Content uploaded by Dongchang Xiao
Author content
All content in this area was uploaded by Dongchang Xiao on Jul 01, 2022
Content may be subject to copyright.
ResouRce
https://doi.org/10.1038/s41593-022-01073-x
1State Key Laboratory of Ophthalmology, Zhongshan Ophthalmic Center, Sun Yat-sen University, Guangdong Provincial Key Laboratory of Ophthalmology
and Visual Science, Guangzhou, China. 2Department of Medicine, Keck School of Medicine, University of Southern California, Los Angeles, CA, USA.
3Guangzhou Regenerative Medicine and Health Guangdong Laboratory, Guangzhou, China. 4Research Unit of Ocular Development and Regeneration,
Chinese Academy of Medical Sciences, Beijing, China. 5Translational Research Institute of Brain and Brain-Like Intelligence and Department of
Anesthesiology, Shanghai Fourth People’s Hospital Affiliated to Tongji University School of Medicine, Shanghai, China. 6European Bioinformatics Institute,
European Molecular Biology Laboratory, Wellcome Genome Campus, Cambridge, UK. 7Guangzhou Laboratory, Guangzhou International Bio Island,
Guangzhou, China. 8Guangdong Province Key Laboratory of Brain Function and Disease, Guangzhou, China. 9These authors contributed equally: Zhao-Zhe
Hao, Jia-Ru Wei, Dongchang Xiao. ✉e-mail: xiangmq3@mail.sysu.edu.cn; yzliu62@yahoo.com; zmiao@ebi.ac.uk; liush87@mail.sysu.edu.cn
There is abundant evidence that, in adult rodents, a substantial
number of new neurons are generated in discrete brain areas,
including the subgranular zone (SGZ) of the hippocampal
dentate gyrus (DG) and the subventricular zone (SVZ) along the
walls of the lateral ventricle1–4. Newborn neurons from the hip-
pocampal DG can integrate into the existing neural circuits5,6 and
contribute to various brain functions, including cognitive flexibility,
emotional control, pattern separation, learning and memory forma-
tion7–9, suggesting their crucial roles in maintaining the network
plasticity7,10.
However, far less is known about whether and how adult neu-
rogenesis occurs in primates. Newly formed neurons have been
revealed in adult primates through labeling of DNA replication and
rodent-derived markers11–15. To what extent adult neurogenesis hap-
pens is still controversial16–22, largely owing to the fact that the con-
clusions were drawn from the immunostaining of a limited selection
of proxy markers. More importantly, the contribution of the newly
formed neurons to the network function is unknown. Differences
between primate and rodent, including their precursor cell propor-
tion23, the maturation process24 and their migration path25,26, have
been documented27. Detailed profiling of the neurogenesis process
in adult primates is necessary to bridge the gaps between rodent and
human adult neurogenesis.
The advent of single-cell RNA sequencing (scRNA-seq) technol-
ogy has enabled an in-depth understanding of mouse adult neuro-
genesis processes, including the quiescence level of stem cells and
their metabolic profile28,29. The molecular dynamics and diversity
of rodent hippocampus DG cell types throughout fetal and early
development and into adulthood have been examined in detail30.
Several subpopulations of precursors, including active RGLs, RGLs
in shallow quiescence state, RGLs in deep quiescence, proliferat-
ing RGLs, neural progenitors, astrocyte progenitors, and oligoden-
drocyte progenitors, have been identified28,31–37. More recently, a
similar approach has successfully profiled these cells in developing
humans38,39 and in adult macaques and pigs40, but a more detailed
molecular characterization of adult primate neurogenesis is needed.
In this work, we performed scRNA-seq of adult primate hippo-
campus, generated a dataset of 207,785 cells and analyzed cell differ-
entiation trajectories and molecular properties during adult primate
neurogenesis. This dataset reveals 34 discriminative cell types that
cover all major hippocampal cell categories, including RGLs, IPCs,
and neuroblasts. Furthermore, we identified stem cell specifica-
tion trajectories for both granule cells (GCs) and neuroglia with
novel intermediate populations. The ability of adult primate neu-
ral precursors to proliferate and differentiate was supported by ex
vivo neurosphere culture experiments. This large-scale dataset can
provide a valuable resource for future studies on molecular mecha-
nisms regulating adult neurogenesis and may serve to bridge the
translation from rodent research to clinical application in human
patients.
Results
Transcriptomic cell type taxonomy. Cells from the hippocampus
(Fig. 1a) with an enrichment of the DG region by microdissection
Single-cell transcriptomics of adult macaque
hippocampus reveals neural precursor cell
populations
Zhao-Zhe Hao 1,9, Jia-Ru Wei 1,9, Dongchang Xiao1,9, Ruifeng Liu1, Nana Xu1,
Lei Tang1, Mengyao Huang1, Yuhui Shen1, Changsheng Xing 2, Wanjing Huang1, Xialin Liu 1,
Mengqing Xiang 1 ✉ , Yizhi Liu 1,3,4 ✉ , Zhichao Miao 5,6,7 ✉ and Sheng Liu 1,8 ✉
The extent to which neurogenesis occurs in adult primates remains controversial. In this study, using an optimized single-cell
RNA sequencing pipeline, we profiled 207,785 cells from the adult macaque hippocampus and identified 34 cell populations
comprising all major hippocampal cell types. Analysis of their gene expression, specification trajectories and gene regulatory
networks revealed the presence of all key neurogenic precursor cell populations, including a heterogeneous pool of radial glia-like
cells (RGLs), intermediate progenitor cells (IPCs) and neuroblasts. We identified HMGB2 as a novel IPC marker. Comparison
with mouse single-cell transcriptomic data revealed differences in neurogenic processes between species. We confirmed that
neurogenesis is recapitulated in ex vivo neurosphere cultures from adult primates, further supporting the existence of neural
precursor cells (NPCs) that are able to proliferate and differentiate. Our large-scale dataset provides a comprehensive adult
neurogenesis atlas for primates.
NATURE NEUROSCIENCE | www.nature.com/natureneuroscience
ResouRce NATuRE NEuROSCIENCE
were collected from eight adult macaque monkeys (Macaca fas-
cicularis) of both sexes. Using a specifically developed scRNA-seq
pipeline (Methods) for adult primates, we established a dataset
containing 207,785 transcriptomes (10x Genomics) of isolated
whole cells that passed the quality control (Methods) (Fig. 1a and
Extended Data Fig. 1). Unsupervised clustering and uniform mani-
fold approximation and projection (UMAP) visualization of the
dataset (Fig. 1b) reveals 34 transcriptomically distinct cell popula-
tions. These populations cover all major cell categories in the hip-
pocampus30,38. These populations comprise: (1) neuroglia, including
immature and mature AQP4+ astrocytes, C1QL1+ oligodendrocyte
progenitors cells (OPCs), BCAS1+ newly formed oligodendrocytes
(NFOLs) and MOG+ mature oligodendrocytes (MOLs); (2) neu-
rons, including GAD2+ inhibitory neurons (GABA), RGS4+ pyra-
midal neurons (Pyr), immature and mature PROX1+ granule cells
(GC_ims and GCs); (3) vascular cells, including FLT1+ endothe-
lial cells and DCN+ vascular leptomeningeal cells (VLMCs); (4)
immune cells, including C1QA+ microglia and LYVE1+ perivascular
macrophages (PVMs); (5) RELN+ Cajal–Retzius cells (CRs); and
(6) CFAP126+ ependymal cells. Especially, five NPC populations
are identified, including two SLC1A3+ RGLs (RGL_1 and RGL_2),
HMGB2+ASCL1+ IPC population (IPC_1), HMGB2+ASCL1− IPC
population (IPC_2) and SOX4+ neuroblast population (Fig. 1c,d).
In addition to the known marker genes, novel precursor and
neuron markers are identified: RFC4 discriminates IPCs; NNAT
(Neuronatin) highlights neuroblasts; and NPY1R is a marker for
GCs (Fig. 1d and Supplementary Information). The identification
of these populations is further validated by their enriched gene
regulatory networks (GRNs, regulons; Extended Data Fig. 2 and
Supplementary Information).
