Downloaded by guest on October 1, 2021 a www.pnas.org/cgi/doi/10.1073/pnas.1814144116 gradients. from macroscopic together of characterized viewpoint been the previously not but have (42–44), data measurements these these of demonstrated reso- pairs has between magnetic association work an Existing function (41–43). resting-state (fMRI) cell/ imaging (vi) nance (v) and (37), (38–40), cytoarchitecture con- axonal density (iv) tract-traced 33–36), (iii) (30, (29), nectivity (32), densities coverage cell interneuron genome-wide are (ii) approximate which with (i) of including expression mice, many in properties, measured been of have range humans, in wide unavailable a (31). of spaces reference maps anatomical Cortical standardized modal- in diverse available from ities datasets experimental with microstructure, spe- organization. functional brain conserved a mammalian in represent of role therefore property may similar and a mouse the play in and cialization gradients exist hierarchical cortex the primate whether of unknown remains areas It cortical 30). mouse of (29, hierarchy global a to has pointed evidence recent nevertheless highly but the 26–28), (22–25). with (10, cortex compared shaping primate hierarchy uniform differentiated in relatively cortical is role cortex their the mouse The to along pointing associ- specialization cortex, transmodal functional primate to in sensory func- areas increasing primary ation of from hierarchy integration, (6, a tional follow properties (5), gradients electrophysiological plasticity Prominent and in 10). (21), variations control through (7), inhibitory of capabilities areas functional cortical specialized different the shape that gradients microstructural 17–20) spatial these as (1–3, continuously of cortex Many the 14–16). across vary 3, properties (1, types cell distribution distinct laminar and of density, size, the and (6–13); structure tic A comparison interspecies hierarchy cortical specialization hierarchical as brains. underlie mammalian well in may orga- that core as a principle as cortex, of areas nizational view cortical human across the variation support in structural results systematic orthologs Our differences. human significant interspecies notable their a with cor- a including mouse tex in in gradients, maps transcriptional agreement cortical gene between interspecies correspondence large-scale report com- of for we reference range species, spatial common across a parison as map imaging magnetic (T1w:T2w) resonance special- T1-weighted:T2-weighted cortical the mouse Using of ization. dimension highlighting spatial cortex, from underappreciated mouse vary of an areas which prefrontal connectivity, to sensory cell axonal primary interneuron long-range expression, and gene densities, a cytoarchitecture, remains reflect large-scale in of organization may similarity gradients topographic brain the or report these we of Here primate which unknown. principle to to mammalian unique extent conserved are The hetero- microcircuits. gradients of brain’s hierarchical gradient the a increas- in by supporting geneity underpinned function areas, integrated association 2018) ingly 16, to August review from sensorimotor for extends that (received hierarchy 2019 primary a 22, displays January cortex approved cerebral and primate MA, The Cambridge, Technology, of Institute Massachusetts Desimone, Robert by Edited China 201210, Shanghai Intelligence, Switzerland; c Fulcher D. cortex Ben mouse across gradients Multimodal erlCnrlo oeetLbrtr,Dprmn fHat cecsadTcnlg,Eidgen Technology, and Sciences Health of Department Laboratory, Movement of Control Neural colo hsc,Sde nvriy yny S 06 Australia; 2006, NSW Sydney, University, Sydney Physics, of School h os sa da oe oivsiaegainso cortical of gradients investigate to model ideal an is mouse The etr 13;meorhtcue(–) edii n synap- and dendritic (3–5); cytoarchi- myeloarchitecture their (1–3); in tecture vary microcircuits cortical brain, the cross d etrfrNua cec,NwYr nvriy e ok Y103 and 10003; NY York, New University, York New Science, Neural for Center | a,1 eeexpression gene onD Murray D. John , | otclgradients cortical b aei Zerbi Valerio , | c n ioJn Wang Xiao-Jing and , b eateto scity aeUiest colo eiie e ae,C 06511; CT Haven, New Medicine, of School University Yale Psychiatry, of Department 1073/pnas.1814144116/-/DCSupplemental ulse nieFbur 9 2019. 19, February online Published (50). Hochberg Corrected specified). pair given otherwise a ues, unless between (40 matched be modalities could of as areas cortical many coefficient, as correlation properties rank Spearman the cortical as shown quantified (49), diverse (ARA) 1 Atlas of Reference Fig. Allen in maps the of spatial areas 40 across the analyzed We Results (48). principle organizing core a of as areas specialization cortical functional across structural variation systematic the with consistent shape circuits, cor- cortical may mammalian of that conservation and interspecies gradients mice an tical between reveal results and ortholog Our humans humans. of and maps macaques, transcriptional mice, with gradients between of correspondence cytoarchitecture a of common demonstrate a to as T1w:T2w reference use spatial We along hierarchy. patterning functional spatial candidate common a a prop- exhibit diverse cortex of hierarchical mouse gradients of of that erties marker show candidate we Here strong (25). gradi- a specialization transcriptional as gene it dominant positioning follows ents, it and of cortex (47), hierarchy human other projections in structural of laminar established T1w:T2w range interareal the wide cortex, feedforward–feedback to a macaque correlated strongly to In is sensitive (46). also properties con- is microstructural myelin T1w:T2w intracortical of (45), marker tent a species. as across interpreted gradients Commonly cortical large-scale in refer- linking spatial measured for common map been a ence providing has humans, that and measurement macaques, mice, MRI noninvasive a is hsatcecnan uprigifrainoln at online information supporting contains article This 1 the under Published Submission.y Direct PNAS a is article This interest.y of conflict no paper. y declare the authors wrote The X.-J.W. and B.D.F. and tools; J.D.M., B.D.F., per- reagents/analytic and V.Z. new data; and contributed analyzed B.D.F. V.Z. X.-J.W. research; and designed J.D.M. X.-J.W. research; formed and J.D.M., B.D.F., contributions: Author owo orsodnemyb drse.Eal e.uce@ynyeua or [email protected] Email: addressed. be may [email protected] correspondence whom To hog hc amla rismypoesinformation. mechanisms process common may brains to brain. mammalian points which primate through species the to across way similarity similar This a puta- along in vary streams brain processing mouse tive the of circuits the local that show of to properties connectivity, axonal and of mea- densities, expression cell of the genes, including range brain, wide mouse a the across combine surements we Here the circuitry. in changes local corresponding brain’s by increasingly underpinned function be of to at specialization thought is This processed levels. integrative which is and along abstract information streams, processing sensory into incoming organized is brain The Significance h ai fT-egtdt 2wihe TwTw images (T1w:T2w) T2-weighted to T1-weighted of ratio The P corr e hnhiRsac etrfrBanSineadBrain-Inspired and Science Brain for Center Research Shanghai A d,e,1 r acltduigtemto fBnaiiand Benjamini of method the using calculated are , and NSlicense.y PNAS PNAS siceTcnsh ohcueZ Hochschule Technische ossische ¨ h iiaiybtentosailmp was maps spatial two between similarity The B. | ac ,2019 5, March . y | o.116 vol. www.pnas.org/lookup/suppl/doi:10. rc,85 Z 8057 urich, ¨ | o 10 no. | urich, ¨ across ρ, 4689–4695 P val-

