Single‐cell transcriptomic and functional characterization of cortical parvalbumin interneurons in a novel conditional knock‐out mouse lacking CSPα/DNAJC5
Abstract
Neurodegenerative diseases are progressive disorders currently without cure. Indeed, the molecular mechanisms underlying synaptic and neuronal degeneration are poorly understood. Cysteine String Protein‐α/DNAJC5 (CSPα/DNAJC5) is a synaptic co‐chaperone associated to neurodegeneration in humans. Adult‐onset autosomal dominant neuronal ceroid lipofuscinosis is caused by mutations in the gene DNAJC5. In mice, the genetic removal of CSPα/DNAJC5 leads to activity‐dependent synaptic degeneration, specially evident in the fast‐spiking GABAergic interneurons that express parvalbumin (PV). The early death of these mice limits the investigation of the natural history of the synaptic and neuronal dysfunction of these cells. We have now generated a conditional KO mouse (PVcre:Ai27D:Dnajc5flox/‐) lacking CSPα/DNAJC5 specifically in parvalbumin‐positive GABAergic interneurons labeled with the optogenetic actuator channelrhodopsin‐2 fused to the fluorescent reporter tdTomato. These mice develop a progressive neurological phenotype characterized by hyperactivity and ataxia without early lethality. At the neocortex, those neurons form disorganized perisomatic synapses with a reduced number of puncta, that, unexpectedly do not compromise the number of PV somata up to 8 months of age. Those neurons are, however, more reluctant to initiate high frequency action potential trains compared to controls. The spontaneous release of GABA from those neurons is altered. The frequency of mIPSCs is decreased, consistent with a lower number of functional synapses. The amplitude of mIPSCs is strikingly reduced and the kinetics modified, perhaps caused by alterations in vesicular GABA content and/or postsynaptic changes in GABA receptors. Further characterization of ion currents and pharmacological and optogenetic dissection of synaptic currents is required to understand the changes in excitability and synaptic transmission. In order to get insight into the homeostatic mechanisms of gene expression associated to synaptic dysfunction in PV interneurons lacking CSPα/DNAJC5, we have carried out the analysis of single cell transcriptomes of tdTomato+ cells isolated from the whole cortex of three different genotypes: PVcre:Ai27D:Dnajc5WT, PVcre:Ai27D:Dnajc5flox/+ and PVcre:Ai27D:Dnajc5flox/‐. In collaboration with Dr. Hjerling‐ Leffler's group (Karolinska Institute), we have combined fluorescence‐activated cell sorting (FACS), the STRT‐seq‐2i method and the WaferGen 9600‐well platform as a novel and powerful strategy for single‐cell RNA sequencing. Next, we have conducted computational analysis using the R toolkit Seurat. We have integrated scRNA‐seq data from the three genotypes (2847 cells) and, based on common sources of variation, we have identified 8 different populations (also known as identities) of PV interneurons and carried out a downstream comparative analysis of gene expression. Three of our PV identities were highly similar to PV subclasses recently described in specific cortical areas and one of them corresponded to chandelier cells. The gene ontology (GO) study based on the differential gene expression analysis suggests genetic dysregulation of metabolism and synaptic function, among other processes, in PVcre:Ai27D:Dnajc5flox/‐ neurons. Amid a wide repertoire of expression changes, our study has identified genes related to brain disorders and to the control of hyperexcitability. Those presumptive changes, however, still require validation with alternative methodologies before they can be used to explain the molecular mechanisms underlying the neuronal and the in vivo phenotypes found in the mice. Overall, the results presented in this thesis throw light and open multiple new perspectives to advance the understanding of molecular and circuit mechanisms underlying synaptic alterations of PV interneurons and the role of CSPα/DNAJC5 in brain disorders.
Full text
DepartamentodeFisiologíaMédicayBiofísica UniversidaddeSevilla Single‐celltranscriptomicandfunctional characterizationofcorticalparvalbumininterneuronsin anovelconditionalknock‐outmouselacking CSPα/DNAJC5 MarinaValenzuelaVillatoro Advisors: Dr.RafaelFernándezChacón Dr.PabloGarcía‐JuncoClemente DoctoralThesis Sevilla,2019
DepartamentodeFisiologíaMédicayBiofísica UniversidaddeSevilla Avd.SánchezPizjuán,4 41009,Sevilla España Dr.RafaelFernándezChacón,CatedráticodelDepartamentodeFisiologíaMédica yBiofísicadelaUniversidaddeSevillayDr.PabloGarcía‐JuncoClemente, InvestigadorRamónyCajaldelDepartamentodeFisiologíaMédicayBiofísicadela UniversidaddeSevilla,CERTIFICANqueMarinaValenzuelaVillatoroharealizado bajosudireccióneltrabajotitulado“Single‐celltranscriptomicandfunctional characterizationofcorticalparvalbumininterneuronsinanovelconditional knock‐outmouselackingCSPα/DNAJC5”quepresentaparaoptaralgradode DoctorporlaUniversidaddeSevilla. Fdo:Dr.RafaelFernándezChacónFdo:Dr.PabloGarcía‐JuncoClemente Sevilla,2019
I TABLEOFCONTENTS TABLEOFCONTENTS......................................................................................................................................I SUMMARYOFTABLES...................................................................................................................................II SUMMARYOFFIGURES................................................................................................................................III SUMMARYOFBOXES...................................................................................................................................IV ABBREVIATIONS.............................................................................................................................................V ABSTRACT......................................................................................................................................................1 INTRODUCTION.............................................................................................................................................3 1.Background:thesynapseandthesynapticvesicle...................................................................................5 2.Thesynapticvesiclecycle..........................................................................................................................7 3.MembranefusionbySNAREproteins.......................................................................................................8 4.Cysteinestringproteinalpha(CSPα).........................................................................................................9 4.1.Localizationandmolecularstructure.................................................................................................9 4.2.ThetrimericcomplexCSPα‐Hsc70‐SGTisessentialforchaperoneactivity.....................................10 4.3.PhenotypiccharacterizationofCSPαknock‐out(KO)mice.............................................................11 4.4.SNAP25isanimportantsubstrateofCSPα......................................................................................12 4.5.RoleofCSPαinGABAergicsynapsesfiringathighfrequency.........................................................13 4.6.ImportanceofCSPαinhumandisease.............................................................................................14 5.Thecerebralcortex..................................................................................................................................15 5.1.Corticalorganization........................................................................................................................15 5.2.Corticalglutamatergicneurons........................................................................................................17 5.3.CorticalGABAergicinterneurons......................................................................................................17 5.3.1.Corticalinterneuronclassification.............................................................................................17 5.4.Corticalconnectivity:pyramidalneuronsversusGABAergicinterneurons.....................................19 6.Parvalbumininterneurons.......................................................................................................................20 6.1.MorphologicalandfunctionalpropertiesofPVdendrites...............................................................21 6.2.MorphologicalandfunctionalpropertiesofPVaxon......................................................................22 6.3.SpecializationsofPVcellsforfastsynapticsignaling.......................................................................23 6.4.PV‐derivedfeedforwardandfeedbackinhibition............................................................................24 6.5.RoleofPVinterneuronsinmodulatinganimalbehavior.................................................................24 6.6.RoleofPVinterneuronsinneurologicaldisorders...........................................................................25 7.Single‐cellRNAsequencing.....................................................................................................................25 7.1.Firstmilestones................................................................................................................................25 7.2.Cellisolationandsequencingmethods............................................................................................26 7.3.Up‐to‐datePVclassificationbyscRNA‐seq......................................................................................27
I 7.4.Transcriptomicsinneurodegenerativediseases..............................................................................27 GOALS..........................................................................................................................................................29 MATERIALSANDMETHODS.........................................................................................................................33 1.Mice.........................................................................................................................................................35 1.1.GenerationofconditionalDnajc5flox/floxmice...................................................................................35 1.2.GenerationofUBC‐Cre‐ERT2:Dnajc5floxmice...................................................................................36 1.3.GenerationofPVcre:Ai27D:Dnajc5floxmice........................................................................................36 2.Fluorescent‐activatedcellsorting(FACS)................................................................................................37 3.ValidationofconditionalDnajc5flox/floxmice............................................................................................39 3.1.GenomicDNA(gDNA)validation......................................................................................................40 3.1.1.gDNAextraction........................................................................................................................40 3.1.2.PCRforDnajc5gDNA.................................................................................................................41 3.2.RNAvalidation..................................................................................................................................41 3.2.1.RNAextraction...........................................................................................................................41 3.2.2.Reverse‐transcription(RT)tocDNA..........................................................................................42 3.2.3.PCRforDnajc5cDNA.................................................................................................................43 3.3.Proteinvalidation.............................................................................................................................43 4.Real‐timequantitativereversetranscription‐PCR(real‐timeqRT‐PCR)..................................................44 4.1.RNAextraction..................................................................................................................................44 4.2.Reverse‐transcriptionandcDNAamplification................................................................................44 4.3.Real‐timeqRT‐PCRandquantification.............................................................................................45 5.Survivalanalysis.......................................................................................................................................45 6.Bodyweightcurve...................................................................................................................................45 7.Openfield................................................................................................................................................46 8.Immunoblotting.......................................................................................................................................46 9.Quantificationofproteinlevels...............................................................................................................47 10.Perfusionandsectioningusingacryostat.............................................................................................48 11.Brigthfieldimmunohistochemistry.......................................................................................................48 12.QuantificationofPVsomatanumber....................................................................................................49 13.Immunofluorescenceinbrainslicesections.........................................................................................49 14.Colocalizationanalysis ...........................................................................................................................49 15.Quantificationofsynapticpuncta.........................................................................................................50 16.QuantificationofPVsomatasize...........................................................................................................50 17.Electrophysiology..................................................................................................................................51 17.1.Preparationofbrainslices..............................................................................................................52 17.2.Electrophysiologicalrecordingsofparvalbumininterneurons......................................................52
I 17.2.1.Intrinsicproperties..................................................................................................................52 17.2.2.Excitatorypostsynapticpotentials(EPSPs)..............................................................................53 17.3.Electrophysiologicalrecordingsofpyramidalneurons..................................................................53 18.Neurondissociation...............................................................................................................................53 19.Single‐cellRNAsequencingprotocol.....................................................................................................55 19.1.Steps1‐2.FACSintoaWaferGen9600microwellplate................................................................56 19.2.Steps2‐3.Lysisandreversetranscription......................................................................................57 19.3.Steps4‐5.IndexedPCRandextraction..........................................................................................59 19.4.Step6.Tagmentationand5′fragmentsisolation...........................................................................60 19.5.Step7.Illuminasequencing............................................................................................................60 19.6.Preliminaryfilteringofrawreads...................................................................................................61 20.Single‐cellRNAsequencingdataanalysis..............................................................................................63 21.Geneontology(GO)analysis.................................................................................................................64 22.Statistics.................................................................................................................................................64 RESULTS.......................................................................................................................................................65 1.CorticalsynapticdysfunctionsinDnajc5conventionalKOmice.............................................................67 2.SpecificCre‐recombinaseactivityattheDnajc5locusinthePVcre:Ai27D:Dnajc5floxmouseline............69 3.tdTomatoexpressionisspecificfortheparvalbumin‐positiveinterneuronpopulation........................74 4.ThespecificdeletionofDnajc5inPVinterneuronsdoesnotcauseincreasedmortalitybutprogressive bodyweightalterationsinPVcre:Ai27D:Dnajc5flox/‐mice.............................................................................77 5.NeurologicalphenotypeupondeletionofDnajc5inPVinterneurons...................................................79 6.CorticalexpressionofpresynapticCSPα/DNAJC5,synaptotagmin‐2andHsc70proteinsare progressivelyreducedinPVcre:Ai27D:Dnajc5flox/‐mice................................................................................84 7.ProgressivelossofSyt2+andPV+presynapticpunctainPVcre:Ai27D:Dnajc5flox/‐micewithsimilarPVcell numberandsize..........................................................................................................................................85 7.1.Age‐dependentlossofPV+andSyt2+synapticpunctainPVcre:Ai27D:Dnajc5flox/‐mice..................85 7.2.TheprogressivedegenerationofpresynapticterminalsinPVcre:Ai27D:Dnajc5flox/‐miceisnot relatedtochangesinthenumberofPVcellsorsomasize.....................................................................90 8.ElectrophysiologicalanalysisofPVinterneuronsshowssimilarintrinsicpropertiesbutchangesin spontaneoussynapticresponsesinPVcre:Ai27D:Dnajc5flox/‐animals...........................................................91 8.1.Intrinsicpropertiesofparvalbumininterneurons............................................................................91 8.1.1.PassivepropertiesofPVcells....................................................................................................92 8.1.2.ActivepropertiesofPVcells......................................................................................................93 8.1.3.Excitatorypostsynapticpotentials(EPSPs)................................................................................99 8.2.Miniatureinhibitorypostsynapticcurrents(mIPSCs)oncorticalpyramidalcells..........................100 9.Single‐cellRNAsequencingofcorticalparvalbumininterneurons.......................................................104 9.1.RobustaveragemappedmRNAreadspercell...............................................................................106
I 10.Analysisoftranscriptomicdataobtainedfromsinglecorticalparvalbumininterneurons.................106 10.1.Areadablematrixtovisualizegenesexpressedbyeverysinglecell...........................................106 10.2.Integrationofsingle‐celltranscriptomicdataacrossdifferentgenotypes..................................107 10.2.1.Cellfiltering:qualitycontrolbyfilteringoutdatafromlowqualitycells.............................108 10.2.2.Normalizationandnoisereductionofindividualgenesexpressioninsinglecells...............109 10.2.3.ClusteringanalysisrevealsdifferentsubpopulationsofPVinterneurons............................112 10.2.3.1.Firstclustering................................................................................................................112 1) SelectionofPvalbmRNAexpression.................................................................................112 2) DeterminationofHighVariableGenes(HVG)...................................................................113 3) Runningmulti‐setCCAandCCselectionforcellclustering..............................................113 4) AlignmentofCCAsubspaces.............................................................................................114 5) Cellclusteringandnon‐lineardimensionalreduction(tSNE)...........................................115 6) Identificationofconservedcelltypemarkers...................................................................117 10.2.3.2.Secondclustering...........................................................................................................119 2) Determinationofhighvariablegenes(HVG)....................................................................119 3) Runningmulti‐setCCAandCCselection...........................................................................120 4) AlignmentofCCAsubspaces.............................................................................................121 5) Cellclusteringandnon‐lineardimensionalreduction(tSNE)...........................................122 6) Identificationofconservedcelltypemarkers...................................................................122 10.2.3.3.Thirdclustering...............................................................................................................124 2) Determinationofhighvariablegenes(HVG)....................................................................124 3) Runningmulti‐setCCAandCCselection...........................................................................124 4) AlignmentofCCAsubspaces.............................................................................................125 5) Cellclusteringandnon‐lineardimensionalreduction(tSNE)...........................................126 6) Identificationofconservedcelltypemarkers...................................................................127 10.2.3.4.Fourthclustering............................................................................................................129 2) Determinationofhighvariablegenes(HVG)....................................................................129 3) Runningmulti‐setCCAandCCselection...........................................................................129 4) AlignmentofCCAsubspaces.............................................................................................131 5) Cellclusteringandnon‐lineardimensionalreduction(tSNE)...........................................132 6) Identificationofconservedcelltypemarkers...................................................................132 10.2.4.Selectionofclustermarkersforthe8detectedPVidentities..............................................132 10.2.5.NumberofcellspergenotypeacrossPVidentities...............................................................145 10.2.6.Controlforspecificcell‐typemarkersacrossgenotypesandPVclusters.............................147 10.2.7.Dnajc5expressioninPVcells................................................................................................151 10.3.Differentialexpression(DE)analysis............................................................................................153
I 10.3.1.DEanalysisincludingallPVcells............................................................................................153 10.3.2.DEanalysisbyPVclusters......................................................................................................158 1)ComparisonbetweenPVcre:Ai27D:Dnajc5flox/‐andPVcre:Ai27D:Dnajc5flox/+PVclusters.....158 2)ComparisonbetweenPVcre:Ai27D:Dnajc5flox/‐andPVcre:Ai27D:Dnajc5WTPVclusters.......163 3)ComparisonbetweenPVcre:Ai27D:Dnajc5flox/+andPVcre:Ai27D:Dnajc5WTPVclusters.......166 DISCUSSION...............................................................................................................................................173 1.ThePVcre:Ai27D:Dnajc5flox/‐asamousemodeltostudyCSPα/DNAJC5inPVinterneurons.................175 2.PVcre:Ai27D:Dnajc5flox/‐micesufferfromaprogressiveage‐dependentneurologicalphenotypewith hyperactivityandataxia............................................................................................................................176 3.SelectivereductionofpresynapticproteinsinPVcre:Ai27D:Dnajc5flox/‐mice........................................177 4.MolecularsignsofpresynapticdegenerationinthecortexofPVcre:Ai27D:Dnajc5flox/‐mice................178 5.SubtlealterationsofexcitabilityincorticalPVinterneuronslackingCSPα/DNAJC5............................179 6.NormalexcitatorysynapticinputsonPVinterneuronslackingCSPα/DNAJC5.....................................181 7.GABAreleasealterationsinPVinterneuronsfromPVcre:Ai27D:Dnajc5flox/‐mice..................................181 8.TheuseofUMIsandSTRT‐seq‐2i/WaferGentechnologyasanovelandpowerfulmethodfor sequencing.................................................................................................................................................183 9.Priortodataanalysis:readdepth,rawdataclassificationandcellfilteringbyQCmetrics.................184 10.Thebasistosupportanintegratedanalysisandreasonsforsuccessiveclusteringsteps..................186 11.TheimportanceofanappropriateCCselectionforclustering...........................................................188 12.PVcellsclusterinto8differentidentities...........................................................................................189 13.Dnajc5expressioninthePVpopulationofdifferentgenotypes........................................................191 14.DifferentialexpressionanalysisrelatedtotheabsenceofCSPα/DNAJC5..........................................192 14.1.Geneontology(GO)analysisrevealsalterationsinsynapticandmetabolicpathways...............193 14.1.1.DysregulationofsynapticprocessesintheabsenceofCSP/DNAJC5.................................193 14.1.2.Up‐regulationof"Glycolysis/Gluconeogenesis"intheabsenceofCSP/DNAJC5...............193 14.2.Up‐regulationofthegeneencodingtheneuropeptideY(Npy)..................................................194 14.3.Expressionofsex‐specificgenes...................................................................................................196 14.4.Up‐regulatedgeneswithconsistentexpressionchanges............................................................196 14.5.Up‐regulatedgenespotentiallyrelatedtoCSP/DNAJC5...........................................................197 14.6.Down‐regulatedgenesrelatedtoneuronalexcitabilityandGABAergictransmission................198 CONCLUSIONS...........................................................................................................................................201 BIBLIOGRAPHY...........................................................................................................................................205
IV SUMMARYOFBOXES BOX1.........................................................................................................................................................104 BOX2.........................................................................................................................................................105 BOX3.........................................................................................................................................................106 BOX4.........................................................................................................................................................107 BOX5.........................................................................................................................................................108 BOX6.........................................................................................................................................................113 BOX7.........................................................................................................................................................117 BOX8.........................................................................................................................................................153
V ABBREVIATIONS 5HT3aR Ionotropicserotoninreceptor aa Aminoacids ACSF Artificialcerebrospinalfluid AD Alzheimer’sdisease ADP Adenosinediphosphate AHP Afterhyperpolarization AMPA α‐amino‐3‐hydroxy‐5‐methyl‐4‐isoxazolepropionicacid ANCL Adult‐onsetneuronalceroidlipofuscinosis AP Actionpotential ATP Adenosinetriphosphate bp Basepairs BSA Bovineserumalbumin CC Correlationvector CCA Canonicalcorrelationanalysis CCK Cholecystokinin cDNA Complementarydeoxyribonucleicacid CEF Cellexpressionformat ChR2 Channelrhodopsin‐2 CNS Centralnervoussystem CSPα Cysteinestringproteinalpha CSV Comma‐separatedvalues DE Differentiallyexpressed DNA Deoxyribonucleicacid dNTP Deoxynucleotidestriphosphate EDTA Ethylenediaminetetraaceticacid EPSP Excitatorypostsynapticpotential
V ER Endoplasmicreticulum ERCCs ExternalRNAcontrolsconsortium ERT2 Estrogenreceptorligandbindingdomain ES Embryonicstem ESCG Eukaryoticsinglecellgenomics FACS Fluorescent‐activatedcellsorting FBS Fetalbovineserum FC Foldchange FS Fast‐spiking FSC Forwardscatter GABA Gamma‐aminobutyricacid gDNA Genomicdeoxyribonucleicacid GO Geneontology Hemi Hemizygous Het Heterozygous Homo Homozygous HPD Histidine,prolineandaspartic(acidresidues)motif Hsc70 Heat‐shockcognate70kDaprotein HVG Highvariablegenes IHC Immunohistochemistry KI KarolinskaInstitute KNN K‐Nearestneighbors KO Knock‐out mEPSCMiniatureexcitatorypostsynapticcurrent mIPSC Miniatureinhibitorypostsynapticcurrent mRNA Messengerribonucleicacid NGFC Neurogliaformscells
V NMDA N‐methyl‐D‐aspartate NMDG N‐methyl‐D‐glucamine NSF N‐ethylmaleimidesensitivefactor NPY NeuropeptideY ns Non‐significant NT Neurotransmitter O/N Overnight PB Phosphatebuffer PBS Phosphatebufferedsaline PCA Principalcomponentsanalysis PCR Polymerasechainreaction PFA Paraformaldehyde PPT1 Palmitoyl‐proteinthioesterase1 PSD Postsynapticdensity PV Parvalbumin PVDF Polyinylidenedifluoride QC Qualitycontrol qRT‐PCRQuantitativereversetranscriptionpolymerasechainreaction RIN RNAintegritynumber RMP Restingmembranepotential RNA Ribonucleicacid RPKM Readsperkilobasepermillionmappedreads rRNA Ribosomalribonucleicacid RT Reverse‐transcription SBD Substrate‐bindingdomain scRNA‐seqSingle‐cellRNAsequencing SDS Sodiumdodecylsulfate
V SDS‐PAGESodiumdodecylsulfatepolyacrylamidegelelectrophoresis SEM Standarderrorofthemean SGT Smallglutamine‐richTRP‐containingprotein Smart Switchingmechanismatthe5′endoftheRNAtranscript SNAP25Synaptosomal‐associatedprotein25KDa SNAPs SolubleNSF‐attachmentproteins SNARE SolubleNSF‐attachmentproteinreceptors SNN Sharednearestneighbor SSC Sidescatter Sst Somatostatin STRT Single‐celltaggedreversetranscription SV Synapticvesicle TBS‐T Tris‐bufferedsalineandtween20 tdT tdTomato TMX Tamoxifen TRP Tetratricopeptiderepeat tSNE t‐distributedstochasticneighborembedding TTX Tetrodotoxin UBC UbiquitinC UMI Uniquemolecularidentifier UV Ultraviolet VGat VesicularGABAtransporter VGlut Vesicularglutamatetransporter Vip Vasoactiveintestinalpeptide WB Westernblot WT Wildtype
1 ABSTRACT Neurodegenerativediseasesareprogressivedisorderscurrentlywithoutcure.Indeed,the molecularmechanismsunderlyingsynapticandneuronaldegenerationarepoorlyunderstood.Cysteine StringProtein‐α/DNAJC5(CSPα/DNAJC5)isasynapticco‐chaperoneassociatedtoneurodegenerationin humans.Adult‐onsetautosomaldominantneuronalceroidlipofuscinosisiscausedbymutationsinthe geneDNAJC5.Inmice,thegeneticremovalofCSPα/DNAJC5leadstoactivity‐dependentsynaptic degeneration,speciallyevidentinthefast‐spikingGABAergicinterneuronsthatexpressparvalbumin(PV). Theearlydeathofthesemicelimitstheinvestigationofthenaturalhistoryofthesynapticandneuronal dysfunctionofthesecells.WehavenowgeneratedaconditionalKOmouse(PVcre:Ai27D:Dnajc5flox/‐)lacking CSPα/DNAJC5specificallyinparvalbumin‐positiveGABAergicinterneuronslabeledwiththeoptogenetic actuatorchannelrhodopsin‐2fusedtothefluorescentreportertdTomato.Thesemicedevelopa progressiveneurologicalphenotypecharacterizedbyhyperactivityandataxiawithoutearlylethality.At theneocortex,thoseneuronsformdisorganizedperisomaticsynapseswithareducednumberofpuncta, that,unexpectedlydonotcompromisethenumberofPVsomataupto8monthsofage.Thoseneurons are,however,morereluctanttoinitiatehighfrequencyactionpotentialtrainscomparedtocontrols.The spontaneousreleaseofGABAfromthoseneuronsisaltered.ThefrequencyofmIPSCsisdecreased, consistentwithalowernumberoffunctionalsynapses.TheamplitudeofmIPSCsisstrikinglyreducedand thekineticsmodified,perhapscausedbyalterationsinvesicularGABAcontentand/orpostsynaptic changesinGABAreceptors.Furthercharacterizationofioncurrentsandpharmacologicalandoptogenetic dissectionofsynapticcurrentsisrequiredtounderstandthechangesinexcitabilityandsynaptic transmission.Inordertogetinsightintothehomeostaticmechanismsofgeneexpressionassociatedto synapticdysfunctioninPVinterneuronslackingCSPα/DNAJC5,wehavecarriedouttheanalysisofsingle celltranscriptomesoftdTomato+cellsisolatedfromthewholecortexofthreedifferentgenotypes: PVcre:Ai27D:Dnajc5WT,PVcre:Ai27D:Dnajc5flox/+andPVcre:Ai27D:Dnajc5flox/‐.IncollaborationwithDr.Hjerling‐ Leffler'sgroup(KarolinskaInstitute),wehavecombinedfluorescence‐activatedcellsorting(FACS),the STRT‐seq‐2imethodandtheWaferGen9600‐wellplatformasanovelandpowerfulstrategyforsingle‐cell RNAsequencing.Next,wehaveconductedcomputationalanalysisusingtheRtoolkitSeurat.Wehave integratedscRNA‐seqdatafromthethreegenotypes(2847cells)and,basedoncommonsourcesof variation,wehaveidentified8differentpopulations(alsoknownasidentities)ofPVinterneuronsand carriedoutadownstreamcomparativeanalysisofgeneexpression.ThreeofourPVidentitieswerehighly similartoPVsubclassesrecentlydescribedinspecificcorticalareasandoneofthemcorrespondedto chandeliercells.Thegeneontology(GO)studybasedonthedifferentialgeneexpressionanalysissuggests geneticdysregulationofmetabolismandsynapticfunction,amongotherprocesses,in PVcre:Ai27D:Dnajc5flox/‐neurons.Amidawiderepertoireofexpressionchanges,ourstudyhasidentified
2 genesrelatedtobraindisordersandtothecontrolofhyperexcitability.Thosepresumptivechanges, however,stillrequirevalidationwithalternativemethodologiesbeforetheycanbeusedtoexplainthe molecularmechanismsunderlyingtheneuronalandtheinvivophenotypesfoundinthemice.Overall,the resultspresentedinthisthesisthrowlightandopenmultiplenewperspectivestoadvancethe understandingofmolecularandcircuitmechanismsunderlyingsynapticalterationsofPVinterneurons andtheroleofCSPα/DNAJC5inbraindisorders. Acknowledgments ThisthesiswassupportedbytheSpanishMinistryofEconomyandCompetitiveness(MINECO) (GrantsBFU2013‐47493‐PandBFU2016‐76050‐P,FPIFellowshipBES2014‐070405,andshortscientificvisit EEBB‐I‐16‐10890),CIBERNED(InstitutodeSaludCarlosIII),ITRIBIS,JuntadeAndalucía(GrantsP12‐CTS‐ 2232andCTS‐600)andFondoEuropeodeDesarrolloRegional.EScellswereobtainedfromEUCOMM.