The neurogenesis drops sharply during juvenile development for
both humans and macaques17,41 and maintains constant in macaques
after puberty14. Consistently, we found the existence of all the cell
populations in both young and middle-aged adults (Extended Data
Fig. 3a). For most cell populations, including RGLs, the proportions
within each sample are not significantly different between the young
(4–6 years old, n = 5 animals and 17 samples) and the middle-aged
adults (13–15 years old, n = 3 animals and 8 samples). This is con-
sistent with the low but sustainable level of neurogenesis found in
adult mice28, suggesting the broad existence of neurogenesis across
age and sex groups in adult macaques.
To systematically compare the gene expression profiles between
the two age groups, we performed the differentially expressed
gene (DEG) analysis for each neurogenic-related population. For
the neuroblast population (Extended Data Fig. 3b), of the 19,888
total detected genes, 1,435 are significantly altered between the two
age groups (false discovery rate (FDR) < 0.05; Methods), whereas
only seven genes show more than 0.5 log2 fold change (log2FC),
with the RPL31 being the most upregulated genes for young adult
(log2FC = 0.68) and NR2F2 being the most upregulated genes for
the middle-aged adult (log2FC = −0.59). Similarly, the DEG analysis
revealed very limited genes that display more than a two-fold change
(log2FC > 1) for all neurogenic-related populations (Extended Data
Fig. 3c). Consistently, the neurogenic-related genes (n = 2,081;
Supplementary Table 1) for the key neurogenic populations are
highly correlated (Extended Data Fig. 3d), without any genes show-
ing more than two-fold differences. There also exist genes that are
slightly shifted (>0.5 fold change; Extended Data Fig. 3e), which
may reflect the subtle shift across ages.
The cell composition and gene expression differences between
male (n = 6) and female (n = 2) macaques (Extended Data Fig. 4)
are also subtle. There are no major differences in the cell distribu-
tions or gene expression between the two groups, suggesting that
the neurogenic process is highly similar between male and female
macaques.
Dissecting the NPC populations. It is challenging to identify
NPCs in primates using a limited number of rodent-derived marker
genes. Here, taking advantage of the high-throughput dataset, we
validate the precursor cell populations from four perspectives: (1)
the expression of canonical marker genes and regulons; (2) cell pro-
liferation and metabolic processes; (3) cross-species data projection
from hippocampus datasets of developing, juvenile and adult mice
or humans30,33,37–39; and (4) differentiation direction inferred by the
RNA velocity42,43. Five key NPC populations are identified, cover-
ing the complete neural genesis process, starting from the putative
neural stem cell (NSC) populations (RGL_1 and RGL_2), via two
actively cycling populations (IPC_1 and IPC_2), to the differentiat-
ing neuroblast population. The neuroblast leads to the immature
granule cells (GC_im) and then mature granule cells (GC_1–3),
further inferring the granule cell maturation process. In addition
to the GCs, RGLs also give rise to the astrocytes (Astro_1–4) via
immature astrocytes (Astro_im1, 2). The lineages and marker genes
of these populations are summarized in a schematic illustration
(Supplementary Information).
Identification of the RGL populations. The RGLs are putative
NSCs found in the adult hippocampus44–46. The RGLs (RGL_1,
the red population at the center of Fig. 1b, and RGL_2, the dark
red population adjacent to RGL_1) are located at the center of the
UMAP visualization of the dataset. The RGLs are closely related
to astrocytes, expressing both the canonical astrocyte marker gene
(GFAP)17,47 and the neural precursor marker genes SOX2 (ref. 28)
and SLC1A3 (ref. 36) (Fig. 1d). Immunofluorescent staining con-
firms the existence of RGLs in the SGZ of macaque hippocampus
DG (Fig. 1f). These SOX2+GLAST+(SLC1A3)GFAP+ RGLs show
typical radial morphology of adult RGLs28. Immunostaining of the
canonical neural progenitor marker NESTIN revealed NPCs at the
SGZ layers with the apical processes reaching the molecular lay-
ers (Extended Data Fig. 5a). Immunostaining of NESTIN with the
Fig. 1 | Dissecting the adult macaque hippocampal cells and molecular signatures. a, Schematic illustration of the workflow. The fluorescent Nissl
staining image illustrates the macaque hippocampus with a typical DG structure. Scale bar, 1 mm. Schematic drawings were created with BioRender. b,
UMAP visualization of 207,785 cells from adult macaque DG and attaching areas. The 34 discriminative cell populations are labeled and colored using the
same color scheme in d. Labels in UMAP: RGL, radial glia-like cells; IPC, intermediate progenitor cells; NB, neuroblasts; OPC, oligodendrocyte progenitor
cells; MOL, mature oligodendrocytes; NFOL, newly formed oligodendrocytes; Astro, Astrocytes; Astro_im, immature astrocytes; CR, Cajal–Retzius cells;
GC_im, immature granule cells; GC, granule cells; Pyr, pyramidal neurons; GABA, GABAergic neurons; PVM, perivascular macrophages; VLMC, vascular
leptomeningeal cells; Unk, Unknown. c, UMAP visualizations colored by the expression of the known marker genes for the selective cell population. Dots,
individual cells; gray, no expression; red, relative expression (log-normalized gene expression). Colorbar indicates the log-normalized gene expression
level. d, The violin plot demonstrates the expression of the discriminative marker genes in the 34 cell populations. Exp, the maximum expression level
of log-normalized gene expression. e, Heatmap of selective marker genes for each cell population with their enriched functional annotations. Exp, the
maximum expression level of log-normalized gene expression. f, Immunostaining of the RGL markers (GFAP and SLC1A3/GLAST) and neural precursor
marker SOX2 in DG show the long apical processes. The arrowhead indicates the nucleus of the RGL. The arrows indicate the apical processes from the
same cell. DAPI, nuclei. Scale bar, 10 μm. n = 6 brains.
NATURE NEUROSCIENCE | www.nature.com/natureneuroscience
ResouRce
NATuRE NEuROSCIENCE
c
d
GFAP MKI67 SOX4
AQP4
GAD2 RGS4
RGL_1
RGL_2
IPC_2
Neuroblast
Ependymal
Endothelial
Microglia
MOL
NFOL
OPC
Astro
Astro_im
GC
IPC_1
PVM
Wnt signaling pathway
Adherens junction
Cell–cell signaling
Endocannabinoid signaling
Protein synthesis
MDM2/MDM4 family protain binding
Mitotic cell cycle
DNA replication
Generation of neurons
Cell projection morphogenesis
Gliogeneisis
Glycolysis
Neurotransmitter secretion
Synaptic vesicle exocytosis
Gliogenesis
Fatty acid metabolism
Glial cell differentiation
Chondroitin sulfate proteoglycan metabolic
Axon ensheathment
Oligodendrocyte differentiation
Ensheathment of neurons
Myelin assembly
Phagosome
Complement and coagulation cascades
Antigen processing and presentation
Hematopoitic cell lineage
Tube morphogenesis
Microtuble-based movement
Cilium organization
1.0–1.0
Normalized z-score
Blood vessel morphogenesis
0
0
3.0
Exp
a
e
UMAP2
UMAP1
VLMC
Pyr_2
PVM
Unk
Microglia_2
Microglia_1
Astro_1,2,3,4
GC_2
RGL_1
RGL_2RGL_2
Pre_OPC
OPC_1, 2 MOL
IPC_2
NFOL
GABA
Pyr_1
CR Endothelial
GC_3
GC_im
Ependymal
GC_1
NB
IPC_1
Astro_im1
Astro_im2
b
C1QA
PROX1
CSPG4
SGZ
GCL
SOX2/GFAP/GLAST(SLC1A3)/DAPI
f
5.0
SLC1A3
2.5
SOX2
5.0
GFAP
2.5
HMGB2
2.5
ASCL1 2.5
RFC4
2.5
MKI67
2.5
SOX4 5.0
NNAT
2.5
PROX1
2.5
NPY1R
2.5
ERC2
2.5
CPLX2
2.5
RGS4
2.5
ELAVL2
2.5
GAD2
2.5
CSPG4
2.5
BCAS1
2.5
MOG 5.0
C1QA 5.0
RELN 5.0
FLT1
2.5
FOXJ1
0
2.5
LYVE1
RGL_1
RGL_2
Astro_im1
Astro_im2
Astro_1
Astro_2
Astro_3
Astro_4
IPC_1
IPC_2
NB
GC_im
GC_1
GC_2
GC_3
Pyr_1
Pyr_2
GABA
Pre_OPC
OPC_1
OPC_2
NFOL
MOL
PhgMG_RGL
PhgMG_IPC2
PhgMG_Pyr
Microglia_1
Microglia_2
CR
Endothelial
Ependymal
PVM
VLMC
Unk
5.0
DCN
Bioinfomatics analysis10x single-cell RNA-seq
Neurosphere
Single-cell suspension
Macaque Hippocampus
Beads
Cells Oil
0
4.0
0
4.0
0
6.0
0
3.0
0
7.5
0
1.5
0
4.0
0
3.0
NATURE NEUROSCIENCE | www.nature.com/natureneuroscience
ResouRce NATuRE NEuROSCIENCE
proliferating marker MKI67 further validates the existence of pro-
liferating NPCs derived from SGZ layers (Extended Data Fig. 5b).
