University of Groningen
Functional parcellation of human and macaque striatum reveals human-specific connectivity
in the dorsal caudate
Liu, Xiaojin; Eickhoff, Simon B; Caspers, Svenja; Wu, Jianxiao; Genon, Sarah; Hoffstaedter,
Felix; Mars, Rogier B; Sommer, Iris E; Eickhoff, Claudia R; Chen, Ji
Published in:
Neuroimage
DOI:
10.1016/j.neuroimage.2021.118006
IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from
it. Please check the document version below.
Document Version
Publisher's PDF, also known as Version of record
Publication date:
2021
Link to publication in University of Groningen/UMCG research database
Citation for published version (APA):
Liu, X., Eickhoff, S. B., Caspers, S., Wu, J., Genon, S., Hoffstaedter, F., Mars, R. B., Sommer, I. E.,
Eickhoff, C. R., Chen, J., Jardri, R., Reetz, K., Dogan, I., Aleman, A., Kogler, L., Gruber, O., Caspers, J.,
Mathys, C., & Patil, K. R. (2021). Functional parcellation of human and macaque striatum reveals
human-specific connectivity in the dorsal caudate. Neuroimage, 235, [118006].
https://doi.org/10.1016/j.neuroimage.2021.118006
Copyright
Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).
Take-down policy
If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.
Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the number of authors shown on this cover page is limited to 10 maximum.
ContentslistsavailableatScienceDirect
NeuroImage
journalhomepage:www.elsevier.com/locate/neuroimage
Functional
parcellation
of
human
and
macaque
striatum
reveals
human-specific
connectivity
in
the
dorsal
caudate
Xiaojin
Liu
a,b,
Simon
B.
Eickhoff
a,b,
Svenja
Caspers
c,d,
Jianxiao
Wu
a,b,
Sarah
Genon
a,b,
Felix
Hoffstaedter
a,b,
Rogier
B.
Mars
e,f,
Iris
E.
Sommer
g,
Claudia
R.
Eickhoff
c,h,
Ji
Chen
a,b,
Renaud
Jardri
i,
Kathrin
Reetz
j,k,
Imis
Dogan
j,k,
André Aleman
l,
Lydia
Kogler
m,
Oliver
Gruber
n,
Julian
Caspers
c,o,
Christian
Mathys
o,p,q,
Kaustubh
R.
Patil
a,b,∗a Institute of Neuroscience and Medicine (INM-7), Heinrich Heine University Düsseldorf, Düsseldorf, Germany b Institute of Systems Neuroscience, Research Centre Jülich, Jülich, Germany
c Institute of Neuroscience and Medicine (INM-1), Research Centre Jülich, Jülich, Germany d Institute for Anatomy I, Medical Faculty, Heinrich Heine University Düsseldorf, Düsseldorf, Germany
e Wellcome Centre for Integrative Neuroimaging, Centre for Functional MRI of the Brain (FMRIB), Nuffield Department of Clinical Neurosciences, John Radcliffe
Hospital, University of Oxford, Oxford, United Kingdom
f Donders Institute for Brain, Cognition and Behaviour, Radboud University Nijmegen, Nijmegen, Netherlands g Department of Biomedical Sciences of Cells & Systems, University Medical Center Groningen, Groningen, Netherlands h Institute of Clinical Neuroscience and Medical Psychology, Medical Faculty, University of Düsseldorf, Düsseldorf, Germany i Division of Psychiatry, University of Lille, CNRS UMR9193, SCALab & CHU Lille, Fontan Hospital, CURE platform, Lille, France j JARA-BRAIN Institute Molecular Neuroscience and Neuroimaging, Forschungszentrum Jülich, RWTH Aachen University, Aachen, Germany k Department of Neurology, RWTH Aachen University, Aachen, Germany
l Department of Neuroscience, University Medical Center Groningen, University of Groningen, Groningen, Netherlands m Department of Psychiatry and Psychotherapy, Medical School, University of Tübingen, Germany
n Section for Experimental Psychopathology and Neuroimaging, Department of General Psychiatry, Heidelberg University, Germany o Department of Diagnostic and Interventional Radiology, Medical Faculty, University of Düsseldorf, Düsseldorf, Germany p Research Center Neurosensory Science, Carl von Ossietzky Universität Oldenburg, Oldenburg, Germany
q Institute of Radiology and Neuroradiology, Evangelisches Krankenhaus, University of Oldenburg, Oldenburg, Germany
a
r
t
i
c
l
e
i
n
f
o
Keywords: Non-human primate Cross-species comparison Striatum Connectivity-based parcellation Parkinson’s disease Schizophreniaa
b
s
t
r
a
c
t
Awidehomologybetweenhumanandmacaquestriatumisoftenassumedasinboththestriatumisinvolved incognition,emotionandexecutivefunctions.However,differencesinfunctionalandstructuralorganization betweenhumanandmacaquestriatummayrevealevolutionarydivergenceandshedlightonhumanvulnerability toneuropsychiatricdiseases.Forinstance,dopaminergicdysfunctionofthehumanstriatumisconsideredtobe apathophysiologicalunderpinningofdifferentdisorders,suchasParkinson’sdisease(PD)andschizophrenia (SCZ).Previousinvestigationshavefoundawidesimilarityinstructuralconnectivityofthestriatumbetween humanandmacaque,leavingthecross-speciescomparisonofitsfunctionalorganizationunknown.Inthisstudy, resting-statefunctionalconnectivity(RSFC)derivedstriatalparcelswerecomparedbasedontheirhomologous cortico-striatalconnectivity.Thegoalherewastoidentifystriatalparcelswhoseconnectivityishuman-specific comparedtomacaqueparcels.Functionalparcellationrevealedthatthehumanstriatumwassplitintodorsal, dorsomedial,androstralcaudateandventral,central,andcaudalputamen,whilethemacaquestriatumwas dividedintodorsal,androstralcaudateandrostral,andcaudalputamen.Cross-speciescomparisonindicated dissimilarcortico-striatalRSFCofthetopographicallysimilardorsalcaudate.Weprobedclinicalrelevanceofthe striatalclustersbyexaminingdifferencesintheircortico-striatalRSFCandgraymatter(GM)volumebetween patients(withPDandSCZ)andhealthycontrols.WefoundabnormalRSFCnotonlybetweendorsalcaudate,but alsobetweenrostralcaudate,ventral,centralandcaudalputamenandwidespreadcorticalregionsforbothPD andSCZpatients.Also,weobservedsignificantstructuralatrophyinrostralcaudate,ventralandcentralputamen forbothPDandSCZwhileatrophyinthedorsalcaudatewasspecifictoPD.Takentogether,ourcross-species comparativeresultsrevealedsharedandhuman-specificRSFCofdifferentstriatalclustersreinforcingthecomplex organizationandfunctionofthestriatum.Inaddition,weprovidedatestablehypothesisthatabnormalitiesina regionwithhuman-specificconnectivity,i.e.,dorsalcaudate,mightbeassociatedwithneuropsychiatricdisorders.
∗Correspondingauthor.
E-mailaddress:k.patil@fz-juelich.de(K.R.Patil).
https://doi.org/10.1016/j.neuroimage.2021.118006
Received15June2020;Receivedinrevisedform10February2021;Accepted17March2021 Availableonline2April2021
1. Introduction
Animal models provide important perspectives on neural func-tions,structures anddisease,withseveral studiesreportingan over-allfunctional andstructuralsimilaritybetween non-humanprimates andhumans(Goulasetal., 2014; Marsetal., 2011,2018; Miranda-Dominguezetal.,2014;vandenHeuveletal.,2019).However,thesame studieshavealsoreportedregion-specificdivergencesrevealing human-specificfeaturespotentiallyrelatedtohuman-specificdisorders. Conse-quently,comparingbrainorganizationbetweenhumansandnon-human primatescanelucidatedifferentiationinbrainorganizationpotentially rootedintheprocessofspeciesevolutionandenrichourunderstanding ofspecializationsinhumanbrainorganization(deSchottenetal.,2019;
Rilling,2014;VandenHeuveletal.,2016)[alsosee:Friedrich(2021)in thisissue].
The striatum is a crucialcomponent of the basal ganglia which worksinconcertwiththecerebralcortextoplanandexecute behav-iors(Haberetal.,2006; HaberandKnutson,2010; Marquandetal., 2017;Smeetsetal.,2000).Throughdiverseafferentprojectionsfrom thecerebralcortex,thestriatum is embeddedin multiple basal gan-gliacircuitsandmediatesmotivationsandemotionsthat drive plan-ning,cognitionthatgenerateappropriatestrategy,andaction execu-tion(Haber, 2003;Nakano etal., 2000).Theclassicalunderstanding of functionaland structural organizationof the striatum is primar-ilyderived from anatomicalandphysiological findings in macaques (AlexanderandCrutcher,1990;Alexanderetal.,1986;Künzle,1975;
SelemonandGoldman-Rakic,1985).Theinternal capsule was artifi-ciallyconsidered asa functionalandstructural boundaryseparating thestriatum intocaudate andputamen withthecaudate considered tobe primarily involved in cognition while the putamen in motor-relatedfunctions.However,functionalandstructuralcomplexityofthe striatumbased oncortico-striatalcircuitsgoes beyondthissimplistic demarcationasdemonstratedbyseveralstudiesinmacaquemonkeys (Calzavaraetal.,2007;Ferryetal.,2000;YeterianandPandya,1991). Forinstance,theheadandthetailcircuitsofthecaudateareinvolvedin short-termandlong-termvaluation,respectively(Hikosakaetal.,2014). Previousstudiesinmacaqueshaveshownthatthroughintegrating var-iousinformationwithin thecortico-striatal circuits,boththecaudate andtheputamenparticipatedirectly in rewardguidedbehaviorand learning(Apicellaetal., 2011;Hassani etal.,2001; Hikosakaetal., 2014,1989;Histedetal.,2009;Watanabeetal.,2003)andthe cortico-striatalrewardcircuitrystructureandfunctionisconservedacross hu-mansandprimates (Haber andKnutson,2010).In primates, projec-tions from anteromedial prefrontal cortex to dorsoanterior striatum mediates learningprocesses relatedto reward-related actions,while projectionsfromsensorimotorcortextodorsoposteriorstriatum medi-atedprocessesrelatedtoacquisitionofhabits(Balleineetal.,2007). Theseprojectionsroutetheinformationflowthroughsubstantianigra parsreticularis(SNr)andglobuspallidusinternal segment(GPi)and backtothecerebralcortex.Thispriorknowledgefromnon-human pri-matesandsubsequentin-vivoneuroimagingfindings in humans sug-geststhatthehumanstriatumispotentiallydividedintoseveral struc-turalandfunctionalsubregions(i.e.parcels)basedontheirinvolvement inmultiplecortico-striatalcircuits(Choietal.,2012;Lehetal.,2007;
Tziortzietal.,2014).