INTRODUCTION
Introductionkk 10 etal.,1996;CoppolaandGundersen,1996),suggestingageneralfunctioninsecretionforCSPproteins. WhileC.elegansandDrosophilahaveonlyasinglegenecodingforCSP(dnj‐14andCsp,respectively), mammalsexpressthreedifferentCSPproteins(CSPα,CSPβ,CSPγ)encodedbyDNAJC5a,bandggenes, presentingisoformsαandβhighhomologyintheiraminoacidsequences,whileCSPγislessrelated.CSPα istheprincipalisoforminthebrain,locatedinthesynapticvesiclemembrane(Fernández‐Chacónetal., 2004;SchmitzandFernández‐Chacón,2009),whereashaircellsfromtheinnerearpresentbothCSPαand CSPβ(Schmitzetal.,2006),andtestisexpressthethreeCSPisoforms(Fernández‐Chacónetal.,2004). ThemolecularstructureofCSPsischaracterizedbydistinctdomains: 1. N‐terminalDNA‐Jdomain:thisregionistypicaloftheHsp40familyofco‐chaperones,andpresents homologywiththebacterialDnaJprotein,whichcomprisesasequenceofapproximately70amino acidshighlyconservedwithinspecies(atleast78%identity).ThisdomaincontainsaconservedHPD (histidine,proline,andasparticacidresidues)motifrequiredforbindingtoHsc70proteintotypically actintherefoldingordisaggregationofclientproteins(Braunetal.,1996;BurgoyneandMorgan, 2015;ChamberlainandBurgoyne,2000;SchmitzandFernández‐Chacón,2009). 2. Linkerregion:thisarea,constitutedofaround20aminoacidsapproximately,ishighlyconserved duringevolutionandisimportantforexocytosis(Zhangetal.,1999). 3. Centralcysteine‐richstring:thisregionnamestheprotein.Thestringcontains13‐15cysteineresidues withinatotalof25aminoacids,beingmostofthempalmitoylated,whichisessentialformembrane targetingofCSPtosynapticvesicles(ChamberlainandBurgoyne,1998). 4. VariableC‐terminaldomain:itissubjectedtodifferentialsplicingandgeneratestruncatedproteinsin Drosophilaandmammals.Itsfunctionisnotwell‐understoodyet,althoughitwashypothesizedto participateinexocytosisandprotein‐proteininteractions(Boaletal.,2004). 4.2.ThetrimericcomplexCSPα‐Hsc70‐SGTisessentialforchaperoneactivity OneofthebestknownfunctionsforCSPαisitsroleasaco‐chaperone(BraunandScheller,1995; ChamberlainandBurgoyne,2000;Tobabenetal.,2001;Zinsmaieretal.,1994).Firstobservations determinedthatCSPαinteractswithmembersoftheheat‐shockproteinfamilyHsp70(mostlyactingas chaperones),especiallyactivatingtheHsc70ATPase(Braunetal.,1996;ChamberlainandBurgoyne,1997; Stahletal.,1999).Later,anovelCSPα bindingpartnerwasidentified,SGT(smallglutamine‐rich tetratricopeptiderepeat‐containingprotein),whichconsistsofthreetetratricopeptiderepeatdomains (TRPs)intandem.Together,thesethreeproteinsconstituteatripartitecomplexthatactsasachaperone machineryonthesynapticvesiclesurface(Figure3),andwasfoundtorefolddenaturedluciferaseinvitro (Tobabenetal.,2001).
Introduction 11 Specifically,theDNA‐JdomainofCSPαmediatesthebindingtotheconstitutivelyexpressedheatshockcognateproteinHsc70,recruitingHsc70tothesynapticvesiclesurface.Hsp70proteinsincludeaN‐ terminaladenosinetriphosphate(ATPase)domainandaC‐terminalsubstrate‐bindingdomain(SBD).The ATPasedomainbindsATP/ADP,butitsATPaseactivitystayslowuntiltheinteractionwiththeDNA‐J domainofCSPα.OnceATPisboundandtheDNA‐JdomainofCSPα interactswiththeHsc70ATPase domain,thehydrolysisofATPswitchestheconformationalstateofHsc70intoahigh‐affinitylockedstate. Moreover,theATPaseactivityisadditionallyenhancedbySGT,thatsimultaneouslybindstotheC‐terminal domainofHsc70andtothecysteine‐stringregion(andC‐terminus)ofCSPα,actingasamolecularbridge betweenCSPαandHsc70,andmakingtherefoldingofdenaturedproteinspossible.AnewbindingofATP promotesthereleaseofrenaturedclientproteins,leavingHsc70readyforanewroundinthechaperone cycle(SchmitzandFernández‐Chacón,2009;Tobabenetal.,2001). 4.3.PhenotypiccharacterizationofCSPαknock‐out(KO)mice CSPαKOmiceweregeneratedbyhomologousrecombinationofaregionincludingexon2,which isreplacedwithaneomycinresistancegenecassette(Fernández‐Chacónetal.,2004;Tobabenetal.,2001) (seelaterFigure16).Immediatelyafterbirth,KOanimalsdonotshowanyobviousdefects,butaftertwo orthreeweeksofpostnatalage,theybegintoexperiencebody‐weightloss,evidentmuscleweakness, Figure3.CSPα‐Hsc70‐SGTtrimericcomplex. TheDNA‐JdomainofCSPαbindstotheATPaseregionofHsc70whileTPRrepeatsofSGTbindtotheC‐terminus ofHsc70aswellastothecysteine‐richstringandtheC‐terminusofCSPα.Altogetherconformsthetripartite chaperonecomplex.ATP,adenosinetriphosphate;CSP,cysteine‐stringprotein;SBD,substrate‐bindingdomain; SGT,smallglutamine‐richtetratricopeptide;TPR,tetratricopeptiderepeat(TakenfromSchmitzandFernández‐ Chacón,2009).
Introductionkk 12 deficiencyinlocomotoractivityandprogressiveblindness(Fernández‐Chacónetal.,2004;Schmitzand Fernández‐Chacón,2009;Schmitzetal.,2006).Furthermore,CSPαKOmicepresentadditionalphenotypic characteristics,forexample,atendencytoclasphindlimbswhensuspendedbythetail,apoorerposture, andalegpositionthatsimulates“invertedchampagnebottles”.Deeperexaminationsshow neurodegenerativechangesandanage‐dependentdeteriorationinsynaptictransmissionofthe neuromuscularjunctionsandCalyxofHeldsynapses.CSPαKOmicediearound1‐2monthsafterbirth (Fernández‐Chacónetal.,2004)(Figure4). 4.4.SNAP25isanimportantsubstrateofCSPα OneoftheclientproteinsfortheCSPα‐Hsc70‐SGTcomplexisSNAP25,aSNAREcomplex‐forming protein(Chandraetal.,2005;Sharmaetal.,2011).Thetrimericcomplexbindsdirectlytomonomeric SNAP25andpreventsitsdegradation,allowingSNAREcomplexformation.AbnormalSNAP25conformers thatpresumablyarisespontaneouslyduringsynapticactivityinhibittheformationofSNAREcomplexes andareubiquitylatedforproteasomaldegradation(Figure5).Interestingly,thisdegradationisenhanced bydeletionofCSPα anddecreasedifoverexpressed,sothetrimericCSPα‐Hsc70‐SGTcomplexacts refoldingSNAP25topreventitsdegradationandensuringSNARE‐complexformation(Sharmaetal.,2011). IthasbeenalreadyshownthatSNAP25levelsare40%decreasedinmicelackingCSPα andthatthe impairedSNARE‐complexassembly(50%reduction)provokesneurodegeneration(Sharmaetal.,2012). Interestingly,overexpressionofα‐synuclein(mutationsinα‐synucleinhavebeenlinkedtoParkinson’s disease)abolishesthelethalphenotypecausedbydeletionofCSPαinmicebypartiallycorrectingSNARE complexassembly,suggestingacooperatingchaperone‐mechanismbetweenCSPα andα‐synucleinto maintainSNAREcomplex(Chandraetal.,2005). A B C Figure4.CSPαKOmicephenotype. A.CSPαKO(CSP‐/‐)miceshowasignificantsmallersizeandseveremotordeficitscomparedtocontrols(CSP+/+) (TakenfromSchmitzandFernández‐Chacón,2009).B.ChangesinbodyweightofWT(CSP+/+),heterozygous (CSP+/‐)andKO(CSP‐/‐)micewithage.C.PostnatalsurvivalcurvecomparingthethreedifferentCSPαgenotypes previouslymentioned(TakenfromFernández‐Chacónetal.,2004).
Introduction 13 ThesefindingspositionedCSPα‐Hsc70‐SGTasachaperonecomplexthatensuresthelong‐term operationoftheSNAREmachinery,whichseemsespeciallycrucialfornerveterminalsthatworkathighfrequencies,sincetheysuffermultipleroundsofexocytosisalongthelifeofaneuron. 4.5.RoleofCSPαinGABAergicsynapsesfiringathighfrequency PreviousobservationssuggestthatCSPα mightbecriticallyrequiredatsynapsessubjectedto heavymembranetraffickingduetointensesynapticvesiclecycling.Basedonthis,García‐Junco‐Clemente etal.,2010studiedCSPαfunctioninsmallcentralGABAergicsynapsesthatworksunderahighactivity regime.Theyrevealedthatsynaptotagmin‐2‐expressing(Syt2+)GABAergicsynapsesformedbyfastspiking(200Hz)parvalbumin(PV)interneuronsdependedonCSPα tomaintaintheirfunctionand integrityalongtime.PVinterneuronssufferedprogressivedegenerationofpresynapticterminalsupon CSPαremoval,whichwasshownbyanage‐dependentreductioninSyt2+andPV+synapticpunctaatthe hippocampusofCSPαKOmice,butwithnochangesinPVsomatanumber.Electrophysiologicalrecordings inhippocampalculturedneuronsfromCSPαKOmiceconfirmedthepreviousdata,showingaprogressive decreaseofmIPSCsfrequencywithtimeinculture,whichagreeswithareductionintheGABAergic synapticinput.Interestingly,glutamatergicsynapseswereapparentlynotaffected,confirmedbythe absenceofchangesinmEPSCsfrequencyorglutamatergicpuncta.SinceglutamatergicneuronstriggerAPs atlowerfrequencies(1Hz)comparedtoPVinterneurons,theproposedhypothesiswasthatPVcellsare moresensitivetonetworkactivity.Importantly,theinhibitionoftheexcitatorytransmissionincellcultures wassufficienttorescueSyt2+synapses,suggestingaspecificactivity‐dependentdegenerationoffast‐ Figure5.CSPα‐Hsc70‐SGTchaperonecomplexfunctioning. AfterSNAP25isdissociatedfromtheSNAREcomplexbyNSFandSNAPs,thereisabalancebetweenSNARE‐ complexformationusingcompetentconformers(SNAP25)andinhibitionofSNARE‐complexassemblydueto misfoldedconformers(SNAP25*).GeneratedabnormalSNAP25*conformersareubiquitylated(Ub)and degradedbytheproteasome,however,thetrimericCSPα‐Hsc70‐SGTcomplexactsrefoldingSNAP25*into competentSNAP25,preventingitsdegradationandensuringSNARE‐complexformation.ATP,adenosine triphosphate;ADP,adenosinediphosphate;Pi,inorganicphosphate;Syb2,synaptobrevin‐2;Synt‐1;syntaxin‐1 ( TakenfromSharma,BurréandSüdhof,2011 ) .
Introduction kk 14 spikingsynapticterminals.Itisimportanttoremarkthatthelossofinhibitionwascompensatedbya homeostaticmechanismthatconsistsofadownscalingofpostsynapticallyexpressedAMPAreceptors (García‐Junco‐Clementeetal.,2010)(Figure6). 4.6.ImportanceofCSPαinhumandisease TwospecificmutationsinhumanDNAJC5/CLN4genehavebeenrelatedtoahumandiseasecalled autosomal‐dominantadult‐onsetneuronalceroidlipofuscinosis(ANCL),characterizedbyalysosomal accumulationofmisfoldedandproteolysis‐resistantmaterial(namedlipofuscin)inneuraltissuesrelated toneurodegeneration.Thisdiseasestartsduringtheadulthood,around30yearsold.Themutationsin DNAJC5/CLN4genearecausedbyageneticdeletion(p.Leu116del)orsubstitution(p.Leu115Arg)intothe cysteine‐stringdomainofCSPα,affectingthepalmitoylation‐dependentsorting.Indeed,asingle‐copy mutationofthegeneproducesadecreaseintheamountofsynapticCSPαwild‐typeproteininneuronal cells,becausethemutationsleadtoaggregatescomposedofbothmutantandWTproteins.Individuals Figure6.PredictedmodelofinhibitoryandexcitatorysynapticchangesuponCSPαremovalinhippocampal cultures. InCSPαWTcultures,thereisabalancebetweenexcitatoryandinhibitoryinputs.Twotypesofinhibitoryones arerepresented:terminalscomingfromfast‐spikinginterneurons(synaptotagmin‐2‐(Syt2)andparvalbuminpositive(PV)basketcells)andterminalscomingfromlessactiveinterneurons(calretinin‐positive(CR)). Terminalsoffast‐spikingneurons,whichworkunderahighactivityregime,needCSPαtoovercomemolecular stressderivedfromcontinuousactivity,andmightinitiateadying‐backdegenerationprocessinanage dependentmanner.CR+interneuronsandpyramidalneuronsarelessdependentonCSPαbecausetheywork underalow‐activityregime.Thesynapticnetworkseemstosensethisinhibitorysynapticinputreductionand triggershomeostaticplasticitymechanisms:excitatorysynapsesbecomeweakerbypostsynapticfunctional downscalingofAMPA‐typereceptors(TakenfromGarcía‐JuncoClementeetal.,2010).
Introduction 15 affectedbyANCLdevelopseizures,movementdisorders,cognitivedeteriorationsandprogressive dementia,andalsohaveasignificantdecreaseinlifespan(Diez‐Ardanuyetal.,2017;Greavesetal.,2012; Noskováetal.,2011).Interestingly,PPT1(palmitoyl‐proteinthioesterase1),aproteinthatremovesacyl chainsfrompalmitoylatedproteinsduringdegradation,wasrecentlyfoundtobeincreasedandmislocalizedinCSPαANCLhumancells,althoughshowingdramaticallyreducedactivity.Importantly,ithas beendemonstratedthatCSPαisasubstrateofPPT1,soitisdepalmitoylatedbyPPT1(Hendersonetal., 2016). 5.Thecerebralcortex Theincreaseinbrainsurfacehasbeenconsideredasthemajordevelopmentduringevolution;for example,asignificantvariabilityisobservedinbrainsizebetweenmammalianspecies(DeFelipe,2011). Thissizeincrementisincloserelationwiththehugeexpansionofthemammaliancortexregardingtothe numberofcells,layersandfunctionalareas,comparedtootherbrainregions.Moreover,thegrowthof corticalregionsisassociatedwiththeacquisitionofrefinedcognitive,sensoryandmotorfunctions.In largemammalsandhumans,thecerebralcortexhasafoldedshapethatisconstitutedbyfissures separatedbyelevatedregions;itisthoughtthatthisfoldinghappenedtoharborthegrowingnumberof neurons.Strikingly,corticalthicknessdoesnotvarysubstantiallyamongspeciesanditisalwaysaround2‐ 4mm.Itcanreachupto4.5mminhumans,whileinthefissuresmayhave1mmthick.Nevertheless,in mammalswithsmallbrains(e.g.ratsandmice),cortexisnotfoldedbutsmooth(DeFelipe,2011;Kandel etal.,2001).Itwasacrucialeventduringtheevolutionofthemammaliantelencephalontheemergence ofahighlycomplexmulti‐laminatedcortex:theneocortex.Theactivityofthisstructureisstraightlyrelated totheappearanceoftheabilitiesthatdistinguishhumansfromothermammals,e.g.,perceptual awareness,thought,languageorconsciousness(DeFelipe,2011). 5.1.Corticalorganization Basedonworksaboutcorticalcytoarchitecture,connectivity,functionalityandgeneexpression (DeFelipe,2011;Glasseretal.,2016;Ngetal.,2009;Tasicetal.,2018),approximately200corticalareas havebeendescribedinhumans,anddozensinrodents.Thesearelocatedwithinthefourmaincortical lobes:frontal,parietal,temporalandoccipital.Althoughcorticalareashavespecificactivitypatternsand informationprocessingroles(e.g.sensoryormotor),thecortexshowssimilarlayeredstructurecontaining preciseneuronalpopulations.Thenumberoflayersandtheirorganizationvarybetweencorticalareas, butthetypicalneocorticalstructurecontainssixlayersthatarenamedfromthemostsuperficialtothe whitematter.Besides,fromafunctionalpointofview,thelayersofthecerebralcortexcanbedividedinto threeparts.
Introductionkk 16 1. ThesupragranularlayersconsistoflayerItoIII.Theyarethemainoriginandterminationof intracorticalconnections,whichareeitherassociational(i.e.,withotherareasofthesame hemisphere)orcommissural(i.e.,connectionstotheoppositehemispherethroughthecorpus callosum). ‐ LayerIisanacellularmolecularstratum,whichisoccupiedbydendritescomingfromcells localizedathigherdepths,andbyaxonsthatmakeeithersynapticconnectionsinthislayeror gothroughit. ‐ LayerIIisbasicallycomposedbysmallsphericalcellscalledgranules,soitisnamedexternal granularstratum.ItisnormallydescribedandanalyzedtogetherwithlayerIII. ‐ LayerIIIcomprisesdiversecelltypes,includinglotsofpyramidal‐shapecells.Inthedeeper regionofthislayer,neuronsaregreaterinsizethanthoselocalizedmoresuperficially.Itis namedexternalpyramidalstratum.LayersIIandIIIcontaindendritesfromcellslocalizedin layersVandVI. 2. LayerIVismainlycomposedbycellssimilartogranulesoflayerII,andhenceitiscalledinternal granularstratum.Itreceivesthalamocorticalconnections,especiallyfromspecificthalamicnuclei, beingmoreprominentinprimarysensorycortices. 3. Theinfragranularlayers,whichprimarilyconnectthecerebralcortexwithsubcorticalregions.These layersaremostdevelopedinmotorcorticalareas. ‐ LayerVcomprisescellswithpyramidal‐likeshapethatarenormallybiggerthanthosefrom layerIII.Itisknownasinternalpyramidalstratum. ‐ LayerVIisaquiteheterogeneouslayerofneuronsanditisknownaspolymorphicstratum.It ismergewiththewhitematter,limitingthecortexandcarryingaxonsfrom/tothecortex. LayersVandVIcontaindendritesfromneuronslocalizedatlayerIIIandIV. TherearemanyexceptionstotheclassicaldivisionoftheneocortexintoI‐VIlayers.Forexample, inmotorandpremotorareasofmanyspecies(includingmice)thereisnolayerIV;ontheotherhand,an extremelyprominentlayerVIisobservedinprimaryvisualcortexthatcouldbesubdividedinatleast3 morelayers(DeFelipe,2011;Kandeletal.,2001).Itisalsowell‐knownthatthedispositionofneuronsin thebarrelcortexofrodentsisdifferentcomparedtoothercorticalregions:thebarrelsconsistoflayerIV aggregatesofsmallspinyneurons(DeFelipe,2011;Feldmeyer,2012). Furthermore,inthe1950s,Mountcastleintroducedtheexpressionof“corticalcolumns”or “modules”todescribetheconceptofverticalinformationprocessingincortex.Thecortexisorganized horizontallyintosixlayers,butalsoverticallyintogroupsofcellssynapticallylinkedacrossthehorizontal
Introduction 17 layers.Thisstructureismoreobviousinsensoryandmotorareasoftheneocortex(Feldmeyer,2012; Mountcastle,1997),andrecentlywasalsofoundinvisualcortex.Nevertheless,columnsareneitheran obligatorynoruniquecorticalfeature(DeFelipe,2011). 5.2.Corticalglutamatergicneurons Glutamatergic(principal)neuronsarethemostabundantcorticalneurons(around80‐90%)and aredistributedinallcorticallayersexceptforlayerI.Theyhaveatypicalpyramidal‐likeshape.Theiraxons makeintracorticalconnections,involvingthemainsourceofexcitatorysynapsesinthecortex,butthey alsorepresentthemajorityofprojectionneurons.Theirdendriticspinesaretheprincipalpostsynaptic targetofexcitatorysynapsesinthecortex.Nevertheless,thestructureofthesecellschangessubstantially amongcorticalareasandspecies,regardingseveralfeatures:thesizeandcomplexityofdendritic arborization,thedensityofdendriticspines,andthetotalnumberofspines,beingthesedifferencescritical forthefunctionalprocessingofcorticalinformation(DeFelipe,2011).Importantly,corticalareascontain multiplehighlystructuredsubnetworksofinterconnectedprincipalneurons(HarrisandMrsic‐Flogel, 2013). 5.3.CorticalGABAergicinterneurons GABAergicinterneuronsarepresentedinalowerproportionincortexcomparedtopyramidalcells (10‐15%inrodentsand25%inmonkeys).Theyarelocalizedinalmostalllayers,andmostlymakelocal connections.AlthoughGABAergicinterneuronsrepresentaminority,theirroleisvitalforbrainfunctioning becausetheyregulatetheactivityofprincipalneuronsusingGABAasNT(DeFelipe,2011;HarrisandMrsic‐ Flogel,2013;Tamamakietal.,2003;Tremblayetal.,2016).CorticalGABAergicinterneuronsarebornin themedialorcaudalganglioniceminences(MGE,CGE)andaresubjectedtospecifictranscriptionfactors forcorticalmigration,suchasNkx2.1andLhx6(Huetal.,2014;KelsomandLu,2013;Natalieetal.,2015; WondersandAnderson,2006). 5.3.1.Corticalinterneuronclassification TheGABAergicpopulationishighlyheterogeneous,thustheycanbedescribedbyseveralcriteria, suchas(1)morphology,(2)theexpressionofmolecularmarkerslikeneuropeptides(somatostatin, cholecystokinin,vasoactiveintestinalpeptide(VIP)andneuropeptideY(NPY))andcalcium‐binding proteins(parvalbumin,calretininandcalbindin),(3)theexpressionofreceptors(acetylcholine,serotonin, noradrenalineanddopamine),(4)functionalproperties(concerningtotheelectrophysiological characteristicsoftheneurons),(5)connectivityand(6)invivopatternsofactivity(Tremblay,LeeandRudy, 2016;Hu,GanandJonas,2014). Basedontheseobservations,classicalclassificationofneocorticalinterneuronsdividesthemin threemajorgroups:parvalbumin(PV),somatostatin(Sst)and5HT3aR(ionotropicserotoninreceptor)
Introductionkk 18 interneurons,althoughtheyarealsointrinsicallyheterogeneous(Figure7).Thesethreepopulations accountfornearly100%ofallGABAergiccorticalinterneurons(KelsomandLu,2013;Pfefferetal.,2013; Tremblayetal.,2016). 1. PVinterneurons.PVcellsarethelargestinterneuronpopulationintheneocortex.Itincludesfastspiking(FS)basketcellsandchandelieroraxo‐axoniccells. ‐ FSbasketcells:theypreferentiallymakeperisomatic“basket”synapticconnectionsonthe somaandproximaldendrites(somato‐dendritic)ofprincipalcellsandinterneurons.Indeed, FSbasketcellsarethelargestpopulationofinterneuronsintheneocortex.Basketcellshave animpressivecollectionofmolecularandcellularspecializationstoguaranteeafast,reliable, strongandtemporallypreciseinhibitionontheirtargetcells.Thespeedandprecisionoftheir signalingisextraordinary(Tremblayetal.,2016). ‐ Chandeliercells:theyformuniquecandlestick‐likesynapticterminalarraystospecifically targettheaxoninitialsegment(axo‐axonic)ofpyramidalcells.Chandeliercellsarealso consideredfast‐spikingbuttheyshowsomedifferencesinintrinsicelectrophysiological propertiescomparedtoPVbasketcells(Tremblayetal.,2016). 2. Sstinterneurons.Theyaredendritictargetinginterneurons.Theycanbedividedintotwomain subgroups: ‐ Martinotticells:theypresentanaxonalplexusinLIandtheytargetthetuftdendritesof pyramidalcells,makingsynapsesondendriticspines.Martinotticellsaremainlylocatedin LII/IIIandLV/VI.BesidesarborizinginLI,asignificantproportionoftheiraxonalarbor,probably contactingthebasaldendritesofotherneurons,ispresentinthelayerwherethesomais located(Tremblayetal.,2016). ‐ Non‐Martinotticells:althoughsharingmanypropertieswithMartinotticells,theylack significantaxonalarborizationinLI,makingsynapsesontodifferentsubcellularcompartments orcelltypes(Tremblayetal.,2016). 3. 5HT3aRinterneurons.Thisgrouprepresentaround30%ofallneocorticalinterneuronsandare thoughttobemorediversethanPVandSstcells.Theyallexpress5Ht3aandnicotinicreceptors.They areenrichedinsupragranularlayersandaredividedintotwosubgroupsbasedonwhetherornotthey expresstheneuropeptideVIP. ‐ Vipinterneurons:theyrepresentabout40%ofall5HT3aRcellsandtheyarelocated preferentiallyinLII/III.ThemajorityofVipinterneuronshaveaverticallyoriented,bipolar‐like
Introduction 19 dendriticmorphology(therestaremultipolar).Theirdendritictreestendtobenarrowand crossseverallayers(Tremblayetal.,2016). ‐ Non‐Vip5HT3aRinterneurons:theyrepresentabout60%ofall5HT3aRcellsandabout90% ofallinterneuronsinLI.Theyincludetheneurogliaformcells(NGFC)(containingalso multipolarneuropeptideY(NPY)‐expressingcells,althoughNPYisalsopartiallyexpressedin SstandPVinterneurons)(Karagiannisetal.,2009;Leeetal.,2010),cholecystokinin(CCK)‐ expressinginterneuronsandotherlessclearlydefinedtypes(Tremblayetal.,2016). 5.4.Corticalconnectivity:pyramidalneuronsversusGABAergicinterneurons Theneocortexreceivesafferentinputfromthethalamus,andfromothercorticalregions,besides strongneuromodulatoryinputsfromdifferentbrainnuclei.Theefferentinputfromneocortexaredriven tootherregionsoftheencephalon,suchasneocorticalareasatbothsidesoftheencephalon,thebasal ganglia,andthespinalcord.Thelayeredstructureofneuronsincortexprovidesanefficientwayto organizeafferent‐efferentrelationsofneocorticalneurons(Kandeletal.,2001). Theneuronalnetworkatthebrainfunctionsthroughsynapticinteractionsbetweenglutamatergic principalneuronsandGABAergicinterneurons.Ithasbeenclassicallyconsideredthattheconnectivity betweeninterneuronsandprincipalcellsmaybenonspecific,althoughrecentstudieshavedefinedunique interneurontargetingontospecificcelltypesand/orsubcellularlocations(García‐Junco‐Clementeetal., 2017,2019;Pfefferetal.,2013;Tremblayetal.,2016).Pfefferetal.,2013revealedthat,indeed,thereis aspecificnetworkofinhibitorysynapticconnectionsinthecortex.Thisnetworkisrevealednotonlyby corticalinhibitoryinterneuronscontactingprincipalcellsbutalsointeractingeachother,inallcortical layers.Forexample,theconnectionprobabilitytoneighboringpyramidalcellsiscloseto100%for parvalbumin‐andsomatostatin‐expressingGABAergicinterneurons(HarrisandMrsic‐Flogel,2013),but Figure7.Diversity,classificationandpropertiesofneocorticalGABAergicinterneurons. Almostalltheinterneuronsinneocortexexpressoneofthemainthree,non‐overlappingmarkers:Parvalbumin (PV,blue),somatostatin(Sst,red)andtheionotropicserotoninreceptor5HT3a(5HT3aR,green‐yellow). Additionalsubdivisionswithineachmainpopulationexistbasedondifferentfeatures;hereweshow morphologicalstructureandcellularandsubcellulartargetingbiases(ModifiedfromTremblayetal.,2016).BC, basketcells;SBC,singlebouquetcells;NGFC,neurogliaformcells;CCK,cholecystokinin;PV,parvalbumin;Sst, somatostatin ; 5HT3aR , ionotro p icserotoninrece p tor ; L1 , corticalla y erI.