Similarly, co-staining of the canonical neural progenitor marker
HOPX with SOX2 and GFAP reveals RGLs in the SGZ with the typi-
cal RGL morphology (Extended Data Fig. 5c and Supplementary
Information), whereas the ASCL1+SOX2+GFAP+ labels both RGLs
and proliferating neural progenitors (Extended Data Fig. 5d).
The RGLs interact extensively with the local environment cues
to restrict their proliferation at a proper level. Consistently, the cell–
cell signaling pathways are upregulated in the RGLs (Fig. 1e and
Supplementary Tables 2 and 3), including the Wnt signaling path-
way, which is one of the principal regulatory pathways for RGLs33.
The RGL marker genes are also enriched for genes that encode pro-
teins involving adherens junction, which is crucial for newborn cell
migration48, and the endocannabinoid signaling49, providing one
possible mechanism of how the RGL processes receive signals from
local neurons, glia and vasculature to regulate their fate.
To further validate our identification of RGLs, we aligned our
dataset to a well-annotated mouse transcriptome dataset with
identified hippocampal RGLs30. The cell type homology analysis
quantitatively compares the cell taxonomy across species50. In brief,
the macaque and mouse datasets were integrated using scVI51 and
Harmony52 and then clustered in the corrected principal compo-
nent analysis (PCA) latent space (Fig. 2a and Methods). A good
cell type alignment (higher alignment score in Fig. 2b; Methods)
can be found between cells with shared gene expression patterns
and distributed at similar coordinates in the UMAP and Louvain
clusters (Fig. 2c,d), whereas cells that do not share gene expression
patterns will have low similarity in their distributions (lower align-
ment score in Fig. 2b). The heatmap of the cross-species taxonomy
matrix (Fig. 2b and Supplementary Table 4) reveals that most of
the macaque hippocampal cell types are aligned exclusively to their
mouse counterparts, demonstrating that the hippocampal cells are
conserved between primates and rodents. The macaque RGLs align
with the adult and young mouse RGLs or immature astrocytes, sup-
porting our identification of the RGLs in macaques. Using similar
approaches, we aligned our dataset to additional developing or adult
mouse and human datasets30,33,37–39 (Fig. 2e,f and Extended Data
Figs. 6 and 7). Similarly, most macaque hippocampal cell types are
aligned exclusively to their mouse, macaque, and human counter-
parts, whereas the macaque RGLs are aligned to RGLs or astrocytes
(Extended Data Figs. 6 and 7).
With the high similarity of marker gene expression between
astrocytes and RGLs (Fig. 3e,f), data integration cannot distinguish
RGLs from astrocytes. To distinguish RGLs from immature astro-
cytes, RNA velocity43 of the RGLs and astrocytes was calculated
using scVelo algorithm42 to infer the differentiation trajectories
among these populations. The scVelo can infer each cell’s expres-
sion gradient in neurogenesis and provide a working hypothesis to
disentangle the subpopulations during the neurogenesis by compar-
ing the abundance of newly transcribed, unspliced pre-mRNAs with
mature, spliced mRNAs, and lineage tracing experiments are neces-
sary to consolidate this inferred lineage hierarchy. Using this algo-
rithm, we inferred the astrocyte genesis trajectory from the RGLs
to astrocytes (Fig. 3a). The differentiation trajectory, together with
the pseudotime derived from the RNA velocity (Fig. 3b), revealed
that the astrocyte genesis trajectory starts with the RGL_1, via the
RGL_2 and immature astrocytes, and eventually to the mature
astrocytes, supporting our identification of the RGLs. The putative
driver genes along the differentiation trajectory were identified (Fig.
3c,d and Supplementary Table 5), suggesting tentative marker genes
to distinguish RGLs from astrocytes.
The DEG analysis between the RGL_1 and RGL_2 suggests that
the two populations may correspond to the level of quiescence in
the neural stem cell pools. Compared to the RGL_1, several cell sig-
naling and proliferation-related genes are upregulated in the RGL_2
population (Fig. 3g and Supplementary Table 6). These genes include
WIF1 (Wnt inhibitory factor 1) that regulates the Wnt pathway;
RGCC (regulator of cell cycle) that regulates cell cycle and prolif-
eration and participates in balancing self-renew and differentiation
in NSCs53; and TPT1, which encodes tumor protein, translationally
controlled 1 protein for cell proliferation. Of note, one of the unan-
notated genes, ENSMFAG00000043121, is upregulated in RGL_1.
The annotation of this gene in the future may further understand-
ing of the RGL population. The Gene Ontology (GO) and Kyoto
Encyclopedia of Genes and Genomes (KEGG) of the DEGs between
RGL_1 and RGL_2 (Fig. 3g) further highlight the status differences
between the RGL_1 and RGL_2. The upregulated genes in RGL_1
are enriched with protein synthesis and cell–matrix interaction (Fig.
3g, left, and Supplementary Table 7), whereas the RGL_2 popula-
tion is involved in reactive oxygen species (ROS) response54 (per-
oxisome/cysteine and methionine metabolism/HIF-1 signaling
pathway) and de novo lipogenesis (fatty acid metabolism/elonga-
tion) (Fig. 3g, right, and Supplementary Table 7), suggesting that
the RGL_2 population may be in a shallower quiescent status than
the RGL_1.
Identification of proliferating NPCs. Two proliferating and puta-
tive IPC populations (IPC_1 and IPC_2) are identified. The IPC_1
population is aligned to IPCs in developing, juvenile and adult mice
(Fig. 2a–d and Extended Data Fig. 6) and progenitor cells in devel-
oping humans (Extended Data Fig. 7), supporting its identification.
Pathway enrichment analysis (Fig. 1e) reveals that the IPC_1 upreg-
ulated genes (Supplementary Tables 2–3) are involved in the mitotic
cell cycle and DNA replication.
