<%BANNER%>

Hippocampal Circadian Rhythms in Temporal Lobe Epilepsy

MISSING IMAGE

Material Information

Title:
Hippocampal Circadian Rhythms in Temporal Lobe Epilepsy
Physical Description:
1 online resource (144 p.)
Language:
english
Creator:
Stanley, David A
Publisher:
University of Florida
Place of Publication:
Gainesville, Fla.
Publication Date:

Thesis/Dissertation Information

Degree:
Doctorate ( Ph.D.)
Degree Grantor:
University of Florida
Degree Disciplines:
Biomedical Engineering
Committee Chair:
Carney, Paul Richard
Committee Co-Chair:
Talathi, Sachin S
Committee Members:
Harris, John Gregory
Khargonekar, Pramod P

Subjects

Subjects / Keywords:
circadian -- epilepsy -- hippocampus
Biomedical Engineering -- Dissertations, Academic -- UF
Genre:
Biomedical Engineering thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract:
For over a century epileptic seizures have been known to cluster at specific times of the day. Recent studies have suggested that the circadian regulatory system may become permanently altered in epilepsy, but little is known about how this affects neural activity and the daily patterns of seizure recurrence. To investigate, we tracked long-term changes in the rate of spontaneous hippocampal EEG spikes (SPKs) in a rat model of temporal lobe epilepsy. In healthy animals, SPKs oscillated with near 24-hour period; however, following injury by status epilepticus (SE), a persistent phase shift of approximately 12 hours emerged in animals that later went on to develop chronic spontaneous seizures. Additional measurements showed that this phase shift affected other features of hippocampal activity, including 24-hour modulation of gamma- and beta-frequency rhythms, but did not affect global 24-hour rhythms, including core body temperature and theta state transitions. Based on this, we hypothesized that the phase shift might be due to locally impaired circadian input to the hippocampus. This was investigated using a biophysical computer model in which we showed that subtle changes in the relative strengths of circadian inputs could produce a phase shift in hippocampal neural activity. Additional EEG analysis provided evidence for altered circadian input strengths by showing that the amplitude of 24-hour modulation of certain EEG rhythms, particularly hippocampal theta, was attenuated following SE injury. Furthermore, MRI provided evidence that the medial septum, a putative circadian relay center, exhibited signs of damage and therefore could contribute to local circadian impairment. Our results suggest that balanced circadian input is critical to maintaining natural circadian phase in the hippocampus, and that damage to the medial septum may disrupt this balance.
General Note:
In the series University of Florida Digital Collections.
General Note:
Includes vita.
Bibliography:
Includes bibliographical references.
Source of Description:
Description based on online resource; title from PDF title page.
Source of Description:
This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility:
by David A Stanley.
Thesis:
Thesis (Ph.D.)--University of Florida, 2013.
Local:
Adviser: Carney, Paul Richard.
Local:
Co-adviser: Talathi, Sachin S.

Record Information

Source Institution:
UFRGP
Rights Management:
Applicable rights reserved.
Classification:
lcc - LD1780 2013
System ID:
UFE0045234:00001

MISSING IMAGE

Material Information

Title:
Hippocampal Circadian Rhythms in Temporal Lobe Epilepsy
Physical Description:
1 online resource (144 p.)
Language:
english
Creator:
Stanley, David A
Publisher:
University of Florida
Place of Publication:
Gainesville, Fla.
Publication Date:

Thesis/Dissertation Information

Degree:
Doctorate ( Ph.D.)
Degree Grantor:
University of Florida
Degree Disciplines:
Biomedical Engineering
Committee Chair:
Carney, Paul Richard
Committee Co-Chair:
Talathi, Sachin S
Committee Members:
Harris, John Gregory
Khargonekar, Pramod P

Subjects

Subjects / Keywords:
circadian -- epilepsy -- hippocampus
Biomedical Engineering -- Dissertations, Academic -- UF
Genre:
Biomedical Engineering thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract:
For over a century epileptic seizures have been known to cluster at specific times of the day. Recent studies have suggested that the circadian regulatory system may become permanently altered in epilepsy, but little is known about how this affects neural activity and the daily patterns of seizure recurrence. To investigate, we tracked long-term changes in the rate of spontaneous hippocampal EEG spikes (SPKs) in a rat model of temporal lobe epilepsy. In healthy animals, SPKs oscillated with near 24-hour period; however, following injury by status epilepticus (SE), a persistent phase shift of approximately 12 hours emerged in animals that later went on to develop chronic spontaneous seizures. Additional measurements showed that this phase shift affected other features of hippocampal activity, including 24-hour modulation of gamma- and beta-frequency rhythms, but did not affect global 24-hour rhythms, including core body temperature and theta state transitions. Based on this, we hypothesized that the phase shift might be due to locally impaired circadian input to the hippocampus. This was investigated using a biophysical computer model in which we showed that subtle changes in the relative strengths of circadian inputs could produce a phase shift in hippocampal neural activity. Additional EEG analysis provided evidence for altered circadian input strengths by showing that the amplitude of 24-hour modulation of certain EEG rhythms, particularly hippocampal theta, was attenuated following SE injury. Furthermore, MRI provided evidence that the medial septum, a putative circadian relay center, exhibited signs of damage and therefore could contribute to local circadian impairment. Our results suggest that balanced circadian input is critical to maintaining natural circadian phase in the hippocampus, and that damage to the medial septum may disrupt this balance.
General Note:
In the series University of Florida Digital Collections.
General Note:
Includes vita.
Bibliography:
Includes bibliographical references.
Source of Description:
Description based on online resource; title from PDF title page.
Source of Description:
This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility:
by David A Stanley.
Thesis:
Thesis (Ph.D.)--University of Florida, 2013.
Local:
Adviser: Carney, Paul Richard.
Local:
Co-adviser: Talathi, Sachin S.

Record Information

Source Institution:
UFRGP
Rights Management:
Applicable rights reserved.
Classification:
lcc - LD1780 2013
System ID:
UFE0045234:00001


This item has the following downloads:


Full Text

PAGE 1

HIPPOCAMPALCIRCADIANRHYTHMSINTEMPORALLOBEEPILEPSY By DAVIDARTHURSTANLEY ADISSERTATIONPRESENTEDTOTHEGRADUATESCHOOL OFTHEUNIVERSITYOFFLORIDAINPARTIALFULFILLMENT OFTHEREQUIREMENTSFORTHEDEGREEOF DOCTOROFPHILOSOPHY UNIVERSITYOFFLORIDA 2013

PAGE 2

c r 2013DavidArthurStanley 2

PAGE 3

Tomyparents,JanandLeonard,andmybrother,Geoffrey 3

PAGE 4

ACKNOWLEDGMENTS Firstandforemost,Iwouldliketothankmymentors,Dr.Paul Carney,Dr.William Ditto,andDr.SachinTalathi.Theyprovidedmewithguidanc ethroughoutmystudies andwerefullydedicatedtoenablingmetosucceedasaresear cher.Theyheldme tothehigheststandardsofscholarlywork.Forthephonecal ls,Skypemeetings,and passionatescienticdebates,manyofwhichcarriedonlate intotheevenings,Iammost sincerelygrateful.Ialsowishtoacknowledgemycommittee membersandcollaborators atArizonaStateUniversityandtheUniversityofFloridawh oprovidedoutstanding feedbackandadvicethroughoutmyPhD:Dr.KevinBennett,Dr .JohnHarris,Dr. LeonidasIasemidis,Dr.PramodKhargonekar,Dr.MarkL.Spa no,andDr.Stephen HelmsTillery. IhavemanypeopletothankforwhereIamtoday.Myelementary schoolfriend CarsonMokencouragedmetobecomeinterestedinmathematic satayoungage.Mr. ColinWackett,myGrade8teacheratMountAlbert,wasveryin spirationalandinstilled inmealoveoflearning.Ihavetothankmysubsequentmathand scienceteachers atDr.JohnM.Denisonfortheirinstructionandfororganizi ngtheMathandScience Olympicsteams,includingMrs.S.Dalrymple,Mrs.C.MacIsa ac,Mr.OlavRandoja,Mr. Churchill,Mr.Law,andMr.DarrenLluoma.Iamalsoindebted toMr.Oberfrankforhis guidancethroughoutOACEnglish. IhadseveralmentorsattheUniversityofToronto,includin gDr.PeterH.Backx,who tookmeintohistopcalibercardiacelectrophysiologylabw henIwasstillahighschool studentandwhoexposedmetoscienticresearchfortherst time.Ialsomustthank Dr.MarkJellinekandDr.NazirP.Kheraniforinvitingmeint otheirlabsandforproviding mewithfurtherearlyexposuretoscienticresearch.Dr.Ma rkJellineksupervisedmy researchonthephysicsofviscousmixingduringthesummero f2004.Dr.NazirKherani supervisedmyEngineeringScienceBachelor'sthesisresea rch,andhelpedmeprepare andpublishmyrstresearchpaper.Iamalsoverygratefulfo rmyopportunitytowork 4

PAGE 5

withDr.MartinRenqiangMin.HetaughtmewhatIknowaboutma chinelearning,and Ideeplyenjoyedourcollaboration.Finally,Iowesinceres tgratitudetomyMaster's advisor,Dr.BerjBardakjian,forencouragingmetoenterth eeldofcomputational neuroscience.WhileIwasalwaysinterestedinthebrain,hi sanimatedclassroom lecturesrevealedthebeautifulmathematicsthatunderlie itsfunction.Hehasbeena superbteacher,mentorandfriend. WhileIwasattheUniversityofToronto,myfriendErnestHow ascompletinghis PhDinneuroscienceandtaughtmemanyimportantlessonsdur ingourdinnertime conversations.Healsoprovidedvaluablefeedbackonmywor kandinspiredmuchofmy writingstyleforthisthesis–Iwillpayhomagewherepossib le.OfmymanyCanadian friends,ShawnBrown,GrishaBoyko-Vekin,LorneChi,Sinis aColic,JoshDian,Saigin Govender,CynthiaLong,KirkHooper,JeromeHu,StephenHut chison,SaraKrowicki, RyanJanzen,MichaelSimmonds,andTristaHurley-Waxaliha veallshapedthisworkin specicways.DespitemyleavingCanada,Iamverygratefult hatyouhaveremained intouchandhavesupportedmethroughoutmyPhD.Ialsowisht oacknowledgemy closefriendsfromArizonaStateUniversity,AnnaDari,Beh namKia,DavidGuffrey,and SrinidhiKuntaegowdanahalli,fortheirsupport,advice,a ndwit.Veryspecialthanksgoto PengpengCaoforbeinganamazingfriendandforalwaysbeing there(especiallywith our“interlibraryloans”). FollowingmymovetotheUniversityofFlorida,Ifoundanewh omeamongsta newgroupofcolleagues.IwishtothankEricBennett,Fernan doDelgado,Francisco Casanova,RobCastellucci,SonamChheda,DianaGu,MariaHo uria,Svetlana Kantorovich,QianyingLin,GowriNatarajan,MansiParekh, AliPiracha,StefanoRe Fraschini,MattShore,DeguangWang,RabiaZafar,RujutaMu nje,ClaireCox,A.M. MerelBoers,andJunliZhoufortheirfriendshipandadvice. Also,specialthanksto MansiforherhelpandtutelagewithMRIanalysis,Ericforhe lpingmegetsetupin 5

PAGE 6

Gainesville,andTinyMcDonaldforsurmountingallthebar rierstomytransferringto UF. Finally,thisworkwouldhavenotbeenpossiblewithoutthes upportofmyfamily. Myspectacularmother,Janet,spentcountlesshoursexposi ngmeatayoungage tomyriadmathproblems,brainteasers,puzzles,andlitera ryworks.Shealsodid herbesttoeducatememusicallybyteachingmepiano,andcon tinuestoeditmy paperstothisday.Myfather,Leonard,helpedmepursuemych ildhoodpassionfor disassemblingvariouselectronicdevices,whichIsuspect relatestomyinterestin dynamicalsystems.Hehasalwayskeptawatchfuleyeonthebi gpictureandhas providedmewithinsightfulguidancethroughoutmylife.To UncleSteve,thankyoufor yoursupportandencouragement,forsharingyourloveofpuz zles,andfor“takingmeto school”meatcards.Tomygrandparents,SallyandJim,youga vemeagreatchildhood andinspiredmetoliveafulllingandbalancedlifeaccordi ngtostrongcorevalues.It isforyouthatIwill“believeinallIcanachieve.”Ihavecom etorealizethatitlargely tracesbacktomylateGrandpaStanthatIamnowpursuinganac ademiccareer.Iam gratefultomylateGrandmaBarbaraforallherkindnessandf orthevaluesheplacedon education.MylateGrandpaArthadepilepsyhimself.Althou ghIdidnotknowhim,he haslargelyshapedwhoIamtoday.Finally,tomybrotherGeof frey,youthinkdeeplyand youareboundtoaccomplishgreatthings.Icanalwaysdepend onyouandlookforward tothefutureandallitsadventures.Let's“tryforthesun.” Inadditiontomyfriendsandmentors,noneofthisworkwould havebeen possiblewithoutfunding.Iamdeeplygratefultothefundin gagencieswhosupported myPhDstudies,inparticular,theNationalSciencesandEng ineeringResearch CouncilofCanada(NSERC),theNationalInstitutesofHealt h(NIH),andtheOfce ofNavalResearch(ONR).Additionally,boththeQueenEliza bethIIAimingforthe TopScholarshipandtheHeartandStrokeFoundationhelpedt ofundmyexposureto researchpriortobeginningmyPhD. 6

PAGE 7

Researchinmanybiologicaleldsinvolvestheuseandsacri ceofanimals,and ethicalargumentscanbemadebothforandagainstanimalexp erimentation.Thiswork didnotinvolvethedirectsacriceofanimals;however,itd idbenetfromdatacollected inpreviousanimalstudies.Iamgratefultothosewhoprovid edmewithexperimental data,includingDanielCordiner,Dr.Dong-UkHwang,Dr.Wen dyNorman,Dr.Mansi Parekh,Dr.SachinS.Talathi,andDr.JunliZhou,andIamres pectfulfortheanimals sacricedtoprovidethisdata. Muchofthemodelingworkdoneinthisthesiswasmadepossibl ebytheopen-source GENESISsoftwarepackagedevelopedbyDaveBeemanandJames Bower.Thisthesis istypesetinL A T E X2 .ReferenceswereimportedtoBibTexusingBIBTEXFORMAT,an incrediblyusefultoolwrittenbyBenBulheller.Imageswer eprocessedusingGIMP.Will M.D.FoeprovidedtheL A T E Xtemplatethatisusedinwritingthisthesis. 7

PAGE 8

TABLEOFCONTENTS page ACKNOWLEDGMENTS .................................. 4 LISTOFTABLES ...................................... 11 LISTOFFIGURES ..................................... 12 ABSTRACT ......................................... 14 CHAPTER 1INTRODUCTION ................................... 16 2BACKGROUNDANDHYPOTHESIS ........................ 19 2.1WhyStudyEpilepsy? ............................. 19 2.2EpilepsyOverview ............................... 21 2.2.1SeizureTypes .............................. 21 2.2.2ClassicationofEpilepsies ....................... 21 2.2.3Etiology ................................. 22 2.2.4TemporalLobeEpilepsy ........................ 23 2.3HippocampalElectroencephalogram ..................... 24 2.3.1HippocampalGammaandThetaRhythms .............. 25 2.3.2HippocampalSharpWavesandInterictalSpikes .......... 26 2.4OnSeizuresandCycles ............................ 27 2.4.1Effectof24-HourRhythmsonSeizures ................ 28 2.4.2EffectofEpilepsyon24-HourRhythms ................ 30 2.5Motivation .................................... 32 2.6Hypotheses ................................... 33 2.7OrganizationofthisDissertation ....................... 34 3THEMAMMALIANCIRCADIANSYSTEM ..................... 36 3.1Overview .................................... 36 3.2AHierarchyofOscillators ........................... 36 3.3CircadianRegulationoftheWake-SleepCycle ............... 38 3.424-HourRhythmsintheBrain ......................... 40 3.524-HourRhythmsintheHippocampus .................... 40 3.6CircadianInputstotheHippocampus ..................... 40 4GENERALMETHODS ................................ 48 4.1Overview .................................... 48 4.2AnimalHusbandry ............................... 48 4.3Long-TermDataCollection .......................... 49 4.4SurgicalMethods ................................ 49 8

PAGE 9

4.5InductionofSelf-SustainingStatusEpilepticus ............... 50 4.6SeizureDetection ............................... 51 4.7EEGandCBTDataAcquisition ........................ 52 4.8ThetaEpochDetection ............................. 52 4.9DataAnalysis .................................. 53 4.9.1SmoothingandDetrending ...................... 53 4.9.2CosinorAnalysis ............................ 53 5HIPPOCAMPALEEGSPIKES:24-HOURRHYTHMPHASESHIFT ...... 54 5.1Overview .................................... 54 5.2Methods ..................................... 54 5.2.1AnimalMethodsandDataCollection ................. 54 5.2.2Spike(SPK)Detection ......................... 55 5.2.3Statistics ................................. 55 5.3Results ..................................... 55 5.3.1PhaseShiftin24-HourRhythmofEEGSPKs ............ 55 5.3.2CoreBodyTemperatureAnalysis ................... 57 5.3.3ThetaActivityAnalysis ......................... 57 5.4Discussion ................................... 62 5.4.1SummaryofResults .......................... 62 5.4.2VariabilityofDatainPre-InjuryTimePeriodofNon-S eizingAnimals 62 5.4.3InterictalSpikesasaDriverforthePhaseShift ........... 63 6MECHANISMSFORCIRCADIANRHYTHMPHASESHIFTING ........ 64 6.1Overview .................................... 64 6.2Methods ..................................... 64 6.2.1ModelingOverview ........................... 64 6.2.2CA3NetworkModel .......................... 66 6.2.3NetworkProperties ........................... 68 6.2.4BackgroundActivity .......................... 69 6.2.5CircadianModulation .......................... 70 6.2.6SensitivityAnalysis ........................... 72 6.3Results ..................................... 73 6.4Discussion ................................... 74 6.4.1GeneralityoftheComputerModeltoAlternativeSourc esofCircadian Perturbation ............................... 74 6.4.2RelationshipBetweenExperimentalSPKPhaseShiftan dComputer ModelingResults ............................ 76 7STRUCTURALCHANGEINTHECIRCADIANSYSTEM ............ 78 7.1BackgroundandOverview ........................... 78 7.2Methods ..................................... 79 7.2.1AnimalMethods ............................. 79 7.2.2ExperimentalTimeline ......................... 79 9

PAGE 10

7.2.3MRIDataCollectionandAnalysis ................... 79 7.3Results ..................................... 81 8HIPPOCAMPALEEGRHYTHMS:DISRUPTED24-HOURREGULATION ... 83 8.1BackgroundandOverview ........................... 83 8.2Methods ..................................... 86 8.2.1AnimalMethodsandDataCollection ................. 86 8.2.2ExtractionofPowerSpectralDensities ................ 86 8.2.3DataAnalysis .............................. 87 8.2.4AnalysisofCorrelationBetweenPhaseandAmplitude ....... 88 8.2.5PrincipalComponentAnalysis ..................... 88 8.2.6Statistics ................................. 89 8.3Results ..................................... 89 8.3.1PhaseShiftin24-HourModulationofBetaandGammaFre quency Rhythms ................................. 89 8.3.2ImbalancedCircadianInputasaDriverforthePhaseSh ift ..... 91 8.3.3Phase–AmplitudeRelationshipof24-HourRhythms ........ 91 8.3.424-HourRhythmsareMultidimensional ................ 93 8.3.5Altered24-HourModulationofThetaRhythmPower ........ 94 8.3.624-HourRhythmsinNon-SeizingRats ................ 95 8.4Discussion ................................... 96 8.4.1RelationshiptoPreviousCircadianLiterature ............ 99 8.4.2InterpretationofResults ........................ 105 8.4.2.1Changesin24-HourRegulationofHippocampalTheta 105 8.4.2.2EffectofThetaStateonPhaseShift ............ 107 8.4.3UnderlyingPhysiologicalMechanisms ................ 107 8.4.4AlternativeMechanisms ........................ 109 8.4.5FutureWork ............................... 109 8.5ClosingRemarks ................................ 110 9DISCUSSIONANDCONCLUSIONS ........................ 112 9.1Overview .................................... 112 9.2ImplicationsofAlteredCircadianRhythms .................. 113 9.2.1CognitiveImpairment .......................... 113 9.2.2EmergenceofSeizures ........................ 113 9.3ClosingRemarks ................................ 114 APPENDIX ACOMPLETEEEGSPKDATA ............................ 117 BCOMPLETEEEGRHYTHMDATA ......................... 119 REFERENCES ....................................... 129 BIOGRAPHICALSKETCH ................................ 144 10

PAGE 11

LISTOFTABLES Table page 3-1Brainregionsexhibiting24-hourrhythms invivo ................. 41 3-2Experimentalmeasurementofhippocampal24-hourrhyth ms .......... 45 3-3Synapticcircadianinputstothehippocampus ................... 47 3-4Non-synapticcircadianinputstothehippocampus ................ 47 6-1Maximalsynapticconductances(nS) ........................ 69 6-2CA3networkconnectivityexpressedintermsofsynaptic convergence ..... 70 6-3Backgroundsynapticactivity ............................ 70 11

PAGE 12

LISTOFFIGURES Figure page 3-1Mammaliancircadiansystem ............................ 39 4-1ExperimentaltimelineforEEG ........................... 49 4-2Electrodeplacement ................................. 51 5-1SPKshapeproles .................................. 56 5-2PhaseshiftinEEGSPK24-hourrhythm ...................... 58 5-3PhaseofCBT24-hourrhythms ........................... 59 5-4Extractedthetaepochs ............................... 60 5-5Phaseoftheta24-hourrhythms ........................... 61 6-1Modelnetworkconnectivity ............................. 65 6-2Biophysicalmodelofphaseshift .......................... 75 7-1ExperimentaltimelineforMRI ............................ 79 7-2MRIanalysisofmedialseptumandmbria .................... 82 8-1EEGfrequencycontent ............................... 96 8-2ExtractionofEEGPSDrhythms .......................... 97 8-3Phaseshiftinbetaandgammafrequencybands ................. 98 8-4Phenomenologicalmodel .............................. 99 8-5PCAsuggestsmultiplecircadianinputs ...................... 100 8-6Phase–amplitudecorrelationof24-hourrhythms ................. 101 8-7Thetarhythmalteration ............................... 102 8-8Phaseshiftisenhancedinnon-thetastates .................... 103 8-9Stabilityofphaseandamplitudefornon-seizingrats ............... 104 9-1Proposedroleofphaseshiftinepilepsy ...................... 115 A-1Phaseshiftin24-hourrhythmsforallseizingrats ................. 117 A-2Phaseshiftin24-hourrhythmsforallnon-seizingrats .............. 118 B-1FullPSDandphaseestimatesfromRat1 ..................... 120 12

PAGE 13

B-2FullPSDfromRat2 ................................. 121 B-3FullPSDfromRat3 ................................. 122 B-4FullPSDfromRat4 ................................. 123 B-5FullPSDfromRat5 ................................. 124 B-6FullPSDfromRat6 ................................. 125 B-7SinusoidaltsforRats1and2 ........................... 126 B-8SinusoidaltsforRats3and4 ........................... 127 B-9SinusoidaltsforRats5and6 ........................... 128 13

PAGE 14

AbstractofDissertationPresentedtotheGraduateSchool oftheUniversityofFloridainPartialFulllmentofthe RequirementsfortheDegreeofDoctorofPhilosophy HIPPOCAMPALCIRCADIANRHYTHMSINTEMPORALLOBEEPILEPSY By DavidArthurStanley August2013 Chair:PaulR.CarneyCochair:SachinS.TalathiMajor:BiomedicalEngineering Foroveracenturyepilepticseizureshavebeenknowntoclus teratspecictimes oftheday.Recentstudieshavesuggestedthatthecircadian regulatorysystemmay becomepermanentlyalteredinepilepsy,butlittleisknown abouthowthisaffects neuralactivityandthedailypatternsofseizurerecurrenc e.Toinvestigate,wetracked long-termchangesintherateofspontaneoushippocampalEE Gspikes(SPKs)inarat modeloftemporallobeepilepsy.Inhealthyanimals,SPKsos cillatedwithnear24-hour period;however,followinginjurybystatusepilepticus(S E),apersistentphaseshift ofapproximately12hoursemergedinanimalsthatlaterwent ontodevelopchronic spontaneousseizures.Additionalmeasurementsshowedtha tthisphaseshiftaffected otherfeaturesofhippocampalactivity,including24-hour modulationofgamma-and beta-frequencyrhythms,butdidnotaffectglobal24-hourr hythms,includingcore bodytemperatureandthetastatetransitions.Basedonthis ,wehypothesizedthat thephaseshiftmightbeduetolocallyimpairedcircadianin puttothehippocampus. Thiswasinvestigatedusingabiophysicalcomputermodelin whichweshowedthat subtlechangesintherelativestrengthsofcircadianinput scouldproduceaphase shiftinhippocampalneuralactivity.AdditionalEEGanaly sisprovidedevidencefor alteredcircadianinputstrengthsbyshowingthattheampli tudeof24-hourmodulation ofcertainEEGrhythms,particularlyhippocampaltheta,wa sattenuatedfollowingSE injury.Furthermore,MRIprovidedevidencethatthemedial septum,aputativecircadian 14

PAGE 15

relaycenter,exhibitedsignsofdamageandthereforecould contributetolocalcircadian impairment.Ourresultssuggestthatbalancedcircadianin putiscriticaltomaintaining naturalcircadianphaseinthehippocampus,andthatdamage tothemedialseptummay disruptthisbalance. 15

PAGE 16

CHAPTER1 INTRODUCTION “Nowsailorsalltakemyadvice—Letsteamshipsbeyourmotta —Andnevergo toseaagain—InthesailingshipCalcutta!” ThiswaswrittenbyJohnFisheratage 14,amidshipman-to-beaboardtheCalcutta.Duringthecour seofhislife(1841-1920) seafaringvesselsexperiencedamonumentaltransformatio n,fromwoodenships propelledbysailstosteel-hulledshipsdrivenbysteamtur bines.TheBritishNavyofthe time,beinganoldandimmenselysuccessfulorganization,w ashighlyresistanttosuch change;indeed,astandardnavyinstructionbookstates:“T hereisnogreaterfallacy thantosupposethatshipscanbenavigatedonlongvoyageswi thoutmastsandsails” [ Nares 1868 ]. ForFisher,however,itwasatimeofinnovation.Hefeltnono stalgiaforthesailing shipsofNelson'stime,andinsteadsoughttoleveragethete chnologicaladvances oftheindustrialrevolutiontotheirfullestpotential.He proposedusingtheinternal combustionengine,asopposedtosteamboilers,topowershi pturbines.Hepredicted theimportance(andthreat)ofsubmarinetechnology,which wasperceivedbyhis contemporariesas“underhanded,unfair,anddamnedun-Eng lish”[ Morris 2010 ].Early inhiscareer,hepublishedapaperonelectrictorpedoguida nce[ Fisher 1871 ].He wasanenthusiastforaviationandthefatherofthe1906Drea dnought,whichsetthe templatefor20thcenturybattleshipdesign.Heenvisioned anoilpipelineunderthe Englishchannel,aprojectthatcametofruitionin1944when itwasusedtosupply troopsinNormandy. 1 TherevolutioninnavalarchitectureduringFisher'stimem irrorstheconnectionoften seeninthesciences,wherebyscienticdiscoveriesarelin kedtoadvancesinequipment 1 AnengagingaccountofAdmiralJohnFisher'slife,beyondhi stechnicalwork,is providedby Morris [ 2010 ]. 16

PAGE 17

ortechnique.ClassicalexamplesincludeGalileo'stelesc ope,whichledtothediscovery ofJupiter'smoons,andRamonyCajal'semploymentofsilver staining,whichyielded thefundamentaltenetsofneuroscience.Fisherhimselfund erstoodthedisruptivenature oftechnologyandthedangerofresistingchange.Heisremem beredtodayasperhaps thesecondmostimportantgureinBritishnavalhistory,af terLordNelson.Importantly, despitereachingthehighestranksoftheadmiralty,hisren ownisnotforhisrolein navalengagements,butratherforhisforward-thinkingref orms.Thisunderscoresthe importanceofhisroleasanearlypromoteroftechnicalinno vation. Presently,theeldofneuroscienceisenteringaperiodofd iversication.Its mainstaytechniqueofelectrophysiologicalrecordinghas remainedrelativelyunchanged forover50years,withthenotableexceptionofpatchclampi ngdevelopmentsin the1980s.Nowhowever,withadvancesincellbiology,newsu itesofopticaland moleculartechniquesareenablingthestudyofneuronswith greaterresolution andspecicitythanpreviouslypossible.Theburgeoninge ldofsyntheticbiologyis hopedtopresentsophisticatedwaystointeractwithneuron soncellularandmolecular levels. 2 Ambitioushardwareprojectsseektoengineerhighdensitye lectroderecording arrays.Non-invasivebrainimagingisimprovingrapidly,a ndhasshownatrendof doublingitsresolutioneverytwelvemonths[ Kurzweil 2005 ].Thesedevelopments havenecessitatedthecreationofnewmathematicalmodels[ Truccoloetal. 2005 ]and haverenedexistingmodelsbyallowingimprovedmeasureme ntofmodelparameters [ GerstnerandNaud 2009 ].Together,theseadvanceshaverecentlypromptedboth 2 Syntheticbiology,whichaimstodevelopstandardizedmeth odsandcomponentsfor constructingbiologicalsystems,hasbeencomparedtothed evelopmentofstandardized partsandprocessesthatoccurred,forexample,duringthei ndustrialrevolutionand, morerecently,duringthesemiconductorrevolution[ Bakeretal. 2006 ].Withsuch aspirations,the21stcenturyhasbeenoptimisticallydubb edthe“centuryofbiology,” followingofthecenturyofchemistry(19thcentury)andthe centuryofphysics(20th century).Thehistoricaloriginofthesetermsisdiscussed by Hingstonetal. [ 2008 ]. 17

PAGE 18

AmericanandEuropeaninitiativestobetterunderstandthe brainandneurological disease.LiketheindustrialrevolutionofFisher'stime,t hesenoveltechniquesaregreatly transformingtheeldofneuroscience. Theresearchpresentedinthisdissertationfollowsthethe meofleveragingnew technology.Aswehavewitnessedoverthelastdecade,impro vementsincomputer processingpowerandstoragecapacityhavebeenimpressive ,andthishasmadeit feasibletocollectandanalyzevastamountsofneuraldata. Infact,itisnowpossible totrackmillisecondtimescaleactionpotentials,thefund amentalunitsofneuralactivity, continuouslyoverdaysandevenweeks.Thismakesslowtimes caleprocesses,such as24-hourcircadianrhythms,attractivetargetsofstudy. Slowrhythmshavebeen examinedextensivelyinisolation,butrelativelylittlei sknownabouthowtheyinteract withotherbrainfunctions,especiallyinrelationtodisea se.Inthisthesiswewillexamine 24-houroscillationsinthebrain,andwewillstudythemwit hinthecontextofepilepsy, apathologythatexhibitsbothrapidelectroencephalogram abnormalitiesandalso slowtimescaleevolution.Itisourmodesthopethatthiswor kwillencouragefurther explorationofslowtimescaleprocesses,alongsidethosem oretraditionallystudiedin neuroscience,leadingtoanimprovedunderstandingofther elationshipbetweenthese vastlydifferenttimescales.Moregenerally,wehopethatt hisworkwillconveytothe readerasenseofexcitementforthepotentialoftechnology toextendthefrontiersof neuroscienceresearchinthecomingyears. 18

PAGE 19

CHAPTER2 BACKGROUNDANDHYPOTHESIS 2.1WhyStudyEpilepsy? Epilepsyisoneoftheoldestdiseasesknowntohumanity,hav ingbeendescribed over2000yearsagoinBabylonianstonetabletinscriptions [ ReynoldsandWilson 2008 WilsonandReynolds 1990 ]. 1 Thesetabletsarepartofaseriescalledthe Sakikku ananthologyofdiseasethatdatesasearlyas718B.C.E.Thes criptonthesetablets providesarigorousdescriptionofmanyclassicalsymptoms ofepilepsy,including paranoia,hallucinations,andmooddisorders.Inaddition ,thesescriptsattemptto explaintheunderlyingsourceoftheepilepticattacks,att ributingthemtodemonsand othersupernaturalforces.ManyancientGreekssharedthis beliefaboutthemystical natureofepilepsy,andreferredtoitasthe“sacreddisease .”However,arefutationofthis mysticismisprovidedbyatreatiseintheHippocraticCorpu s: “Menregarditsnatureandcauseasdivinefromignoranceand wonder, becauseitisnotatallliketootherdiseases.Andthisnotio nofitsdivinity iskeptupbytheirinabilitytocomprehendit...Butifitisr eckoneddivine becauseitiswonderful,insteadofonetherearemanydiseas eswhich wouldbesacred;for,asIwillshow,thereareothersnolessw onderfuland prodigious,whichnobodyimaginestobesacred.” 2 Importantly,ithasbeenarguedthatthisrefutationmaybeo neoftheearliest recordeddistinctionsbetweenmysticismandscience[ RiggsandRiggs 2005 ]. 1 PartsofthischapterareinpresswiththeJournalofNeuroph ysiology:Stanley,D.A., Talathi,S.S.,Parekh,M.B.,Cordiner,D.,Zhou,J.,Mareci ,T.H.,Ditto,W.L.,Carney,P.R. Localphaseshiftinthe24-hourrhythmofhippocampalEEGsp ikingactivityinarat modeloftemporallobeepilepsy[ Stanleyetal. 2013 ]. 2 From OntheSacredDisease ,apartoftheHippocraticCorpus,translatedby FrancisAdamsandpublishedbyKaplan[ Fischer 2008 ]. 19

