nature,206(4986),pp.833–4.

Rall,W.,1959.Branchingdendritictreesandmotoneuronmembraneresistivity.ExperimentalNeurology,1(5),pp.491–527.

Ramakers,G.M.J.&Storm,J.F.,2002.ApostsynaptictransientK(+)currentmodulatedbyarachidonicacidregulatessynapticintegrationandthresholdforLTPinductioninhippocampalpyramidalcells.ProceedingsoftheNationalAcademyofSciencesoftheUnitedStatesofAmerica,99(15),pp.10144–9.

Remy,S.,Beck,H.&Yaari,Y.,2010.Plasticityofvoltage-gatedionchannelsinpyramidalcelldendrites.CurrentOpinioninNeurobiology,20(4),pp.503–509.

Remy,S.,Csicsvari,J.&Beck,H.,2009.Activity-dependentcontrolofneuronaloutputbylocalandglobaldendriticspikeattenuation.Neuron,61(6),pp.906–16.

Sheffield,M.E.J.&Dombeck,D.A.,2015.Calciumtransientprevalenceacrossthedendriticarbourpredictsplacefieldproperties.Nature,517(7533),pp.200–204.

Shitaka,Y.etal.,1996.BasicFibroblastGrowthFactorIncreasesFunctionalL-TypeCa2+ChannelsinFetalRatHippocampalNeurons:ImplicationsforNeuriteMorphogenesisInVitro.J.Neurosci.,16(20),pp.6476–6489.

47

Sholl,D.A.,1953.Dendriticorganizationintheneuronsofthevisualandmotorcorticesofthecat.Journalofanatomy,87(4),pp.387–406.

Sjöström,P.J.etal.,2010.DendriticExcitabilityandSynapticPlasticity.PhysiologicalReviews,(November2009),pp.1–28.

Smith,S.L.etal.,2013.Dendriticspikesenhancestimulusselectivityincorticalneuronsinvivo.Nature,503(7474),pp.115–120.

Spencer,W.A.&Kandel,E.R.,1961.ELECTROPHYSIOLOGYOFHIPPOCAMPALNEURONS:IV.FASTPREPOTENTIALS.JNeurophysiol,24(3),pp.272–285.

Spruston,N.etal.,1995.Activity-dependentactionpotentialinvasionandcalciuminfluxintohippocampalCA1dendrites.Science(NewYork,N.Y.),268(5208),pp.297–300.

Staff,N.P.etal.,2000.RestingandactivepropertiesofpyramidalneuronsinsubiculumandCA1ofrathippocampus.Journalofneurophysiology,84(5),pp.2398–2408.

Stuart,G.&Häusser,M.,2007.Dendrites,OxfordUniversityPress.Stuart,G.J.,Dodt,H.U.&Sakmann,B.,1993.Patch-clamprecordingsfromthesomaand

dendritesofneuronsinbrainslicesusinginfraredvideomicroscopy.PflügersArchiv :Europeanjournalofphysiology,423(5-6),pp.511–8.

Stuart,G.J.&Häusser,M.,2001.DendriticcoincidencedetectionofEPSPsandactionpotentials.Natureneuroscience,4(1),pp.63–71.

Stuart,G.J.&Sakmann,B.,1994.Activepropagationofsomaticactionpotentialsintoneocorticalpyramidalcelldendrites.Nature,367(6458),pp.69–72.

Svoboda,K.etal.,1997.Invivodendriticcalciumdynamicsinneocorticalpyramidalneurons.Nature,385(6612),pp.161–5.

Szobota,S.etal.,2007.Remotecontrolofneuronalactivitywithalight-gatedglutamatereceptor.Neuron,54(4),pp.535–45.

Takagi,H.,2000.RolesofionchannelsinEPSPintegrationatneuronaldendrites.Neuroscienceresearch,37(3),pp.167–71.

