Gliogenesis in the outer subventricular zone promotesenlargement and gyrification of the primate cerebrumBrian G. Rasha, Alvaro Duquea, Yury M. Morozova, Jon I. Arellanoa, Nicola Micalia, and Pasko Rakica,b,1

aDepartment of Neuroscience, Yale University, New Haven, CT 06520; and bKavli Institute for Neuroscience at Yale, Yale University, New Haven, CT 06520

Contributed by Pasko Rakic, February 8, 2019 (sent for review January 4, 2019; reviewed by Christopher Kroenke and Zoltan Molnar)

The primate cerebrum is characterized by a large expansion ofcortical surface area, the formation of convolutions, and extraor-dinarily voluminous subcortical white matter. It was recentlyproposed that this expansion is primarily driven by increasedproduction of superficial neurons in the dramatically enlargedouter subventricular zone (oSVZ). Here, we examined the devel-opment of the parietal cerebrum in macaque monkey and foundthat, indeed, the oSVZ initially adds neurons to the superficiallayers II and III, increasing their thickness. However, as the oSVZgrows in size, its output changes to production of astrocytes andoligodendrocytes, which in primates outnumber cerebral neuronsby a factor of three. After the completion of neurogenesis aroundembryonic day (E) 90, when the cerebrum is still lissencephalic, theoSVZ enlarges and contains Pax6+/Hopx+ outer (basal) radial glialcells producing astrocytes and oligodendrocytes until after E125.Our data indicate that oSVZ gliogenesis, rather than neurogenesis,correlates with rapid enlargement of the cerebrum and develop-ment of convolutions, which occur concomitantly with the forma-tion of cortical connections via the underlying white matter, inaddition to neuronal growth, elaboration of dendrites, and ampli-fication of neuropil in the cortex, which are primary factors in theformation of cerebral convolutions in primates.

cerebral cortex | brain development | corticogenesis |brain convolutions | glia

Cortical development is characterized by the orderly, se-quential production of neurons followed by glia, and upon

their generation, these cell types must migrate long distances totheir final destinations (1–5). The principal stem cells for excit-atory neurons and glial cells are the radial glial cells (RGCs)whose bodies are situated in the ventricular zone (VZ) (4–6). Inall mammals, including marsupials, daughter cells of RGCs giverise to progenitors that lose their apical attachment to the VZsurface and populate the subventricular zone (SVZ), which issmall in rodents but much larger and more complex in carnivoresand primates (7–10). In many gyrencephalic mammals, the SVZcan be divided into the inner (iSVZ) and outer (oSVZ) layers,which are separated by an inner fiber layer (IFL) and differ interms of gene expression and complement of neural progenitorsubtypes. The oSVZ compartment becomes very prominent inprimates, including humans, and contains detached RGCs (11),recently renamed as outer radial glia (oRG) or basal radial glia(bRG) (5). Knowledge about the number, types, and sequencesof neurons and glia generated from cortical progenitor cells isessential for understanding cortical development and evolutionas well as deciphering the mechanisms of neurodevelopmentaldiseases, including those that affect gyrification, e.g., lissence-phaly and polymicrogyria (12, 13).Recently it has been postulated that the remarkable enlarge-

ment of the oSVZ and its addition of neurons to the superficiallayers (III and II) is critical for the 1,000-fold expansion of thecortical surface during evolution and the main cause of cerebralgyrification in carnivores, nonhuman primates, and humans (3,14–16). However, this idea has been challenged (17–19). Actu-ally, there is compelling evidence which confirmed classic studiesby Retzius, Cajal, and others stating that detached fetal glial cells

(presently called oRG or bRG) (20) are, at the later stages ofprenatal development, the main sources of astrocytes and oli-godendrocytes (21). However, some recent studies in ferret,monkey, and human have proposed that the oSVZ almost ex-clusively produces neurons (7, 8, 22). In addition, one prominentrecent hypothesis postulated that localized areas of the oSVZwith high numbers of progenitors give rise to gyri—but not theintervening sulci (3). These “hot spots” of neurogenesis byoRGCs are proposed to cause a “fanning-out” of the cortex atspecific positions, causing the cortex to bend to form a gyrus (23,24). However, whether the timing of neurogenesis in primatescould account for such a theory, as well as whether such hot spotsactually exist and correlate with specific gyri in primates, is un-clear (19). Here, we examined the cells produced in the oSVZtoward the end of neurogenesis and beyond, in the developingmacaque monkey. We show that gliogenesis, rather than neu-rogenesis, is a principal function of a robust oSVZ afterembryonic day (E)92, which coincides with the initiation ofgyrification between E100 and E125.