PAGE 20

Itisappropriatethatepilepsywasatthecentreofthisearl ydebateonthe distinctionbetweenmysticismandscience.Indeed,itisst illamatterofdebatetoday astowhetheroperationsofthebrain,particularlythegene rationoftheconscious experience,iswithinthescopeofscientictheory.Itseem sdifculttoimaginehow acollectionofatomsobeyingthelawsofphysicscouldgiver isetothesubjective experience[ Ho 2011 ]. 3 Whileresearchontheneuralbasisofconsciousnessis progressing, 4 thisisstillverymuchanopenquestion.Epilepsyisoftenre ferredto asthemodeldiseasetostudyforpurposesofunderstandingt hebrain.Inparticular, researchintothelossofconsciousnessaccompanyingcerta intypesofepileptic seizuresmayyieldcriticalinsightstothesequestions. Inadditiontothedesiretounderstandthebrain,manyscien tistschooseto studyepilepsyduetothedemandfortreatment.Epilepsyist hethirdmostcommon neurologicaldisorderafterstrokeandAlzheimer'sdiseas e[ Hirtzetal. 2007 ],withone in26peopledevelopingepilepsyatsomepointintheirlives ,andonein10people experiencingaseizure.Overall,approximately30-40%ofe pilepsycasesarerefractory topresentpharmacologicaltherapy[ KwanandBrodie 2000 ].Forthosepatientsthat aresuccessfullytreated,therearenumeroussideeffectsw ithwhichtheymustcontend. Epilepticsaremorelikelytosufferfrommentalandcogniti vedecits,andcanalso experiencesleepdisordersandothercomorbiditiesthatdi minishtheirqualityoflife. Itisdifculttoassessandquantifythesocietalimpactsof epilepsy.Oneapproachis toexaminethedirectandindirectmonetarycostsassociate dwiththedisease.Amajor studyby Begleyetal. [ 2000 ]estimatestheprojectedannualcostofepilepsyintheUS 3 Chalmers [ 1995 ]commentsthatanyphysicalexplanationforanexperience, for example,ofthecolourred,suffersthecriticismthatsucha physicalprocesscouldalso occurintheabsenceofsuchanexperience. 4 NobellaureateFrancisCrickdedicatedmuchofhislaterlif etodevelopinga frameworkforstudyingconsciousness[ CrickandKoch 2003 ]. 20

PAGE 21

isinexcessof$12.5billion.Importantly,indirectcosts, suchascomorbiddiseasesand socioeconomiceffects,includingforegoneearnings,hous ecare,andunemployment, accountfor85%ofthesecosts. 5 Giventhemagnitudeoftheseindirectcosts,even therapiesthatdonotoutrightcuretheseizures,butrather makethemmoremanageable ormitigatetreatmentsideeffects,canhavealargeimpacto nreducingexpendituresand onimprovingqualityoflife. Theresearchpresentedinthisdissertationfocusesonthef undamentalmechanisms associatedwithepilepsy.Whilethisworkmaysomedaybeimp ortantfordevelopinga cure,themoreimmediateimplicationswillbefortacklingt heseindirectcostsandside effects,suchasimprovingseizurepredictability.Adetai leddiscussionofimplicationsof thisworkwillbeprovidedinsubsequentchapters. 2.2EpilepsyOverview 2.2.1SeizureTypes Epilepsyisaneurologicaldisordercharacterizedbyrecur rentspontaneous seizures.Aseizureisdenedasanabnormalsynchronousdis chargeofneural activityinthebrain.Epilepticseizuresmaybeeitherpart ial(alsoknownasfocal)or generalized.Partialseizuresexhibitalocalizedsiteofo nset,whereasgeneralized seizuresaredistributed.Partialseizurescanbeeithersi mpleorcomplex,depending whetherornotalossofconsciousnessoccurs.Theymayalsoe xhibitsecondary generalization,progressingtoinvolvethewholecortex.2.2.2ClassicationofEpilepsies Themostcommonepilepsyclassicationschemewasproposed bytheInternational LeagueAgainstEpilepsy(ILEA)in1981[ ILEA 1981 ].Thisschemeinvolvesclassifying 5 Lowerpercentagesforindirectcostshavebeenreportedout sideoftheUS. Strzelczyketal. [ 2008 ]havecompiledtheresultsof22studiesfromavarietyof geographicregions. 21

PAGE 22

epilepsysyndromesbasedonseizurelocalization,specic allywhethertheseizureis localized(partial,focalseizures)orgeneralized.Athir dcategoryisreservedforseizures thatareunclassied,eitherbecauseofinadequatedata,or becauseofsomeattributes thatdefyclassicationintoeitherofthetwoaforemention edgroups.Thisclassication wasrevisedbyILEAin1989intoadual-axisclassicationsc heme,withoneaxis describinglocalization,asabove,andtheotheraxisdescr ibingetiology[ ILEA 1989 ]. Fortherstaxis,anepilepticsyndromemaybeeither1)loca lization-related(local, partial,orfocal),2)generalized,3)special(suchassitu ation-relatedseizures),or4) undeterminedwhetherfocalorgeneralized.Fortheseconda xis,anepilepticsyndrome maybeA)idiopathic(age-related,usuallyofgeneticinher itance),B)symptomatic involvingaknowninjury,orC)cryptogenic.Cryptogenicep ilepsiesareepilepsiesfor whichtheetiologyisunknown.Theyarepresumedtobesympto matic,buttheyare alsoage-related.Localization-relatedsymptomaticepil epsies(category1-B)include commonepilepsysyndromessuchastemporal,parietal,fron tal,andoccipitallobe epilepsies.Absenceepilepsiestypicallyfallintocatego ry2-A,generalizedidiopathic syndromes.CertaininstancesofLennox-Gastautsyndromea reconsideredcryptogenic [ ILEA 1989 ]. 2.2.3Etiology Theunderlyingetiologyofepilepsyisvaried.Hereditarye pilepsieshavebeen associatedwithspecicproteinmutations,includingindi vidualvoltage-gatedand ligand-gatedchannels[ Biervertetal. 1998 Meisleretal. 2005 ].Thesemutations alterneuronalproperties,oftenmakingthemhyperexcitab le.Ontheotherhand, approximately40%ofepilepsiesareacquiredthroughinjur ytothenervoussystem [ HauserandHesdorffer 1990 Razaetal. 2004 ].Insuchcasesthebrainentersa silentperiodlastingmonthsoryearsfollowingtheinitial insult[ Angelerietal. 1999 Aroniadou-Anderjaskaetal. 2008 ].Thisisreferredtoasthe latency periodofepilepsy. Duringthistimeseizuresdonotoccurbutunderlyingpathol ogicalprocessesareat 22

PAGE 23

work,eventuallyculminatingintheemergenceofrecurrent spontaneousseizures. Followinginjuryandthroughoutthelatencyperiod,thebra inissaidtobeinastate of epileptogenesis .Epileptogenicprocessesmaycontinuetooccurevenaftert he beginningofchronicseizures. Theprecisemechanismsthatcausethehealthybraintotrans itiontotheepileptic stateareunknown.Onamostgenerallevel,epilepticseizur esarethoughttoresultfrom animbalancebetweenexcitationandinhibition[ El-Hassaretal. 2007 Sloviter 2005 ]. 6 Thisimbalancemayresultfromanumberofmechanisms,forex ample,sproutingof recurrentexcitatoryconnections,lossofinhibitorycell s,disinhibition(i.e.inhibitionof inhibitorycells),changesinreceptordensity,andinhibi tory-drivenhyper-synchrony. Theseconceptsaresummarizedindetailinanumberofreview articles[ Briggsand Galanopoulou 2011 DudekandSpitz 1997 ].Epilepticseizuresmaybedependent notonlyonlocalchangesinexcitability,butalsoonintera ctionamongstmanydistant brainregions.Forexample,ithasbeenproposedthat,eveni nthecaseofepilepsies classiedasfocal,multipledisparatebrainregionsmustb econnectedfortheseizureto occur[ Bertram 2009 ].Thisinvolvementofmultiplesitesmayexplaintherecurr enceof seizuresyearslaterfollowingresectionofwhatwasonceco nsideredtheprimaryseizure focus.2.2.4TemporalLobeEpilepsy TemporalLobeEpilepsy(TLE)isthemostcommontypeoflocal ization-related epilepsy[ Wiebe 2000 ]andisalsoamongthemostrefractorytopresentmedical treatment[ Engel 2001 ].TLEpatientsexhibitsimplepartialseizures,complexpa rtial seizures,andalsosecondarygeneralizedseizures.Seizur esmayoccurinclusters, 6 Thisimbalanceneednotnecessarilybeexcitatory.Forexam ple,ithasbeen suggestedthatenhancedinhibitionmayinfactcontributet otheepileptogenicprocess throughinhibition-inducedsynchrony[ DudekandSpitz 1997 ]. 23

PAGE 24

inintervals,orinisolation.TLEcanbedividedintotwotyp es:mesialTLE,which involvesthehippocampus,parahippocampalgyrus,and/ora mygdala;andlateralTLE, whichinvolvesthetemporallobeoftheneocortex.Asasympt omaticepilepsy,TLE resultsfromspecicbraininsults,suchastraumaticbrain injury,infection,stroke, andprolongedseizure(statusepilepticus),andemergesfo llowingalatencyperiod [ Aroniadou-Anderjaskaetal. 2008 ].AclassicalhallmarkofTLEistheinterictalspike [ Buzsakietal. 1991 Leung 1988 ],apathologicalpopulationdischargethatappears spontaneouslyinpathologicalEEG.Highfrequencyoscilla tions(HFOs)(80-500Hz)are alsoassociatedwithpathologicalactivityinTLE[ Braginetal. 1999 Stabaetal. 2002 ]. GiventheprevalenceofTLEandalsoitsrelativeintractabi lity,TLEwillbetheepilepsy syndromethatisthefocusofthisdissertation. 2.3HippocampalElectroencephalogram Duringthenaturalcourseofitsoperation,thebrainexhibi tsmanypatternsof activity.Inadditiontotheringofindividualneuronstha taccountsformostinformation processing,thecombinedactivityofmanyneuronsformsreg ularpatternsthatcan bedetectedandclassiedintoseveralspecicstates.Meas urementofthisactivity istypicallyperformedbyelectroencephalography(EEG).T hisinvolvesusingscalpor depthelectrodes 7 todetecttheelectriceldsproducedbyhundredsofsurroun ding neurons.PatternsofactivitypickedupbytheEEGaretypica llyassociatedwiththe behaviouralstateoftheorganism.Thesepatternscanbegro upedintotwocategories: 1)EEGrhythms,whicharerecurrentoscillationsintheEEGs ignalwithwell-dened frequencyrangesand2)EEGspikes,whicharehigh-amplitud etransientsintheEEG signalthatreectthesynchronousdischargeoflargepopul ationsofneurons.Belowwe willreviewthemajorpatternsofactivityspecictothehip pocampus. 7 Inourstudies,depthelectrodesareusedtorecordEEGfromw ithinthe hippocampus. 24

PAGE 25

2.3.1HippocampalGammaandThetaRhythms Withinthehippocampus,thetwomaintypesofrhythmsarehip pocampaltheta andgammarhythms.Thethetarhythm(5-10Hz)iscontinually presentduring statesofarousal,particularlyawakemobilityandexplora tion,andalsoduringREM sleep.Itisahighlypervasiveoscillation,encompassingt heentirehippocampusand extendingtootherconnectedstructures,includingtheent orhinalcortex,themedial septumandthesupramammillarynucleus[ Buzs aki 2002 ].Ithasbeenassociated primarilywithencodingandretrievalofmemory[ Hasselmo 2005 ]andalsospatial positioning[ O'KeefeandRecce 1993 ].Initiationofthethetarhythmiscontrolledby neuromodulatoryinputfromtheascendingarousalcentreso fthebrainstem,which projecttothehippocampusandseptalareas[ Vertesetal. 1993 ].Theseinputspromote thecircuitryofthehippocampusandconnectedstructurest oenterintothetafrequency oscillations.Theactualmechanismbywhichthesethetaosc illationsaregeneratedis thoughttodependcriticallyoninteractionbetweenthehip pocampusandthemedial septum,asstudieshaveshownthatlesioningtheseptumabol isheshippocampal theta[ Petscheetal. 1962 ]. 8 Inparticular,resonantinteractionsamongstGABAergic interneuronsinthehippocampusandmedialseptumhavebeen shownusingcomputer modelstoreproducetheta-frequencyoscillations[ Hajosetal. 2004 Wang 2002 ]. However,recent invitro studieshaveshownthatitisalsopossibletoreproducethet a oscillationsinhippocampalslicewithoutseptalinput,su ggestingthatthehippocampus alonepossessesthenecessarycircuitrytogeneratehippoc ampaltheta[ Goutagnyetal. 2009 ]. 8 Therearemultiplecelltypespresentinthemedialseptumth atcouldcontributeto hippocampalthetageneration.Thetwomaintypesarecholin ergicandGABAergiccells. Studieshaveshownthatselectivepharmacologicallesioni ngofseptalcholinergiccells reduces,butdoesnoteliminate,thethetarhythm.Thissugg estsseptalGABAergiccells aresufcienttomaintainthethetaoscillation,andthatse ptalcholinergiccellsregulate thethetarhythm'smagnitude[ Leeetal. 1994 ]. 25

PAGE 26

Gammarhythms(25-140Hz)arealsopresentinthehippocampu s.Unlikethe thetarhythm,whichissustainedthroughoutspecicbehavi ours,gammarhythms tendtooccurinshorttransientburstsofonlyseveralcycle s[ Buzsaki 2009 ].Gamma rhythmscanoccuratspecictimesinthethetacycleandalso outsideoftheta.They arebelievedtoplayaroleinbindingtogethercellassembli esforinformationprocessing. Twomainmechanismsareinvolvedwiththegenerationofhipp ocampalgamma oscillations:interneuronnetworkgamma(ING)andpyramid alinterneuronalnetwork gamma(PING).ForING,thegammaoscillationisgenerateden tirelybyinhibitory interactionsamongstinterneurons,ashasbeendemonstrat edininterneuron-only networkmodels[ WangandBuzsaki 1996 ].WhileINGisoftenstudiedin invitro preparations,itlikelydoesnotoccurinisolation invivo .Pyramidalcellinvolvementis demonstratedbyPING,inwhichpyramidalcellsprovidefast excitationtoinhibitory neuronstomodulatetheirexcitationandthereforecontrib utetotheirsynchronization. Detailsofthesemechanismsareprovidedbyanumberofrevie ws[ Buzs akiandWang 2012 Whittingtonetal. 2000 ]. 2.3.2HippocampalSharpWavesandInterictalSpikes Whenthehippocampusisnotinthethetastate,itcanexhibit severalothermodes ofactivity,thepredominantonebeinglargeamplitudeirre gularactivity(LIA)[ Ho 2011 ]. SuperimposedonthisLIAcanemergehippocampalpopulation dischargescalledsharp waves(SPWs).Theyappearasshort(30-120ms)highamplitud etransientsintheEEG signal[ Buzsaki 1986 ].SPWsarethoughttobeimportantformemoryconsolidation andthetransferofinformationbetweenthehippocampusand theneocortex[ Buzsaki 1996 ].Theyareformedbyrecurrentcircuitryofthehippocampal CA3regions,andit isthoughtthatperisomatic-targetinginterneuronsplaya centralroleintheirinitiation [ Ellenderetal. 2010 ].Interictalspikes(IS)aresimilartoSPWsandareassocia tedwith pathologicalbrainchangesseeninepilepsy.SPWsandISocc urduringthenon-theta state,whichincludesthebehaviourssuchasawakeimmobili tyandslow-wavesleep 26

PAGE 27

[ Kramisetal. 1975 ].Awakeimmobilityreferstoactivitiessuchasdrinking,e ating,and grooming[ SuzukiandSmith 1987 ].WhileSPWsoccurexclusivelyinnon-thetastates, IScanalsooccurinthethetastate,althoughtheirrateisse verelydepressed[ Buzsaki etal. 1991 Leung 1988 SuzukiandSmith 1987 ].Inthisdissertation,wewillreferto spikingeventssuchasSPWsandIScollectivelyashippocamp alEEGspikes(SPKs). 2.4OnSeizuresandCycles Epilepsyisfundamentallylinkedtobrainrhythms.Likebra inrhythms,epileptic seizuresareaproductofunderlyingneuralactivityandare dependentonthesynchronization oflargenumbersofneurons.Justasneuromodulatoryinputs associatedwithspecic behaviouralstatescontroltheemergenceofcertainrhythm s,sotoocanbehavioural statespromoteneuromodulatoryconditionsfavourablefor theemergenceofseizures. Forexample,anexperimentalstudyinratsshowedthattempo rallobeseizureswere reducedduringepisodesofthetaactivity,suggestingthat thethetastateisunfavourable fortheformationofseizures[ Colometal. 2006 ]. 9 Similarly,statechangesassociated withdifferentstagesofsleepcanalsoaffecttheemergence ofepilepticseizures[ Malow etal. 1997 ]. Thefocusofthisdissertation,however,isontherelations hipbetweenepilepsy andrhythmsthathavemuchlargertimescales,ontheorderof 24-hours.Priorstudies haveshownthatthereisactuallyabidirectionalrelations hipbetweenepilepticseizures and24-hourrhythms,withseizuresshowing24-hourpattern sofrecurrence,andalso epilepsyaffectingthe24-hourrhythmsofbiologicalproce sses.Theserelationshipsare reviewedinthefollowingsections. 9 Thisreductionwasobservedforthreedifferentcases:fors pontaneousepisodesof theta,forpharmacologicallyinduedtheta(carbachol),an dalsoforthetainducedbytail pinch.Thissuggeststhattheseizurereductionisdependen tonthepresenceoftheta state,regardlessofitsmechanismofinduction[ Colometal. 2006 ]. 27

PAGE 28

2.4.1Effectof24-HourRhythmsonSeizures Ithaslongbeenknownthat,althoughepilepticseizuresapp eartoberandom andspontaneousevents,theyactuallyshowpredictablepat ternswhenstudiedover verylongtimescales.Thiswasrstdocumentedinthelate19 thcenturybySirWilliam Gowers,whogroupedepileptic”ts”bytheirtendencytoocc urmostfrequentlyeither duringtheday,duringthenight,ordiffuselythroughoutbo ththedayandnight[ Gowers andSchlesinger 1885 ].Thiswasoneoftherstindicationsofaconnectionbetwee n epilepticseizuresanddailyrhythms.Sincethen,the24-ho urrhythmofseizureshas beenclassiedinavarietyofepilepsysyndromes,includin gTLE[ HofstraanddeWeerd 2009 Hofstraetal. 2010 ]. Therearemanydailyrhythmspresentinthebody.Theirorigi nandfunctioncan betracedtotheextrinsic24-hourlight-darkcycle,whichd eterminesoptimaltimesfor resting,feeding,andprocreating.Mostorganismsonthepl anetentraintothisdaily cycle.Mammalsshow24-hourrhythmsintheirgloballyregul atedfunctions,including digestiveactivity,wake-sleepstate,rest-activitycycl e,andcorebodytemperature [ BeersmaandGordijn 2007 ].Correspondingly,togeneratetheseglobalrhythms,basi c physiologicalprocesses,includingneuralactivity,meta bolicactivity,genetranscription, andhormonesecretion,mustalsoberegulatedona24-hourcy cle.Together,these rhythmsarecoordinatedtoalargeextentbythecircadianti mingsystem,ahierarchyof entrainable24-hourclocksthatoscillateevenintheabsen ceofexternalcues. Althoughsleepandcircadianrhythmsarecloselyrelated,t hereareimportant differencesbetweenthemaswell.Theyarecontrolledbydif ferent(althoughinterconnected) brainregionsandutilizedifferentneuromodulatorsandne urohormones.Whilethey areusuallycoupled,incasessuchasjetlag,theycanbecome misaligned. 10 The 10 Forceddesynchronyprotocolscanseparatetheendogenousc ircadianrhythm fromother24-hourrhythmsinthebody[ Cambrasetal. 2007 Carskadonetal. 1999 ]. 28

PAGE 29

wake-sleepcycleissometimesusedasasurrogatemarkerfor circadianrhythms,but itisoftendesirabletodifferentiatebetweenthetwo,espe ciallywhentryingtoelucidate underlyingmechanisms.Additionaldetailsonmechanismsu nderlyingcircadianand wake-sleepregulationareprovidedinChapter 3 Whilethereisstrongevidencethatwake-sleepstatecanin uenceepileptiform activity[ Malowetal. 1997 ],anumberoflinesofevidencesuggestthatthe24-hour patternofseizuresisdrivenprimarilybycircadianrhythm s.First,ithasbeenshown inananimalmodeloflimbicepilepsythatthe24-hourrhythm ofseizuresappearsfor bothanimalskeptinacontrolledlight-darkenvironment,a ndalsoforthosekeptin constantdarkness[ Quiggetal. 2000 ].Additionally,arecentclinicalstudyexamined ahumanpatientwithtwoseparateepilepticfoci,onelimbic andoneparietal.This investigationshowedthat,althoughseizuresfrombothfoc ishowed24-hourrhythms, theserhythmswereoutofphasewitheachother,withtempora lseizurespeakingnear noonandparietalseizurespeakingintheearlymorninghour s[ QuiggandStraume 2000 ].Finally,studieshavecomparedhumanmesialTLEtoasimil arepilepsymodel inrats.Itwasshownthatthe24-hourseizurerhythmsinrats andhumanswerein phase,bothpeakingintheearlyafternoon,despitethefact thathumansarediurnal (sleepingatnight)andratsarenocturnal[ Quiggetal. 1998 ].Althoughwake-sleepcycle isinvertedacrossnocturnalanddiurnalorganisms,24-hou rrhythmsofthecircadian systemarephase-conservedacrossthesespecies.Forexamp le,melatonin,oneofthe primaryhormonaloutputsofthecircadiansystem,issecret edatnightforbothnocturnal anddiurnalorganisms[ Quigg 2000 ].Thesendingssuggestthattheprimarydriver Anotherimportantdifferencebetweencircadianrhythmsan dthewake-sleepcycleis thatcircadianrhythmsaretruerhythms,whereasthewake-s leepcycleisaseriesof discretetransitionsbetweenwakefulnessandstagesofsle ep. 29

PAGE 30

ofseizure24-hourrhythmsforTLEmaynotbethewake-sleepc ycle,butratherthe circadiansystem. Despitetheevidencefortheinvolvementofthecircadiansy stemintheregulation of24-hourrhythms,littleisknownaboutmechanismsunderl yingthisregulation. Traditionallyitisthoughtthatthe24-hourseizurerhythm sresultfrommodulationofthe epilepticbrainregionbyneuromodulatorsand/orneurohor monesthatarereleasedwith 24-hourrhythmicity[ Quigg 2000 ].Thus,similartohowcertainbehaviouralstatesexhibit favourableconditionsforseizureformation,certaintime sofdaymayalsocontainlevels ofneuromodulatorsthatpromotetheepilepticbraintobemo stpronetoseize.Itiswell establishedthatthereleaseofmanyneuromodulatorsandne urohormonesisregulated ona24-hourbasis;additionally,manyofthesealsohavethe abilitytopromoteorinhibit epilepticseizures.Anexampleofsuchaneurohormoneismel atonin,whichisreleased primarilyatnight.Althoughitisaputativeanticonvuslan t,ithasalsobeenshownto promoteepileptiformactivityinslice[ MusshoffandSpeckmann 2003 ].Furthermore, asmentionedabove,melatoninisahormonethatisphase-con servedacrossnocturnal anddiurnalspecies.Othercompounds,suchasvasopressina ndhormonesofthe hypothalamic/pituitary/adrenalaxishavebeenproposedt oplayrolesinthe24-hour rhythmofseizures[ Quigg 2000 ].Morerecentstudieshavealsohypothesizedthat molecularmechanismsmightplayasimilarrole[ Choetal. 2012 ].However,asofyet, acausallinkagebetweentheseneuromodulators,neuralact ivity,andseizure24-hour rhythmshasnotbeendemonstrated.2.4.2EffectofEpilepsyon24-HourRhythms Afundamentalquestionincircadianresearchishowcertain diseasesaffect thecircadiansystemandimpairitsabilitytocoordinate24 -hourrhythms.Studies suggestthatsevereneurodegenerativedisorders,includi ngAlzheimer'sdisease, Huntington'sdisease,Parkinson'sdisease,andmultiples clerosis,affectbrainregions associatedwithsleepandcircadianrhythmsandthisresult sindisrupted24-hour 30

PAGE 31

rhythms[ BarnardandNolan 2008 Wulffetal. 2010 ].Interestinthisismotivatedby thefactthatsuchdisrupteddailyrhythmscanhaveextensiv epathologicalimplications, includingemotionaldisorders,cognitivedeciency,impa iredresistancetodisease,and increasedrisksofcardiovasculardiseaseandcancer[ Hastingsetal. 2003 Wulffetal. 2010 ].Studiesonshiftworkershaveshownincreasedriskofdiab etes,ulcers,cancer andcardiovasculardiseases,andchronicjetlaghasbeende monstratedtoproduce temporallobeatrophyandcognitivedefects[ Cho 2001 Choetal. 2000 ]. Evidenceisgrowingforthecasethatepilepsy,too,mightaf fectthecircadian system.Therearetwodistinctavenuesbywhich24-hourrhyt hmsmightbecome disturbedinepilepsy.First,theacuteeffectsofepilepti cseizuresmightpromote transient24-hourrhythmperturbations.Thisissupported byastudyshowingthat dailyrhythmsofcorebodytemperatureexperienceslightan dtransientphaseshifts followingseizures[ Quiggetal. 2001 ].Secondly,ithasbeenproposedthatpermanent structuralchangesinthebrainassociatedwiththechronic epilepsystatemayblock orimpaircircadianrhythms[ Quigg 2000 ].Thisissupportedbyanumberofstudies ongloballyregulatedfunctions,whichhaveshownabnormal itiesofthewake-sleep cycle[ Bastlundetal. 2005 Shouseetal. 1996 ],increasedvariabilityincorebody temperature(CBT)24-hourrhythms[ Quiggetal. 1999 ],andalteredmelatoninrelease [ HofstraanddeWeerd 2009 Quigg 2000 ].Onestudyalsoreportedaltered24-hour behaviouralrhythms,butsuggestedthatthesemightbecaus edbypostictalhyperactivity [ StewartandLeung 2003 ].Avarietyofbrainregionshavebeenproposedtounderlie thesechanges.Forexample,reportedcelllossinthedorsom edialhypothalamus,a regionrelevantforsleepregulation,hasbeensuggestedas thecauseforabnormalities inwake-sleepcycle[ Bastlundetal. 2005 ].Likewise,changesinthemedialpreoptic nucleusreportedby[ Quiggetal. 1999 ],mayunderlietheincreasedcomplexityin corebodytemperaturerhythms.Therearelikelyotherbrain structuresinvolvedinthe regulationofcircadianrhythmsmaybealteredintheepilep ticstateandthatremainto 31

PAGE 32

beidentied.Insummary,preliminaryevidencesupportsth enotionthatthecircadian systemisalteredinepilepsy;however,unlikethecaseforn eurodegenerativediseases, itisnotclearexactlywhichmechanismsandbrainregionsar einvolved.Thisresults partlyfromthelargenumberofepilepsysyndromes,andalso fromepilepsy'snatureof affectingawiderangeofbrainstructures[ Parekhetal. 2010 ]. 2.5Motivation Thusfar,wehaveoutlinedadualrelationshipbetweenepile psyand24-hour rhythms:rst,thepatternofrecurrenceofepilepticseizu resismodulatedona24-hour cycle;second,epilepsycanaffectthecircadianregulator ysystemtoaltertheexpression of24-hourrhythms.Whilethesephenomenahavebeencharact erizedbyanumber ofstudies,muchworkremainstobedonetoelucidatetheunde rlyingmechanisms. Aswehavediscussed,descriptionsarestilllargelyphenom enologicalandbasedon system-levelstudies;fewstudiesonthistopicdelveintop erforminglong-termneuralormolecular-levelmeasurementsduringepilepsy. 11 Suchmeasurements,however, willconstrainthesetofpotentialphenomenologicalmodel sandalsoyielddatathat canguidefuturequantitativemodelingefforts.Thisworkw illultimatelysuggestfuture experimentalstudiestoconrmspecicmechanismsforboth seizure24-hourrhythms andforthecircadiandisruptionseeninepilepsy. Fromamorepracticalperspective,abetterunderstandingo ftherelationship betweenepilepsyand24-hourrhythmswillbebenecialfora numberofreasons. First,byrevealinghowdailychangesinthebrainaffectsei zurelikelihood,itwillyield abetterunderstandingforthebrainnaturallyregulatesse izures.Thisknowledgewill thenbeusefulfordevelopingnewpharmacologicaltherapie stoinhibitepilepticseizures. 11 Anexceptiontothisisthestudyby Matzenetal. [ 2012 ],whichcharacterizes changesinexcitationandinhibitionthroughsingleandpai redpulseelectrophysiological measurements. 32

PAGE 33

Secondly,beingabletounderstandandpredictdailyseizur erhythmswillallowfor optimizationofexistingepilepsytherapies.Forexample, chronotherapies,suchas differentialdosingofantiepilepticdrugsaccordingtoth etime-of-day,optimizationof sleepschedules,improvementofsleephygiene,anddailymo dulationofelectrical treatmentssuchasvagalnervestimulationcouldallimprov eseizurecontrolandreduce treatmentsideeffects.Third,asmentionedabove,clinica lstudieshaveshownthat,in focalepilepsy,thetimeofpeakseizureoccurrenceisdepen dentonthelocationofthe seizurefocus.BymininglargeamountsofEEGdata,itmaybep ossibletooptimize epilepsydiagnosisbyincorporatinginformationonseizur e24-hourrhythms.Fourth, seizurepredictionstrategies,whichaimtogiveadvancedw arningofwhenepileptic seizureswilloccur,couldbenetfromincorporatingtimeo fdayinformation.Theseand otherchronotherapystrategiesarereviewedindetailby Loddenkemperetal. [ 2011 ]. Asmentionedabove,therearealsomanynegativesideeffect sanddiseaserisks associatedwithdisorganizedsleepandcircadianrhythms. Abetterunderstandingofthe waysinwhichtheserhythmsarealteredistherststeptowar dstreatingthesleepand circadiandisordersthatepilepticsoftenexperience. 2.6Hypotheses Inthisdissertation,wewillinvestigate24-hourrhythmso fhippocampalneural activityinaratmodelofTLE.Ourapproachwillconsistofbi ophysicalcomputer modeling,analysisoflargeneuraldatasets, 12 andMRIanalysis. Basedonthecurrentstateoftheliterature,weformulateth efollowinghypotheses: 1. Featuresofhippocampalneuralactivity,includingEEGrhy thmsandpopulation spikingevents,willshow24-hourrhythmicityintheiracti vity. 2. Theserhythmswillbedisruptedinepilepticanimals. 12 Long-termneuralrecordingswerecollectedby Talathietal. [ 2009 ]. 33

PAGE 34

3. Altered24-hourrhythmscanbelinkedtostructuraldamagei nthecircadian system. 4. StructuraldamagewillbedetectableusingMRIanalysis. Bybeginningwithlong-termneuralmeasurements,ourappro achwillcomplement existingsystems-levelstudies,elucidatinghow24-hourr hythmsarealteredonaneural level.Shouldhypothesis#2beshowntrue,and24-hourneura lrhythmsbecome disruptedinepilepsy,itispossiblethattheassociatedch angeinhomeostasiscould inuencetheemergenceofseizures.Inthiscase,circadian disruptionandseizure 24-hourrhythms,twoaspectsofepilepsythatarenormallys tudiedindependently, mayinfactberelated.Suchandingcouldshapetheparadigm offuturestudies, shiftingawayfromtheideathatseizure24-hourrhythmsare passivelyentrainedto normalcircadianrhythms,andtowardstheideathatcircadi anabnormalitiesencourage theemergenceofseizures.Moregenerally,ourneural-leve lapproachwillleadto thedevelopmentofmoremechanistictheoriesfortherelati onshipbetweencircadian rhythmsandepilepsyand,hopefully,haveimpactsrangingf romthedevelopmentof chronotherapiestoimprovedfundamentalunderstandingof thedisease. 2.7OrganizationofthisDissertation Thisdissertationaimstoprovideacompleteinvestigation of24-hourrhythms inspontaneousneuralactivity,derivedfrompreviouslyco llectedlong-termneural recordings.Followingouranalysisofthisdata,wewillpro posespecicmechanismsto explainourresultsandtestthesebydevelopingabiophysic alcomputermodel.Thiswill leadtofurtherhypotheses,whichwillbetestedthroughsub sequentEEGandMRIdata analysis. Priortopresentingtheseresults,however,someadditiona lbackgroundonthe circadiansystemisrequired.ThisispresentedinChapter 3 .Chapter 4 outlines thegeneralmethodsusedinthisstudythatarerelevanttosu bsequentchapters. Chapters 5 8 addressourspecichypothesesstatedabove.Resultsofour analysis 34