To investigate the possible transition from the IPCs to gran-
ule cells, we analyzed the linkage among the IPC_1, neuroblasts,
and neurons, using mutual k-nearest neighbor (mKNN) in the
Harmony-corrected latent space (Fig. 4a and Methods). In an mKNN
graph, cells are linked by edges when they are each other’s nearest
neighbors, suggesting possible differentiation transitions based on
their gene expression pattern. The mKNN analysis recovers the tra-
jectory from IPC_1 to immature GCs through neuroblasts, whereas
there are hardly any edges between neuroblasts and pyramidal neu-
rons. The mKNN graphs visualize the expression profile of important
genes along the differentiation trajectory, including the expression of
MKI67 (a proliferation marker gene), SOX2 (a neural precursor cell
marker gene), ASCL1 (a transcription regulator-encoding gene for
Fig. 2 | Validation of macaque hippocampus cell populations using annotated datasets. a, UMAP visualization of the integration between the macaque
(n = 15,878 cells) and mouse (n = 9,810 cells, Hochgerer_1 dataset) datasets using scVI followed by the Harmony algorithm. b, Cell type homologies
between macaques and mice, predicted based on the shared cluster membership. The color of each cell represents the alignment score between mouse
and macaque cells. The larger value (darker in the heatmap) indicates a better alignment. Rows show macaque populations, and columns show mouse
populations. Colorbar indicates the alignment score. c, UMAP visualization of macaque cell annotations using the same UMAP coordinates in a, stained
with the same color scheme as in Fig. 1b. d, UMAP visualization of mouse cell annotations using the same UMAP coordinates in a. Labels in mouse
UMAP: Astro_im, immature-Astro; CR, Cajal–Retzius; GABA_im, immature_GABA; GC_im, immature_GC; GC_juv, GC_juvenile; NB, neuroblast; nIPC_p,
nIPC-perin; Pyr_im, immature-Pyr; RGL_yng, RGL_young. e, The integration between the macaque neural genesis cells and neurons (n = 800 cells) and the
mouse NPCs (n = 132 cells). Colorbar indicates the alignment score. f, The integration between the macaque (n = 15,878 cells) and the human embryonic
cells (n = 6,323 cells). Colorbar indicates the alignment score.
NATURE NEUROSCIENCE | www.nature.com/natureneuroscience
ResouRce
NATuRE NEuROSCIENCE
Astro_juv
Astro_adu
Ependymal
Pyr
OPC
NFOL
MOL
CR
nIPC_p RGL
Astro_im
RGL_yng
NB
Pyr_im
Microglia
nIPC
VLMC
PVM
GC_im
GC_adu
GC_juv
Endothelial
GABA_im
GABA
CR
VLMC Endothelial
PVM
Microglia
Unk
IPC_2
Astro_im
RGL
Astro
Ependymal
IPC_1
GABA
Pyr
OPC
Pre_OPC
NFOL
MOL
GC
GC_im
NB
a
Macaque Mouse
UMAP2
UMAP1 1.0
0
Mouse_cluster
RGL_yng
RGL
Astro-adu
Astro-juv
Astro_im
nIPC-p
nIPC
NB
Pyr_im
GC_im
GC-juv
GC-adu
GABA_im
GABA
Pyr
OPC
NFOL
MOL
Microglia
CR
Endothelial
Ependymal
PVM
VLMC
RGL
Astro
Astro_im
IPC_1
IPC_2
NB
GC_im
GC
GABA
Pyr
Pre_OPC
OPC
NFOL
MOL
Microglia
CR
Endothelial
Ependymal
PVM
Unk
RGL
Astro
Astro_im
IPC_1
IPC_2
NB
GC_im
GC
GABA
Pyr
OPC
NFOL
MOL
MG
CR
Endothelial
Ependymal
PVM
b
d
c
Macaque_cluster
S1 S2 S3 S4 S5 SA
RGL_1
IPC_2
IPC_1
NB
GC_im
GC_1
GABA
Pyr_1
0
1.0
Macaque_cluster
Mouse_cluster
e
Shin et al. 2015, adult mouse Nes-CFP
RGL
IPC_2
IPC_1
Hochgerner et al. 2018, mouse E16.5 - p132
NPC1
Astro
ExN
InN
OPC
Oligo
MG
Endo
RGL
IPC_1
NB
IPC_2
Astro_im
Astro
Pyr
GC_im
GC
GABA
OPC
NFOL
Microglia
Endothelial
Macaque_cluster
0
1.0
Human_cluster
Li et al. 2018 Human 5–20 PCW
RGL
IPC_1
NB
IPC_2
Astro_im
Astro
Pyr
GC_im
GC
GABA
OPC
NFOL
Microglia
Endothelial
NPC2
NPC3
NPC4
f
RGL
CR
PVM
VLMC
NB
Microglia
Endothelial
Ependymal
Macaque
OPC
NFOL
MOL
GABA
Pyr
Astro
GC
GC_im
Astro_im
IPC_1
IPC_2
Pre_OPC
Unk
RGL
CR
PVM
VLMC
NB
Microglia
Endothelial
Ependymal
Mouse
OPC
NFOL
MOL
GABA
Pyr
RGL_yng
Astro_juv
Astro_adu
nIPC_p
GC_juv
GC_adu
GABA_im
nIPC
Pyr_im
GC_im
Astro_im
NATURE NEUROSCIENCE | www.nature.com/natureneuroscience
ResouRce NATuRE NEuROSCIENCE
0
0.9
e
ML
GCL
Hilus
SGZ
SOX2/GFAP/DAPI SOX2/GFAP/DAPI
f
0
4.0
SLC14A1CALY
a b
c
d
0
1.0
GCL
ML
SGZ
Hilus
Pseudotime
Cell population
Maturation-related genes
2
SLC1A3
SLC1A2
AQP4
WIF1
F3
ENSMFAG00000043121
TPT1
HPCA
0
50
100
150
RGL_2
RGL_1
–2 0
log2 fold change
–log
10
P
RGL_2 RGL_1
g
EMP1
DCN
SOX7
CENPF
ZNHIT6
TNNT2
SYNPR
ID3
CSRNP1
SCIN
MACF1
ID3 ACSL6
Cytokine binding
Protein targeting to ER
Cytosolic ribosome
Structural constituent of ribosome
Ubiquitin protein transferase regulator
Regulation of actin cytoskeleton
Anchoring juction
Focal adhesion
Cell adhesion molecule binding
Extracellular organelle
Fatty acid elongation
Fatty acid metabolism
Peroxisome
Cysteine and methionine metabolism
Biosynthesis of amino acids
Propanoate metabolism
Carboxylic acid metabolic process
HIF-1 signaling pathway
PPAR signaling pathway
Pluripotency of stem cells pathway
–log10 FDR –log10 FDR
0 >3015 0 5 10
RGL_1
RGL_2
Astro_1
0
4.0
0
3.0
0
5.0
Astro_im2
Astro_im1
Fig. 3 | RGLs and astrocytes are distinct populations. a, The highlighted RGL and astrocyte maturation-related populations are re-visualized as a UMAP
(RGL_1, n = 1,107 cells; RGL_2, n = 1,246; Astro_im1, n = 1,661; Astro_im2, n = 698; Astro_1, n = 12,322), whereas the velocities are visualized as arrows.
b, The UMAP visualization is colored according to the scVelo pseudotime from early (blue) to late (red) and the marker genes for different maturation
stages of the astrocytes. Colorbar, pseudotime. c, The heatmap shows the expression of the pseudotime-related genes as Viridis (light yellow as a
high expression), whereas the colorbar on top shows the cells in the astrocyte maturation stages as in a. Selective genes are labeled to the right of the
heatmap. Colorbar indicates gene expression level. d, The UMAP visualization is colored according to the genes that distinguish immature astrocytes
from astrocytes. Colorbar indicates the log-normalized gene expression level. e, Immunostaining of the RGL markers (SOX2 and GFAP) in DG reveals the
long apical processes (white box) branching into the molecular layer. Scale bar, 20 μm. n = 8 brains. f, Immunostaining of GFAP reveals astrocytes (yellow
arrows) in hilus with a morphology that is distinct from RGLs. Scale bar, 20 μm. n = 8 brains. g, 5,575 DEGs between RGL_1 and RGL_2. These genes are
visualized using the volcano plot, and the selected GO and KEGG terms that are significantly (Fisher’s exact test implemented in g:Profiler; FDR-adjusted
P value for each term was indicated as the color-coded squares) enriched for the RGL_1 upregulated (left) and RGL_2 upregulated (right) genes are listed.
ER, endoplasmic reticulum.
NATURE NEUROSCIENCE | www.nature.com/natureneuroscience
ResouRce
NATuRE NEuROSCIENCE
IPC marker), EOMES (an IPC and neuroblast marker that indicates
the neuronal fate), NEUROD2 (a transcription regulator-encoding
gene specifying the neuronal fate) and PROX1 (a canonical GC
marker). This is consistent with the neurogenesis process from cell
proliferation to neuron specification (Fig. 4b). Additional DEGs for
each population are shown in Fig. 4c. Co-immunostaining of neu-
ral precursor marker SOX2 and proliferating marker MKI67 in the
brain sections captures the cell division process before, during and
after the nuclear division, further validating the existence of prolifer-
ating NSCs in adult primates (Fig. 4d).