ResultsWe examined samples of the dorsal parietal cerebral wall in thedeveloping macaque monkey (n = 11) at E69, E70, E90, E92,E120, E125, E145, and E149, and postnatal day (P) 7, P65, andP91. Immunohistochemistry for Pax6 demonstrated the pres-ence of the oSVZ at E70 and near disappearance by E145 (Fig.1 B–D). Large numbers of oSVZ Pax6+ progenitors were foundat E70 and E92, but it further expands at E125 (Fig. 1 B–D).We correlated these data with the distribution of tritiatedthymidine ([3H]dT)-labeled cells after acute injection followedby 1-h survival, which reveals the location of cells undergoing S

Significance

The formation of cortical convolutions of primates, includinghumans, is one of the most important subjects in de-velopmental neuroscience, but the underlying mechanisms—considered mostly solved toward the end of the 20th century—have become controversial in the last decade. Recent studiessuggest that a stem cell zone called the outer subventricularzone (oSVZ) induces gyri to form directly through neuro-genesis. Here, we provide evidence in macaque monkey dem-onstrating that oSVZ neurogenesis is actually complete beforegyrification begins and that a major function of the oSVZ is toproduce the glial cells associated with the rapid expansion andfolding of the cortical surface.

Author contributions: B.G.R. designed research; B.G.R., A.D., Y.M.M., and N.M. performedresearch; J.I.A. contributed new reagents/analytic tools; B.G.R., A.D., and Y.M.M. analyzeddata; B.G.R. and P.R. wrote the paper; and P.R. supervised the project.

Reviewers: C.K., Oregon Health & Science University; and Z.M., University of Oxford.

The authors declare no conflict of interest.

Published under the PNAS license.1To whom correspondence should be addressed. Email: pasko.rakic@yale.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1822169116/-/DCSupplemental.

Published online March 20, 2019.

www.pnas.org/cgi/doi/10.1073/pnas.1822169116 PNAS | April 2, 2019 | vol. 116 | no. 14 | 7089–7094

NEU

ROSC

IENCE

Dow

nloa

ded

by g

uest

on

Nov

embe

r 29

, 202

0

phase of the cell cycle. We found that the distribution of [3H]dT-positive cells generally matched the position of cortical progeni-tors in the VZ, iSVZ, and oSVZ zones between E70 andE92, but [3H]dT+ cells were also present in the intermedi-ate zone (IZ), subplate (SP), and cortical plate (CP) at E69,E90, and E120 (Fig. 1 A and E–H). We observed fewer [3H]dT+

cells in the periventricular iSVZ, as well as the oSVZ at E120.At E125, Pax6+ cells were present mainly at the cortico-striatalboundary region, and only rarely in the nascent ependymalzone underlying the corpus callosum (Fig. 1D). At E145, Pax6+

cells in the remnant germinal zones were very rare and ex-pression was almost undetectable.Neurogenesis is the primary output of proliferative activity in

the dorsal parietal VZ/SVZ at E75 (25), as shown by thecolabeling of the majority of cortical BrdU+ cells with NeuN inspecimens injected at E75 and killed at P91 (SI Appendix, Fig. S1A and B). However, both NeuN+ and NeuN− BrdU+ corticalcells were found (SI Appendix, Fig. S1C). In monkeys injectedwith [3H]dT at E102 and killed postnatally on P65, we foundlabeled cells of exclusively glial morphology, indicating that

neurogenesis had come to a close in dorsal parietal cortex beforeE102 (2, 25, 26). Therefore, continued presence of progenitormarkers such as Pax6, EGFR, and Olig2 beyond E92 in theparietal cortical wall indicates gliogenesis (8, 26).To investigate the output of the oSVZ, we stained macaque

tissue of different embryonic ages with various neuronal and glialprogenitor markers. Together with [3H]dT labeling (Fig. 1), be-tween E69 and E92, our analysis indicates that both neurons forthe superficial cortical layers, as well as some glial cells, aregenerated simultaneously (Fig. 2). These late-generated neuronscontribute to the complexity and thickness of the superficialcortical layers in primates, but do not add additional radialcolumns, which extend across all layers. During this period,postmitotic neurons move to the CP and settle in cortical col-umns following the radial unit model (2, 21). However, we foundthat the oSVZ also begins to produce glial cells, which migrate tothe CP but primarily to the future white matter. We found thatone of the earliest astroglial precursor markers, EGFR (27–30),costained numerous Olig2+ cells exiting the germinal zones aswell as many Pax6+ RGCs in the VZ and many progenitors of the