PAGE 35

ofneuraldata,withspecicfocusonEEGspikingevents,are presentedinChapter 5 .Inthischapter,wedemonstratethat,forepilepticanimal s,aphaseshiftemerges inthe24-hourrhythmofEEGspikingevents(Hypotheses#1an d#2).Basedonthis, weproposethatcertainbrainregionsresponsibleforprovi dingcircadianregulation becomedysfunctional.WetestthisproposedmechanisminCh apter 6 bydesigninga neuralnetworkmodelof24-hourrhythmsandusingittodemon stratehowstructural changes,inparticulartothemedialseptum,canreproduceo bservedcircadian anomalies(Hypothesis#3).Thisleadstoinvestigationsof structuralchangesinthe medialseptuminChapter 7 andoffunctionalchangesinthethetarhythminChapter 8 Specically,Chapter 7 uses invivo andexcisedMRItomeasurechangesintheseptum inepilepticanimals(Hypothesis#4).Chapter 8 returnstoHypotheses#1and#2by showingthattheamplitudesofEEGrhythmsaremodulatedona 24-hourtimescale and,furthermore,themodulationofcertainrhythms(betaa ndgamma)experiencea phaseshiftfollowinginjury.Chapter 8 alsoprovidesevidencethattherearemultiple circadianinputsthatmaybealteredfollowinginjury(Hypo thesis#3).Finally,wediscuss implicationsofthecollectivendingspresentedinthiswo rkinChapter 9 ,specicallyin relationtoseizure24-hourrhythms.Thematerialpresente dinChapters 5 6 ,and 7 is derivedfromourinpressmanuscripttitled“Localphaseshi ftinthe24-hourrhythmof hippocampalEEGspikingactivityinaratmodeloftemporall obeepilepsy.” 35

PAGE 36

CHAPTER3 THEMAMMALIANCIRCADIANSYSTEM 3.1Overview Thereprimarygoalofthischapteristoprovidereaderswith thenecessaryand sufcientbackgroundinformationaboutthecircadianregu latorysysteminorderto interpretresultspresentedinsubsequentchaptersofthis dissertation. 1 Wewillreview whatisknownaboutthestructuralorganizationofthecirca diansystemandfocus specicallyonhowthiscaninuenceneuralactivityintheb rain. 3.2AHierarchyofOscillators Circadianrhythmsare24-houroscillationsthatareubiqui tousinbiologicalsystems, rangingfrombacteriatothehumanbrain.Coordinationofci rcadianrhythmsisachieved bythecircadiantimingsystem,whichconsistsofthreegene ralcomponents:aninput, apacemaker,andanoutput[ Loddenkemperetal. 2011 Ohdo 2010 ].Inmammals, theprimarycircadianpacemakerislocatedinthesuprachia smaticnucleus(SCN) [ GerstnerandYin 2010 ].WeshalldiscusstheSCNandalsotheothercomponentsof thecircadiansysteminthefollowingparagraphs. Thereareseveralimportantinputstothecircadiansystem. Aprimaryinput arisesfromthephotoreceptorsintheretina,whichtransmi tinformationaboutthe environmentallight-darkcycle.Thesephotoreceptorsinc ludebothconventionalrods andconesandalsoretinalganglioncells(RGCs),whichserv easthenon-imageforming photoreceptorsforthebrain.TheyprojectdirectlytotheS CNandalsotomanyother brainstructures.Theyrespondprimarilytobluelight,aso pposedtogreenorviolet light[ Vandewalleetal. 2007 ].RGCs,rods,andconestraversetheretinohypothalamic 1 PartsofthischapterareinpresswiththeJournalofNeuroph ysiology:Stanley,D.A., Talathi,S.S.,Parekh,M.B.,Cordiner,D.,Zhou,J.,Mareci ,T.H.,Ditto,W.L.,Carney,P.R. Localphaseshiftinthe24-hourrhythmofhippocampalEEGsp ikingactivityinarat modeloftemporallobeepilepsy[ Stanleyetal. 2013 ]. 36

PAGE 37

tracttoreachtheSCN,wheretheirinputactstoentrainSCNc ellstotheenvironmental light-darkcues.AdditionalimportantinputstotheSCNari sefromtheintergeniculate leaet(IGL),fromtheraphenuclei,andfromseveralothersi tesinthebrainthattransmit non-photicinformationtotheSCNaboutfeedingandtempera turepatterns[ Ohdo 2010 ]. TheSCNconsistsofapproximately20,000neurons.Inslice, intactSCNtissue hasbeenshowntobeintrinsicallyoscillatory,exhibiting sustained24-hourrhythmsof neuralringactivityinisolationofotherinputs.These24 -houroscillationsarenotan emergentpropertyofthenetwork,asisolatedSCNcellsincu lturehavealsoshown sustained24-hourrhythms[ Hastingsetal. 2003 ].Rather,SCNringactivityoscillations arethoughttobegeneratedby24-houroscillationsingenet ranscription.Clockgenes, including Per Cry ,and Rev-erb ,havebeenshowntoparticipateinafeedbackcircuit thatoscillateswithnear-24-hourperiod[ KoandTakahashi 2006 Shearmanetal. 2000 ].ThismolecularclockmachineryisnotpresentinallSCNce lls,butonlyincertain “clock”cells.Thesecellspossess24-hourrhythmsofring activity,andalsoexhibit uniqueelectrophysiologicalproperties.Thedailychange sintheirelectricalbehaviour maybelinkedtomolecular24-houroscillationsthroughmod ulationofspecicpotassium currents[ Belleetal. 2009 ]. WhiletheSCNservesasthemasterclockofthecircadiansyst em,itdoesnot exhibitagreatmanyefferentprojections.Rather,thecirc adiansystemreliesona numberofintermediarynucleithatareorganizedintoahier archical-typenetwork(Figure 3-1 )[ GerstnerandYin 2010 ]. 2 ThesenucleireceivetheSCN'sclocksignal,integrate itwithotherinformation,andthentransmititthroughoutt hebody.Astandardexample 2 Thishierarchicalnetworkincludesbothfeedforwardandfe edbackconnections. Additionally,anumberofrecentstudieshaveidentiedind ependentperipheralcircadian oscillatorsthatformpartofthecircadiannetworkandthat ,liketheSCN,maintain 24-houroscillationsinisolation[ GuildingandPiggins 2007 KoandTakahashi 2006 ]. 37

PAGE 38

isthecircuitthatcontrolsthereleaseofmelatonin.While melatoninisaprincipaloutput ofthecircadiansystem,itisproducedbythepinealglandan ditsreleaseisindirectly controlledbytheSCN.Specically,SCNreliesonapolysyna pticpathwayviathe paraventricularnucleus(PVN)andthesuperiorcervicalga nglion(SCG)tonallyreach thepinealgland[ Richteretal. 2004 ].Melatoninthenisreleasedintothebloodstream toexerteffectsthroughoutthebodyandalsotoexertafeedb ackeffectontheSCNitself [ Saperetal. 2005 ]. 3.3CircadianRegulationoftheWake-SleepCycle Anotherimportantfunctionofthecircadiansystemisthere gulationofthe wake-sleepcycle.Thecircuitcontrollingsleepandwakefu lnessprovidesanimportant exampleofhowtheSCNsignalisintegratedwithotherinform ation.Specically,the wake-sleepcircuitconsistsofcontrastingbraincentresp romotingarousalandsleep, respectively.Keyarousalareasaremonoaminergicnuclei, suchasthelocuscoeruleus (LC),theraphenuclei,andthetuberomammillarynucleus(T MN);theseareactivated inpartbyorexincontainingcells(ORX)ofthelateralhypot halamicarea(LHA).The mainsleep-promotingregionistheventrolateralpre-opti cnucleus(VLPO).Adetailed reviewthisarchitectureisprovidedby Saperetal. [ 2005 ]but,essentially,thesetwo regionsaremutuallyinhibitoryandfunctionasaip-opsw itch.Manyofthesenuclei, suchastheVLPOandtheORXcellsreceiveinputfromtheSCN[ Abrahamsonetal. 2001 Deurveilheretal. 2005 Loddenkemperetal. 2011 ].However,whiletheSCN exertsaninuence,theswitchdoesnotstrictlyfollowtheS CN's24-hourrhythm.Rather, wake-sleeptransitionsrespondtomanyadditionalfactors ,includinghomeostaticdrive (extendedperiodsofsleepwillfollowdeprivation)andals oexternalstimulation.Inthis manner,thewake-sleepcircuitintegratestheSCNclocksig nalwithmanyotherinputs, allowingittoresponddynamicallytoabreadthofdemands. 38

PAGE 39

Figure3-1.Asummaryofthehierarchicalarrangementofthe mammaliancircadian system.TheSCNdoesnotdirectlycontrol24-hourrhythmsin thebody,but actsthroughanumberofintermediarynucleithattransmitt heclocksignal whilstintegratingitwithotherinformation.Themajority ofSCNoutput traversestheventralanddorsalsubparaventricularzone( vSPZanddSPZ). dSPZneuronsprojecttothemedialpreopticnucleus(MPO)to regulate 24-hourrhythmsofbodytemperature.vSPZneuronspasstoth e dorsomedialnucleusofthehypothalamus(DMH).TheDMHthen projectsto theventrolateralhypothalamicnucleus(VLPO),thelatera lhypothalamicarea (LHA),andtheparaventricularnucleusofthehypothalamus (PVH)to inuence,respectively,sleep,arousal,andreleaseofcor ticosteroidsand melatonin.Anumberofadditionalnucleiandcircadian-reg ulatedfunctions, suchasfeedingandthyroidrelease,arenotshown,butarere viewed elsewhere[ GerstnerandYin 2010 Loddenkemperetal. 2011 ]. Reproducedwithpermissionfrom Saperetal. [ 2005 ]. 39

PAGE 40

3.424-HourRhythmsintheBrain Thehierarchicalstructureofthecircadiansystemallowsf ortheregulationof 24-hourrhythmspervasivelythroughoutthebody.Ithasrec entlybeenestimatedthat morethan10%ofgenesaresubjecttocircadianregulationin mammalianperipheral tissue[ Storchetal. 2002 ].Inthebrain,circadianregulationishighlypervasiveas well,and24-hourrhythmsofneuralactivitycanbedetected atmostrecordingsites usingavarietyofrecordingtechniques.Asummaryofpublic ationsdescribingbrain 24-hourrhythmsispresentedTable 3-1 .Inthistable,wedidnotincludedataonthe hippocampus,whichisinsteadpresentedinthenextsection 3.524-HourRhythmsintheHippocampus Investigationsontherelationshipbetween24-hourrhythm sandepilepsyinthis dissertationwillfocusspecicallyontemporallobeepile psy(TLE)inthehippocampus. Thismakesthehippocampusanimportantstructureuponwhic htofocusinourreview ofthecircadianliterature.Fortunately,asanimportants itefortheformationoflearning andmemory,thehippocampushasbeenofinterestfornumerou scircadianstudies. Thisinterestarisesmainlyfromtheobservationthatmanyb ehaviours,particularlythose involvingmemoryandcognitiveperformance,showtime-ofdaydependency[ Barnes etal. 1977 GerstnerandYin 2010 ].Wehavepresenteddetailedsummaryofstudies onhippocampal24-hourrhythmsinTable 3-2 .Itisimportanttonotethat,whilethereare numerousstudieson24-hourrhythmsinthehippocampus,the reisapaucityofstudies thathaveexaminedtheserhythmsinthediseasedstate. 3.6CircadianInputstotheHippocampus FromTable 3-2 ,itisclearthatmultiplemeasuresofhippocampalneuralac tivity show24-hourrhythms,includingbothspontaneousandevoke dactivity.Evidence fromasubsetofthesestudies[ Chaudhuryetal. 2005 Wangetal. 2009 Westand Deadwyler 1980 ]suggeststhattheserhythmsare,inpart,trulycircadianthatis, 40

PAGE 41

Table3-1.Brainregionsexhibiting24-hourrhythms invivo BrainRegionMeasurement a SpeciesReference AccumbensnucleusMUAHamster Yamazakietal. [ 1998 ] AnteriorHypothalamusHist.Rats Mochizukietal. [ 1992 ] b CaudatenucleusMUARat InouyeandKawamura [ 1979 ] 5-HT,Hist.,NERat FriedmanandWalker [ 1968 1969 ] CaudateputamenMUAHamster Yamazakietal. [ 1998 ] LateralhypothalamicareaMUARat Inouye [ 1983 ] LateralseptumMUAHamster Yamazakietal. [ 1998 ] LocuscoeruleusSUARat Aston-Jonesetal. [ 2001 ] 5-HT,NERat Agrenetal. [ 1986 ] MedialanddorsalrapheMUARat InouyeandKawamura [ 1979 ] 5-HTRat Pinatoetal. [ 2004 ] c 5-HT,NERat Agrenetal. [ 1986 ] MedialpreopticregionMUAHamster Yamazakietal. [ 1998 ] MedialseptumMUAHamster Yamazakietal. [ 1998 ] Mid-brain5-HT,Hist.,NERat FriedmanandWalker [ 1968 1969 ] MidbrainreticularformationMUARat InouyeandKawamura [ 1979 ] PreopticareaMUARat Inouye [ 1983 ] ReticularformationMUARat Inouye [ 1983 ] StriamedullarisMUAHamster Yamazakietal. [ 1998 ] SubstantiaNigraMUARat Inouye [ 1983 ] SuprachiasmaticnucleusMUARat InouyeandKawamura [ 1979 ] MUARat Inouye [ 1983 ] MUAHamster Yamazakietal. [ 1998 ] ThalamusMUARat Inouye [ 1983 ] VentrolateralthalamicnucleusMUAHamster Yamazakietal. [ 1998 ] VentromedialhypothalamicnucleusMUARat Inouye [ 1983 ] Hamster Yamazakietal. [ 1998 ] VisualcortexMUARat InouyeandKawamura [ 1979 ] a MUA=multiunitactivity,SUA=singleunitactivity,5-HT=s erotonin,NE=norepinephrine,Hist.=histamine b Histaminergicneuronsinthebrainarelocatedinthetubero mmaillarynucleusoftheposteriorhypothalamus.However, theanteriorhypothalamicareacontainsthehighestconcen trationofhistaminergicbres. c Variationwasbimodal.41

PAGE 42

independentofbehaviouralorlight-darkcyclechanges.Th israisesafundamental question:howdotheserhythmsarise? WhiletherearenodirectprojectionsfromtheSCNandthehip pocampus,there arenumerousindirectpathwaysbywhichcircadiansignalsm ayberelayedtothe hippocampus.Unfortunately,however,thereislittleevid enceastowhichofthese pathwaysaremostrelevant.Wehaveconductedadetailedlit eraturesearchofthese potentialpathwaysandhavegroupedthemintothreecategor ies.First,thehippocampus receivesafferentsynapticprojectionsfrommanybrainreg ionsandmostofthese regionsshow24-hourrhythms(Table 3-1 ).Thus,thesebrainregionscouldpotentially berelevantfordrivinghippocampal24-hourrhythms.Major hippocampalafferents thatshow24-hourrhythmicityaresummarizedinTable 3-3 3 .Second,neuronsinthe hippocampushavereceptorsformanyneurohormonesthatare modulatedona24-hour cycle,suchasmelatoninandadenosine[ Liuetal. 2000 Wanetal. 1999 ].These couldalsoinuencehippocampalneuralactivity;theactio nofsuchneurohormonesis summarizedinTable 3-4 .Third,anumberofrecentstudieshaveshownthatmolecular clockspresentinhippocampalneuronsmightinuencesynap ticactivity[ Valnegri etal. 2011 Wangetal. 2009 ].Takentogether,itisclearthattherearemanypossible pathwaysbywhich24-hourrhythmscanpropagatetothehippo campus.However,itis notclearwhichofthesepathwaysare functionally relevantforgeneratinghippocampal 24-hourrhythms.Inordertoanswersuchaquestion,additio nalexperimentalstudies, specicallythosequantifyingtheeffectsoflesioningput ativecircadianpathways,are needed.Unfortunately,althoughsuchstudieshavebeencon ductedfortheprimary 3 Themajorityoftheseafferentsarriveinthehippocampusvi atworoutes,onedorsal, oneventral.Thedorsalroutetraversesthembria-fornix, cingulum,whereastheventral routetraversestheamygdala[ DaSilvaetal. 1990 Mongeauetal. 1997 ] 42

PAGE 43

outputsoftheSCN[ InouyeandKawamura 1979 ],verylittleisknownaboutthe functionalcircadianinputstothehippocampus. Despitethepaucityofdirectevidencethroughlesioningex periments,there existsomestudiesthatprovideindirectindicationsofwhi chpathwaysmightbemost relevantforgeneratinghippocampal24-hourrhythms.Assu ggestedbyTables 3-1 and 3-3 ,theseptummaybeanimportantrelaycentrebetweentheSCNa ndthe hippocampus.Multiunitrecordingshaveshownthatthesept umpossessesarobust circadianrhythmthatpersistsintheabsenceofexternalli ght/darkcues[ Yamazakietal. 1998 ].Furthermore,itiswellestablishedfromtracingstudies thattheseptum,unlike thehippocampus,receivesheavyinnervationfromtheSCN[ Morinetal. 1994 ].Inturn, themedialseptumisperhapsthemostcriticalsubcorticali nputtothehippocampus. Arecentstudyonhippocampaldependentlearningsuggested thattheseptummay playafunctionalroleinregulatinghippocampalcircadian rhythms[ Rubyetal. 2008 ]. Specically,theauthorsshowedthathamsterswitharrhyth micSCNdisplayedimpaired hippocampallearning.Theyproposedthatthismayresultfr omchronicinhibitionof theseptumbythearrhythmicSCN,sincepharmacologicalinh ibitionoftheseptumhad similareffectsonlearning.Suchchronicinhibitionofthe septumcouldproducethese effectsthroughsuppressionofseptalexcitationofthehip pocampus. Anotherstudyinvestigatedtheeffectsofmelatoninonsyna ptictransmissionand LTPinthehippocampus,comparingcontrolmicetomelatonin knockouts[ Chaudhury etal. 2005 ].Theyfoundthatmelatoninknockoutaltered,butdidnotel iminate 24-hourrhythmsofsynaptictransmission.Thissuggeststh at24-hourrhythmsinthe hippocampusaregeneratedbytheoverlappinginuenceofmu ltiplecircadianinputs,of whichmelatoninisone. Astudyonhippocampalplacecellactivityhasshownthat,al thoughhippocampal placecellringexhibits24-hourpatterns,thesepatterns arenotphase-lockedtothe day-nightcycle.Rather,itwasproposedthattheyareentra inedtootherexternalcues, 43

PAGE 44

suchasfoodavailability.Therefore,itwasproposedthata foodentrainable24-hour oscillatoroutsidetheSCNmightalsobearelevantcircadia ninputtothehippocampus [ MunnandBilkey 2012 ].Finally,anumberofotherfunctionallyrelevantpathway shave beenproposed,suchashypothalamic-regulatedhormonerel ease[ Matzenetal. 2012 ] andindirectinputfromtheorexin-secretingcellsinthela teralhypothalamus[ Gerstner andYin 2010 ].However,atpresent,thereislittledirectevidencefort heirfunctional relevance. Insummary,wehaveprovidedareviewoftheliteratureconce rning24-hour rhythmsinthehippocampus.Wehavealsoconductedadetaile dliteraturesurveyon thepossibleinputpathwaysthatcouldleadtothegeneratio nofthese24-hourrhythms. Althoughthereislittleevidenceforwhichofthesepathway sarestrongestandmost functionallyrelevant,preliminarystudieshavesuggeste dthatbothmelatoninandinput fromthemedialseptummayplayimportantrolesingeneratio nofhippocampal24-hour rhythms.Thenotionthatmultiplecircadianinputsmaycont ributetothegeneration ofhippocampal24-hourrhythmsisconsistentwiththerequi rementthatthecircadian hierarchymustdynamicallyregulate24-hourrhythmstores pondtoavarietyofexternal cues. 44

PAGE 45

Table3-2.Experimentalmeasurementofhippocampal24-hou rrhythms StudyStimulationRecordingSpeciesQuantity a Comments b sitesiteMeasured Barnesetal. [ 1977 ] PerforantpathDentategranule cells Rat,Squirrelmonkey invivo fEPSP,PSamplitude PeakfEPSPoccurredoutofphaseforrat(nocturnal)andmonkey(diurnal) Westand Deadwyler [ 1980 ] PerforantpathDentategranule layer Rat invivo fEPSP,PSamplitude CircadianrhythmofPS,peakingatnight,independentofbehaviour,light,orcorticosterone.NodifferencesinfEPSP. Caulleretal. [ 1985 ] Perforantpath(angularbundle) DentatehilusRat invivo fEPSPslope, PSamplitude fEPSPpeaksduringnight,PSduringday Harrisand Teyler [ 1983 ] Dentatestratummoleculare Dentategranulecells RatsliceLTPofPSamp.LTPpeaksinnightindentate CA1stratumradiatum CA1stratumpyramidale LTPpeaksinnightinCA1 Danaand Martinez [ 1984 ] PerforantpathDentatehilusRat invivo LTPofPSamp.LTPpeaksatnightin controls,peaksduringdayinadrenalectomizedrats Bruneland deMontigny [ 1987 ] Nostim.CA3pyramidal cells Rat invivo Singlecellringrate(spontaneous) Peakringrateinthemorning,withseasonaldependence Raghavanetal. [ 1999 ] CA3stratumradiatum CA1pyramidalSyrianhamster slice LTPofPSaftertetanus SlicespreparedduringdayandtestedatnighthavehighestLTP a fEPSP=eldexcitatorypostsynapticpotential(evoked),P S=populationspike(evoked),VC=voltageclamp, LTP=long-termpotentiation,HVA=high-voltage-activate d,s/mAHP=slow/mediumafterhyperpolarization,ADP= afterdepolarization. b Foreachstudy,24-hourvariationinmeasuredquantitiesim pliedunlessotherwisestated.45

PAGE 46

Table 3-2 .Continued StudyStimulationRecordingSpeciesQuantityComments sitesiteMeasured Koleetal. [ 2001 ] CA3pyramidalcells CA3pyramidalcells RatsliceHVACa 2+ currents,mAHP,sAHP,ADPduringVC Ca 2+ currentsgreatestafter onsetofdarkphase.mAHPreducedandADPincreasedatnight;nochangesAHP.Spikefrequencyaccommodationreducedatnight. Chaudhury etal. [ 2005 ] CA1stratumradiatum CA1pyramidallayer MousesliceLTPoffEPSP slope LTPdecaysmoreslowlyduringnight CA1dendriticlayer LTPofPSslope LTPofPSgreateratnight CA1pyramidallayer PSamplitudeNodifferenceforcontrol mice,butgreateratnightformelatoninknockout 46

PAGE 47

Table3-3.Synapticinputstothehippocampusfrombrainreg ionsshowing24-hour modulation BrainregionNeurotransmitterReference HypothalamusMultiple PasquierandReinoso-Suarez [ 1978 ] LocuscoeruleusNorepinephrine Gageetal. [ 1983 ], HaringandDavis [ 1985 ], PasquierandReinoso-Suarez [ 1978 ] MedialseptumAcetylcholine, GABA Gageetal. [ 1983 ], Petersonetal. [ 1987b ] RaphenucleiSerotonin AzmitiaandSegal [ 1978 ], Pasquierand Reinoso-Suarez [ 1978 ] SubstantianigraDopamine Scattonetal. [ 1980 ] Tuberomammillarynuclei Histamine Panulaetal. [ 1989 ] Ventraltegmentalarea Dopamine Scattonetal. [ 1980 ] Table3-4.Non-synapticinputstothehippocampusshowing2 4-hourmodulation ModulatorSpecies preparation PeakRelease EffectReference AdenosineRatsliceNightInhibitory(inhib.of glutamaterelease) Liuetal. [ 2000 ] BDNFRatsliceNodataBDNFpotentiates synapticresponse KangandSchuman [ 1995 ] BDNFRatsliceNight24-hourrhythm, peakingatnight,indentateandCA3.NochangeinCA1 Schaafetal. [ 2000 ] MelatoninRatsliceNightAttenuates GABAergiccurrents Wanetal. [ 1999 ] MelatoninRatsliceNight1 Mmel.increases neuronringrate Musshoffetal. [ 2002 ] MelatoninRatsliceNight1 Mmel.increases epileptiformactivity Musshoffand Speckmann [ 2003 ] MelatoninRatsliceNightMelatonin-associated reductioninLTP Ozcanetal. [ 2006 ] 47

PAGE 48

CHAPTER4 GENERALMETHODS 4.1Overview Thischapterdescribestheexperimentalmethodsthatarege neraltoChapters 5 7 and 8 1 Thisincludesacompletedescriptionofhowthethreemainex perimental datasetsusedinthisdissertationwerecollected.Therst istheEvolutionintoEpilepsy EEGdataset,whichwascollectedovermultipleyearsbyDr.D ong-UkHwang,Dr. SachinS.Talathi,Dr.JunliZhou,andothers.Theseconddat asetisthecorebody temperature(CBT)data,collectedbyDanielCordiner.Thet hirdistheMRIdata, collectedbyMansiB.Parekh.Additionally,thischapterwi lldescribegeneraldata analysismethodsthatareusedinChapters 5 and 8 4.2AnimalHusbandry Animalstudieswereconductedon2-month-oldmaleSpragueD awleyratsweighing 200-265g.ProtocolswereapprovedbytheUniversityofFlor idasInstitutionalAnimal CareandUseCommittee(IACUCprotocolnumberD710).Allani malswerehousedin a24hsymmetriclight-dark(LD)environmentwithlightstag ecenteredat12:00noon andwithconstanttemperatureandhumiditylevels.Animals wereprovidedwithfood, water,andcleaningatregularintervalsandmonitoredwith continuoustime-locked video.Injurywasinducedusingawell-characterizedmodel forchronictemporallobe epilepsy(TLE)[ Lothmanetal. 1989 1990 Sanchezetal. 2006 ],inwhichelectrical inductionofSEisusedtobringaboutastateofrecurrentspo ntaneousseizuresafter alatencyperiodof2–4weeks.Animalsweremonitoredduring pre-injury,latency,and spontaneouslyseizingstagesofepileptogenesis.Thelate ncyperiodwasincludedin 1 PartsofthischapterareinpresswiththeJournalofNeuroph ysiology:Stanley,D.A., Talathi,S.S.,Parekh,M.B.,Cordiner,D.,Zhou,J.,Mareci ,T.H.,Ditto,W.L.,Carney,P.R. Localphaseshiftinthe24-hourrhythmofhippocampalEEGsp ikingactivityinarat modeloftemporallobeepilepsy[ Stanleyetal. 2013 ]. 48

PAGE 49

thisanalysisbecausespontaneousseizureshavebeenprevi ouslyshowntopromote transientperturbationstocircadianrhythms[ Quiggetal. 2001 ].Thebeginningof thespontaneouslyseizingstagewasmarkedbytherstrecor dedgrade3orgreater seizure. 4.3Long-TermDataCollection EEGandCBTDatawerecontinuouslycollectedfromanimalsth roughout epileptogenesis.Duetotheexperimentaldemandsofcontin uouslongitudinalrecording, itwasnecessarytousealownumberofanimalsforEEGandCBTr hythmanalysis. However,eachanimalprovidedmanycircadiancyclesofdata ( > 3weeksperanimal), whichallowedforclearreconstructionof24-hourrhythmsa ndhighlycondent estimationsofphase.Theexperimentaltimelineissupplie dinFigure 4-1 Surgical,electrophysiological,anddataanalysistechni quesaredescribedbelow. Figure4-1.ExperimentaltimelineforEEGandcorebodytemp erature(CBT)recordings. ForEEG,dataweresplitintopre-injury(Pre),post-injury latency(Post-L) andpost-injuryspontaneouslyseizing(Post-SS)stages.T hebeginningof thePost-SSstagewasdenedasfollowingtherstspontaneo usgrade3or greaterseizure.SinceCBTrecordingsdidnotincludesimul taneousEEG duetointerferencefromtheCBTtransponder,wedidnotatte mpttosplit post-injurydataintolatencyandspontaneouslyseizingst ages. 4.4SurgicalMethods Priortoallsurgeries,ratswereanesthetizedbysubcutane ousinjectionof10mg/kg xylazine(WebsterVeterinary)andmaintainedanesthetize dby1.5%isourane(Akron Inc.).Allratswerestereotaxicallyimplantedintheright ventralhippocampus(5.3mm caudaltoBregma,4.9mmlateral(right)ofBregma,and5mmve ntralfromdura)with aTeon-coatedbipolartwistedelectrode(330mdiameter), whichwaslaterusedfor 49

PAGE 50

inductionofSE.Additionally,forcollectionofEEGdata,s ixteenchannelOmnetics basedmicrowirerecordingelectrodearrays(TuckerDavisT echnologies,Alachua, FL)wereimplantedbilaterallyintothedentategyrusandCA 1regions(Figure 4-2 ). Electrodesweremadeoftungstenwithpolyimideinsulation ,had2mmtiplength,and were50mindiameter.Themicrowirearraywasanchoredtothe skullwithfour0.8 mmstainlesssteelscrews.TwowereAP2mmtothebregmaandbi lateral2mm, andtwowereAP-2mmtothelambdoidalsutureandbilateral2m m.Thesetwo pairsservedasthegroundandreference,respectively.Ele ctrodeplacementwas veriedusingMRIafterbrainshadbeenexcised(seebelow). ForCBTrecording,a radio-frequencyTAE-Mittertransponderwasinsertedinth eabdominalcavityand poweredbyaER4000Energizer/Receiverunit(bothfromMini MitterCo.,Bend,OR) thatwasplacedunderneaththecages. 4.5InductionofSelf-SustainingStatusEpilepticus ToinduceSE,10-secondpulsetrainswereappliedtothebipo lartwistedstimulating electrodesusingaModel2100Stimulator(A-MSystems,Sequ im,WA),consisting ofbiphasicsquarewaveswithfrequency50Hz,pulseduratio n1ms,andamplitude 250-400A,with2-secondintervalsbetweentrains[ Lothmanetal. 1989 1990 Sanchez etal. 2006 ].Theprotocolforvaryingthestimuluscurrentamplitudew asasfollows: Theinitialstimuluscurrentwas50A.Thiswasincreasedin2 5Aincrementsuntileither threeconsecutivegrade5seizureswereobservedoruntilth eupperlimitstimulus currentof600Awasreached.Thestimulationappliedtoalls pontaneouslyseizing animalsfellwithintherangeof250-400A.Animalsdemonstr atedwetdogshakes andseizuresthroughoutthedurationofthestimulationpro cedure,whichlastedfor 60-90mins.Uponterminationofstimulation,EEGrecording sdocumentedintermittent, self-sustaining30-60sseizuresandinterictal2-5HzEEGa ctivitylastingfor24hours. Followingalatencyperiodof2-4weeks,ratsdevelopedspon taneousrecurrentlimbic seizures.Recordingsofthelatencyperiodbegan2daysfoll owingSEstimulus.ForEEG 50

PAGE 51

Figure4-2.Placementofstimulatingandrecordingelectro des.Channelnumber2, correspondingtotheipsilateralCA1region,wasusedforEE Ganalysisin thisstudy. recordings,atotalofN=6animalswerestimulated.Thesewe redividedintotwogroups: thosethatexhibitedspontaneousseizures(N=3)andthoset hatdidnotexhibitseizure activityforatleast4weeksfollowingstimulus(N=3).ForC BTandMRIanalysis,only ratsthatdevelopedspontaneousseizureswereanalyzed. 4.6SeizureDetection Anin-houseseizuredetectionsystem[ Talathietal. 2008 ]wasusedtoscanEEG datasetstoidentifypotentialseizureepochs.Thesedatas etswerethenvisuallyscored byanexpertepileptologist(PRC).TheRacinegradesofthes eizureswereconrmed fromvideorecordings[ Racine 1972 ].Seizureinformationwasusedtoassessthe 51

PAGE 52

beginningofthespontaneouslyseizingstageofepileptoge nesis,whichwasmarkedby therstrecordedgrade3orgreaterseizure. 4.7EEGandCBTDataAcquisition TheEEGdatafromtherecordingelectrodeswaschanneledthr ougha16-channel commutatortoapairof16channelRA16PAMedusaPreAmpswith frequencyresponse between1.5Hzand7.5Hz(TuckerDavisTechnologies,Alachu a,FL).TheEEGsignal wasdigitizedattherateof12kHzandpassedontothedigital signalprocessing(DSP) unit,thePentusaRX-5acquisitionboard(TuckerDavisTech nologies,Alachua,FL).The datafromtheDSPunitwerestreamedontoaPCandstoredforfu rtherprocessing.CBT datawerestreamedthroughtheER4000receiverandstoredon aPCusingVitalView software(MiniMitterCo.,BendOR)atasamplingrateofoned atapointpersecond. DuetointerferencefromtheE-Mittertransponderinthehip pocampalEEGrecordings, itwasnecessarytocollectEEGandCBTdataindependentlyfr omseparatesetsof animals. 4.8ThetaEpochDetection Theta-activityinthehippocampusisdistributedthrougho utthewake-sleepcycle. Giventhefactthattheta-statetransitionscanaffecttheo ccurrenceofmanyhippocampal activities,includingSPWs,IS,andgammarhythms[ Buzsakietal. 1991 Leung 1988 SuzukiandSmith 1987 ],wesoughttoquantifythedistributionofthetaactivity throughouttheday.Epochsofhippocampalthetaactivitywe reidentiedusingan automatedroutine[ Belluscioetal. 2012 Csicsvarietal. 1998 ].Wecalculatedtheratio ofthepowerintheta(6–10Hz)tothepowerindelta(1–6Hz)in 2-secondepochsand classieddataaspredominantlythetawhenthisratioexcee dedaspeciedthreshold. Thethresholdwas2.0bydefault,althoughthesensitivityo fourresultstoarangeof thresholdvaluesbetween1.0and3.0instepsof1.0wasexami ned.Similarly,our resultswerefoundtobeinsensitivetoothercommonchoices ofthetaanddeltabands, suchas5–10Hzand2–4Hz,respectively. 52