Introductionkk 26 Consortium,2018;Usoskinetal.,2015;Zeiseletal.,2015).Thefirsttranscriptomicexperimentproduced bysingle‐cellRNAsequencingwaspublishedbyTangetal.,in2009usingamouseblastomere,afterwhat awiderangeofstudiesbeganinthisfieldandwerepublishedonlyfewyearslater. 7.2.Cellisolationandsequencingmethods TosequencethewholemRNAfromasinglecell,twoprincipalchallengesneedtobeovercome: (1)isolation/capturingofsinglecellsand(2)amplificationofaminimalamountofmRNApercell.Forthe firstgoal,severalapproacheshavebeenestablished,suchasthewidely‐usedfluorescent‐activatedcell sorting(FACS).Moreover,somecompanieshavedevelopedadvancedcelldispensertechnologiesto captureindividualcellsbeforesequencing.Forexample,theFluidigmC1platformwasdesignedtoisolate cellsusingamicrofluidicchipwherecellsremainarrestedincapturesites;later,throughimaging techniques,itispossibletoselectfluorescent/healthycells(Islametal.,2014;Marquesetal.,2016; Muñoz‐Manchadoetal.,2018;Pollenetal.,2014;Zeiseletal.,2015).Therearealsomicrodopletmethods, suchthoseusedforDrop‐Seqor10XGenomics,whichinvolvecellisolationusingaqueousdropletsina non‐aqueoussuspension(GrünandVanOudenaarden,2015;Kolodziejczyketal.,2015;Macoskoetal., 2015;Poulinetal.,2016).Othermethodsalsousedforcellisolationarelasercapturemicrodissection, cytoplasmicaspiration(suchasPatch‐Seq),andmanualorautomatedmicropipetting(GrünandVan Oudenaarden,2015;Kolodziejczyketal.,2015;Muñoz‐Manchadoetal.,2018;Poulinetal.,2016).More recently,theWaferGenICELL8platformhasemergedtoincreasethethroughputofmicrofluidiccell capture(FluidigmC1)byincreasingthenumberofcellsperchip(9600wells). DifferentmethodsforRNAsequencinghavebeendeveloped,beingSmart‐seq(ortheimproved methodSmart‐seq2byPicellietal.,2013)andSTRT‐seq(Islametal.,2011)(ortheadvancedversionSTRT‐ seq‐2ibyHochgerneretal.,2017)themostusedprotocols(GrünandVanOudenaarden,2015; Kolodziejczyketal.,2015;Poulinetal.,2016).Thesetechniquesemploydifferentapproachestolabeland countmRNAmolecules,includingUMIs(uniquemolecularidentifiers)orRPKM(readsperkilobaseper millionmappedreads)forSTRT‐seqandSmart‐seq2protocols,respectively.Nowadays,bothmethodsare usuallyrunonIlluminaHiSeqtechnology. Isolatingwholesingle‐cellsmightresultdifficultinhighlyinterconnectedtissues,suchasthebrain. Indeed,somecelltypesaremorevulnerabletotissuedissociationthanothers,includingparvalbumin interneurons,whichaddstheneedtoselectivelyenrichthispopulationusingCre‐driverlinestoensurea sufficientsampling(Bakkenetal.,2018;Lakeetal.,2017).scRNA‐seqincludesmorelikelythecytoplasmic RNAcontent,butalsonuclearRNAmightbesequencedsinceusingthewholecell.Besides,nucleolar, mitochondrialandotherorganelle‐specificRNAsmightbealsocaught.Nowadays,thereareemerging worksshowingacomparableresultusingsingle‐nucleustranscriptomics,whichareapparentlymore resistanttocellisolation‐basedtranscriptionalartifactsderivedfromwholecells(Abdelmoezetal.,2018;
Introduction 27 Bakkenetal.,2018;Grindbergetal.,2013;Lakeetal.,2017).Nevertheless,scRNA‐seqcanrevealRNA abundancywithhighquantitativeprecisionbut,sofar,alimitationisthatweonlycaptureastaticsnapshot atatimepointinacell(LaMannoetal.,2018). 7.3.Up‐to‐datePVclassificationbyscRNA‐seq Normalbrainfunctionismaintainedbyavariedsetofdifferentiatedcelltypesthatdevelop specializedrolestosupportneuralprocesses.Incortex,neuronalnetworkintegrationisessentialto discernactivityandbehaviorpatterns,involvingahighdiversityofGABAergicinterneurons.Severalclasses ofGABAergicinterneuronshavebeendescribedbasedonsingle‐cellRNAsequencinginspecificcortical, hippocampalorstriatalregionsbymultiplepublishedworks(Harrisetal.,2018;Hochgerneretal.,2017; Muñoz‐Manchadoetal.,2018;Pauletal.,2017;Poulinetal.,2016;Tasicetal.,2018;Zeiseletal.,2015). Interestingly,alsoseveralsubpopulationsofPVinterneuronshavebeendefinedbysomeofthese publications:e.g.,Harrisetal.,2018differentiatedbetween7subclassesofhippocampalCA1PVcells,and Tasicetal.,2018described10PVpopulationsinvisualandanteriorlateralmotorcortex(theycanbealso differentiatedbycorticallayerorigin).TheseworksareattemptingtodefinediscretePVinterneurons classification,althoughdemonstratingavariabilitythatdependsonreaddepth,samplesize,sequencing methodandclusteringalgorithms,whichshouldbeconsideredforacompleteunderstanding.PVbasket andchandeliercellsexpressionpatternsbasedonsynapsespecificityduringdevelopment(Favuzzietal., 2019)havebeenalsocomparedwithdefinedclustersfromscRNA‐seqdata(Pauletal.,2017;Tasicetal., 2016;Zeiseletal.,2015).Hopefully,futureinvestigationswillrevealnewinsightsintoregion‐,layer‐and synapse‐specificPVidentities. 7.4.Transcriptomicsinneurodegenerativediseases Transcriptomestudiesaboutneurodegenerativedisordersarenotcurrentlyveryabundant.Gene expressionprofilingshavebeenusedtoquantifyexpressionlevelsofdozenstothousandsofgenes betweencontrolandpatientbrainstouncoverthemolecularbiologyofneurodegenerativediseases, includingAlzheimer’sdisease,Parkinson’sdiseaseandschizophrenia(Cooper‐Knocketal.,2012;Labadorf etal.,2018;Mufsonetal.,2006).However,theseapproachesanalyzealimitednumberofgenesandthus provideonlysetsofdifferentiallyexpressed(DE)genesthatmaybeinvolveinsomebiologicalpathways. Thesingle‐cellRNAseqtechnologyprovidesahigh‐throughputtooltoexhaustivelyidentifyDEgenesalong thewholetranscriptomethatcouldariseaspowerfulcandidatemarkersoftheneurodegenerative condition.Nevertheless,themajorityofscRNA‐seqpublicationsarenotfocusedonthecomparison betweentwoconditions(e.g.controlversusinjured/knock‐out/treatedmice,humantissuesorcells)but oncelltypeclassification.Fewworksarenowemergingonthistopic,wheretheycompareaspecificcell typepopulation,findingdifferentiallyexpressedgenesintheimpairedconditionand/ortheemergenceor disappearanceofcertainsubpopulations(Butleretal.,2018;Mungubaetal.,2019;Tayetal.,2018).We
Introductionkk 28 considerthatscRNAsequencingprovidesarobustmethodthatcanbeappliedtodiscerngeneexpression patternsinneurodegenerativedisorders.
GOALS
Goals 31 Ourknowledgeofthemolecularmechanismsunderlyingneurodegenerationisstillpoor,atleast tobeabletodesigntherapiestoameliorateneuronalandsynapticdysfunction.Neurodegenerationisa multistepprogressiveprocessthatdoesnotstartasageneralizedsynchronizedneuronalphenomena withinthebrain.Sickneuronsco‐existwithhealthyneuronsand,likely,thehealthybrainreactsuponthe neurodegenerationofdamagedneuronalsubpopulations.Ontheotherhand,atthesinglecelllevel, neuronaldegenerationisprobablenotamerepassive“wearandtear”phenomenonbut,instead,anactive processinvolvingprogressivehomeostaticmolecularchangesingeneexpression.Indeed,itiscurrently unknownhowjustsynapticdegenerationorevenlongtermsynapticdysfunctionimpactonthedynamics oftheneuronaltranscriptome.AsdescribedintheIntroduction,previousstudieshaveidentified CSPα/DNAJC5asasynapticvesicleproteinthat,bysofarnotwellunderstoodmechanisms,promotesthe maintenanceofthesynapsesofhighlyactiveneuronssuchasfast‐spikingGABAergicinterneurons. GeneticallymodifiedKOmicelackingCSPα/DNAJC5diearoundonemonthofageand,therefore,present limitationstostudymolecularmechanismsofneurodegenerationbeyondearlyadulthood.Thisthesis pursuestheinvestigationoftheneuralandmolecularmechanismsalongtheadultlifeunderlyingsynaptic degenerationuponthegeneticdeletionofCSPα/DNAJC5specificallyinPVGABAergicinterneurons. Accordingly,themaingoalsofthisthesisare: 1. GenerationandvalidationofanovelconditionalKOmousespecificallylackingCSPα/DNAJC5inPV GABAergicinterneuronsexpressing,inthesameneurons,afluorescentmarkerandanoptogenetic actuatorsuitableforlight‐mediatedstimulation. 2. Phenotypiccharacterizationofthemice,includingneurological,biochemical, immunohistochemicalandelectrophysiologicalanalysisofPVGABAergicneuronsandsynapses. 3. Investigationofgeneexpressionchangesassociatedtosynapticandneuronaldysfunctioninthe absenceofCSPα/DNAJC5basedonsingle‐celltranscriptomicscomparativeanalysis.
MATERIALSANDMETHODS
MaterialsandMethods 35 1. Mice Tofulfiltheexperimentaltasksproposedinthisthesis,severalmouselineswererequired.For initialexperiments,weusedconventionalDnajc5knockout(KO)animals(TheJacksonlaboratory,stockno. 006392,ThomasC.Südhof)(Fernández‐Chacónetal.,2004),althoughconditionalmouselinescarrying Dnajc5floxedallelesweregeneratedforthemajorityoftheproceduresdescribedinthiswork.For experimentsinvolvingconditionalDnajc5flox/floxmice,twoCre‐recombinase‐expressingmouselineswere purchasedforbreeding:UBC‐Cre‐ERT2(TheJacksonLaboratory,stockno:007001,EricBrown)andPVcre (TheJacksonLaboratory,stockno:008069,SilviaArber)(Hippenmeyeretal.,2005).Finally,wehavealso takenadvantageoftheAi27Dmouseline(TheJacksonLaboratory,stockno:012567,HongkuiZeng) (Madisenetal.,2012)tointroducethechannelrhodopsin‐2/tdTomato(ChR2/tdTomato)fusionproteinin theexperimentalPVcremouseline,and,then,beingabletolabelPVcellsusingtdTomatoasreporter.The strategiesforthegenerationoftheDnajc5flox/floxandthetwoCre‐recombinase‐expressingmouselinesare describedindetailbelow. AllproceduresinvolvinganimalresearchwereperformedinaccordancewiththeEuropeanUnion Directive2010/63/EUabouttheprotectionofanimalsusedforscientificpurposesandapprovedbythe CommitteeofAnimalUseforResearchattheUniversityofSeville.Animalswerekeptinanimalfacilities (IBiSandcampusMacarena)witha12/12‐hlight/darkcycleandunrestrictedaccesstofoodandwater. 1.1.GenerationoftheconditionalDnajc5flox/floxmouseline Embryonicstem(ES)cellswereacquiredfromtheEuropeanConditionalMouseMutagenesis Program(EUCOMM).Themutantcellsweregeneratedbyrecombinantinsertionofaknock‐outfirst mutantconstructcontainingloxPsitesflankingexon3ofthemouseDnajc5gene(ENSMUSE00000170833 chr2:181282045‐181282258).ThefullsequenceofthemousetargetedKO‐first,conditionalready,lacZ‐ taggedmutantalleleDnajc5:tm1a(EUCOMM)fromtheWellcomeTrustSangerInstituteisavailableat GenBank(accessionnumberJN952788).ThreedifferentEScellcloneswereexaminedbasedonthe karyotypeandtheirmorphologicalandgrowingpropertiesinculture.Oneofthem(JM8A3.N1‐E01)was chosenandinjectedintoC57BL/6JOlaHsdmouseblastocystsattheUnitatd’AnimalsTransgènics (UniversitatAutònomadeBarcelona).Atotalof12chimericmicewereobtainedwithvariabledegreeof chimerismandamongthemonemalewithhighchimerism(90%)wasusedforbreedingwithC57BL/6NHsd femalemiceandgavegermlinetransmission.Interbredofresultantgermlinemutantmice(Tm1a heterozygousmice)yieldedTm1ahomozygousmicethatlackedDnajc5anddevelopedthetypicalearly neurologicalphenotypedescribedforconventionalDnajc5KOmice.Next,germlinemutantmicewere crossbredwithROSA26:FLPeknock‐inmice(TheJacksonLaboratory,Stockno.003946)toremovetheFRT‐ flankedspliceacceptorsite,β‐galreporter,andneomycinresistancecassette.ResultantDnajc5flox/+mice wereinterbredtogeneratehomozygousmutantmicewithconditionalknockoutpotential(Dnajc5flox/flox).
MaterialsandMethods 42 andUBC‐cre‐ERT2:Dnajc5flox/floxanimals(P100).TheRNAwasextractedbyRNeasyMiniKit(74104,Qiagen) followingmanufacturer’sinstructions.Fewmodificationsweremadetothisprotocol,suchasusingTRIzol reagent(15596026,Ambion),chloroform(22711.260,AnalaRNORMAPUR)and70%ethanol(1009832500, EMSURE)toefficientlylysate,separateandclear,respectively,samplesbeforetransferringthemto columnsandcontinuingwiththecustomQiagenprotocol.Ontheotherhand,tdT+andtdT‐cellsfrom PVcre:Ai27D:Dnajc5flox/+(n=3)andPVcre:Ai27D:Dnajc5flox/‐(n=3)littermatesweredirectlysortedinto200µl ofRNAlaterTMSolution(AM7020,Invitrogen)toimproveRNAstorageandstabilizationat‐80°Cuntil processing.Forsortedcells,wecarriedoutaspecificprotocolthatimprovedRNAobtaining.First,frozen cellswerethawedonice,centrifugedat2600gfor1minat4°Candsupernatantcarefullydiscardedusing apipette,leavingaminimalvolumeinthetubeduetonotvisiblepellet.500µlofTRIzolwasaddedto samplesandhomogenizedbyvortexfor1min,thenlefttorestatroomtemperatureatleast5min. Chloroformwasthensupplemented,andsamplesvigorouslyshakenbyhandfor15secandthenleftto incubateatroomtemperaturefor2min.Acentrifugationstepat12000gfor15minat4°Cwasperformed tohelpphenol/chloroform‐basedphaseseparation.Upperaqueousphasewascollectedinanewtube, and250µl2‐Propanol100%(I9516,Sigma)and0.5µlglycogen(stock20mg/ml,10901393001,Roche) wereaddedbeforeshakingbyhand,andincubatedO/Nat‐20°CtofacilitateRNAprecipitation.Nextday, sampleswerecentrifugedat12000gfor10minat4°C,supernatantdiscardedandpelletwashedwith1 ml75%ethanolbyvortexing.Sampleswereagaincentrifugedat7500gfor10minat4°C,supernatant removedandpelletwasairdriedduring10‐15min;then,RNAwasresuspendedin15µlRNase‐freewater andincubatedat55°Cduring15min.2µlRNAwasusedtotestRNAqualityandquantityby2100 Bioanalyzer(EukaryoteTotalRNAPico,Agilent)beforefinallykeptat‐80°Cuntilreverse‐transcription.RNA IntegrityNumber(RIN)forallsampleswerearound7‐8,whichwerequiteoptimumconcerningsorted cells,andconcentrationsrangedfrom1‐5ng/µl,hencenocorrectionsweremadetoequalizesample concentrationduetolowquantity. 3.2.2.Reverse‐transcription(RT)tocDNA TheRNAreverse‐transcription(RT)wasperformedusingatwo‐stepprotocolbasedonthe TranscriptorFirstStrandcDNAsynthesiskit(04379012001,Roche).Startingmaterialforthefirstreaction was1µgofRNA.Becausewecouldnotobtainsuchamountinoursamples,thefirstreactionmixwas changedtoincludeasmuchstartingmaterialaspossible;inthiscase,thetotalvolumeofourremaining RNA(12µlRNA),wasmixedto1µlofRandomhexamerprimer600µMtoobtainthemaximumcDNA quantity.Thismixturewasincubatedinathermalcyclerfor10minat65°Cfordenaturationandcooled downonice.Thesecondreactionwasthenpreparedbymixing4µlTranscriptorRTReactionBuffer5X, 0.5µlRNaseInhibitor40U/µl,2µldNTPMix10mMeachand0.5µlTranscriptorReverseTranscriptase20 U/µl,andfinally7µlofthismixturewasaddedtotheprimer‐templatemixturetube.Reagentsweremixed
MaterialsandMethods 43 andcentrifugedtocollectsamplesatthebottom.Reverse‐transcriptionwasrunusingthefollowing programsettings:25°C10min;50°C60min;85°C5mintoinactivatetranscriptase;4°Choldtostop reaction.Samplesweredirectlyfrozenat‐20°Cuntiluse,presumablycontainingthesameamountofcDNA asstartingRNA(from12to60nginatotalof20µl). 3.2.3.PCRforDnajc5cDNA PCRconditionsandmixwerefirstsetupwithcontrolsamplesandanextrasortedsample.PCRmix wasprepared,including17.3µlmiliQwater,2.5µlBufferPCR10X,1µlMgCl250mM,0.5µlMLM‐2005‐ 0110µM,0.5µlMLM‐2005‐0210µM(Table2;S1),0.2µldNTPsmix100mMand1µlhandmadeTaq polymerase.2µlofcDNAsampleswasaddedto23µlofPCRmix,reachingatotalvolumeof25µlper reaction.Reactionswererunaccordingtotheseprogramsettings:95°C5min;35cycles:95°C30sec,60°C 30sec,72°C30sec;72°C2min;10°Chold.Sampleswerethensupplementedwithloadingbuffer5Xand separatedina2%‐agarosegelwithethidiumbromidebyPowerPac200(Bio‐Rad)at100mVduring40min forbetterresolution.GelwasvisualizedusingtheBioDoc‐It™220ImagingSystem(UVP)withanM‐26 ultraviolet(UV)transilluminatorandpicturesweretaken. LocusPrimer nameSequence5’3’Primer typeTemplateSize Conditional Dnajc5flox/flox SLB‐ 2017‐01CACGTCAGATAGTCTTGGGGTCTForwardgDNAWild‐type=765bp Tm1c=1029bp Tm1d=245bp SLB‐ 2017‐02GCAAATGGTTGTGCTCCATGTACGReversegDNA MLM‐ 2005‐01ACCATGTTCTTGGACTGGACAForwardmRNA (cDNA) Wild‐typeorfloxed withoutrecombination =348bp Mutant=134bp MLM‐ 2005‐02AGCAGCAGTTAAAGCAACAGCReversemRNA (cDNA) 3.3.Proteinvalidation AccordingtotheInternationalMousePhenotypingConsortium(IMPC),theanalysisofthemutant alleleDnajc5tm1a(EUCOMM)WtsipredictsthatthefourmRNAtranscriptvariantsforDnajc5wouldgeneratetwo truncatedproteinproductsafterremovalofthefloxedregion.ThethreemRNAtranscriptscodingthe samewild‐typeproteincontaining198aa(ENSMUST00000072334,ENSMUST00000108797, ENSMUST00000108796),afterremovalofDnajc5,wouldproducethefollowingpeptidewith112aa: MADQRQRSLSTSGESLYHVLGLDKNATSDDIKKSYRRCLLFVASSPAATAAAVCAVALTAAVGNASPRHLRVRRQNSTY PLKTWRHSCSLMKGRLQTHRSSYSQHLPQRPPS*,withanestimatedmolecularweightof12.38kDa(S1).The fourthwild‐typenon‐codingtranscript(ENSMUST00000116365),afterremovalofDnajc5,wouldgenerate thefollowingpeptideconsistingof112aa: Table2.PrimersforDnajc5 gDNAandmRNAdetection. Summaryincludingname,sequence,type,templateandproductsizeforprimersusedtodetectDnajc5genomic DNA(gDNA)andmRNA(cDNA),tovalidateCre‐dependentrecombinationatthislocus.
MaterialsandMethods 44 MADQRQRSLSTSGESLYHVLGLDKNATSDDIKKSYRRCLLFVASSPAATAAAVCAVALTAAVGNASPRHLRVRRQNSTY PLKTWRHSCSLMKGEGTDTVPRVFVVASGTVEV*,withanestimatedmolecularweightof12.07KDa(S1).For bothpredictedtruncatedproteins,thefirst36aawouldcorrespondtotheN‐terminusofCSPα/DNAJC5 whiletherestwouldbeanaberrantpolypeptide.Ifthosepolypeptideswerestable,theDnaJdomain (aminoacids15to84)wouldbedisruptedbecauseonlythefirst22aminoacidsofsuchadomainwouldbe preserved,accordingtotheConservedDomainstoolfromNCBI(S1).Hence,inordertocheckifindeed anytruncatedversionofCSPα/DNAJC5wasproduced,weanalyzedproteinexpressionaround12KDain theUBC‐Cre‐ERT2:Dnajc5flox/floxandPVcre:Ai27D:Dnajc5floxmouselinesusingimmunoblotswithR807,an antibodyraisedagainstaGST‐fusionproteinofCSPα/DNAJC5thatbindstothefull‐lengthversionofthe protein(Fernández‐Chacónetal.,2004;Tobabenetal.,2001)(Table4).See8.Immunoblottingsectionfor methodologydescription. 4. Real‐timequantitativereversetranscription‐PCR(real‐timeqRT‐PCR) 4.1.RNAextraction ThesameprotocolforRNAextractionontdT+andtdT‐sortedcellswascarriedoutforreal‐time qRT‐PCRinlittermatepairsofPVcre:Ai27D:Dnajc5flox/+andPVcre:Ai27D:Dnajc5flox/‐animals(n=3).RNAquality andconcentrationwereassessedby2100Bioanalyzer,andRINswerefoundoptimal(around7‐8.5)forall samples,exceptfortwosampleswheretheRINwasnotpossibletobedetermined.Anyway,allsamples werestoredat‐80°Candusedfordownstreamanalysis. 4.2.Reverse‐transcriptionandcDNAamplification Becausewepreviouslyobservedverylowamount(1‐5ng/μl)ofstartingRNAmaterialfromsorted cells,andsamplesweredepletedafteracoupleoftests,wedecidedtomaximizepre‐amplifiedRNAbefore reverse‐transcriptionusingtheQuantiTectWholeTranscriptomeKit(207043,Qiagen)andfollowing manufacturer’sinstructions.Briefly,10ngofeachsamplewaspreparedinatotalvolumeof5µlnucleasefreewater,and5µlofRTmix(4µlT‐ScriptBuffer,1µlT‐ScriptEnzyme)wasaddedbypipettingandthen centrifuged.Thismixturewasincubatedinathermocyclerforreverse‐transcription:37°C30min;95°C5 min;22°Chold.Secondly,theadditionof10µlligationmix(6µlLigationBuffer,2µlLigationReagent,1µl LigationEnzyme1,1µlLigationEnzyme2)totheRTreactionandlaterincubationat22°Cduring2h producedcDNAligation.Finally,30µlofamplificationmix(29µlREPLI‐gMidiReactionBuffer,1µlREPLI‐ gMidiDNAPolymerase)wasaddedtotheligationreaction,andthenincubatedat30°Cduring8hfora high‐yieldreaction(upto40µgcDNA),at95°Cfor5mintoinactivateallenzymesand10°Chold.Every samplewasruninan2%‐agarosegeltocheckforthepresenceofamplifiedcDNAbeforeperformingthe real‐timeqRT‐PCR,andmeasuredbyNanoDrop2000,whichrevealedconcentrationsbetween1700‐1900
MaterialsandMethods 45 ng/µlandgoodqualityratios.Samplesweredilute½inTE2X(20mMTris‐HClpH8,2mMEDTA)and storedat‐20°Cuntiluse. 4.3.Real‐timeqRT‐PCRandquantification Thereal‐timeqRT‐PCRwascarriedoutina384‐wellplateusingreducedvolumeperreaction:2.5 µlTaqManFastAdvancedMasterMix(4444557,AppliedBiosystems),0.25µlTaqManGeneExpression Assay(Table3)and2.25µlcDNAofa1/250dilutionfromtheoriginalconcentration.Triplicatereactions wererunusingthe7900HTFastReal‐TimePCRSystem(AppliedBiosystems)for45cycles:95°C15sec, 60°C1min.AverageCt(cyclethreshold)valuesforPvalbfromtriplicateswerestandardizedwitha housekeepinggene(beta‐actin)foreachsample(ΔCt)andbothsortedfractionswerecompensatedby substractionofthemeanoftdTomato‐negativefractionsvalues(ΔΔCt).Datawerefinallyrepresented usingthemean2‐ΔΔCtvaluesnormalizedagainstthemean2‐ΔΔCtvaluefromalltdTomato‐negativefractions, separatelyforPVcre:Ai27D:Dnajc5flox/+andPVcre:Ai27D:Dnajc5flox/‐mice,usingGraphPadPrism6software. ProbeGeneExonboundaryReference Mm00443100_m1ParvalbuminProbespansexons1‐24453320,Applied Biosystems Mm02619580_g1Actin,betaProbemapswithinexon34334482,Applied Biosystems 5.Survivalanalysis Inourmousecolony,wemaintainedareasonablenumberofmiceforbothPVcre:Ai27D:Dnajc5flox/+ andPVcre:Ai27D:Dnajc5flox/‐genotypesuntil8months,whichwereusedfordistinctexperimentsduringthis workandthereforemakingitpossibletoelaborateasurvivalcurvefrom0to8months.29malesand29 femaleswereusedforthatpurpose.DatawereanalyzedbyGraphPadPrism6usingaKaplan‐Meier survivalcurveandthestatisticsprovidedbythistest(Log‐rankMantel‐Coxtest). 6.Bodyweightcurve Atotalof5malesand5femalesforPVcre:Ai27D:Dnajc5flox/+mice,and5malesand4femalesfor PVcre:Ai27D:Dnajc5flox/‐ werereservedtoperformabodyweightstudy.Becausethephenotypical differencesbetweencontrolandmutantmicebecameclearlyvisiblearoundP30,thisassaywasdoneby weightinganimalsonceaweekfromP30toP79(around3months),andonceamonthfromP79toP240 (8months).AnalysiswascarriedoutwithGraphPadPrism6,whereeachtimepointwascompared betweencontrolandmutantmice,separatingdatabysexduetophysiologicaldifferencesinbodyweight inmalesandfemales. Table3.TaqManGeneExpressionAssayprobes. DescriptionofTaqmanprobesusedforreal‐timeqRT‐PCRexperiments.