Furthermore,recentstudieshaveprovidedevidencefora correspon-dencebetweencortico-striatalfunctionalnetworksandtheirgene ex-pressionprofiles(Andersonetal.,2018)whichcombinedwiththelack ofhuman-specifictranscriptionalsignatureofneotenyinthestriatum (Bakkenetal.,2016)suggestastrongevolutionarilyconservedgenetic makeupofthehumanstriatum.However,recentstudieshave identi-fiedasubstantialnumberofdifferentiallyexpressedgenesprimarily re-latedtodopaminebiosynthesisinthehumanstriatumcomparedtoother species(Raghantietal.,2016;Sousaetal.,2017).Theseevidences fur-thersupportourinvestigationinsharedandhuman-specificfunctional organizationofthestriatum.
Due to its key roles in cognitive, emotional, executive and mo-tor functions,thehumanstriatumhasbeen implicatedinthe patho-physiologyofseveraldiseases.Amongthem,Parkinson’sdisease(PD) (Owenetal.,1992;Zhaietal.,2018)andschizophrenia(SCZ)(Lietal., 2020;Simpsonetal.,2010)aretwomajorsocio-economicallyrelevant disorderswithaclearlinktodopaminergicdeficitswithinthestriatum. Striatalfunctionalandstructuralabnormalitieshavebeenfoundin pa-tients comparedtohealthycontrols (HC)(Heetal.,2019; Xuetal., 2016).Generally,animal modelsprovide relevantinsightinto poten-tialpathophysiologicalmechanisms,optionsformedicaltreatmentand clinicalapplicationsofsuchexperimentsfortheseneuropsychiatric dis-eases (ChoudhuryandDaadi,2018; Qiuetal.,2019).Especially, ro-dentandmacaquemodelsarewidelyusedtoinvestigatethesedisorders (Castneretal.,2004;CenciandCrossman,2018),however,animal mod-elsareinsufficientiftheydivergefromhumanbrainorganization. Fur-thermore,manyneuropsychiatricdiseasesprimarilyaffectinghumans mightbeinfluencedbyvariousfactorsspecifictohumanslike, person-ality,familyandsocialenvironmentwhichinturnareassociatedwith brainorganization.Itisnotpossibletotransfersuchcomplex character-isticsandtheirinteractionswithbrainorganizationtoanimalmodels. Hence,itisnecessarytoinvestigatewhetherfunctionalandstructural abnormalitiesofthestriatalorganizationspecifictohumansarerelated toneuropsychiatricdiseases.
Recentadvancesinnon-invasivein-vivoneuroimagingtechniquesin humans andnon-humanprimate makesdirectcross-species compari-sonpossible(deSchottenetal.,2019)(alsoseeInthisissue:Friedrich etal.).Comparisonofthestructuralorganizationofthecortico-striatal circuitsbetween humanandmacaque striatumhavebeen conducted usingprobabilisticdiffusiontractography(PDT)basedondiffusionMRI (Neggersetal.,2015;Xiaetal.,2019).Neggersetal.(2015)compared connectionsbetweencorticalmotorareasandthestriatumin human andmacaqueusingPDT.Theyfoundthatthefrontaleyefields (FEF) connectedwiththeheadofthecaudateandanteriorputamen,andthe primarymotorcortex(M1)connectedwithmoreposteriorpartsofthe caudateandputameninmacaque.However,inhuman,theconnectivity ofFEFandM1islargelywiththeposteriorputamenandtoasmaller degree withthecaudate.Xia etal.(2019)alsousedPDT toidentify theventralstriatuminhumansandmacaquesbasedontheir cortico-striatalstructuralconnectivity,andthenexaminedinterspecies differ-encesinthestructuralconnectivityfingerprintsofthisregion.These re-sultsshowthatthestructuralconnectivityforsubregionsoftheventral striatummightbedissimilarbetweenhumansandmacaques.
Althoughpreviousstudieshaveprovidedrichanddiverse informa-tionaboutcortico-striatalstructuralconnectivity(Neggersetal.,2015;
Xiaetal.,2019),littleisknownaboutthefunctionalconnectivity(FC)of themacaquestriatumandwhetheritdiffersfromhumans.FCuses cor-relationoftimeseriesfrombloodoxygenationleveldependent(BOLD) functionalMRIsignals,whichreflectsthetemporalsynchronyof neu-ronalactivationpatternsbetweenbrainregions[forreview,seeVanDen HeuvelandPol(2010)]includingwhentheFCismeasuredatrest—i.e. resting-stateFC(RSFC)(Biswaletal.,2010).RSFChasbeenwidelyused inhumanresearchtoinvestigateintrinsicneuronalactivationpatternof thestriatumtorevealitsfunctionalorganization(Barnesetal.,2010;
Janssenetal., 2015;Jaspersetal.,2017; Jungetal., 2014).For in-stance,Jungetal.(2014)examinedtheparcellationofthehuman stria-tum basedonits RSFCtothewholebrain. Primarily,thecaudateis dividedintothree subregionsalong theanterior-ventralaxisandthe posterior-dorsalaxis.Thespatialpatternofthesethreeclusters corre-spondtothehead,body,andtailofthecaudate.Similarly,theputamen wassplitintothreesubregionsbutalongananterior-posterioraxis.In ourownwork(Liuetal.,2020),weidentifiedjointmulti-modal parcel-lationofthestriatumwhichincludedRSFCasonemodality.Wefound thatthestriatumwassplitintosubregionsalongtherostro-caudaland ventro-dorsalaxesfromcoarse(k=3)tofine-grained(k=9) parcella-tionsbasedonitsintrinsicRSFC.However,similaritiesanddifferencesin thefunctionalparcellationofthestriatumbetweenhumanandmacaque
areunclear.Suchacross-speciesanalysiscanshedlightonsimilarities ofhumanbrainorganizationwithourphylogeneticallycloserelatives, and,moreimportantly,revealorganizationspecifictohumans.In ad-dition,itremainsanopenquestionwhetherthehuman-specificstriatal organizationisinvolvedincomplexneuropsychiatric disorderswhich areoftenspecifictohumans.
In this study, we capitalize on in-vivo neuroimaging data to di-rectlycomparethefunctionalorganizationofthestriatumbetween hu-mansandmacaques,andsubsequentlyinvestigatedcortico-striatalRSFC andstructuralalterationinthestriatalsubregionsintwo neuropsychi-atricdiseases(PD,SCZ). Thehumanneuroimaging datafor parcella-tionwasassessedfromtheHumanConnectomeProject(HCP),while macaqueneuroimagingdatawasassessedfromthePRIMatEData Ex-change(PRIME-DE).Thereis agrowinginterestin comparativeMRI investigationoffunctionalandstructuralorganizationofthebrain be-tweenhumansandmacaques(Balstersetal.,2020;Marsetal.,2018;
Vanduffeletal.,2014;Xuetal.,2019).However,mostpreviousstudies onlyrecruitedverysmallsamples(n<10)ofnon-humanprimate sub-jects.Althoughtechnologicalandmethodologicaladvanceshave pro-motedthedevelopmentofnon-humanprimateresearchduringthepast two decades,neuroimaging data collectionremained limiteddue to lackofnecessaryfacilitiesandcapabilities.PRIME-DEaddressedthese challengesbyaggregatingindependentlyacquirednon-humanprimate MRIdatasetsandopenlysharingthemthroughtheInternational Neu-roimagingData-sharingInitiative(INDI)(Milhametal.,2020,2018). Benefitingfrom these two opendatasets, we directly compare func-tionalaspectsofbrainorganizationbetweenhumanandmacaque.To comparefunctionalorganizationofhumanandmacaquestriatum,we firstemployedconnectivity-basedparcellation(CBP)[forreview,see
Eickhoff etal.(2015),Eickhoff etal.(2018)]thatcanidentify subre-gionsbasedontheirsimilaritiesinconnectivity,suchasRSFCandPDT (Genonetal.,2018;Plachtietal., 2019a).WeusedRSFCas connec-tivitymeasuretoseparatelyperformCBPinhumanandmacaque,and thencomparedtheresultingparcelsbasedontheirconnectivitytoaset ofhomologouscerebralregions.ThePearsoncorrelationdistancewas usedtoestimatethedissimilarityofconnectivityfingerprintsbetween humanandmacaquestriatalsubregions.Finally,wecompared differ-encein cortico-striatalRSFCandinacross-modal analysisalsoused voxel-basedmorphometry (VBM)toinvestigatefunctionaland struc-turalalterationsinthehumanstriatalsubregionsbetweenpatients(PD, SCZ)andhealthycontrols(HC).