Fig. 1. (A) Developmental series of macaque co-ronal brain sections at the indicated fetal ages dem-onstrating the development of the oSVZ. (A) Nisslstain. (B–D) Pax6 immunostaining shows the positionof the oSVZ, as well as the commencement of EGFRexpression by E70 in the VZ, progressing to the oSVZ,with numerous EGFR+ cells invading the white matterand subplate by E92–E125. (D) By E125, the oSVZ is indecline. (E–H) Tritiated thymidine injections with 1-hsurvival demonstrate the progression of proliferativeactivity in the developing cortical wall. (G and H) Notethat at E90, tritiated thymidine-positive cells arecommon in the subplate (SP) and cortical plate (CP),and by E120, tritiated thymidine positive cells aredistributed almost evenly across the cortical wall, in-dicating that proliferative activity is not confined tothe classical germinal zones. 2/3, cortical layers 2 and3; 5/6, cortical layers 5 and 6; CTX, cortex; IFL, innerfiber layer (40); INS, insular cortex; iSVZ, inner SVZ; IZ,intermediate zone; LV, lateral ventricle; oSVZ, outerSVZ; STR, striatum; VZ, ventricular zone; WM, whitematter. B–D are composite images. (Scale bars: A andE, 1 mm; B–D, 30 μm; F–H, 50 μm.).

7090 | www.pnas.org/cgi/doi/10.1073/pnas.1822169116 Rash et al.

Dow

nloa

ded

by g

uest

on

Nov

embe

r 29

, 202

0

SVZ at E70, E92, and E125 (Figs. 1–3), but EGFR did notcolocalize with the neuronal markers, Tbr1 or βIII-tubulin (SIAppendix, Fig. S2). Furthermore, at E70, many migrating EGFR+

cells coexpressed the oligodendrocyte precursor cell (OPC)marker, PDGFRα (Fig. 2F). Together, these data strongly indicatethat many cells maintaining EGFR expression after their exit fromthe VZ/SVZ have glial fates, many of which are likely OPCs. AtE70 and through E145, Olig2/EGFR double-positive cells werefound streaming out of the dorsal parietal oSVZ radially into theIZ, SP, and CP, and many of the displayed migratory morphology,with orientation parallel to developing large axonal tracts such aslateral longitudinal bundle and the corpus callosum (Figs. 1C and2D). Migratory EGFR+ cells positioned outside the germinalzones, roughly bipolar and displaying a characteristic leadingprocess, were Olig2+ (Fig. 2E), whereas more mature EGFR+

cells with more elaborated processes at E125 were typically Olig2−

(Fig. 2E). This suggests that, as putative OPCs cease migrating andmature, they lose Olig2 expression but retain EGFR for sometime. EGFR was detected in extremely rare cells at P7, and not atall at P91 (SI Appendix, Fig. S1), indicating that EGFR expressionin cells emanating from the germinal zones is a characteristicfeature of developing, but not mature, oligodendrocytes.GFAP+ radial fibers of RGCs were visible throughout the

cortical wall at E92, but GFAP+ astrocytes were difficult todistinguish among the many RGC fibers at this age (Fig. 2G). Incontrast, at E125 and E145, GFAP+ astrocytes were numerous inthe SP, white matter, and oSVZ (Fig. 2H). In our material,EGFR+ glial precursors outside the germinal zones did notcolocalize with the robust GFAP expression in putative astro-cytes, indicating that they are likely to be developing oligoden-drocytes (Fig. 2 E and H). At E149, using electron microscopy,we detected the formation of myelin among initially myelin-freeimmature oligodendrocytes in the white matter underlying cin-gulate cortex (Fig. 2I), representing the earliest detection ofmyelination in the developing monkey. Although Olig2+ OPCswere observed invading the CP by E70 and were abundant atE125 and E145, Olig2 expression was not detected at P7 and