MaterialsandMethods 46 7.Openfield Fortheanalysisoflocomotoractivity,weutilizedtheopenfieldtest.Beforestarting,itwasneeded tosetupseveralparametersonBiobservesoftwaretooptimizemotionidentification (http://www.biobserve.com/behavioralresearch/products/viewer/):(1)anappropriatemousebody detection(Figure11),(2)calibrationofcagedimensions(conversion:pixelstocm,squareboxes),(3) configurationoftheworkingarea(thewholecage),(4)establishmentofzones(inthiscase,onlythe workingarea)and(5)selectionofadegreeofsensitivitytoavoidbackgroundcontamination.Allrecordings werere‐analyzedonalabcomputerbyViewer3todefinethesameparametersandmaketrackcorrections incaseofnoisysignals.Wecarriedoutthisexperimenton44mice(23females:n=9PVcre:Ai27D:Dnajc5flox/+ andn=14PVcre:Ai27D:Dnajc5flox/‐;and21males:n=11PVcre:Ai27D:Dnajc5flox/+andn=10 PVcre:Ai27D:Dnajc5flox/‐)during30minindark.UVcameraslocatedontopoffourcagesrecordedfourmice perexperiment.Automatically,Biobservesoftwaregeneratedrecordingdataandasummaryofbehavioral parameters.Amongthem,weselectedvelocity(cm/s),activity(percentage(%)oftimethemouse overcametheactivitythresholdof1cm/s),tracklength(cm),walldistance(cm),andnumberofheadbobs, headstretches(Figure11),tailmovesandambulations(spontaneousshort‐termaccelerationsthat overcametheambulationthresholdof60cm/s).AllparameterswereanalyzedusingGraphPadPrism6, separatelyformalesandfemales,andalltogether. 8.Immunoblotting Cortexand/orliversampleswereextractedfrom(1)Dnajc5WTandKOlittermatemiceatP15and P30,(2)PVcre:Ai27D:Dnajc5flox/+andPVcre:Ai27D:Dnajc5flox/‐littermatemiceat2‐month(adultmice)and8‐ month(oldmice)postnatalage,and(3)Dnajc5flox/floxandUBC‐Cre‐ERT2:Dnajc5flox/floxlittermatemiceat P100. Micewereanesthetizedwith2%tribromoethanol,andfreshtissues,flash‐frozeninliquidnitrogen, werekeptat‐80°Cuntilprocessing.TissuelysisandhomogenizationwerecarriedoutinRIPAbuffer(20 mMTris‐HClpH7.4,150mMNaCl,1mMEDTA,1%IGEPAL,0.1%SDS)supplementedwithproteaseand Figure11.Mousebodyidentificationandtypesofheadmovementsanalyzedduringtheopenfield experiments. DrawingsextractedfromtheBiobserveopenfielddocumentationshowingmouseidentification(head,body,tail) (left)andnaturalheadmovements(right)thatwereanalyzedduringvideorecording.
MaterialsandMethods 47 phosphataseinhibitors(05892791001and04906837001,Roche).Sampleswerethenmaintainedin continuousagitationfor2hoursoniceandvortexedevery15‐30minutes.Afterthat,sampleswere centrifugedatmaximumspeed(14000rpm)for15minutesat4°C,andsupernatantswerecollectedas lysates.ProteinconcentrationoflysateswasestimatedusingPierceBCAProteinAssayKit(23227,Thermo Scientific),followingmanufacturer’sinstructions.LysateswerethenprocessedwithLaemmlibuffer4X supplementedwith0.4Mbeta‐mercaptoethanol,reachingafinalconcentrationofLaemmliat1X(50mM Tris‐HClpH6.8,10%glycerol,2%SDS,0.0067%bromophenolblue,0.1Mbeta‐mercaptoethanol),and heatedat95°Cfor10‐15minutesinathermomixerwithoutagitation.Percentageofacrylamideforgels werechosendependingondesiredproteinmolecularweights.Weused10%‐acrylamidegelsfor separationofmedium‐molecularweightproteinsamong25‐75KDaand14%forseparationofsmall proteinsamong10‐20KDa.Equivalentamountsofprotein(40µg)andvolume(30µl)wereanalyzedby SDS‐PAGEusingRunningBuffer1X(25.5mMglycine,2.5mMTris,0.1%SDS,pH8.4±0.2).Sampleswere runfirstat50mVforapproximately30minandfinallyat100mVfor1hourandahalfmore.Afterthat, samplesinacrylamidegelsweretransferredtoPVDFmembranes(1620177,Biorad)intocoldTransfer Buffer1X(25.5mMglycine,2.5mMTris,20%methanol,with0.1%SDSincaseofmedium‐molecular weightproteins,orwithoutSDSifsmallproteins)at200mAfortwohoursifmedium‐proteinsortwohours andahalfifsmallproteins.MembraneswerethenwashedthreetimesinTris‐BufferedSalineandTween 20(TBS‐T)1X(13.7mMNaCl,10mMTris,0.1%Tween20,pH7.5)andblockedwith3%BSAor3%milkin TBS‐Tduring1houratroomtemperaturewhilemildlyagitating.Then,membraneswereimmunoblotted withprimaryantibodies(Table4)inthesameblockingsolutionO/Nat4°Cwhilemildlyagitating. Posteriorly,theywerewashedthreetimeswithTBS‐Tandincubatedduring1houratroomtemperature whilemildlyagitatingwithappropriatehorseradishperoxidase(HRP)‐conjugatedsecondaryantibodiesin thesameblockingsolution:goatanti‐rabbitIgG(1:10000;111‐035‐144,JacksonImmunoResearch)and goatanti‐mouseIgG(1:10000;115‐035‐166,JacksonImmunoResearch).Mouseanti‐β‐actinwasusedas loadingcontrol.ClarityTMWesternECLSubstrate(170‐5060,BIORAD)wasusedforchemiluminescent westernblotdetection.ImagesweretakenusingtheChemidocTouchImagingSystem(ThermoFisher Scientific). 9.Quantificationofproteinlevels Afterimmunoblotting,proteinquantificationwasperformed.Imageswereacquiredusingthe sameexposuretimesandbandswereanalyzedusingImageJ(https://imagej.nih.gov/ij/download.html), anopensourceimageprocessingprogram.Briefly,bandswereselectedwitharectangularbox(Ctrl+1) andlaneswereplotted(Ctrl+3).Weusedstraightlinestoseparatebetweenbandareasandfromthe background,andwequantifiedeachbandareausingthewand(tracing)tool.Datawerecollectedand
MaterialsandMethods 48 compensatedwiththehousekeepingprotein(β‐actin)expressionlevelforeachwell.Datawere representedinbardiagramsusingnormalizeddataversuscontrolmiceinGraphPadPrism6software. 10. Perfusionandsectioningusingacryostat Anesthesiausedwas2%tribromoethanolon(1)Dnajc5WTandKOlittermatemiceatP30and(2) PVcre:Ai27D:Dnajc5flox/‐andPVcre:Ai27D:Dnajc5flox/+littermatemiceat2‐monthand8‐monthpostnatalage. Miceweretranscardiallyperfusedwith1XPhosphateBufferedSaline(PBS)(137mMNaCl,2.7mMKCl,10 mMNa2HPO4,2mMKH2PO4,pH7.4)toeliminatebloodfromtissues,andthenwith4%paraformaldehyde (PFA)inPhosphateBuffer1X(PB)(19.98mMNa2HPO4,79.59mMKH2PO4,pH7.4)tofixtissues.Brains wereremovedandpost‐fixedO/Nin4%PFAat4°C.Brainswerethencryoprotectedin30%sucroseinPB (with0.01%sodiumazide)andaLeicaCM1950cryostatwasutilizedforserialsectioningusingcoronal planes;oneoutofeverysixthbrainslices(40µmthickness)wasanalyzedforimmunohistochemistryand immunofluorescenceexperiments.Sliceswerestoredat‐20°C(50%glycerolinPBS1X)untiluse. 11. Brigthfieldimmunohistochemistry CoronalslicesofthemotorcortexfromPVcre:Ai27D:Dnajc5flox/+andPVcre:Ai27D:Dnajc5flox/‐mice werewashedwithPBS1Xandincubatedduring20minutesatroomtemperaturewith3%H2O2and10% methanolinPBStoblocktheendogenousperoxidase.SliceswerethenwashedwithPBSandblockedina solutioncontaining3%FBSand0.3%TritonX‐100inPBSfor1houratroomtemperature,whilemildly agitating.SectionswereincubatedO/Nwithprimaryantibodies(Table4)at4°Cusingthesameblocking solution.SeveralwashesinPBSwereperformedbeforeincubationusingappropriatesecondary biotinylatedantibodies(1:1000dilutionin0.3%Triton/PBS)for1houratroomtemperaturewhilemildly agitating:donkeyanti‐rabbitIgG(711‐065‐152,JacksonImmunoResearch)anddonkeyanti‐mouseIgG (711‐065‐151,JacksonImmunoResearch).SliceswerewashedseveraltimesusingPBS,andthenincubated inanavidin‐biotin‐peroxidasesolution(ABCkit,VECTASTAIN,VectorLaboratories)for1hourwhilemildly agitatingatroomtemperature.Afterwashing,adevelopingsolutioncontaining0.02%diaminobenzidine aschromogen,0.04%nickelammoniumsulfateand0.01%H2O2inPBwasaddedtorevealthe immunoperoxidaselabeling,onemakereachtime.Finally,PBwasaddedtostopthereactionwhen appropriate,followedbyseveralwashesinPB.Sectionsweremountedonpreviouslygelatinizedglass slidesandletdry.Slideswerethenpassedthroughanalcoholbatterytodehydrateallsections(30sec, water;30sec,ethanol70%;30sec,ethanol80%;30sec,ethanol90%;30sec,ethanol100%twice)and throughxylenetoclearthesamples(30sec,xylene;10min,xylene).SlideswerecoverslippedwithDPX (Sigma),andletdry.ImagesweretakenusingaNewCASTBX‐61microscope(Olympus)anda4Xobjective lens(Olympus).
MaterialsandMethods 49 12.QuantificationofPVsomatanumber ImagestakenfromimmunohistochemistryexperimentswereusedforPVsomatanumber quantification.AtotalofthreeXYplaneswithsamedimensionswerecroppedfromeachbrightfieldimage selectingregionsthatcontainedonlylayerII/IIIneuronsfrommotorcortex.PVsomatawasmanually quantifiedbythe“multi‐point”toolfromImageJ,andmeanvalues(n=27fieldspergenotype,from3 PVcre:Ai27D:Dnajc5flox/+andPVcre:Ai27D:Dnajc5flox/‐mice)wereplottedandanalyzedusingGraphPadPrism 6.NumberofPVcellswererepresentedpermm2. 13.Immunofluorescenceinbrainslicesections Approximatelyfifteenserialsliceswereselectedfromeachmousetoanalyzethewholemotor cortex(AP:since2.58mmto1.42mm,frombregma)throughimmunofluorescencetechniques.After severalwasheswithPBS,sectionswereincubatedinblockingsolution(3%FBSand0.3%TritonX‐100in PBS)for1hatroomtemperaturewhileagitating.Primaryantibodies(Table4)werepreparedinblocking solutionandslicesincubatedO/Nat4°C.SectionswererinsedthreetimeswithPBS,andincubatedindark withappropriatefluorescentlylabelledsecondaryantibodiesin0.3%TritonX‐100inPBSfor2hatroom temperature,whileagitating:RhodamineReddonkeyanti‐chickenIgG(1:300,703‐295‐155,Jackson ImmunoResearch),AlexaFluor488donkeyanti‐rabbitIgG(1:500,711‐545‐152,Jackson ImmunoResearch),Cy3donkeyanti‐rabbitIgG(1:500,711‐165‐152,JacksonImmunoResearch),Alexa Fluor488donkeyanti‐mouseIgG(1:500,715‐545‐151,JacksonImmunoResearch)and/orAlexaFluor647 goatanti‐rabbitIgG(1:250,111‐605‐144,JacksonImmunoReseach).Sliceswerethenwashed,mounted andcoverslippedwithDakoFluorescenceMountingMedium(S3023,Dako),andletdryat4°Cindarkfor 1‐2days.ImagesweretakenwithaNikonA1R+confocalmicroscope. 14.Colocalizationanalysis Confocalimagesofmotorcorticalregionsweretakenusinga20Xglycerolimmersionobjective lens(Nikon)tostudythedegreeofcolocalizationbetweenPV+andtdTomato+neurons.Thesamelaser intensitywassetupforimagingacquisitioninallmice.ImageJwasusedforquantification.Briefly,regions havingthesameXYdimensionswerechosentoidentifymotorcorticalneuronsoflayerII/III,forevery image.Then,fromatotalof15‐16serialslicespermouse,weselectedtwoZ‐stacksof1µmfor quantification,obtainingaround30‐32fieldspermouse.Attheend,n≈90fieldspergenotype(fromatotal of3PVcre:Ai27D:Dnajc5flox/+andPVcre:Ai27D:Dnajc5flox/+mice)wereseparatelyandmanuallyquantifiedfor PV‐positive(PV+)somata(greenchannel)andtdTomato‐positive(tdTomato+)perisomallabelling(red channel),usingthe“multi‐point”tool.Pointsdetectedinbothgreenandredchannelswereconsideredas colocalizationvalues.Bythisway,weobtainedatotalaveragenumberofPV+andtdTomato+cells,and thereforewewereabletocalculatethepercentageofPV‐tdTomatocolocalization,andthepercentages
MaterialsandMethods 50 ofPV+cellsandtdTomato+cellsthatdidnotcolocalize.WeusedRStudio1.1.442andRfreesoftware version3.4.4toplotthedatawithaVenndiagramusingthefunctiondraw.pairwise.vennfromthepackage VennDiagram. 15.Quantificationofsynapticpuncta Confocalimagesofmotorcorticalregionswereacquiredusinga60Xoilimmersionobjectivelens (Nikon)toquantifysynapticpunctalabelledbySyt2andPV.Laserintensitywasalwaysthesameforeach age(2monthsor8months)andZ‐planedistancewasestablishedat1µm.TwoplaneswiththesameXY dimensionswereacquiredfrom16‐30motorcortexslices(40µm‐thickness)permouseandseparately quantifiedforPV+(redchannel)andSyt2+(greenchannel)puncta(PV:n=82or67imagesfrom PVcre:Ai27D:Dnajc5flox/+andn=79or72imagesfromPVcre:Ai27D:Dnajc5flox/‐animalsat2and8months, respectively;Syt2:n=76or67imagesfromPVcre:Ai27D:Dnajc5flox/+andn=79or72imagesfrom PVcre:Ai27D:Dnajc5flox/‐animalsat2and8months,respectively;fromatotalof3micepergenotype).For thisquantification,weusedImageJtocreatetwodifferentmacrosforredandgreenchannels.Both macroscontainedthefollowingstepsandwererunineveryimage:(1)selectionofatotalof1‐30stacks inZ‐plane,(2)imageswereconvertedto8‐bits,(3)athresholdlevelwasadjustedforeachchannelfor binaryconversionand(4)punctadensitywascalculatedwith“Analyzeparticles”moduleusingthe thresholdedimageofeveryZ‐stack.WeexcludedPVsomassettingthemaximumpunctasizeto200pixels, andthebackgroundcontaminationsettingtheminimumpunctasizeto10pixels.ForSyt2+puncta,no maximumpunctasizewasstablished.CollecteddatawereanalyzedusingGraphPadPrism6.Average punctadensitywasrepresentedper100µm2. 16.QuantificationofPVsomatasize Usingtheredchannel(PV)oftheconfocal60Ximagesusedforsynapticpunctaquantification,we measuredPVsomatasize(n=104PVcellsfromPVcre:Ai27D:Dnajc5flox/+andn=181PVcellsfrom PVcre:Ai27D:Dnajc5flox/‐animalsat2months;n=152PVcellsfromPVcre:Ai27D:Dnajc5flox/+andn=142PVcells fromPVcre:Ai27D:Dnajc5flox/‐at8months,fromatotalof3micepergenotype).TwoZ‐stacksof1µmfrom 4‐5imagespermousewerequantifiedwithBitplaneImaris.Minimumandmaximumsizedetection parameterswereadjustedseparatelyfor2monthsand8months,torecognizeonlytheneuronsomata labelling(notdendritic)andensuretheseparationofsomasincasetheyappearedsidebyside.PVsomata areadataweredeterminedinµm2andanalyzedwithGraphPadPrism6.
MaterialsandMethods 51 ProteinPrimaryantibodyHost speciesMethodConcentration Reference CSPα anti‐CSPα(C‐terminus)Rabbit WB 1:2000 (BSA) EnzoCat.No.ADI‐VAP‐ SV003‐E CSPα anti‐GST‐fusionprotein ofCSPα(full‐length)Rabbit1:1000 (BSA) (Fernández‐Chacónet al.,2004;Tobabenet al.,2001) Synaptotagmin‐1 anti‐synaptotagmin‐1 cytoplasmictail,clone 41.1 Mouse1:2000 (BSA) SynapticSystemsCat. No.105011 Synaptotagmin‐2znp‐1Mouse1:750 (Milk) DevelopmentalStudies HybridomaBank Hsc70anti‐UncoatingATPase (hsc70),clone3C5Mouse1:1500 (BSA) SynapticSystemsCat. No.149011 SNAP25anti‐SNAP25,clone SMI81Mouse1:1000 (BSA)Biolegend Parvalbuminanti‐parvalbumin,PV 27Rabbit1:750 (Milk)Swant Synaptobrevin‐2anti‐synaptobrevin‐2/ VAMP2,clone69.1Mouse1:2000 (Milk) SynapticSystems,Cat. No.104211 β‐actinanti‐β‐actin,cloneAC‐ 74Mouse1:5000 (BSAormilk)Sigma‐Aldrich,A2228 CSPα anti‐CSPα Rabbit IHC 1:1000 EnzoCat.No.ADI‐VAP‐ SV003‐E Parvalbuminanti‐parvalbumin,PV 27Rabbit1:1000 Swant,PV27 Synaptotagmin‐2znp‐1Mouse1:500 DevelopmentalStudies HybridomaBank SNAP25anti‐SNAP25,clone SMI81Mouse1:5000 Biolegend Parvalbuminanti‐parvalbumin,PV 27Rabbit IF 1:1000Swant tdTomatoCPCA‐mCherryChicken1:1000EncorBiotechonology Inc. Synaptotagmin‐2znp‐1Mouse1:500 DevelopmentalStudies HybridomaBank 17.Electrophysiology Electrophysiologicalrecordingswereperformedattwodifferentpostnatalages,2monthsand8 months.Foreachtimelapse,PVcre:Ai27D:Dnajc5flox/+(n=3)andPVcre:Ai27D:Dnajc5flox/‐(n=3,withthe exceptionofn=4forintrinsicpropertiesat2months)animalswereused.Intrinsicelectrophysiological propertiesandexcitatorypostsynapticpotentials(EPSPs)ofparvalbumininterneurons,andminiature inhibitorypostsynapticcurrents(mIPSCs)onglutamatergiccellswereanalyzedatbothagestoevaluate thedegreeofinhibitioninthemotorcortexofourmousemodel. Table4.Primaryantibodies. Summaryofprimaryantibodiesemployedduringthisworkinwesternblot(WB),immunohistochemistry(IHC) andimmunofluorescence(IF).
MaterialsandMethods 58
MaterialsandMethods 59 around2‐3Kb(ESCGdata).Afterthat,theP1B‐UMI‐RNA‐TSOprimer,containing3guaninesatthe3’end, annealedtothe3incorporatedcytosinesatthecDNA,andmadeatemplateswitchingusedtointroduce adistinctbarcode(composedof6‐bprandomsequence(UMI))toeverysynthetizedcDNAmolecule.In thisway,the6‐bpUMIiscapableofdistinguishingupto4096differentmRNAmolecules,whichissufficient forasinglemammaliancell(Islametal.,2014).Theplatewascentrifugedfor1minuteatmaximumspeed (>2000g)afteranydispenseandincubationsteptoguaranteeaccuratemixingofreagents.MicroSealA film(BioRad)wasemployedtosealthearraywhennecessary(Figure14:steps2,3). 19.3.Steps4‐5.IndexedPCRandextraction 25indexprimers(DI‐P1A‐idx[1‐25]‐P1B)(Table5)wereaddedfollowingthereversetranscription tocarryoutthefirstPCR.Inthisway,eachwellfromevery10x10quadrantofthechipgotauniquebarcode (BC1).Theseindexeswerelateremployedtoidentifytowhichcell/welleachtranscriptbelongs.Each primerwasdispensedin100nlintoadifferentwell(from1to25wellswithineach96quadrant),using washeswith2%bleachbetweeneachsetofprimers,andreachingafinalprimerconcentrationof200nM inthePCRreaction.Finally,570nlofPCRmixwereaddedtoeachsample(finalconcentrationof1XKAPA HiFiReadyMix,supplementedwith0.2mMdNTPand100nMDI‐PCR‐P1A),andPCRreactionwasrunwith thefollowingprogram:95°C3min;5cycles:98°C30sec,67°C1min,72°C6min;15cycles:98°C30sec, 68°C30sec,72°C6min;72°C5min;10°Chold.Eachpairofprimers(thesamecommonDI‐PCR‐P1Aprimer witheachoneoftheDI‐P1A‐idx[1‐25]‐P1Bprimers)amplifiedallUMI‐taggedcDNAmoleculesfromasingle cell/well.Afterthat,acleaningstepwithAMPureXPbeadsat0.7:1ratiowasperformedtopurifycDNA fromprimers,andcDNAwaselutedinEBbuffer(Qiagen).Finally,anassemblybetweentheWaferGen plateandaclean96‐wellplatewasmountedandcentrifugedduring5minatmaximumspeed(>3000g)to transfertheWaferGenplatecontentintothe96‐wellplate.Attheend,eachwellfromthe96‐wellplate containedpooledindexedamplifiedcDNAfrom25wellsbelongingtoanentire10x10quadrantfromthe originalchip.ThisplatewasanalyzedbyBioanalyzer(HighSensitivityDNAAssay,Agilent)todetermine cDNAquantityandquality,andfinallysealedandkeptat‐20°C(Figure14:steps4,5). _____________________________________________________________________________________ Figure13.GraphicrepresentationofwellpositionpatternsforcellsortingintoaWaferGenplatform. PictureofamatrixsimulatingtheWaferGenplatecomposedby9600microwells.Itisformedby96quadrants (12columnquadrantsx8rowquadrants),andeachquadrantisinturnconstitutedby100wells(10columnsx 10rows).Only25positionscanbeusedduetosequencingbarcodelimitation,leadingtoatotalof2392useable wells(orangepositions).NotethedifferentsortingtemplateinH12quadrant(17suitablewells).Purplewells representpositionswheresamplechangingtookplaceatFACS;therightwellisthelastpositionsortedfrom theprevioussampleandtheleftwellcorrespondstothefirstpositionsortedfromthenewsample.Theeleven highlightedblackpositions(row4,inrowCquadrant;row2,inrowFquadrant;andH12‐W01well)indicate wellslabelledas“Undetermined”duringReadData.Rscriptandlaterremovedfromdownstreamanalysisto avoidpotentialcontamination.Sortingrowdirectionisspecifiedontheright.