2. Methods
Briefly,wefollowedatwo-stepprocedure(seeFig.1).Inthefirst stepweperformedCBPbasedontheRSFCofthestriatalvoxelswith thewhole-braingraymattervoxelstouncoverthefunctional organiza-tionofthehumanandmacaquestriatumseparatelyandchose parcella-tionschemesbasedondata-drivenmodelselection.Inthesecondstep, weperformedcross-speciescomparisonofthestriatalclustersfromthe firststepbasedontheirconnectivitywithasetofhomologouscortical regions.Thestriatalclustersweretheninvestigatedfordifferences be-tweenpatients(PDandSCZ)andHCintheircortico-striatalRSFCand structuralatrophybasedongraymatter(GM)volume.
2.1. Dataset
ForCBP,both humanandmacaque resting-statefunctional mag-netic resonance imaging (fMRI) datasets were obtained from open access sources. We used the Human Connectome Project (HCP) young adult sample (https://www.humanconnectome.org), assessing 324 unrelated subjects (male/female: 158/166, age range: 22–37 years). We also selected another dataset with 206 unrelated sub-jects (male/female: 100/106, age range: 22–36 years) from HCP in order to retest functional parcellation results (Supplementary Material). The macaque monkey dataset was obtained from the
recently established PRIMatE Data Exchange (PRIME-DE) project (http://fcon_1000.projects.nitrc.org/indi/indiPRIME.html). PRIME-DE currently contains 219 macaquemonkeys from25 institutes. Inthis study,weselected56macaquesubjectsfromfourinstitutes;University ofOxford(male/female:20/0,agerange:2.41–6.72years),Instituteof Neuroscience,China(male/female:7/1,age range:3.80–5.99years), NewcastleUniversity(male/female:12/2,agerange:3.90–13.14years) and University of California, Davis (male/female: 0/19, age range: 18.50–22.50years).Weonlyusedthesefourmacaquedatasetsbecause theyproviderelativelylargeresting-statefMRIdatasamplesizes.Details ofscanningandimagingofHCPandPRIME-DEdatasetsarelistedin Ta-bleS1andTableS2.GivenallmacaquedatacontributedtoPRIME-DE projectweremadeavailableregardlessofdataquality,weperformed qualitycontrolandexcluded7macaquesubjectsbeforedataanalysis (SupplementaryMaterials).
Forthe clinical datasets,we collected resting-statefMRIand T1-weighted images of Parkinson’s disease (PD) patients from Hein-rich Heine University Düsseldorf and RWTH Aachen University (Pläschke et al., 2017). Together, these two datasets included 101 patients (female: 47, age: 63.09 ± 10.06) and 96 healthy controls (HC, female:45,age:58.87± 9.81).Wecollected resting-statefMRI of schizophrenia (SCZ) patients and controls from RWTH Aachen University, CenterforBiomedical ResearchExcellence(Mayer etal., 2013), the University of Groningen (Chen et al., 2020a, 2020b;
Vercammenetal.,2010),theUniversityof Göttingen,theUniversity of Lille (Lefebvre et al., 2016), andUtrecht University(Clos et al., 2014).ThepooledSCZdatasetincluded142patients(female:41,age: 34.94±11.72)and136HC(female:40,age:33.82±11.11).Wealso collectedT1-weightedstructuralMRIdataofSCZpatientsfromthe Cen-terforBiomedicalResearchExcellence,theUniversityofGroningen,the UniversityofLille,theTechnicalUniversityofMunichandUtrecht Uni-versity.These datasettotallyincluded159patients(female: 54,age: 35.92±12.08)and166HC(female:64,age:34.32±11.94).No sig-nificantsexdifferencewasobservedbetweenpatientsandHC[𝜒2-test:
p=0.96forPDandHC,p=0.92forSCZandHC(resting-statefMRI datasets),p=0.39forSCZandHC(T1-weightedimagesdatasets)].A significantdifferencewasfoundinagebetweenPDandHC(two-sample t-testp<0.01),butnotbetweenSCZandHC(p=0.41forresting-state fMRIdatasets,p=0.23forT1-weightedimagesdatasets). Additional informationonMRscanningparameterforabovedatasetscanbefound inSupplementaryMaterials.
ForPRIME-DE,allexperimentalprocedureswereapprovedbylocal ethicsboardspriortoanydatacollection.UKmacaquedatasetswere obtainedwithHomeOfficeapprovalandabidewiththeEuropean Di-rectiveon theprotectionof animalsusedin research(2010/63/EU). FortheNINPrimateBrainBank/UtrechtUniversitydataset,postmortem specimenswereloanedfromtheNetherlandsInstituteofNeuroscience PrimateBrainBank(PBB;http://www.primatebrainbank.org/).No in-dividualsweresacrificedforPBBbrainissue.Instead,brainswere col-lectedfromindividualsthatdiedfromnaturalcausesorthathadtobe humanelyeuthanizedforreasonsunrelatedtothetissuecollection.
Theethicsprotocolsforanalyses ofthese datawereapprovedby theHeinrichHeineUniversityDüsseldorfethicscommittee(No.4039, 4096).
2.2. Datapreprocessing
Human resting-state fMRI data preprocessing was performed us-ing SPM12 (Wellcome Trust center for Neuroimaging, London,
http://www.fil.ion.ucl.ac.uk/spm/software/spm12).Thepreprocessing was consistentacross thehumandatasets.FortheHCP weused the minimallypreprocessedvolumetricdataintheMontrealNeurological Institute(MNI)space(Glasseretal.,2013;VanEssenetal.,2012).We performedadditionalstepsonHCPdataandtheotherhumandatasets werepreprocessedtooptimallyalignfMRItime-seriesinthecommon MNIspaceandremovemotionartifactsandnuisancesignals.Foreach
Fig. 1. A sketch of our proposed pipeline depicting its use on the left striatum. (A) Connectivity-basedparcellation(CBP)of hu-manandmacaquestriatumbasedontheRSFC between the striatal voxelsand whole-brain graymattervoxels.Eachsubject’sRSFC con-nectivitymatrixisthensubjectedtok-means clusteringwhile varying numberof clusters. Thefinalnumberofclusterswasselected us-ingseveraldata-drivenmodelselection crite-ria.Thesubject-levelclusterswherethen ag-gregated into group-level clustering for hu-mans and macaques separately. (B) Cross-speciescortico-striatalconnectivity:TheRSFC betweenstriatalvoxelsandregionsofregional map(RM)wascalculatedforeachsubjectand thenaveraged.Thedistancebetweeneach hu-manstriatumvoxelandeachmacaquestriatum voxel was calculatedusing the Pearson cor-relation(1-R)betweenthecorrespondingRM connectivityprofiles.(C)Comparisonof CBP-derivedparcelscross-species.Thepermutation basedZ-scoreswerecalculatedforcomparing the cortico-striatalRSFC of the CBP-derived parcels(shownherefork=6).
subject,weexcludedthefirstfourechoplanarimaging(EPI)volumesto allowtheMRIsignaltoreachsteadystatefollowedbyrealignmenttothe firstandsuccessivelytothemeanimage.Next,themeanEPIwas coreg-isteredtotheGMprobabilitymapandnormalizedtoMNIspaceusing theunifiedsegmentationalgorithm(AshburnerandFriston,2005). Sub-sequently,weappliedthisnon-linearlytransformationtoallEPIimages beforesmoothingwithakernelof5mmfullwidthathalfmaximum (FHWM).Finally,weperformed timeseriesdenoisingusingmultiple regressionofmeanwhitematter(WM)andcerebrospinalfluid(CSF) signals,and24motionparameters(Satterthwaiteetal.,2013)before band-passfiltering(0.01–0.08Hz)theresiduals.Notethatwedidnot usetheindividualT1imagesfornormalizationastheEPIbased align-menthasbeenshowntoprovide goodresults(Calhounetal., 2017;
Dohmatobetal.,2018),andthesegmentationofthemeanEPIswith unifiedsegmentationusestissueprobabilitymapsaspriorshelping de-lineationinsub-corticalregions.
Macaque resting-state fMRI data preprocessing was per-formed using SPM12 and the FMRIB Software Library (FSL,
https://fsl.fmrib.ox.ac.uk/fsl/fslwiki). The first four volumes of theEPIimageswerediscarded.AfterbrainextractionusingFSLBET,all imageswereheadmotioncorrectedbyaligningtothefirstEPIimage. Next,T1imagewasregisteredtoYerkes19template,andthemeanEPI imagewascoregisteredtoT1image.TheEPIimageswerenon-linearly normalizedtoYerkes19 template with voxel sizeof 1× 1× 1𝑚𝑚3 by
usingthetransformationparametersfromlaststepinFSL.ThenallEPI imagesweresmoothedwithaFHWMof3mm.NextweperformedWM, CSFsignalandheadmotionregressionaswellasband-pass filtering similartothehumandata.
WealsotestedtherobustnessofourhumanCBPresultsusingthe FMRIB’sICA-basedXnoiseifier(FIX)basedartifactremovalontheHCP data.Forallhumanandmacaquesubjects,wefurthercalculated
voxel-wisetemporalsignal-to-noise(tSNR)forallthevoxelsinthestriatum (SupplementaryMaterials).
2.3. Regionofinterest(ROI)definition
Humanstriatum:TheregionofInterest(ROI)forthehumanleftand therightstriatumwereextractedusingtheHarvard-Oxfordsubcortical structuralprobabilityatlasavailableviaFSL.Weextractedthecaudate andputamenwithavoxelsizeof2mmx2mm x2mmbasedona probabilitythresholdof25%,andthencombinedthesestructuresinto onehumanstriatalROIforeachhemisphere.Thisprocedureresultedin aleftstriatumROIwith1286voxels(caudate:487,putamen:799)and arightstriatumROIcomprising1307voxels(caudate:511,putamen: 796).