NEUROSCIENCE T1w:T2w A 1.18 RSPd B RSPagl VISamVISpm SSp-tr 1.1 SSp-ll PTLp VISp somatomotor 0.60 SSp-ul 1 medial SSp-bf VISal SSp-un VISl temporal ACAd visual AUDpo VISpl anterolateral PL MOs AUDd 0.9 MOp SSp-n FRP AUDp prefrontal

SSp-m T1w:T2w 0.8 ORB SSs AUDv TEa 0.7 ECT Superior AId GU VISC PERI 0.6 Right Posterior PL AIv GU AIv AId ApI ILA TEa AIp SSs VISl

Anterior ECT FRP MOs MOp VISp VISC Left PERI ORBl VISpl PTLp VISal RSPv ACAv AUDv RSPd ACAd AUDp SSp-ll AUDd SSp-n SSp-tr ORBvl ORBm VISpm SSp-ul VISam Inferior SSp-m AUDpo SSp-un RSPagl SSp-bfd Cortical area 10-3 0.7 7 40 ACAd AUDp SSp-bfd SSp-m SSp-n AIv C D E AId F TEa 4 MOp SSp-ul SSp-bf 35 MOs d 6 ACAv VISC SSp-m 0.6 VISpl VISp SSp-ll VISp SSs ORBvl SSp-un ORBl 30 SSp-tr VISal AUDpo VISam SSs RSPd 5 AUDv SSp-tr SSp-ll AUDd 0.5 AUDp AUDd ORBl VISpm 3 ACAv ACAd MOs PTLp 25 PL VISC SSp-ul FRP FRP SSp-n PTLp AId VISC 4 PL RSPagl VISpl MOs VISal FRP VISam ORBvl RSPv AId VISal 0.4 RSPagl ORBvl 20 VISl VISam GU AUDpo ILA ORBm VISpm 3 VISpl ORBl RSPd SSp-tr GU AIv AUDv VISpm 2 AIv AUDp 15 MOp TEa 0.3 ORBm TEa GU MOp VISl ORBm TEa SSp-bfd 2 VISC MOs VISal RSPv SSs Cytoarchitecture type AIp VISp 10 ORBvl AId ILA PERI RSPaglAUDv VISam VISpl SSp-m PL SSp-n PTLp Inferred hierarchical level ACAv AIp ORBl RSPv 0.2 ECT RSPd AUDd RSPd FRP ILA ACAv PERI VISpm AUDd AIp SSp-ul 1 AIp 1 SSp-ul SSs 5

PV:(PV+SST) cell density (L2/3) VISl SSp-m SSp-ll ORBm AIv ACAd ACAd SSp-bfd SSp-n ILA PLECT RSPagl Afferent axonal connection density RSPv SSp-tr VISp PERI 0.1 0 SSp-ll 0 AUDp 0.7 0.8 0.9 1.0 1.1 0.7 0.8 0.9 1.0 1.1 0.7 0.8 0.9 1.0 1.1 0.7 0.8 0.9 1.0 1.1 T1w:T2w T1w:T2w T1w:T2w T1w:T2w

Fig. 1. The spatial map of the MRI measurement, T1w:T2w, is correlated with spatial maps of other structural properties. (A) Variation in T1w:T2w across mouse cortical areas (visualization through medial sections is in SI Appendix, Fig. S3). (B) T1w:T2w broadly decreases across families of connectivity-based groupings of cortical areas (30), from somatomotor areas to anterolateral and prefrontal areas. Groups are ordered by decreasing T1w:T2w, as are areas within each group. (C–F) Scatter plots are shown for T1w:T2w (horizontal axis) vs. (C) cytoarchitecture type (37), (D) relative density of PV(PV + SST) cells in L2/3 (29), (E) sum of incoming interareal axonal projection weight (33), and (F) hierarchical level inferred from feedforward–feedback laminar projection patterns (30). Circles are colored according to the connectivity modules in B and have sizes scaled by T1w:T2w. A small amount of vertical noise has been added to points in C to aid visualization (cytoarchitecture type takes discrete values, indicated as dotted horizontal lines). Abbreviations for cortical areas are defined in SI Appendix.

Cortical Gradients Follow T1w:T2w. We first investigated whether statin (SST)-containing interneurons; the ratio PV : (PV + SST) the T1w:T2w map is informative of functionally relevant macro- orders cortical areas along a candidate functional hierarchy scopic gradients of cortical variation in mice, as in macaques (29). We computed the correlation between T1w:T2w and layer and humans (25). As shown in Fig. 1A, the T1w:T2w map 2/3 density of each of three measured interneuron cell types, of the mouse cortex displays nontrivial anatomical specificity PV-, SST-, and vasoactive intestinal peptide (VIP)-containing on top of a broad spatial embedding, increasing along the cells, as well as the proposed hierarchy marker, PV:(PV + SST) inferior–superior axis, |ρ| = 0.53 (P = 5 × 10−4), perhaps in part (29), across 36 matching cortical areas. PV:(PV + SST) is posi- −4 reflecting a neurodevelopmental gradient of cortical matura- tively correlated with T1w:T2w, ρ = 0.58 (Pcorr = 5 × 10 , cor- tion (51) (see SI Appendix for more details). Fig. 1B shows that recting for four independent comparisons), as shown in Fig. 1D. T1w:T2w broadly decreases across five spatially localized con- The result is consistent with existing studies that have found a nectivity modules (30), from somatomotor to prefrontal areas, covariation of PV neuron density with myelin content (5) and consistent with a decreasing trend from primary unimodal to the degree of laminar differentiation (15), leading to a charac- transmodal areas in macaques and humans (25). T1w:T2w is teristic variation of inhibitory control (2, 7, 48). Given a loose not significantly correlated to variation in cell density, ρ = 0.21 interpretation of SST interneurons as having a putative “input- (P = 0.2) (39), or neuron density, ρ = −0.08 (P = 0.6, across 39 modulating” function relative to the more “output-modulating” matching cortical areas) (40), making it well positioned to cap- PV interneurons (29), this trend is also consistent with more ture density-independent differences in neural architecture (1) functionally integrative areas (lower T1w:T2w) requiring greater (alternative data sources gave consistent results; SI Appendix). input control. We also compared the interareal variation of Intracortical myelin, which T1w:T2w is sensitive to (45), T1w:T2w to cell densities (mm−3) reported by Ero¨ et al. (40) increases with laminar differentiation (4, 5), with areas lower for , excitatory cells, inhibitory cells, modulatory cells, astro- in the cortical hierarchy exhibiting a greater degree of laminar cytes, , and microglia across 39 matching corti- elaboration (22). Accordingly, T1w:T2w captures the variation cal areas. We found a significant correlation between T1w:T2w in cytoarchitecture in macaques, Kendall’s τ = 0.87 (across eight and the density of glia, ρ = 0.48 (Pcorr = 0.01); inhibitory cells, cytoarchitectonic types), and humans, ρ = 0.74 (using layer IV ρ = −0.47 (Pcorr = 0.01); microglia, ρ = 0.37 (Pcorr = 0.04); and gene markers) (25). As shown in Fig. 1C, we report a similar oligodendrocytes, ρ = 0.37 (Pcorr = 0.04) (scatter plots shown in trend in mouse cortex, with T1w:T2w increasing from dysgran- SI Appendix, Fig. S6). ular to eulaminar areas, Kendall’s τ = 0.51 [P = 2 × 10−6, using We next investigated whether T1w:T2w is related to prop- five cytoarchitectonic categories assigned to 38 matching cortical erties of interareal axonal connectivity, measured using viral areas (37)]. tract tracing (33), focusing on the normalized axonal connection w Interneuron densities vary across mouse cortical areas (29). density projected to (weighted in-degree, kin) and from w In layer 2/3, sensory-motor areas contain a greater proportion (weighted out-degree, kout) each cortical area. Across 38 match- w of (PV)-containing interneurons, while associa- ing areas, T1w:T2w is significantly correlated with kin, ρ = −0.40 tion and frontal areas contain a greater proportion of somato- (Pcorr = 0.03, correcting for two independent comparisons),