Talbot,M.J.&Sayer,R.J.,1996.IntracellularQX-314inhibitscalciumcurrentsinhippocampalCA1pyramidalneurons.Journalofneurophysiology,76(3),pp.2120–4.

Tippens,A.L.etal.,2008.Ultrastructuralevidenceforpre-andpostsynapticlocalizationofCav1.2L-typeCa2+channelsintherathippocampus.TheJournalofcomparativeneurology,506(4),pp.569–83.

Tochitsky,I.etal.,2012.Optochemicalcontrolofgeneticallyengineeredneuronalnicotinicacetylcholinereceptors.Naturechemistry,4(2),pp.105–11.

Tran-Van-Minh,A.etal.,2015.Contributionofsublinearandsupralineardendriticintegrationtoneuronalcomputations.Frontiersincellularneuroscience,9,p.67.

Trigo,F.F.,Corrie,J.E.T.&Ogden,D.,2009.Laserphotolysisofcagedcompoundsat405nm:Photochemicaladvantages,localisation,phototoxicityandmethodsforcalibration.JournalofNeuroscienceMethods,180(1),pp.9–21.

Trimmer,J.S.&Rhodes,K.J.,2004.Localizationofvoltage-gatedionchannelsinmammalianbrain.Annualreviewofphysiology,66(1),pp.477–519.

Vetter,P.,Roth,A.&Häusser,M.,2001.Propagationofactionpotentialsindendritesdependsondendriticmorphology.Journalofneurophysiology,85(2),pp.926–937.

Volgraf,M.etal.,2006.Allostericcontrolofanionotropicglutamatereceptorwithanopticalswitch.Naturechemicalbiology,2(1),pp.47–52.

Waters,J.,Schaefer,A.&Sakmann,B.,2005.Backpropagatingactionpotentialsinneurones:Measurement,mechanismsandpotentialfunctions.ProgressinBiophysicsand

48

MolecularBiology,87(1SPEC.ISS.),pp.145–170.Wathey,J.C.etal.,1992.ComputersimulationsofEPSP-spike(E-S)potentiationin

hippocampalCA1pyramidalcells.TheJournalofneuroscience :theofficialjournaloftheSocietyforNeuroscience,12(2),pp.607–618.

Wei,D.-S.,2001.CompartmentalizedandBinaryBehaviorofTerminalDendritesinHippocampalPyramidalNeurons.Science,293(5538),pp.2272–2275.

Williams,S.R.,2004.Spatialcompartmentalizationandfunctionalimpactofconductanceinpyramidalneurons.Natureneuroscience,7(9),pp.961–7.

Wong,R.K.,Prince,D.A.&Basbaum,aI.,1979.Intradendriticrecordingsfromhippocampalneurons.ProceedingsoftheNationalAcademyofSciencesoftheUnitedStatesofAmerica,76(2),pp.986–990.

Xu,N.etal.,2012.Nonlineardendriticintegrationofsensoryandmotorinputduringanactivesensingtask.Nature,492(7428),pp.247–51.

Yang,S.,Emiliani,V.&Tang,C.,2014.ThekineticsofmultibranchintegrationonthedendriticarborofCA1pyramidalneurons.Frontiersincellularneuroscience,8(May),p.127.

Yaroslavsky,A.N.etal.,2002.Opticalpropertiesofselectednativeandcoagulatedhumanbraintissuesinvitrointhevisibleandnearinfraredspectralrange.Physicsinmedicineandbiology,47(12),pp.2059–73.

Yasuda,R.etal.,2004.Imagingcalciumconcentrationdynamicsinsmallneuronalcompartments.Science’sSTKE :signaltransductionknowledgeenvironment,2004(219),p.pl5.

Yizhar,O.etal.,2011.Optogeneticsinneuralsystems.Neuron,71(1),pp.9–34.Zhang,F.etal.,2007.Multimodalfastopticalinterrogationofneuralcircuitry.Nature,

446(7136),pp.633–9.