Macaquestriatum:TheROIforthemacaqueleftandrightstriatum wereextractedusingtheINIA19template(Rohlfingetal.,2012).We ex-tractedbothcaudateandputamenandresampledthemtotheYerkes19 spacewithavoxelsizeof1mmx1mmx1mm.Wealsocombined thecaudateandputamenintoonemacaquestriatalROIforeach hemi-sphere.Thenumbersofvoxelswas1457(caudate:551,putamen:906) intheleftand1445(caudate:550,putamen:895)intherightstriatum ofthemacaques.Differentvoxelsizesresultedinasimilarresolution foreachspeciesasreflectedinthesimilarnumberofstriatalvoxelsin humanandmacaqueROIs.
2.4. Connectivity-basedparcellationusingfunctionalconnectivity 2.4.1. Whole-brainresting-statefunctionalconnectivity
We estimatedtheresting-statefunctionalconnectivity (RSFC) be-tweenthestriatalvoxelsandthewhole-brainvoxels.Tothisend,we usedpreprocessedresting-statefMRIdatatocalculatethePearson
cor-relationbetweenthetimeseriesofeachvoxelwithinthestriatumandall otherGMvoxelsforeachhumanandmacaquesubject.Thecorrelation coefficientswerethenFisher-Ztransformed.Oneresting-statefunctional connectivitymatrixwascalculatedperindividual.
2.4.2. Clusteringalgorithm
Inlinewithpreviousconnectivity-basedparcellation(CBP)studies (Crippaetal.,2011;Genonetal.,2018;Jungetal.,2014;Plachtietal., 2019a; Xu etal., 2020a),voxelswithin a ROI aregroupedinto dis-tinct clusters(i.e., subregions) through a clusteringalgorithm based ontheirsimilarityinRSFCpatterns.Generally,k-meansclustering di-videsagivenROIintoapreselectednumberofknon-overlapping clus-ters (Nanetti et al.,2009). Thek-means algorithm is knownto pro-videaccurateparcellationresultscomparedtootherclusteringmethods (Thirionetal.,2014).Inthisstudy,weappliedthek-meansclusteringas implementedintheyaelpackage(https://gforge.inria.fr/projects/yael) ontheindividualRSFCmatrix.Thenumbersofpotentialsubdivisions from2to7clusterswereinvestigated.Basedonreportedresultsinthe literature,weassumedthatameaningfulorganizationofthestriatum canbeobservedatlowandmediumresolutionbutnotatveryhigh res-olution(i.e.notinmorethan7subdivisions).Foreachk-meansrun, thebestsolutionbasedonthesumofsquaresfrom100initializations witharandomlyplacedinitialclustercenterswereused.Importantly, foreachsolution,k-meansclusteringwasperformedattheindividual level.Resultingindividual-levelclustersolutionswerethencombined intoasinglegroup-levelparcellationbycomputingthemostfrequent clusterassignmentforeachstriatalvoxelacrossallhumanormacaque subjectsseparately.Notethatweusethetermsclusterandparcel inter-changeably.
Thegroup-levelparcellationfromtheindividual-levelclusterswere calculatedasfollows.First,aconsensusclusteringoftheindividual-level clusteringsolutionsiscalculatedusingthehierarchicalclustering algo-rithmwiththeHammingdistancetoaccountforthearbitrarinessofthe clusteridsacrosssolutionsandwiththenumberofclustersequaltothe numberofclustersintheindividualsolutionsunderconsideration.Each individual-levelsolutionisthenmatchedwiththehierarchical cluster-ingusingapermutationthatmaximizesthematchoftheclusterids.The finalgroup-levelsolutionisthencalculatedasthemodeofthealigned individual-levelclusteringsolutions.Thisprocedureidentifiesa cluster-ingthatisrepresentativeofthewholegroup.
2.4.3. Clusterselectioncriteria
As clustering is an unsupervised process, it is difficult to know which model selectioncriteria to use (Friedman et al., 2001). We, therefore,selectedtheclustersolutionsbasedonfivedifferentcriteria (Eickhoff etal.,2015):twotopologicalcriteria(percentageof misclas-sifiedvoxels,andhierarchyindex),aninformation-theoreticcriterion (variationofinformationacrossclustersolutions)andtwocluster sepa-rationcriteria(changeininter/intraclusterdistanceandthesilhouette index).Detailedinformationabouteachcriterionandtheselection pro-cedurecanbefoundintheSupplementaryMaterials.
2.5. Cross-speciescomparison
2.5.1. Homologouscortico-striatalRSFC
Wecalculatedcortico-striatalRSFCofhumansandmacaquesbased onthecorticalROIsselectedfromtheRegionalMapparcellation(RM) definedbyKötterandWanke(2005). TheRMis basedon cytoarchi-tectonic, grossanatomical,andfunctionalcriteria,minimizing cross-speciesdiscrepanciesinontology.Reidetal.(2016)compared diffusion-basedstructuralconnectivitystrengthinhumanswithneuronal tracer-basedstructuralconnectivity strengthin macaquesbasedon theRM showingamoderatelyhighcorrespondence.Goulasetal.(2014) em-ployed the RM to compare the structural connectivity between macaquesandhumansanddemonstratedagoodoverallcorrespondence providingadditionalvalidationoftheRMforcross-speciescomparison.
TheRMthusprovidesagoodwayforcross-speciescomparisonbetween humansandmacaquemonkey.TheRMparcellationcontains82cortical regions(seeFig.S1andTableS3).Foreachhumanandmacaque sub-ject,wecalculatedthePearsoncorrelationbetweentheaveragedtime seriesacrossallvoxelswithineachcorticalRMROI andthetime se-riesofeachstriatalvoxeltocreateacortico-striatalRSFCmatrix.Each rowofthismatrixrepresentstheconnectivitypatternofastriatalvoxel witheachofthe82RMROIs.Weaveragedtheconnectivitymatrices acrosshumansubjectsandacrossmacaquesubjectstogenerate group-representativehomologouscortico-striatalRSFCmatricesseparatelyfor bothspeciesandbothhemispheres(Fig.4).
2.5.2. Cluster-basedcross-speciescomparison
AstheRMregionsareconsideredhomologous,wecouldestimate thedissimilaritybetween ahumanstriatalvoxelandamacaque stri-atalvoxelusingthePearsoncorrelationdistance(1−𝑟)betweentheir connectivitypatternwiththeRMregions.Bycalculatingdistances be-tweenallpairsofhumanandmacaquestriatalvoxelsweobtainedtwo cross-speciesdistancematrices(D),onefortheleftandonefortheright striatum.Inthismatrixanelement𝐷𝑖𝑗 quantifiesthedissimilarity
be-tweenthecortico-striatalRSFCpatternofahumanstriatalvoxeliwith that ofa macaquestriatalvoxel j. Asouraimis tocomparehuman andmacaquestriatalclustersobtainedbyfunctionalCBP,weuseda voxel-wisedistancematrixDtogetherwiththeclusterlabelsfromthe clusteringresultstoestimatethedistancebetweenallhuman-macaque parcelpairs.Thedistancebetweenahuman-macaqueparcelpairH-M wascalculatedastheaverageofallhumanandmacaquestriatal voxel-wisedistancesassignedthehumanclusterHandthemacaqueclusterM, i.e.𝑀𝑒𝑎𝑛𝑡𝑟𝑢𝑒(𝐻−𝑀)=∑𝑖∈𝐻,𝑗∈𝑀𝐷𝑖𝑗∕|𝐻| × |𝑀|.Ahuman-macaque
par-celpaircanbe deemedtohaveasignificantlysimilarcortico-striatal RSFCpatternifthedistancebetweenthemislessthanwhatcanbe ex-pectedbychance.Toachievethis,weusedapermutationtestinwhich theclusterlabelsassignedtothevoxelswerepermuted.Asahuman parcelcanbe similartomultiplemacaqueparcelsandviceversa,we performedpermutationsintwoways.First,wetestwhetherahuman parcelissimilartorandomlygeneratedmacaqueparcelbyshufflingthe clusterassignmentofmacaquestriatalvoxels(thecolumnsofD)and cal-culatingthedistancebetweenahumanparcelandtheshuffledmacaque parcel.Byrepeatingthispermutation5000timesweobtainanempirical distribution[Mean1, 2,3… 5000(H−M)].Themeanandstandard
devia-tionofthisempiricaldistributionisthenusedtocalculatetheZ-score ofthetruedistancebetweenparcelsHandM:
𝑍𝐻−𝑀
𝐷 =
𝑀𝑒𝑎𝑛𝑡𝑟𝑢𝑒(𝐻−𝑀)−𝑀𝑒𝑎𝑛𝑝𝑒𝑟𝑚𝑢𝑡𝑒𝑑(𝐻−𝑀)
𝑆𝑡𝑑𝑝𝑒𝑟𝑚𝑢𝑡𝑒𝑑(𝐻−𝑀)
A lower valueof 𝑍𝐻−𝑀
𝐷 reflects highersimilarity of the
cortico-striatalRSFCbetweenahumanparcelHandamacaqueparcelM com-paredtorandomlyselectedmacaquestriatalvoxels.Inthesameway, wecalculate𝑍𝑀−𝐻
𝐷 byshufflingtheclusterassignmentsofthestriatal
humanvoxelswhichreflectsrelativesimilarityoftheclusterpairwith respecttorandomlyselectedhumanstriatalvoxels:
𝑍𝑀−𝐻
𝐷 =
𝑀𝑒𝑎𝑛𝑡𝑟𝑢𝑒(𝑀−𝐻)−𝑀𝑒𝑎𝑛𝑝𝑒𝑟𝑚𝑢𝑡𝑒𝑑(𝑀−𝐻)
𝑆𝑡𝑑𝑝𝑒𝑟𝑚𝑢𝑡𝑒𝑑(𝑀−𝐻)
WeemployedastrictcriterionthatbothZ-scoresmustbebelowa thresholdvaluetodeclarethecorrespondingclusterpairtohave signif-icantlysimilarcortico-striatalRSFC.Wesetthesignificancethreshold at2standarddeviations(2SD=−1.96).