PAGE 53

4.9DataAnalysis 4.9.1SmoothingandDetrending Forpurposesofanalyzing24-hourrhythmsinSPKrate,EEGrh ythmspower,theta epochoccurrence,andCBTtimeseries,alltimeserieswerea lltreatedinasimilar manner.First,timeseriesdatawerepooledinto1hnon-over lappingtimebinsandthe averagevalueforeach1htimebinwascalculated.Dataweret hensmoothedusing 6-hour,90%-overlappingmovingaveragewindows.Weobserv edthat,forsometime series,datavalueswoulddriftoverthecourseofdaysorwee ks.Thiswasparticularly thecaseforSPKrates,aswaspreviouslyinvestigated[ Talathietal. 2009 ],andwerefer tothisasbaselinedrift.Todetrenddatabyremovingthesel ong-timescalechanges, weestimatedthebaselinedriftbyaveragingdatawithina1daymovingwindow.Then, wesubtractedthesebaselinevaluesfromtheoriginaldata. Ingureswheredetrended (baseline-subtracted)dataareshown,weusethesymbol todenotethattheplotted valueisadeviationfrombaseline.Wealsoexploredtheuseo falternativewindowsizes andoverlapvaluesforsmoothingandbaselinecalculations ,andfoundthesechanges hadminimaleffectontheresults.4.9.2CosinorAnalysis Toobtainphaseinformation,thedetrendedtimeseriesdata werethencompressed intoasingle24-hourtimewindow.Thiswasachievedbyapply ingthemappingT n T n modulo24hours,whereT n isthetimepointinhoursassociatedwitheachdatavalue X n inthedetrendedtimeseries.Datawerethenttedtosinusoi dalfunctionsoftheform f = A cos (2 ( T T 0 ) = 24) usingleast-squaresminimization.Toconrmthatthedata wereindeedsinusoidal,weconductedthezero-amplitudete st[ Nelsonetal. 1979 ].The 95%condenceintervalsassociatedwithphaseestimateswe reestimatedaspreviously described[ Nelsonetal. 1979 ]. 53

PAGE 54

CHAPTER5 HIPPOCAMPALEEGSPIKES:24-HOURRHYTHMPHASESHIFT 5.1Overview Tostudy24-hourrhythmsofneuralactivity,weobtained invivo EEGrecordingsthat wereperformedcontinuouslythroughoutthepre-injury,la tency,andchronicstagesofa rattemporallobeepilepsy(TLE)model. 12 Fromthisdata,weextractedspontaneous large-amplitudeEEGeventsorSPKs,whichincludedbothsha rpwaves(SPWs)and interictalspikes(IS)[ Buzsakietal. 1983 SuzukiandSmith 1987 ].Together,SPWs andIShavepreviouslybeenusedtoquantify24-hourrhythms ofhippocampalneural activity[ Talathietal. 2009 ].Duetotheirgenerationwithinthehippocampus(Buzski 1986)andthesimilaritiesintheirunderlyingmechanismso fgeneration[ Buzsakietal. 1991 Leung 1988 SuzukiandSmith 1987 ],theyareanidealmarkerforspontaneous neuralactivityoriginatingwithinthehippocampus.Inthe followingsections,wereport ouranalysisof24-hourrhythmsofthesehippocampalSPKs. 5.2Methods 5.2.1AnimalMethodsandDataCollection Animalmethods,experimentaltimelines,andproceduresfo racquisitionofrawdata aredescribedintheGeneralMethodschapter(Chapter 4 ).Anexperimentaltimeline issuppliedinFigure 4-1 oftheGeneralMethods.Ourproceduresforttingof24-hour 1 PartsofthischapterareinpresswiththeJournalofNeuroph ysiology:Stanley,D.A., Talathi,S.S.,Parekh,M.B.,Cordiner,D.,Zhou,J.,Mareci ,T.H.,Ditto,W.L.,Carney,P.R. Localphaseshiftinthe24-hourrhythmofhippocampalEEGsp ikingactivityinarat modeloftemporallobeepilepsy[ Stanleyetal. 2013 ]. 2 ThisworkinvolvedanalysisoftheEvolutionintoEpilepsyE EGdatasets,whichwere collectedovermultipleyearsbyDr.Dong-UkHwang,Dr.Sach inS.Talathi,Dr.Junli Zhou,andotherexperimentalistswhopreviouslyworkedwit hDr.Carney.Dr.Talathi alsosuppliedcodeforextractionandsortingofspikedata. DanielCordinercollected CBTdata. 54

PAGE 55

rhythmdatatocosinefunctionsandforextractionofthetae pochsarealsodiscussedin thatchapter. For invivo EEGrecordings,atotalofN=5animalswereusedforforEEGre cording (3spontaneouslyseizing,2stimulatedbutnon-seizing).A dditionally,N=3ratswere usedforCBTrecording(allspontaneouslyseizing).5.2.2Spike(SPK)Detection FromEEGdata,weextractedlarge-amplitudeEEGevents,whi chincludedboth hippocampalsharpwaves(SPWs)andinterictalspikes(IS)[ Buzsakietal. 1983 Suzuki andSmith 1987 ].WeshallrefertoSPWsandIScollectivelyashippocampalE EG spikes(SPKs).OurprocedureforidentifyingandsortingSP Ksisdetailedpreviously [ Talathietal. 2009 ].Briey,datafromasinglemicrowirechannelinCA1weredi vided into1-hournon-overlappingepochs.High-amplitudeevent sweredetectedwhenthe signalexceededathresholdof5 ,where isthestandarddeviationofthedatainthe epoch.Eventswerecenteredandnormalizedwithina0.45swi ndowandwerethen inputintoacustomizedversionofanestablishedspikeclus teringalgorithm[ Feeetal. 1996 ].Thisalgorithmensuredspikeswithconsistentwaveformw eretrackedthroughout theexperiment(Figure 5-1 ). Subsequentsmoothingofdataandmeasurementof24-hourrhy thmsisas describedinSection 4.9 ofChapter 4 5.2.3Statistics Allstatisticaltestsreportedarepaired-samplesttests, unlessotherwisespecied. Inallcases,signicanceisconsideredtobep < 0.05.Errorbarsrepresentstandarderror ofthemean(SEM)unlessotherwisestated. 5.3Results 5.3.1PhaseShiftin24-HourRhythmofEEGSPKs SpontaneoushippocampalEEGSPKsweretrackedcontinuousl ythroughoutthe dayforthepre-injury,latency,andchronicstagesofTLEmo delrats.Thiswasperformed 55

PAGE 56

Figure5-1.EEGspike(SPK)prolesforpre-injury(Pre),la tency(Post-L)and spontaneouslyseizing(Post-SS)stagesofepileptogenesi s.Solidlineshows meanshapeproleanddottedlinesare200randomlyselected SPKsfroma 24-hourinterval. foreachanimal,anddatawereanalyzedlongitudinally.SPK ratesineachanimal showeddistinct24-houroscillations(Figure 5-2 A).However,followinginjury,aphase shiftemergedof12h(Figure 5-2 B).Thisphaseshiftwasstatisticallysignicantand persistedthroughoutthelatencyperiod(p=0.010,N=3rats )andaftertheadventof spontaneousseizures(Figure 5-2 C,D)(p=0.0078,N=3rats).Goingfrompre-injuryto latencyperiods,thephaseshiftsexperiencedbythesethre eanimalswere11.9 1.4, 9.1 1.1,and9.0 0.7hours(mean 95%condence).Asimilarphaseshiftwas reportedpreviously[ Talathietal. 2009 ];however,theoriginofthephaseshiftwas notclearand,furthermore,itwasnotclearwhetherthephas eshiftwasrelatedtothe emergenceofseizures.Therefore,wealsoanalyzedSPKrate sinanimalsthatwere stimulatedintoSE,butthatdidnotsuccessfullyexhibitch ronicspontaneousseizures. Theseanimalsweremonitoredforatleast4weeksfollowingi nitialSEtoconrmthe absenceofseizures.Wefoundthat,althoughtherewassomed riftinthephase,itwas muchlessseverethanfortheseizinganimals(Figure 5-2 E,F).Specically,thephase shiftsobservedforthetwonon-seizinganimalswere5.6 2.3and2.5 1.5hours(mean 95%condence).Thesedatasuggestthatthephaseshiftisco rrelatedwiththe 56

PAGE 57

emergenceofseizuresfollowingSEstimulation;however,f urthertestsneedtobedone todeterminewhetherthephaseshiftplaysacausalroleinth eemergenceofseizures. Whiletheaforementionedanalysiswasbasedonalowsamples ize(N=3seizing, N=2non-seizinganimals),theresultswereconsistentinth ateachofthethreeseizing animalsshowedaphaseshiftbetween9and12hours.Sincepha seswereestimated overmanycircadiancyclesfrombaseline-subtracteddata, itwaspossibletoobtain highlycondentestimatesofphaseforeachanimal;the95%c ondenceintervals associatedwithacrophaseestimateswereatmost 1.3hoursforseizinganimals, andallreportedanimalspassedthezero-amplitudetestwit hp < 0.0001[ Nelsonetal. 1979 ].Thus,wecanbecondentoftherobustappearanceofaphase shiftineachof thethreeanimalsexamined.5.3.2CoreBodyTemperatureAnalysis GiventhatthemastercircadianclockintheSCNplaysacentr alroleinentraining 24-hourrhythmsthroughoutthebody,wetrackeddailyrhyth msofCBT,awell-validated biomarkerforSCNrhythms[ HofstraanddeWeerd 2008 ].Overall,thephaseofCBT circadianrhythmsdidnotshiftsignicantlyduringtheepi leptogenicperiod(Figure 5-3 )(p=0.68,N=3rats),althoughthevariabilityinthesignal didincrease,ashasbeen previouslyreported Quiggetal. [ 1999 ].ThissuggeststhattheSPKphaseshiftcannot beattributedtoaphaseshiftintheSCNmastercircadianclo ck. 5.3.3ThetaActivityAnalysis Perturbationstoactivityrhythmsandsleeparchitectureh avebeenreportedin animalandhumanepilepsy[ Bastlundetal. 2005 Shouseetal. 1996 Stewartand Leung 2003 ],andmightalsocontributetotheSPKphaseshift.Activity andsleep statecaninuenceSPKsbyaffectingthetransitionofthehi ppocampusbetweentheta (activated)andnon-theta(deactivated)states(Buzski19 96).Thethetastateoccurs duringawakemobilityandrapideyemovement(REM)sleep,wh ilethenon-thetastate occursduringawakeimmobilityandslow-wavesleep[ Kramisetal. 1975 ].Itiswell 57

PAGE 58

Figure5-2.Phaseshiftinthe24-hourrhythmofspontaneous hippocampalEEGspikes (SPKs).(A,B)ChronictrackingofSPKratesforpre-injury( Pre),latency (Post-L),andspontaneouslyseizing(Post-SS)stagesofep ileptogenesisfor asinglerat.Daysaremarkedfromthestartofpre-injuryrec ordingin(A)and fromthedayofstatusepilepticusin(B).Redverticallinei ndicatestherst spontaneousseizureanddottedverticallinescorrespondt o00:00 (midnight).Thegapsinthetracesreectmissingdatadueto technical problems.(C)ThebaselinedriftinSPKratewassubtractedo utandthe resultingdetrendedtimeserieswascombinedintoa24-hour timewindow (modulo24).ThiswasdoneforPre,Post-L,andPost-SSstage sof epileptogenesis.Cosinets(red)revealaphaseshiftpost -injury.(D) EstimatingtheaveragetimeofminimumSPKactivityacrossa llanimals showedastatisticallysignicantphaseshiftduringPostLandPost-SS stages.ThetimeofminimumSPKactivitywasmeasuredtoavoi dthe discontinuitybetween23:59and00:00.(E)SPKratesaresho wnfromarat thatwasstimulatedintostatusepilepticus,butthatdidno tsuccessfully developspontaneousseizuresfollowingfourweeksofmonit oring.Pre-injury (Pre)andpost-injury(Post)stagesareshown.Cosinets(r ed)showaslight driftinphasefollowinginjury.(F)Foranimalsthatdidnot develop spontaneousseizures,thephaseshiftpost-injurydidnotr eachsignicance. Valuesaremean SEM.*p < 0.05byttest. 58

PAGE 59

Figure5-3.Circadianphaseofcorebodytemperature(CBT)r hythmisunperturbed followinginjury.(A,B)Trackingof24-hourrhythmsofCBTi sshownfora singlespontaneouslyseizingrat,pre-andpost-injury,re spectively.Daysare markedfromthestartofpre-injuryrecordingin(A)andfrom thedayof statusepilepticusin(B).(C)Datawerecompressedintosin gle24-hourtime windowsasinpreviousgures.Therewasnoobservedphasesh iftfollowing injury,althoughthevariabilityofthesignaldidincrease .(D)Thelackof phaseshiftinCBTactivityfollowinginjurywasconsistent acrossrats examined.Valuesaremean SEM. establishedthatSPWsareabsentduringthetaandthatISrat esaresubstantially reducedduringtheta[ Buzsakietal. 1991 Leung 1988 SuzukiandSmith 1987 ]. Therefore,suppressionofSPKscanemergeatdifferenttime sofdayresultingfrom activitiesthatpromotehippocampaltheta. Toinvestigatewhetherhippocampalthetaisalteredinsuch awaythatcouldaffect thedailypatternofSPKs,wemeasuredchangesintheaverage timespentinthetheta statethroughoutthecircadiancycle.Thetawasdistinguis hedfromnon-thetabasedon theratiobetweenthepowerinthethetaband(6–10Hz)andthe powerinthedelta band(1–6Hz)in2-secondepochsofdata[ Belluscioetal. 2012 Csicsvarietal. 1998 ] (seeMaterialsandMethods).ExampletracesofeachEEGstat eareshowninFigure 5-4 .Weobservedthatthedistributionofthetaactivityoverth ecircadiancycleshowed 59

PAGE 60

a24-hourrhythmthatdidnotshiftphasefollowinginjury(F igure 5-5 A–C),aresultthat wasconsistentacrossallratsexamined(Figure 5-5 D)(p=0.71andp=0.31forlatency andspontaneouslyseizingstages,respectively;N=3rats) .ThissuggeststhattheSPK phaseshiftisnotdrivenbyaphaseshiftinthedailydistrib utionofhippocampaltheta. Similarndingswereobtainedwhenvaryingtheratioforthe tadetectionbetween1.0 and3.0. SPKsmayincludebothSPWsandISand,asdiscussedabove,whi leSPWsare absentduringtheta,IScanoccurduringthetastates[ Buzsakietal. 1991 Leung 1988 SuzukiandSmith 1987 ].Therefore,wehypothesizedthattheemergenceof IScouldcontributetothephaseshiftbypromotingtheemerg enceofSPKsduring timesofdaywhenthetaisprevalent.Toinvestigatethis,we trackedtherateofSPKs occurringexclusivelyinthenon-thetastate(Figure 5-5 E,F).Still,aphaseshiftof12 hoursappeared,aresultthatwasconsistentacrossallrats examined(p=0.0073and p=0.00094forlatencyandspontaneouslyseizingstages,re spectively;N=3rats).This suggeststhatthephaseshiftisnotdependentonISoccurrin gduringthethetastate and,ingeneral,isnotduetotheoccurrenceofthetastatetr ansitions. Figure5-4.Extracted2-secondepochsofthetaactivityand non-theta,respectively.SPK eventisindicatedbyarrow.Ratioofthetatodeltapoweris3 .28intheta(A) and0.09innon-theta(B). 60

PAGE 61

Figure5-5.InteractionofSPKswiththe24-hourrhythmofth etaactivitycannotexplain SPKphaseshift.(A,B)Fractionoftimespentinthetastatef orpre-injury (Pre),latency(Post-L),andspontaneouslyseizing(PostSS)stagesof epileptogenesisforasinglerat.Daysaremarkedfromthest artofpre-injury recordingin(A)andfromthedayofstatusepilepticusin(B) .Redvertical linemarkstherstspontaneousseizure.Thegapsinthetrac esreect missingdataduetotechnicalproblems.(C)Datawerecompre ssedintoa single24-hourtimewindowasdescribedinpreviousgures. Thetheta activitytrendsshowa24-hourrhythmthatdoesnotshiftpha sebetweenPre, Post-L,andPost-SSstages.Cosinetsareindicatedbysoli dredlines.(D) Examiningthetimeofminimumthetaactivityacrossallrats showedthat therewasnosignicantchangeinthethetarhythms24-houro scillation.(E) ExaminationofSPKsexclusivelyinthenon-thetastateallo wedfortracking ofSPKratesindependentoftransitionsbetweenthetaandno n-thetastates. Asbefore,aphaseshiftwasobservedfollowinginjury.(F)T hisanalysiswas repeatedforallratsandweobservedastatisticallysigni cantphaseshift betweenthepre-injuryandpost-injurystagesofepileptog enesis.Valuesare mean SEM.*p < 0.05byttest. 61

PAGE 62

5.4Discussion 5.4.1SummaryofResults Wehaveshownthatthereemergesaphaseshiftof12hoursinth e24-hourrhythm ofhippocampalspikesinananimalmodeloflimbicepilepsy. Thisphaseshiftwas evidentinallseizinganimalsexamined,andwasseverelyre ducedinanimalsthatwere stimulatedbutthatdidnotdevelopspontaneousseizures.W emeasuredthisphaseshift relativetoboththeCBTrhythmandthe24-hourrhythmofhipp ocampalthetaactivity, andfoundthatneitheroftheserhythmsexhibitedaphaseshi ftsufcienttoaccountfor thechangesinhippocampalSPKs.5.4.2VariabilityofDatainPre-InjuryTimePeriodofNon-S eizingAnimals Itwasnotedthatpre-injurydataforthenon-seizinganimal sshowedasignicant amountofvariability,asevidencedbythesizeoftheerrorb arinFigure 5-2 F.Thismay haveresultedinpartfromthefactthattechnicaldifculti espreventedthefullsevendays ofpre-injurydatafrombeingcollectedforoneofthesenonseizinganimals.Thedata fromtheanimalinquestionareshownasanexampleinFigure 5-2 E.Althoughthis animalstillpassedthezero-amplitudetest(p=0.000096), thepaucityofdataincreased thecondenceintervalofthisanimalsT SPKmin to 2.3hours(95%condenceinterval); thisisalmosttwicethelargestcondenceintervalforthes eizinganimals, 1.3hours. Whilethepaucityofdatawasonefactorcontributingtothev ariabilityofthepre-injury data,anothercontributingfactormaybethepresenceofaci rcadiandriverthatentrains nottotheday-nightcycle,buttoanothercycleintheenviro nment.Ithaspreviouslybeen proposedthatentryintoanewenvironmentandassociatedfo odavailabilitycouldaffect aputativefood-entrainablecircadianoscillator[ MunnandBilkey 2012 ].Theinuence ofsuchacircadiandrivercouldcontributetovariabilityi ntheSPKacrophaseinthe pre-injuryperiod.Furthertestsneedtobedonetodetermin ewhethersuchvariability inthepre-injuryacrophaseaffectsthelikelihoodofdevel opingspontaneousseizures followingSE. 62

PAGE 63

5.4.3InterictalSpikesasaDriverforthePhaseShift TosummarizeouranalysisoftheSPKphaseshiftinrelationt othetarhythms,we observedthefollowing:1)thatthetarhythmsdonotshiftph asefollowinginjury;and2) thatSPKactivityexclusivelywithinthenon-thetastateex hibitsaphaseshift.Whilethe secondpointdiscountsthepossibilitythatISmaycontribu tetothephaseshiftbyleaking intothethetastatefollowinginjury,thereisyetanotherm echanismbywhichIScould beresponsibleforthephaseshift.Specically,itispossi blethatthephaseshiftcould simplybeduetoanincreaseintheoverallfrequencyofISfol lowinginjury,provided thattheseISoccuratadifferentcircadianphasethantheSP Ws.Unfortunately,this possibilitycannotbeconrmedordeniedwithoutdirectmea surementofSPWsand IS.However,forseveralreasons,weproposethatbothSPWsa ndIStogethershift towardspeakingduringthedaypost-injury.First,giventh esimilaritiesinSPWandIS mechanismsofgeneration[ Buzsaki 1986 SuzukiandSmith 1987 ],itismostlikelythat theyareaffectedbycircadiandriversinasimilarmanneran d,therefore,exhibitsimilar circadianphases.Secondly,ouranalysisofEEGrhythmshas shownthatcircadian modulationofEEGrhythmamplitudeinthebetaandlowgammaf requencyranges alsoexhibitsaphaseshiftfollowinginjury(Chapter 8 ).Therefore,thisidenticationofa secondaryEEGfeaturethatalsophaseshiftssuggeststhatp rocessesotherthansimply theemergenceofISareinvolved. 63

PAGE 64

CHAPTER6 MECHANISMSFORCIRCADIANRHYTHMPHASESHIFTING 6.1Overview Inthischapter,wedevelopadetailedcomputermodelincorp oratingrealistic circadianinputsinordertoexplainourexperimentalobser vations. 1 Theseobservations areoutlinedinthepreviouschapter,inwhichwereportapha seshiftinthe24-hour rhythmofhippocampalspikingactivityinepilepticrats.W ealsofoundthatthecircadian phasesofCBTandhippocampalthetaactivityrhythmsstatio nary,suggestingthat thephaseshiftdidnotresultfromalterationtotheSCNrhyt hmorwake-sleepcycle. Therefore,wehereinvestigatealternatemechanismsthatc ouldaccountfortheSPK phaseshift.Asdiscussedearlier,thecircadiansystemiso rganizedintoahierarchyof relaycenters.WhiletheSCNitselfdoesnotprojectdirectl ytothehippocampus,the hippocampusreceivescircadianinputfrommanyotherrelay centers.Itispossiblethat permanentdamagetothesecenters,ashasbeenhypothesized tooccurinmultiple typesofepilepsy[ Quigg 2000 ],couldcontributetotheexperimentallyobservedphase shift.Theideasinthischapterwereinitiallypresentedat OCNS2011[ Stanleyetal. 2011a ]. 6.2Methods 6.2.1ModelingOverview Toinvestigatethishypothesis,weimplementedadetailedc omputermodelof circadianregulationinahippocampalneuralnetwork.Thre etypesofhippocampal neuronsweremodeled,namely,pyramidalcells,basketcell s,andO-LMcells(Figure 6-2 A).Inaddition,weincludedapopulationofMSGneurons.The modelsparameters 1 PartsofthischapterareinpresswiththeJournalofNeuroph ysiology:Stanley,D.A., Talathi,S.S.,Parekh,M.B.,Cordiner,D.,Zhou,J.,Mareci ,T.H.,Ditto,W.L.,Carney,P.R. Localphaseshiftinthe24-hourrhythmofhippocampalEEGsp ikingactivityinarat modeloftemporallobeepilepsy[ Stanleyetal. 2013 ]. 64

PAGE 65

andnetworkconnectivitywerebasedontheCA3network(Figu re 6-1 ),whichcontrols theinitiationoftheCA1SPKs[ Buzsaki 1986 Ellenderetal. 2010 ]. Figure6-1.Networkconnectivityintheneuralnetworkmode l.Sitesofcircadianinputto regionCA3andtothemedialseptumareindicatedbytildemar ks. Additionally,allGABAergicsynapsesaresubjecttocircad ianmodulationby melatonin.BC=basketcells;MSG=medialseptalGABAergici nterneurons; O-LM=stratumoriensinterneuronsprojectingintolacunos ummoleculare; PYR=pyramidalneuron. Representationofcircadianinputinthemodelwasbasedone videncethatmultiple sourcesofcircadiandriveinteractinthehippocampustopr oduceanoverall24-hour 65

PAGE 66

rhythm.Forexample,itwasrecentlyshownthatdailyvariat ionsinCA1evoked responsesarealteredbutnoteliminatedbymelatoninknock out[ Chaudhuryetal. 2005 ],suggestingtheoverlappinginuenceofmultiplecircadi anfactors.Threecircadian inputsweremodeled:circadianmodulationofmedialseptum neuralactivity,nighttime releaseofmelatonin,anddiurnalmodulationofhippocampa lpyramidalcells.These inputsarebasedonknownphysiologyandthemodelwasvalida tedtoreproduce ndingsfromseveralexperimentalstudies:rhythmiccycli ngofpyramidalcellringrate, peakingduringtheday[ BrunelanddeMontigny 1987 ];increasednighttimeinhibitory drivefollowingSEinjury[ Matzenetal. 2012 ];andalsoincreasedneuralactivityin responsetoelevatedmelatonin[ Musshoffetal. 2002 ]. 6.2.2CA3NetworkModel Ourinvivoanalysistracked24-houroscillationsintherat eofspontaneous hippocampalSPKs.TheSPKsrecordedexperimentallyinCA1a reknowntooriginate asaresultofsynchronouspopulationburstsinCA3circuitr y[ Buzsaki 1986 ].Therefore, modelingeffortswerefocusedoncircadianregulationofne uralactivityinregionCA3. Thenetworkmodelcontained200CA3pyramidalcells,25bask etcells,and25stratum oriensinterneuronsthatprojectintothelacunosummolecu lare(O-LM).Inaddition, torepresentcircadianinputfromthemedialseptum,wealso includedapopulationof medialseptalGABAergic(MSG)neurons.Wefocusedspecica llyonseptalGABAergic ratherthanseptalcholinergicinputbecausepharmacologi cal[ Buzsaki 1986 Suzuki andSmith 1988b ]andlesioning[ Leeetal. 1994 ]studieshaveshownlittlecholinergic effectonSPKratesinvivo. Thepyramidalneuronmodelwascomprisedof19compartments ,with8compartments forthebasaldendrites,10fortheapicaldendrites,andasi nglesomaticcompartment [ Traubetal. 1991 ].Eachcompartmentcontainedactiveandpassivechannelco nductances [ HodgkinandHuxley 1952 ]andwasconnectedtoothercompartmentsbypassive 66

PAGE 67

resistances[ Rall 1962 ].Thecableequationsforthemembranevoltage, V i ,in compartment i aredescribedbelow. C m i dV i dt = 19 X j =1 G i j ( V j V i ) I int i I syn i I leak i + I inj i (6–1) I int i = I Na + I K DR + I Ca + I K AHP + I K C + I K A (6–2) Intheseequations, G i j isthespecicaxialconductancebetweencompartments i and j ( G i j =0 forunconnectedcompartments; G i i =0 forall i )and C m i isthespecic membranecapacitance. I int i representsthesumofthefollowingintrinsicioniccurrent s: sodium(Na);delayedrectierpotassium(K DR );A-typetransientpotassium(K A );calcium (Ca);calcium-dependentpotassium(K AHP );andcalcium-andvoltage-dependent potassium(K C ).ThesecurrentsaredescribedbytheHodgkin-Huxleyforma lism, I k = g k g k ( V E k ) ,where g k and E k areionchannelmaximalconductanceandreversal potential,respectively. g k isaproductofHodgkin-Huxleytypegatingvariables x with rstorderkineticsoftheform dx ( V t ) = dt = ( V )(1 x ) ( V ) x ( V ) and ( V ) arederivedfromempiricaltstoexperimentaldata[ Traubetal. 1991 ].Similarly, I syn representsthesumofallsynapticcurrents, I syn = P k g syn k s k ( V E syn k ) .Eachtermin thissumreectsaspecicpresynapticneuronandnetworkpr opertiesareprovidedin thefollowingsection.Variables g syn k and E syn k arechannelmaximalconductanceand reversalpotential,respectively.Thegatingvariablefor thesynapse, s k ,iscalculated bythe synchan objectinGENESIS[ BowerandBeeman 2003 ],andusestheimpulse responsefunction A ( e t = 1 e t = 2 ) = ( 1 2 ) .Inthisequation, A isanormalizingfactor and 1 and 2 aresynaptictimeconstants.Theyarespecictothetypeofs ynapseand aredescribedinthesectionbelow. I leak i representsleakthroughthecompartment's membraneand I inj i isdirectcurrentinjectionappliedtothesomaandwassetto 500pA. 67

PAGE 68

Allbiophysicalparameters,includingvaluesforratecons tants,maximalconductances, reversalpotentials,calciumaccumulation,andcompartme ntalgeometries,aretaken directlyfromtheGENESISimplementationof Traubetal. [ 1991 ]. Forallinterneurons,weusedasinglecompartmenttorepres entthecell.Basket cellsarebasedon WangandBuzsaki [ 1996 ],whileO-LMandMSGcellsarebasedon Wang [ 2002 ].Codeforallinterneuronswasderivedfromtheimplementa tionby Hajos etal. [ 2004 ],whichisavailableonModelDB(accessionnumber116567)[ Hinesetal. 2004 Miglioreetal. 2003 ].Thevoltageequationforallinterneurontypesassumesth e formbelow. C m i dV i dt = I int I syn I leak + I inj (6–3) Intrinsiccurrents, I int ,arespecictoeachinterneurontypeasfollows:basket I int = I Na + I K DR ;O-LM I int = I Na + I K DR + I H + I CA + I K Ca ;MSG I int = I Na + I K DR + I K S .Here,Naand K DR aresodiumanddelayed-rectierpotassium,respectively, whileCaishigh-voltage activatedcalcium,K Ca iscalcium-dependentpotassium,K s isslowlyinactivating potassium,andhisthehyperpolarization-activatedcatio ncurrent.Fulldescriptions oftherateconstantsunderlyingthesecurrentshavebeenpr eviouslypublished[ Hajos etal. 2004 Wang 2002 WangandBuzsaki 1996 ].Defaultcurrentinjectionvalues, I inj forbasket,O-LM,andMSGwere0pA,-10pA,and21pA,respecti vely. SimulationswereruninGENESIS2.3[ BowerandBeeman 2003 Boweretal. 2002 ]onaMacXServcluster,anddatawereanalyzedinMatlabR201 1a.Thecomplete GENESIScodeisavailableonModelDB,accessionnumber1421 04. 6.2.3NetworkProperties TheCA3networkpropertieswerespeciedintermsofsynapti cstrengths,synaptic timeconstants,andsynapticconnectivity.Synapticcurre ntswereimplementedwiththe standarddouble-exponentialmechanism,usingtheGENESIS synchan object[ Bower 68

PAGE 69

andBeeman 2003 ](detailsabove).PyramidalcellsformedAMPAsynapses,wh ileall interneuronsformedGABA A synapses.Table 6-1 listssynapticstrengths,specifying themaximalconductanceofagivensynapticconnection.The sevalueswerebased onexperimentallymeasuredunitaryexcitatory/inhibitor ypostsynapticpotentialvalues [ TraubandMiles 1991 Traubetal. 1999a ],whereavailable,aswellasonprevious models[ Hajosetal. 2004 Neymotinetal. 2011 Taxidisetal. 2012 ].Synaptictime constantswerethoseusedin[ Neymotinetal. 2011 ],withAMPAsynapseshaving 1 = 0.05msand 2 =5.3ms.AllGABAergicsynapseshad 1 =0.07and 2 =9.1ms,with theexceptionofsynapseswithpresynapticO-LMcells.Insu chcases, 1 =0.2and 2 = 20ms(Neymotinetal.2011).Aschematicdiagramofthenetwo rkconnectivityisshown inFigure 6-1 andinTable 6-2 wereportnetworkconnectivityintermsofthenumberof presynapticneuronsconvergingoneachpostsynapticneuro n.Connectivityvalueswere derivedfrompairedrecordingsofpresynapticandpostsyna pticcells[ TraubandMiles 1991 ]aswellasfrompreviousmodelingstudies[ Hajosetal. 2004 Neymotinetal. 2011 Taxidisetal. 2012 Traubetal. 1996 ].Pyramidalcellsreceivedinnervationfrom otherpyramidalcellsinthebasaldendrites(4thbasalcomp artmentfromthesoma). Basketcellsinnervatedthesomaofpyramidalcells,wherea sO-LMcellsinnervatedthe distalapicaldendrites(9thapicalcompartmentfromtheso ma). Table6-1.Maximalsynapticconductances(nS) Postsynaptic PresynapticPyramidalBasketO-LMMSG Pyr2.60.050.05Basket9.20.125O-LM8.3MSG0.50.50.25 6.2.4BackgroundActivity ThebackgroundnetworkactivitywasmodeledbyPoisson-dis tributedexcitatoryand inhibitorypostsynapticpotentials.Thesesynapsesimpin gedonthesomaofpyramidal cellsandinterneurons,andalsoonthemostdistalapicalde ndriticcompartmentof 69

PAGE 70

Table6-2.CA3networkconnectivityexpressedintermsofsy napticconvergence Postsynaptic PresynapticPyramidalBasketO-LMMSG Pyr102010Basket1525O-LM10MSG5510 pyramidalcells.ThemeanfrequencyofAMPAandGABAbackgro undeventswas1000 Hz,andthemaximalbackgroundsynapticconductancesarepr ovidedinTable 6-3 Thesevalueswerechosentoprovidesymmetricrestingmembr anepotentialuctuations at-65mVwithstandarddeviationsbetween1and2mV(Destexh eetal.2003). Table6-3.Backgroundsynapticactivity CellSynapseSectionConductance(nS) PyramidalAMPASoma1.0PyramidalAMPADendrite1.0BasketAMPASoma0.03O-LMAMPASoma0.03PyramidalGABA A Soma2.5 PyramidalGABA A Dendrite2.5 BasketGABA A Soma0.01 O-LMGABA A Soma0.01 6.2.5CircadianModulation Torepresentcircadiandrive,weincorporatedthreemajori nputstothemodel thatweresubjectto24-hourmodulation(Figure 6-1 ).Theseinputsweredenedas afunctionofthe24-hourcircadiantimevariable,T,withT= 0hcorrespondingto midnightandT=12htonoon.Eachcircadianinputwascontrol ledbyacircadian scalingfactor(Figure 6-2 B),denedbelow.Sincecircadianchangeshappenonmuch slowertimescalesascomparedtotheneuralnetworksdynami cs,atotalof16separate simulationswererun,inwhichTwasvariedasamodelparamet er.Eachsimulation produced10secondsofdata,andtherstsecondwasdiscarde dpriortoanalysisin ordertoremovetransienteffects. 70