4690 | www.pnas.org/cgi/doi/10.1073/pnas.1814144116 Fulcher et al. Downloaded by guest on October 1, 2021 Downloaded by guest on October 1, 2021 |ρ cortex, human h rtP fcria rncito ssgicnl correlated significantly map, is T1w:T2w transcription the cortical (55); of with PCA PC probabilistic first using The values missing (accounting most variation for the transcriptional we of estimate maps (52), to spatial genes (PCA) explanatory brain-expressed analysis 1,055 components of principal set used a of gradients tional nomtv raiaino otclaes ri ra r hw scrlswt ai cldb 1:2 n smerzd xnlpoetos(3 are (33) projections axonal (symmetrized) in defined and are T1w:T2w areas by cortical al. for et scaled Abbreviations Fulcher radii eye. the functionally with guide a circles to yielding as manually space, added the shown been in are has close shading profiles areas Background transcriptional Brain possible. similar where areas. with annotated areas cortical cortical of places (52) organization genes informative brain-expressed 1,055 of components principal 2A. and Fig. T1w:T2w in between plotted correlation expression, are negative genes strong marker The glutamate interneuron of subset and selected and subunit a T1w:T2w of between levels coefficients transcriptional Correlation results). full for P2ry12, genes, subunit tor marker, myelin the Calb2); and Htr2c, (Grin3a, subunits are Grik2, these genes receptor 24 of glutamate (P of including many T1w:T2w maps with between transcriptional correlated the intercorrelation significantly that high found correc- we the testing conservative genes), to for (a due correcting (50) cell-type tion After hypotheses and Appendix ). subunit independent (SI receptor multiple 86 (54) of genes set We selected marker cor- a (32). in on obtain (listed focus (AMBA) to genes first 4,181 criteria Atlas for maps quality Brain expression stringent tical Mouse applied Allen and hybridiza- developed situ the in form using resolution tion, cellular with measured brain, Transcription. Gene 1F macaques, Fig. in reported in correlation shown correlation negative as mice in weak 0.09), level perfectly a hierarchical is and found 1 T1w:T2w and we between nonhierarchical Here hierarchy is a (30). 0 with (where hierarchical) hierarchy, 0.126 structural just global of does a score cortex into mouse neatly The fit (30). areas not hierar- cortical candidate mouse assign to to levels used chical been recently have in patterns tion inputs axonal T1w:T2w). of lower (with strength areas aggregate integrative functionally greater more a reflects trend cte ltof plot Scatter (B) markers. cell-type interneuron and subunits receptor glutamate coding genes ( streams. of processing maps meaningful transcriptional into them the vs. organizes T1w:T2w and profiles expression T1w:T2w gene between their ρ, using areas cortical of projection 2. Fig. with not but 1E, Fig. in plotted A oudrtn o 1:2 eae odmnn transcrip- dominant to relates T1w:T2w how understand To opeesv aao aia-pcfi nrcria projec- intracortical laminar-specific on data Comprehensive 06-. 02002040.6 0.4 0.2 0 -0.2 -0.4 -0.6 0 = opr with compare .29; Grik4, rncitoa aso ltmt eetrsbntaditrernmre ee oaywt h 1:2 a,adapicplcomponents principal a and map, T1w:T2w the with covary genes marker interneuron and subunit receptor glutamate of maps Transcriptional P2ry14, z ρ soenraie eesof levels normalized -score ;itrerncl-yemres(Pvalb markers cell-type interneuron Htr5b); = Grm2, Spearman correlationwithT1w:T2w | ρ| −0. Cnr1, 0 = 63 eetasrpinlmp ftemouse the of maps transcriptional Gene Trhr, ;srtnnrcpo uuis(Htr1a, subunits receptor serotonin Grm5); .81 and Oxtr, Grin2a Grin2b (P Pvalb Grm3 Grm4 Gria2 Gria4 Grin1 2)(h orlto ihP2i weaker, is PC2 with correlation (the (25) .Fg 2C Fig. S4 ). Fig. Appendix, SI corr |ρ Mc4r, Gria1 Gria3 Grik1 Grik2 Grik3 Grik4 Grik5 Grin2c Grin2d Grin3a Grin3b Grm1 Grm2 Grm5 Grm8 Calb2 Sst Vip ncnrs otesrn negative strong the to contrast in , n ag fohrrecep- other of range a and Mobp; 5 = 0 = P2ry2 k × .53 out w Chrm5, ρ 10 Grin3a Glutamate: metabotropic , = (P ρ −4 (see Interneuron markers −0.76 Glutamate: kainate 0 = Glutamate: NMDA Glutamate: AMPA ,i hw nFg 2B. Fig. in shown is ), 6 = corr transcription, Galr2, .14 al S2 Table Appendix, SI × < (25). (P 10 .We Appendix). SI 0. corr ρ −4 Hcrtr2, 05, Grin2d, = IAppendix). SI ,a ti in is it as ), 0 = Grin3a relative expression −0.29 ≥ |ρ ρ -2 -1 0 1 2 3 ...... 1.2 1.1 1.0 0.9 0.8 0.7 0.6 ipasa displays = .This .4). B 3(P −0.63 Grin3a Hcrtr1, ILA Grik1, ACAv ORBm 0.39), (P VISpl and = TEa ORBvl PL corr AIp ORBl ECT AId = AIv r ,23 ,5 a n b(2 coste2 ee identi- genes 24 shown the T1w:T2w, across to (32) correlated 6b S7 3B significantly and Fig. be 6a, in to 5, above 4, fied 2/3, 1, ers layer. cortical individual any in T1w:T2w 2/3, to layer in T1w:T2w 10 with density cell −0.62 SST of cell 5, PV correlation found layer and layer—we T1w:T2w in cortical between density each correlation positive in significant type a cell (37 hypoth- are (50)—each independent) Results tests 2/3 (assumed areas). esis areas), 15 (35 across (37 6 Correcting 3A and 1 (column). areas), Fig. layers: (36 in cortical 5 plotted five areas), interneu- (21 of of 4 each types areas), in three of (29) densities rons cell and overall T1w:T2w this between to correlated highly T1w:T2w of were (values 1–5 measurement; layers layers cortical each in all for computed combining estimated by was area T1w:T2w investigated brain (29). (32) We density transcription gene interneuron layers? of and maps cortical layer-specific using specific question this of specialization the Specificity. Laminar (18). fMRI resting-state human in visually characterized been and recently have that streams (22) hierarchies primary–transmodal processing parallel resembles one functional to different close clearly separates areas profiles cortical of transcriptional organization transcriptional PCs, This similar another. leading two with the of areas space placing the into areas cortical of projection taeta arsoi rdet faelseilzto r not are specialization areal demon- of gradients results macroscopic gene-expression that and strate cell-density These layers). hti etitdt pcfi otcllyr eg,tepositive the (e.g., layers T1w:T2w cortical with specific correlation with to overall correlation restricted an is show with that genes association (e.g., other an layers consistent while display individual mostly all genes in is Some T1w:T2w T1w:T2w layers. ρ ( and cortical correlation expression across overall its high between a tion and 3A) in pat- correlation Fig. 0.82; laminar strongest compare the similar 5, (with a layer correlations exhibit T1w:T2w of measurements tern two the density, T1w:T2w RSPagl 5 enx netgtdgn-xrsinpten ncria lay- cortical in patterns gene-expression investigated next We coefficients, correlation Spearman computed first We −4 .Cnitn with Consistent ). × GU PERI VISl SSp-un VISC ACAd VISp RSPv AUDv 10 .VPcl est i o xii infiatcorrelation significant a exhibit not did density cell VIP ). RSPd .Fragvngn,tedrcino correla- of direction the gene, given a For Appendix). SI −4 SSp-tr (P SSp-m SSp-ul FRP MOs .(C ). AUDpo VISpm corr AUDp SSp-bfd SSp-ll rslsfral8 ee r in are genes 86 all for (results SSp-n rjcino ri ra notesaeo h w leading two the of space the into areas brain of Projection ) PTLp VISal VISam 4 = MOp SSs i.S5). Fig. Appendix, SI AUDd Mobp × r ag-cl otclgainsdie by driven gradients cortical large-scale Are ρ PNAS 10 o ahcria ae rw n eltype cell and (row) layer cortical each for 0 = Pvalb