In addition,we also visually checked spatialcorrespondence be-tweentheclusterpairs(i.e.withinthewholestriatum)andifahuman striatalclusterwasnotdeemedsimilartoanyspatiallycorresponding macaquestriatalclusterthenthisclusterwaslabeledasshowing dis-similarcortico-striatalRSFCbetweenhumanandmacaque—i.e.a clus-terwithhuman-specificcortico-striatalRSFC.Macaque-specificclusters wereidentifiedinasimilarway.
Fig.2. Connectivity-basedparcellation(CBP) ofleft(A)andright(B)humanandmacaque striatumatdifferentlevels(k)ofsubdivision.
2.5.3. Human-specificcortico-striatalRSFC
Wefurtherinvestigatedhowthecortico-striatal RSFCdiffered be-tweenhumans andmacaquesfor thehuman-specificstriatal clusters identifiedinourpreviousanalysis.Forthis,firsttheaveraged cortico-striatalRSFCmatricesbasedoneachstriatalvoxelandthe82cortical RMregionsforallhumanandallmacaquesubjectswerecalculated.The averagecortico-striatalRSFCacrossagivenhuman-specificcluster vox-elsandcorrespondingspatiallysimilarmacaque-specificclustervoxels wascalculated.Themacaque-specificconnectivitywassubtractedfrom thehuman-specificconnectivityandthedifferencewasZ-scoredfor vi-sualizationpurposes.Notethatahigherscorehereindicatesastronger connectivityinhumanwhilealowerscoreindicatesastronger connec-tivityinmacaque.
2.6. Cortico-striatalRSFCandvoxel-basedmorphometry(VBM)analysis indisease
To gain insights into which striatal clusters, and especially the human-specificstriatalclusters,arerelatedtoclinicallyrelevant func-tionalandstructuralalterationsinhumans,weinvestigateddifferences incortico-striatalRSFCandGMvolumeofstriatalclustersbetween pa-tients(PDandSCZ)andHC.
Resting-statefunctionalimages werepreprocessedwith thesame stepsasdescribedin“Datapreprocessing”.WecalculatedthePearson cor-relationbetweentheaveragedtimeseriesofthevoxelswithinagiven striatalclusterandaveragedtimeseriesofallvoxelswithineachcortical regionbasedonRM.Finally,weexamineddifferencesinRSFCofeach striatalclusterandeachRMregionbetweenpatientsandHCbyusing atwo-samplet-testwhilecontrollingfor“sex”,“age” and“sites”.The resultingp-valueswereFDRcorrected.Thisanalysiswasperformedfor leftandrightstriatalclustersseparately.
We alsoinvestigated differencesin averagedGM volume of stri-atal cluster between patients and HC. T1-weighted images were preprocessed using the Computational Anatomy Toolbox (CAT12,
http://www.neuro.uni-jena.de/cat/) in SPM12. All imageswere first segmentedintoGM,WM,andCSFusingthestandardunified segmenta-tion.WethenprocessedtheimagesusingthestandardsettingsinCAT12, includingDARTELnormalization,spatiallyadaptivenon-linearmeans denoising,aMarkovrandomfieldweightingof0.15,bias regulariza-tion(0.0001)andFWHMcutoff (60mm).TheresultingnormalizedGM segmentsweremodulatedonlyforthenon-linear componentsofthe deformation,whichmeansweonlyusedlocalandnon-linear deforma-tionstoadjusttheheadsizetoestimatetheGMvolume.Next,we
ex-tractedtheaverageGMvolumeforthehuman-specificstriatalclusters foreachsubjectandexaminediftheydifferbetweenpatientsandHC. Considering thatdifferencesin GMvolume mightbeassociatedwith sex,age,hemisphereandsites,weappliedasix-wayanalysisof vari-ance(ANOVA)thatincludednotonly“diseasestatus”,butalso“striatal clusters”,“sex”,“age”,“hemisphere” and“sites” asfactors.
3. Results
Wefirstinvestigatedthefunctionalparcellationofthehumanand macaquestriatumseparatelyfromlowtohighlevelsofsubdivisionand examinedhowthedifferentclustersolutionsweresupportedbythefive data-drivenmodelselectioncriteria.ForeachstriatumROI(humanand macaque,leftandrightside),weidentifiedtheappropriatecluster so-lutionthatwassupportedbythemajorityofthecriteria.
3.1. Humanstriatum
Wefoundthatthehumanstriatumwassplitintoclusters(i.e.parcels) alongthedorso-ventralandrostro-caudalaxisforbothleftandrightside fromlowtohighlevelsofsubdivision(i.e.k=2–7clusters,Fig.2).At k=2,thehumanstriatumwasdividedintocaudateandputamen.At k=3,theputamenwassubdividedintoarostralandcaudalcluster.At k=4,fortheleftside,theputamenwasfurthersplitintoventral, cen-tralandcaudalclusters.However,fortherightstriatum,thecaudate wassplitintorostralanddorsalcluster.Atk=5,theleftcaudatewas dividedintorostralanddorsalclusters,whichsimilartothatofright caudateatk=4.Inturn,therightputamenwassplitintothreeclusters, whichsimilartothatofleftputamenatk=4.Similarstriatalclusters werefoundbetweenleftandrightsideatk=6,includingrostral, dor-somedial,dorsalcaudateandventral,central,caudalputamen.Atthe highestlevelofsubdivision(k=7),fortheleftside,thecaudal puta-menwasdividedintoadorsalandaventralpart.However,fortheright side,wefoundanadditionalsmallclusterlocatedbetweenthecentral andventralputamen.
3.2. Macaquestriatum
Asexpected,atk=2,bothmacaqueleftandrightstriatumwere dividedintocaudateandputamensimilarlytowhatwefoundin the humanstriatum(Fig.2).Atk=3,thecaudatewasdividedintorostral anddorsalparts,whileatk=4,wefoundaventralclusterthatwas derivedfromtheputameninthe2-clustersolution.Atk=5,fortheleft
side,therostralcaudatefromthe3-clustersolutionwassplitintotwo clustersincludingthemedialandlateralparts.Forrightside,the ven-tralputamenobtainedfromk=4wasdividedintotwopartsalongthe dorso-ventralaxis.Atk=6,theputamenwasdividedintorostraland caudalclustersforleftside,whilewasdividedintorostrodorsal, rostro-ventralandcaudalclustersforrightside.Inaddition,fortheleftside, wefoundadorsomedialclusterthatderivedfromthedorsalcaudate.At thehighestlevelofsubdivision(k=7),thedivisionoftheputamenin theleftsidewassimilartothatoftherightsideat6-clustersolution.For therightside,wefoundanadditionalsmallclusterthatderivedfrom caudalputamen.
3.3. Selectionofclustersolutions
Wetheninvestigatedhowthesefunctionalparcellationsolutionsof thestriatumfromlowtohighlevelsofsubdivisionweresupportedby thedataitselfbasedonseveralclusterselectioncriteria.Fig.S4-S7and TableS5showtheresultsofclustersolutioncriteriaforbothhumanand macaquestriatum.
Humanstriatum(Fig.S4-S5):boththeHierarchyindexandvariation ofinformationacrossclusterscriteriasuggestedthe6-clustersolution overtheothersforleftside. TheHierarchyindexalsosupportedthe 3-clustersolutionforbothside,and6-clustersolutionsforrightside.
Macaquestriatum(Fig.S6-S7):boththepercentageofmisclassified voxelsandsilhouettevaluecriteriasupportedthe6-clustersolutionfor bothside.TheHierarchyindexcriterionsuggestedk=3forbothside, whilethevariationofinformationcriterionsupporteda5-cluster solu-tionfortheleftside,and3-clustersolutionfortherightside.
Mostcriteriasupported3and6clustersolutionswhichcanbe re-gardedasthestablesolutionsinbothleftandrightstriatum inboth humanandmacaqueCBPanalysis.Thus,forsubsequentanalyseswe fo-cusedontheseparcelsandcomparedtheircortico-striatalRSFCacross humanandmacaque.
3.4. Threeandsixparcelsofhumanandmacaquestriatum
Fig.3showsdetailedlocationinformationofhumanandmacaque striatalparcelswith3and6clusters.Fork=3, thehumanleftand rightstriatumweresubdividedintocaudate,rostralandcaudal puta-men(Fig.3A).Inmacaque,leftandrightstriatumweresplitinto puta-men,rostralanddorsalcaudateatthissolution(Fig.3B).Fork=6, theparcellationof thehumanleftandrightstriatumwerealso simi-lar,whichincludedorsal,dorsomedialandrostralcaudateandventral, central,andcaudalputamen(Fig.3A).Inmacaques,theparcellationof theleftandrightstriatumwereslightlydifferentforthe6-cluster solu-tion(Fig.3B).Wefounddorsal,rostrodorsal,rostroventralcaudateand caudalputamenforbothleftandrightside.Theydifferedonlyinthat theleftsidehadanadditionaldorsomedialcaudate,whiletherightside hadanadditionalrostroventralputamen.Thedetailsforthebetween subjectandindividual-levelandgroup-levelparcelsareprovidedinthe SupplementaryMaterials(Fig.S8).