P91. Thus, Olig2+/EGFR+ putative OPCs are likely grossly di-minished or are no longer being produced by P7 and have dif-ferentiated into more mature states with accumulation of myelinin their cytoplasmic processes enveloping the axons.To determine whether oRGCs give rise to glial cells, we

coimmunostained sections for the oRG marker, Hopx (31), andEGFR at E70, E125, and E145. At E70, Hopx labeled numerous,

Fig. 2. The majority of EGFR+ cells exiting the ger-minal zones are Olig2+ glial precursors. Coronalsections of fetal macaque dorsal parietal cortex atE70 (A–D and F), E125 (E), E92 (G), or E145 (H),stained by immunohistochemistry for the indicatedgenes. (A–D) Colabeling of migratory EGFR+ cellswith Olig2. B–D are higher magnification images ofthe boxed regions in A. (E’’’’’) Most cells occupyingthe white matter at E125 are developing oligoden-drocytes or astrocytes. Oligodendrocyte precursorsare labeled by Olig2, overlapping substantially withEGFR expression (E’’’’; arrowheads). Olig2 labelssome oligodendrocyte precursors that were EGFRnegative (green arrowhead in E’’’’). EGFR labelssome putative oligodendrocytes that have appar-ently lost Olig2 expression (arrows in E’’’’). No over-lap of EGFR or Olig2 expression with GFAP was found(arrow in E’’’’’; H). (F) Many EGFR+ cells entering thefuture white matter and subplate zones coexpress theOPC marker, PDGFRα. (G) GFAP+ radial fibers were stillabundant at E92. (H) However, by E145, numerousGFAP+ astrocytes became apparent. (I) Representativeelectron micrograph from E149 white matter un-derlying cingulate cortex. While the majority of axonsare nonmyelinated, some show initial steps of myeli-nation (I; red asterisks). A and F are composite images.(Scale bars: A, 50 μm; B–D and F, 30 μm; E, G, and H,20 μm; I, 1 μm.).

Fig. 3. Differentiation of EGFR+ putative oligodendrocytes from Pax6+ cells.Coronal sections of macaque dorsal parietal cerebral wall were stained byimmunohistochemistry for Pax6 (green) and EGFR (red); DAPI is blue. (A, D,and G) By E70, many EGFR+ cells in the VZ express high levels of Pax6 andsome also in the oSVZ, but few reside in the SP. (B, E, and H) By E92, EGFRexpression is found in more cells in the oSVZ and many costain for Pax6,while many are now visible in the SP. (C, F, and I) The VZ is mostly dissipatedby E125, with a large reduction in the number of Pax6+ cells in the oSVZ;EGFR+ cells begin to differentiate into putative early oligodendrocytes in theoSVZ, IZ, and SP, predominantly with nonmigratory morphology, withelaborate cellular processes, and primarily located in nascent white matter.(E) Ependymal zone. (Scale bar: 50 μm.)

Rash et al. PNAS | April 2, 2019 | vol. 116 | no. 14 | 7091

NEU

ROSC

IENCE

Dow

nloa

ded

by g

uest

on

Nov

embe

r 29

, 202

0

radial process-bearing, Ki67+ progenitors in the oSVZ, althoughnot all Hopx+ oRGCs were Ki67+ (Fig. 4 A, C, and D). We foundthat some Hopx+ oRG were EGFR+ in the oSVZ (Fig. 4C),although most were EGFR− (Fig. 4 A, C, and D). By E125, manyHopx+ oRGCs had lost their basal radial fiber, were EGFR+/Ki67−,and had acquired a more elaborated, nonmigratory morphologycharacteristic of differentiating glia (Fig. 4F), indicating thatHOPX+/EGFR+ cells in the oSVZ may directly differentiate intooligodendrocytes without dividing further. Some EGFR+ cellswere also Ki67+ in the large white matter tracts, subplate, and theiSVZ and oSVZ at E70 and E125, indicating continued pro-duction of oligodendrocytic fates by early EGFR+ OPCs (Fig. 4 A,B, and D). Since many HOPX+/EGFR+ cells were Ki67−, the dataappear to support both direct astroglial differentiation fromoRGCs as well as indirect differentiation via proliferating OPCs.By E145, a few Hopx+/Ki67− cells were found in the peri-

ventricular remnant iSVZ, enriched at the corticoseptal bound-ary region, which displayed long radial fibers oriented latero-ventrally toward the temporal lobe. This is consistent with con-tinued, but spatially localized and possibly cortical area-specific,gliogenesis of generally nonproliferative oRGCs. Thus, pro-duction of oligodendrocytes, while initially occurring relativelyevenly across the tangential dimension at E70–E92, appears tobe more robust in lateral cortical regions than in dorsal/medialregions in macaque at late fetal stages around E125–E145,during the period of initial gyrification. At E145, putative