The IPC_2 is another proliferating population (Fig. 5). A sub-
set of these cells express EZH2, encoding the neural fate epigen-
etic regulator (Fig. 5a), inferring that their fate is shifting from
stem cell maintenance to differentiation55. The IPC_2 population
has a low expression level of the neuronal fate specification tran-
scription factor gene ASCL1 (Fig. 1d), distinguishing it from IPC_1
population30. It is aligned to the mouse IPCs (Fig. 2a–d) but distinct
from IPC_1 in the macaque datasets (Fig. 1b,d). Pathway enrich-
ment analysis (Fig. 1e and Supplementary Tables 2 and 3) suggests
that the IPC_2 cells are characterized by the protein synthesis and
MDM2/MDM4 protein binding that involves the cell cycle check-
point control, together with the metabolic pathways, such as NADH
dehydrogenase activity, oxidoreductase activity and cellular nitro-
gen compound metabolic process. These functions suggest that
IPC_2 can be in a different status from the actively proliferating
IPC_1 population.
a b
Pyr_2
0
4.0
0
2.0
0
1.5
0
3.0
GC_im NB IPC_1
Pyr_1
IPC_1
GC_im
MKI67
ASCL1 EOMES
NEUROD2 PROX1
0
0.2
0.4
0.6
0.8
dSOX2/MKI67/DAPI
SGZ
GCL
SGZ
GCL
SOX2/MKI67/DAPI SOX2/MKI67/DAPI
SGZ
GCL
SGZ
GCL
SOX2/MKI67/DAPI
SOX2
0
3.0
0
3.0
NB
c
BCAN
ANXA5
PTN
CCND1
RGCC
NNAT
TUBB2B
MARCKSL1
RPS13
TMSB10
NRN1
CALB1
LAMP5
TNNT2
C1QL3
SNAP25
SGZ
GCL
Fig. 4 | Dynamic neurogenesis processes in the adult macaque hippocampal DG. a, The neurogenic-related population and the immature GCs, which
are highlighted in the UMAP visualization at the top, are shown with the mKNN graph plot (bottom). IPCs (n = 163 cells, indicated by the black circles)
link to neuroblasts (n = 1,904 cells, indicated by the color of mustard) and GC_im cells (n = 7,547, indicated by burgundy), whereas pyramidal neurons
are separated (Pyr_1 and Pyr_2, n = 1,384 and 123, respectively, indicated by turquoise). Nodes represent the cells. The edges represent the mKNNs.
b, mKNN graphs, as in a, stained for the expression of genes that are highly expressed in proliferating cells (MKI67), NPCs (SOX2), IPCs (ASCL1),
neuroblasts (EOMES and NEUROD2) and immature GCs (PROX1). Colorbar indicates the log-normalized gene expression level. c, The heatmap illustrates
the expression of the marker genes that are highly expressed in each cell population. The x axis represents the genes, whereas the y axis represents the
cells. Colorbar, log-normalized gene expression. d, Immunostaining validates the existence of proliferating precursor cells in the SGZ of adult macaques,
capturing the process before (left panel), during (middle panels), and after (right panel) the cell division. SOX2, precursor cell marker; MKI67, proliferating
cell marker; DAPI, nuclei. Scale bars, 20 μm (low magnification) and 10 μm (high magnification). n = 5 brains.
NATURE NEUROSCIENCE | www.nature.com/natureneuroscience
ResouRce NATuRE NEuROSCIENCE
IPCs can be identified by HMGB2 (Figs. 1d and 5b), which
encodes a chromatin-related transcriptional activator that is crucial
for the transition of NSCs from quiescence to proliferative status56.
Immunostaining results of HMGB2, GFAP and SOX2 are consis-
tent with the existence of a subset of GFAPLowHMGB2High NPCs.
Within the same section (Fig. 5c,d), we captured several putative
NPC types, including a typical RGL that is GFAP+HMGB2− with
radial processes, a GFAPLowHMGB2High, a SOX2+ NPC and two cells
with both GFAP and HMGB2 expression at intermediate levels. The
HMGB2HighGFAPLow and HMGB2LowGFAPHigh cells can be found
close to each other (Fig. 5e) or even within the same proliferating
clusters (Fig. 5f), further supporting the transition between RGLs
and IPCs.
Dissecting the GC genesis. Neuroblasts are the first differen-
tiation step after exiting the cell cycles30. The macaque neuroblast
population is identified by its transcriptomic similarity to mouse
neuroblast (Fig. 2 and Extended Data Fig. 6) or progenitor cells in
developing humans (Fig. 2 and Extended Data Fig. 7). The RNA
velocity field embedded on the UMAP suggests a GC specification
trajectory, from neuroblasts to the immature GCs and eventually
to the mature GCs (Fig. 6a). Resolving the transcriptional cascade
suggests the putative driver genes (Fig. 6b and Supplementary Table
8) and further visualizes each step of the GC maturation process
(Fig. 6e).
To profile the differences between mature and immature GCs,
we performed a DEG analysis between the immature GC popula-
tion (GC_im, n = 7,547) and the largest mature GC population
(GC_1, n = 23,491) (Fig. 6c and Supplementary Table 9). The path-
way enrichment analysis reveals that the upregulated genes for the
mature GCs are involved in the neuron functions, such as chemi-
cal synaptic transmission, neuron projection morphogenesis and
SNARE binding pathways. This is consistent with the function of
mature GCs. On the contrary, the top enriched pathways for imma-
ture GCs include protein targeting the endoplasmic reticulum, oxi-
dative phosphorylation, the ephrin receptor signaling pathway and
Alzheimer’s disease. These terms are consistent with the biologi-
cal process during neuron migration and maturation (Fig. 6c and
Supplementary Table 10). The adult macaque hippocampus has a
substantially higher proportion of immature GCs than that in the
adult mouse hippocampus30. Co-immunostaining of neuroblast and
newborn neuron marker DCX and GC marker PROX1 confirms
the abundance of immature GCs in macaques (Fig. 6d). This abun-
dance of immature GCs may be explained by the prolonged matu-
ration period, reflecting the species divergence between mice and
macaques24,57,58. Still, it is likely that some of these immature GCs
are not newborns while still maintaining a higher level of plasticity.
Further study using scVelo-derived driver genes may help further
elucidate the GC maturation process during adult neurogenesis.
A continuous molecular cascade of adult macaque neurogen-
esis. After the major neurogenic-related populations were estab-
lished, we tried to reconstruct the whole neurogenic lineages that
cover all major neurogenic populations (Extended Data Fig. 8).
After removing the cell-cycle-related genes, these populations were
reclustered and formed a continuous lineage that recapitulates the
neurogenic process (Extended Data Fig. 8a,b and Methods). The
scVelo analysis inferred a lineage from RGL to IPC, via neuroblast,
to GC_im (Extended Data Fig. 8c,d, Supplementary Table 11 and
Supplementary Information). This lineage relationship is consistent
with that inferred by Slingshot (Extended Data Fig. 8e,f), providing
additional support for the neurogenic process in adult macaques.
Divergent gene expression between rodents and primates. To
explore the tentative differences between the macaque and mouse
neurogenic cells, we compared the marker gene expression pat-
terns between the two species (Extended Data Fig. 9). Many
canonical neural progenitor markers are conserved across spe-
cies, such as SLC1A3, SOX2, SOX4, and DPYSL3 (Extended Data
Fig. 9a), or enriched only in macaques or mice (Extended Data
Fig. 9b). Furthermore, we compared the expression of macaque
neurogenic-related genes (n = 2,081; Methods) to those in adult
mice30. For each neurogenic population, the comparison revealed
multiple genes that had at least two-fold changes between the two
species (Extended Data Fig. 9c). For RGLs (Extended Data Fig.
9c), 43 genes were increased in macaques than in mice, including
FOS, PPP3CA, RTN4, CHN1 and YWHAH, whereas 12 genes were
higher in mice, including MT3, DBI, GPM6B, ATP1B2, and DCLK1.