3.5. CBProbustnesschecks
Weperformedseveralcontrolanalysistochecktherobustnessofour clusteringsolutions.First,weperformedsplit-halfanalysistocheckthe effectofsampleselectionandfoundthegroup-levelclusteringsolutions tobestableacross1000randomsplitsofthedata.Wethenreanalyzed theHCPdatabutthistimeafterapplyingtheFIX-baseddenoisingwhich providesanalternatewaytoremovemotionartifacts.Wefoundahigh levelofmatchusingtheadjustedrandindex(ARI)betweenthereported parcelsandtheFIX-denoisedparcels;0.86(k=3)and0.69(k=6)for thehumanleftstriatumand0.88(k=3)and0.75(k=6)forthehuman rightstriatumfortheselectedclusteringsolutions.Wethenperformed tworeplicationanalyses.Theseresultswerehighlysimilarwiththe solu-tionsobtainedinourmainanalysis[alladjustedrandindex(ARI)above
0.58].ThesecondreplicationwasperformedonaseparateHCPsample whichshowedallARIabove0.56(TableS4).Tofurthervalidateour hu-manstriatumparcellation,weestimatedtheRSFCbetweenthestriatal clusters(atk=6)andsevencorticalnetworks(Yeoetal.,2011)based onaveragedRSFCacrossallsubjects.WeobservedthattheRSFC pat-ternsbetweenourstriatalclustersandthesevencorticalnetworkswas similartothatofChoietal.(2012)(Fig.S10)showingthatourRSFC basedfunctionalparcellationisinlinewithapreviouslarge-scalestudy. DetailsofthesecontrolanalysesareprovidedintheSupplementary Ma-terials.
Wealsoperformedrobustnesschecksforthemacaqueclustering so-lutions.Specifically,weperformedsplit-halfanalysis,andcomparedthe mainsolutions(k=3and6)withthatobtainedusingonlyanesthetized subjects(allARIabove0.9)andwithahold-outsampleof12subjects (allARIabove0.3).Severaladditionaltestswereperformedtocheckthe validityofthehumanandmacaqueclusteringsolutionsincludinga dif-ferentwaytoobtainthegroup-levelparcellation,permutationtestsand removalofbordervoxels(seeSupplementarymaterialsandTableS6). Wefurthertestedthesensitivityofthegroup-levelclusteringtosample sizeandourresultsshowthatlargersamples,asusedhere,providea bettersignalresultinginmorestablegroup-levelparcellation(Fig.S9). These analysesindicatedthatourclusteringsolutions arerobust(see Supplementarymaterials).
3.6. Cross-speciescomparison
Weusedtheregionalmap(RM)-derived82homologouscortical re-gionstocalculatecortico-striatalRSFCcomparableacrosshumanand macaque. Fig.4A showscortico-striatal RSFCbetween each left and rightstriatalvoxelandcorticalregionsseparatelyaveragedacross hu-manormacaquesubjects.WeadoptedthePearsoncorrelationdistance metrictoestimatethedissimilarityofconnectivitybetweenanygiven humanandmacaquestriatalvoxels(Fig.4B1leftstriatum,Fig.4B2right striatum).
3.6.1. Cluster-basedcross-speciescomparison
We calculated permutation-based Z-scores to assess cross-species similarityofthefunctionalparcels.Notethattheweobtainedtwo Z-scores,oneafterpermutingmacaqueclusterassignmentsandtheother afterpermutinghumanclusterassignments,andahuman-macaque clus-ter pair was deemed toshow significant similarityin their cortico-striatalRSFConlyifbothZ-scoreswerebelowthesignificance thresh-old(seeMethods).Wefocusedonthe3-and6-clustersolutionsasthese weresupportedbyvariousclusterselectioncriteria.Forthe3-cluster so-lution,wefoundsignificantlysimilarcortico-striatalRSFCbetween hu-manleftandrightcaudalputamenandmacaqueleftandrightputamen, aswellasbetweenhumanleftandrightrostralputamenandmacaque leftandrightrostralcaudate(Fig.5A).Forthe6-clustersolution, signif-icantlysimilarcortico-striatalRSFCwasfoundbetweenthehumanand macaqueleftcaudalputamen(Fig.5B).Meanwhile,thecortico-striatal RSFCofhumanleftventralandcentralputamenweresignificantly sim-ilar tothatof macaque left rostral,dorsomedialcaudate androstral putamen.Wealsoobservedsignificantsimilarityincortico-striatalRSFC betweenhumanleftrostralcaudateandmacaqueleftrostrolateral cau-date.Fortherightstriatum,similarresultswereobservedforhumanand macaquecaudalputamen.Thecortico-striatalRSFCofhumanright ven-tralandcentralputamenweresignificantlysimilartothatofmacaque rostroventralcaudateandrostralputamen.Inaddition,wealsofound thehumanrightdorsomedialandrostralcaudatehavesignificant sim-ilarcortico-striatalRSFCwithmacaquerightrostralcaudate.In sum-mary,wefoundsimilarcortico-striatalRSFCindorsomedial,rostral cau-dateandputamenbetweenhumansandmacaquesat3-and6-cluster solutions.
Interestingly,wefoundthatthedorsalcaudateclusterinhumansdid notmatchwithanyofthemacaqueclusters,especiallynotwithspatially
Fig.3. Thelocationofeachclusterforhuman (A)andmacaque(B)striatumat3and6 clus-terssolutions.Thesagittal,coronalandcross sectionviews(C)providedetailedlocalization oftheclusters.
similarmacaquecluster.Thisclusterwasthereforelabeledasshowing human-specificcortico-striatalRSFC.
3.6.2. Cross-speciesdifferenceincortico-striatalRSFCofdorsalcaudate Tofurtherinvestigatethecorticalconnectivityofthehuman-specific parcel,wecalculateddifferenceintheconnectivityofthehumanand macaquedorsalcaudateparcelsbasedonitsconnectivitywiththe cor-respondinghomologous82corticalRMROIs(Fig.6).Forbothleftand rightside,wefoundthehumandorsalcaudatetobemorestrongly con-nectedtoprefrontalregions whilethemacaquedorsalcaudate tobe morestronglyconnectedtothevisualareasandtothepre/postcentral gyri.Thepre/postcentralgyriareapartof thesensorimotorcircuits,
withtheprecentralgyrusmainlyrelatedtomotorfunctionswhilethe postcentral gyrus corresponds to the primary somatosensory cortex (Johns,2014).
3.7. Differenceincortico-striatalRSFCofstriatalclustersbetweenpatients andHC
Wefoundsignificantdifferencesincortico-striatalRSFCofmultiple striatalclustersforbothPDandSCZcomparedtoHC(Fig.7).
ForPD,cortico-striatalRSFCofleftandrightventral,central, cau-dalputamenandrostralcaudatewassignificantlydifferentthanHC(p< 0.05,FDRcorrected).SignificantlyweakerRSFCbetweencentral
puta-Fig.4. Cortico-striatalresting-statefunctional connectivity(RSFC)forhumanandmacaque (A).Pearsoncorrelationdistanceofeachvoxel basedoncortico-striatalconnectivityofhuman andmacaqueforcross-speciescomparison(B).
Fig. 5. Permutation based Z-score of Pear-soncorrelationdistanceforcluster-based cross-speciescomparison.Thehumanandmacaque clusters were generated from connectivity-basedparcellationatk=3(A)and6(B).
Fig. 6. Differencein cortico-striatal resting-state functional connectivity (RSFC) of dor-salcaudate (at k = 6) between human and macaque. The cortical regions with posi-tivevaluerepresentstrongerconnectionwith thehuman dorsalcaudate, while a negative value reflects stronger connectivity between macaquedorsalcaudateandhomologous corti-calregions.ThedifferenceinRSFCvalueswere Z-scoredforvisualizationease.
Fig. 7.Significant differences in cortico-striatalRSFCofstriatalclusters(atk=6) be-tweenA) PDpatientsversusHC,B)SCZ pa-tientsversusHC.Thecolorboxesrepresent sig-nificantpvalue(p<0.05,FDRcorrected).The solidboxesrepresent RSFCinHC> Patient, whilethedashedboxesrepresentRSFCin Pa-tient>HC.Abbreviation:PD,Parkinson’s dis-ease;SCZ,schizophrenia;HC,healthycontrols; N.S.,nosignificantdifference.
menandcaudalputamenwithinferiorparietalcortex(IPC)werefound inPDcomparedtoHC.AsignificantlystrongerRSFCinPDcompared toHCwasfoundbetweenrightdorsalcaudate(i.e.,thehuman-specific cluster)andIPC,anteriorcingulatecortex(ACC),secondary somatosen-sorycortex(S2),andsecondaryauditorycortex(A2).
ForSCZ,thehuman-specificdorsalcaudateclustershowedthemost significantdifferencesbetweenpatientsandcontrolsonbothleftand rightsides,18and26respectively(Fig.7),withthemajorityofthem higherinSCZ(17and26,respectively).Forthiscluster,SCZshoweda significantlyhigherRSFCwiththetemporalcorticesincludingsuperior temporalcortex(STCandVTC),visualareas(V1andV2,dVACand ven-tralpartvVAC),A2,primarysomatosensorycortex(S1),andsubgenual cingulatecortex(SSC).SignificantlylowerRSFCbetweencaudal puta-menandventrolateralprefrontalcortex(vlPFC),ventrolateral premo-torcortex(vlPMC),butasignificantlyhigherRSFCbetweenthisstriatal clusterandtemporalcortices(inferiorandventral,ITCandVTC),and anteriorvisualarea(dorsalpart,dVAC)werefoundinSCZcomparedto HC.SignificantlylowerRSFCbetweencentralputamenandvlPMC,but significantlystrongerRSFCbetweenthisclusterandposteriorcingulate cortex(PCC)andvVACwerefoundinSCZthaninHC.Forventral puta-men,wefoundsignificantlylowerRSFCinSCZpatientsascompared toHCbetweenthisstriatalclusterandvlPFC,vlPMC,medialpremotor cortex(MPMC),andACC.However,wealsofoundsignificantlyhigher RSFCbetweenthisstriatalclusterandPCC,andvVAC.,Wefound sig-nificantdifferencesbetweentheconnectivityofleftsiderostralcaudate withprefrontal,parietal,andpremotorcortices,whilesignificant differ-encesbetweenconnectivityofrightrostralcaudatewithvisualand au-ditoryareasandcentraltemporalcortex(CTC).Inaddition,althoughwe foundnosignificantdifferenceincortico-striatalRSFCofthe dorsome-dialcaudatebetweenPDandHC,somesignificantresultsweredetected betweenSCZandHC.Forexample,significantlylowerRSFCbetween dorsomedialcaudateandIPC,MPMC,anddorsolateralprefrontal cor-tex(dlPFC),whilesignificantlyhigherRSFCbetweenthisstriatalcluster andSSCanddVACwerefoundinSCZcomparedtoHC.