MaterialsandMethods 60 19.4.Step6.Tagmentationand5′fragmentsisolation OncecDNAwasamplified,itwassimultaneouslyfragmentedandbarcodedby“tagmentation”. Thetagmentationstepwasperformedusing96transposomes,eachonecontainingadistinctindex barcode(BC2)tolabelcDNAfromeachsubarray(96intotal).Toassembletransposomestocks,barcoded adapters(6.25μM)(STRT‐Tn5‐Idx[1–96]+STRT‐TN5‐U)(Table5),Tn5transposase(6.25μM)andglycerol (40%)weremixedduring1hat37°C.ForTn5reactions,3µloftheprevioustransposomemixwasadded to2µlofamplifiedcDNA,inatotalvolumeof20µlof1XCutSmartBuffer(NewEnglandBiolabs),andthen incubatedduring20minat55°C.Duringthetagmentationstep,theTn5transposase(assembledwiththe barcodedadapters)randomlycuttheamplifieddouble‐strandedcDNAandpastedtheannealedbarcoded adaptersattheplacewherethecDNAmoleculewascleaved.Twopossibleproductsaregenerated,but onlytheonecontainingtheUMIandBC1atthe5’end,andtheBC2atthe3’endwillbeamplifiedduring thesecondPCRstep. Aftertagmentation,100µlofDynabeadsMyOneStreptavidinC1(ratio1:20inBBbuffer:10mM TrisHClpH7.5,5mMEDTA,250mMNaCl,0.5%SDS)wereaddedtothetagmentationreaction(ratio1:1) andincubatedatroomtemperatureduring15min.Then,allsampleswerepooledinasingletube,washed twiceinTNT(20mMTrisHClpH7.5,50mMNaCl,0.02%Tween‐20)andfinallyresuspendedin50µlof TNT.ExoSAPIT(Affymetrix)wasusedtocleantheremainingadaptersusinganincubationstepof15min at37°C,andthenwashedtwiceinTNTandonceinEBbuffer.Thesingle‐strandedlibrarywaselutedin50 µlofnuclease‐freewaterbyincubationduring10minat70°C,afterwhatthesupernatantwascollected inacleantube.Next,1.5XvolumesofAMPureXPbeads(Beckman)wereemployedtobindtothelibrary, thesupernatantdiscardedandbeadsresuspendedinthesecondPCRmix(1XKAPAHiFiReadyMix supplementedwith200mMP1_2nd_PCR,200nMP2_2nd_PCR)(Table5)toperformthefollowingcycle: 95°C2min;8cycles:98°C30sec,65°C10sec,72°C20sec;72°C5min;10°Chold.Threeconsecutive stepstocleanandbindthecDNAwithAMPureXPbeads(0.7X,0.5Xand1Xvolumes,respectively)were carriedouttofinallyelutethelibraryinEBbuffer(Figure14:step6). 19.5.Step7.Illuminasequencing Beforeproceedingwithsequencing,libraryconcentrationandqualitywasassessedusingthe Bioanalyzer(HighSensitivityDNAAssay,Agilent).IlluminaHiSeq2000or2500equipmentwasusedfor high‐throughputsequencingusingaSingle‐End50cyclekit,andDI‐Read1‐Seq(Read1),STRT‐Tn5‐U(Index 1) andDI_idxP1A‐Seq(Index2)primers(Table5).45bp‐readswereproducedbySTRT‐seq‐2i,whichwere composedbya6‐bpUMIfollowedby3guanidinesandfinallythe5’transcript.Thetwoindexreads contained8and5bp,correspondingtoIndex1(subarraybarcodeforeach96pooledwellfroma10x10 quadrant)andIndex2(wellbarcode(1‐25)withineach10x10quadrant)(Figure14:step7).Sampleswere
MaterialsandMethods 61 sequencedonceinanonlyrun.OnelaneoftheIlluminasequencingflowcellwasusedperchip;intotal,4 laneswereused. ProcessPrimerSequence5’3’ Lysis/reverse transcription STRT‐P1‐T315′Bio‐AATGATACGGCGACCACCGATCG‐ TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTT P1B‐UMI‐RNA‐TSO5′Bio‐rCrTrArCrArCrGrArCrGrCrTrCrTrTrCrCrGrArTrCrT‐ rNrNrNrNrNrN‐rGrGrG PCR1 DI‐PCR‐P1A5′Bio‐AATGATACGGCGACCACCGA DI‐P1A‐idx[1–25]‐P1B5′Bio‐AATGATACGGCGACCACCGAGATCTACAC‐XXXXX‐ CTACACGACGCTCTTCCGATC TagmentationSTRT‐Tn5‐Idx[1–96]CAAGCAGAAGACGGCATACGA‐YYYYYYYY‐ GCGTCAGATGTGTATAAGAGACAG STRT‐TN5‐U5′PHO‐CTGTCTCTTATACACATCTGACGC PCR2P1_2nd_PCRAATGATACGGCGACCACCGAGATC P2_2nd_PCRCAAGCAGAAGACGGCATACGAGAT Sequencing DI‐Read1‐SeqATGATACGGCGACCACCGAGATCTACAC‐NNNNNN‐ CTACACGACGCTCTTCCGATCT STRT‐Tn5‐U5′PHO‐CTGTCTCTTATACACATCTGACGC DI_idxP1A‐SeqAATGATACGGCGACCACCGAGATCTACAC 19.6.Preliminaryfilteringofrawreads First,IlluminaHiSeqsoftwareperformedanintrinsiccontrolannotatingreadsasinvalid.Second, 3’baseshavingabadqualityscorewerediscarded.Third,allthesixUMIbasesinthe5’endhadtosatisfy ascorethatsavedthereadaftercuttingawaytheUMI,elsethereadwasremoved.Fourth,acontrol checkingwhetherthethreefollowingbaseswereguanidines(GGG)wascarriedouttokeeptheread. ThesethreeGwereremoved,besidesanyadditionalGuptonine,aspotentialtemplate‐switchderived. Fifth,thereadwasdiscardedaccordingtothefollowingassumptions:(1)iftheremainingsequenceofthe readendedinapolyAsequenceleaving<25bases,(2)ifitcontainedlessthansixnon‐Abasesor(3)ifit containedadinucleotiderepeatwithlessthansixotherbasesateitherend.Finally,readswereorganized bywell/cell,givenbythetwoindexbarcodes.Afterqualitycontrol,Bowtiealignerwasusedforalignment, onlypermittingupto3base‐mismatchesandupto24alternativemappings.Whenareadfoundno alignment,itwasrealignedtoanartificialchromosomethatincludedeverysplicejunctionexistentforan exon(UCSC,GenomeBrowser).Thecoordinatesofalignedsplicejunctionsweretranslatedbacktothe correspondinggenomicpositions..Atthispoint,weestimatedtheread/sequencingdepthbyaverage mRNAtotalmappedreadspercell. Table5.CustomprimersforSTRT‐seq‐2iprotocol. PrimersusedforfullSTRT‐seq‐2imethodaredisplayedinthistableasdescribedinHochgerneretal.2017. Numbersbetween[]representatotalof25primersneededforeachwellidentificationina10x10quadrant, andatotalof96primersrequiredtoidentifythe96subarrays.UMI,BC1andBC2sequencesaredenotedin green,orangeandred,respectively,accordingtoFigure14.Bio,biotin;PHO,phosphorylationof5’end.
MaterialsandMethods 62 Figure14.DetailedSTRT‐seq‐2imethodusing FACStosortcellsintoaWaferGenplate. SchematicSTRT‐seq‐2imethodmodifiedfromHochgerneretal.2017,includingpicturesforvisualdescriptionof thefull‐stepsingle‐cellRNAsequencingprocess(describedinthe19.Single‐cellRNAsequencingprotocolsection).
MaterialsandMethods 63 UMIcountswerecalculatedbythenumberofreadscontainingeachUMIsequenceforeach genomicpositionandstrandcombination,separatelyforeachwell.Beforethat,theexonsofeachlocus withalltranscriptvariantsweremergedtoacombinedmodelrepresentingtotalexpressionofthelocus. Ifanymultiread(areadthatmapstoseveraldistinctlocations)mappedtosomerepeatoutsideexons,it wasannotatedasoneoftheserepeatsbyrandomanditwasnotconsideredasanexon‐derivedtranscript. However,whetherthemultireadmappedonlytosomeexons,itwasassignedtothemostsimilarexonto thetranscriptmodel5’end.Ifitdidnotmaptoanyexon,arandomassignationtoanymappingwasmade. Afterthoseassignations,thenumberofdistinctUMIswascountedtogeneratethetotalnumberof moleculesateachmappingposition;nevertheless,ifonlyasinglereadwasproducedfromanyUMI,this UMIwasdiscardedwiththepurposeofminimizingtheprobabilityoffindingfalsemoleculesderivedfrom PCRorsequencingmistakes.RawUMIcountswerefinallycorrectedforcollisionprobability. 20. Single‐cellRNAsequencingdataanalysis Single‐cellRNAsequencinggenerateshugedatasetsthatneedtobeanalyzedwithbioinformatics. Packagesforsingle‐cellsequencinganalysisarepredominantlydevelopedinRprogramminglanguage, thusweanalyzedoursequencingdatausingRfreesoftware(https://www.r‐project.org/).WeinstalledR forWindows(version3.4.4)(https://cran.r‐project.org/bin/windows/base/old/3.4.4/)andalsoRStudio (version1.1.442)(https://www.rstudio.com/products/rstudio/download/),aninterfacethatintegratesa setoftoolsdesignedtohelpwithRprogramming.ItcontainsaneditortowriteRscriptbeforeexecution, aconsole,andtoolstomanagedifferentparameterssuchashistory,environment(wherevariablesare), installedpackages,files,plotsandhelp,withinothers.WepreferredtoworkwithRStudiobecauseit generatesaneasyandintuitiveplatformthathelpsbeginnerstoworkwithR,insteadofusingaconsole. ThemainRpackageutilizedforoursingle‐cellsequencingdataanalysiswastheSeuratpackage (version2.3.4),developedbySatijalab(‘https://github.com/satijalab/seurat’),inadditiontowell‐known typicalpackagesalreadycontainedinRsuchasbaseandgraphics,withinothers.Anyway,allused packagesthatwerenotinstalledbydefaultinRwereproperlyinstalledandcalledwhennecessaryatthe beginningofanyRscript. Itusuallyexistssomegenesthatcanbeidentifiedindifferentlocations(loci)atthegenome;the providedgenelistcontainedbydefaultalltheseknownlocations,whichweredesignatedby“_locN”after thegenename.Inouranalysis,fewofthedetectedclustermarkersanddifferentiallyexpressedgenes (DE)correspondedtogeneswithspecificlocations,sothegenenamecontainedalocussuffixthat appearedinthegeneratedfigures(e.g.volcanoplots,violinplotsorDEtables).However,compiled bibliographicinformationregardingthosegeneswasdonewithouttakingintoaccounttheexactgene location,usingjustthesinglegenename,andthisishowitappearswrittendowninS11andS15.
MaterialsandMethods 64 ThescRNAsequencinganalysiswasentirelydevelopedduringthisthesisandinvolvedanewand laboriousstudyforusthatconsumedanimportanttimeofthiswork.Forthisreason,thebioinformatics analysisofthescRNA‐seqstudyhasbeenincludedintheResultssectioninadditiontothematerialsand methodologyforthedissociationandgenomicprotocolsdescribedhere. 21. Geneontology(GO)analysis ClueGOv2.5.1pluginofCytoescapev3.6.1softwarewasusedforGeneOntology(GO)analysis (Bindeaetal.,2009),usingonlythegenessignificantly(p‐value<0.05)detectedasup‐ordown‐regulated ineachgroupforbothclusteredortotalPVcellstudies(Supp.Data:Rscriptsandfiles>Seuratintegrated analysis>Up‐anddown‐regulatedgenesforGOanalysis.R;Supp.Data:Up‐anddown‐regulatedgenesfor GO(CSV)).DatabasesemployedforGOanalysisinvolvedKyotoEncyclopediaofGenesandGenomes (KEGG)andGeneOntologyConsortium(GO),whichincludepathways,functions,processesandreactions (namedterms)alreadyknown.Wesetuptheanalysistofindpathwayswithap‐value≤0.05(significant) towhichatleast3geneswerefoundrelated.Kappascores(≥0.04)wereusedforfunctionalgrouping.We selectedtheGOtermfusionoptiontokeeponlythemostinterestingtermforGOparent‐childrelationto simplifyresults.Moreover,termrepetitions(thesametermcanbeobtainedfrommorethanone database)werediscardedtoavoidforredundancyandonlythefirstresultforatermwasmaintained. StatisticalsignificancewasmeasuredbyEnrichment/Depletion(two‐sidedhypergeometrictest)with Bonferronistepdownp‐valuecorrection,thedefaultoptioninClueGO.Finally,ineveryGOanalysis,the groupleadingterm(themostsignificanttermwithinagroup)wasselectedasthemostrepresentative termforeachgroup.ThoseleadingtermswerethenorderedbasedontheirGOgroupp‐value(fromthe mostsignificanttothelesssignificant).Noexactgenelocation(loci)wastakenintoaccountneitherforGO analysis. 22. Statistics Statisticalsignificancewasmeasuredbasedontheresultofthenormalitytestappliedtoeachpair ofcomparedsamples.ThenormalitytestusedforthemajorityofanalyseswasD’Agostino‐Pearson(the recommendedoptioninGraphPadPrism6).Whenitwasnotpossibletousethistestduetothesample size(n<8),theShapiro‐Wilktest(forthebodyweightcurveanalysis)orAnderson‐Darlingtest(for immunoblottingandqRT‐PCRanalyses)wereappliedtotestfornormalitydistribution.Dependingonthe resultofthenormalitytest,statisticalsignificancewasmeasuredusingStudent’st‐testorMann‐Whitney Utestineachpairofcomparedsamples.
RESULTS
Results 67 1. CorticalsynapticdysfunctionsinDnajc5conventionalKOmice PreviousworkofourgroupshowedthatthecompletedeletionofCSP/DNAJC5(Dnajc5 conventionalKOmouse)producedanactivity‐dependentdegenerationofGABAergicsynapses,mainlyin PVinterneurons(García‐Junco‐Clementeetal.,2010).Thisworkwasperformedinhippocampus,abrain structurecomposedbywell‐characterizedandeasilydistinguishablelayersthatpresentdefinedneuronal connectivity.OneplausiblehypothesisisthatduringtheneurodegenerationofGABAergicterminalscould existdifferentialgeneexpressioninPVcellslackingCSP/DNAJC5comparedtocontrolmice,sousingthe high‐throughputsingle‐cellRNAsequencingtechniquewecoulddetectcandidategenesfor neurodegeneration.WhenwefirststartedwithhippocampalpreparationstosortPVcells,wefoundthe limitationofcollectingverylowyieldofcells,whichmadesortingtimeverylong,compromisingcell viabilityandmakingsubsequentexperimentsverydifficulttooptimize,suchasgDNAandRNAextraction. Furthermore,oncewerealizedthatsortingefficiencywouldbeextremelyharmedandcellviabilityhighly compromised,‐evenmore,itwouldhaverepercussionsinanimportantobjectiveofthisthesis,whichwas thesingle‐cellRNAsequencing‐,wetookthedecisiontoswitchtothecerebralcortex,withimproved sortingresults.Inperspective,amajoradvantageofthecortexisthepossibilityofanalyzingcircuit dynamicsofGABAergicneuronsusinginvivotwo‐photonmicroscopyinawakebehavingmice.Inaddition, thealterationsthatwemightfindcouldhelp,atleastpartially,tounderstandthemotorphenotypefound inthesemutantmice(seelater).Withtheideaofperformingamoredetailedstudycenteredinaspecific corticalregion,wefocusedonthemotorcorticallayerII/III,soalltheimmunohistochemistryand immunofluorescencequantifications,aswellastheelectrophysiologicalrecordings,werecarriedoutin thiscorticalregion.LayerII/IIIwaschosenduetothehighestpercentageofPVinterneurons(togetherwith layerV)inthisparticulararea(Tamamakietal.,2003). SeveralpublicationsregardingCSP/DNAJC5havealsofocusedoncorticalneurons,evenonthe wholebrain(Sharmaetal.,2011,2012),supportingourdecision.Althoughaneurodegenerativecortical dysfunctionhasbeenalreadyreportedinthesepublications,weaimedtoconfirmitinourhands,looking forasimilardysfunctionalsynapticphenotypecomparedtothehippocampusthatwouldsupporttofocus oureffortsincorticalPVneuronsduringthiswork. Toassessthisquestion,wefirstanalyzedbyimmunohistochemistrydifferentpresynapticproteins involvedinsynapticvesiclefusionduringNTreleaseatthepresynapticterminals,usingP30Dnajc5WT andKOmice(n=2micepergenotype).Withthistechnique,weaimedtohaveaglobalviewof CSPα/DNAJC5,PV,SNAP25andSyt2expressioninthemotorcorticalregionusingbrainslices.Weobserved ageneralreductioninSNAP25andSyt2puncta,aswellasareductioninPVaxonallabellingofDnajc5KO mice,butnotinoverallPVsomata.Moreover,KOmicecompletelylackedCSPα,asexpected(Figure15A) (Fernández‐Chacónetal.,2004;Tobabenetal.,2001).Toconfirmtheseobservations,weperformed
Results 74 Ourlaboratorydisposesofanantibody(R807)thatrecognizesthefull‐lengthCSPα/DNAJC5 protein(Tobabenetal.,2001),enablingaproperdetectionofapotentiallyCSPα/DNAJC5truncated protein.Nonetheless,althoughtheR807antibodydidrecognizethewildtypeproteininliversfromDnajc5 WTandDnajc5flox/floxmice(34KDa),therewerenodifferencesinexpressionpatternsbetween Dnajc5flox/floxandUBC‐cre‐ERT2:Dnajc5flox/floxsamplesatthelower‐molecular‐weightsectionoftheblot, neitherwhenweisolatedandhighlyexposedthatregion(from10to19KDaapproximately)(n=3miceper genotype)(Figure18B). OnceweconfirmedtheabsenceofaberrantversionsofCSPαproteinswithlowmolecularweight inliversamples,atissuethatpreviouslyshowedastrongdeletionoffloxedDnajc5mRNA,wecarriedout thesametestusingtheR807antibodyincorticallysatesfromPVcre:Ai27D:Dnajc5flox/+and PVcre:Ai27D:Dnajc5flox/‐mice(n=3micepergenotype).Abandwithamolecularweightcloseto30KDa, correspondingtothewildtypeprotein,wasfoundinallsamplesexceptfortheDnajc5KOnegativecontrol. WealsoobservedasmallreductioninCSPα/DNAJC5expressioninPVcre:Ai27D:Dnajc5flox/‐ samples,a surprisingresultbecausetheDnajc5floxedallelewasonlyremovedfromPVcells.Nevertheless,wedid notfindsignsoftruncatedproteins,evenuponlongtimeexposures,atthemolecularrangebetween10‐ 19KDa(Figure19). 3. tdTomatoexpressionisspecificfortheparvalbumin‐positiveinterneuron population IntheprevioussectionweconfirmedacorrectCre‐recombinaseactivityatDnajc5locus,sothe nextstepwastovalidatethespecificityofthetdTomato+labellingonPVinterneurons.Forthispurpose, wecarriedoutimmunofluorescenceexperimentstoquantifyPV‐tdTomatocolocalizationin PVcre:Ai27D:Dnajc5flox/+andPVcre:Ai27D:Dnajc5flox/‐miceat2monthsold(n=90fieldspergenotypefroma totalof3micepergenotype).PV+somataandtdTomato+perisomallabellingwereconsideredfor _____________________________________________________________________________________ Figure18.AnalysisofCSPα/DNAJC5protein inliverlysatesfromUBC‐cre‐ERT2:Dnajc5flox/floxmice. ImmunoblotswithequivalentamountsofproteinfromcorticalbrainorlivertissueidentifyingCSPα/DNAJC5with twodifferentantibodies.Dnajc5WTandKOareusedaspositiveandnegativecontrols,respectively.Samplesare obtainedfromDnajc5WTmice(WT),Dnajc5conventionalKOmice(KO)andDnajc5flox/floxmice(F/F)treatedwith tamoxifen,without(‐)orwith(+)atransgenedrivingtheubiquitousexpressionofCre‐recombinase(UBC‐cre‐ ERT2).A.BlotanalyzedwithanantibodyraisedagainsttheC‐terminusofCSPα;itshowsthattheliverfromUBC‐ cre‐ERT2:Dnajc5flox/floxmicetreatedwithtamoxifenlacksCSPα.B.Blotanalyzedwithanantibodyraisedagainst aGST‐fusionproteinofCSPα(R807);itshowsthattheliverfromUBC‐cre‐ERT2:Dnajc5flox/floxmicetreatedwith tamoxifenlacksCSPα(arrow).Therightpanelrepresentsahigherexposureimageofthesameblotshowingno evidencefortheexistenceofatruncatedlowermolecularweightversionofCSPα uponCre‐recombinase mediatedgenomicdeletion.NotethattheliverCSPα/DNAJC5presentsahighermolecularweightcomparedto thecorticalprotein.β‐actinisusedasloadingcontrol.
Results 75
Results 76 quantification.Weobtainedaround90%ofcolocalizationforbothPVcre:Ai27D:Dnajc5flox/+and PVcre:Ai27D:Dnajc5flox/‐genotypes,whilepercentagesofnon‐colocalizationwere7and10%forPV+neurons and4and1%fortdTomato+cellsincontrolsandmutants,respectively(Table7;Figure20A).No differenceswerefoundbetweenPVcre:Ai27D:Dnajc5flox/+andPVcre:Ai27D:Dnajc5flox/‐miceintotalPV‐ tdTomatocolocalization(p=0.3369,ns,Mann‐WhitneyUtest),asexpectedfromthesamepromotordrivenCre‐activity(PVcre). AgeGenotypePV‐tdTomato colocalization(%) PV+cellswithno colocalization(%) tdTomato+cellswith nocolocalization(%) 2monthsDnajc5flox/+88.73±0.97 7.41±0.524.52±0.94 Dnajc5flox/‐88.78±0.6410.42±0.660.95±0.19 Figure19.AnalysisofCSPα/DNAJC5proteinincorticallysatesfrombothPVcre:Ai27D:Dnajc5flox/+ and PVcre:Ai27D:Dnajc5flox/‐mice. Immunoblotcontainingequivalentamountsofproteinfromcorticalbraintissueanalyzedwithanantibodythat recognizestheGST‐fusionproteinofCSPα(R807).Dnajc5WTandKOareusedaspositiveandnegativecontrols, respectively.SamplesareobtainedfromDnajc5WTmice(WT),Dnajc5conventionalKOmice(KO), PVcre:Ai27D:Dnajc5flox/+mice(F/+)andPVcre:Ai27D:Dnajc5flox/‐mice(F/‐).ThisblotshowsasmallreductioninCSPα expression(arrow)incorticalbraintissueofPVcre:Ai27D:Dnajc5flox/‐mice.Rightpaneldisplaysahigherexposure imageofthesameblotshowingnoevidencefortheexistenceofatruncatedlowermolecularweightversionof CSPαuponCre‐recombinasemediatedgenomicdeletion.β‐actinisusedasloadingcontrol. Table7.SummaryofcolocalizationvaluesbetweenparvalbuminandtdTomatoexpressionintheexperimental mice. Meancolocalizationvalues(%)±SEMbetweenparvalbuminandtdTomatoinPVcre:Ai27D:Dnajc5flox/+and PVcre:Ai27D:Dnajc5flox/‐miceat2monthsold.ThetablealsoshowsthepercentageofPV+cellsandtdTomato+ cellsthatdidnotcolocalize.Valuesareroundedusingtwodecimals.ReferredtoFig.20A.
Results 77 Furthermore,wealsoinvestigatedthespecificityofcolocalizationinvivobysortingtdTomato+ cells,namelywhetherwewereobtaininganenrichedpopulationofPVneuronswhensortingtdTomato+ cellscomparedtotdTomato‐cells.Weperformedareal‐timequantitativePCRtostudymRNAlevelsof parvalbuminbothintdTomato‐positiveandnegativesortedfractionsfromcontrolandmutantanimals (n=3foreachgenotype).Resultswerenormalizedversusnegativefractionsfromeachgenotypeand representedas2(‐ΔΔCt),normalizedmRNAlevels.Inbothgenotypes,parvalbuminmRNAwasstrongly increasedintdTomato+fractionscomparedtotdTomato‐ fractions,althoughanimportantvariability withinsamplesofthesamegenotypewasfound,generatinglargestandarderrorofthemean(SEM)values andnonsignificantresults(p=0.29,ns,Student’st‐testforcontrols;p=0.30,ns,Student’st‐testformutants) (Table8).DifferenceswerenotsignificanteitherbetweentdTomato‐fractions(p>0.99,ns,Student’sttest)orbetweentdTomato+fractions(p=0.29,ns,Student’st‐test)ofcontrolsversusmutants.Thesedata provedthatwewerenotlosingmanyPVcellsintdTomato‐negativefractions,butinsteadwewere obtainingahighlyenrichedpopulationofPVneuronswhensortingtdTomato+cells,confirmingaspecific tdTomato+labellinginPVcells(Figure20B). AgeGenotypetdTomato‐sortedcellstdTomato+sortedcells 2monthsDnajc5flox/+1.00±0.32 991440±810214 Dnajc5flox/‐1.00±0.841971±1638 4.ThespecificdeletionofDnajc5inPVinterneuronsdoesnotcauseincreased mortalitybutprogressivebodyweightalterationsinPVcre:Ai27D:Dnajc5flox/‐mice PVcre:Ai27D:Dnajc5flox/+andPVcre:Ai27D:Dnajc5flox/‐micedidnotshowneitherearlylethalitynora distinguishablephenotypeimmediatelyafterbirth.WedevelopedacontinuoussurvivalcurvefromP0 until8monthspostnatally.Nosignificantdifferencesinmortalityweredetectedbetweenmutantand controlmice(96.552%survivalat8monthsforPVcre:Ai27D:Dnajc5flox/+mice;93.103%survivalat8months forPVcre:Ai27D:Dnajc5flox/‐mice;p=0.5614,ns,Log‐rankMandel‐Coxtest;n=29pergenotype)(Figure21A). Furthermore,weobservedthatmutantanimalscouldsurviveevenforlongerperiods(atleastuntil15 months)withoutshowingdramaticlethalitycomparedtocontrols;however,itwasfrequentlynecessary toprovidefoodandwateronthecagefloortofacilitatefeedingaccessduetohindlimblocomotor deterioration(seenextsection). Table8.SummaryofexpressionlevelsforPVmRNAinsortedcellfractionsobtainedfromtheexperimental mice. Normalizedmeanvalues±SEMofmRNAexpressionintdTomato‐negative(tdTomato‐ cells)andtdTomatopositive(tdTomato+cells)sortedcellsinPVcre:Ai27D:Dnajc5flox/+andPVcre:Ai27D:Dnajc5flox/‐miceat2monthsold. Valuesareroundedusingtwodecimals.ReferredtoFig.20B.