Similarly, the comparisons for the IPCs, neuroblasts, and GCs also
revealed multiple DEGs (Extended Data Fig. 9c). These divergences
in the gene expression profiles suggest the differences in the neuro-
genic processes between rodents and primates.
Ex vivo neurosphere supports the existence of NSCs. To fur-
ther confirm the presence of NSCs in the hippocampus of adult
macaques, we sought ex vivo evidence by culturing hippocampal
cells in an NSC medium. During the culture, a small number of cells
with distinct morphologies were found to adhere to the culture dish
(Fig. 7a and Extended Data Fig. 10a), proliferated for many cycles
and formed a neurosphere in 3–4 weeks (Fig. 7b–e and Extended
Data Fig. 10b–d) and continued to expand (Fig. 7f). We seeded
these neurospheres in separate plate wells and found them gradu-
ally growing into cells with NSC morphology. Notably, they could
be further expanded and passaged after dissociation into single
Fig. 5 | HMGB2 correlates with the transition between RGL and IPCs. a, UMAP visualization of the IPCs and the expression level of proliferating marker
MKI67 and epigenetic regulator gene EZH2. The IPCs are indicated by the black circles. Colorbars indicate the log-normalized gene expression level. b,
Violin plot of GFAP and HMGB2 gene expression distribution of RGLs and IPCs. c, Immunostaining of the IPC marker HMGB2 and the canonical RGL marker
GFAP capture cell populations consistent with the possible transition between RGLs and IPCs. Box 1 and white arrowheads: a typical RGL with strong
GFAP expression but no HMGB2 expression. Box 2 and yellow arrowheads: an intermediate cell with weak GFAP and weak HMGB2 expression. Box 3 and
magenta arrowheads: an HMGB2+ cell with strong HMGB2 expression but very weak GFAP expression. Scale bar, 20 μm. n = 4 brains. d, The schematic
illustration of the three cell types shown in the boxes and their counterpart of RGLs (Box 1 and the green cell, GFAP+HMGB2−), IPCs (Box 3 and the red
cell, GFAP−HMGB2+) and an intermediate state between RGLs and IPCs (Box 2 and the light green cell, GFAP+HMBG2+). e, Immunostaining of HMGB2
and GFAP reveals a GFAP−HMGB2+ IPC cell (magenta arrowheads) and a cell at the transition stage with the typical RGL morphology (yellow arrowheads)
that express both HMGB2 and GFAP. Compared to the IPC cell, the transition stage cell has a lower HMGB2 expression level and a higher GFAP expression
level. Tissue collected from a different animal as in c. Scale bar, 25 μm. Schematic cells on the right illustrate the cell type and their gene expression pattern
as in d. White dashed circles indicate the cell bodies of the cells of interest. n = 4 brains. f, Immunostaining of HMGB2 and GFAP reveals a proliferating
cluster containing four precursor cells: a GFAPHighHMGB2Low RGL (white arrowhead), a GFAPLowHMGB2High IPC (magenta arrowheads) and two cells at the
transition stage with an intermediate level of both GFAP and HMGB2 expression (yellow arrowheads). Tissue collected from the same animal as in c. Scale
bar, 20 μm. Schematic cells on the right illustrate the cell type and their gene expression pattern. White dashed circles indicate the cell bodies of the cells
of interest. n = 4 brains.
NATURE NEUROSCIENCE | www.nature.com/natureneuroscience
ResouRce
NATuRE NEuROSCIENCE
cells. At about passage 4, most neurosphere-derived cells grew into
a monolayer of typical NSCs (Fig. 7g–j).
To confirm the NSC identity of neurospheres, we profiled the
transcriptomes of 6,988 neurosphere cells by scRNA-seq. In the
UMAP visualization, we identified cells with elevated expression
of NSC markers NES, VIM, and PAX6, the proliferating marker
MKI67, and the proliferating neural progenitor marker HMGB2
(Fig. 7k and Extended Data Fig. 10e). The population adjacent
HMGB2
ML
GCL
SGZ
GFAP
a b
IPC MKI67 EZH2
Normalized expression
GFAP
0
2.0
4.0
1.0
3.0
5.0
HMGB2
RGL
IPC_1
0
2.0
4.0
1.0
3.0
5.0
c
d
0
4.0
123
HMGB2/SOX2/GFAP/DAPI
1
2
3
RGL Intermediate IPC
IPC_2
DAPI SOX2HMGB2 GFAP Merge
DAPI SOX2HMGB2 GFAP Merge
e
f
GCL
SGZ
GCL
SGZ
IPC
GFAP–HMGB2+
Intermediate
GFAP+HMGB2+
RGL
GFAP+HMGB2–
0
3.0
NATURE NEUROSCIENCE | www.nature.com/natureneuroscience
ResouRce NATuRE NEuROSCIENCE
to the proliferating cells expresses neuroblast markers NNAT and
SOX4 and immature neuron marker TUBB3 (TUJ1) (Extended
Data Fig. 10e). Consistent with the scRNA-seq data, neurosphere
cells were immunoreactive for NESTIN, PAX6, MKI67, HMGB2,
VIMENTIN, NNAT, and SOX4 (Fig. 7l–p and Extended Data Fig.
10f–h). In addition, a small number of cells in the neurosphere
were labeled with both TUJ1 and 5-ethynyl-2′-deoxyuridine (EdU)
(Extended Data Fig. 10i). We asked whether neurosphere cells had
the potential to differentiate into neurons, astrocytes and oligoden-
drocytes as typical NSCs would, and we found that, when induced
in proper culture conditions, they were able to differentiate into
cells that were immunoreactive for neuron marker MAP2, astro-
cyte marker GFAP or oligodendrocyte marker CNPase (Fig. 7q–s).
These results indicate that the neurosphere cells derived from the
adult primate hippocampus have differentiation potencies.
To compare the transcriptome of ex vivo neurospheres and
in vivo cells, we applied SCCAF, a logistic-regression-based machine
learning algorithm (Methods), after data integration with Harmony.
A cell type classifier was trained with the in vivo macaque monkey
data and projected to the ex vivo data (Fig. 7t). This algorithm clas-
sified each ex vivo cell to its most similar cell type from the in vivo
data. The projection predicted the MKI67+ precursor population as
a combination of IPCs and neuroblasts. This population differenti-
ates into several clusters, including the SLC1A3+ and PAX6+TUJ1+
populations that are similar to the astrocytes and immature GCs,
respectively. The projection from the mice yields similar results
(Extended Data Fig. 10j). These results revealed the transcriptomic
differences among the ex vivo cells, which supports the existence of
proliferating precursor cells in the hippocampus of adult macaques,
with differentiation potencies. Our ex vivo cell culture method can
NB GC_im
GC_1
0
1.0
DEGs
GC_im GC
Protein targeting to ER
Oxidative phosphorylation
Ephrin receptor signaling
Extracellular space
Alzheimer disease
Chemical synaptic transmission
Neuron projectin morphogenesis
Regulation of ion transport
SNARE binding
Synaptic signaling
1.5–1.5
Normalized z-score SGZ
GCL
DAPI DCX PROX1 Merge
SGZ
GCL
ML
d
0
2.0
TMEM159 NOVASAP2 COTL1
a b
c
e
NADH dehydrogenase activity
Oxidoreductase activity
Cell adhension
Synapes
0
2.0
0
1.5
0
1.5
Fig. 6 | The maturation processes of the GCs in the adult macaque hippocampal DG. a, The GC genesis-related populations (neuroblast, n = 1,898 cells;
GC_im, n = 7,547 cells; and GC_1, n = 23,491 cells, highlighted in the UMAP visualization) are re-analyzed with UMAP, whereas the RNA velocities are
visualized as arrows. b, The heatmap of tentative driver genes during GC maturation process derived by RNA velocity. Colorbar on top of the heatmap
represents the cell types, using the same color scheme as in a. c, The normalized mean z-core of the DEGs is grouped by the immature GC (GC_im) and
mature GC (GC), with red representing upregulation and blue representing downregulation. Selective pathway enrichment terms for the immature and
mature GC DEGs are labeled at the right of the heatmap, respectively. d, Immunostaining validates the existence of immature GCs in the SGZ. DCX,
immature neuron marker; PROX1, GC marker; DAPI, nuclei. Scale bar, 20 μm. n = 5 brains. e, The UMAP visualization of the novel marker gene expression.