Principal Component 2 (brain-expressed genes) −4 -0.8 -0.6 -0.4 -0.2 0.2 0.4 0.6 sdie ylyr4adinfragranular and 4 layer by driven is 0 .52 n ae 6, layer and ) 15- 0500.5 0 -0.5 -1 -1.5 |

Principal Component1(brain-expressedgenes) prefrontal C ac ,2019 5, March FRP xrsina akro Vcell PV of marker a as expression (P ORBvl corr pamncreaincoefficients, correlation Spearman A)

7 = medial ORBm ORBl Trhr, VISpm × | RSPv ρ ILA 10 o.116 vol. = RSPd ACAv −3 PL and Grin3a, −0.65 AId Fig. Appendix, SI RSPagl ,adanegative a and ), MOs a ECT ntereolatera ACAd SSp-tr | AIv IAppendix SI MOp VISp o 10 no. VISpl PERI (P TEa VISam AIp SSp-ul GU VISC corr PTLp VISl l AUDpo SSp-ll VISal v SSp-un SSp-n SSs SSp-m Htr2c), i SSp-bfd

AUDd

somato motor AUDp sua AUDv

| 4 = temporal 4691 l . ρ × = = ρ ,

NEUROSCIENCE L1 (37 areas) A L1 (39 areas) B 0.6

L2/3 (39 areas) 0.4 L2/3 (37 areas)

L4 (22 areas) 0.2 L4 (21 areas) L5 (38 areas) 0 Layer Layer L5 (36 areas) L6a (37 areas) -0.2

L6 (35 areas) L6b (30 areas) -0.4

all (39 areas) all (40 areas) -0.6 PV VIP Trhr SST Oxtr Cnr1 Mc4r Grik1 Grik4 Htr2c Htr1a Grik2 Htr5b Galr2 Mobp Pvalb Grm5 Grm2 P2ry2 Calb2 Hcrtr1 Hcrtr2 Grin3a Chrm5 Grin2d P2ry14 Interneuron cell type P2ry12 Gene

Fig. 3. Cell densities and exhibit distinctive laminar patterns of covariation with T1w:T2w across the mouse cortex. (A and B) Spearman correlation coefficients, ρ, of T1w:T2w are plotted as colors for (A) cell density of three types of interneurons (29) and (B) transcriptional level of a given gene (column) in a given cortical layer (row). Overall correlations, obtained from combining all layers of each area, are shown in the bottom row of each plot. Transcriptional levels (32) are shown for the 24 genes with significant overall correlations to T1w:T2w, ordered by their overall correlation to T1w:T2w (results for all 86 brain-related genes are in SI Appendix, Fig. S7).

dominated by particular cortical layers; all layers exhibit macro- from each data type: gene expression, intracortical axonal con- scopic gradients of areal specialization, for distinct microcircuit nectivity, T1w:T2w, and interneuron cell density. These proper- properties. ties were visualized together by plotting the cortical variation of each measurement type with a distinct color map, as shown in A Common Hierarchical Gradient. The variation of T1w:T2w mir- Fig. 4. Measurements (columns) with highly correlated variation rors the large-scale variation of microstructural properties along across cortical areas (rows) were placed close to each other using a putative functional hierarchy of cortical areas. To more directly linkage clustering (SI Appendix). A common gradient emerges understand relationships between the diverse cortical gradients from the covariation of these diverse measurements of cortical characterized above, we combined representative measurements structure, estimated as the first principal component of the data

1.5

1 Measurement 0.5 type Correlation distance PL AIv 516397824 ILA cell density transmodal ORBm ORBvl ACAv ECT TEa AIp T1w:T2w PERI AId FRP ORBl ACAd cytoarchitecture GU RSPagl AUDpo VISC VISpl intracortical AUDv MOs connectivity RSPd VISam VISpm

Cortical area RSPv VISl gene expression PTLp AUDp VISal MOp VISp gene expression AUDd SSp-tr principal component SSs SSp-un SSp-ll primary SSp-ul structural SSp-n hierarchy somatosensory SSp-bfd SSp-m

T1w:T2w expression expression expression

PV cell density Grik2 hierarchical level PC1 transcription Pvalb Grin3a axonal input strength cytoarchitecture type Cortical properties

Fig. 4. Diverse measurements of cortical areas share a common gradient of variation. Cortical areas form rows, and measurements form columns, with different color maps used to code different datasets (from low values, light colors, to high values, dark colors), as labeled: density of PV (29) (green); T1w:T2w (53) (yellow/red); cytoarchitecture (37) (red); weighted in-degree, kw , of normalized axonal projection density (33) (orange); expression of three selected genes Grin3a, Grik2, and Pvalb (32) (blue); and the first principal component of brain-related genes (green-blue) and estimated hierarchical level (30) (purple). Areas (rows) are ordered according to the first principal component of this multimodal matrix of cortical properties and are colored according to their membership in one of six connectivity-based groupings (30) (compare with Fig. 1B). Measurements (columns) are ordered according to average linkage clustering on correlation distances according to the annotated dendrogram. Measurements cluster into two groups that either increase (Left cluster) or decrease (Right cluster) along a candidate functional hierarchy, from primary somatosensory to transmodal prefrontal areas. Missing data are shown as black rectangles.