Results 78 Duringgrowth,certainphenotypiccharacteristicsbecamenotoriousinmutantmice.AroundP30, werealizedthatPVcre:Ai27D:Dnajc5flox/‐ miceapparentlyhadreducedbodysizecomparedtotheir littermatecontrols,sowedecidedtocarryoutabodyweightcurvefromP30(1month)toP240(8months). Maleandfemalebodyweightsarenaturallydifferent,sothestudywasmadeseparatelyforeachsex(n=5 Figure20.Motorcorticalneuronsshowhigh colocalizationo f PVandtdTomatolabelling. A.Left,representativeconfocalimagesshowingmotorcorticalneuronsofPVcre:Ai27D:Dnajc5flox/+and PVcre:Ai27D:Dnajc5flox/‐ micelabelledwithanti‐parvalbumin(green)andanti‐mCherry(red)antibodies.White squaresidentifymagnifiedimagesofcorticallayerII/IIIneurons.ColocalizationbetweenPVandtdTomatois pointedoutusingwhitearrowsinthemergepanel.ThefusionoftdTomatowithChR2generatestheexpected perisomalandaxonallabellinglocalizedtothemembrane.Right,Venndiagramsdisplayingthepercentageof colocalizationbetweenPVandtdTomatoinmotorcorticallayerII/IIIforbothgenotypes,andalsothepercentages ofPV+cellsandtdTomato+cellsthatdonotcolocalize(n≈90fieldspergenotype,3micepergenotype).B.Levels ofparvalbuminmRNA(2(‐∆∆Ct))obtainedfromreal‐timeqPCRofFAC‐sortedtdTomato+andtdTomato‐cellsfrom PVcre:Ai27D:Dnajc5flox/+andPVcre:Ai27D:Dnajc5flox/‐ mice.Asexpected,thereisapredominanceofPVcellsin positivefractionsofbothgenotypes.ErrorbarsindicateSEMandstatisticalsignificanceisdeterminedbyMann‐ WhitneyUtestforthecolocalizationanalysisandStudent’st‐testforqRT‐PCR.
Results 79 formaleandfemalecontrolmice,n=5formalemutantmiceandn=4forfemalemutantmice).Datashowed thatmalePVcre:Ai27D:Dnajc5flox/‐miceexperiencedbodyweightdecreasecomparedtomalecontrolmice, whichbecamestatisticallysignificantfromP37toP240(e.g.P30:p=0.6994,ns,Student’st‐test;P128: ***p=0.001,Student’st‐test;P240:*p=0.0159,Mann‐WhitneyUtest)(Table9;Figure21B,D).Female PVcre:Ai27D:Dnajc5flox/‐micealsoshowedadecreaseinbodyweightsimilartomutantmales.Comparedto femalecontrolmice,theyexhibitedstatisticallysignificancefromP51toP240(e.g.P30:p=0.3576,ns, Student’st‐test;P128:****p=0.0001,Student’st‐test;P240:**p=0.0017,Student’st‐test)(Table9;Figure 21C,D). Males(g)Females(g) AgeDnajc5flox/+Dnajc5flox/‐Dnajc5flox/+Dnajc5flox/‐ P30 15.66±1.1515.15±0.5914.97±0.4613.89±1.01 P3719.38±0.6916.39±0.1716.67±0.5115.43±0.65 P4421.92±0.4018.67±0.3717.85±0.6216.26±0.49 P5124.19±0.4320.68±0.5219.72±0.6917.00±0.48 P5825.18±0.5422.32±0.3620.74±0.6518.24±0.62 P6526.45±0.8022.36±0.6520.94±0.68 17.27±0.36 P7226.83±0.7522.37±0.4221.83±0.70 17.98±0.50 P7928.48±0.8623.48±0.7622.88±0.5818.78±0.45 P10032.26±1.5624.51±1.0924.10±0.6819.03±0.32 P12835.46±1.6724.78±1.2825.69±0.54 20.24±0.10 P15637.89±2.2523.70±0.7427.88±1.5319.80±0.45 P18438.21±2.4223.09±0.9128.05±1.2019.19±0.20 P21239.41±2.6623.48±1.7228.56±1.6219.02±0.33 P24041.93±2.4024.88±0.6929.05±1.6519.55±0.47 5. NeurologicalphenotypeupondeletionofDnajc5inPVinterneurons Afterbirth,controlandmutantmicewereindistinguishable.However,longtermobservations duringgrowthrevealedacharacteristicneurologicalphenotypeinmutantmice.AroundP30,the phenotypewasmildlyobservedinPVcre:Ai27D:Dnajc5flox/‐,butitwasatP60whenitbecamenoticeableand easilydetected,soweestablished2months(adultmice)asasuitableagetoinvestigatephenotypes.At thistimepoint,mutantmicepresentedthefollowingphenotypicalcharacteristics:(1)ahighlocomotor activitylevel,includingdisorganizedmovementsthroughoutthecagecomposedoftime‐variable patternedrunningepisodes;(2)ahindlimbdeteriorationthatgeneratedacharacteristicataxicgaitand hyperreflexia,withtendencytostretchhindlimbsduringtailsuspension;(3)aslightbackhump;(4)lumbar lordosis,whichwasrevealedbytheirabilitytocontorteasily;and(5)abnormalheadmovements,bothat restandduringlocomotion,whichincludedspontaneousheadtossingmovementsandupwardgazesthat sometimesweresustainedduringseconds(calledstargazing),similartothemovementsreportedfor Table9.Summaryofbodyweightvaluesintheexperimentalmice(malesandfemales). Meanvalues±SEMofbodyweight(g)inPVcre:Ai27D:Dnajc5flox/+andPVcre:Ai27D:Dnajc5flox/‐maleandfemalemice fromP30toP240(1monthto8monthsold).Valuesareroundedusingtwodecimals.ReferredtoFig.21B,C,D.
Results 80 Figure21.Survivalcurveandbodyweightscreening ofPVcre:Ai27D:Dnajc5flox/+andPVcre:Ai27D:Dnajc5flox/‐mice until8monthspostnatally. A. AnalysisoflatesurvivalshowstheabsenceofdifferencesinmortalitybetweenPVcre:Ai27D:Dnajc5flox/+and PVcre:Ai27D:Dnajc5flox/‐miceover8months.StatisticalsignificanceiscalculatedusingtheLog‐Rank(Mantel‐Cox) test(p=0.5614).B.BodyweightcurveformalemicefromP30toP240(8months)betweenPVcre:Ai27D:Dnajc5flox/+ andPVcre:Ai27D:Dnajc5flox/‐ genotypes.SignificantchangesareobservedfromP37.C.Bodyweightcurvefor femalemicefromP30toP240(8months)betweenPVcre:Ai27D:Dnajc5flox/+andPVcre:Ai27D:Dnajc5flox/‐genotypes. SignificantchangesareobservedfromP51.D.Plotrepresentingbodyweightcurvesformaleandfemalemice frombothupperplots.ErrorbarsindicateSEM.Student’st‐testandMann‐WhitneyUtestareusedforstatistical significancedetermination.*p=0.05‐0.01;**p=0.01‐0.001;***p=0.001‐0.0001.
Results 81 stargazerknock‐outmice(Letts,2005;Lettsetal.,1998;Noebelsetal.,1990;OstenandStern‐Bach,2006). Moreover,incertainlitters,itwasalsoobserved(6)arepeatedbackwardsomersaultingactivity.These observationsledustoconcludethatPVcre:Ai27D:Dnajc5flox/‐ micedevelopedaneurologicalphenotype, easilyobservedatP60(Supp.Data:Neurologicalphenotype(videos)). Asimpairedmovementswereobservedinmutants,wequantifiedlocomotoractivityinthese mice.Toevaluatethisbehavior,weperformedtheopenfieldtestseparatelyonmalesandfemales(n=11 controlmales,n=10mutantmales,n=9controlfemales,n=14mutantfemales).Thelocomotorbehaviorof bothmaleandfemalePVcre:Ai27D:Dnajc5flox/‐miceproducedanon‐stopmotiontrackaroundthewhole cagecomparedtocontrols(Figure22A).Thedataanalyzedfromvideorecordingsrevealedthat,compare tocontrols,maleandfemalePVcre:Ai27D:Dnajc5flox/‐ animalsdisplayedsignificantincreasesinvelocity (males:****p<0.0001,Mann‐WhitneyUtest;females:***p=0.0001,Student’st‐test),tracklength(males: ****p<0.0001,Mann‐WhitneyUtest;females:***p=0.0001,Student’st‐test)andactivity(males: ****p<0.0001,Mann‐WhitneyUtest;females:****p<0.0001,Student’st‐test).Onlymutantmales showedsignificantlyincreasednumberofambulations(spontaneousshort‐termacceleration)(males: ***p=0.0003,Mann‐WhitneyUtest;females:p=0.0558,ns,Student’st‐test).Ontheotherhand,normal headmovements,suchasheadbobs(males:****p<0.0001,Student’st‐test;females:****p<0.0001, Student’st‐test)andheadstretches(males:****p<0.0001,Student’st‐test;females:****p<0.0001, Student’st‐test),andalsoregulartailmoves(males:****p<0.0001,Student’st‐test;females:***p<0.0004, Student’st‐test)weresignificantlyreduced,bothinPVcre:Ai27D:Dnajc5flox/‐malesandfemalescompared tocontrolanimals.Nevertheless,wedidnotfindapparentdifferencesinanxietybetweengenotypes,a behaviorthatismildlyassessedintheopenfieldassaybywalldistance,neitherinmales(p=0.1905,ns, Mann‐WhitneyUtest)norfemales(p=0.279,ns,Student’st‐test)(Table10;Figure22B). Males Females AgeDnajc5flox/+Dnajc5flox/‐Dnajc5flox/+Dnajc5flox/‐ Velocity(cm/s)6.29±0.3717.16±1.935.93±0.4813.59±1.28 Tracklength(cm)11335±664.6030881±347410698±866.9024233±2293 Activity(%)48.95±1.7781.28±2.1945.96±1.9271.92±2.99 #Ambulations2.18±0.8741±14.2711.22±3.6490.36±30.98 Walldistance(cm)5.01±0.315.40±0.414.37±0.314.96±0.38 #Headbobs1827±73.99791.10±88.812226±97.401113±106 #Headstretches2652±94.02986.40±111.702669±140.401317±124.70 #Tailmoves559.90±36.19284.20±36.30776.40±68.19422.70±50.84 Table10.Summaryoftheopenfieldassaysperformedintheexperimentalmice(malesandfemales). Meanvalues±SEMforvelocity,tracklength,activity,ambulations,walldistance,headbobs,headstretchesand tailmovesinPVcre:Ai27D:Dnajc5flox/+andPVcre:Ai27D:Dnajc5flox/‐miceat2monthsold,separatingmalesand females.Valuesareroundedusingtwodecimals.ReferredtoFig.22B.
Results 82 Maleandfemaleresultswereverysimilar,sotheywerealsoanalyzedtogetherbymixingcontrol andmutantrawdata:incrementsinvelocity(****p<0.0001,Mann‐WhitneyUtest),tracklength (****p<0.0001,Mann‐WhitneyUtest),activity(****p<0.0001,Mann‐WhitneyUtest)andambulations (***p=0.0007,Mann‐WhitneyUtest)weredetectedinPVcre:Ai27D:Dnajc5flox/‐mice.Decrementsin headbobs(****p<0.0001,Student’st‐test),headstretches(****p<0.0001,Student’st‐test)andtailmoves (****p<0.0001,Student’st‐test)werefoundintotalPVcre:Ai27D:Dnajc5flox/‐ micecomparedtototal PVcre:Ai27D:Dnajc5flox/+animals(Table11;Figure22C).Nochangeswereobtainedinwalldistancebetween genotypesusingtotalmice(p=0.255,ns,Student’st‐test)(Figure22C). Totalmice AgeDnajc5flox/+Dnajc5flox/‐ Velocity(cm/s)6.13±0.2915.08±1.13 Tracklength(cm)11048±524.9027120±2035 Activity(%)47.60±1.3175.82±2.16 #Ambulations6.25±1.9569.79±19.38 Walldistance(cm)4.72±0.235.15±0.28 #Headbobs2007±73.88979.10±78.04 #Headstretches2660±79.371179±91.18 #Tailmoves657.40±43.25365±35.64 Althoughthemutantmicelivedlong(Figure21A),theywereprogressivelylosingmobilityalong time.At8months,fewPVcre:Ai27D:Dnajc5flox/‐animalsneededtobehumanlysacrificedduetohigh mobilityloss,especiallyinhindlimbs.Itwasalsodistinctivetoobserve8‐montholdmutantmicereclined ontheirstretchedhindlimbsand,astheywerenotabletofullycontroltheirownbodybalance,often fallingdown,evenduringlocomotion(Supp.Data:Neurologicalphenotype(videos)).Becauseweobserved thistime‐dependentprogressivephenotype,wedecidedtoestablishtwotemporalintervalsforfuture experiments(P60:2months,andP240:8months),inordertostudyboth,earlyandlatestagesofthe neurologicalphenotype. _____________________________________________________________________________________ Table11.Summaryoftheopenfieldassaysperformedintheexperimentalmice(alltogether). Meanvalues±SEMforvelocity,tracklength,activity,ambulations,walldistance,headbobs,headstretchesand tailmovesinPVcre:Ai27D:Dnajc5flox/+andPVcre:Ai27D:Dnajc5flox/‐miceat2monthsold,mixingmalesandfemales. Valuesareroundedusingtwodecimals.ReferredtoFig.22C. Figure22.NeurologicalphenotypeupondeletionofDnajc5 inPVcre:Ai27D:Dnajc5flox/‐miceat2monthsofage. A.RepresentativetrackexamplesfromPVcre:Ai27D:Dnajc5flox/+andPVcre:Ai27D:Dnajc5flox/‐miceduringa30 minute‐recordingopenfieldexperiment.Mutantmiceshowamotiontrackcharacterizedbynon‐stop movementsthroughoutthewholecage.B.PVcre:Ai27D:Dnajc5flox/+andPVcre:Ai27D:Dnajc5flox/‐miceareanalyzed inopenfieldexperimentsduring30minutes,indark,andquantificationofvelocity(cm/s),tracklength(cm), activity(%),ambulations(spontaneousshort‐termacceleration),walldistance(cm),andnumberofheadbobs, headstretchesandtailmovesareperformed.C.SameanalysisasBbutmixingmalesandfemalesduetotheirdata similarity.ErrorbarsindicateSEMandstatisticalsignificanceisobtainedbyStudent’st‐testandMann‐Whitney Utest.***p=0.001‐0.0001;****p<0.0001.
Results 83
Results 90 7.2.TheprogressivedegenerationofpresynapticterminalsinPVcre:Ai27D:Dnajc5flox/‐ miceisnotrelatedtochangesinthenumberofPVcellsorsomasize Aftertheglobalscreeninginexpressionpatternsofpresynapticproteins,wedirectedourattention specificallyintoPVneurons,tofigureoutifthedegenerativephenotypeisonlyrestrictedtosynapsesor if,inaddition,itisaccompaniedbysomaticalterations,i.e.,inthenumberorsizeofPVinterneurons.We usedtheimmunohistochemistryimageslabelledwithparvalbuminantibodiestoquantifyPVsomata numberinmotorcorticallayerII/III(n=27fieldsfromatotalof3micepergenotypeandage). Unexpectedly,wefoundanincreasednumberofPVcellsinPVcre:Ai27D:Dnajc5flox/‐miceat2months comparedtocontrols(*p=0.0216,Student’st‐test),butmostimportantly,nochangeswereobservedat8 monthsbetweengenotypes(p=0.5583,ns,Mann‐WhitneyUtest)(Table15;Figure26A).Thatsuggested thatneurodegenerationwasaffectingjustthepresynapticterminals,notreachingthePVsomata.Next,to testifthereweremorphologicalalterationsinvolvingPVsomatasize,theareaofPVsomaswasmeasured usingconfocalimagesstainedwithanti‐parvalbuminantibodies(n=104PVsomatafrom19images collectedfrom3controlmiceandn=181PVsomatafrom19imagescollectedfrom3mutantmiceat2 months;n=152PVsomatafrom21imagescollectedfrom3controlmiceandn=142PVsomatafrom21 imagescollectedfrom3mutantmiceat8months).WefoundnochangesinPVsomatasizebetween PVcre:Ai27D:Dnajc5flox/‐andPVcre:Ai27D:Dnajc5flox/+mice,neitherat2months(p=0.6344,ns,Mann‐Whitney Utest)norat8months(p=0.5472,ns,Mann‐WhitneyUtest)(Table15;Figure26B). AgeGenotype#PVsomas/mm2PVsomasize(µm2) 2monthsDnajc5flox/+215.70±12.74696.20±24.07 Dnajc5flox/‐262±14.82672.10±16.90 8monthsDnajc5flox/+196.30±13.26892.50±40.38 Dnajc5flox/‐191.70±16.12915.50±40.51 _____________________________________________________________________________________ Table15.Summaryofthequantitativeanalysisregarding PVcellnumber andsomasizeintheexperimental mice. Meanvalues±SEMindicatingthenumberofPVsomasandPVsomasizeinlayerII/IIIcorticalslicesfor PVcre:Ai27D:Dnajc5flox/+andPVcre:Ai27D:Dnajc5flox/‐miceat2and8monthsold.Valuesareroundedusingtwo decimals.ReferredtoFig.26. Figure26.QuantificationofthenumberandsomasizeofPVinterneuronsfromPVcre:Ai27D:Dnajc5flox/+ and PVcre:Ai27D:Dnajc5flox/‐miceat2and8monthspostnatally. A.BarplotsshowingthenumberofPVsomaspermm2measuredat2(left)and8(right)months,using immunohistochemistryimagesofmotorcorticallayerII/IIIlabelledwithanti‐parvalbuminantibodiesin PVcre:Ai27D:Dnajc5flox/+andPVcre:Ai27D:Dnajc5flox/‐animals.B.BarplotsdisplayingthequantificationofPVsoma size(µm2)at2(left)and8(right)months,usingconfocalimagesofimmunofluorescenceofmotorcorticallayer II/IIIlabelledwithanti‐parvalbuminantibodiesfromPVcre:Ai27D:Dnajc5flox/+andPVcre:Ai27D:Dnajc5flox/‐animals. ErrorbarsindicateSEM.StatisticalsignificanceisobtainedusingStudent’st‐testandMann‐WhitneyUtest. *p=0.05‐0.01.
Results 91 8.ElectrophysiologicalanalysisofPVinterneuronsshowssimilarintrinsic propertiesbutchangesinspontaneoussynapticresponsesin PV cre :Ai27D:Dnajc5 flox/‐ animals OurpreviousresultshaveshownasignificantlossofPV+andSyt2+synapticpunctathatmay suggestadegenerationofparvalbuminpresynapticterminalsinvivo.Toinvestigatethefunctional meaningoftheseresults,weaimedtostudytheintrinsicpropertiesofthistypeofneurons,aswellastheir inhibitoryeffectontoexcitatoryneuronsthroughelectrophysiologicalrecordingsinmotorcorticallayer II/IIIneurons,usingbrainslicesfromPVcre:Ai27D:Dnajc5flox/+andPVcre:Ai27D:Dnajc5flox/‐genotypesatboth 2and8monthspostnatalage. TheelectrophysiologicalexperimentshavebeenentirelycarriedoutbyDr.PabloGarcía‐Junco‐ ClementeandDr.JoséLuisNieto‐González. 8.1.Intrinsicpropertiesofparvalbumininterneurons Todeterminefunctionalintrinsicpropertiesofcorticalparvalbumininterneurons,bothpassive andactivepropertieswereanalyzed.
Results 92 8.1.1.PassivepropertiesofPVcells Thestudyofthepassivepropertiesofneuronsinformsabouthowtransientchangesinmembrane potentialdissipateastheypropagateinspaceandtime(localpotential).Accordingtothis,passive propertiesofPVcellswereexamined.Restingmembranepotential(RMP)wassimilarinboth PVcre:Ai27D:Dnajc5flox/+andPVcre:Ai27D:Dnajc5flox/‐genotypesat2and8months(2months:p=0.4449,ns, Mann‐WhitneyUtest;8months:p=0.2175,ns,Student’st‐test),evenwhenwecomparedbetween controlsandmutantsatdifferentages(controls:p=0.6681,ns,Student’st‐test;mutants:p=0.9715,ns, Mann‐WhitneyUtest).Afterthat,wecalculatedthetimeconstant(tau,τ)applyingsingle50pA hyperpolarizingcurrentinjectiontodeterminethevelocityofthemembranepotentialchange(see MaterialsandMethodssection).Weobservednodifferencesbetweencontrolandmutantanimalsneither at2and8monthsofage(2months:p=0.1385,ns,Student’st‐test;8months:p=0.0764,ns,Student’sttest),norwhencomparingcontrolsandmutantsatdifferentages(controls:p=0.0641,ns,Student’st‐test; mutants:p=0.0779,ns,Student’st‐test)(Figure27A).Next,weappliedbothhyperpolarizingand subthresholddepolarizingcurrentinjectionsusing10pAsteps(seeMaterialsandMethodssection)to determinetheinputresistanceofthePVcells(Figure27B).Intensity‐voltageplotswererepresentedfor eachneuron,andinputresistancewascalculatedaccordingtoOhm’slaw:R=V/I.NochangesinPVinput resistancewasobservedat2monthsbetweencontrolsandmutants(p=0.4075,ns,Mann‐WhitneyUtest); however,PVcellsfrom8‐monthPVcre:Ai27D:Dnajc5flox/‐ miceunderwentasignificantreduction (*p=0.0463,Mann‐WhitneyUtest)ofinputresistancecomparedtoPVcre:Ai27D:Dnajc5flox/+mice(Figure 27C).Aftercomparingcontrolsandmutantsatdifferentages,wecouldnotobserveanyvariationininput resistance(controls:p=0.25,ns,Mann‐WhitneyUtest;mutants:p=0.6554,ns,Mann‐WhitneyUtest). Moreover,thecapacitancevaluesofPVcells(timeconstant/inputresistance)didnotdisplayvariations betweengenotypesorage(2months:p=0.6483,ns,Mann‐WhitneyUtest;8months:p=0.4914,ns, Student’st‐test;controls:p=0.8777,ns,Student’st‐test;mutants:p=0.8402,ns,Mann‐WhitneyUtest) (Figure27D).MeanvaluesforpassivepropertiesofPVcellsareshowninTable16(2months:n=30control PVcellsandn=43mutantPVcells;8months:n=35controlPVcellsandn=34mutantPVcells). AgeGenotypeRMP(mV)Timeconstant (ms) Inputresistance (MΩ)Capacitance(pF) 2monthsDnajc5flox/+‐71.14±1.505.03±0.21144.50±12.7439.49±2.31 Dnajc5flox/‐‐70.21±0.88 4.66±0.15125.20±7.4641.83±2.25 8monthsDnajc5flox/+‐71.88±0.68 5.55±0.19151±9.0039.97±2.1 Dnajc5flox/‐‐70.42±0.95 5.08±0.19126.60±6.0741.90±1.82 Table16.SummaryofthepassivepropertiesofPVcellsintheexperimentalmice. Meanvalues±SEMofrestingmembranepotential(RMP),timeconstant,inputresistanceandcapacitancein PVcre:Ai27D:Dnajc5flox/+andPVcre:Ai27D:Dnajc5flox/‐PVcellsfrommotorcorticallayerII/IIIat2and8monthsold. Valuesareroundedusingtwodecimals.ReferredtoFig.27.
Results 93 8.1.2.ActivepropertiesofPVcells Anactiveneuronalsignalinvolvesachangeinmembranepotentialthatismaintainedoveralong distance(actionpotential).Accordingtothis,thestudyoftheactionpotential(AP)kineticcanprovide essentialevidenceabouthowneuronspropagatetheelectricalinformation.Basedonthispremise,several parametersweremeasuredfromAPofPVcells:amplitude,rise‐time,half‐widthandafter hyperpolarization(AHP)(Figure28A),whichwillbeexplainedindetailduringtheanalysis. First,weanalyzedtheAPparametersforbothgenotypesandages.Forthatpurpose,weapplied theminimumcurrentinjection(seeMaterialsandMethodssection)toevokeasingleactionpotentialin everycell,andafterwardswecalculatedanaveragefromallAPsproducedwithinallPVcells.APthreshold showednovariationsbetweengenotypesorage(2months:p=0.4408,ns,Student’st‐test;8months: p=0.7813,ns,Student’st‐test;controls:p=0.9672,ns,Student’st‐test;mutants:p=0.3288,ns,Student’sttest).Afterthat,APamplitudewascalculatedfromthresholdpeak(Figure28A).Wecouldnotobserveany differenceinAPamplitudebetweenPVcre:Ai27D:Dnajc5flox/+andPVcre:Ai27D:Dnajc5flox/‐genotypes,neither at2nor8months(2months:p=0.4176,ns,Student’st‐test;8months:p=0.9869,ns,Student’st‐test). Interestingly,theAPamplitudewasdecreasedat8monthscomparedto2months,showingsignificant reductionswhencomparingbothcontrols(**p=0.0011,Student’st‐test)andbothmutants (****p<0.0001,Student’st‐test)(Figure28B).RegardingAPrise‐timeandhalf‐width,bothparameters wereverysimilarinallanalyzedcases,althoughshowingsignificantchangesonlybetweencontrolsatthe twodifferentages(APrise‐time:**p=0.0074,Mann‐WhitneyUtest;APhalf‐width:**p=0.0034,Mann‐ WhitneyUtest).MeanvaluesforAPparametersareshowninTable17(2months:n=30controlPVcells andn=43mutantPVcells;8months:n=35controlPVcellsandn=34mutantPVcells). Theafterhyperpolarizationphase(AHP)ofaneuron’sactionpotentialprovidesinformationabout themembranepotentialfallbelowtheRMPduringtherepolarizationstep(alsoknownasundershoot phase)(Figure28A).ForeachPVneuron,theAHPamplitudewascalculatedfromRMPtohyperpolarization peakofanactionpotential,andtheAHPdecaywasmeasuredusinganormalizedAPforeveryparvalbumin AgeGenotypeAPthreshold (mV) APamplitude (mV) APrise‐time (ms) APhalf‐width (ms) 2monthsDnajc5flox/+‐33.91±1.1685.33±2.670.086±0.0030.149±0.005 Dnajc5flox/‐‐35.63±1.6787.81±1.720.091±0.0030.158±0.006 8monthsDnajc5flox/+‐33.97±1.0474.21±1.940.095±0.0030.174±0.006 Dnajc5flox/‐‐33.55±1.1074.25±1.920.097±0.0030.162±0.005 Table17.SummaryofAPpropertiesofPVcellsintheexperimentalmice. Meanvalues±SEMofAPthreshold,amplitude,rise‐timeandhalf‐widthinPVcre:Ai27D:Dnajc5flox/+and PVcre:Ai27D:Dnajc5flox/‐PVcellsfrommotorcorticallayerII/IIIat2and8monthsold.Valuesareroundedusing twoorthreedecimals.ReferredtoFig.28B.