oligodendrocytes were dispersed in massive numbers, populatingthe entirety of the IZ, which is transforming into large volumesof subcortical white matter (SI Appendix, Fig. S3). ImmatureEGFR+/Olig2+ cells within the white matter and SP were usuallyKi67+ at E70, and many also at E125, indicating that they continueto proliferate outside the original germinal zones, likely corre-sponding to many [3H]dT-labeled cells present in the cortical wall(Fig. 1). Thus, these cells, together with GFAP+ astrocytes,appeared to occupy a large portion of the cerebral white matter,which is in primates much greater in volume than the cortex andbecomes populated by glial cells that are more numerous thancortical neurons by E125–E145 (Fig. 5 and SI Appendix, Fig. S3).

DiscussionOur findings demonstrate that the large oSVZ of the dorsalparietal cerebrum of macaque monkey becomes primarily glio-genic after E92—before the onset of gyrification—and is a majorsource of the astrocytes and oligodendrocytes that will out-number cerebral neurons by about threefold in macaque andhuman (32, 33). Although oRGCs have previously been charac-terized as almost exclusively neurogenic (7), our data in macaquedemonstrate a gradual transition of this zone from neurogenesis tooligodendrogenesis between E70 and E92, a period when upperlayer II and III neurons are being generated (25, 26). This period ofoverlapping neurogenesis and gliogenesis underscores the initial co-existence of heterogeneous groups of neural progenitor cells—with

Fig. 4. oRG generate putative oligodendrocytes inmacaque dorsal parietal cortex. Coronal sections ofdeveloping macaque cerebral wall at the indicatedages, stained via immunohistochemistry for the in-dicated genes. Hopx marks the VZ/SVZ as well as oRGwithin the oSVZ, and large numbers of these cellswere double labeled for EGFR. (A and B) At E70,most EGFR+ cells entering the white matter/SP wereKi67+. Some EGFR+ cells in the oSVZ were Hopx+ (C),while others were Hopx− (D). By E125, many Hopx+

cells in the oSVZ (E) were EGFR+ and displayed dif-ferentiating morphology characteristic of glia (F).After exit from the germinal zones, EGFR+ putativeoligodendrocyte precursors were either negative forHopx or expressed low levels. (E) At E125, the lateraltip of the lateral ventricle showed the largest con-centration of Hopx+ cells. A and E are compositeimages. (Scale bars: A and E, 100 μm; B, D, and F,20 μm.)

7092 | www.pnas.org/cgi/doi/10.1073/pnas.1822169116 Rash et al.

Dow

nloa

ded

by g

uest

on

Nov

embe

r 29

, 202

0

some RGCs neurogenic and others gliogenic (21). Accordingly,we found that the timing of onset of EGFR expression in RGCscorrelates with initial gliogenesis, supporting the view that MAPK-related growth factor signaling, initially provided by other pathwayssuch as fibroblast growth factor (FGF) signaling (34), continues incortical RGCs via EGF signaling to promote their proliferationduring the gliogenic phase (27).The emergence of EGFR+/Olig2+ cells by E70 indicates that

increasing numbers of RGCs and oSVZ precursors, principallyoRGCs, evolve to produce only glial fates after E92. In agree-ment with our findings, high EGFR expression has been repor-ted to be an early oligodendrocyte fate determinant in corticalRGCs of the mouse (27), and increasing numbers of both VZRGCs and oRG have been found to be Olig2+ at the terminalstage of neurogenesis in monkey and human (27, 35).It is likely that the antibody to EGFR can label both neuronal

and glial progenitors in the germinal zones. Indeed, we cannotexclude the possibility that between E70 and E90, some of the