4692 | www.pnas.org/cgi/doi/10.1073/pnas.1814144116 Fulcher et al. Downloaded by guest on October 1, 2021 Downloaded by guest on October 1, 2021 IAppendix, (SI data gene-expression mouse S9 Fig. the correspondence to mouse–human criteria applied quality-control were signal, stringent more progressively meaningful as increased a of ment ρ 0.40 0.38 uce tal. et Fulcher (P genes marker myelin and and (−0.44,0.33); genes gene, marker receptor includ- interneuron tosin consistency, interspecies the strong ing exhibit genes key other icismyrflc ilgclsbtaeo ucinlspe- of functional integration of and processing substrate efficient biological enables micro- a cortical that of cialization reflect makeup may the in circuits variations spatial Macroscale Discussion 5, Fig. genes, in 2,951 shown all genes (i) brain-related for 8 the reproduced was to but limited not was cortex: exhibit human cortex in mouse −0.63, variation the in (post- with similar T1w:T2w genes humans the a with of and correlations Two spa- adults). strongest hybridization) six mice in from between microarray situ modalities mortem differences in measurement vast distinct (high-throughput and noise, scale, measurement tial given striking given a cortex, correla- of human the map gra- in compared transcription the gene cortex, the mouse we are in and the this, gene but T1w:T2w between investigate between (25), conserved To tion genes cortex species? brain-related human two specific and of mouse dients in Transcrip- dients Gene and Gradients. T1w:T2w tional Between with Consistency than Mouse–Human T1w:T2w with S8). den- Fig. correlated cell Appendix, (SI (PV strongly level one hierarchical more but properties were all T1w:T2w sity) and between level, magnitude hierarchical in and similar were corre- 35 better coefficients Spearman the measurements). lation (across both (30) how 4 to Fig. common compared in al. areas shown cortical we properties et the gradients, with correlates cortical Harris each multimodal of the ordering explains shown hierarchical clusters structural anticorrelated two consen- the 4. this Fig. forming in along gradient, decreases individual or sus Each increases areas. either prefrontal gradient measurement primary consensus integrative from This hierarchy, to 4. functional somatosensory Fig. putative a of along rows areas the orders reorder to used and ssrn orltoscnb nepee seiec o a for much evidence as as respondence, relationship interpreted a be of can absence correlations relationship. an level strong as the as interpreted at (low be noisy humans not either be with Allen correlations in can weak and T1w:T2w consequently, (56) experiments; (32) (AHBA) individual AMBA of Atlas the Brain criteria, Human quality-control rigorous of measurements as by independent reflected the human is ρ between have gradients correspondence that transcriptional a Inter- in analyzed datasets. two consistency genes the between species brain-related matched be could 86 and orthologs our of 70 mh m × oivsiaewehrodrn ra yTwTwo the or T1w:T2w by areas ordering whether investigate To ssoni i.5 efidsgicn neseiscor- interspecies significant find we 5, Fig. in shown As ρ 2 = 10 and mh Mbp 0 = (P (P −42 ). Grik1 lhuhw aebe aeu odvlpadapply and develop to careful been have we Although . × ρ .65 6 = ρ 3 = h .Sgicn os–ua correspondence mouse–human Significant 0.45). (0.34, 10 0 ri-xrse ee (52), genes brain-expressed 806 (ii) ); h = hc emaue sacreainars genes across correlation a as measured we which , −19 × (P × and −0.65) (−0.56,0.54), 10 ρ 10 7 = 0atoyeerce genes, -enriched 60 (iii) ); mh −7 1:2 olw oiattasrpinlgra- transcriptional dominant follows T1w:T2w −3 × ;ad( and ); 0 = 4 ernerce genes, neuron-enriched 143 (iv) ); 10 Oxtr .44 ρ −6 m ρ Pvalb/PVALB h 5) ossetwt h enhance- the with Consistent (57). ) n o h ua rhlgo that of ortholog human the for and , 1oioedoyeerce genes, -enriched 41 v) (P .8;guaaereceptor glutamate (−0.41,0.48); 2) eetn h aclto for calculation the repeating (25), Grik2 1 = Calb2 ρ × h rmc (low mice or ) 10 .2,and (−0.60,0.52), (−0.48, −4 .Arneof range A 0.70). (0.57, Grin3a/GRIN3A .Teareetis agreement The ). Mobp ;teoxy- the −0.45); ρ mh 0 = ρ 3 0.41) (0.43, ρ m mh .25 should ) 0 = ρ ρ (ρ Grik4 (P mh mh m .31 = = = = inpten nseic(.. iul rcsigsras(63), streams processing visual) the (e.g., While specific in cortex. patterns mouse tion projec- of feedforward–feedback hierarchical exhibits areas compar- frontal cortex by mouse homologous and homogeneous between strikingly V1 are ison (27) between that structure pyra- differences properties synaptic 3 cortex, interareal layer strong highly of and rhe- the exhibit morphology in (26) in and example, neurons than For physiology midal pronounced 62). the less 10, monkeys, far (7, sus is cortex 9) primate (8, differentiated mice in erties inter- in brains. specialization its mammalian hierarchical stereotypical as support of that well properties existence as anatomical the map, supports measurement spatial correspondence, microcircuit single common species of a a aspects diverse along by of architecture covariation indexed cortical the be T1w:T2w, of like complexity cannot the clearly While diversity species. large- across index gradients to T1w:T2w scale like measurements noninvasive using usefulness of the demonstrate and hier- specialization of functional pro- underpinning archical anatomical results the of Our understanding connectivity. more vide axonal gene interareal laminar-specific connec- and densities, additional expression, cell report and T1w:T2w we between mice, tions synaptic in encoding invasive and available high-resolution genes Using measurements PCA) types. cortical of cell neuronal using dominant maps and receptors (estimated the transcriptional with variation specific cortex ref- with transcriptional human similarly spatial of and varies common map mouse T1w:T2w noninva- a both comparison; the as in interspecies T1w:T2w, used for map, we to erence contrast (61), need MRI homology the underappreciated sive interspecies Circumventing an an heterogeneity. display define which interareal mice, of cortex large-scale in human level mul- those these and across match macaque that 25) in (25) show 21, observed we 20, gradients Here hierarchical 10, (58–60). 7, timescales (5, tiple information sensory diverse hw sadse leln.Tasrpinlmp fkybrain-related key cortex, of human maps and mouse Transcriptional in line. (P T1w:T2w with blue similarly vary dashed genes a T1w:T2w as between shown correlation mice, the in plot ρ levels We transcription genes. and brain-related of ents 5. Fig. human T1w:T2w = h ereo neaelvraino irsrcua prop- microstructural of variation interareal of degree The -0.8 -0.6 -0.4 -0.2 0.2 0.4 0.6 0.8 1 0 × 08-. 04-. . . . 0.8 0.6 0.4 0.2 0 -0.2 -0.4 -0.6 -0.8 os–ua ossec fTwTwadtasrpinlgradi- transcriptional and T1w:T2w of consistency Mouse–human 10 Grin3a vria xs,fr7 ri-eae ee.Teeult ieis line equality The genes. brain-related 70 for axis), (vertical −4 ). Grik2 Htr2c P2ry12 Grm2 Htr1a Grik1 Calb2 Chrm5 Cnr1 Grm5 Grik4 P2ry14 PNAS Chrm1 Grin2d Oxtr Cnr2 Oprm1 Ntsr1 Grm1 Grin3b Grin2c Gria1 Oprl1 Grik3 Hrh1 | Hrh3 Vip Oprk1 Adora2a Drd4 ac ,2019 5, March Drd1 ρ Grm8 Adrb1 mouse T1w:T2w Sst Sstr2 Gria3 Drd2 Grik5 Vipr2 Calb1 Mchr1 Oprd1 Gria4 Drd3 Adra1d Galr1 Adra1a hrzna xs,adhumans, and axis), (horizontal Hrh2 Sstr4 Chrm4 Adra2a Gria2 Npy1r Grin2a Htr4 Chrm3 | Grm3 Grin2b Htr3b Grm4 o.116 vol. Plekhb1 Chrm2 Gabbr2 Grin1 Mbp P2ry6 Mobp | Hcrtr1 o 10 no. Pvalb ρ mh | = 4693 0.44