Results 94
Results 95 interneuron.WeobservednochangesinAHPamplitudebetweencontrolandmutantPVcellsneitherat 2nor8months(2months:p=0.2913,ns,Student’st‐test;8months:p=0.4425,ns,Student’st‐test),oreven whencomparingbetweencontrolsandmutantsatdifferentages(controls:p=0.4305,ns,Student’st‐test; mutants:p=0.788,ns,Student’st‐test).Nevertheless,itwasdetectedasignificantdecreaseinAHPdecay (calculatedbyadoubleexponentialfitting,usingaweightedtau)at8monthsinparvalbumincellsfrom PVcre:Ai27D:Dnajc5flox/‐ micecomparedtocontrols(*p=0.0149,Mann‐WhitneyUtest),althougha decreasingtendencyat2months(p=0.8265,ns,Mann‐WhitneyUtest)wasalsoevident.Basedonthis, mutantPVcellsrepolarizefasterthancontrolcells,thustheymightbepreferablyreadytofireAPsagain (Figure28C).Indeed,whencomparingPVcre:Ai27D:Dnajc5flox/‐PVcellsatbothages,wefoundasignificant decreaseat8monthcomparedto2months(**p=0.0085,Mann‐WhitneyUtest).Controlneuronsshowed similarAHPweightedtau(p=0.9954,ns,Mann‐WhitneyUtest).MeanvaluesforAHPparametersare displayedinTable18(2months:n=30controlPVcellsandn=43mutantPVcells;8months:n=35control PVcellsandn=34mutantPVcells). AgeGenotypeAHPamplitude(mV)AHPweightedtau(ms) 2monthsDnajc5flox/+23.68±0.7514.76±5.56 Dnajc5flox/‐22.49±0.7812.22±4.07 8monthsDnajc5flox/+22.94±0.5812.56±2.63 Dnajc5flox/‐22.18±0.804.99±0.91 Finally,studiedtheexcitabilityofPVcells,thatis,theircapacitytogenerateactionpotentials.We followedastepprotocolconsistingof20pAdepolarizingcurrentinjections(seeMaterialsandMethods section)toinduceactionpotentials,andnumberofevokedAPsweremeasuredateachcurrentpulse _____________________________________________________________________________________ Table18.SummaryofAHPpropertiesofPVcellsintheexperimentalmice. Meanvalues±SEMofAHPamplitudeandAHPweightedtauinPVcre:Ai27D:Dnajc5flox/+andPVcre:Ai27D:Dnajc5flox/‐ PVcellsfrommotorcorticallayerII/IIIat2and8monthsold.Valuesareroundedusingtwodecimals.Referred toFig.28C. Figure27.PassiveelectrophysiologicalpropertiesoflayerII/IIIparvalbumininterneuronsinmotorcortical slices. A. Left,representativechangeofthecellmembranepotentialduringhyperpolarizingcurrentinjectionsin current‐clampexperimentsofaPVcre:Ai27D:Dnajc5flox/+mouse.Theexponentialfitting(redline)tothemean membranepotentialisusedtocalculatethetimeconstant.Right,barplotsfromPVcre:Ai27D:Dnajc5flox/+and PVcre:Ai27D:Dnajc5flox/‐miceshowingsimilarmembranetimeconstantsat2and8monthsold.B.Left,current stepprotocoltocalculateinputresistanceusingthecurrent‐clampconfiguration.Right,representativeexample ofmembranepotentialchangesduringtheapplicationofthepreviousprotocolinaPVcre:Ai27D:Dnajc5flox/+ mouse.C.Left,exampleofanintensity‐voltage(I‐V)plotinaPVcre:Ai27D:Dnajc5flox/+genotype,wherelinearfitting (dashedline)isusedtocalculatetheinputresistanceaccordingtoOhm’slaw.Right,barplotsshowingtheinput resistancequantificationforPVcre:Ai27D:Dnajc5flox/+andPVcre:Ai27D:Dnajc5flox/‐genotypesat2and8monthsold. D.BarplotsshowingthecapacitancevaluesofPVinterneuronsforbothgenotypesatthetwopostnatalages.No changesincapacitancearedetected.ErrorbarsindicateSEM.Student’st‐testandMann‐WhitneyUtestareused forstatisticalsignificancedetermination.*p=0.05‐0.01.
Results 96 (Figure29A).WefoundsimilarfiringpatternsincontrolPVcellsatbothages(Figure29C). However,whencomparedtotheirrespectivemutantcells,aslightlycurvedisplacementwasobservedat 8monthsinPVcre:Ai27D:Dnajc5flox/‐cells,showingareductioninAPsnumberateachcurrentinjection, whilenovariationsseemedtohappenat2months,withtheexceptionthatmutantPVcellsseemedto saturatefirstthancontrolneurons(accommodation)(Figure29B).Interestingly,whenwecompared mutantPVcellsatthetwoages,wefoundthat2month‐oldPVinterneuronsfiredanincreasingnumber ofAPsperinjectedcurrentstep,but8month‐oldPVinterneuronsfiredasingleAPafterthesameinjected current,althoughtheyfinallyreachedasimilarfiringpatternafterstrongdepolarizingsteps(Figure29C). Theseobservationswereconfirmedbytherheobaseparameter,whichdescribestheminimalcurrent injectionneededtogenerateatleastoneactionpotentialwith50%likelihood.Figure29Dshoweda significantincreaseinrheobaseofparvalbumincellsfromPVcre:Ai27D:Dnajc5flox/‐ animalscomparedto controlcellsat8months(*p=0.0489,Mann‐WhitneyUtest),indicatingtheyneededastronger depolarizingcurrenttofirethefirstAP,sotheywerelessexcitable.Nochangeswerefoundinrheobase neitherat2months(p=0.4805,ns,Mann‐WhitneyUtest)norwhencomparingcontrolormutantcellsat differentages(controls:p=0.7156,ns,Student’st‐test;mutants:p=0.1081,ns,Mann‐WhitneyUtest). MeanvaluesforrheobaseareshowninTable19(2months:n=30controlPVcellsandn=43mutantPV cells;8months:n=35controlPVcellsandn=34mutantPVcells). AgeGenotypeRheobase(pA) 2monthsDnajc5flox/+219.60±19.23 Dnajc5flox/‐254.50±23.67 8monthsDnajc5flox/+228.90±17.06 Dnajc5flox/‐288.50±21.15 _____________________________________________________________________________________ Table19.SummaryofrheobasevaluesofPVcellsintheexperimentalmice. Meanvalues±SEMofrheobaseforPVcre:Ai27D:Dnajc5flox/+andPVcre:Ai27D:Dnajc5flox/‐PVcellsat2and8months old.Valuesareroundedusingtwodecimals.ReferredtoFig.29D. Figure28.ActiveelectrophysiologicalpropertiesoflayerII/IIIparvalbumininterneuronsinmotorcortical slices. A.RepresentativetraceofaPVactionpotentialindicatingalltheelectrophysiologicalmeasurements.B.Left, averagedactionpotentialobtainedforbothforPVcre:Ai27D:Dnajc5flox/+andPVcre:Ai27D:Dnajc5flox/‐genotypes,at 2and8monthsold.Right,barplotsshowingmeanAPamplitudeofPVinterneuronsforbothgenotypesat2and 8monthsold.Nodifferencesbetweengenotypesarefound;atime‐dependentdecreaseinamplitudeisobserved inbothgenotypes.C.Upper,afterhyperpolarization(AHP)traceobtainedafternormalizingtheaveragedAPfor PVcre:Ai27D:Dnajc5flox/+andPVcre:Ai27D:Dnajc5flox/‐ genotypes,at2and8monthsold.Lowerleft,barplots indicatingmeanAHPamplitudeofPVneuronsforbothgenotypes,at2and8monthsold.NodifferencesinAHP amplitudearedetected.Lowerright,barplotsshowingweightedtauofPVcellsforPVcre:Ai27D:Dnajc5flox/+and PVcre:Ai27D:Dnajc5flox/‐animalsat2and8months.Statisticallysignificantdifferencesarefoundat8monthsin mutantmice.ErrorbarsindicateSEM.Student’st‐testandMann‐WhitneyUtestareusedforstatistical significancedetermination.*p=0.05‐0.01,**p=0.01‐0.001,****p<0.0001.
Results 97
Results 98
Results 99 8.1.3.Excitatorypostsynapticpotentials(EPSPs) Excitatorypostsynapticpotentials(EPSPs)aredepolarizinglocalpotentialsthatmakethe postsynapticneuronmorelikelytofireanactionpotential.Thistemporarydepolarizationofthe postsynapticmembraneisaresultofopeningglutamate‐gatedionchannels.TheEPSPswererecordedto studytheexcitatoryinputsontothecorticalPVcells,socurrent‐clamprecordingswereperformedin absenceofinjectedcurrent(atRMP).NospontaneousAPsweredetectedinPVcellsfromneithercontrol normutantmiceatanyage,butwedidobserveEPSPsinbothgenotypesandages(Figure30A).Foreach PVcell,werecordedthespontaneousEPSPsduring30seconds.Wecalculatedthefrequencyofeventsper cell,andthenweaveragedallEPSPstoobtainatemplatetocalculatetheamplitudeandkineticofthe EPSPspergenotypeandage(seeMaterialsandMethodssection).WecouldnotdetectdifferencesinEPSP frequencybetweenPVcre:Ai27D:Dnajc5flox/+andPVcre:Ai27D:Dnajc5flox/‐genotypesneitherat2nor8months (2months:p=0.3965,ns,Mann‐WhitneyUtest;8months:p=0.1788,ns,Mann‐WhitneyUtest),butitwas observedasignificantage‐dependentfrequencyincreasebothincontrolsandmutants(****p<0.0001for bothanalysis,Mann‐WhitneyUtest).RegardingEPSPamplitude,nochangesweredetectedin PVcre:Ai27D:Dnajc5flox/‐miceneitherat2or8months(2months:p=0.3809,ns,Student’st‐test;8months: p=0.1772,ns,Mann‐WhitneyUtest)norwhencomparingbetweencontrolsandmutantsatdifferentages (controls:p=0.8286,ns,Mann‐WhitneyUtest;mutants:p=0.8608,ns,Mann‐WhitneyUtest)(Figure30B). EPSPrise‐timeandhalf‐widthdisplayedsimilarvaluesforbothgenotypesandages(EPSPrise‐timeat2 months:p=0.2523,ns,Student’st‐test;8months:p=0.8721,ns,Mann‐WhitneyUtest;controls:p=0.0930, ns,Mann‐WhitneyUtest;mutants:p=0.2635,ns,Mann‐WhitneyUtest;EPSPhalf‐widthat2months: p=0.4894,ns,Mann‐WhitneyUtest;8months:p=0.2476,ns,Student’st‐test;controls:p=0.2616,Mann‐ WhitneyUtest;mutants:p=0.5364,ns,Student’st‐test).AveragedvaluesforEPSPpropertiesareshown inTable20(2months:n=28controlPVcellsandn=37mutantPVcells;8months:n=35controlPVcellsand n=34mutantPVcells). _____________________________________________________________________________________ Figure29.ExcitabilitypropertiesoflayerII/IIIparvalbumininterneuronsinmotorcorticalslices. A.Schemeofthedepolarizingcurrentstepprotocol(graytrace)andAPfiringresponse(blacktrace)ofPVcells. ThedashedredboxesmarkmagnificationsofAPresponsestospecificcurrentsteps.B.Numberofaction potentialsproducedfollowinginjectionofdepolarizingcurrentsteps,comparingPVcre:Ai27D:Dnajc5flox/+and PVcre:Ai27D:Dnajc5flox/‐PVcellsat2(left)and8(right)months.C.Numberofactionpotentialsproducedfollowing injectionofdepolarizingcurrentsteps,comparingcontrolcellsat2and8months(left)andmutantcellsat2and 8months(right).D.RheobasevaluesofindividualPVcellsandmeanshowingtheminimalcurrentinjection neededtogenerateatleastoneactionpotentialforbothPVcre:Ai27D:Dnajc5flox/+andPVcre:Ai27D:Dnajc5flox/‐ genotypesat2and8months.Astatisticallysignificantincreaseisdetectedinmutantmiceat8monthscompared tocontrols.ErrorbarsindicateSEM.Student’st‐testandMann‐WhitneyUtestareusedforstatisticalsignificance determination.*p=0.05‐0.01.
Results 106 9.1.RobustaveragemappedmRNAreadspercell ThefirstanalysiswecarriedoutafterreceivingourscRNA‐seqdatawastocalculatethereaddepth (BOX3).TheaveragemappedmRNAreadsobtainedpercellforallchipsweresimilartothatfoundin Hochgerneretal.,2017(41000mappedmRNAreads/cell)(Table22). MappedmRNAreadsMappedmRNAreads/cell WG170029314017338954 WG170038283226934643 WG170047861282032879 WG170058324110234814 10.Analysisoftranscriptomicdataobtainedfromsinglecorticalparvalbumin interneurons DuringourRNAsequencingdataanalysis,wecreatedseveralRscriptsincludingallsteps, annotationsandexplanationsoffunctionsandprocessesused.ThescRNA‐seqanalysiswillbeseparated inthreemainsections:thecreationofareadablematrixfordataanalysis,theintegrationofallthedata forclusteringcells,andthedifferentialexpressionanalysisbetweengenotypepairs. 10.1.Areadablematrixtovisualizegenesexpressedbyeverysinglecell Basedonthewelldistributionsalreadymentioned(Figure13),wecreatedtheReadData.Rscript (Supp.Data:Rscriptsandfiles>Readdata)forreadingrawdata,labellingwellsandcombiningdatasets. WecombinedwellsbelongingtoWT,Flox+andFlox‐genotypesfromeverychipprevioustodownstream analysis.Briefly,thesameprocesswasappliedtothefourrawdatasetstoobtain:(1)attributesandgene expressiondatasavedintotwodifferentdataframesperchip,(2)amatrixcontainingwellpositionstofill inwith“genotype”and“animalnumber”labels(BOX4),andfinally(3)themergeofthefourattributes datasetsandthefourgeneexpressiondatasetstogeneratethefilesAttributesFinal.csvand ExpressionFinal.csv(Supp.Data:Rscriptsandfiles>ReadData).Thesefileswerenecessaryforthenext BOX3 AfterscRNA‐seq,theESCGfacilityprovideduswithrawdatatocalculatesequencing/readdepth,whichdefines theaveragenumberofreadsthataligntoaknownreferencegenome.Usingthenumberofmappedreadsper well,weestimatedtheaverageofmappedmRNAreadspercell:wesummedthemappedreadsalongallwells anddivideditpertotalwellssequenced.2392wellsweresequenced,however,H12‐W01wellpositionfrom everyplatewasdiscardedbeforecalculation,soinourcasewedividedby2391sequencedwells/cells,according tosortingandsequencinglimitationsforaWaferGenchip(seeMaterialsandMethodssection;Figure13). Table22.MappedreadsfromscRNA‐seqinWaferGenchips. TabledisplayingtotalmappedmRNAreadsobtainedfromeachofthefoursequencedWaferGenchips (WG17002,WG17003,WG17004,WG17005),andtheestimatedaveragemappedmRNAreadspercell(total mappedgenesdivideby2391sequencedcellsineachchip)(seeBOX3).Numbersareroundedtounits.
Results 107 steps.ExpressionFinal.csvcontainedinformationofatotalof9568cells(2392sequencedwellsperchip) and24487genes(Figure33A). AsexplainedinBOX4,becausethreedifferentgenotypes(seeMaterialsandMethodssection) werecollectedinthechipsWG17002,WG17003andWG17005,atotalof11wells(+5fromCquadrants, +5fromFquadrants,+1fromHA12‐W01well)weredenotedas“Undetermined”perchipandlater discarded.Ontheotherhand,weonlysortedtwogenotypesintheWG17004array(seeMaterialsand Methodssection).Thisdistributionevadedpotentialcontaminationaftersamplechangingbetweenthe twoPVcre:Ai27D:Dnajc5flox/‐miceandonly6wells(only+5fromFquadrants,+1fromHA12‐W01)were denotedas“Undetermined”inthischipandlaterremoved.Inconclusion,from2392sequencedwellsper chip,2381(2392‐11)remainedasvalidcellsfromtheWG17002,WG17003andWG17005chips,and2386 (2392‐6)remainedasvalidcellsfromtheWG17004chip. 10.2.Integrationofsingle‐celltranscriptomicdataacrossdifferentgenotypes WetestedseveralpackagesforthestudyofourscRNA‐seqdata,suchastheSCE(Lunetal.,2016) andSCDE(Kharchenkoetal.,2014)packages.WefinallydecidedtofollowSeuratpackageinstructionsfor BOX4 StartingwithourfourrawdatasetsfromeachWaferGenmicrowell‐arrayplatform(WG17002,WG17003, WG17004andWG17005),wewrotetheReadDatascripttoreadthisrawdatafromfilesin“cellexpression format”(.cef),asprovidedbytheESCGfacility(Supp.Data:Rscriptsandfiles>ReadData).Thisprototypeoffile containssinglecellsincolumnsandgenesinrows,butitalsohasadditionalinformation,suchaswellnames, platenames,barcodes,etc.OurpurposewastocreateasingledatasetcalledExpressionFinal,whereallsingle cellsfromeacharraywerelocalizedincolumns,andallgenes(theywerealwaysthesameineachdataset) appearedinrows.ExtrainformationfromeachwellwascombinedinadatasetnamedAttributesFinal. AstheCEFformathasawell‐definedstructure(https://github.com/linnarsson‐lab/ceftools),wewrotethescript accordingtoknownpositionsrelativetotheinformationrequired(wecouldalsoknowthesepositionsby openingthe.ceffilewithExcel).Moreover,thankstothesortingpatternprovidedbyESCG(Figure13;Supp. data:ReadData>WGWellsIDs.csv),wewereabletoknowtheidentityofthemousecontainedineach position.Thisfileincludedveryimportantandusefulinformationtorecognizewhereeverysinglecellcame from,thusallowingustoputa“genotype”label(“WT”forPVcre:Ai27D:Dnajc5WTanimals,“Flox+”for PVcre:Ai27D:Dnajc5flox/+animals,or“Flox‐”forPVcre:Ai27D:Dnajc5flox/‐animals;onwards,WT,Flox+andFlox‐due tonomenclaturelimitationswithRlanguage)andan“animalnumber”labeltoeachwell/column.Becauseour strategyaimedtosortaround800cellsfromeachmouseperday,itwasquiteingenioustowriteascript distinguishingbetweenadjacentwellsthatbelongedtodifferentanimals.Infact,twoparticularrowsfromthe platewerespeciallytreated:row4,inrowCquadrants,androw2,inrowFquadrants(onlyinWG17002, WG17003andWG17005chips).Thesetworowquadrantscontainedwellswithtwodifferentgenotypes,dueto samplechangingatFACS.ToavoidpotentialcontaminationwhenchangingsamplesduringFACS,wedecidedto discardfivewellsaroundtheselocations:twowellsfromtheprevioussampleandthreewellsfromthenew sampleafterchanging(Figure13).Werejectedanextrawellfromthenewsamplebecausesorting contaminationseemedmorelikelytohappeninthissample.Thesewellsweredenotedas“Undetermined”and werenotusedfordownstreamanalysis.Furthermore,H12‐W01wellwasalsonamedas“Undetermined”and removedfromtheanalysis,duetoamisunderstandingwiththespecialsortingpatterninH12quadrantthatleft thiswellempty(Figure13).
Results 108 anintegratedanalysis(Butleretal.,2018).Thisisanovelpackagedesignedto(1)explorefirsttwoormore datasetsbyexaminingdataseparately(inthiscasebygenotype)followedby(2)powerfuldataclustering basedoncombining(integrating)thedifferentdatasets.Ourintegrateddataset,containingahighnumber ofcellsuponcombiningdifferentgenotypeswere,therefore,especiallysuitabletoyieldafruitful clusteringbasedonintegratedanalysis.ThismethodisdescribedinButleretal.,2018andexplainedin SatijaLabwebsite(https://satijalab.org/seurat/immune_alignment.html). 10.2.1.Cellfiltering:qualitycontrolbyfilteringoutdatafromlowqualitycells Bydefault,ourdatasetscontainedseveralgroupsofgenesnotsuitableforouranalysisstrategy thatweeliminated.ThesegenesincludedfluorescentmarkersorRNAspike‐inandERCC(ExternalRNA ControlsConsortium)genes(BOX5).Atotalof109geneswereremoved,leaving24378genes(Figure33A). Inaddition,weexcluded“Undetermined”wells,whicheliminated39cells(11+11+11+6)andledtoatotal of9529cells(2381cellsperthreechipsand2386cellsperonechip)(Figure33A).Afterthisinitialfiltering, wesplitExpressionFinal.csvandAttributesFinal.csvbygenotypeintothreeseparateddataframes(2381 WTcells,3169Flox+cells,3979Flox‐cells)(Figure33A),whichwerepreparedforindividualqualitycontrol. Next,weproceededwithcellfiltering,animperativeinitialsteptoremovelow‐qualitycellsin ordertoensurethattechnicalnoisedonotimpairthedownstreamanalysis(Butleretal.,2018;Lunetal., 2016;Mayeretal.,2018;Zeiseletal.,2015).Rcodeforcellfiltering,normalizationandscalingofdatais describedinSeuratintegratedanalysis‐Cellfiltering.Rscript(Supp.Data:Rscriptsandfiles>Seurat integratedanalaysis).Commonqualitycontrol(QC)metricsinvolve:(1)thenumberoftotalUMIs(nUMI), whichisdefinedasthesumofcounts(inthiscase,UMIs)throughallgenesforasinglecell,(2)thenumber ofexpressedgenes(nGene),whichcorrespondstothenumberofgeneswithnon‐zerocountsforasingle cell,and(3)thepercentageofmitochondrialgenesdetectedpersinglecell(percent.mito).AllQC parametersprovideageneralviewofcellquality.CellswithlownUMIornGeneareconsideredlowquality cells,probablybecausemRNAhasnotbeenwellcapturedtobeconvertedtocDNAandamplified,or becausethediversepopulationoftranscriptshavenotbeencapturedequally,respectively.Conversely, highproportionsofmitochondrialgenesindicatepoorcellquality,possiblyduetoincreasedapoptosis and/orlossofcytoplasmicRNAuponcelllysis(Lunetal.,2016). BOX5 Spike‐inandERCCfeaturesareRNAtranscriptsofknownsequenceandquantitythatareusedascontrolgenes tocalibratesequencingmeasurements(Lunetal.,2016;Kharchenkoetal.,2014).ERCCsareusuallyusedto calculatethefittingcurveinsomeanalyses(forexample,SCE‐orSCDE‐basedtutorials),however,theSeurat packageincludesothermethodstoavoidnoisethatdonotrequireERCCs.Instead,Seuratcalculateshighvariable genes(seelater2.DeterminationofHighVariableGenes(HVG)duringtheclusteringsteps).
Results 109 Firstofall,wecreatedaSeuratobject(CreateSeuratObjectfunction)ofeachgenotype‐based dataframe,ausefultooltostorebigdatainanorganizedvariable.Thisfunctionautomaticallydiscarded cellsthatpresentedatotalcountsumequaltozeroacrossallgenes,whichleft2190WTcells,2833Flox+ cellsand3434Flox‐cells(Figure33A).OncetheSeuratobjectwascreated,werepresenteddistributions ofnUMI,nGeneandpercent.mitoforWT,Flox+andFlox‐cells,usingtheVlnPlotfunctionfromSeuratto drawviolinplots,usefulgraphicsthatshowprobabilitydensity.ByexaminingQCmetricswithviolinplots, weobservedthatcelldensitylookedpreferentiallyatthebottominnUMIandnGeneplotsinallgenotypes (Figure33B).Celldensityinpercent.mitoplotslookedquitehighinallgenotypes,andevensomecells reachedto100%(Figure33B).Thesethreeresultssuggestedthatourdatacontainedcellswithpoor quality.TofurtherevaluateQCmetrics,wecalculatedtheaveragenumberfornUMI,nGeneand percentageofmitochondrialgenesbeforefiltering;theyshowedsimilarresultsamonggenotypes(Table 23). AccordingtorecentpublicationsusingUMIs(Butleretal.,2018;Hochgerneretal.,2017;Zeiselet al.,2015),weestablishedaccuratethresholdsfornUMI,nGeneandpercent.mitotoensureanappropriate cellqualitycontrolforourdata:(1)cellswithmorethan1000detectedUMIs,(2)morethan350detected genesand(3)lessthan20%expressedmitochondrialgeneswereretainedusingtheFilterCellsfunction fromSeurat(Figure33B,C).Thisstepleft946WTcells,1470Flox+cellsand1600Flox‐cells(Figure33A) andhighernewmeanvaluesfornUMI,nGeneandpercent.mito(Table23;Figure33C),whichwereagain verysimilarbetweengenotypesandnowcomparabletothatobtainedinHochgerneretal.,2017(3686 genes,8706molecules). BeforecellfilteringAftercellfiltering nUMInGenePercent.mitonUMInGenePercent.mito WT31221106196734233112 Flox+37501286156872232311 Flox‐ 35651141157373231211 10.2.2.Normalizationandnoisereductionofindividualgenesexpressioninsinglecells Aftercellfiltering,theexpressionofeverygeneateverysinglecellneedstobenormalizedtothe totalgeneexpressionineverysinglecell,sothateachcellcanbecomparedagainsteachother. NormalizeDataistheglobalnormalizationfunctionemployedinSeurat.Itnormalizesgeneexpressionfor everysinglecellbyitstotalgeneexpression,thenusesadefaultfactor(10000)tomultiplytoandmakesa log‐transformationoftheproduct.Afternormalization,datawerescaled.Datascalingisanindividualgene Table23.SummaryshowingtheaveragenumbersfornUMI,nGeneandpercent.mitopergenotype. Meanvaluesofthenumberofmolecules(nUMI),thenumberofgenes(nGene)andthepercentageof mitochondrialgenes(percent.mito,%)detectedbeforeandaftercellfilteringwerecalculatedbymeanfunction fromthebasepackageinRforeachgenotype.Valuesareroundedtounits.
Results 110
Results 111 Figure33.FilteringoutlowqualitycellsfromscRNAsequencingdata. A.Sequentialschemedisplayingstepbystepcellselection.Ingreen,thenumberofcellsremoved;inred,cells selectedduringeachstep(blue)alongthebioinformaticsstudy.Thefirstpartoftheanalysisisdetailedinthe ReadData.Rscriptandthesecondpart,inCellfiltering.Rscript.B.Violinplotsshowingnumberofgenes(nGene), numberofUMIs(nUMI)andpercentageofmitochondrialgenes(percent.mito)detectedinWT,Flox+andFlox‐ singlecellsbyscRNAsequencingbeforecellfiltering.Blackdotsrepresentsinglecells.Theorangeareasindicate distributionanddensityoftheanalyzeddata,beingthickenedwheremostofcellsarelocated.VlnPlotfunction isusedtomaketheplots.MinimalthresholdsfornGeneandnUMI(350and1000,respectively)andamaximal thresholdforpercent.mito(20%)areestablishedforcellqualitycontrol(redlines);cellswithlessthan350nGene and1000nUMI,andwithmorethan20%percent.mitoarediscardedasindicativeofpoorcellquality.C.Violin plotsshowingnumberofgenes(nGene),numberofUMIs(nUMI)andpercentageofmitochondrialgenes (percent.mito)detectedinWT,Flox+andFlox‐singlecellsbyscRNAsequencingaftercellfiltering.Again,black dotsrepresentsinglecells,andtheorangeareasindicatedistributionanddensityofanalyzeddata.VlnPlot functionisusedtomaketheplots,representingcellsthatpassthecellqualitycontroldescribedinB.Notethat densityinallparametersisnowdisplacedtohighervalues.