Takentogether,we foundsignificant differencein cortico-striatal RSFCofcentral,ventralputamenandrostralcaudatebetweenpatients (PDandSCZ)andHC.Interestingly,forthedorsal caudate(i.e.,the human-specificcluster),almostallsignificantresultsshowedahigher cortico-striatalRSFCinpatients(PDandSCZ)comparedtoHC.In addi-tion,forSCZ,mostofthesignificantdifferenceswereobservedbetween thisstriatalclusterandvisualareas,auditoryandsomatosensorycortex. 3.8. AlterationinGMvolumeofstriatalclustersinPDandSCZ
Weanalyzedhow“diseasestatus”,“striatalclusters”,“age”,“sex”, “hemisphere” and“sites” arerelatedtotheaverageGMvolumeofthe striatumbyapplyingasix-wayANOVA.
Maineffect
Wefoundsignificantmaineffectsof“diseasestatus”,“striatal clus-ters”,“sex”,“age”,“hemisphere” and“sites” ontheaverageGMvolume inbothdisorders(Fig.8).BothPDandSCZpatientsshowedsignificantly lowerGMstriatalvolumecomparedtoHC(p<0.001).When combin-ingPDandHC,malesubjectshadasignificantlowerGMvolumethan femalesubjects,butwefoundaninverseresultinSCZandHCdataset. YoungersubjectshadahigherGMvolumethanoldersubjects.The cor-relationanalysisshowedasignificantnegativecorrelationbetweenthe GMvolumeofthestriatumandage(PDandHC:r=−0.301,p<0.001, SCZandHC:r=−0.389,p<0.001).Inaddition,wefoundsignificant higherGMstriatalvolumeinlefthemispherecomparedtoright hemi-sphere(p<0.001).
Interactioneffects
Wethenfocusedontheinteractioneffectsofthefactor“disease sta-tus” (PDandSCZseparately)withotherfactors(i.e.,“striatalclusters”, “sex”,“age” and“hemisphere”)(Fig.8).WefoundsignificantlowerGM volumeofallstriatalclustersinPDcomparedtoHC.Infemalesubjects, PDpatientshavelowerGMvolumeofthewholestriatumthanHC. Sig-nificantlowerGMvolumeofcentral,ventralputamenandrostral cau-datewerefoundinSCZcomparedtoHC.Nosignificantdifferencein GMvolumeofthehumanspecificstriatalcluster(dorsalcaudate)was foundbetweenSCZandHC.Moreover,wealsofoundsignificant inter-actionsbetween“diseasestatus” and“age”,showingasignificant neg-ativecorrelationbetweenGMvolumeofthestriatumandageinboth SCZpatients(r=−0.316,p<0.001)andHC(r=−0.459,p<0.001,
Fig.8).
4. Discussion
This study investigated functional parcellation of human and macaquestriatum.Tothisend,wefirstdemarcatedthefunctional re-gionsofhumanandmacaquestriatumusingconnectivitybased parcel-lation(CBP)basedonwhole-brainRSFC.Threeandsixstriatalcluster solutionsforbothhumanandmacaquewereselectedbasedonvarious data-drivenmodelselectioncriteria.Wefoundadorso-ventral anda rostro-caudaltopographicalorganizationinbothhumanandmacaque striatum.Wethenusedtheseclustersasabasistoestimatecross-species similaritybetweenpairsofhuman-macaqueclustersbasedontheirRSFC with82 RegionalMap (RM)homologouscorticalregions.Significant similarityinthecortico-striatalRSFCwerefoundbetweenthehuman andmacaquerostralcaudateandputamenforbothleftandrightside. However,therewasnosignificantsimilarityinthecortico-striatalRSFC betweenhumandorsalcaudateandanyofthemacaquestriatalclusters. Furtheranalysisofthishuman-specificdorsalcaudateclusterrevealed cross-speciesdifferencesinitscortico-striatalRSFC,especiallywiththe prefrontalregions,somatosensorycortexandvisualareas.When prob-ingforclinicalsignificance,RSFCbetweenthishuman-specificcluster (dorsalcaudate)withvisualarea,auditorycortex,IPC,somatosensory
Fig.8. DifferenceinVBMofstriatalclusters (atk = 6) betweenpatientswith(PD, SCZ) andHC.Abbreviation:PD,Parkinson’sdisease; SCZ,schizophrenia;HC,healthycontrols;L(R), left(right)hemisphere;M(F):male(female).
cortexwerefoundtobestrongerinSCZthanHC.Also,structural atro-phyinthishuman-specificclusterwasfoundinPDpatientscompared toHC.
4.1. Functionalparcellationofthehumanstriatum
Thehumanstriatumwasdividedintocaudateandputamenatthe simplestparcellationwithk=2.Althoughinformativeasabaseline,this solutionwasnotselectedbyourdata-drivenmodelselectionsuggesting amorecomplexfunctionalorganizationofthestriatum.Solutionswith 3and6clusterswereselectedbyvariousselectioncriteria(TableS5) andarediscussedbelowindetail.
Atthe3-clustersolutionforthehumanstriatum,weobserveda cau-datenucleusclusterandtheputamenwassplitintorostralandcaudal parts(Fig.3).Ourparcellationoftheputamenatthissolutionissimilar torecentstudythatparceltheputamenintoanteriorandposteriorpart basedonfunctionalconnectivitygradients(Tianetal.,2020).Finding differentparcelsassociatedwiththecaudatenucleusandtheputamen isinlinewiththemodularviewofthestriatumandthedifferent func-tionsofthesetwomajormodules(Grahnetal.,2008).Althoughthe tra-ditionalviewsuggeststhattheputamenismoreassociatedwithmotor functions,severalstudieshavedemonstratedthatitisalsorelatedto var-iouscognitiveprocessesincludinglearningandmemory(Elletal.,2011;
O’Dohertyetal.,2004).Choietal.(2012)investigatedRSFCbetween striatalsubregionsandsevencorticalnetworks (visual,somatomotor, dorsal attention,ventral attention, limbic,frontoparietal anddefault mode)asidentifiedbyYeoetal.(2011).Theyfoundthattheventral attentionandfrontoparietalnetworkswereconnectedwiththerostral putamen,whilethesomatomotornetworkwasprimarilyconnectedwith thecaudalputamen.Inaddition,Jungetal.(2014)foundtherostral putamenwaspositivelylinkedtoaffectiveandcognitivecontrolcortical regions,whereasthecaudalputamenwaspositivelylinkedtomore mo-torcontrolcorticalregions.Paulietal.(2016)analyzed5809functional imagingstudiesandestimatedtask-basedfunctionalco-activation pat-ternsofthestriatalvoxelswiththecerebralcortex.Accordingtothese co-activationpatterns,theputamenwasdividedintotherostralpartthat wasrelatedtosocialandlanguagefunctions,andthecaudalpart asso-ciatedtosensorimotorprocesses.Inlinewiththesepreviousfindings ourfunctionalparcellationoftheputamensuggestsadifferentiationin intrinsicRSFCandfunctionsbetweenitsrostralandcaudalparts.
The6-clusterresultsweregenerallysimilaracrosshemispheresand showedanoverallsimilaritywithourpreviousmulti-modalCBPofthe humanstriatum(Liuetal.,2020).TheRSFCpatternsbetweenour stri-atalclustersandthesevenYeocorticalnetworkswassimilartothatof
Choi etal.(2012)(Fig.S10).Similarparcellationwereshownin the
Garcia-Garciaetal.(2018)studywhichfoundstablestriatalclusters in-cludingtherostral,dorsalcaudate,caudalputamenacrossthreescans. Anotherstudy(Kimetal.,2013)appliedthetemporalindependent com-ponentanalysis(ICA)toclusterthebasalgangliaandthethalamusinto 31 functionalsubdivisions.Theyfoundthecaudatewas dividedinto head,body andtail parts.Inourparcellation,therostral partof the caudatewasclosetothehead,whilethedorsalanddorsomedialparts approximately matchedthebodyandthetail ofthecaudate, respec-tively.Differencesinfunctionalconnectivityofmeta-analytic connec-tivitymodeling(MACM)betweentheheadandbody/tailofthe cau-datenucleusrevealedthat theheadof thecaudateismore involved in cognitiveandemotionalprocesses compared tothebodyandtail (Robinsonetal.,2012).Inaddition,ourstriatalclustersweresimilarto thatinJanssenetal.(2015)study,whofoundthecaudatewasdivided intodorsal,ventral,rostralpartwhiletheputamenwassplitintodorsal, rostral,caudalpart.Theslightdifferenceislikelycausedbydifferences indataandclusteringmethods.
Ourfunctionalparcelsofhumanstriatumwerepartlyinagreement withpreviousstructuralCBPofthestriatum.Lehetal.(2007)applied DTI tractographytoexamine thestructuralconnectivitybetweenthe frontalcortexandthestriatum.TheyfoundthedlPFCprojectsto dorso-caudalcaudatewhilevlPFCprojectstoventro-rostralcaudate.For puta-men,structuralconnectivitywasfoundbetweensupplementarymotor area(SMA)anddorso-caudalputamen,betweenpremotorareaand me-dialputamen,aswellasbetweenprimarymotorarealateralputamen.In anotherstudy,Tziortzietal.(2014)usedsimilarmethodsandshowed thatthefrontallobeprojectstoalmostthewholecaudate,therostral andcentralputamenwhiletheparietallobeprojectstocaudalcaudate anddorso-caudalputamen.Small-interspersedprojectionsfromthe tem-poralandoccipitallobewereobservedintheventro-caudalputamen. Overall,thesestructuralconnectivitybasedstudiesdividethestriatum alongthedorso-ventralandrostro-caudalaxes.Similartothese find-ings,ourfunctionalparcellationat6-clustersolutiondiscernedrostral, dorsalcaudateandventral,caudalputamen,whichmaysuggesta con-vergentfunctionalandstructuralorganizationofthesestriatalclusters (Liuetal.,2020).