EGFR+ RGCs may be able to give rise to EGFR− superficialneurons as well as to EGFR+ glia. However, since our [3H]dTdata show that after about E92 there is no addition of neurons(25), our present data indicate that EGFR+ progenitors in thedorsal parietal VZ, iSVZ, and oSVZ at that fetal age produceexclusively glial cells. The finding that oRG can generate glialcells is not unexpected, as RGCs have previously been shown totransform into astrocytes, and oRG are quite similar to RGCs ofthe VZ (11, 21). We found that GFAP+ astrocyte differentiationbegan in the dorsal parietal monkey cortex between E92 andE125, although at this time we are not able to determine whethercortical astrocytes predominantly originate from VZ or oSVZprogenitors, as this will require future fate mapping studies. Thedemise of the oSVZ as a discernible layer by E145, principallydue to heavy inundation by transgressing axons, is a logicalconclusion to the late cell fates produced by oRG. Namely, thefinal glial fates leave oRG progeny contained within their mi-gratory target tissue (white matter). Thus, our data fill a gap inour understanding of the development of VZ and oSVZ pro-genitors and more precisely demonstrate the time-course anddevelopmental role of the neuron-oligodendrocyte-astrocyte fateprogression in cortical progenitor dynamics.We found that some of these putative OPCs are proliferative

even after leaving the germinal zones, likely continuing to gen-erate additional glia. Indeed, areal differences in production ofoligodendrocytes, combined with their targeted migration tospecific cortical white matter areas with high densities of pro-jection axons entering or exiting the cortical plate (36), couldfeature prominently in the mechanical mechanisms directingcertain gyral and sulcal formations in primates, and warrantsfurther study (37). For example, development of the longitudinalaxonal bundle of the cingulum is associated with specific for-mation of the cingulate gyrus. Likewise, massive thalamic inputto the somatosensory cortex contrasts with nearby motor cortexaxons traveling the opposite direction to subcortical targets, andassociation areas do not send or receive axons to/from sub-cortical regions. The result is a spatio-temporally patternedgrowth of individual, area-specific axon tracts, which are asso-ciated with accumulation of oligodendrocytes. Surgical manipu-lations including enucleation of the eyes creates changes in whitematter ingrowth to the visual cortex, leading to massive changesin the magnitude and pattern of gyrification (2).While a certain amount of neuronal production and surface area

undoubtedly must be achieved as a prerequisite for gyrification, theinsufficiency of oRG neurogenesis in the oSVZ to produce corticalconvolutions in the common marmoset monkey indicates that otherfactors are at work (18, 38). Similarly, the appearance of an oSVZin a large-brained rodent, the agouti, again does not generallycorrelate with gyral development in that species, since a robustoSVZ is present across the tangential cortical dimension, but onlyone sulcus is visible at the dorsal extremity of the cortex (38), andtherefore any potential role of oSVZ neurogenesis in corticalfolding is debatable in multiple species.Are the production of glia by the oSVZ and associated differences

in white matter development important for gyrification? It appearsthat gyrification requires a certain number of axons filling thedeveloping white matter, attracting large numbers of glial cellsthat are produced by the oSVZ. Thus, while the simple presenceof a large oSVZ is not sufficient to induce gyrification, the con-tribution of oSVZ gliogenesis to gyral development may be com-mensurate with the level of white matter axonal production,amplifying its volume.In conclusion, the expansion of the cerebral cortex and its in-

timately linked process of gyrification are known to involve manymechanical factors, including neuronal growth, elaboration ofdendrites, and general amplification of cortical neuropil (4, 39),and our data do not support a decisive role for neuronal pro-duction in the oSVZ in the folding of the macaque cerebrum.

Fig. 5. EGFR+ and GFAP+ glial cells populate the cerebral white matter byE125–145. (A–C) The early cortical layer 6 and SP are marked by Tbr1 expres-sion. (D) Tbr1 expression dissipated by about E145. Early gliogenesis at E70–E92is characterized by OPCs invading the white matter and subplate with imma-ture migratory morphology (A and B), whereas large numbers of more dif-ferentiated EGFR+ (C–E) and GFAP+ (E) glia are visible at E125–E145. Thedeveloping cortex contained a few EGFR+ and GFAP+ glia, but staining wasgenerally very faint. All locants are composite images. (Scale bar: 100 μm.)

Rash et al. PNAS | April 2, 2019 | vol. 116 | no. 14 | 7093

NEU

ROSC

IENCE

Dow

nloa

ded

by g

uest

on

Nov

embe

r 29

, 202

0

Instead, given the almost completely lissencephalic form of thehuman cerebrum at 25 gestational weeks, as well as that of themacaque monkey at the time of cessation of neurogenesis in pa-rietal cortex by about E92, our data indicate that robust oSVZgliogenesis is needed to support the rapid and patterned growth ofcortical connections in the cerebral white matter, which in ma-caque represents a prominent factor in determining the magnitudeand arrangement of cortical convolutions.