NEUROSCIENCE with a corresponding variation in microstructure (64), inter- computations in heavily myelinated and eulaminar somatosen- areal laminar projections across the whole cortex appear to be sory areas relative to more plastic and adaptive agranular pre- more stereotyped in macaques (65) than in mice (30). Differ- frontal areas (3, 5). While these diverse gradients may reflect ent functional modules of the mouse cortex fit a hierarchical associated differences in myelination, other gradients reported organization to differing extents, with a low hierarchy score here cannot be linked straightforwardly to relative myelination for the prefrontal module (0.03) and higher scores for visual levels, such as the strong transcriptional variation of Grin3a and (0.33) and temporal (0.51) modules, where a maximum score Calb2 with T1w:T2w. The convergence of multimodal cortical of 1 corresponds to an ideal hierarchy (30). Across the whole gradients may therefore reflect deeper organizational mecha- cortex, mouse brain areas fit relatively poorly into a global nisms acting in concert, perhaps through development (51). Our hierarchical organization, attaining a hierarchy score of just results are consistent with existing models of mammalian cor- 0.126, although this index is yet to be calculated in macaques, tical organization based on systematic structural variation as a preventing direct interspecies comparison. Consistent with a core organizing principle (48); in this context, the convergence lesser degree of differentiation in mouse cortex, here we report of diverse gradients found here could be used to predict lam- consistently weaker correlations of T1w:T2w with other spa- inar patterns of interareal connectivity (30) that do not rely tial maps relative to macaques and humans, for hierarchical on a global hierarchical representation of cortical areas. To level (ρmacaque = −0.76, ρmouse = −0.29), cytoarchitectural type understand the functional importance of the dominant cortical (τmacaque = 0.87, τmouse = 0.51), and the leading principal com- gradient reported here, further work is needed to explain how ponent of gene transcription (|ρmouse| = 0.53, |ρhuman| = 0.81) characteristic differences in synaptic structure and inhibitory (25). A small number of genes exhibit mouse–human differ- control may allow for more flexible behavior at the level of ences in the direction of their expression gradients with T1w:T2w information processing within local microcircuits (3, 48). (Fig. 5), but in most of these cases the correlation is weak in As more is learned about how variations in cellular and synap- either mice or humans and may be attributable to measurement tic microstructure shape functional specialization in the cortex, noise in the atlas-based expression data used here. However, it will be important to complement this understanding with our analysis flags candidate genes that further investigation may new mathematical models of brain dynamics that are properly reveal to show robust mouse–human differences in the direction constrained by the breadth and spatial detail of new datasets of expression gradients with T1w:T2w, including NMDA recep- (29, 59, 71, 72). Such approaches will allow theory and exper- tor signaling genes Grin2b/GRIN2B (ρm = 0.19, ρh = −0.63), iment to develop in tandem, with physiologically constrained Grin2d/GRIN2D (ρm = −0.41, ρh = 0.13), and Grin3b/GRIN3B mathematical models making functional predictions that can be (ρm = −0.34, ρh = 0.26). tested experimentally and used to refine the models. Under- As well as interspecies differences, our results also highlight standing how individual cortical areas—each treated as a local an underappreciated spatial dimension of cortical specialization computational unit (73) with distinctive dynamical properties— in mice that matches cortical gradients in primates. The consis- communicate coherently on a whole-brain scale may aid motiva- tency of transcriptional maps of 70 receptor subunit and cell-type tion for the next generation of brain-inspired machine- marker genes with the common reference map, T1w:T2w, in algorithms (74). Our findings offer guidance for the develop- −4 mice and humans (ρmh = 0.44, P = 1 × 10 ) is striking given ment of dynamical and functional models of large-scale cortical vastly different spatial scales and major differences in expression circuits in mammalian species. measurement between mice [high-throughput in situ hybridiza- tion (32)] and humans [microarray data from six postmortem Materials and Methods subjects (56)]. In contrast to the findings above, in which we A summary of our data and analysis methods is provided here, with addi- found a generally weaker relationship of cortical gradients to tional details provided in SI Appendix. For all datasets considered, we T1w:T2w in mice relative to primates, these transcriptional gra- retrieved values for each of 40 brain regions from the Allen Brain Refer- dients were comparable in magnitude between the two species. ence Atlas Common Coordinate Framework (CCFv3) (33, 49), where possible. A 2D projection of mouse cortical areas based on their tran- T1w:T2w data were obtained from the scalable brain atlas (53) in Waxholm scriptional signatures across brain-expressed genes (Fig. 2C) dis- space (75) and rescaled to 25-µm isotropic voxel spacing and normalized 3 tinguishes somatomotor, auditory, and visual processing streams to CCFv3, as shown in SI Appendix, Figs. S1 and S2. Cell densities (/mm ) from anterolateral and prefrontal areas and yields an organiza- of neurons, glia, excitatory cells, inhibitory cells, modulatory cells, astro- cytes, oligodendrocytes, and microglia are from Ero¨ et al. (40). Cell-count tion analogous to parallel processing streams in low-dimensional data are also from CUBIC-Atlas (39). Cytoarchitectonic classification of areas embeddings of human fMRI correlation networks (18, 66). is from Goulas et al. (37). Mouse brain axonal connectivity weights were T1w:T2w is commonly interpreted as a marker of gray-matter taken as normalized connection densities from the statistical model of Oh myelin content (45), although both T1- and T2-weighted images et al. (33). Interneuron cell densities by region and layer are from qBrain are sensitive to a wide range of microstructural properties (29). Transcriptional maps (including for specific layers) are from the AMBA (46). Our results provide transcriptional evidence for a con- (32) and were filtered by applying stringent quality-control criteria to allow nection between T1w:T2w and myelin (67), demonstrating a interpretable spatial maps for 4,181 genes. T1w:T2w correlations to gene significant relationship with the expression of Mobp and other transcriptional maps in human cortex are taken from an analysis of AHBA oligodendrocyte-enriched genes (which also display high mouse– data (56) by Burt et al. (25). human correspondence, ρ = 0.65). Interpreting T1w:T2w as mh ACKNOWLEDGMENTS. We thank Alex Fornito for his detailed and insight- a marker of relative intracortical myelination is consistent with ful comments on the manuscript. B.D.F. is supported by the National Health the increase in T1w:T2w across areas with increasing lami- and Medical Research Council Grant 1089718. J.D.M. is supported by National nar differentiation, which is associated with myelin content (4, Institute of Mental Health (NIMH) Grant R01MH112746. V.Z. is supported by the Swiss National Science Foundation Ambizione PZ00P3 173984. X.-J.W. is 5, 48). Axonal myelination improves transmission speeds (68) supported by NIMH Grant R01MH062349, Simons Collaborative Global Brain and prevents the formation of new synaptic connections (69, Program Grant 543057SPI, and Office of Naval Research Grant N00014-17-1- 70), properties that are consistent with the fast, “hard-wired” 2041.