Results 112 treatmenttoeliminatesourcesofvariationthatcouldaffectdownstreamdimensionalityreductionand clusteringtoinferthecorrectgeneexpressionpatterns.Sourcesofvariationmightcomefromtechnical noise,batcheffect(inourcase,thevariabilityduetodifferentexperimentaldays)ornoiseofbiological origin.ScaleDataistheSeuratfunctionusedforscaling,whichconstructslinealmodelstopredictgene expressionbasedonthevariablesweincluded(toregresson),andthencenters(theexpressionofeach geneineverysinglecelliscenteredbysubtractingtheaverageexpressionforthatgene)andscales(the centeredexpressionvalueforeachgeneineverysinglecellisdividedbyitsstandarddeviation)the resultingvaluesinthedataset.Wetestedtheeffectonclusteringofregressingoutgenesaccordingto nUMI,nGene,thepercentageofmitochondrialgenesandthenumberofplate(forbatcheffect).Wefound usefultoregressoutgenesbasedonnUMIasapotentialsourceofvariation,butweobservedthat regressingouttheothermentionedfeaturesdidnotalterresultssubstantially.Filtered,normalizedand scaleddatafromseparatedgenotypesweresavedfordownstreamanalysis. 10.2.3.ClusteringanalysisrevealsdifferentsubpopulationsofPVinterneurons Anumberofrecentstudiesbasedonsinglecelltranscriptomicshaveadvancedthemolecular taxonomyofneuronsandrevealedtheexistenceofdifferentsubpopulations(clusters)ofPVinterneurons (Harrisetal.,2018;Tasicetal.,2016,2018).Therefore,weproceededtoinvestigatetheexistenceofPV clustersamongourselectedcells.Onceweobtaineddatasetscontaining“healthy”cellsandseparatedby genotype,wecarriedoutasequentialclusteringanalysis. Overall,weperformedfourconsecutiveclusteringstepstofinallyachievearationalandconsistent PVclustering.Afterthefirstthreeclusteringsteps,wegotexpressionofcell‐damage‐relatedgenesornonneuronalgenesasclustermarkers,soweexcludedthesecellpopulationsandstartedagainanew clusteringanalysis.Thissubparagraphhasbeenorganizedinfoursections/clusteringanalyses,whereeach clusteringconsistedoffivemainprocesses.Becausefromthe2ndtothe6thstepwererepeatedper clusteringanalysis,theywillbeonlyfullydescribedinthefirstonetoavoidredundancy. 10.2.3.1.Firstclustering WeappliedseveralcriteriatobuildupafirstsetofPVneuronclusters.Asexplainedbelow,we attendedfirsttotheexpressionofPvalb,wecalculatedhighvariablegenes(HVG),andthenapplied canonicalcorrelationanalysis(CCA)forclustering.ThecorrespondingcodeisinSeuratintegratedanalysis ‐PVclustering1.Rscript(Supp.Data:Rscriptsandfiles>Seuratintegratedanalysis). 1) SelectionofPvalbmRNAexpression WeevaluatedPvalbmRNAexpressionandremovednon‐Pvalb‐expressingcells.Wefoundthatthe majorityoftheremainingcellspergenotypeexpressedPvalbmRNA(Pvalbexpression>0UMI;WT:86%,
Results 113 Flox+:85%,Flox‐83%)whileonlysomecellsdidnot.AfterremovingcellsthatdidnotexpressedPvalb,we obtained817WTcells,1257Flox+cellsand1331Flox‐cells(Figure35A). 2) DeterminationofHighVariableGenes(HVG) Thesecondstepinvolvedthedetectionofhighvariablegenes(HVG).Thesearegenesthat contributestronglytocell‐to‐celltranscriptomicvariationwithinahomogeneouscellpopulationand thereforeareusefultoidentifyclusters.ThedetectionofHVGisrequiredtoapplytheCCA(Canonical CorrelationAnalysis).TheRfunctioninSeuratthatidentifiesoutliergenesisFindVariableGenes(BOX6). WeobtainedHVGseparatelypergenotype(S3):6074genesinWT,5728genesinFlox+and5063 genesinFlox‐cells.However,onlyHVGpresentinatleasttwoofthedatasetswereselectedwithinthe first1000HVGfromeachgenotype,whichfinallyyielded703HVGintotal.Anadditionalrequirementfor agentobeconsideredHVGwastohaveavariancevaluedifferenttozeroineverydataset.Accordingly, theRunMultiCCA(see3.Runningmulti‐setCCAandCCselectionforcellclusteringsection)functionapplied tothepreviousHVGselection,detectedandfilteredoutanygenewithzerovarianceinanyofthedatasets, reducingtheHVGfinalnumberfrom703to698HVG(Figure35A). 3) Runningmulti‐setCCAandCCselectionforcellclustering Thedimensionofourdatasetatthatmomentwas24378genes(with698HVG)by3405cells (Figure35A),butmanygenesandsinglecellswereprobablynotinterestingbecausetheydidnotvarya lot.ThegoalofPrincipalComponentAnalysis(PCA),adimensionalityreductionprocedure,istofindthe vectors(usuallythefirsttwoprincipalcomponents)inwhichthereismostvariance.Inotherwords,the aimofdimensionalityreductionistoremoveredundantinformationtodescribedataonlybyfeatureswith highvariance,i.e.,findingalow‐dimensionsetofaxestodefineourdata (https://towardsdatascience.com/dimensionality‐reduction‐does‐pca‐really‐improve‐classification‐ outcome‐6e9ba21f0a32).Inthisway,cellswithhighcorrelation(similarcells)willappeartogether,while distinctcellswillbeseparated. BOX6 FindVariableGenesusesaplotthatdisplaysinlogarithmicscaletheaverageexpression(weusedmean)versus adispersionmeasurement(weemployedvariancetomeanratio,thatis,thestandarddeviation)foreachgene acrosscells.Thisfunctionprovidesausefulmethodtoidentifygenesthatshowastrongrelationbetween variabilityandaverageexpressionwithincells,basedonpredefinedthresholdsforeachparameter.These limitationswereestablishedaccordingtorecommendedsettingsforUMIdatanormalizedto10000molecules (https://satijalab.org/seurat/pbmc3k_tutorial.html),whichcomprised0.0125and3forminimalandmaximal cutoff,respectively,ofthemeanexpression(Xaxis),and0.5forminimalcutoffofthestandarddeviation(Yaxis). Wewereonlyinterestedinidentifyinggeneswhichexpressionvariessignificantlywithincellsofeachgenotype andthereforetheyshowahighstandarddeviation.
Results 114 AfterselectingHVG,thethreedatasetswerefinallyintegratedinonedataset(socalled“Integrated analysis”)inordertorunamulti‐setdimensionalityreductionwiththeselectedhighvariablegenes. RunMultiCCAfunctionwasusedtoperformdimensionalityreductionwiththethreedatasets(similarto PCA,whichusesonlyonedataset)for30canonicalvectors(CCs)orsubspaces,takinggenotype‐mixedcells andfindingcommonsourcesofvariation.EachCCwouldbasicallyrepresenta“metagene”setthat describesaspecificgroupofcellsinourdata.InFigure34A,B,wecanobservecelldistributionusingaCC1 violinplot,andalsobyplottingCC1againstCC2,thetwofirstcanonicalcorrelationvectors(CCs),after runningCCA.Althoughtwodistinguishablegroupswereeasilyidentifiedinbothplots,singlecells overlappedbetweengenotypesandnoshiftedpopulationsweredetected‐exceptforamilddisplacement ofFlox‐cellsinthesmallergroup‐,suggestingtheabsenceofuniquepopulationsinanyofthem,atleast usingthefirsttwoCCs. Normally,thefirstCCsexplainmostofthevariabilityinthedata,makingtherestofdimensions somehowdispensable.Nevertheless,itisveryimportanttochooseCCscorrectlyfordownstreamanalysis toovercomethetechnicalnoiseinanysinglegene,becauseSeuratemploysdimensionalityreduction scoresforclustering.ForCCselection,wetookadvantageoftwoapproaches.Weusedthe MetageneBicorPlotfunctiontoplotthemeasurementofcorrelationstrengthforthe30CCs(bicor‐plot), inordertofindwhenthecurvesaturates,whichgaveusthestoppointforCCsselectionsincenextCCs wouldnotgiveextrainformation(Figure34B).Inaddition,wealsoexaminedheatmaps(auseful visualizationoftheabovementioned“metagene”)ofthetop10genesdrivingeachoneofthe30CCs, usingDimHeatmapfunction,butusingonly500cellstospeeduptheprocess.Consideringthesetwo visualizations,wedecidedtoreasonablyselectthefirst17CCs,becauseweobservedthatFlox+andFlox‐ curvesinthebicor‐plotextinguishedatCC17(Figure34B),andheatmapsbecameundistinguishablealso afterCC17(S7).Inaddition,wedetectedthatheatmapsfromCC18toCC30displayedthesamegenes, confirmingthatthesedimensionswerenotprovidingadditionalinformationastheycameonlyfromWT data(S7). 4) AlignmentofCCAsubspaces Asafinalprocessbeforeclustering,dataweresubjectedtoaCCAsubspacesalignmenttocorrect scaleanddensitychangesbetweengenotypesthatcouldleadtodifferencesamongCCs.Itwasdoneby theAlignSubspacefunction,whichreturnedanewdimensionalreductionaligned,inourcase,forthe17 selectedCCs,thatwereusedforclustering.AlignedCC1andCC2wereshownbyviolinplotsinFigure34C.
Results 115 5) Cellclusteringandnon‐lineardimensionalreduction(tSNE) AligneddatawereusedasinputforFindClusters,theSeuratfunctioninvolvedingroupingcellsby clusters/identities.Briefly,itfirstcalculatesk‐nearestneighbors(KNN)basedonEuclideandistanceinCCA spaceandconstructsasharednearestneighbor(SNN)plot;thenitadjuststhemodularityfunctionto defineclusters.Thatis,FindClustersincorporatesaresolutionparameterthatestablishestheclustering granularity,wherehighlevelscouldleadtoagreaternumberofidentitiesinlargerdatasets.Foradataset Figure34.Multi‐setCCArunning,CCsselectionandCCAsubspacesalignmentduringthefirstclustering. A.ViolinplotdisplayingWT,Flox+andFlox‐celldistributiondescribedbythefirstsubspaceorcanonicalvector, CC1.Blackdotsrepresentsinglecells,andcolorsindicategenotype.B.Dotplotshowingcelldistribution representedbythetwofirstcanonicalvectors,CC1versusCC2,afterrunningCCAdimensionalreduction.DimPlot isusedtocreatethisplot.Again,eachdotrepresentsasinglecell,andcolorsindicategenotype.Cellsare separatedintwodistinguishablegroups,bothcontainingcellsfromallgenotypes.C.MetageneBicorPlotfunction generatesabicor‐plotshowingmeasuresofcorrelationstrength(Yaxis)foreachCC(Xaxis).Colorsindicate genotype.CCselectionisestablishedatCC17asFlox+andFlox‐curvesextinguishedafterthisCC.D.Violinplots showingcelldistributiondescribedbyCC1(left)andCC2(right)scoresafterCCAsubspacesalignment(ACC).Note thatallgenotypesaresuccessfullyaligned.
Results 122 5) Cellclusteringandnon‐lineardimensionalreduction(tSNE) Next,FindClusterswasrunusingthealignedCCAsubspaces,with22CCs.Aresolutionparameter of0.6revealed9clusters(0‐8),wherecellsfromallgenotypeswereagainspreadwithinthe9identities (Figure39A,B). 6) Identificationofconservedcelltypemarkers ClustermarkersdetectedinthissecondclusteringareavailableatSupp.Data:Conservedmarkers (CSV).Violinrepresentationsofclustermarkersrevealedthatidentitymarkersforpopulations0to4were higherexpressedinthesegroups,althoughtheywerealsodetectedintheotherclusters;whilemarkers forpopulations5to8seemedtobemorespecific.Allofthemshowedap‐value<0.05forthethree genotypes(Figure40). Figure39.SecondclusteringofcellsfromscRNAsequencingdata. A.Sequentialschemedisplayingstepbystepselectionofcellsalongthesecondclusteringevent;inred,number ofcells,genesandCCsmaintainedorselectedduringeachstep(2‐5,blue)alongtheSeuratintegratedanalaysis ‐ PVclustering2.Rscript.Nineclustersarefoundafterthesecondclusteringstep.Thefinaltableshowsthe numberofcellspercluster(0‐8).Theredcrossindicateslaterremovalofcluster8.B.Left,tSNEplotshowingcell populationswithacolorpatternbasedonthe9clustersobtainedafterthesecondclustering.Right,thesame tSNErepresentationcoloredbygenotype.ThethreegenotypesarespreadandpresentinallPVclusters.Identities areranked(0‐8)accordingtohighernumberofcells.HVG,highvariablegenes;CCA,canonicalcorrelation analysis;CCs,canonicalvectors.
Results 123 Figure40.ClustermarkersdefiningPVinterneuronpopulationsafterthesecondclusteringstep. Violinplotsdisplayingclustermarkerexpression(Yaxis)acrossclusters(Xaxis)basedonprobabilitydistributions. AllgenesshowninthisfigurecorrespondtothefirstsignificantconservedmarkerobtainedforeachPVcluster. Markersareselectedbycomputingthemeanofavg_logFCvaluesfromWT,Flox+andFlox‐cellsperpopulation. Theredarrowindicateswhichidentitydefineseachconservedmarker.
Results 124 Nevertheless,afterexploringclustermarkers,werealizedthatidentity8containedgenes(Apoe, Sepp1,Itm2a,Fcer1g,Dbi,Crip1,Cst3,B2m,C1qc,Ctss,etc.)belongingtonon‐neuronalcellpopulations, suchasastrocytes,oligodendrocytes,andmacrophages,withinothers.Thosegeneswerebarelyornot expressedinPVsubpopulations,accordingtomousecelltypesfromtheAllenBrainAtlas (http://celltypes.brain‐map.org/rnaseq/mouse;Tasicetal.,2016,2018)(Figure41A).Whenwelookedat nUMIinallclusters,wefoundnopotentialdoubletsinidentity8(Figure41B),neitheranevidentbatch effect(Figure41C)noracleargenotypicpredominance(Figure41D).Therefore,weremovedcluster8due tonon‐neuronalcellcontaminationthatmighthaveremainedmaskedduringthefirstclustering.Data wereagainseparatedintocellsfromPVcre:Ai27D:Dnajc5WT(726cells),PVcre:Ai27D:Dnajc5flox/+(1074cells) andPVcre:Ai27D:Dnajc5flox/‐(1150cells)mice(Figure43A). 10.2.3.3.Thirdclustering Thethirdclusteringstartedagainfromthe2ndtothe6thstepoftheclusteringprocess.Seurat integratedanalysis‐PVclustering3.RscriptdescribestheRcodeforclustering3(Supp.Data:Rscriptsand files>Seuratintegratedanalysis). 2) Determinationofhighvariablegenes(HVG) FindVariableGeneswasrunagain,identifying6801HVGforWT,6788HVGforFlox+and6590HVG forFlox‐cells(S5).649geneswereselectedasHVGtakingintoaccountthefirst1000genesfromthethree datasets(Figure43A). 3) Runningmulti‐setCCAandCCselection Onemoretime,weobservedtwogroupsofcellsafterplottingCC1andCC2dimensions,withcells overlappinginbothgroupsfromallgenotypes(Figure42A,B).WTandFlox‐ curvesinthebicor‐plot followedanormaldecreasewithavisibleelbowdrivingtosaturationafterCC20,whileFlox+line experimentedahighincreasebeforeCC20(Figure42C).WefollowedthesameguidelinesforCCselection asbefore,soheatmapswerealsoexamined.CCdimensionsvisualizedbyheatmapsbecamemessyand _____________________________________________________________________________________ Figure41.Analysisofidentity8obtainedfromthesecondclustering. A.HeatmapgeneratedfromtheAllenBrainAtlasformousecelltypes.Itshowstheexpressionlevelsofthefirst significantmarkersdefiningcluster8withintheirestablishedPVpopulationsandsomenon‐neuronalsubclasses, includingastrocytes,oligodendrocytes,VLMCandmacrophages.Thiscomparisondemonstratesthatthefirst highlyexpressedgenesfoundintheidentity8belongtonon‐neuronalsubclasses(yellowbox)andnottoPVones. B.Violinplotsexaminingthenumberofmolecules(nUMI)percell(blackdot)inallclusters(0‐8).Theredarrow indicatestheidentitysubjectedtofurtherexamination.C.ViolinplotsshowingnUMIpercellincluster8based onmouseoriginnumber.D.ViolinplotsshowingnUMIpercellincluster8basedonthedifferentgenotypes (PVcre:Ai27D:Dnajc5WT,PVcre:Ai27D:Dnajc5flox/+andPVcre:Ai27D:Dnajc5flox/‐).Thesamecolorsdenotinggenotypes aremaintained.Datasuggestnopotentialdoublets,batcheffectoracleargenotypepredominanceincluster8. Notetheabsenceofcellsfromsomemiceincluster8(10miceoutofn=12totalmice).OPC,oligodendrocyte progenitorcells;VLMC,vascularandleptomeningealcells;PVM,perivascularmacrophages.
Results 125 showedthesamegeneexpressionafterCC23(S9).Thus,23CCswerechosenforalignment. 4) AlignmentofCCAsubspaces Figure42DshowedalignedCC1andCC2afteralignmentofsubspacesusing23CCs.
Results 126 5) Cellclusteringandnon‐lineardimensionalreduction(tSNE) Next,clustermarkersdetectionwascarriedoutusingthefirst23CCsandaresolutionparameter of0.6,whichidentified8clusters(0‐7)fromtheremainingcells.Onemoretime,allgenotypeswere presentineachoftheidentities(Figure43A,B). Figure42.Multi‐setCCArunning,CCsselectionandCCAsubspacesalignmentduringthethirdclustering. A.ViolinplotdisplayingWT,Flox+andFlox‐celldistributiondescribedbythefirstsubspaceorcanonicalvector, CC1.Blackdotsrepresentsinglecells,andcolorsindicategenotype.B.Dotplotshowingcelldistribution representedbythetwofirstcanonicalvectors,CC1versusCC2,afterrunningCCAdimensionalreduction.DimPlot isusedtocreatethisplot.Again,eachdotrepresentsasinglecell,andcolorsindicategenotype.Cellsare separatedintwodistinguishablegroups,bothcontainingcellsfromallgenotypes.C.Bicor‐plotgeneratedby MetageneBicorPlotfunctionshowingmeasuresofcorrelationstrength(Yaxis)foreachCC(Xaxis).Colorsindicate genotype.CCselectionisestablishedatCC23.D.ViolinplotsshowingcelldistributiondescribedbyCC1(left)and CC2(right)scoresafterCCAalignment(ACC).Notethatallgenotypesareproperlyaligned.
Results 127 6) Identificationofconservedcelltypemarkers IdentificationofclustermarkersduringthisthirdclusteringwaspresentedatSupp.Data: Conservedmarkers(CSV).Whenlookingatviolinplotsshowingthefirstidentitymarkerswithp‐value< 0.05forallgenotypes,weobservedthatmarkersforclusters0to3werealsoexpressedintheother populations.However,clustermarkersforpopulations4to7weremainlydetectedintheirownidentities (Figure44). Figure43.ThirdclusteringofcellsfromscRNAsequencingdata. A.Sequentialschemedisplayingstepbystepselectionofcellsalongthethirdclusteringevent;inred,numberof cells,genesandCCsmaintainedorselectedduringeachstep(2‐5,blue)alongtheSeuratintegratedanalysis‐PV clustering3.Rscript.Eightclustersarefoundafterthethirdclusteringstep.Thefinaltableshowsthenumberof cellspercluster(0‐7).Theredcrossindicateslaterremovalofcluster4.B.Left,tSNEplotshowingcellpopulations withacolorpatternbasedonthe8clustersobtainedafterthethirdclustering.Right,thesametSNE representationcoloredbygenotype.ThethreegenotypesarespreadandpresentinallPVclusters.Identitiesare ranked(0‐7)accordingtohighernumberofcells.HVG,highvariablegenes;CCA,canonicalcorrelationanalysis; CCs,canonicalvectors.
Results 128 Figure44.ClustermarkersidentifyingPVinterneuronpopulationsafterthethirdclusteringstep. Violinplotsdisplayingclustermarkerexpression(Yaxis)acrossclusters(Xaxis)basedonprobabilitydistributions. AllgenesshowninthisfigurecorrespondtothefirstsignificantconservedmarkerobtainedforeachPVcluster. Markersareselectedbycomputingthemeanofavg_logFCvaluesfromWT,Flox+andFlox‐cellsperpopulation. Theredarrowindicateswhichidentitydefineseachconservedmarker.
Results 129 Theroutineexaminationofclustergenesdemonstratedthatidentity4surprisinglycontained ribosomalRNA(Rn4.5s)andpredicted“Gm”genes(suchasGm1821),indicatingapotentialcelllysis.This resultwassimilartothatfoundinidentity3duringthefirstclustering,butthistimecluster4alsoenclosed severalneuronalgenesthatweredetectedinotheridentities,forexample,Snca,Kcna1andPthlh, suggestingthatcluster4englobeddamagedPVcells–probablypreviouslymaskedamongdifferent clusters‐thatemergedasaconsequenceofrunninganewclusteringanalysis.Inthatway,re‐clustering usingthesameresolutionparameter(0.6)mighthadcollectedthesecellsandhadputthemtogether becausetheirexpressionpatternsofdamage‐relatedgeneswereverysimilar.Becausethesegeneshave beendetectedtwiceinourdata(seeFirstclustering‐step6),wedecidedtoconfirmthattheywereactually notexpressedinPVinterneuronsusingtheAllenBrainAtlasformousecelltypes(http://celltypes.brainmap.org/rnaseq/mouse).Indeed,itwasfoundnoexpressionofribosomalRNARn4.5sinparvalbumincells, buttheGm1821gene(anubiquitinpseudogene)wasrandomlydetectedamongPVsubclasses(Figure 45A).Wealsodetectedthatidentity4comprisedcellswiththehighestnumberofUMIs(Figure45B), suggestingthatthisclustermightharborsomedoublets,althoughnobatcheffectseemedtohappen (Figure45C)oragenotypicpredominanceincellorigin(Figure45D).Alltheseobservationsconcludedwith cluster4removalfordownstreamanalysis,leadingtoafourthre‐clusteringwiththeremainingcells:712 cellsfromPVcre:Ai27D:Dnajc5WTmice,1027cellsfromPVcre:Ai27D:Dnajc5flox/+miceand1108cellsfrom PVcre:Ai27D:Dnajc5flox/‐mice(Figure47A). 10.2.3.4.Fourthclustering DetailedRcodeforthefourthclusteringisdescribedinSeuratintegratedanalysis‐PVclustering 4.Rscript(Supp.Data:Rscriptsandfiles>Seuratintegratedanalysis),includingfromthe2ndto6thstep. 2) Determinationofhighvariablegenes(HVG) Highvariablegenesinthisoccasioninvolved6822genesforWT,6857genesforFlox+and6564 genesforFlox‐cells(S6).635geneswereselectedasHVGaftertheevaluationofthefirst1000genesfrom thethreedatasets(Figure47A). 3) Runningmulti‐setCCAandCCselection Asbefore,CC1andCC2describedtwodistinctgroups,bothcontainingcellsoverlappingfromall genotypes(Figure46A,B).Similartothethirdclustering,WTandFlox‐curvesinthebicor‐plotfolloweda normaldecreasewithadetectablesaturationtendencyafterCC20,butFlox+curveshowedabigrise aroundCC15(Figure46C).Also,althoughheatmapslookedprettyindistinguishablealmostsinceCC5,it wasafterCC21whenwedetectedthesamegeneexpressionpatterns(S10).Thisfactdirectedourdecision totake21CCsfordownstreamanalysis.
Results 130 Figure45.Analysisofidentity4obtainedfromthethirdclustering. A.HeatmapgeneratedfromtheAllenBrainAtlasformousecelltypes.Itshowstheexpressionlevelsofthefirst twosignificantmarkersdefiningcluster4(Rn4.5sandGm1821)withintheirestablishedPVsubpopulations.No expressionisdetectedforRn4.5s,andrandomexpressionofGm1821isfoundamongPVsubclasses.B.Violinplots examiningthenumberofmolecules(nUMI)percell(blackdot)inallclusters(0‐7).Theredarrowindicatesthe identitysubjectedtofurtherexamination.C.ViolinplotsshowingnUMIpercellincluster4basedonmouseorigin number.D.ViolinplotsshowingnUMIpercellincluster4basedonthedifferentgenotypes(PVcre:Ai27D:Dnajc5WT, PVcre:Ai27D:Dnajc5flox/+andPVcre:Ai27D:Dnajc5flox/‐).Thesamecolorsdenotinggenotypesaremaintained.Data showthatcluster4includescellswiththehighestnUMIvalues,butsuggestnobatcheffectoracleargenotype p redominance.
Results 131 4) AlignmentofCCAsubspaces AlignSubspacefor21CCsproducedthenewdimensionalreductionthatwasshownbyCC1and CC2inFigure46D. Figure46.Multi‐setCCArunning,CCsselectionandCCAsubspacesalignmentduringthefourthclustering. A.ViolinplotdisplayingWT,Flox+andFlox‐celldistributiondescribedbythefirstsubspaceorcanonicalvector, CC1.Blackdotsrepresentsinglecells,andcolorsindicategenotype.B.Dotplotshowingcelldistribution representedbythetwofirstcanonicalvectors,CC1versusCC2,afterrunningCCAdimensionalreduction.DimPlot isusedtocreatethisplot.Again,eachdotrepresentsasinglecell,andcolorsindicategenotype.Cellsare separatedintwodistinguishablegroups,bothcontainingcellsfromallgenotypes.C.Bicor‐plotgeneratedby MetageneBicorPlotfunctionshowingmeasuresofcorrelationstrength(Yaxis)foreachCC(Xaxis).Colorsindicate genotype.CCselectionisestablishedatCC21.D.ViolinplotsshowingcelldistributiondescribedbyCC1(left)and CC2(right)scoresafterCCAalignment(ACC).Notethatallgenotypesareproperlyaligned.