PAGE 71

Therstcircadianinputwasmodulationofthemedialseptum .Multiunitrecordings haveshownthatmedialseptumactivitypossessesarobustci rcadianrhythm,peaking atnight[ Yamazakietal. 1998 ].Modulationofthemedialseptumwasachievedby scalingthetoniccurrentinjection, I inj ,toMSGcells: I inj = I inj S septal .Inthisequation, I inj isthedefaultcurrentinjectionforMSGcells,asdescribed above, S septal =(1+ C septal cos (2 T = 24) istheseptalcircadianscalingfactor,and C septal =0.25istheseptal circadianscalingcoefcient.Thisinputcausedmaximalse ptalactivitytopeakwhen T=0h.Inourmodel,wefocusedspecicallyonseptalGABAerg icinputbecause,as mentionedabove,previousstudieshaveshownlittlecholin ergiceffectonSPKratesin vivo[ Buzsaki 1986 Leeetal. 1994 SuzukiandSmith 1988b ]. Second,melatonin,aprimarysystemicoutputofthecircadi ansystem,isreleased nightlybythepinealgland.Inthehippocampus,melatoninh asanexcitatoryeffect anddecreasesthestrengthofGABAergicinhibition[ Wanetal. 1999 ].Toaccount fortheeffectsofmelatonin,allGABAergicmaximalsynapti cconductances(including backgroundsynapses)weremultipliedbythescalingfactor S mel =1+ C mel cos (2 ( T 12) = 24) ,producingmaximumattenuationofGABAatnight. C mel =0.10withtheexception ofFigure 6-2 E,where C mel isvariedacrossarangeofparameters(describedbelow). Finally,avarietyofotherfactorscanalsoinuenceCA3act ivity,including24-hour variationofCA3pyramidalneuroncalciumcurrents[ Koleetal. 2001 ],adenosine release[ Liuetal. 2000 ],andneuromodulatoryeffects.Asasimplifyingassumptio n, werepresentedthecombinedeffectsoftheseadditionalcir cadianinputsbysinusoidal modulationofpyramidalcellcurrentinjection.Experimen talstudiessuggestthat 24-hourmodulationofCA3pyramidalcellringactivitypea ksduringtheday[ Brunel anddeMontigny 1987 ]and,therefore,pyramidalcurrentinjectionwasmodulate das follows: I inj = I inj S pyr ,where I inj isthedefaultcurrentinjectionforpyramidalcells,as describedabove; S pyr =(1+ C pyr cos (2 ( T 12) = 24) ;and C pyr =0.25.Valuesofthe 71

PAGE 72

septal,melatonin,andpyramidalcircadianscalingsfacto rsasafunctionoftimeofday aresummarizedinFigure 6-2 B. Thedefaultstrengthsofthecircadianscalingcoefcients C septal C mel and C pyr werebasedonexperimentalstudieswhereavailable.Speci cally, C mel =0.10was chosentoproduceanapproximately20%changeinGABA A currents,consistentwith thatobserveduponapplicationofphysiologicallevelsofm elatonininslice[ Wanetal. 1999 ]. C pyr wasconstrainedsothatpyramidalcellringpeakedduringt heday,inorder toensureagreementwithexperimentalmeasurements[ BrunelanddeMontigny 1987 ], althoughwenotethatthisexperimentallymeasureddaytime peakshowedaseasonal dependenceaswell. C septal waschosensuchthatinterneuronswereentrainedtoseptal circadiandrive.Thus,while C mel wasconstrainedbydirectexperimentalmeasurement, C pyr and C septal wereadjustedasparameterstoenablethemodeltoreproduce functional activity.6.2.6SensitivityAnalysis Toevaluatethesensitivityofourmodelingresultstovaria tionintherelative strengthsofcircadianinputs,wesweptmelatoninandsepta lcircadianinputsthrougha rangeofvalues.Thestrengthofthemelatonincircadianinp utwasvariedbysweeping C mel between0.05and0.18.Thestrengthofmedialseptalinnerva tionwasvariedby randomlydisablingMSGcells,soastosimulateinjurytothe septum.Thepercentageof remainingMSGcells, fMSG ,wasvariedbetween0and100%.Allotherparameters wereheldattheirdefaultvalues.Foreach( C mel fMSG )pairofvalues,asetof simulationswereruninwhichcircadiantimeTwasvariedand theresultingring ratedatawerettoasinusoidwith24-hourperiodtoestimat eacrophase.Forsome iterationsofthesimulation,instancesarosewhensinusoi dscouldnotbereliably ttothedata.Generally,theseinstancescorrespondedtos ituationswhereinthe relationshipbetweenringrateandtimewaseithernon-sin usoidalorofverylow amplitude.Therefore,werejectedtsforwhichthemeansqu arederror(MSE)was 72

PAGE 73

greaterthan33%ofthedatasvarianceandforwhichtheampli tudeofthesinusoidal oscillationwaslessthan3%ofthemeanringrate. Wealsoinvestigatedthesensitivityofthemodeltochanges initsinternalnetwork connectivity.Drasticnetworkchangescouldaltertherin gactivitypatternsproduced bythenetwork.Subtlernetworkchangescouldaffecthow24hourrhythmspropagated throughoutthenetwork.However,inmostcases,thenetwork wasstableforchangesin convergencevaluesby5neurons.Wealsofoundthatincreasi ngtheparameters C pyr and C septal ,whichwetunedasfreeparameters,couldoftencounteractt heeffectson network24-hourrhythmsofdecreasednetworkconvergence, andviceversa. 6.3Results Asdescribedinthecomputermodelreceivedthreemajorcirc adianinputs. Examplesofneuralringandcircadianscalingfactorinput sareshowninFigure 6-2 A,B.Whenallthreecircadianinputswereincluded,pyramid al,O-LM,andbasket cellringratesoscillatedinphase,peakingatnoon(Figur e 6-2 C).Torepresentinjury, weexploredtheeffectsofremovingoneofthecircadianinpu ts,specically,theinput fromMSGcells.Thiscausedbasketcellringacrophasetosh ifttothenight,while pyramidalsstillpeakedatnoon(Figure 6-2 D).Pyramidalcelloscillationsalsoshowedan increaseinamplitude,resultinginaclearphasemisalignm entwithbasketcells.O-LM cellsstillpeakedweaklyatnoon,buttheytoocouldexhibit aweaknighttimepeakwith increasedmelatonindrive.Tofurtherinvestigatetheorig insofthebasketcellphase shift,wesystematicallyvariedthestrengthsofbothsepta landmelatonininputs(Figure 6-2 E).Whentheseptalinputwassufcientlystrongerthanthem elatonindrive,basket cellringpeakedduringthedayand,whentheconversewastr ue,itpeakedduring thenight.Increasingtheamplitudeofthepyramidalcellci rcadiandriveelevatedthe boundaryregionvertically(notshown).Ingeneral,inorde rforthebasketcellphase shifttooccur,theseptalcircadianinputmustdominateint hehealthystateandthen besufcientlydamagedsuchthattheboundarybetweenthese twophase-regimes 73

PAGE 74

iscrossed.Thethinnessoftheboundaryregionimpliessens itivitytosmallchanges incircadiandrive;insomecasesevenan 20%reductioninseptalinnervationwas sufcienttoproduceaphaseshift,anamountthatiscompara blewithcelllossreported inotherparahippocampalregions[ Gorteretal. 2003 ].Theseresultsshowthatphase shiftingofhippocampalneuralactivitycanemergefollowi ngachangeinthebalanceof circadianinputs.Asweshalldiscuss,thisphaseshiftingo fneuralringratesmighthave directrelevanceforexplainingthephaseshiftinexperime ntallymeasuredSPKs,and mightalsoberelevantforthedeterminingthedailytimingo fepilepticseizures. 6.4Discussion 6.4.1GeneralityoftheComputerModeltoAlternativeSourc esofCircadian Perturbation WhileourMRIcharacterizationfocusedspecicallyonchan gesinthemedial septum,ourcomputermodeldisplayedaphaseshiftinthecir cadianrhythmofbasket cellringduetochangesinthemelatonincircadianinput(F igure 6-2 E)aswell.Sucha phaseshiftcouldalsobeobservedforchangesinthestrengt hofthepyramidalcircadian input(notshown).Therefore,themechanismforproducinga phaseshiftbyaltering therelativebalanceofcircadiandriveappearsgeneraland independentofthenature ofthespeciccircadiandriversinvolved.OurMRIstructur alcharacterizationofthe medialseptumsuggeststhatdamagetotheseptumisonesourc eforproducingaltered circadiandrive.Damagetotheseptumisalsosupportedbypr eviousstudies[ Gorter etal. 2001 2003 ].However,itispossiblethatothercircadiandriverscoul dalsobe modiedsoastocontributetothephaseshift.Forexample,c hangesinmelatonin levelshavebeenreportedinepileptics,althoughthereisc onictingevidenceas towhethermelatoninisincreasedordecreasedintheabsenc eofseizures[ Bazil etal. 2000 HofstraanddeWeerd 2009 Schapeletal. 1995 ].Likewise,damage toothernucleiundercircadianinuence,suchastheorexin -secretingcellsinthe hypothalamus[ Peyronetal. 1998 Selbachetal. 2004 ],histamine-secretingcellsin 74

PAGE 75

Figure6-2.Damagetothemedialseptumcanproduceacircadi anphaseshiftby alteringthebalanceofcircadianinput.(A)Voltagetraces showingexamples ofneuralactivityproducedbythemodel.Celltypesfromtop tobottomare BC,MSG,O-LM,andPYRandarecoloredblue,green,red,andcy an, respectively.(B)Threecircadianinputswereincluded:ci rcadianmodulation ofpyramidalcellcurrentinjection,circadianmodulation ofmedialseptal GABAergiccellcurrentinjection,andnightlyattenuation ofGABA A synaptic transmissionbymelatonin.Eachinputwassimulatedbyadju stingits respectivescalingfactor,Spyr,Sseptal,andSmel(seeMat erialsand Methods).(C,D)Simulationsofneuralactivitybefore(C)a ndafter(D)the completeremovalofmedialseptuminputshowa1800phaseshi ftinbasket cellcircadianrhythms.MSGcellsareindicatedbydottedgr eenline.(E) Basketcellringacrophaseisafunctionofthepercentageo fseptal innervation,fMSG,andthemelatonincircadianscalingcoe fcient,C mel (see MaterialsandMethods).Crossesmarksimulationcongurat ionsinC,D; regionswheresinusoidscannotbereliablyttedforphasee stimationare shadedwhite.Foraphaseshifttooccur,medialseptalinput mustbe sufcientlydamagedsuchthatthewhiteboundaryregionisc rossed.Error barsmean SEM.BC=basketcells;MSG=medialseptalGABAergic interneurons;O-LM=stratumoriensinterneuronsprojecti ngintothe lacunosummoleculare;PYR=pyramidalneurons. 75

PAGE 76

thetuberomammillarynuclei[ YanovskyandHaas 1998 ],orserotonin-producingcellsof theraphenuclei[ AssafandMiller 1978 Kubotaetal. 2003 ],mayalsoberelevant.In general,itispossiblethatanycircadianinputthatmeetst hefollowingthreecriteriacould berelevantforinuencingtheSPKphaseshift:1)hastheabi litytoinuencetherate ofhippocampalSPKs;2)showsevidenceof24-hourrhythms;3 )experiencesdamage oralterationfollowingSE.Whilethemedialseptummeetsth esecriteria,othersources ofcircadianinputmayberelevantaswell.Itisalsofeasibl ethatconnectivitywithinthe CA3networkmightbealteredduringepileptogenesis.Weinv estigatedthispossibility bychangingnetworkconvergenceswhileholdingotherparam etersconstant,andfound thatthiscouldproduceaphaseshiftundercertaincircumst ances.Forexample,one suchconditionwasareductionintheconvergenceofpyramid alcellsonbasketcells. Aspyramidalcellsreachedpeakringduringtheday,lossof theirinputtobasketcells allowedtheunderlyingmelatonincircadiandrivetopromot epeakringatnight,thereby producingaphaseshift.Thus,lossofprincipalcellinputt obasketcellsduetonecrotic processesmightalsobeacontributingfactortoaphaseshif t.Thisinvestigationalso underscorestheprinciplethatmanyparameterchangescanl eadtoaphaseshiftviathe sameunderlyingmechanism,namely,achangeinthebalanceo fcircadiandrivetoa givencelltype.6.4.2RelationshipBetweenExperimentalSPKPhaseShiftan dComputer ModelingResults Animplicitassumptioninourcomputermodelingworkisthat thephaseshiftinthe circadianrhythmofSPKsisdirectlyrelatedtohippocampal neuralringactivity.While theneuralmechanismsofSPKinitiationarenotyetfullyund erstood,recentstudies haveshownthatSPKratesareinuencedbyneuronring.Inpa rticular,aninvitro studyshowedthatdirectstimulationofindividualperisom atic-targetinginterneurons, butnotothercelltypes,wassufcienttoaffecttheprobabi lityofCA3populationbursts [ Ellenderetal. 2010 ].Interestingly,ourcomputermodelalsopredictedthatap hase 76

PAGE 77

shiftwouldemergespecicallyinbasketcellsringfollow inglossofseptalinput(Figure 6-2 ).Basketcellsweremostsusceptibletophaseshiftingbeca usetheywerestrongly inuencedbyboththeseptalandthemelatonincircadianinp uts.Forexample,while O-LMcellswerealsodrivenbytheseptalcircadianinput,th eywerelessinuencedby melatoninowingtotheirlackofrecurrentGABAergicconnec tivity.Forothervaluesof circadianscalingcoefcients(specicallyincreased C mel and C septal ),weobservedthat phaseshiftscouldoccurforO-LMandpyramidalcellsaswell .Therefore,itispossible thatchangesintheringpatternsofotherneurontypesor,a lternatively,changesin networkproperties,suchassynchronization,mightalsoco ntributetothecircadian phaseshift. 77

PAGE 78

CHAPTER7 STRUCTURALCHANGEINTHECIRCADIANSYSTEM 7.1BackgroundandOverview Ourmodelingworkhassuggestedthatchangesinthestrength ofcircadianinputs canproduceaphaseshiftinhippocampal24-hourrhythms,ev enwhenthephasesof theindividualcircadianinputsremainconstant. 12 Suchalteredcircadiandrivecould arisefromdamagetocircadianrelaycenters,ashasbeenpro posedtooccurinepilepsy [ Quigg 2000 ]. Thereareanumberofcircadianrelaycentersthatcouldbeda magedsoasto contributetothisphaseshift.Wechosespecicallytoexam inethemedialseptum forsignsofstructuralchangeforanumberofreasons.First ,itwasrecentlyproposed basedoncognitivestudiesthattheseptumcouldbeanimport antcircadianrelay centerbetweentheSCNandthehippocampus[ Rubyetal. 2008 ].Multiunitrecordings haveshownthattheseptumpossessesarobustcircadianrhyt hmthatpersistsinthe absenceofexternallight/darkcues[ Yamazakietal. 1998 ].Furthermore,itiswell establishedfromtracingstudiesthattheseptum,unliketh ehippocampus,receives heavyinnervationfromtheSCN[ Morinetal. 1994 ].Inturn,themedialseptumis perhapsthemostcriticalsubcorticalinputtothehippocam pus,andlesioningstudies haveshownthattheseptumstronglyinuencestherateofspo ntaneousgeneration ofSPKs[ Buzsaki 1986 SuzukiandSmith 1988a ].Giventhatthemedialseptum exhibitsbothcircadianrhythmicityandalsotheabilityto regulatehippocampalSPKs,we proceededtoinvestigateanatomicalchangesinthemedials eptumusingMRIstructural 1 PartsofthischapterareinpresswiththeJournalofNeuroph ysiology:Stanley,D.A., Talathi,S.S.,Parekh,M.B.,Cordiner,D.,Zhou,J.,Mareci ,T.H.,Ditto,W.L.,Carney,P.R. Localphaseshiftinthe24-hourrhythmofhippocampalEEGsp ikingactivityinarat modeloftemporallobeepilepsy[ Stanleyetal. 2013 ]. 2 ThisworkwasmadepossiblebytheanalysisofMRIdata,colle ctedbyDr.MansiB. Parekh.Dr.Parekhalsoperformedthebertrackinganalysi s. 78

PAGE 79

analysis.Thedatainthischapterwereoriginallypresente datSleep2012[ Stanleyetal. 2012 ]. 7.2Methods 7.2.1AnimalMethods Generalmethodsforanimalhandlingandinductionofthetem porallobeepilepsy (TLE)modelareasdescribedintheGeneralMethodschapter( Chapter 4 ). 7.2.2ExperimentalTimeline AtotalofN=11animalswereusedforexcisedMRI(3control,8 spontaneously seizing).Ofthe8spontaneouslyseizinganimalsusedforex cisedMRI,N=6also providedusabledataforlongitudinalinvivoimaging.Anim alsweremonitoredforat most60days.Theexperimentaltimelineforratsundergoing MRIisshowninFigure 7-1 Figure7-1.ExperimentaltimelineforratMRI. 7.2.3MRIDataCollectionandAnalysis ForMRI,theratswerecontinuouslyvideo/EEGrecordedfora periodof60 dayspost-SE.Temporalchangesintheratbrainsweremonito redinvivowithMRI pre-SEandalsoatdays3and60post-SEat11.1Tesla.Therats wereinitially anesthetizedwith4%isouranein2.0L/minO 2 andthen4mg/kgofxylazinewas injectedsubcutaneously,alongwith2mLoflactatedRinger 'ssolution(Hospira,Lake Forrest,IL)tomaintainthephysicalconditionoftheratsd uringtheextendedMRI scans.Eachratwasplacedinaproneposition,inacustom-ma deMRI-compatible stereotaxicframeandcradle,toallowrepeatableposition ingandminimizemotion artifacts.Inthemagnet,anesthesiawasmaintainedwith1. 5–2.0%isouraneinO 2 at 79

PAGE 80

1L/min.Respirationandtemperatureweremonitoredandphy siologicaltemperature wasmaintainedusingheatedairowingovertheanimal(SAIn struments,StonyBrook, NY).Magneticresonancewasmeasuredusingacustom-built, saddleshaped470 MHzcoil,forbothexcitationanddetection,positionedont opoftherat'sheadand centeredoverthebrain.Invivoacquisitionparametersfor diffusion-weightedimaging andT2measurementshavepreviouslybeenpublished[ Parekhetal. 2010 ].Diffusion weightedimaging(DWI)datawerecollectedin27directions withab-valueof800 s/mm 2 andin6directionswithab-valueof100s/mm 2 ,andwerethenttedtoarank-2 tensortocalculateAD(averagediffusivity)andFA(fracti onalanisotropy).Aregionof interestwasmanuallydrawnforthemedialseptumtoquantif yAD,FA,andT2relaxation times.Aftertheratswereimagedinvivoatthe60dayspost-S Etimepoint,theywere transcardiallyperfusedandthebrainsexcisedandstoredi n10%formalin.Highangular resolutiondiffusionimaging(HARDI)datawerecollectedf romexcisedbrainsin46 directionswithab-valueof1000s/mm 2 andin6directionswithab-valueof100s/mm 2 [ Parekhetal. 2010 ].ThelogoftheimageintensityforeachvoxelforHARDIdata wastlinearlytoarank-2tensormodelofdiffusion(i.e.di ffusiontensorimaging,DTI) asafunctionofthediffusionweighting,andtheADandFAval ueswerecalculated fromtheresultingtensor[ Basser 1995 ]forthembriaandmedialseptum.Forber tracking,thediffusiondisplacementprobabilitydistrib utionfunctionateachvoxelwas estimatedusingtheMixtureofWisharts(MOW)approachtot acontinuousdistribution ofdiffusiontensors[ Jianetal. 2007 ].Thispost-processingmethodallowedforthe resolutionofcomplexberstructures,suchascrossingand kissingberswithinavoxel. Tractographywasperformedontheentireimagedbrainbysee dingallvoxelswiththe followingparameters:64seedspervoxel;astepsizeof0.5o fthevoxellength,with aturninganglethresholdof50;andanFAthresholdof0.05at eachstep.Thembria regionofinterestwasusedtothenlteroutbersthatpasse dthroughtheregion.Fibers 80

PAGE 81

thattraversedthecorpuscallosumwereexcluded.Allstati sticaltestsreportedare paired-samplesttestsunlessotherwisespecied. 7.3Results Tostudyanatomicalchangesinthemedialseptum,weperform edalongitudinal structuralcharacterizationwithhighelddiffusion-and T2-weightedMRI.From diffusion-weightedMRIdata,wecalculatedmeasuresofADa ndFA,whichprovide informationaboutthemagnitudeanddirectionalityofmole culardisplacement, respectively[ Pierpaolietal. 1996 ].InvivoMRIofthemedialseptumshoweda statisticallysignicantincreaseinADandareductioninT 2relaxometryat60days post-SE(Figure 7-2 A).TheincreaseinADissuggestiveofcellloss,andtheredu ced T2likelyreectsthesequesteringofironbymicrogliaanda strocytes,whichformscars inresponsetoneuronaldamage[ Zatta 2003 ].Changeswereseenat3dayspost-SE, reectinglocaledema.Whilethesechangesdonotreachsign icance,histological studiesinasimilaranimalmodelhaveconrmedearlycelllo ssintheseptalareas followingSE[ Gorteretal. 2001 2003 ].Sincethemedialseptumprojectstothe hippocampusprimarilybywayofthembria[ Petersonetal. 1987a ],weperformed bertrackinganalysisofthembriabasedonexciseddiffus ion-weightedimages. Thisanalysisrevealedastatisticallysignicant32.29.5 %reductioninbervolume 60dayspost-SE(Figure 7-2 B).Thechangesobservedinthemedialseptum(Figure 7-2 A)implythatthislossofbervolumecanbepartiallyaccoun tedforbyWallerian degenerationofmedialseptalprojectionstothehippocamp us[ Pierpaolietal. 2001 ]. Thesendingssuggestthatlossofcircadianinputtothehip pocampusmightemerge throughdegenerationofthemedialseptum. 81

PAGE 82

Figure7-2.Structuralchangesinthemedialseptumandmbr iainepileptogenic animals.A,Quantiedinvivoaveragediffusivity(AD),fra ctionalanisotropy (FA),andT2relaxationvaluesforthemedialseptuminpre-i njury,3days post-statusepilepticus(SE),and60dayspost-SEstagesfo rN=6rats. StatisticallysignicantchangesinADandT2areindicativ eofneuronlossin themedialseptum.*p < 0.05,relativetopre-injurymeasurement;#p < 0.05, relativeto3daysmeasurement.B,Fibertrackingofthembr iainexcised controlandspontaneouslyseizingratbrains60dayspost-S E.Asignicant reductionintotalbervolumeisobservedintheseizingrat s(N=8rats)when comparedtocontrolrats(N=3rats).*p < 0.05,unpairedttestrelativeto controlmeasurement.Valuesaremean SEM. 82

PAGE 83

CHAPTER8 HIPPOCAMPALEEGRHYTHMS:DISRUPTED24-HOURREGULATION 8.1BackgroundandOverview Inthischapter,weinvestigatelong-termchangesinthepow erofhippocampalEEG rhythmsinrattemporallobeepilepsy(TLE).Asdiscussedin Section 2.3 ofChapter 2 ,rhythmicactivityisamainfeatureinthehippocampalEEG. Likethespontaneous hippocampalspike(SPK)eventsstudiedinChapter 5 ,EEGrhythmscanserveasa markerfor24-hourregulationofhippocampalneuralactivi ty.Specically,sinceEEG rhythmsarewell-denedandareassociatedwithspecichip pocampalcelltypes,they canyieldadditionalinformationaboutneural-levelmecha nisms,complementarytoour SPKanalysis. Inourpreviouswork,weprovidedevidencethattheSPK24-ho urrhythmphase shiftwasnotassociatedwithchangesinwake-sleepcycleor suprachiasmaticnucleus (SCN)rhythms.Rather,weproposedamechanismbywhichcirc adianinputstothe hippocampusregulatedtheringactivityofspecicgroups ofneuronsona24-hour basis,andthatsubtlechangesintheseinputscouldproduce aphaseshiftinthe 24-hourrhythmofcellringactivity.Inourmodel,weobser vedthatbasketcellswere particularlysensitivetocircadianinput,andmostreadil yshiftedtheirringphase followingcircadianperturbation.Basketcellsareknownt oplayfundamentalrolesinthe generationofhippocampalgammarhythms(25–140Hz)[ Buzs akiandWang 2012 WangandBuzsaki 1996 Whittingtonetal. 2000 ],whichleadstothepredictionthat theserhythmsinparticularmightbeaffectedinourexperim entaldata.Additionally,in ourpreviousworkwepostulatedthatdamagetothemedialsep tummightbepartly responsibleforalteringhippocampal24-hourrhythms.Asd iscussedinChapter 2 ,the medialseptumisimportantforgenerationofthehippocampa lthetarhythm.Previous studieshaveshownthathippocampalthetarhythmamplitude isreducedinepilepsy andhavesuggestedthattheseptalstructuresresponsiblef orpacingthethetarhythm 83

PAGE 84

maybealtered[ Colometal. 2006 ].Therefore,basedontheseptum'sputativeroleasa circadianinput,weexpectthatthetarhythmpowermayalsob ealteredasafunctionof thetimeofday 1 Traditionally,thephrase“EEGrhythms”canhavetwodistin ctmeanings.First,EEG rhythmscanrefertospecicandoftentransientoscillatio nsthatareassociatedwith speciccognitiveactivities.Theseoscillationsmayoccu pydifferentfrequencyranges dependingonthespeciesandbrainregionsfromwhichtheyar erecorded[ Penttonen andBuzs aki 2003 ].Hippocampalgammaandthetaareexamplesofsuchoscillat ions. Incontrast,EEGrhythmsmayalsorefertoallactivitywithi nadenedfrequencybands, regardlessofbrainstate.Thesefrequencybandarealsoref erredtousingtermssuch asalpha,theta,andgammaandhaveinternationallyaccepte ddenitions[ Chatrian etal. 1974 ].ThedenedEEGfrequencybandswilloften,butnotalways, correlatewith thefrequenciesofspecicneuraloscillations.Forexampl e,thethetabandisdened internationallyas4–7.5Hz,butthetaoscillationsareslo werinlargeranimals(3–5 Hz),suchasthecat,andfasterinsmalleranimalssuchasthe rat(6–9Hz)[ Penttonen andBuzs aki 2003 ]. Inthischapter,ourapproachistoanalyzelong-termchange sinthepowerof hippocampalEEGrhythms.Ideally,wewouldliketotrackcha ngesintheamplitudes ofspecicEEGoscillatoryevents,sincethesecanbeclosel yrelatedtospecictypes ofneuralactivity.However,giventhelargevolumeofdatar equiredtostudy24-hour rhythms,itwasnotpossibletomanuallyannotateeveryEEGe vent.Forthetarhythms, weusedacommonautomatedroutinetoextractepochsoftheta activity[ Csicsvarietal. 1 Notethatthetarhythmoscillationsandthetastate,whichw eanalyzedinChapter 5 arenotequivalent.Thetaoscillationsareaphenomenonoft hethetastate.Forexample, extremecases,suchascompletemedialseptallesion,willa bolishthetaoscillationsbut leaveotherhallmarksofthethetastatestillapparent,suc hassuppressionofSPWs [ Buzsakietal. 1983 SuzukiandSmith 1988a ]. 84

PAGE 85

1998 1999 2000 ].Forotheroscillations,however,therearenot(toourkno wledge) reliableautomatedschemesforclassication.Therefore, wefollowedamorecommon approach,whichwastomeasurethepowerinspecicfrequenc ybandsacrossalldata. Wedid,however,choosefrequencybandstocorrespondtothe knownfrequencyranges ofrathippocampalEEGoscillationsreportedintheliterat ure[ Buzs akiandSilva 2012 Buzs akiandWang 2012 PenttonenandBuzs aki 2003 Traubetal. 1999b Whittington etal. 1997 ].Specically,weexaminedeightmainfrequencybands:del ta(1–5Hz), theta(5–10Hz),beta(12–25Hz),slowgamma(25–50Hz),midfrequencygamma (50–90Hz),fastgamma(90–140Hz),fastoscillations/ripp les(140–200Hz),and ultra-fastoscillations(200–500Hz). Inthefollowingsections,werstextractfrequencybandpo werasafunction oftimeusingpowerspectraldensity(PSD)analysis.Weshow thatallfrequency bandsexhibit24-houroscillations.Then,wedemonstratet hatcertainfrequency bands(namelythoseinthegammaandbetaranges)phaseshift followinginjury.To explainthisphaseshift,weintroduceaphenomenologicalm odelinwhichthereare multiplecircadianinputstothehippocampusandproposeth atthephaseshiftcould beproducedbychangesintherelativestrengthsofthesecir cadianinputs.Thisis supportedbyouridenticationofaformofphase–amplitude coupling,inwhichweshow that24-hourrhythmscanbegroupedinto2categoriesbasedo nphaseandamplitude changes,suggestingtheinuenceoftwoputativecircadian inputtothehippocampus. Furthermore,usingprincipalcomponentanalysis(PCA),we showthatmultipleseparate modesarerequiredtocapture95%ofthevarianceinourdata. Wediscusshowthis mightindicatetheexistencemultiplecircadianinputs.Fi nally,motivatedbyourprevious workandourinterestinthemedialseptum,weprovideadditi onalcharacterizationof 24-hourmodulationofthehippocampalthetarhythm.Weshow specicallythatthe magnitudeofthephaseshiftisreducedinthehippocampalth etastateandproposethat thethetastatemayimprovefunctionalcouplingtoaspecic circadianinput.Wealso 85

PAGE 86

showthatthephaseshiftandchangesinhippocampalthetadi dnotoccurinratsthat wereinjuredbutthatdidnotdevelopTLE.Previously,weper formedasimilaranalysisto thatpresentedhere,onlyusingempiricalmodedecompositi on(EMD)inplaceofPSD, anddemonstratedsimilarresults[ Stanleyetal. 2011b ]. 8.2Methods 8.2.1AnimalMethodsandDataCollection Animalmethods,experimentaltimelines,andproceduresfo racquisitionofrawdata aredescribedintheGeneralMethods(Chapter 4 ).Anexperimentaltimelineissupplied inFigure 4-1 oftheGeneralMethods.Aswithourpreviousanalysis,weana lyzeddata exclusivelyfromChannel2ofFigure 4-2 .Ourproceduresforttingof24-hourrhythms tocosinefunctionsandextractionofthetaepochsarealsod escribedtherein.Incontrast tothedatausedinthepreviousEEGchapter(Chapter 5 ),wenowhavedataforN=4 seizingratsandN=2non-seizingrats.Forconvenience,wew illlabelthefourseizing rats1–4andthenon-seizingrats5–6.Theadditionalseizin ganimal,Rat4,was notavailableforpreviousspikeanalysisbecausethespike sortingalgorithmfailedto classifyspikesofconsistentshape.8.2.2ExtractionofPowerSpectralDensities ToextractchangesinEEGpowerasafunctionoftime,werstd ividedtheEEG signalinto2-secondnon-overlappingepochs(Figure 8-2 A).Wethenapplieda 2-secondhammingwindowandcalculatedthePSD(Figure 8-2 B).Themeanpowers ineachoftheeightfrequencybandswererecorded.Thisproc esswasrepeatedfor every2-secondepochofdataandyielded,foreachrat,eight timeseriesdescribing EEGpowerasafunctionoftimeforeachoftheeightfrequency bands.Inouranalysis valuesreportedarethemeanPSDofafrequencyband,rathert hanthetotalintegrated power.ThisyieldsvaluesinunitsmV 2 /Hzandallowspowervaluestobeindependentof thewidthoftheirchosenfrequencyband. 86

PAGE 87

Insomecases,wedesirednerspectralresolution.Toimpro vespectralresolution, ratherthanusingthestandardfrequencybandslistedabove ,wesplitdataintomany narrowerfrequencybands.Specically,wesampleddelta(1 –5Hz)andtheta(5–10 Hz)asbefore,butthensubsequentfrequenciesweredivided into19logarithmically spacednon-overlappingbandsbetween10and500Hz.Deltaan dthetafrequency bandswereleftunchangedsincethesefrequencybandswereu sedfortheidentication ofthetaepochs.Itisindicatedinthegurecaptionscasesf orwhichtheselogarithmic frequencydivisionsareused.8.2.3DataAnalysis Subsequentsmoothingofdataandmeasurementof24-hourrhy thmsisas describedintheGeneralMethods(Section 4.9 ),butwithsomeimportantadditions. First,asbefore,timeseriesaresmoothedusing6-hour90%o verlappingwindows, yieldingtracesshowing24-hourvariation(Figure 8-2 C,dottedtrace).Detrendingis performedbysubtractingoutbaselinedrift,estimatedby2 -daywindows(Figure 8-2 C, solidgreenlines).Theresultingdetrendedtimeseriesare denotedwiththesymbol Dataarethenseparatedintopre-injury(Pre),post-injury latency(Post-L)and post-injuryspontaneouslyseizing(Post-SS)stagesandco sinoranalysis(seeGeneral Methods,Section 4.9 )isusedtoextracttheamplitudeandphaseof24-houroscill ations (Figure 8-2 D).Zeroamplitudetestsarealsoappliedtoensurethatdata areindeed oscillatory[ Nelsonetal. 1979 ]. Consideringthatourdatasetsweredrawnfromonly4animals and,yet,contained over23weeksofcombineddata,wehaveinsomecasesassumedt hatdataineach ofthethreeexperimentalstages(Pre,Post-L,orPost-SS)w ereergodic.Underthis assumption,wesplitdetrendeddataintonon-overlapping2 -daybins,conductedcosinor analysisoneachbin,andcollectedtheoutputforPre,PostL,andPost-SSexperimental stages.Outputdatafromallratswerepooledtogether,prov idingalargenumberofdata points.Althoughthedataislikelynottrulyergodic,espec iallyduringthelatencyperiod, 87