Materials and MethodsAll animal work was performed in accordance with Yale InstitutionalAnimal Care and Use Committee guidelines. Timed-pregnant fetal andpostnatal monkey brains were obtained in-house at the Yale AnimalResource Center (YARC). Brains were immerse fixed in formaldehyde andprocessed into cryosections according to standard methods. Fluores-cence immunohistochemistry utilized a standard citrate-based antigenretrieval step, with imaging on a Zeiss LSM 510 confocal microscope with

Coherent Chameleon titanium sapphire 2-photon laser. [3H]dT caseswere part of the MacBrainResource, and no additional animals werekilled for those data. BrdU injections were given at 50 mg/kg. Stereo-logical counts of [3H]dT+ cells utilized a StereoInvestigator system.Electron microscopy utilized standard protocols including embedding inEpon-Araldite mixture, with thin sections prepared on a Reichert ul-tramicrotome and imaged with a JEM 1010 electron microscope. Fullmethods information, including all antibodies and dilutions, is availableonline in SI Appendix.

ACKNOWLEDGMENTS.We thankMariamma Pappy for laboratory help includingthe restoration of [3H]dT-labeled macaque brain specimens, Albert Ayoub for col-lecting fetal monkey brain specimens, YARC for macaque breedings, and Yaleveterinary clinical services for surgery services related to fetal monkey C-sections.We thank the MacBrainResource (https://medicine.yale.edu/neuroscience/macbrain/)for access to archived macaque tissue specimens. We thank the Kavli In-stitute for Neuroscience at Yale and National Institutes of Health (NIHGrants R01 MH113257, R01 DA02399, and R01 EY002593) for funding.

1. Rakic P (1972) Mode of cell migration to the superficial layers of fetal monkey neo-cortex. J Comp Neurol 145:61–83.

2. Rakic P (1988) Specification of cerebral cortical areas. Science 241:170–176.3. Kriegstein A, Noctor S, Martínez-Cerdeño V (2006) Patterns of neural stem and pro-

genitor cell division may underlie evolutionary cortical expansion. Nat Rev Neurosci 7:883–890.

4. Rakic P (2009) Evolution of the neocortex: A perspective from developmental biology.Nat Rev Neurosci 10:724–735.

5. Dehay C, Kennedy H (2007) Cell-cycle control and cortical development. Nat RevNeurosci 8:438–450.

6. Noctor SC, Flint AC, Weissman TA, Dammerman RS, Kriegstein AR (2001) Neuronsderived from radial glial cells establish radial units in neocortex. Nature 409:714–720.

7. Hansen DV, Lui JH, Parker PR, Kriegstein AR (2010) Neurogenic radial glia in the outersubventricular zone of human neocortex. Nature 464:554–561.

8. Martínez-Cerdeño V, et al. (2012) Comparative analysis of the subventricular zone inrat, ferret and macaque: Evidence for an outer subventricular zone in rodents. PLoSOne 7:e30178.

9. Molnár Z, et al. (2014) Evolution and development of the mammalian cerebral cortex.Brain Behav Evol 83:126–139.

10. Cheung AF, et al. (2010) The subventricular zone is the developmental milestone of a6-layered neocortex: Comparisons in metatherian and eutherian mammals. CerebCortex 20:1071–1081.

11. Schmechel DE, Rakic P (1979) A golgi study of radial glial cells in developing monkeytelencephalon: Morphogenesis and transformation into astrocytes. Anat Embryol(Berl) 156:115–152.

12. Gleeson JG, et al. (1998) Doublecortin, a brain-specific gene mutated in human X-linked lissencephaly and double cortex syndrome, encodes a putative signaling pro-tein. Cell 92:63–72.

13. Bae BI, et al. (2014) Evolutionarily dynamic alternative splicing of GPR56 regulatesregional cerebral cortical patterning. Science 343:764–768.

14. Namba T, Huttner WB (2017) Neural progenitor cells and their role in the develop-ment and evolutionary expansion of the neocortex. Wiley Interdiscip Rev Dev Biol 6:e256.