4.2. Functionalparcellationofthemacaquemonkeystriatum
Data-drivenmodelcriteriaalsosuggested3and6clustersolutions for the macaque striatum, which are discussed here.We found the macaquestriatumtobedividedintorostral,dorsalcaudateand puta-meninthe3-clustersolution(Fig.3).Apreviousstudyfoundthefocal projectionsfromlateralBrodmannArea9(BA9),apartofthefrontal cortex,terminatedinthedorsalcaudate,whileprojectionsfromBA46 terminatedintheputamen (Calzavaraetal., 2007).Previousstudies (Johnsonetal.,1968;KempandPowell,1970)investigatedthefiber degenerationandrevealedarostro-caudalorganizationforcortical ter-minalinstriatalregions.Ourparcellationofmacaquestriatumis gener-allyconsistentwiththesereports.Therostralcaudateprobablyincludes themajorpartoftheheadofthecaudate.Thisregionisassignedtothe cortico-striatalloopthatreceivesafferentprojectionsfromdorsolateral prefrontalcortexandlateralorbitalfrontalcortex,whichisrelatedto emotions,motivationandhighercognitiveprocesses(Alexanderetal., 1986;Haber, 2003).Ourparcellationresultconfirmsthisdifferential RSFC.Theparcellationresultsdifferedbetween humanandmacaque striatumatthisclusteringgranularity.Inhumans,theputamenwassplit intorostralandcaudalparts,whileinmacaques,thecaudatewas di-videdintorostralanddorsalparts.However,itshouldbenotedthat theinformationthatcanbegainedfromalowgranularityparcellation islimited,giventhecomplexfunctionalorganizationofthestriatum. Wehencecomparedhumanandmacaquestriatumparcellationatthe highergranularityof6clusters.
Atthehighergranularityof6clusters,themacaquerostralcaudate fromthe3-clustersolutionwasfurtherdividedintodorsalandventral parts(Fig.3).Previousnon-humanprimatestudieshasshownthatthe dorsolateralpartof thecaudate headconnectswith thedorsolateral prefrontalregions,whiletheventromedialpartconnectswiththe or-bitalfrontalregions(Alexanderetal.,1986;GoldmanandNauta,1977;
Künzle, 1978;Selemon andGoldman-Rakic,1985;Yeterian andVan Hoesen,1978).Thedorsolateralpartofthecaudateheadalsoreceives projectionsfromthearcuatepremotorareaandposteriorparietal cor-tex(BA7)(Künzle,1978;SelemonandGoldman-Rakic,1985).This stri-atalregionisrelatedtogolddirectedactions,suchasworkingmemory (BonelliandCummings,2007).Ourresultsarepartlyinlinewithand extendthesepreviousfindings,suggestingaconvergentfunctionaland structuralconnectivityoftherostralcaudate/headcaudate.Inaddition, comparedwiththeparcellationofthehumanstriatumatthis6-cluster solution,thehumanrostralcaudatethatwasnotsplit.Thissuggestsa morehomogenousRSFCwithinthehumanrostralcaudatethanwithin macaquerostralcaudate.
Takentogether,based on theCBPanalyseswe founddifferential functionalparcellationofhumanandmacaquestriatumbasedontheir RSFC,especiallyconcerningtherostralcaudate.Theseresultsmaybe relatedtodisproportionatevolumetricdifferencesintheregions func-tionallyconnectedwith thestriatum,suchas humanprefrontal cor-tex(Carlén,2017; Smaersetal., 2017),hippocampus andamygdala (Bargeretal.,2014) duringprimateevolution,andmayhavealtered RSFCofthesestriatalregionsandgenerateddifferentialfunctional par-cellationsinhumansandmacaques.
4.3. Cross-speciescomparisonbasedonhomologouscortico-striatalRSFC Wecomparedhumanandmacaquestriatumparcellationresultsin adata-drivenfashion,whichallowedustoquantifyandlocalizethe extentofsimilarityanddifferences.
Similarityinthecortico-striatalRSFCoftherostralcaudateand puta-menwas observed between human andmacaque (Fig. 5). Although correspondence between RSFC andmicrostructural connectivity fea-turesofsubcorticalregionsremainspoorlyunderstood(Moereletal., 2014; van den Heuvel et al., 2015),our findings supplement previ-ousstudiesshowingsimilaritiesinthecellularandmolecular composi-tionanddistribution,aswellasfunctionsofthestriatumacrossspecies
(Betarbetetal.,1997;HaberandKnutson,2010;Hardmanetal.,2002;
Lohrenzetal.,2016).Forexample,thecaudalputamenislikelymore relatedtomotorfunctionsinhumans(Marchandetal.,2008). Similar-ityincortico-striatalRSFCofthecorrespondingclusterbetweenhuman andmacaquesuggestsasimilarmechanismandabilityofprimary ac-tionexecution.Thisresultisinlinewithrecentfindingsthatfunctional involvementofcaudal/lateralputamenincortico-striatalmotorcircuits aresimilaracrosshuman,macaqueandmouse(Balstersetal.,2020). Bothhumans andmonkeysadoptsaccadic eyemovements tosearch forobjectsin acrowdedscene(Henderson,2003;Sheinbergand Lo-gothetis,2001).Thesaccadepatternsmaychangewithdifferentgoals andthoughts(Yarbus,2013).Thecaudateinmacaquemonkeys partic-ipatesinthevisualselectionbasedonthevalueofthevisualobjects (Hikosakaetal.,2014).Ourfindingofsignificantcross-species similar-ityofcortico-striatalRSFCoftherostralcaudatemayreflectsomewhat comparablemechanismrelatingtorewardvalue-basedselection behav-ior.Inaddition,severalstriatalclusters(e.g.,rostrodorsaland rostro-ventralcaudate)andtheircortico-striatalRSFCweresimilarbetween humanandmacaque albeitnothighlysimilar basedon permutation Z-scores(Fig.5B).Perhapsthisisduetothesimilarconnectivityand functionoftheseadjacentstriatalclusters.Hence,similarcortico-striatal RSFCofhumanrostralcaudatewithmacaquerostralcaudate,aswell aswithmacaquerostralputamencouldbedetected,givenstriatal clus-ters intheselocationsarerelatedtoemotionandcognitivefunctions (Liuetal.,2020).
Intriguingly,therewasnosignificantsimilarityinthecortico-striatal RSFCofthetopographicallysimilardorsalcaudateinthetwospecies (Fig.5B).Thismaysuggestafunctionalmodificationofthedorsal cau-dateduringevolution.Thecaudatenucleusplaysanimportantrolein integratingvisualinformationandrewardcontextindecision-making (Doietal.,2020).ThedorsalcaudateconnectswithdlPFC(Choietal., 2012;Robinsonetal.,2012),andthiscircuitisassociatedwith gener-ationofmotivation,includingtheexpectedrewardofaction,and pre-dictionofaction-outcomecontingency(Balleineetal.,2007;Haberand Knutson,2010;Muccietal.,2015).Inhumans,theactivationofdorsal caudatehasbeenobservedduringanticipationofreward,whichwas re-latedtoreal-lifemotivation(Muccietal.,2015).Basedontheexpected reward,thedorsalcaudatemaymediateactionselectionaswellas asso-ciatingtheseactionswithoutcomeingoal-directedbehaviors. Accumu-latedevidences(Balleineetal.,2007;Burtonetal.,2015;Hiebertetal., 2017;Hollermanetal.,1998)suggestthatthestriatumisacrucialpart ofacircuitrelatedtorewardanddecision-makinginbothhumanand non-humanprimates.However,thecaudateplaysadifferentrolethan theputamenasdiscussedaboveandwhetherinvolvementofthe dor-salcaudateingoal-directedbehaviorsisaffectedbyvarious complexi-tiesinhumansocialinteractionsandinducealterationinitsfunctional andstructuralconnectivitybetweenhumanandnon-humanprimateis stillnotknown.Duringprimateevolution,humanshavepresumably en-countereddifferentrewards,decision-makingandsocialcommunication tasksthannon-humanprimate(SantosandRosati,2015),whichmay haveinduceddifferentialneuralactivityofthedorsalcaudatereflected inthepresentlyobserveddissimilarcortico-striatalRSFCbetween hu-manandmacaque.Inapreviousstudy(Neubertetal.,2015), differ-entialfunctionalandstructuralconnectivitybetweencorticalregions relatingtoreward-guided learninganddecision-makingbetween hu-manandmacaquehavebeenreported. Ourfindingsshowed dissimi-larcortico-striatalRSFCofdorsalcaudaterelatingtorelevantreward anddecision-makingfunctionscan be detectedin cross-species com-parisonsupplementingprevious studies.Inaddition,furtherpost-hoc analysisrevealedthemacaquedorsalcaudatetobestronglyconnected tosomatosensorycortex(pre/postcentralgyri)andvisualareas,while thisregioninhumanwasmorestronglyconnectedtoprefrontalregions (Fig.6).Thepre/postcentralgyriarerelatedtoprimarymotorand sen-soryfunctions,whiletheprefrontalregionsareinvolvedemotion, com-plexcognitivecontrolandmotorfunctions.Thisdifferencein cortico-striatalRSFCofthedorsalcaudatemaysuggestitsmoreinvolvement