PAGE 88

wejustifythisassumptionbecausewefoundthatdatapoints forPost-LandPost-SS stageswerequitesimilar,especiallywhencomparedtopreinjurydata.Therefore, althoughthesystemischangingduringepileptogenesis,th isdoesnotappeartoexert majoreffectsontheparametersbeingmeasured.Thesesimil aritiesbetweenPost-Land Post-SSdatapointswereapparentbothwithandwithoutassu mingergodicity.Wenote ingurecaptionsinstancesinwhichweassumeergodicityan dusepooleddata. 8.2.4AnalysisofCorrelationBetweenPhaseandAmplitude Ourextractionof24-hourrhythmsandsubsequentcosinoran alysisyielded measurementsforboththephaseandtheamplitudeof24-hour rhythms.Thisanalysis wasconductedforeachfrequencybandinourinitialspectra ldecomposition.Inaddition toquantifyinghowthesevalueschangedasafunctionofinju ry,wealsosoughtto understandhowchangeinamplitudeandphasewerecorrelate dacrossanimals. Specically,forasingleratinthePost-Lstage,weconstru ctedvectors P =[ 1 ... M ] and A =[ A 1 ... A M ] .Thesestored,respectively,thephaseandamplitudeestim atesfor eachoftheMfrequencybands.Weusedthecorrelationcoefc ientbetweenthesetwo vectorsasanestimateforphase–amplitudecorrelation.Th iswasperformedforboth Post-LandPost-SSstagesforall4rats.Wedescriberesults forthisanalysisapplied tothefullsetofM=21logarithmicallyspacedfrequencyban ds(describedabove). However,resultswereobtainedusingthesetofM=8defaultf requencybands. 8.2.5PrincipalComponentAnalysis Principalcomponentanalysis(PCA)wasappliedinordertoi nvestigatethe dimensionalityofthedata.Datafromeachfrequencybandwe redetrended,as describedabove,andnormalizedtoastandarddeviationof1 .0andmeanofzero. Weshallrefertosuchdatafromthe i th frequencybandas x i ( t ) .Wethencalculated thecovariancematrixusing X i j = < ( x i ( t ))( x j ( t )) > .PCAwasappliedbynding eigenvaluesandeigenvectorsofthecovariancematrix,suc hthat X i = i i where i isthe i th eigenvalue, i isthe i th (column)eigenvector,and denotesmatrix 88

PAGE 89

multiplication.Forvisualizationpurposes,PCAanalysis wasappliedtodatafromthefull setofM=21logarithmicallyspacedfrequencybands.Simila rresultswereobtainedfor theM=8defaultfrequencybands.Eigenvectorsandeigenval uesweresortedinorderof descendingeigenvalue. The i th mode,denedastheprojectionofdataalongthe i th eigenvector,was calculatedas y i ( t )= x(t) i ,where x(t) isarowvector [ x 1 ( t ), x 2 ( t ),... x N ( t )] .We estimatedthefractionalcontribution 2 K oftherstKmodestothetotalvarianceofall frequencybandsby 2 K = K X i =1 i = (8–1) Inthisequation, = P i i isthetotalvarianceacrossallfrequencybands. 8.2.6Statistics AllstatisticaltestsreportedareWilcoxonranksumtests, unlessotherwise specied.Signicanceisconsideredtobep < 0.05.Errorbarsrepresent95%condence intervals. 8.3Results 8.3.1PhaseShiftin24-HourModulationofBetaandGammaFre quencyRhythms Tobeginouranalysis,werstusedanidealnotchltertospl ittherawdatainto eightfrequencybands.Thisprovidedavisualizationofthe intrinsicrhythmicactivity intheEEGsignal.ThisisillustratedinFigure 8-1 .Inparticular,thetabandactivityis apparentandisdetectedbythethetaepochdetectionalgori thm(indicatedinred). Secondly,weillustrateourprocedurefortrackingoflongtermchangesinEEG. InFigure 8-2 A–CshowsanexamplePSDanddemonstrateshowlong-termchan ges inthebetafrequencyband(12–25Hz)powerareextractedand smoothed(see Methods).Theresultingtimeseriesexhibitsclear24-hour modulation.Figure 8-2 D 89

PAGE 90

showshowthisbandexhibitsaphaseshiftfollowinginjury, similartoourndingsfor EEGSPKs. WhileFigure 8-2 providesanexampleofthephaseshiftforasinglefrequency band,ourcompletecharacterizationincludedallfrequenc ybandsforallrats,(Figure 8-3 ).Inparticular,wefoundthat24-houroscillationswereub iquitousforallfrequency bands.Lowerfrequencyrhythms(delta,theta)wereentrain edsuchthattheiramplitudes peakednearnoon,whilehigherfrequencyrhythmspeakedclo setomidnight.Thiswas thecasethroughoutallexperimentalstages(Pre,Post-L,a ndPost-SS).Inthemiddle frequencyranges,generallybetaand/orgamma,aphaseshif tappearedfollowinginjury, inwhichthepeakshiftedfromnoontomidnight.Sampledataf romRat1areshownin Figure 8-3 A,B.Inparticular,itisclearthat24-hourrhythmscanbecl usteredintotwo groups:thosethatpeakduringthedayandthosethatpeakdur ingthenight(dottedred line).Additionally,weobservedthatthephaseshiftappea redveryearlyinthelatency stage,almostimmediatelyfollowingstatusepilepticus(S E)injuryforallrats(Appendix B ,Figures B-1 – B-4 ). Statisticswereconductedonthephaseshiftforallanimals .Figure 8-3 Cshows thatthephaseshiftwassignicantforbeta,slowgamma,and mid-frequencygamma frequencybands.Statisticsareasfollows:beta( p =0.001 and p =2.3 10 5 for latencyandspontaneouslyseizingstages,respectively), slowgamma( p =0.0016 and p =3.1 10 5 ),andmid-frequencygamma( p =0.01 and p =0.0055 ). Wealsoanalyzedchangesintheamplitudesofcosinetsfoll owinginjury.We foundstatisticallysignicantreductionsinamplitudeth atoccurredexclusivelyforthelow frequencybands,delta,theta,andbeta(Figure 8-3 D).Statisticsareasfollows:delta ( p =0.17 and p =0.0047 forspontaneouslyseizing),theta( p =0.0036 and p =0.0088 ), andbeta( p =0.0049 and p =0.048 ). 90

PAGE 91

8.3.2ImbalancedCircadianInputasaDriverforthePhaseSh ift Wehavethusfarobservedthefollowing:24-hourmodulation oflowfrequency rhythmspeakedduringtheday,24-hourmodulationofhighfr equencyrhythmspeaked atnight,andmiddlefrequencyrhythmsshiftedfrompeaking duringthedaytopeakingat night.Furthermore,lowfrequencyrhythmsexhibitedastat isticallysignicantreduction intheamplitudeoftheir24-hourmodulationfollowinginju ry.Herewewilldiscussthese ndingsandproposeasimplephenomenologicalmodel. Whilethehippocampusclearlypossesses24-hourrhythms,i tlikelydoesnot receivedirectprojectionsfromtheSCN.Rather,anumberof indirectpathwaysfor relayingcircadianinputhavebeenproposed[ GerstnerandYin 2010 ].Inthisstudy, wehaveobservedthat24-hourrhythmscanbegroupedintotwo categories:those thatpeakduringthedayandthosethatpeakduringthenight. Weproposethatthese mightcorrespondtotwoseparatecircadianinputs(Figure 8-4 ).Ifthisisthecase,then thephaseshiftmightbeexplainedbyatransferincontrolfr omthe“ Day ”inputtothe “ Night ”input.Thisissupportedbytheobservationthatthelowfre quencyrhythms, whichcorrespondedtothedaytimepeak,showedareductioni ntheiramplitudeof 24-hourmodulation.Suchatheoryisinlinewithourpreviou smodelingwork(Chapter 6 ; Stanleyetal. [ 2011a 2013 ]),whichsuggeststhatthephaseshiftmightresultfrom subtleimbalancesincircadianinput.Insubsequentsectio ns,weshallrefertothesetwo putativecircadianinputsas Day and Night 8.3.3Phase–AmplitudeRelationshipof24-HourRhythms Thephenomenologicalmodeldescribedaboveproposesthatt herearetwomain circadianinputstothehippocampus,eachwithuniquephase s(daytimeandnighttime peaks)andamplitudes.Therefore,weexpecttheretobesome relationshipbetween phaseandamplitudeinthe24-hourrhythmsrecordedexperim entally.Figure 8-6 A,B displaysamplitudeandphasedatafromall4rats.Toshowtha tourresultsarenot dependentonthechoiceoffrequencybands,asopposedtousi ngthetraditionalEEG 91

PAGE 92

frequencybands,wenowsplitdatainto19logarithmicallys pacedfrequencybands, plusthetaanddelta. 2 Inthisdata,weobservedapatternwherebyfrequencybands possessingadaytimepeakinpowergenerallyexhibitedadec reaseinamplitudeoftheir 24-hourmodulationpost-injury.Theconversewastrueforn ighttimepeakingfrequency bands,whichshowedanapparentincreaseintheiramplitude post-injury.Thereare someexceptionstothistrend,notablythehighfrequencies ofRat1andtheentire latencyperiodofRat4. 3 Toquantifythistrend,weexaminedthecorrelationcoefci ent betweenthephaseandamplitudechanges(seeMethods,Secti on 8.2.4 )andfounda positivecorrelationacrossall4rats.Thatis,ahigherT PSDmax (meaninganactivitypeak closerto24:00)correspondedtostronger24-hourrhythmsp ost-injury.Incontrast,lower T PSDmax (valuesgenerallydidnotdropbelownoon)correspondedtor educedamplitude 24-hourrhythms.Thiswasthecaseduringbothlatencyandsp ontaneouslyseizing stagesforall4rats,withcorrelationvaluesrangingbetwe en0.33and0.85andaverage valuescloseto0.5.Together,thesendingssuggestthatth erearetwomaincircadian inputstothehippocampus,andthattheirstrengthisdiffer entiallymodulated,withone increasingandtheotherdecreasingfollowinginjury. 4 2 Similarresultswereobtainedusingtraditionalfrequency bands. 3 ThefactthatthelatencyperiodofRat4showedauniformredu ctioninthe amplitudeof24-hourmodulationacrossallfrequencybands mightrelatetothefact thatRat4exhibitedarelativelyshortlatencyperiodandth atall24-hourrhythms appearedtobetemporarilyreducedfollowingSE.Theamplit uderecoveredandthe trendre-appearedforthespontaneouslyseizingstageofep ileptogenesis.SeeAppendix B ,Figure B-4 4 Wenotethat,whilehighfrequencybandsdidshowarelativel yconsistentincrease inpowerfollowinginjuryineachanimal(Figure 8-6 B)thisdidnotreachsignicance whendatawerepooledandaveragedacrossallanimals(seeFi gure 8-3 fordescription). 92

PAGE 93

8.3.424-HourRhythmsareMultidimensional Inthissection,weexaminecorrelationrelationsamongstt hevariousfrequency bands'24-hourrhythmsusingPCA.PCAisadimensionalityre ductiontechniquethat identiesasetoflinearlyuncorrelatedvariablesthatpro videareduceddescriptionofa multidimensionalsignal.Thesevariablesdescribemotion inavectorspaceofreduced dimension(givenbyprincipalcomponenteigenvectors),wi theachvariabledeningan uncorrelatedmodeoftheoriginalsignal.Asweshalldiscus s,suchuncorrelatedmodes mayrevealinformationabouttheunderlyingcircadianinpu tstothesystem. OurPCAanalysis(Figure 8-5 )revealedthatatotalof3modeswererequiredto capture95%ofthesignalvariance(Figure. 8-5 A).Theseincludeamodethatpeaks duringthenightandcontributestooscillationsofthehigh frequencybands,amodethat peaksduringthedayandcontributestolowfrequencybands, andadiffusemodethat exhibitsaslightpeakintheeveninghours(Figure 8-5 B,C).Allthreeprincipalmodes forallratspassedthezeroamplitudetest[ Nelsonetal. 1979 ]. 5 Itisimportantto emphasizethatwechosetoanalyzeasinglestageofepilepto genesis,ratherthanthe wholedataset,inordertoensurethatthedatawererelative lystationary.Wechosethe spontaneouslyseizingstagebecauseitwasthelongestandt husprovidedthemost data. TheseresultsstatethattheEEG24-hourrhythmscanbelarge lyreconstructed bythelinearsummationofthreeuncorrelatedmodes.Theexi stenceofmultiple uncorrelatedmodessupportsthenotionofmultiplecircadi aninputs,since,abstractly, 5 Itissignicantthatthesemodesbe24-hoursinusoidalrhyt hmsbecausedifferent EEGrhythmsaregeneratedbyspecicunderlyingmechanisms ,andtheirpoweris modulatedbymanynon-circadianfactorsthatoperateonava rietyoftimescales. Dependentonourchoiceoflteringtechniques,thesemodul ationswillalsoappearin oursignalsalongside24-houroscillations.Sincethereis noreasontoexpectthatthese otherfactorsshouldbecorrelated,theymayappearasaddit ionaldimensionsinour data.Therefore,itisimportanttoconrmthatthemodesare 24-hourrhythms. 93

PAGE 94

wewouldexpectrhythmsproducedbymultipleindependentos cillatorswouldbe lesscorrelatedthanrhythmsproducedbyinputfromasingle oscillator,suchasthe SCN.Oneinterpretationisthattheseuncorrelatedmodesma ycorrespondtospecic circadianinputs,suchasthosedepictedinFigure 8-4 .Specically,thefactthatthe rsttwomodesconcentratetheircontributionstothehigha ndlowfrequencybands, respectively,supportstheorganizationof Day and Night driversthatwehavepredicted. Wenotethatthisresultisnon-trivialbecause,basedonFig ure 8-3 ,wemighthave expectedPCAtorevealthatasingle24-hourrhythmicmodewa ssufcienttocapture 95%ofthesignalvariance.Suchasinglemodecouldproduced ayandnightpeaking 24-hourrhythmssimplybyinvertingthesignofeigenvector entrieswhereneeded.Such andingwouldinsteadsuggestthepresenceofasingledomin antcircadianinput. 8.3.5Altered24-HourModulationofThetaRhythmPower Whilewehaveproposedabstractentitiesforthecircadiani nputsshownin Figure 8-4 ,anumberofstudieshavesuggestedspecicentitiesthatmi ghtbetheir physiologicalcorrelates.ThesearesummarizedinChapter 3 ,Section 3.6 .Inparticular, ourpreviousworksuggestedthatcircadianinputviathemed ialseptummightbecome impairedandcontributetothephaseshiftingofhippocampa lSPKs.Thiswassupported bybothadetailedcomputermodel(Chapter 6 )andanMRIstudythatsuggested damagetothemedialseptumpost-injury(Chapter 7 ). Themedialseptumisrmlyestablishedthrough invivo studiestobecriticalfor thegenerationofhippocampaltheta[ Leeetal. 1994 Petscheetal. 1962 ].Givenits importance,ourhypothesisthatthemedialseptumisimpair edwouldbefalsiedbya lackofalterationtothethetarhythm.Onthecontrary,whil echangesinhippocampal thetarhythmmayindicatethepossibilityofmedialseptumd amage,changestothe hippocampusorotherstructurescouldalsoproducesimilar effects Colometal. [ 2006 ]. Preliminaryanalysisofthethetafrequencybandsuggested thatits24-hourmodulation wasreduced(Figure 8-3 D).Here,wewillinvestigatethetarhythmsinmoredetail. 94

PAGE 95

Weusedastandardtechnique,describedinChapter 4 ,toextractepochsoftheta rhythmactivitybasedontheratioofpowerinthetaanddelta frequencybands.InFigure 8-7 A,weinvestigatedthepowerinthethetafrequencyspecica llyduringepochs ofthetaactivity.Wefoundthatthetapowerwassignicantl yreducedatalltimesof theday.Thisisinagreementwithndingsinasimilarepilep symodel[ Colometal. 2006 ]andmayreectdamagetotheunderlyingthetageneratingst ructures.The greatestreductioninthetapoweroccurednearnoon,andthe implicationsofthiswillbe addressedinthediscussion.Figure 8-7 Bshowsastatisticallysignicantreductionin the24-hourmodulationofhippocampalthetarhythms. 6 Therefore,inadditiontothe thetarhythmitselfbeingreduced,its24-hourmodulationi salsoimpaired. InFigure 8-8 ,wesplitdataintothetaandnon-thetastatesandrepeatedo ur characterizationofEEGphase(asinFigures 8-2 and 8-3 ).Weobservedthatthe phaseshiftstillappearedforbetaandgammafrequencyrhyt hmsinboththethetaand non-thetastates;however,thephaseshiftwasmorepronoun cedinthenon-thetastates. Specically,post-injurythebetaandgammabandspeakedsi gnicantlyearlierinthe dayinthethetastate,ascomparedtonon-thetastate(Figur e 8-8 B,C).Intermsofour phenomenologicalmodel(Figure 8-4 ),thismightsuggestthat,inthethetastate,the circadianinputspromotingdaytimeactivitypeakwererela tivelyenhanced. 8.3.624-HourRhythmsinNon-SeizingRats Finally,fornon-seizingrats(Rats5and6),werepeatedthe majorfacetsofour characterizationabove.ThisissummarizedinFigure 8-9 .DetrendedPSDdatafor selectedfrequencybandsareshownand,quantifyingthepha seofthisactivity,we observedthatthephaseshiftmagnitudewasmuchreducedfor non-seizinganimals (Figure 8-9 A,B).Thechangesinphasedidnotreachstatisticalsignic ance,nordid 6 ThisisalsovisibleinFigure 8-7 A,butlessclearlybecausedatawerenot detrended. 95

PAGE 96

thechangesinamplitude(Figure 8-9 C,D).Finally,thethetarhythmproleshowed consistentpatternspre-andpost-injury(Figure 8-9 E,F).Together,theseresultssuggest thatthechangesin24-hourmodulationandhippocampalthet aareassociatedwith epileptogenesisandarenotmerelyanepiphenomenonofinju ry. Figure8-1.FrequencycontentofrawEEGdata.A,Twodaysofc ontinuousrawEEG datatakenfromthepre-injurystage.B,Toptraceshows12se condsofraw data.Bottomeighttracesshowdecompositionofrawdataint ofrequency bands(indicatedinHz).Theamplitudeofthetoptwofrequen cybandshas beenmagniedbyafactorof5forimprovedvisualization.Ar rowsindicate 2-secondepochsofdatadetectedastheta-statebasedonthe theta/delta powerratiomethod.Datawereseparatedintofrequencyband susingan ideal(noncausal)lter. 8.4Discussion OurEEGstudieshaveproducedthefollowingndings.First, wehaveshownthat thepowerofhippocampalEEGrhythmsismodulatedona24-hou rcycle.Secondly, theserhythmsarealteredinepilepsy,withbetaandgammafr equencyrhythms exhibitingaphaseshiftanddelta,beta,andthetafrequenc yrhythmsexhibiting 96

PAGE 97

Figure8-2.Long-termtrackingofpowerspectraldensity(P SD)revealsaphaseshift followinginjury.A,A12-secondsampleofrawdatatakenfro mthepre-injury stage.A2-secondepochofdataishighlightedingreen.B,PS Dofthegreen 2-secondepochofdata.Eighthorizontalbarsindicatethem eanpowerin eachofthefrequencybandsusedinouranalysis.Thehorizon talpositionof thebardenotesthefrequencyrangeoftheband.C,Longtermc hangesin thebeta-rangefrequencyband(12–25Hz)powerwereestimat edforthe entiredatasetandsmoothedby6-hour90%overlappingslidi ngwindows (dotted).Additionally,baselinedrift(solidgreen)issh ownasestimatedusing a2-dayslidingwindow.Dataareshownforeachofthethreeex perimental stages:pre-injury(Pre),post-injurylatency(Post-L)an dpost-injury spontaneouslyseizing(Post-SS).D,Thebaselinedriftwas subtractedout andtheresultingdetrendedtimeserieswasexpressedinasi ngle24-hour window(modulo24).Cosinetsareshowninred.Aphaseshift of 180 degreesisvisiblebetweenPreandPost-LstagesandPreandP ost-SS stagesforthisparticularfrequencyband.Alldatashownis fromRat1. 97

PAGE 98

Figure8-3.Phaseandamplituderelationshipamongstfrequ encybands.A)Cosinets todetrendedPSDdata( PSD)forselectedfrequencybandsforRat1. ModulationofEEGrhythmsinthethetaanddeltafrequencyba ndsis entrainedtopeaknearnoon,whilethegammafrequencybands peaknear midnight.Thebetafrequencybandexhibitedaphaseshift,f rompeaking duringthedaytopeakingatnight.Reddottedlineillustrat esdivision betweendayandnightpeakingrhythms.DataarefromRat1.Ot herrats showedphaseshiftsalsooccupyinggammafrequencybands.B )Timeof maximum PSD(T PSDmax ),asextractedfromcosinetsforallfrequency bands.FrequencybandsareasdenedinFigure 8-1 .C)Examinationofthe phaseshiftacrossallratsrevealedthatthephaseshiftwas statistically signicantforbeta(12–25Hz),slowgamma(25–50Hz),andmid-frequencygamma(50–90Hz)frequencybands.D)Examina tionofthe amplitudeofcosinetsshowedthatthemagnitudeof24-hour modulationin delta,theta,andbetafrequencybandsreducedfollowingin jury.Certain higherfrequencybandsshowedanincreaseinamplitudepost -injury,but thesedidnotreachsignicanceandarenotshownduetospace limitations. *p < 0.05byWilcoxinranksumtestonpooleddata(N=16,20,and55 2-daybinsforPre,Post-L,andPost-SSdata,respectively) .Pre,Post-L,and Post-SSrefertopre-injury,post-injurylatency,andpost -injuryspontaneously seizingstagesofepileptogenesis,respectively. 98

PAGE 99

Figure8-4.Alteredcircadianinputasamechanismforthe24 -hourrhythmphaseshift. TheobservationthatEEGrhythmscouldbeclearlydividedin totwogroups, thosethatpeakduringthedayandthosethatpeakatnight,su ggeststhat theserhythmsaredrivenbytwomaincircadianinputstotheh ippocampus. Wehypothesizedthatreductioninthestrengthofthedayinp utmight produceaphaseshiftbypromotingatransferincontroltoth enightcircadian input.Redarrowsindicatethepossibilitythat24-hourmod ulationofthe middlefrequencymaybetransmittedindirectlythroughcou plingtoother EEGrhythmgenerators,ratherthanbydirectinputfromcirc adianrelay centers. areductionintheiramplitudeof24-hourmodulation.Throu ghanalysisofdata dimensionalityandalsophase–amplituderelationships,w ehaveprovidedevidence thattherearetwomaincircadianinputstothehippocampusa ndhavesuggestedthat theirstrengthisalteredfollowinginjury.Finally,wehav eshownthat24-hourregulation ofthetarhythmsisalsoalteredinepilepsyandthatthethet astateappearstopromote activitytopeakearlierintheday.Inthissection,wewilld iscusstheseresultsinrelation topreviousliterature,proposeaspecicphysiologicalme chanismthatcouldexplain thesendings,andsuggestfuturestudies.8.4.1RelationshiptoPreviousCircadianLiterature Therehavebeenfewpreviousinvestigationson24-hourrhyt hmsofhippocampal neuralactivityinepilepsy.Ourpreviousworkinthisareas howedthatanotherEEG feature,hippocampalSPKs,alsoexhibitedaphaseshift(Ch apters 5 – 7 ; Stanleyetal. [ 2013 ]).Inthispreviousstudy,weacknowledgedthepossibility thattheappearanceof 99

PAGE 100

Figure8-5.PrincipalcomponentanalysisofEEGfrequencyb andssuggestsmultiple circadianinputs.A)Principalcomponentanalysis(PCA)wa sappliedtothe detrendedtimeseriesforeachrat(Appendix B ,Figure B-1 – B-4 ). 2 K representsthefractionalcontributionoftherstKmodest othetotal varianceinallfrequencybands(seeMethods).Therstthre emodesare sufcienttocapture95%ofthevariance(dottedline).B)Th ethreeprincipal modes, y 1 ( t ) ... y 3 ( t ) ,whichrepresentdataviewedintheeigenvectorspace. Therstofthesemodespeaksatmidnight,thesecondatnoon, andthe thirdisdistributed.C)Theeigenvectorsassociatedwithe achofthethree principalmodesshowconsistentpatterns,with 1 contributingprimarilyto highfrequencybands, 2 tolowfrequencybands,and 3 toallbutthe middlefrequencybands.In(A,C)populationmeans(red)are shown alongsidevaluesfromindividualrats(black).Datafroman examplerat(Rat 1)isshownin(B).PCAwascalculatedspecicallyusingdata fromthe spontaneouslyseizingstage,whichwaschosenbecauseitwa sthelongest ofthethreeexperimentalstages. 100

PAGE 101

Figure8-6.Post-injurychangesinamplitudearecoupledto phaseA)Timeofmaximum PSD(T PSDmax ,asextractedfromcosinets)forRats1-4(i–iv).B)Percen t changeinamplitudeofPSD24-hourmodulation(A PSD ,asextractedfrom cosinets),relativetopre-injuryperiod.Weobservedtha tforregionswhere A PSD decreasedfollowinginjury,thecorrespondingregionsoft hephase plots(A)showedrhythmsweregenerallyentrainedtoanoontimepeak. Similarly,regionswhereA PSD increasedfollowinginjuryappeared associatedwithmidnightT PSDmax in(A).C)Wequantiedthiscorrelationby measuringthecorrelationcoefcientbetweenamplitudech angeandphase forallfourratsandforthetwodifferentstagesofepilepto genesis.Apositive correlationwasshownforallrats(crosses),withmeanvalu es > 0.5(bars). Whiteshadingindicatesdataforwhichcircadiantscouldn otbereliably obtained(zeroamplitudetestfailure). 101

PAGE 102

Figure8-7.Altereddailymodulationofhippocampalthetap owerfollowinginjuryA)We measuredmeanthetaPSDduringeight3-hourtimewindowsthr oughoutthe day.Weobservedastatisticallysignicantreductioninth etathatwas greatestclosetonoon.*p < 0.05byWilcoxinranksumtestonpooleddata (StatisticsbasedonatleastN=29,39,and933-hourtimewin dowsforPre, Post-L,Post-SSdata,respectively.Nforeachtimewindows varieddueto gapsindata).B)Amplitudeofthe24-hourrhythminthetheta frequency band,asestimatedfromcosinets.Astatisticallysignic antreductionwas foundpost-injury.Similarresultswerefoundforthedelta andbetabands.*p < 0.05byWilcoxinranksumtestonpooleddata(N=16,20,and55 2-day binsforPre,Post-L,andPost-SSdata,respectively).Forb oth(A)and(B), datawereanalyzedspecicallyfromepochsofdataidentie dasbeing theta-state,accordingtoourthetadetectionalgorithm(s eeMethods).Pre, Post-L,andPost-SSrefertopre-injury,post-injurylaten cy,andpost-injury spontaneouslyseizingstagesofepileptogenesis,respect ively. theSPKphaseshiftmightinfactbecausedbytheemergenceof interitalspikesrather thananactualchangeincircadianrhythms.Ouridenticati onhereinofasecondary featureinthehippocampalEEGsuggeststhatthephaseshift isnotafunctionofour choiceoffeature.Secondly,thisstudyalsoproposedthats ubtleimbalancesincircadian inputstrengthcouldgenerateaphaseshift,andthiswasdem onstratedinabiophysical computermodel.Thisproposedmechanismissupportedbyour presentndings,which provideevidenceformultiplecircadianinputsthataredif ferentiallyregulatedfollowing injury. Otherstudieshavealsosupportedthenotionofchangesinci rcadianregulation inepilepsy.Inparticular, Matzenetal. [ 2012 ]usedsingle-andpaired-pulseresponse 102

PAGE 103

Figure8-8.Phaseshiftisenhancedinnon-thetastates.Two -secondepochsofdata wereclassiedaseitherthetaornon-thetausingastandard classication method(seeMethods).Then,thetimeofmaximum PSDwascalculatedas inFigure 8-3 Cforeachofthethetaandnon-thetadatasets T PSD j stage .A-C) pre-injury(Pre),post-injurylatency(Post-L)andpost-i njuryspontaneously seizing(Post-SS)stagesareshown,respectively.Aswitht hecasewhen thetaandnon-thetaepochswerepooledtogether(Figure 8-3 C),aphase shiftwasobservedinthemiddlefrequencyranges.However, theshiftin phaseforthethetastatedatawassignicantlylessthantha tfornon-theta. Frequencybands1–8areasdenedinFigure 8-1 .*,#p < 0.05byWilcoxin ranksumtestonpooleddata(N=16,20,and552-daybinsforPr e,Post-L, Post-SSdata,respectively).#indicatessignicancerela tivetopre-injury datapointand*indicatesnon-thetadatapointissignican tlydifferentfrom correspondingthetadata. 103

PAGE 104

Figure8-9.Analysisofphaseandamplitudechangesfornonseizingrats.A)Cosinets todetrendedPSDdataforselectedfrequencybandsforRat6. B)Timeof maximum PSD(T PSDmax ),asextractedfromcosinetsforRats5-6(i–ii). C)Examinationofphasedataacrossbothnon-seizingratsre vealedno signicantchangeinphasepostinjury.Frequencybandsasi nFigure 8-3 C. D)Likewise,amplitudedataacrossdidnotrevealanysigni cantchange.E) AsinFigure 8-7 A,meanthetaPSDwasmeasuredduringeight3-hourtime windowsthroughouttheday.Therewasnosignicantreducti onintheta,and thedistributionmaintaineditspeaknearnoonfollowingin jury.F)The amplitudeofthe24-hourrhythminthethetafrequencybandw asestimated basedoncosinets.Unlikeforthecaseofseizingrats(Figu re 8-7 B),we didnotobserveasignicantreductionpost-injury.Forbot h(E)and(F),data wereanalyzedspecicallyfromepochsofdataidentiedasb eing theta-state(seeMethods).Wilcoxinranksumtestswereapp liedtopooled datafromnon-seizingrats,andsignicantdifferencesbet weenpre-and post-injurystateswerenotobserved(N=4and92-daybinsfo rPreandPost forpanels(C),(D)and(F);Datainpanel(E)obtainedfromat leastN=6and 163-hourtimewindowsforPreandPost.Nforeachtimewindow svaried duetogapsindata).Pre,Post-L,andPost-SSrefertopre-in jury,post-injury latency,andpost-injuryspontaneouslyseizingstagesofe pileptogenesis, respectively. 104

PAGE 105

measurementstosuggestthathippocampalexcitationwasen hancedduringtheday andthatinhibitionwasenhancedatnightinananimalepilep symodel.Thesechanges, theyproposed,mightbecausedbyalteredrhythmsofhormone releasefromthe hypothalamus.Additionally,asmentionedabove, Chaudhuryetal. [ 2005 ]conducted experimentsonmelatoninknockoutmiceandshowedthat24-h ourmodulationof hippocampalactivitywasaltered,butnoteliminated.This suggeststhat,inaddition tomelatonin,otherhippocampalinputswerecontributingt othegenerationof24-hour rhythms.8.4.2InterpretationofResults8.4.2.1Changesin24-HourRegulationofHippocampalTheta Hippocampalthetaactivitywaspreviouslyinvestigatedin apilocarpinestatus epilepticusanimalmodelby Colometal. [ 2006 ].Conrmingtheirobservations,we haveshownareductionintheamplitudeofhippocampaltheta powerfollowinginjury. Furthermore,quantifyingthesechangesasafunctionofthe timeofday,wefoundthat thegreatestreductioninhippocampalthetapoweroccurred closetonoon. Abstractly,therearetwogeneralmechanismsthatmightacc ountforthereduction ofthetapoweratnoon.First,assumethat24-hourrhythmsof thetapoweraredrivenby acircadianinputthatpeaksatnoonandthatpromotesincrea sedthetaactivitypower. Then,lossofthisinputcouldproduceareductioninthetath atisgreatestatnoon. Alternatively,assumethat24-hourrhythmsofthetapowera redrivenbyacircadian inputthatis inhibited atnoonandthatthisinputactstodiminishthetaactivitypo wer.In thiscase,lossofthisnoon-timeinhibition(i.e.disinhib ition)mightalsocontributetoan increaseinthetaactivity. Therearemanycandidatesforcircadianinputsthatcouldin uencethetarhythm power.Onecandidateisthemedialseptum,thecholingergic andGABAergicseptal 105