1. Charvet CJ, Finlay BL (2014) Evo-devo and the primate isocortex: The central 3. Goulas A, Zilles K, Hilgetag CC (2018) Cortical gradients and laminar projections in organizing role of intrinsic gradients of neurogenesis. Brain Behav Evol 84:81–92. . Trends Neurosci 41:775–788. 2. Beul SF, Hilgetag CC (2015) Towards a “canonical” agranular cortical microcircuit. 4. Sanides F (1970) Functional architecture of motor and sensory cortices in primates Front Neuroanat 8:7021. in the light of a new concept of neocortex evolution. The Primate Brain Advances

4694 | www.pnas.org/cgi/doi/10.1073/pnas.1814144116 Fulcher et al. Downloaded by guest on October 1, 2021 Downloaded by guest on October 1, 2021 9 a e evlM,YoBT(07 ptih nbign irsaeand microscale bridging on spotlight A (2017) BTT Yeo MP, Heuvel gra- den principal van a along 19. network in default-mode gradients the Situating thickness (2016) Cortical al. et (2015) DS, PC Margulies Fletcher 18. IM, Goodyer L, Ronan variations. K, and Wagstyl Themes circuit: 17. neocortical The (2015) GMG Shepherd KD, Harris 16. distinguishes architecture Quantitative (2001) parval- H Barbas CC, Hilgetag -binding SM, Dombrowski the 15. of distribution Cellular (1999) myelo-, al. Cyto-, et layers: PR, Cortical Hof 2017) 14. 12, (August K lateral in Zilles strength N, synaptic glutamatergic Palomero-Gallagher of 13. Diversity (2015) JI Luebke M, Medalla dendritic of 12. development spinogenesis, Postnatal (2014) Postnatal N Ichinohe development: I, Fujita T, cell Oga H, Pyramidal Aoi T, Sasaki (2014) 11. I Fujita GN, Elston 10. 7 olsA yig B,Hlea C(06 rnilso piaea n contralateral and ipsilateral of Principles (2016) CC Hilgetag HBM, Uylings A, Goulas 37. brain. mouse adult the in expression gene of atlas Genome-wide (2006) al. et E, Lein 32. inhibitory of diversity and in Strength (2017) gradients JI Luebke Large-scale JY, Wang (2017) JP, DS Gilman M, Margulies Medalla PL, 21. Bazin PL, Bazin JM, Huntenburg 20. 9 uaaiT,e l 21)Atredmninlsnl-elrslto whole- single-cell-resolution three-dimensional A (2018) al. et TC, func- in Murakami neurons 39. of Distribution (2013) G Paxinos CR, Watson S, Herculano-Houzel 38. projectome. cortico-striatal mouse The G (2016) 36. al. brain. neocortex. et mouse mouse the H, the of networks of Neural Hintiryan (2014) connectome al. 35. et mesoscale B, Zingg A 34. (2014) al. et SW, Oh 33. cell and layer by cortical F connections intracortical cell-type-based 31. of organization stereotyped The (2018) reveal al. et maps JA, Brain-wide Harris (2017) 30. al. et cortical Y, of blocks Kim building generalizable 29. not are neurons Pyramidal (2017) JI Luebke 28. excitatory of features ultrastructural Comparative (2017) M Medalla JI, Luebke A, neurons— Hsu pyramidal of 27. features Area-specific (2016) JI cortex Luebke human M, Medalla across JP, specialization Gilman transcriptomic 26. of Hierarchy (2018) al. brain. the et in JB, wiring and Burt Segregation (2012) 25. K within Amunts pathways K, sensory Zilles converging of 24. study anatomical An (1970) T Powell cognition. E, to Jones sensation From 23. (1998) MM Mesulam 22. uce tal. et Fulcher .BnvdsPcin ,Hme-ihn ,Ballesteros-Y F, Hamzei-Sichani R, Benavides-Piccione 9. Ballesteros-Y 8. .Eso N(03 otx onto n h el e nihsit h pyramidal the into insights New cell: the think and to spinous cognition more Cortex, the (2003) All GN lobe: frontal Elston the 7. of cells Pyramidal (2000) GN Elston 6. Garc 5. est n opooyo edii pnsi os neocortex. mouse in spines dendritic of 138:403–409. morphology and Density arsaehmnbanarchitecture. brain human macroscale organization. cortical macroscale of dient hierarchies. structural Neurosci monkey. rhesus the in and systems Phylogenetic cortical prefrontal mammals: of neocortex areas. the in cortical patterns. calretinin developmental and human calbindin, bumin, in architecture synaptic 10.1016/j.neuroimage.2017.08.035. and receptor- monkey. rhesus the in cortices visual primary versus prefrontal marmoset. of cortex prefrontal medial the Funct in Struct neurons Brain pyramidal III layer of structure . and growth, axon regions. growth, dendritic cortical mouse among differs neurons pyramidal Cortex of size Dendritic function. prefrontal and neuron cortex. with. prefrontal primate in indicators stability 46:2392–2405. and plasticity of 137–208. pp York), New Corporation, Division/Meredith Primatology in otc-otclcnetvt ntemouse. the in connectivity cortico-cortical profiles. connectivity distinct by specificity 97:698–715. maintains graph, cortical dense resolution. cortex. prefrontal lateral from cingulate Neurosci anterior primate differentiates signaling organization. cortical human ri ta sn UI- xaso irsoyadtsu clearing. tissue and microscopy expansion CUBIC-X 21:625–637. using cortical atlas different brain quantitatively reveals cortex cerebral mouse the zones. of areas tional 19:1100–1114. 207–214. 2018. 18, April posted Preprint, bioRxiv:292961. brain. mouse the in class dimorphism. sexual subcortical and architecture networks. monkey. and mouse adult the of Neurol cortices frontal and visual the in monkey. rhesus and mouse in study comparative a topography. neuroimaging structural by captured monkey. the of cortex cerebral the rhD ta.(07 nitrciefaeokfrwoebanmp tcellular at maps whole-brain for framework interactive An (2017) al. et D, urth ¨ am ˘ aCbzsM Cabezas ´ ıa nt ,e l 21)Temuecria oncoe hrceie ya ultra- an by characterized connectome, cortical mouse The (2018) al. et R, anut¸ ˘ Neurosci J rn Neuroanat Front 16:990–1001. 525:2175–2191. 445:168–176. 18:170–181. 37:4717–4734. rn Neuroanat Front a Neurosci Nat a ´ e ,BnvdsPcin ,Eso N ut ,DFlp (2006) J DeFelipe R, Yuste GN, Elston R, Benavides-Piccione I, nez ˜ d oakC,Mnan Apeo-etr-rfsEducational (Appleton-Century-Crofts W Montagna CR, Noback eds , 20:1–4. ,JyeMP onY,Zkpuo ,Bra 21)Mro trends Mirror (2017) H Barbas B, Zikopoulos YJ, John MKP, Joyce A, ´ 220:3245–3258. NeuroImage 7:35. 21:139–149. hmNeuroanat Chem J 11:13644. TICS ee Cortex Cereb 22:21–31. 111:241–250. Brain Neuron 93:793–820. rcNt cdSiUSA Sci Acad Natl Proc ri tutFunct Struct Brain 16:77–116. 