The marker genes are for neuroblasts (ASAP2 and TMEM159), immature GCs (COTL1) and mature GCs (NOV). Colorbar indicates the log-normalized gene
expression level. ER, endoplasmic reticulum; NB, neuroblast.
NATURE NEUROSCIENCE | www.nature.com/natureneuroscience
ResouRce
NATuRE NEuROSCIENCE
k
PAX6
HMGB2
MKI67
NES
0
2.0
IPC_1 NB GC_im GC_1 GC_2
RGL_2 Astro_1 Astro_2
SCCAF projection
Astro_im1 Astro_im2
NESTIN
HMGB2
MKI67
PAX6 CNPaseMAP2
EdU/NESTIN
GFAP
l m n o
p q r s
t
D3 D14 D21 D30
a b c d
P1 P2 P3 P4
f g h
e
ij
P0
D30
0
2.5
0
2.5
0
0.5
Fig. 7 | Morphology and gene expression of cultured hippocampal neurospheres. a–j, In vitro recording process of two cultured neurosphere
clones from the adult macaque hippocampi. Scale bars, a–d and f, g, 40 μm; e, 160 μm; h–j, 80 μm. n = 4 clones. D, day; P, passage. k, UMAP
visualization of the cultured sphere cells (n = 6,988 cells), stained by the expression of selected marker genes. NES, RGL marker; HMGB2, IPC
marker; MKI67, proliferating cell marker; PAX6, neural progenitor marker. Colorbar indicates the log-normalized gene expression level. i–s,
Immunostaining analysis of EdU long-term labeling of the cultured neurospheres from most cells were labeled by fluorescence for both NESTIN
and EdU in m. The cultured neurosphere cells from macaque hippocampus were immunoreactive for NSC marker NESTIN in i, proliferating neural
progenitor marker HMGB2 in n, proliferating cell marker MKI67 in o, neural progenitor marker PAX6 in p, neuronal marker MAP2 in q, astrocyte
marker GFAP in r and oligodendrocyte marker CNPase in s. Scale bars, 40 μm. n = 4 clones for each immunostaining combination. t, Left: The
UMAP visualization demonstrates the genetic heterogeneity of cultured sphere cells that proliferated from isolated adult macaque hippocampal cells.
The colors represent the regression-based SCCAF projection results from the macaque, using the same color scheme as in Fig. 1d. Right: The UMAP
visualization stained by the SCCAF projection results in selective populations. Red dots represent the distribution of the predicted cell types. Astro,
astrocyte; NB, neuroblast.
NATURE NEUROSCIENCE | www.nature.com/natureneuroscience
ResouRce NATuRE NEuROSCIENCE
maintain the self-proliferating neural progenitor population for a
long time while keeping its differentiation potency. To our knowl-
edge, this is the first ex vivo system that captures the neurogenesis
process at the single-cell transcriptome level, providing additional
evidence for the existence of NPCs in the adult primate hippocam-
pus DG with the potency of proliferation and differentiation.
Discussion
Using scRNA-seq, we profiled the cell types in the adult macaque
hippocampus and identified key NPCs that are hallmarks of adult
hippocampal neurogenesis. The existence of RGLs in quiescence
status and the IPC population that is actively proliferating pro-
vides strong support for the robustness of neurogenesis in the adult
primate hippocampus. Although the proportions of RGLs, IPCs,
and neuroblasts in our datasets are similar to rodents (Extended
Data Fig. 9d)30, major differences between the two species exist.
Comparison with mouse single-cell transcriptomic data revealed
that the marker gene expression patterns in macaques are substan-
tially different from those of rodents. The adult macaque hippo-
campus has a substantially higher proportion of immature GCs than
that in the adult mouse hippocampus30, further suggesting the dif-
ferences in the neurogenic process between primates and rodents.
It is unclear whether and how the adult animals maintain their
NSC pools59. The existence of the self-renewable RGL population,
which prevents the NSC pools from exhaustion, has been controver-
sial44,45,60. We find a distinct population, the HMGB2+ IPC_2, that is
proliferating and may participate in the process. In contrast to the
IPC_1 that is dedicated to mitosis, the IPC_2 is complicated. The
expression of MKI67 and the GRN MYBL2 suggest that the IPC_2
is involved in proliferation. This population is also enriched in GRN
FOXM1 crucial for NSC self-renewal61 and MEF2C that involves in
the NSC differentiation (Extended Data Fig. 2). Consistently, scVelo
(Extended Data Fig. 8c) infers that IPC_2 gives rise to IPC_1.
Lacking the lineage tracing validation, it is too early to identify the
exact function and the lineage hierarchy of IPC_2. Using the marker
genes ASCL1 and HMGB2 may help.
Echoing the challenge in the immunostaining validation of
the neurogenesis cells in humans using rodent-derived marker
genes17,18,40, the marker gene expression patterns in macaques are
substantially different from those of rodents (Extended Data Fig. 9
and Supplementary Information). For example, compared to NPCs
from developing hippocampus, the expression level of classical neu-
rogenesis marker genes at the transcriptomic level, such as NES or
DCX, are low in adult primates. Instead, SLC1A3 (also known as
GLAST) works better than NES or other canonical developmental
NSC markers28. In adult rodents, Slc1a3-derived lineage contributes
more to the adult neurogenesis than Nestin lineage does62. Validating
and using marker genes that are specifically picked from adult cell
populations may provide additional information to understand
adult neurogenesis. Although this is consistent with another recent
publication focusing on the adult primates, pigs, and mice40, valida-
tion and interrogation of these populations may provide extra sup-
port for this divergence across species. The consistent results from
integration between our data with mice (Fig. 2 and Extended Data
Fig. 6) and developing humans (Fig. 2 and Extended Data Fig. 7),
however, suggest that the fundamental molecular architecture can
be comparable across species.
Our description of primatesʼ neurogenesis is largely based on the
knowledge accumulated from rodent research. Future research on
the species divergence, and the extension of our understanding of
adult neurogenesis in general, may fine-tune our annotation. For
example, the altered gene expression pattern of the IPC_1 suggested
the differences in the molecular cascade during the neurogenesis
process across species. The proliferating population IPC_2, which
forms a distinct cluster from the IPC_1, may reflect transitions dur-
ing the complicated neurogenesis process in adult primates. Further
interrogation of the real nature of these key populations may deepen
our understanding of adult neurogenesis. It can be crucial for the
development of therapeutic approaches to treat currently incurable
brain disorders, such as stroke, visual impairment, and Alzheimer’s
and Parkinson’s diseases63. The key point may be that, if adult pri-
mate neurogenesis exists, it will be considerably different from the
other species. Comparative studies between primates and rodents
may be necessary for a further understanding of underlying mecha-
nisms for adult neurogenesis in primates.
Online content
Any methods, additional references, Nature Research report-
ing summaries, source data, extended data, supplementary infor-
mation, acknowledgements, peer review information; details of
author contributions and competing interests; and statements of
data and code availability are available at https://doi.org/10.1038/
s41593-022-01073-x.
Received: 24 August 2021; Accepted: 7 April 2022;
Published: xx xx xxxx
References
1. Gonçalves, J. T., Schafer, S. T. & Gage, F. H. Adult neurogenesis in the
hippocampus: from stem cells to behavior. Cell 167, 897–914 (2016).
2. Kaplan, M. S. & Hinds, J. W. Neurogenesis in the adult rat: electron
microscopic analysis of light radioautographs. Science 197, 1092–1094 (1977).
3. Doetsch, F., Caillé, I., Lim, D. A., García-Verdugo, J. M. & Alvarez-Buylla, A.
Subventricular zone astrocytes are neural stem cells in the adult mammalian
brain. Cell 97, 703–716 (1999).