PAGE 106

outputsofwhichhavebeensuggestedtopromoteincreasedth etaactivitypower. 7 Thus,assumingthemedialseptalinputpeaksatnoon,damage tothisinputcould explainthenoon-timereductioninhippocampaltheta. Wenotethatastudyon24-hourrhythmsby Yamazakietal. [ 1998 ]showedthat multi-unitactivity(MUA)inthemedialseptumpeakedatnig htratherthanduringthe day.However,itisdifculttotranslatetheresultsofthei rstudytooursbecausetheir datawascollectedcontinuously,regardlessofbehavioura lstate.Sincebehavioural states,suchaswakefulnessandsleep,canalsoinuenceneu ralactivity, 8 itisdifcult todistinguish24-hourrhythmsdrivenbybehaviourfromtho sedrivenbytheSCN. 9 Intheirdata,themedialseptumMUAappearswellcorrelated withwheelrunning activity(seeFigure2Bof Yamazakietal. [ 1998 ]).Thissuggeststhatthe24-hour rhythmstheyhavereportedmaybemainlydrivenbybehaviour alactivity.Incontrast, ourmeasurementsoftheta-activitypowerwereperformedex clusivelyinthethetastate and,hence,shouldbelessdependentonbehaviour.Ifourpre dictionfortheroleofthe medialseptumiscorrect,wewouldexpect24-hourrhythmsof medialseptumactivity, exclusivelywithinthethetastate,topeakduringthedayas opposedtoduringthe night.Experimentalmeasurementofseptal24-hourrhythms insuchastate-dependent 7 Thisissupportedbybothlesioningandpharmacologicalstu dies:1) Leeetal. [ 1994 ]showedthatselectivecholinergiclesionofthemedialsep tumreducestheta rhythmpower.2) Smytheetal. [ 1992 ]foundthatbicuculine,whichattenuatesGABA, wasabletorestorethetafollowingseptalinactivation.Th eyarguedthereforethatlossof septalGABAergicinputmightreducehippocampalthetabypr omotinghyperinhibition. 3) Buzs akietal. [ 1986 ]showedurethane,whichingeneralpotentiatesinhibitory transmitters(thusalsopromotinghyperinhibition)andat tenuatesexcitatorytransmitters [ HaraandHarris 2002 ],reducesthetapower. 8 Indeed, Hassanietal. [ 2009 ]haveshownthatdifferentcellspresentinthemedial septumeachexhibitacomplexarrayofringpatternsacross wake–sleepbehavioural stages. 9 Thisisonemotivationforinvestigating24-hourrhythmsin boththethetastate (Figures 8-7 and 8-8 )andinthefulldataset(Figure 8-3 ). 106

PAGE 107

mannerwouldbeusefulforinterpretingourresultsandmays uggesthowtheinuence oftheSCNintegrateswithother24-hourrhythms.8.4.2.2EffectofThetaStateonPhaseShift Wehaveshownthat,followinginjury,thepeakactivityinbe taandgamma frequencybandswaslessshiftedtowardsmidnightinthetas tates,ascomparedto deltastates(Figure 8-8 ).EEGrhythmssuchasthetaservethepurposeofincreasing functionalcouplingacrossbrainregions[ Buzsaki 2009 ].Alargenumberofstructures participateinthehippocampalthetaoscillation,suchast hemedialseptum,the supramammillaryregion,theamygdala,andalargenumberof corticalstructures [ Buzs aki 2002 ].Oneinterpretationofthisndingisthatthethetastatet emporarily increasesthecouplingbetweenthehippocampusandtheputa tivecircadianinput responsibleforgeneratingthe Day rhythm(Figure 8-4 ),therebyencouragingapeak inactivitythatisearlierintheday.Analternateinterpre tation,isthat,sincethere aredifferencesbetweenthetaandnon-thetagamma[ Leung 1998 TraubandMiles 1991 ],thedifferenceinpeaktimescouldsimplybetheresultoft hecircadiansystem differentiallyregulatingthesetwotypesofgamma.8.4.3UnderlyingPhysiologicalMechanisms BasedontheanalysisofEEGsignals,wehaveprovidedeviden cethatthereare twomaincircadianinputstothehippocampus.Thisraisesac entralquestion:what arethephysiologicalcorrelatesofthesetwocircadianinp uts?Thereareagreatmany candidatecircadianinputstothehippocampus,summarized inChapter 3 ,Section 3.6 However,inadditiontosimplyprovidinga24-hourrhythmic signaltothehippocampus, thecircadianinputinquestionmustalsoA)havetheability tomodulateEEGrhythms andB)becomealteredduringepileptogenesis.Whilethispl acessomerestrictionson whichinputscouldberelevant,damageassociatedwithSEis widespreadandmany brainregionsbecomealtered[ Parekhetal. 2010 ]. 107

PAGE 108

Themedialseptum,givenitsrichinteractionwithhippocam paldynamics,meetsthe criteriaofbeingabletomodulatehippocampalEEGrhythms. Anumberofstudieshave alsosuggestedthatitmaybedamagedoralteredduringepile ptogenesis.Concerning itsinuenceonneuralrhythms,themedialseptumhasbeenin vestigatedinanumber ofempiricalstudies.Itisrmlyestablished invivo thatthemedialseptumiscritical forthegenerationofhippocampaltheta[ Leeetal. 1994 Petscheetal. 1962 ]. Invivo studieshavealsosuggestedthatbothseptallesionandphar macologicalmodulation viaatropine,acholinergicantagonist,caninuencehippo campalbetaandgamma oscillations(20–50Hz)[ Leung 1998 ].GABAergicseptalactivityhasalsobeenshown toinuencegammafrequencyrhythmsby invivo pharmacologicalmodulationvia muscimol,aGABAagonist[ MaandLeung 1999 ].Fromawiringperspective,the medialseptumisamajorsourceofcholinergicinputandthis projectswidelythrough thehippocampus.Furthermore,GABAergicinputfromthemed ialseptumappearsto specicallytargethippocampalinterneurons,andtherefo reshouldplayacriticalrolein thegenerationandregulationofEEGrhythms[ FreundandAntal 1988 ].Thesetargets includebothO-LMcells,whichareassociatedwithhippocam paltheta[ Hajosetal. 2004 Wang 2002 ],andbasketcells,whichareassociatedwithgamma[ Buzs akiand Wang 2012 ]. InadditiontoourpriorMRIstudies,otherinvestigationsh aveprovidedstructural andfunctionalevidenceformedialseptumalteration.Usin ghistology, Gorteretal. [ 2001 2003 ]reportedcelllossinthemedialseptumearlyfollowingSE. Additionally, Colometal. [ 2006 ]reportedalteredringofmedialseptalneuronsduringepi leptogenesis, withrhythmicfastringneurons(putativeGABAergicandgl utamatergic)increasingtheir ringrate.Theysuggestedthisincreasedringratecouldb erelatedtobothdestruction ofseptalGABAergicneuronsandalsodegenerationofinhibi torysepto-hippocampal projections.Changestothemedialseptummightalsoexplai ntheobservedthetarhythm 108

PAGE 109

alterationsand,aswehavepreviouslydiscussedatlength, theobservedchangesin hippocampalSPK24-hourrhythms. Figure 8-6 alsoshowsthathippocampal24-hourrhythmsinhighfrequen cybands increaseinamplitudefollowinginjury.Whileadecreasein a24-houroscillationcanbe explainedsimplybylossordamagetothecorrespondinginpu t,itismoredifcultto explainhowanincreasein24-hourrhythmsmightarise.Chan gesinmelatonin24-hour rhythmshavebeeninvestigatedinanumberofstudiesand,wh ileevidenceismixed, somehavesuggestedthatitsreleasecanbeenhancedintheep ilepticstate[ Hofstra anddeWeerd 2009 Quigg 2000 ].Melatoninalsohaswellestablishedmechanistic roleswithinthehippocampus,attenuatingGABA A mediatedcurrents[ Wanetal. 1999 ] andincreasingoverallexcitability[ MusshoffandSpeckmann 2003 Musshoffetal. 2002 ]. 8.4.4AlternativeMechanisms Herewediscussapossiblealternativemechanismtothatpro posedinFigure 8-4 Insteadofthephaseshiftbeingproducedbyatransferincon trolofcircadiandrivers,the phaseshiftcouldsimplyappearbecauseofourchoiceofusin gxedfrequencybands inouranalysis.Specically,ifanEEGrhythmthatisnormal lyentrainedtopeakduring thenightisalteredpost-injurysuchthatitentersalowerf requencyband,thiscould producetheappearanceofaphaseshift.Thus,adaptivetech niques,suchasempirical modedecomposition,whichdonotrequirepre-speciedfreq uencydivisionsmight beusefulinfutureanalysis.However,aswecanseeforsomea nimalsinFigure 8-6 particularlyRats2and4,thephaseshiftcanoccupyaverywi derangeoffrequency bands.ItisunlikelythatanEEGrhythm'sfrequencywouldsh iftsodrasticallyfollowing injury,makingthisexplanationnotviablefortheseanimal s. 8.4.5FutureWork Basedonthisstudy,wecanidentifythreeareaswherefuture workisneeded. First,additionalsignalanalysisworkisneededtofurther characterizechangesin 109

PAGE 110

24-hourrhythms.Specically,sincegammarhythmsarenotc ontinuouslyoccurring eventsbut,rather,transientphenomena,itispresentlyno tpossibletosaywhether the24-hourfrequencybandmodulationwewitnessedisdueto changesinthe amplitudeofeachgammaeventorduetochangesinhowoftenth eyoccur.Ideally, futurestudiesemployingautomatedtrackingtechniqueswo uldtrackboththeamplitude oftheseoscillationsandalsotheirrateofoccurrence,sim ilartoourapproachhere forstudyingtheta.Secondly,ourdatasetscontainaccompa nyingvideorecordings. Sincebehaviouralstatescaninuencebrainrhythms,itwou ldbeidealtoclassify dataaccordingtobehaviouralstateandremovethisvariabl efromouranalysis. Finally,weshouldlookatthe24-hourmodulationofmeasure mentsofcoherenceand cross-frequencycoupling.Thiswillbeusefulforidentify ingwhether24-hourmodulation ofmiddlefrequencyrhythmsisdrivenbycouplingtootherEE Grhythms(redarrows, Figure 8-4 )orbydirectcircadianinput(blackarrows,Figure 8-4 ).Suchstudiescould beaccompaniedbythegenerationofdetailedsepto-hippoca mpalmodelsthatcanbe usedtostudyhowvariouscircadianinputswouldmodulateEE Grhythms.Finally,to identifythephysiologicalidentityofthespeciccircadi aninputs,weproposeobtaining additionalchronicEEGrecordingsunderconditionsofeith ermedialseptallesionand pinealectomy. 8.5ClosingRemarks Insummary,wehaveshownthatthereemergesaphaseshiftint he24-hour modulationofEEGrhythmsinthebetaandgammafrequencyban ds.Wehavealso providedevidencethattherearemultiplecircadianinputs tothehippocampusandthat thesechangeinstrengthfollowingepileptogenicinjury.A dditionally,wehaveprovideda detailedcharacterizationofthechangesthatoccurtothet aoscillations.Thesendings supportthepredictionofourpriormodelingwork,whichpre dictedthataphaseshift couldemergeasaresultofchangesintherelativestrengths ofhippocampalcircadian 110

PAGE 111

inputs.Additionally,itisalsoconsistentwithourpropos althatthemedialseptummight beacircadianrelaycenterthatisalteredtounderlietheph aseshift. 111

PAGE 112

CHAPTER9 DISCUSSIONANDCONCLUSIONS 9.1Overview Inthisdissertation,wehavestudiedhippocampal24-hourr hythmsandhave characterizedtheiralterationintemporallobeepilepsy( TLE). 1 Wecharacterizedthat twoprimarymarkersofthehippocampalEEG,spikes(SPKs)an dbetaandgamma frequencyrhythmsexhibitaphaseshiftfollowingepilepsy -inducinginjury.These changesappeartobedirectlyassociatedwithepilepsy,asr atsthatreceivedinjurybut didnotdevelopspontaneousseizuresdidnotexhibittheses ymptoms.Toinvestigate theunderlyingmechanisms,weconstructedadetailedmodel ofthesepto-hippocampal neuralnetwork,andcharacterizedhowsubtlechangesinthe relativestrengthsof circadianinputscouldleadtoaphaseshift.Thismodelpred ictedthatthereareatleast twocircadianinputsdrivinghippocampal24-hourrhythms. Throughadditionalanalysis ofEEGrhythms,wedemonstratedfunctionalevidencefortwo maincircadianinputs basedonbothdimensionalityanalysisandonthecharacteri zationofphase–amplitude relationships.Finally,weinvestigatedstructuralchang esinthemedialseptum,a putativecircadianinput,andshowedthatitshowedsignsof injuryinepilepticanimals. Weproposethatcircadianinputviathemedialseptummightb ealteredsoasto contributetotheobservedhippocampal24-hourrhythmdisr uptions. Inthischapter,wewilldiscussimplicationsofourndings inrelationtobroader epilepsyandcircadianresearch. 1 PartsofthischapterareinpresswiththeJournalofNeuroph ysiology:Stanley,D.A., Talathi,S.S.,Parekh,M.B.,Cordiner,D.,Zhou,J.,Mareci ,T.H.,Ditto,W.L.,Carney,P.R. Localphaseshiftinthe24-hourrhythmofhippocampalEEGsp ikingactivityinarat modeloftemporallobeepilepsy[ Stanleyetal. 2013 ]. 112

PAGE 113

9.2ImplicationsofAlteredCircadianRhythms 9.2.1CognitiveImpairment Circadianalterationmaycontributetoepilepsy-associat edcognitivesymptoms.For example,circadianrhythmshavebeenshowntoaffectmemory formation[ Gerstnerand Yin 2010 ],andthereisevidencethatprocessessuchaslong-termpot entiation(LTP) areregulatedona24-hourcycle[ Chaudhuryetal. 2005 ].Hippocampalsharpwaves areimportantforLTP[ Buzsaki 1996 Selbachetal. 2004 ]andthetaandgammawaves playcentralrolesinmemoryencoding[ ColginandMoser 2010 ].Thus,itispossiblethat thephasemisalignmentofhippocampalactivitiesinvestig atedinthisstudyheremight berelevantformemoryimpairmentinepilepsy.9.2.2EmergenceofSeizures Itistraditionallythoughtthatprocessessuchasneuronlo ssandsubsequent circuitrewiringcontributetotheemergenceofseizuresby creatingexcitatory-inhibitory imbalances[ BriggsandGalanopoulou 2011 DudekandSpitz 1997 El-Hassar etal. 2007 Sloviter 2005 ].Itispossiblethatalteredcircadianregulationcould facilitatethecreationofsuchexcitatory-inhibitoryimb alances.Ourmodelingwork showedthatreducedseptalcircadianinputcanproduceapha semisalignmentin the24-hourrhythmsofpyramidalcellsandbasketcells(Fig ure 6-2 C,D);wepropose thatthiscreatesanoptimaltimewindowforseizureoccurre nce.Adescriptionofthis phenomenon,andhowitmightinteractwithotherepileptoge nicprocessestotriggerthe emergenceofseizures,isprovidedinFigure 9-1 .Specically,thephaseshiftappears duringthelatencyperiodresultingfromstructuraldamage associatedwithSE(Figure 9-1 ,i-ii).Althoughtheoptimaltimewindowexistsduringthel atencyperiod,celldeath andotherchangesresultingfromSEsuppressringsuchthat excitationisstillbalanced byinhibition(Figure 9-1 ,iii).Overtime,epileptogenicprocessesrestoremeanri ng activitytopre-seizurelevels(dottedline),buttheycann otrestorethecorrectbalanceof circadiandrive(Figure 9-1 ,iv).Thus,evenifthedailymeanringrateofexcitatoryan d 113

PAGE 114

inhibitorycellsdoesnotexceedpre-seizurelevels,thepr esenceofthephaseshiftwill stillpromoteatemporary,yetpotentiallypotent,excitat ory-inhibitoryimbalanceduring theoptimaltimewindow(Figure 9-1 ,v). Thepresenceofthisoptimaltimewindowmayreectthe24-ho urrhythmof epilepticseizures,whichareobservedtoclusterintheaft ernooninbothrodentand humanTLE[ HofstraanddeWeerd 2009 Loddenkemperetal. 2011 Quigg 2000 ]. Previouslyithasbeenhypothesizedthatthe24-hourrhythm ofepilepticseizuresresults frompassiveentrainmenttothe24-hourrhythmsofneuromod ulators[ Quigg 2000 ]. However,thendingsofaphaseshiftsupportthenotionthat the24-hourrhythmof seizuresmaybeactivelydrivenbyabnormalcircadianregul ationthatpromotesperiods ofexcitatory-inhibitoryimbalancethroughouttheday. Anadditionalpropertyofthemechanismsunderlying24-hou rseizurerhythmscan bediscernedbasedoncross-speciescomparison.Specical ly,althoughrodentsand humanswithTLEhavepeakseizureoccurrenceintheafternoo n,theyhaveopposing activitypatterns(nocturnalvs.diurnal),thusindicatin gthatdailyrhythmsofseizures arenotdrivendirectlybysleeporwakeactivity.Rather,th eunderlyingmechanismmust bephase-conservedacrossthesespecies.Twenty-fourhour rhythmsofmedialseptum ringactivity,toourknowledge,havenotyetbeenquantie dinhumans;however, melatoninisknowntobephase-conserved[ Quigg 2000 ].Whethertheunderlying mechanismispassiveentrainment,ashasbeenpreviouslypr oposed,oranactive responsetoaphaseshift,wepredictthatthephasesoftheos cillationsinvolvedshould beconservedacrossspecies. 9.3ClosingRemarks Basedonanumberoflargedatasets,wehaveconductedathoro ughanalysis of24-hourrhythmsofhippocampalneuralactivityinTLE.Wh ilewehopethatour investigationshavemadeasignicantcontributiontothe eldofresearchoncircadian rhythmsandepilepsy,webelievetherearemanyexcitingopp ortunitiesfordiscoveryin 114

PAGE 115

Figure9-1.Schematicofcircadianchangesduringepilepto genesisandtheirproposed inuenceonepilepticseizures.(A,B)Functionalandstruc turalchanges duringepileptogenesis,respectively.Pyramidal(PYR)an dbasketcell(BC) ringratesareshownandkeyeventsarenumbered.Dottedlin eindicates thepre-injuryaverageringrateforpyramidalcells.Soli dlinesserveasa guidetotheeyeforchangesintheratioofexcitationtoinhi bition.Keyevents (i-v)duringepileptogenesisare:i)Priortoinjury,PYRan dBCactivityare balancedthroughouttheday.ii)Injuryfromstatusepilept icus(SE) attenuatescircadianinputtothehippocampusfromregions suchasthe medialseptum,creatinganimbalanceincircadiandrive.ii i)Thisproducesa phaseshiftintheringofbasketcellactivity.Additional damagecausedby SE,suchashippocampalcellloss(B,ii),alterstheaverage dailyring activityandsetsepileptogenicprocessesinmotion.iv)Ep ileptogenic processes,includingfurthercellloss,sprouting,andoth erhomeostatic mechanisms,drivetheemergenceofspontaneousseizures.v )These processesactinparttorestorethedailyaveragelevelsofe xcitatoryand inhibitoryactivity(dottedline).However,theydonotres torethecorrect balanceofcircadianinputand,thus,thephaseshiftpersis ts.Thecircadian phaseshiftproducesatimewindowduringwhichtheratioofe xcitationto inhibitionisincreased,increasingthelikelihoodofseiz ures. 115

PAGE 116

thisarea.Forexample,ouranalysisfocusedonasinglechan nelinadatasetcontaining atotalof32electrodes,meaningweareonlyusingasmallfra ctionofthetotalavailable data.Giventhattheseelectrodesarebilaterallyspacedth roughoutthehippocampus anddentategyrus,futureworkinvestigating24-hourregul ationofspatialcouplingis possible.Additionally,withthedevelopmentofnewtechni quesforautomatedextraction ofneuralfeatures,itwillbepossibletoinvestigatemores pecicneuralfeatures, includingindividualepisodesofgammaandbetaoscillatio ns. Finally,thisdissertationmayserveasatestamenttothepo tentialofthe“big data”movementwithintherealmofneuroscience.Ironicall y,throughtheuseof signalprocessing,dataanalysis,andmodelingtechniques ,wehavemanagedto advancetheeldofresearchinalargelyexperimentaldomai nwithoutconducting asinglenewanimalexperiment.Asthequantity,quality,an drichness,ofdatathat singleexperimentalistscancollectcontinuestogrow,wea rehopefulforincreased collaborationbetweenexperimentalistsandtheoretician sinneuroscience. 116

PAGE 117

APPENDIXA COMPLETEEEGSPKDATA ThisappendixexpandsonthedatapresentedinChapter 5 ,andshowsthedata forallanimalsusedtoobtainstatisticsforSKP24-hourrhy thms.Figure A-1 shows extracted24-hourrhythmsforallN=3seizingratsusedtoge neratestatisticsinFigure 5-2 D.Likewise,Figure A-2 showsextracted24-hourrhythmsforallN=3seizingrats usedtogeneratestatisticsinFigure 5-2 F. FigureA-1.(A-C),Shownarethe24-hourrhythmsofSPKactiv ityforallratsthat developedspontaneousseizures.Ratshownin(C)correspon dstoexample animalshowninFigure 5-2 C 117

PAGE 118

FigureA-2.(A,B)Shownarethe24-hourrhythmsofSPKactivi tyforthetworatsthatdid notdevelopespontaneousseizures.Ratshownin(A)corresp ondsto exampleanimalshowninFigure 5-2 E. 118

PAGE 119

APPENDIXB COMPLETEEEGRHYTHMDATA Thisappendixprovidestwosetsofgures.Therstset(Figu re B-1 to B-6 )shows thedetrendedEEGPSDtimeseriesthatwereusedintheanalys isinChapter 8 .They alsoprovidecontinuousestimatesofthetimeofdaythatPSD activityismaximal (T PSDmax ).ThesecondsetshowsdetrendedEEGPSDtimeseriescondens edinto single24-hourtimewindowsandttosinusoids. 119

PAGE 120

FigureB-1.DetrendedPSDdata( PSD)isshownforallfrequencybandsforRat1. Timeofmaximum PSD(T PSDmax )wasestimatedforsliding2-daywindows (50%overlapping),providingarollingestimateofphase.B lackverticalline indicatestimeofinductionofstatusepilepticus(SE)andr edverticallines indicateseizuretimings. 120

PAGE 121

FigureB-2.Rat2data,asinFigure B-1 121

PAGE 122

FigureB-3.Rat3data,asinFigure B-1 122

PAGE 123

FigureB-4.Rat4data,asinFigure B-1 123

PAGE 124

FigureB-5.Rat5data,asinFigure B-1 .Thisisanon-seizinganimal. 124

PAGE 125

FigureB-6.Rat6data,asinFigure B-1 .Thisisanon-seizinganimal. 125

PAGE 126

FigureB-7.(A,B)Sinusoidalts(red)ondetrendeddatafor Rats1and2.Datahas beencondensedinto24-hourtimewindows(modulo24). 126

PAGE 127

FigureB-8.(A,B)Sinusoidalts(red)ondetrendeddatafor Rats3and4.Datahas beencondensedinto24-hourtimewindows(modulo24). 127

PAGE 128

FigureB-9.(A,B)Sinusoidalts(red)ondetrendeddatafor Rats5and6.Datahas beencondensedinto24-hourtimewindows(modulo24).These are non-seizinganimals. 128

PAGE 129

REFERENCES E.Abrahamson,R.Leak,andR.Moore.Thesuprachiasmaticnu cleusprojectsto posteriorhypothalamicarousalsystems. Neuroreport ,12(2):435–440,2001. H.Agren,M.Koulu,J.M.Saavedra,W.Z.Potter,M.Linnoila, etal.Circadiancovariation ofnorepinephrineandserotonininthelocuscoeruleusandd orsalraphenucleusin therat. Brainresearch ,397(2):353–358,1986. F.Angeleri,J.Majkowski,G.Cacchio,A.Sobieszek,S.D'ac unto,R.Gesuita, A.Bachleda,G.Polonara,L.Krolicki,M.Signorino,etal.P osttraumaticepilepsy riskfactors:One-yearprospectivestudyafterheadinjury Epilepsia ,40(9):1222–1230, 1999. V.Aroniadou-Anderjaska,B.Fritsch,F.Qashu,andM.F.Bra ga.Pathologyand pathophysiologyoftheamygdalainepileptogenesisandepi lepsy. Epilepsyresearch 78(2):102–116,2008. S.Y.AssafandJ.J.Miller.Theroleofarapheserotoninsyst eminthecontrolofseptal unitactivityandhippocampaldesynchronization. Neuroscience ,3(6):539–50,1978. G.Aston-Jones,S.Chen,Y.Zhu,andM.L.Oshinsky.Aneuralc ircuitforcircadian regulationofarousal. Natureneuroscience ,4(7):732–738,2001. E.C.AzmitiaandM.Segal.Anautoradiographicanalysisoft hedifferentialascending projectionsofthedorsalandmedianraphenucleiintherat. JournalofComparative Neurology ,179(3):641–667,1978. D.Baker,G.Church,J.Collins,D.Endy,J.Jacobson,J.Keas ling,P.Modrich,C.Smolke, andR.Weiss.Engineeringlife:buildingafabforbiology. ScienticAmerican ,294(6): 44–51,2006. A.R.BarnardandP.M.Nolan.Whenclocksgobad:neurobehavi ouralconsequencesof disruptedcircadiantiming. PLoSGenetics ,4(5):e1000040,2008. C.A.Barnes,B.L.McNaughton,G.V.Goddard,R.M.Douglas,a ndR.Adamec. Circadianrhythmofsynapticexcitabilityinratandmonkey centralnervoussystem. Science ,197(4298):91–2,1977. P.J.Basser.Inferringmicrostructuralfeaturesandtheph ysiologicalstateoftissuesfrom diffusion-weightedimages. NMRBiomed ,8(7-8):333–44,1995. J.F.Bastlund,P.Jennum,P.Mohapel,S.Penschuck,andW.P. Watson.Spontaneous epilepticratsshowchangesinsleeparchitectureandhypot halamicpathology. Epilepsia ,46(6):934–8,2005. C.W.Bazil,D.Short,D.Crispin,andW.Zheng.Patientswith intractableepilepsyhave lowmelatonin,whichincreasesfollowingseizures. Neurology ,55(11):1746–8,2000. 129

PAGE 130

D.G.BeersmaandM.Gordijn.Circadiancontrolofthesleep– wakecycle. Physiology& behavior ,90(2):190–195,2007. C.E.Begley,M.Famulari,J.F.Annegers,D.R.Lairson,T.F. Reynolds,S.Coan, S.Dubinsky,M.E.Newmark,C.Leibson,E.So,etal.Thecosto fepilepsyinthe unitedstates:Anestimatefrompopulation-basedclinical andsurveydata. Epilepsia 41(3):342–351,2000. M.D.Belle,C.O.Diekman,D.B.Forger,andH.D.Piggins.Dai lyelectricalsilencingin themammaliancircadianclock. ScienceSignaling ,326(5950):281,2009. M.A.Belluscio,K.Mizuseki,R.Schmidt,R.Kempter,andG.B uzsaki.Cross-frequency phase-phasecouplingbetweenandoscillationsinthehippo campus. JNeurosci ,32 (2):423–35,2012. E.H.Bertram.Temporallobeepilepsy:wheredotheseizures reallybegin? Epilepsy& behavior ,14(1):32–37,2009. C.Biervert,B.C.Schroeder,C.Kubisch,S.F.Berkovic,P.P ropping,T.J.Jentsch,and O.K.Steinlein.Apotassiumchannelmutationinneonatalhu manepilepsy. Science 279(5349):403–406,1998. J.BowerandD.Beeman. ThebookofGENESIS:Exploringrealisticneuralmodelswith theGEneralNEuralSImulationSystem .Springer-Verlag,NewYork,NY,2003. J.Bower,D.Beeman,andM.Hucka. Thegenesissimulationsystem ,volume2.2002. A.Bragin,J.Engel,C.L.Wilson,I.Fried,andG.W.Mathern. Hippocampaland entorhinalcortexhigh-frequencyoscillations(100–500h z)inhumanepilepticbrain andinkainicacid-treatedratswithchronicseizures. Epilepsia ,40(2):127–137,1999. S.W.BriggsandA.S.Galanopoulou.Alteredgabasignalingi nearlylifeepilepsies. NeuralPlast ,2011:527605,2011. S.BrunelandC.deMontigny.Diurnalrhythmsintheresponsi venessofhippocampal pyramidalneuronstoserotonin,norepinephrine,gamma-am inobutyricacidand acetylcholine. BrainResBull ,18(2):205–12,1987. G.Buzsaki.Hippocampalsharpwaves:theiroriginandsigni cance. BrainRes ,398(2): 242–52,1986. G.Buzsaki.Thehippocampo-neocorticaldialogue. CerebCortex ,6(2):81–92,1996. G.Buzs aki.Thetaoscillationsinthehippocampus. Neuron ,33(3):325–340,2002. G.Buzsaki. RhythmsoftheBrain .OxfordUniversityPress,USA,2009. G.Buzs akiandF.L.d.Silva.Highfrequencyoscillationsintheint actbrain. Progressin Neurobiology ,98(3):241–249,2012. 130

PAGE 131

G.Buzs akiandX.-J.Wang.Mechanismsofgammaoscillations. Annualreviewof neuroscience ,35:203–225,2012. G.Buzsaki,L.W.Leung,andC.H.Vanderwolf.Cellularbases ofhippocampaleegin thebehavingrat. BrainRes ,287(2):139–71,1983. G.Buzs aki,J.Czopf,I.Kondakor,andL.Kellenyi.Laminardistrib utionofhippocampal rhythmicslowactivity(rsa)inthebehavingrat:current-s ourcedensityanalysis,effects ofurethaneandatropine. Brainresearch ,365(1):125–137,1986. G.Buzsaki,M.Hsu,C.Slamka,F.H.Gage,andZ.Horv ath.Emergenceand propagationofinterictalspikesinthesubcorticallydene rvatedhippocampus. Hippocampus ,1(2):163–80,1991. T.Cambras,J.R.Weller,M.Angl es-Pujor as,M.L.Lee,A.Christopher, A.Dez-Noguera,J.M.Krueger,andO.Horacio.Circadiand esynchronizationof corebodytemperatureandsleepstagesintherat. ProceedingsoftheNational AcademyofSciences ,104(18):7634–7639,2007. M.A.Carskadon,S.E.Labyak,C.Acebo,andR.Seifer.Intrin siccircadianperiodof adolescenthumansmeasuredinconditionsofforceddesynch rony. Neuroscience letters ,260(2):129–132,1999. L.J.Cauller,Z.Boulos,andG.V.Goddard.Circadianrhythm sinhippocampal responsivenesstoperforantpathstimulationandtheirrel ationtobehavioralstate. BrainRes ,329(1-2):117–30,1985. D.J.Chalmers.Facinguptotheproblemofconsciousness. Journalofconsciousness studies ,2(3):200–219,1995. G.Chatrian,L.Bergamini,M.Dondey,D.Klass,M.Lennox-Bu chthal,andI.Petersen. Aglossaryoftermsmostcommonlyusedbyclinicalelectroen cephalographers. ElectroencephalogrClinNeurophysiol ,37(5):538–548,1974. D.Chaudhury,L.M.Wang,andC.S.Colwell.Circadianregula tionofhippocampal long-termpotentiation. JBiolRhythms ,20(3):225–36,2005. H.Cho,X.Zhao,M.Hatori,T.Y.Ruth,G.D.Barish,M.T.Lam,L .-W.Chong, L.DiTacchio,A.R.Atkins,C.K.Glass,etal.Regulationofc ircadianbehaviour andmetabolismbyrev-erb-[agr]andrev-erb-[bgr]. Nature ,485(7396):123–127,2012. K.Cho.Chronic'jetlag'producestemporallobeatrophyand spatialcognitivedecits. NatNeurosci ,4(6):567–8,2001. K.Cho,A.Ennaceur,J.C.Cole,andC.K.Suh.Chronicjetlagp roducescognitive decits. JNeurosci ,20(6):1–5,2000. L.L.ColginandE.I.Moser.Gammaoscillationsinthehippoc ampus. Physiology ,25(5): 319–329,2010. 131