15. Borrell V (2018) How cells fold the cerebral cortex. J Neurosci 38:776–783.16. Stahl R, et al. (2013) Trnp1 regulates expansion and folding of the mammalian ce-

rebral cortex by control of radial glial fate. Cell 153:535–549.17. Hevner RF, Haydar TF (2012) The (not necessarily) convoluted role of basal radial glia

in cortical neurogenesis. Cereb Cortex 22:465–468.18. Kelava I, et al. (2012) Abundant occurrence of basal radial glia in the subventricular

zone of embryonic neocortex of a lissencephalic primate, the common marmosetCallithrix jacchus. Cereb Cortex 22:469–481.

19. Kroenke CD, Bayly PV (2018) How forces fold the cerebral cortex. J Neurosci 38:767–775.

20. Bystron I, Blakemore C, Rakic P (2008) Development of the human cerebral cortex:Boulder committee revisited. Nat Rev Neurosci 9:110–122.

21. Levitt P, Cooper ML, Rakic P (1981) Coexistence of neuronal and glial precursor cells inthe cerebral ventricular zone of the fetal monkey: An ultrastructural immunoperox-idase analysis. J Neurosci 1:27–39.

22. Reillo I, Borrell V (2012) Germinal zones in the developing cerebral cortex of ferret:Ontogeny, cell cycle kinetics, and diversity of progenitors. Cereb Cortex 22:2039–2054.

23. Zilles K, Palomero-Gallagher N, Amunts K (2013) Development of cortical foldingduring evolution and ontogeny. Trends Neurosci 36:275–284.

24. Reillo I, de Juan Romero C, García-Cabezas MA, Borrell V (2011) A role for in-termediate radial glia in the tangential expansion of the mammalian cerebral cortex.Cereb Cortex 21:1674–1694.

25. Rakic P (1974) Neurons in rhesus monkey visual cortex: Systematic relation betweentime of origin and eventual disposition. Science 183:425–427.

26. Rakic P (1995) A small step for the cell, a giant leap for mankind: A hypothesis ofneocortical expansion during evolution. Trends Neurosci 18:383–388.

27. Sun Y, Goderie SK, Temple S (2005) Asymmetric distribution of EGFR receptor duringmitosis generates diverse CNS progenitor cells. Neuron 45:873–886.

28. Aguirre A, Dupree JL, Mangin JM, Gallo V (2007) A functional role for EGFR signalingin myelination and remyelination. Nat Neurosci 10:990–1002.

29. Aguirre A, Gallo V (2007) Reduced EGFR signaling in progenitor cells of the adultsubventricular zone attenuates oligodendrogenesis after demyelination. Neuron GliaBiol 3:209–220.

30. Sugiarto S, et al. (2011) Asymmetry-defective oligodendrocyte progenitors are gliomaprecursors. Cancer Cell 20:328–340.

31. Pollen AA, et al. (2015) Molecular identity of human outer radial glia during corticaldevelopment. Cell 163:55–67.

32. Azevedo FA, et al. (2009) Equal numbers of neuronal and nonneuronal cells make thehuman brain an isometrically scaled-up primate brain. J Comp Neurol 513:532–541.

33. Gabi M, et al. (2010) Cellular scaling rules for the brains of an extended number ofprimate species. Brain Behav Evol 76:32–44.

34. Rash BG, Lim HD, Breunig JJ, Vaccarino FM (2011) FGF signaling expands embryoniccortical surface area by regulating Notch-dependent neurogenesis. J Neurosci 31:15604–15617.

35. Mo Z, Zecevic N (2009) Human fetal radial glia cells generate oligodendrocytesin vitro. Glia 57:490–498.

36. Price DJ, et al. (2006) The development of cortical connections. Eur J Neurosci 23:910–920.

37. Van Essen DC (1997) A tension-based theory of morphogenesis and compact wiring inthe central nervous system. Nature 385:313–318.

38. García-Moreno F, Vasistha NA, Trevia N, Bourne JA, Molnár Z (2012) Compartmen-talization of cerebral cortical germinal zones in a lissencephalic primate and gyr-encephalic rodent. Cereb Cortex 22:482–492.

39. Welker W (1990) Why does cerebral cortex fissure and fold? Cerebral Cortex, edsJones EG, Peters A (Springer, Boston), Vol 8B, pp 3–136.

40. Molnár Z, Clowry G (2012) Cerebral cortical development in rodents and primates.Prog Brain Res 195:45–70.

7094 | www.pnas.org/cgi/doi/10.1073/pnas.1822169116 Rash et al.

Dow

nloa

ded

by g

uest

on

Nov

embe

r 29

, 202

0