93:1248–1251. 13:1124–1138. Cell ee Cortex Cereb Brain a Neurosci Nat ee Cortex Cereb a 171:456–469.e22. ´ e ,DFlp ,YseR(2006) R Yuste J, DeFelipe I, nez ˜ rn Neuroanat Front 121:1013–1052. 252:1–15. Science Neurosci J 11:975–988. Cell 21:1251–1259. 27:2078–2094. 113:12574–12579. 156:1096–1111. 335:1582–1584. u Neurosci J Eur a Neurosci Nat Neuroscience a Neurosci Nat NeuroImage, Nature 8:13644. 35:112–127. Comparat J Neuron Cereb 508: Nat J 8 absH(05 eea otcladseilperna oncin:Picpe from Principles connections: prefrontal special and cortical General (2015) H Barbas 48. 5 ono A ta.(00 ahl pc:A mg-ae eeec o coordinating for reference image-based An space: Waxholm (2010) al. et GA, Johnson 75. neocortex. of ways The matrix: the Mapping (2007) KAC Martin RJ, Douglas 73. brain large-scale of modeling Biophysical (2018) A Anticevic M, Demirtas cortex JD, human Murray across heterogeneity 72. Hierarchical 2019) 7, con- (February al. functional et M, between Demirtas relationship 71. systematic A (2017) al. et JM, myeloarchitectonics. on Huntenburg of note axons A 70. ensheathes (1962) myelin V neocortical Braitenberg of fraction 69. large A MRI (2016) al. of et characterization KD, Micheva Transcriptomic 68. (2018) L French the SP, of gradients Pantazatos Functional J, (2018) SS Ritchie Ghosh 67. JD, Gabrieli for JD, matrix Schmahmann connectivity of X, interareal directed Guell Recruitment and 66. weighted (2016) A A (2012) al. Burkhalter et NT, Q, Markov Wang 65. P, Bista AM, Meier RD, cortico-cortical D’Souza selective for 64. organization laminar A (2017) A Burkhalter neuron RD, in D’Souza variation cross-cortex 63. Systematic, (2015) BL Finlay DJ, Cahalane are CJ, Charvet patterns the expression 62. and gene Neuron-enriched patterning (2013) Synaptic P (2017) Pavlidis L, A French Morrison PPC, K, Tan Zilles 61. A, Seeholzer R, cir- Duarte large-scale 60. A (2015) XJ Wang H, Kennedy MA, Gariel cortex. K, primate Knoblauch across R, timescales Chaudhuri intrinsic of 59. hierarchy A (2014) al. et JD, Murray oligo- and 58. neurons, , for database transcriptome A (2008) al. human et adult JD, the Cahoy the of atlas in 57. comprehensive analysis anatomically An component (2012) principal al. et to MJ, Hawrylycz approaches Practical 56. (2010) T Raiko A, Ilin 55. Janu update. 54. 2017 (GXD): database K expression P, Tiesinga gene R, Bakker mouse The 53. (2017) al. maturation. et cortical JH, hierarchical Finger of 52. Mechanisms (2017) B and practical Hu A T, rate: Chomiak discovery false 51. the Controlling (1995) Y Hochberg Y, Benjamini 50. (2008) HW Dong 49. architectures. counterstream high-density Cortical (2013) al. et on NT, changes: Markov based morphological brain vivo spurious 47. of in origin Neurobiological areas (2016) cortical al. et human S, Lorio Mapping 46. (2011) the DC in connectivity Essen Van hub MF, of signature Glasser transcriptional 45. A basis (2016) structural A The Fornito (2017) M BD, Rudina Fulcher N, 44. Wenderoth connec- J, Structural Balsters V, (2017) Zerbi BD J, Grandjean Fulcher A, 43. Fornito N, Wenderoth V, Zerbi SS, with Sethi brain mouse 42. the Mapping (2015) N Wenderoth M, Rudin J, Grandjean V, Zerbi 41. Er 40. 4 abetn H an ,KrigK 21)Twr nitgaino eplearning deep of integration an Toward (2016) KP Kording G, Wayne AH, Marblestone 74. tutr ofunction. to structure 342:1238406. n ua brain. mouse human adult the and in patterns oligodendrocyte-enriched with anti-correlated regionally dynamics. cortical of timescales cortex. primate the in processing 88:419–431. dynamical hierarchical for mechanism cuit Neurosci function. and development brain understanding for Neurosci resource new A dendrocytes: transcriptome. brain values. missing of presence Neurosci. BMC content. related and atlases brain public to Res Acid Nucleic Neurosci testing. multiple to approach powerful Mouse Male study. MRI quantitative A MRI. T2-weighted and T1- by revealed as content myelin connectome. mouse mouse. the in connectivity functional large-scale of brain. mouse the in dynamics signal BOLD regional 27:047405. to relates topology tome identification. network functional for 123:11–21. pipeline optimized An rs-fMRI: Neuroinf Front os ri research. brain mouse Neuroimag psychiatry. computational for applications and dynamics cortex. dynamics. cerebral neural large-scale shapes human the in myelin 27:981–997. intracortical and nectivity cortex. cerebral the neurons. inhibitory in local ratio T1-w/T2-w the on 174:504–517. focused contrast, cerebellum. of hierarchy cortex. cerebral the macaque on depends cortex visual mouse areas. across interconnecting excitation and inhibition communication. primates. and rodents in numbers n neuroscience. and 56:226–238. ,Gwli O elrD aka 21)Acl ta o h os brain. mouse the for atlas cell A (2018) H Markram D, Keller MO, Gewaltig C, o ¨ oi 21)Arcpo-ae nlsso oa cssesi h ua brain. human the in ecosystems local of analysis receptor-based A (2017) S sonis ˇ 17:1661–1663. 28:264–278. 11:150–158. 3:777–787. eLife Wly ooe,NJ). Hoboken, (Wiley, 18:551. 12:e17727. rn Neuroanat Front 45:D730–D736. h le eeec ta:ADgtlClrBanAlso h C57Bl/6J the of Atlas Brain Color Digital A Atlas: Reference Allen The 7:568. rn Neurosci Front rn optNeurosci Comput Front rcNt cdSiUSA Sci Acad Natl Proc Nature nuRvNeurosci Rev Annu te 21)Tesaal ri ta:Isatwbbsdaccess web-based Instant atlas: brain scalable The (2015) R otter ¨ NeuroImage eLife PNAS eLife ee Cortex Cereb u ri Map Brain Hum ahLanRes Learn Mach J 5:1025. 489:391–399. 5:e15784. | urOi Neurobiol Opin Curr 7:5. 11:411. ac ,2019 5, March ee Cortex Cereb 53:365–372. 10.1016/j.neuron.2019.01.017. Neuron, 24:17–36. ttScB Soc Stat R J 38:269–289. 10:406. 37:1801–1815. NeuroInf 113:1435–1440. 11:1957–2000. 25:147–160. | Neurosci J 43:156–165. opNeurol Comp J 57:289–300. 13:353–366. o.116 vol. ilPyhar o Neurosci Cog Psychiatry Biol Neurosci J 37:8092–8101. | 31:11597–11616. o 10 no. 118:141–156. ee Cortex Cereb NeuroImage NeuroImage rn Cell Front | Neuron Neuron Science Chaos 4695 Nat J

NEUROSCIENCE