PAGE 132

L.V.Colom,A.Garca-Hern andez,M.T.Casta neda,M.G.Perez-Cordova,andE.R. Garrido-Sanabria.Septo-hippocampalnetworksinchronic allyepilepticrats:potential antiepilepticeffectsofthetarhythmgeneration. Journalofneurophysiology ,95(6): 3645–3653,2006. F.CrickandC.Koch.Aframeworkforconsciousness. Natureneuroscience ,6(2): 119–126,2003. J.Csicsvari,H.Hirase,A.Czurko,andG.Buzsaki.Reliabil ityandstatedependenceof pyramidalcell-interneuronsynapsesinthehippocampus:a nensembleapproachin thebehavingrat. Neuron ,21(1):179–89,1998. J.Csicsvari,H.Hirase,A.Czurk o,A.Mamiya,andG.Buzs aki.Oscillatorycouplingof hippocampalpyramidalcellsandinterneuronsinthebehavi ngrat. TheJournalof neuroscience ,19(1):274–287,1999. J.Csicsvari,H.Hirase,A.Mamiya,G.Buzs aki,etal.Ensemblepatternsofhippocampal ca3-ca1neuronsduringsharpwave-associatedpopulatione vents. Neuron ,28(2):585, 2000. F.L.DaSilva,M.Witter,P.Boeijinga,andA.Lohman.Anatom icorganizationand physiologyofthelimbiccortex. Physiologicalreviews ,70(2):453–511,1990. R.C.DanaandJ.L.Martinez.Effectofadrenalectomyonthec ircadianrhythmofltp. Brainresearch ,308(2):392–395,1984. S.Deurveilher,K.Semba,etal.Indirectprojectionsfromt hesuprachiasmaticnucleus tomajorarousal-promotingcellgroupsinrat:implication sforthecircadiancontrolof behaviouralstate. Neuroscience ,130(1):165–183,2005. F.E.DudekandM.Spitz.Hypotheticalmechanismsforthecel lularand neurophysiologicbasisofsecondaryepileptogenesis:pro posedroleofsynaptic reorganization. JClinNeurophysiol ,14(2):90–101,1997. L.El-Hassar,M.Milh,F.Wendling,N.Ferrand,M.Esclapez, andC.Bernard.Cell domain-dependentchangesintheglutamatergicandgabaerg icdrivesduring epileptogenesisintheratca1region. JPhysiol ,578(Pt1):193–211,2007. T.J.Ellender,W.Nissen,L.L.Colgin,E.O.Mann,andO.Paul sen.Primingof hippocampalpopulationburstsbyindividualperisomatictargetinginterneurons. J Neurosci ,30(17):5979–91,2010. J.Engel.Intractableepilepsy:denitionandneurobiolog y. Epilepsia ,42(s6):3–3,2001. M.S.Fee,P.P.Mitra,andD.Kleinfeld.Automaticsortingof multipleunitneuronal signalsinthepresenceofanisotropicandnon-gaussianvar iability. JNeurosci Methods ,69(2):175–88,1996. C.Fischer. TheCorpus:TheHippocraticWritings .Kaplan,2008. 132

PAGE 133

J.A.F.Fisher. ATreatiseonElectricity:AndtheConstructionandManagem entof ElectricalandMechanicalTorpedoes .Grifn,GreatBritain,1871. T.F.FreundandM.Antal.Gaba-containingneuronsinthesep tumcontrolinhibitory interneuronsinthehippocampus. Nature ,336(6195):170–3,1988. A.H.FriedmanandC.A.Walker.Circadianrhythmsinratmidbrainandcaudate nucleusbiogenicaminelevels. TheJournalofphysiology ,197(1):77–85,1968. A.H.FriedmanandC.A.Walker.Ratbrainamines,bloodhista mineandglucoselevels inrelationshiptocircadianchangesinsleepinducedbypen tobarbitonesodium. The JournalofPhysiology ,202(1):133–146,1969. F.H.Gage,A.Bj ¨ orklund,andU.Stenevi.Reinnervationofthepartiallydea fferented hippocampusbycompensatorycollateralsproutingfromspa redcholinergicand noradrenergicafferents. BrainResearch ,268(1):27–37,1983. J.R.GerstnerandJ.C.Yin.Circadianrhythmsandmemoryfor mation. NatRev Neurosci ,11(8):577–88,2010. W.GerstnerandR.Naud.Howgoodareneuronmodels? Science ,326(5951):379–380, 2009. J.A.Gorter,E.A.vanVliet,E.Aronica,andF.H.LopesdaSil va.Progressionof spontaneousseizuresafterstatusepilepticusisassociat edwithmossybresprouting andextensivebilaterallossofhilarparvalbuminandsomat ostatin-immunoreactive neurons. EurJNeurosci ,13(4):657–69,2001. J.A.Gorter,P.M.GoncalvesPereira,E.A.vanVliet,E.Aro nica,F.H.Lopesda Silva,andP.J.Lucassen.Neuronalcelldeathinaratmodelf ormesialtemporal lobeepilepsyisinducedbytheinitialstatusepilepticusa ndnotbylaterrepeated spontaneousseizures. Epilepsia ,44(5):647–58,2003. R.Goutagny,J.Jackson,andS.Williams.Self-generatedth etaoscillationsinthe hippocampus. Natureneuroscience ,12(12):1491–1493,2009. W.R.GowersandE.B.Schlesinger. Epilepsyandotherchronicconvulsivediseases: theircauses,symptomsandtreatment .Wood'sLibraryofstandardmedicalauthors. WilliamWoodandCo.,NewYork,1885. C.GuildingandH.D.Piggins.Challengingtheomnipotenceo fthesuprachiasmatic timekeeper:arecircadianoscillatorspresentthroughout themammalianbrain? EuropeanJournalofNeuroscience ,25(11):3195–3216,2007. M.Hajos,W.E.Hoffmann,G.Orb an,T.Kiss,andP.Erdi.Modulationof septo-hippocampalthetaactivitybygabaareceptors:anex perimentaland computationalapproach. Neuroscience ,126(3):599–610,2004. 133

PAGE 134

K.HaraandR.A.Harris.Theanestheticmechanismofurethan e:theeffectson neurotransmitter-gatedionchannels. Anesthesia&Analgesia ,94(2):313–318,2002. J.H.HaringandJ.N.Davis.Differentialdistributionoflo cuscoeruleusprojectionstothe hippocampalformation:anatomicalandbiochemicaleviden ce. Brainresearch ,325 (1):366–369,1985. K.M.HarrisandT.J.Teyler.Agedifferencesinacircadiani nuenceonhippocamaplltp. Brainresearch ,261(1):69–73,1983. O.K.Hassani,M.G.Lee,P.Henny,andB.E.Jones.Dischargep rolesofidentied gabaergicincomparisontocholinergicandputativeglutam atergicbasalforebrain neuronsacrossthesleep–wakecycle. TheJournalofNeuroscience ,29(38): 11828–11840,2009. M.E.Hasselmo.Whatisthefunctionofhippocampalthetarhy thm?linkingbehavioral datatophasicpropertiesofeldpotentialandunitrecordi ngdata. Hippocampus ,15 (7):936–949,2005. M.H.Hastings,A.B.Reddy,andE.S.Maywood.Aclockworkweb :circadiantimingin brainandperiphery,inhealthanddisease. NatRevNeurosci ,4(8):649–61,2003. W.A.HauserandD.C.Hesdorffer. Epilepsy:frequency,causes,andconsequences EpilepsyFoundationofAmericaNewYork,NY,USA,1990. M.L.Hines,T.Morse,M.Migliore,N.T.Carnevale,andG.M.S hepherd.Modeldb:A databasetosupportcomputationalneuroscience. JComputNeurosci ,17(1):7–11, 2004. P.F.Hingston,L.C.Barone,andZ.Michalewicz. DesignbyEvolution:Advancesin evolutionarydesign .Springer,2008. D.Hirtz,D.Thurman,K.Gwinn-Hardy,M.Mohamed,A.Chaudhu ri,andR.Zalutsky. Howcommonarethecommonneurologicdisorders? Neurology ,68(5):326–337, 2007. E.C.Y.Ho. Ifyouwanttobeslowyouhavetobefast:Controlofslowpopul ation activitiesbyfast-spikinginterneuronsvianetworkmulti stability .PhDthesis,University ofToronto,2011. A.L.HodgkinandA.F.Huxley.Aquantitativedescriptionof membranecurrentandits applicationtoconductionandexcitationinnerve. JPhysiol ,117(4):500–44,1952. W.A.HofstraandA.W.deWeerd.Howtoassesscircadianrhyth minhumans:areview ofliterature. EpilepsyBehav ,13(3):438–44,2008. W.A.HofstraandA.W.deWeerd.Thecircadianrhythmanditsi nteractionwithhuman epilepsy:areviewofliterature. SleepMedRev ,13(6):413–20,2009. 134

PAGE 135

W.A.Hofstra,M.C.Gordijn,J.C.v.H.-v.d.Poel,J.v.d.Pal en,andA.W.DeWeerd. Chronotypesandsubjectivesleepparametersinepilepsypa tients:alarge questionnairestudy. Chronobiologyinternational ,27(6):1271–1286,2010. ILEA.Proposalforrevisedclinicalandelectroencephalog raphicclassicationofepileptic seizures. Epilepsia ,22(4):489–501,1981. ILEA.Proposalforrevisedclassicationofepilepsiesand epilepticsyndromes. Epilepsia ,30(4):389–399,1989. S.T.Inouye.Doestheventromedialhypothalamicnucleusco ntainaself-sustained circadianoscillatorassociatedwithperiodicfeedings? BrainRes ,279(1-2):53–63, 1983. S.T.InouyeandH.Kawamura.Persistenceofcircadianrhyth micityinamammalian hypothalamicislandcontainingthesuprachiasmaticnucle us. ProcNatlAcadSciUS A ,76(11):5962–6,1979. B.Jian,B.C.Vemuri,E.Ozarslan,P.R.Carney,andT.H.Mare ci.Anoveltensor distributionmodelforthediffusion-weightedmrsignal. Neuroimage ,37(1):164–76, 2007. H.KangandE.M.Schuman.Long-lastingneurotrophin-induc edenhancementof synaptictransmissionintheadulthippocampus. Science ,267(5204):1658–1662, 1995. C.H.KoandJ.S.Takahashi.Molecularcomponentsofthemamm aliancircadianclock. Humanmoleculargenetics ,15(suppl2):R271–R277,2006. M.H.Kole,J.M.Koolhaas,P.G.Luiten,andE.Fuchs.High-vo ltage-activatedca2+ currentsandtheexcitabilityofpyramidalneuronsinthehi ppocampalca3subeldin ratsdependoncorticosteroneandtimeofday. NeurosciLett ,307(1):53–6,2001. R.Kramis,C.H.Vanderwolf,andB.H.Bland.Twotypesofhipp ocampalrhythmical slowactivityinboththerabbitandtherat:relationstobeh aviorandeffectsofatropine, diethylether,urethane,andpentobarbital. ExpNeurol ,49(1Pt1):58–85,1975. D.Kubota,L.L.Colgin,M.Casale,F.A.Brucher,andG.Lynch .Endogenouswavesin hippocampalslices. JNeurophysiol ,89(1):81–9,2003. R.Kurzweil. Thesingularityisnear:Whenhumanstranscendbiology .Penguinbooks, 2005. P.KwanandM.J.Brodie.Earlyidenticationofrefractorye pilepsy. NewEngland JournalofMedicine ,342(5):314–319,2000. M.G.Lee,J.J.Chrobak,A.Sik,R.G.Wiley,andG.Buzsaki.Hi ppocampalthetaactivity followingselectivelesionoftheseptalcholinergicsyste m. Neuroscience ,62(4): 1033–47,1994. 135

PAGE 136

L.S.Leung.Generationofthetaandgammarhythmsinthehipp ocampus. Neuroscience&BiobehavioralReviews ,22(2):275–290,1998. L.W.Leung.Hippocampalinterictalspikesinducedbykindl ing:relationstobehavior andeeg. BehavBrainRes ,31(1):75–84,1988. D.K.Liu,R.L.Horner,andJ.M.Wojtowicz.Timeofdaydeterm inesmodulationof synaptictransmissionbyadenosineintherathippocampals lices. NeurosciLett ,282 (3):200–2,2000. T.Loddenkemper,S.W.Lockley,J.Kaleyias,andS.V.Kothar e.Chronobiologyof epilepsy:diagnosticandtherapeuticimplicationsofchro no-epileptology. JClin Neurophysiol ,28(2):146–53,2011. E.W.Lothman,E.H.Bertram,J.W.Bekenstein,andJ.B.Perli n.Self-sustaininglimbic statusepilepticusinducedby'continuous'hippocampalst imulation:electrographicand behavioralcharacteristics. EpilepsyRes ,3(2):107–19,1989. E.W.Lothman,E.H.Bertram,J.Kapur,andJ.L.Stringer.Rec urrentspontaneous hippocampalseizuresintheratasachronicsequelatolimbi cstatusepilepticus. EpilepsyRes ,6(2):110–8,1990. J.MaandL.-W.S.Leung.Medialseptummediatestheincrease inpost-ictalbehaviors andhippocampalgammawavesafteranelectricallyinduceds eizure. Brainresearch 833(1):51–57,1999. B.A.Malow,R.Kushwaha,X.Lin,K.J.Morton,andM.S.Aldric h.Relationshipof interictalepileptiformdischargestosleepdepthinparti alepilepsy. Electroencephalographyandclinicalneurophysiology ,102(1):20–26,1997. J.Matzen,K.Buchheim,andM.Holtkamp.Circadiandentateg yrusexcitabilityinarat modeloftemporallobeepilepsy. ExpNeurol ,234(1):105–11,2012. M.H.Meisler,J.A.Kearney,etal.Sodiumchannelmutations inepilepsyandother neurologicaldisorders. JournalofClinicalInvestigation ,115(8):2010,2005. M.Migliore,T.M.Morse,A.P.Davison,L.Marenco,G.M.Shep herd,andM.L.Hines. Modeldb:makingmodelspubliclyaccessibletosupportcomp utationalneuroscience. Neuroinformatics ,1(1):135–9,2003. T.Mochizuki,A.Yamatodani,K.Okakura,A.Horii,N.Inagak i,andH.Wada.Circadian rhythmofhistaminereleasefromthehypothalamusoffreely movingrats. Physiology &behavior ,51(2):391–394,1992. R.Mongeau,P.Blier,andC.DeMontigny.Theserotonergican dnoradrenergicsystems ofthehippocampus:theirinteractionsandtheeffectsofan tidepressanttreatments. BrainResearchReviews ,23(3):145–195,1997. 136

PAGE 137

L.P.Morin,N.Goodless-Sanchez,L.Smale,andR.Y.Moore.P rojectionsofthe suprachiasmaticnuclei,subparaventricularzoneandretr ochiasmaticareainthe goldenhamster. Neuroscience ,61(2):391–410,1994. J.Morris. Fisher'sFace .Faber&Faber,2010. R.G.MunnandD.K.Bilkey.Theringrateofhippocampalca1p lacecellsismodulated withacircadianperiod. Hippocampus ,22(6):1325–37,2012. U.MusshoffandE.J.Speckmann.Diurnalactionsofmelatoni nonepilepticactivityin hippocampalslicesofrats. LifeSci ,73(20):2603–10,2003. U.Musshoff,D.Riewenherm,E.Berger,J.D.Fauteck,andE.J .Speckmann.Melatonin receptorsinrathippocampus:molecularandfunctionalinv estigations. Hippocampus 12(2):165–73,2002. G.S.Nares. Seamanship .JamesGrifnandCo.,1868. W.Nelson,Y.L.Tong,J.K.Lee,andF.Halberg.Methodsforco sinor-rhythmometry. Chronobiologia ,6(4):305–23,1979. S.A.Neymotin,M.T.Lazarewicz,M.Sherif,D.Contreras,L. H.Finkel,andW.W. Lytton.Ketaminedisruptsmodulationofinacomputermodel ofhippocampus. J Neurosci ,31(32):11733–43,2011. S.Ohdo.Chronotherapeuticstrategy:rhythmmonitoring,m anipulationanddisruption. Advanceddrugdeliveryreviews ,62(9):859–875,2010. J.O'KeefeandM.L.Recce.Phaserelationshipbetweenhippo campalplaceunitsand theeegthetarhythm. Hippocampus ,3(3):317–330,1993. M.Ozcan,B.Yilmaz,andD.O.Carpenter.Effectsofmelatoni nonsynaptictransmission andlong-termpotentiationintwoareasofmousehippocampu s. Brainresearch ,1111 (1):90–94,2006. P.Panula,U.Pirvola,S.Auvinen,andM.Airaksinen.Histam ine-immunoreactivenerve bersintheratbrain. Neuroscience ,28(3):585–610,1989. M.B.Parekh,P.R.Carney,H.Sepulveda,W.Norman,M.King,a ndT.H.Mareci.Early mrdiffusionandrelaxationchangesintheparahippocampal gyrusprecedetheonset ofspontaneousseizuresinananimalmodelofchroniclimbic epilepsy. ExpNeurol 224(1):258–70,2010. D.A.PasquierandF.Reinoso-Suarez.Thetopographicorgan izationofhypothalamic andbrainstemprojectionstothehippocampus. BrainResearchBulletin ,3(4): 373–389,1978. M.PenttonenandG.Buzs aki.Naturallogarithmicrelationshipbetweenbrainoscil lators. Thalamus&RelatedSystems ,2(2):145–152,2003. 137

PAGE 138

G.M.Peterson,L.R.Williams,S.Varon,andF.H.Gage.Losso fgabaergicneuronsin medialseptumaftermbria-fornixtransection. NeurosciLett ,76(2):140–4,1987a. G.M.Peterson,L.R.Williams,S.Varon,andF.H.Gage.Losso fgabaergicneuronsin medialseptumaftermbria-fornixtransection. Neuroscienceletters ,76(2):140–144, 1987b. H.Petsche,C.Stumpf,andG.Gogolak.Thesignicanceofthe rabbit'sseptumasa relaystationbetweenthemidbrainandthehippocampusi.th econtrolofhippocampus arousalactivitybytheseptumcells. Electroencephalographyandclinicalneurophysiology ,14(2):202–211,1962. C.Peyron,D.K.Tighe,A.N.vandenPol,L.deLecea,H.C.Hell er,J.G.Sutcliffe,and T.S.Kilduff.Neuronscontaininghypocretin(orexin)proj ecttomultipleneuronal systems. JNeurosci ,18(23):9996–10015,1998. C.Pierpaoli,P.Jezzard,P.J.Basser,A.Barnett,andG.DiC hiro.Diffusiontensormr imagingofthehumanbrain. Radiology ,201(3):637–48,1996. C.Pierpaoli,A.Barnett,S.Pajevic,R.Chen,L.R.Penix,A. Virta,andP.Basser.Water diffusionchangesinwalleriandegenerationandtheirdepe ndenceonwhitematter architecture. Neuroimage ,13(6Pt1):1174–85,2001. L.Pinato,Z.Ferreira,R.Markus,andM.Nogueira.Bimodald ailyvariationinthe serotonincontentintheraphenucleiofrats. BiologicalRhythmResearch ,35(3): 245–257,2004. M.Quigg.Circadianrhythms:interactionswithseizuresan depilepsy. EpilepsyRes ,42 (1):43–55,2000. M.QuiggandM.Straume.Dualepilepticfociinasinglepatie ntexpressdistinct temporalpatternsdependentonlimbicversusnonlimbicbra inlocation. Annalsof neurology ,48(1):117–120,2000. M.Quigg,M.Straume,M.Menaker,andE.H.Bertam.Temporald istributionofpartial seizures:comparisonofananimalmodelwithhumanpartiale pilepsy. Annalsof neurology ,43(6):748–755,1998. M.Quigg,H.Clayburn,M.Straume,M.Menaker,andE.H.Bertr am.Hypothalamic neuronallossandalteredcircadianrhythmoftemperaturei naratmodelofmesial temporallobeepilepsy. Epilepsia ,40(12):1688–96,1999. M.Quigg,H.Clayburn,M.Straume,M.Menaker,andE.H.Bertr am.Effectsofcircadian regulationandrest-activitystateonspontaneousseizure sinaratmodeloflimbic epilepsy. Epilepsia ,41(5):502–9,2000. M.Quigg,M.Straume,T.Smith,M.Menaker,andE.H.Bertram. Seizuresinducephase shiftsofratcircadianrhythms. BrainRes ,913(2):165–9,2001. 138

PAGE 139

R.J.Racine.Modicationofseizureactivitybyelectrical stimulation.ii.motorseizure. ElectroencephalogrClinNeurophysiol ,32(3):281–94,1972. A.V.Raghavan,J.M.Horowitz,andC.A.Fuller.Diurnalmodu lationoflong-term potentiationinthehamsterhippocampalslice. Brainresearch ,833(2):311–314,1999. W.Rall.Theoryofphysiologicalpropertiesofdendrites. AnnNYAcadSci ,96:1071–92, 1962. M.Raza,R.E.Blair,S.Sombati,D.S.Carter,L.S.Deshpande ,andR.J.DeLorenzo. Evidencethatinjury-inducedchangesinhippocampalneuro nalcalciumdynamics duringepileptogenesiscauseacquiredepilepsy. ProceedingsoftheNational AcademyofSciencesoftheUnitedStatesofAmerica ,101(50):17522–17527, 2004. E.H.ReynoldsandJ.V.K.Wilson.Psychosesofepilepsyinba bylon:theoldest accountofthedisorder. Epilepsia ,49(9):1488–1490,2008. H.G.Richter,C.Torres-Farfan,P.P.Rojas-Garcia,C.Camp ino,F.Torrealba,and M.Seron-Ferre.Thecircadiantimingsystem:makingsenseo fday/nightgene expression. BiologicalResearch ,pages11–28,2004. A.J.RiggsandJ.E.Riggs.Epilepsy'sroleinthehistorical differentiationofreligion, magic,andscience. Epilepsia ,46(3):452–453,2005. N.F.Ruby,C.E.Hwang,C.Wessells,F.Fernandez,P.Zhang,R .Sapolsky,andH.C. Heller.Hippocampal-dependentlearningrequiresafuncti onalcircadiansystem. Proc NatlAcadSciUSA ,105(40):15593–8,2008. J.C.Sanchez,T.H.Mareci,W.M.Norman,J.C.Principe,W.L. Ditto,andP.R.Carney. Evolvingintoepilepsy:Multiscaleelectrophysiological analysisandimaginginan animalmodel. ExpNeurol ,198(1):31–47,2006. C.B.Saper,T.E.Scammell,andJ.Lu.Hypothalamicregulati onofsleepandcircadian rhythms. Nature ,437(7063):1257–63,2005. B.Scatton,H.Simon,M.LeMoal,andS.Bischoff.Originofdo paminergicinnervationof therathippocampalformation. Neuroscienceletters ,18(2):125–131,1980. M.J.Schaaf,R.Duurland,E.R.deKloet,andE.Vreugdenhil. Circadianvariation inbdnfmrnaexpressionintherathippocampus. Molecularbrainresearch ,75(2): 342–344,2000. G.J.Schapel,R.G.Beran,D.L.Kennaway,J.McLoughney,and C.D.Matthews. Melatoninresponseinactiveepilepsy. Epilepsia ,36(1):75–8,1995. O.Selbach,N.Doreulee,C.Bohla,K.S.Eriksson,O.A.Serge eva,W.Poelchen,R.E. Brown,andH.L.Haas.Orexins/hypocretinscausesharpwave -andtheta-related 139

PAGE 140

synapticplasticityinthehippocampusviaglutamatergic, gabaergic,noradrenergic, andcholinergicsignaling. Neuroscience ,127(2):519–28,2004. L.P.Shearman,S.Sriram,D.R.Weaver,E.S.Maywood,I.Chav es,B.Zheng,K.Kume, C.C.Lee,G.T.vanderHorst,M.H.Hastings,etal.Interacti ngmolecularloopsinthe mammaliancircadianclock. ScienceSignaling ,288(5468):1013,2000. M.N.Shouse,A.M.daSilva,andM.Sammaritano.Circadianrh ythm,sleep,and epilepsy. JClinNeurophysiol ,13(1):32–50,1996. R.S.Sloviter.Theneurobiologyoftemporallobeepilepsy: toomuchinformation,not enoughknowledge. CRBiol ,328(2):143–53,2005. J.W.Smythe,L.V.Colom,andB.H.Bland.Theextrinsicmodul ationofhippocampal thetadependsonthecoactivationofcholinergicandgaba-e rgicmedialseptalinputs. Neuroscience&BiobehavioralReviews ,16(3):289–308,1992. R.J.Staba,C.L.Wilson,A.Bragin,I.Fried,andJ.Engel.Qu antitativeanalysisof high-frequencyoscillations(80–500hz)recordedinhuman epileptichippocampusand entorhinalcortex. Journalofneurophysiology ,88(4):1743–1752,2002. D.Stanley,P.Carney,M.Parekh,T.Mareci,S.Talathi,andW .Ditto.Phaseshift inhippocampalcircadianrhythmduringthelatentperiodof epilepticrats. BMC Neuroscience ,12(Suppl1):P76,2011a. D.Stanley,S.Talathi,X.Ni,L.Huang,Y.Lai,W.Ditto,andP .Carney.Circadianrhythms ofgammaoscillationsfollowingstatusepilepticus:Impli cationsforcognition. American EpilepsySociety ,2011b. D.Stanley,M.Parekh,S.Talathi,W.Ditto,T.Mareci,andP. Carney.Magnetresonance diffusionchangesinthehypothalamusandmedialseptumacc ompanyepileptogenic phaseshiftinhippocampalcircadianrhythms.In SLEEP ,volume35,pages A210–A210.AmerAcadSleepMedicineOneWestbrookCorporat eCtr,STE920, Westchester,IL60154USA,2012. D.A.Stanley,S.S.Talathi,M.B.Parekh,D.J.Cordiner,J.Z hou,T.H.Mareci,W.L. Ditto,andP.R.Carney.Phaseshiftinthe24-hourrhythmofh ippocampaleegspiking activityinaratmodeloftemporallobeepilepsy. Journalofneurophysiology ,2013. L.S.StewartandL.S.Leung.Temporallobeseizuresalterth eamplitudeandtimingof ratbehavioralrhythms. EpilepsyBehav ,4(2):153–60,2003. K.-F.Storch,O.Lipan,I.Leykin,N.Viswanathan,F.C.Davi s,W.H.Wong,andC.J. Weitz.Extensiveanddivergentcircadiangeneexpressioni nliverandheart. Nature 417(6884):78–83,2002. A.Strzelczyk,J.P.Reese,R.Dodel,andH.M.Hamer.Costofe pilepsy. Pharmacoeconomics ,26(6):463–476,2008. 140

PAGE 141

S.S.SuzukiandG.K.Smith.Spontaneouseegspikesinthenor malhippocampus.i. behavioralcorrelates,laminarprolesandbilateralsync hrony. Electroencephalogr ClinNeurophysiol ,67(4):348–59,1987. S.S.SuzukiandG.K.Smith.Spontaneouseegspikesinthenor malhippocampus. iv.effectsofmedialseptumandentorhinalcortexlesions. ElectroencephalogrClin Neurophysiol ,70(1):73–83,1988a. S.S.SuzukiandG.K.Smith.Spontaneouseegspikesinthenor malhippocampus. v.effectsofether,urethane,pentobarbital,atropine,di azepamandbicuculline. ElectroencephalogrClinNeurophysiol ,70(1):84–95,1988b. S.S.Talathi,D.U.Hwang,M.L.Spano,J.Simonotto,M.D.Fur man,S.M.Myers,J.T. Winters,W.L.Ditto,andP.R.Carney.Non-parametricearly seizuredetectioninan animalmodeloftemporallobeepilepsy. JNeuralEng ,5(1):85–98,2008. S.S.Talathi,D.U.Hwang,W.L.Ditto,T.Mareci,H.Sepulved a,M.Spano,andP.R. Carney.Circadiancontrolofneuralexcitabilityinananim almodeloftemporallobe epilepsy. NeurosciLett ,455(2):145–9,2009. J.Taxidis,S.Coombes,R.Mason,andM.R.Owen.Modelingsha rpwave-ripple complexesthroughaca3-ca1networkmodelwithchemicalsyn apses. Hippocampus 22(5):995–1017,2012. R.TraubandR.Miles. Neuronalnetworksofthehippocampus .CambridgeUniversity Press,Cambridge,MA,1991. R.Traub,J.Jefferys,andM.Whittington. FastOscillationsinCorticalCircuits .TheMIT Press,Cambridge,MA,1999a. R.D.Traub,R.K.Wong,R.Miles,andH.Michelson.Amodelofa ca3hippocampal pyramidalneuronincorporatingvoltage-clampdataonintr insicconductances. J Neurophysiol ,66(2):635–50,1991. R.D.Traub,M.A.Whittington,S.B.Colling,G.Buzsaki,and J.G.Jefferys.Analysis ofgammarhythmsintherathippocampusinvitroandinvivo. JPhysiol ,493(Pt2): 471–84,1996. R.D.Traub,M.A.Whittington,E.H.Buhl,J.G.Jefferys,and H.J.Faulkner.On themechanismofthe r frequencyshiftinneuronaloscillationsinducedinrat hippocampalslicesbytetanicstimulation. TheJournalofneuroscience ,19(3): 1088–1105,1999b. W.Truccolo,U.T.Eden,M.R.Fellows,J.P.Donoghue,andE.N .Brown.Apoint processframeworkforrelatingneuralspikingactivitytos pikinghistory,neural ensemble,andextrinsiccovariateeffects. Journalofneurophysiology ,93(2): 1074–1089,2005. 141

PAGE 142

P.Valnegri,M.Khelfaoui,O.Dorseuil,S.Bassani,C.Lagne aux,A.Gianfelice, R.Benfante,J.Chelly,P.Billuart,C.Sala,etal.Acircadi anclockinhippocampus isregulatedbyinteractionbetweenoligophrenin-1andrev -erb[alpha]. Nature neuroscience ,14(10):1293–1301,2011. G.Vandewalle,C.Schmidt,G.Albouy,V.Sterpenich,A.Dars aud,G.Rauchs,P.-Y. Berken,E.Balteau,C.Degueldre,A.Luxen,etal.Brainresp onsestoviolet,blue,and greenmonochromaticlightexposuresinhumans:prominentr oleofbluelightandthe brainstem. PLoSOne ,2(11):e1247,2007. R.Vertes,L.Colom,W.Fortin,andB.Bland.Brainstemsites forthecarbacholelicitation ofthehippocampalthetarhythmintherat. Experimentalbrainresearch ,96(3): 419–429,1993. Q.Wan,H.Y.Man,F.Liu,J.Braunton,H.B.Niznik,S.F.Pang, G.M.Brown,and Y.T.Wang.Differentialmodulationofgabaareceptorfunct ionbymel1aandmel1b receptors. NatNeurosci ,2(5):401–3,1999. L.M.Wang,J.M.Dragich,T.Kudo,I.H.Odom,D.K.Welsh,T.J. O'Dell,andC.S. Colwell.Expressionofthecircadianclockgeneperiod2int hehippocampus:possible implicationsforsynapticplasticityandlearnedbehaviou r. ASNneuro ,1(3),2009. X.J.Wang.Pacemakerneuronsforthethetarhythmandtheirs ynchronizationinthe septohippocampalreciprocalloop. JNeurophysiol ,87(2):889–900,2002. X.J.WangandG.Buzsaki.Gammaoscillationbysynapticinhi bitioninahippocampal interneuronalnetworkmodel. JNeurosci ,16(20):6402–13,1996. M.O.WestandS.A.Deadwyler.Circadianmodulationofgranu lecellresponseto perforantpathsynapticinputintherat. Neuroscience ,5(9):1597–602,1980. M.Whittington,R.Traub,N.Kopell,B.Ermentrout,andE.Bu hl.Inhibition-based rhythms:experimentalandmathematicalobservationsonne tworkdynamics. InternationalJournalofPsychophysiology ,38(3):315–336,2000. M.A.Whittington,R.D.Traub,H.J.Faulkner,I.M.Stanford ,andJ.G.Jefferys. Recurrentexcitatorypostsynapticpotentialsinducedbys ynchronizedfastcortical oscillations. ProceedingsoftheNationalAcademyofSciences ,94(22):12198–12203, 1997. S.Wiebe.Epidemiologyoftemporallobeepilepsy. TheCanadianJournalofNeurologicalSciences ,27(1),2000. J.WilsonandE.Reynolds.Textsanddocuments.translation andanalysisofa cuneiformtextformingpartofababyloniantreatiseonepil epsy. Medicalhistory 34(2):185,1990. 142

PAGE 143

K.Wulff,S.Gatti,J.G.Wettstein,andR.G.Foster.Sleepan dcircadianrhythm disruptioninpsychiatricandneurodegenerativedisease. NatRevNeurosci ,11(8): 589–99,2010. S.Yamazaki,M.C.Kerbeshian,C.G.Hocker,G.D.Block,andM .Menaker.Rhythmic propertiesofthehamstersuprachiasmaticnucleusinvivo. JNeurosci ,18(24): 10709–23,1998. Y.YanovskyandH.L.Haas.Histamineincreasesthebursting activityofpyramidalcells intheca3regionofmousehippocampus. NeurosciLett ,240(2):110–2,1998. P.Zatta. MetalIonsandNeurodegnerativeDisorders .WorldScientic,London,UK, 2003. 143

PAGE 144

BIOGRAPHICALSKETCH OriginallyfromToronto,Canada,DavidisthesonofJanetan dLeonardStanley andhasoneyoungerbrother,Geoffrey.DavidattendedDr.Jo hnM.DensionHigh SchoolandpursuedhisundergraduatedegreeinEngineering ScienceattheUniversity ofToronto.Inhisthirdyear,hespecializedinthephysicso ption,andfocusedhis bachelor'sthesisonquantumdotsolarcellsunderthesuper visionofDr.Nazir P.Kherani.Davidgraduatedwithhonoursin2007.Hethenmov edintotheeldof computationalneuroscience,joiningthegroupofDr.BerjL .BardakjianattheUniversity ofToronto,andworkedoncharacterizingmechanismsunderl yingsynapticnoise.David completedhisMAScinElectricalEngineeringin2009. MovingtoArizonaandjoiningthenewlabofDr.WilliamL.Dit to,Davidbegan studyingtherelationshipbetweencircadianrhythmsandep ilepsyin2009.He transferredfromArizonaStateUniversitytotheUniversit yofFloridain2011,where hecompletedhisdoctoralresearchundertheguidanceofDr. PaulR.CarneyandDr. SachinS.Talathi.HereceivedhisPh.D.fromtheUniversity ofFloridainthesummer of2013.Followinggraduation,Davidplanstopursuepostdo ctoralstudiesandalsoto continuetocollaboratewithhisformeradvisorsonproject srelatedtoneuronalnoise, circadianrhythmsand,hopefully,syntheticbiology. 144