4. Altman, J. & Das, G. D. Autoradiographic and histological evidence of
postnatal hippocampal neurogenesis in rats. J. Comp. Neurol. 124, 319–335
(1965).
5. Yang, C.-H. et al. Circuit integration initiation of new hippocampal neurons
in the adult brain. Cell Rep. 30, 959–968 (2020).
6. Song, J., Olsen, R. H. J., Sun, J., Ming, G.-L. & Song, H. Neuronal circuitry
mechanisms regulating adult mammalian neurogenesis. Cold Spring Harb.
Perspect. Biol. 8, a018937 (2016).
7. Jessberger, S. et al. Dentate gyrus-specic knockdown of adult neurogenesis
impairs spatial and object recognition memory in adult rats. Learn. Mem. 16,
147–154 (2009).
8. Kempermann, G., Georg Kuhn, H. & Gage, F. H. More hippocampal neurons
in adult mice living in an enriched environment. Nature 386, 493–495 (1997).
9. van Praag, H., Kempermann, G. & Gage, F. H. Running increases cell
proliferation and neurogenesis in the adult mouse dentate gyrus. Nat.
Neurosci. 2, 266–270 (1999).
10. Livneh, Y., Adam, Y. & Mizrahi, A. Odor processing by adult-born neurons.
Neuron 81, 1097–1110 (2014).
11. Eriksson, P. S. et al. Neurogenesis in the adult human hippocampus. Nat.
Med. 4, 1313–1317 (1998).
12. Spalding, K. L. et al. Dynamics of hippocampal neurogenesis in adult
humans. Cell 153, 1219–1227 (2013).
13. Gould, E. et al. Hippocampal neurogenesis in adult Old World primates. Proc.
Natl Acad. Sci. USA 96, 5263–5267 (1999).
14. Kornack, D. R. & Rakic, P. Continuation of neurogenesis in the hippocampus
of the adult macaque monkey. Proc. Natl Acad. Sci. USA 96, 5768–5773
(1999).
15. Yuan, T.-F., Li, J., Ding, F. & Arias-Carrion, O. Evidence of adult
neurogenesis in non-human primates and human. Cell Tissue Res. 358, 17–23
(2014).
16. Ngwenya, L. B., Heyworth, N. C., Shwe, Y., Moore, T. L. & Rosene, D. L.
Age-related changes in dentate gyrus cell numbers, neurogenesis, and
associations with cognitive impairments in the rhesus monkey. Front. Syst.
Neurosci. 9, 102 (2015).
17. Sorrells, S. F. et al. Human hippocampal neurogenesis drops
sharply in children to undetectable levels in adults. Nature 555, 377–381
(2018).
18. Moreno-Jiménez, E. P. et al. Adult hippocampal neurogenesis is abundant in
neurologically healthy subjects and drops sharply in patients with Alzheimer’s
disease. Nat. Med. 25, 554–560 (2019).
19. Kempermann, G. et al. Human adult neurogenesis: evidence and remaining
questions. Cell Stem Cell 23, 25–30 (2018).
20. Tobin, M. K. et al. Human hippocampal neurogenesis persists in aged adults
and Alzheimer’s disease patients. Cell Stem Cell 24, 974–982 (2019).
21. Boldrini, M. et al. Human hippocampal neurogenesis persists throughout
aging. Cell Stem Cell 22, 589–599 (2018).
NATURE NEUROSCIENCE | www.nature.com/natureneuroscience
ResouRce
NATuRE NEuROSCIENCE
22. Terreros-Roncal, J. et al. Impact of neurodegenerative diseases on human
adult hippocampal neurogenesis. Science 374, 1106–1113 (2021).
23. Aizawa, K., Ageyama, N., Terao, K. & Hisatsune, T. Primate-specic
alterations in neural stem/progenitor cells in the aged hippocampus.
Neurobiol. Aging 32, 140–150 (2011).
24. Seki, T. Understanding the real state of human adult hippocampal
neurogenesis from studies of rodents and non-human primates. Front.
Neurosci. 14, 839 (2020).
25. Curtis, M. A. et al. Human neuroblasts migrate to the olfactory bulb via a
lateral ventricular extension. Science 315, 1243–1249 (2007).
26. Sanai, N. et al. Corridors of migrating neurons in the human brain and their
decline during infancy. Nature 478, 382–386 (2011).
27. Snyder, J. S. Recalibrating the relevance of adult neurogenesis. Trends
Neurosci. 42, 164–178 (2019).
28. Harris, L. et al. Coordinated changes in cellular behavior ensure the lifelong
maintenance of the hippocampal stem cell population. Cell Stem Cell 28,
863-876 (2021).
29. Artegiani, B. et al. A single-cell RNA sequencing study reveals cellular and
molecular dynamics of the hippocampal neurogenic niche. Cell Rep. 21,
3271–3284 (2017).
30. Hochgerner, H., Zeisel, A., Lönnerberg, P. & Linnarsson, S. Conserved
properties of dentate gyrus neurogenesis across postnatal development
revealed by single-cell RNA sequencing. Nat. Neurosci. 21, 290–299 (2018).
31. Huang, W. et al. Origins and proliferative states of human oligodendrocyte
precursor cells. Cell 182, 594–608 (2020).
32. Kriegstein, A. & Alvarez-Buylla, A. e glial nature of embryonic and adult
neural stem cells. Annu. Rev. Neurosci. 32, 149–184 (2009).
33. Shin, J. et al. Single-cell RNA-seq with waterfall reveals molecular cascades
underlying adult neurogenesis. Cell Stem Cell 17, 360–372 (2015).
34. Suh, H. et al. In vivo fate analysis reveals the multipotent and self-renewal
capacities of Sox2 neural stem cells in the adult hippocampus. Cell Stem Cell
1, 515–528 (2007).
35. Berg, D. A. et al. A common embryonic origin of stem cells drives
developmental and adult neurogenesis. Cell 177, 654–668 (2019).
36. Llorens-Bobadilla, E. et al. Single-cell transcriptomics reveals a population of
dormant neural stem cells that become activated upon brain injury. Cell Stem
Cell 17, 329–340 (2015).
37. Joglekar, A. et al. A spatially resolved brain region- and cell type-specic
isoform atlas of the postnatal mouse brain. Nat. Commun. 12, 463 (2021).
38. Zhong, S. et al. Decoding the development of the human hippocampus.
Nature 577, 531–536 (2020).
39. Li, M. et al. Integrative functional genomic analysis of human brain
development and neuropsychiatric risks. Science 362, eaat7615 (2018).
40. Franjic, D. et al. Transcriptomic taxonomy and neurogenic trajectories of
adult human, macaque, and pig hippocampal and entorhinal cells. Neuron
110, 452–469 (2022).
41. Eckenho, M. F. & Rakic, P. Nature and fate of proliferative cells in the
hippocampal dentate gyrus during the life span of the rhesus monkey. J.
Neurosci. 8, 2729–2747 (1988).
42. Bergen, V., Lange, M., Peidli, S., Wolf, F. A. & eis, F. J. Generalizing RNA
velocity to transient cell states through dynamical modeling. Nat. Biotechnol.
38, 1408–1414 (2020).
43. La Manno, G. et al. RNA velocity of single cells. Nature 560, 494–498 (2018).
44. Pilz, G.-A. et al. Live imaging of neurogenesis in the adult mouse
hippocampus. Science 359, 658–662 (2018).
45. Bonaguidi, M. A. et al. In vivo clonal analysis reveals self-renewing and
multipotent adult neural stem cell characteristics. Cell 145, 1142–1155
(2011).
46. Bonaguidi, M. A., Song, J., Ming, G.-L. & Song, H. A unifying hypothesis on
mammalian neural stem cell properties in the adult hippocampus. Curr. Opin.
Neurobiol. 22, 754–761 (2012).
47. Miller, J. A. et al. Conserved molecular signatures of neurogenesis in the
hippocampal subgranular zone of rodents and primates. Development 140,
4633–4644 (2013).
48. Denoth-Lippuner, A. & Jessberger, S. Formation and integration of new
neurons in the adult hippocampus. Nat. Rev. Neurosci. 22, 223–236 (2021).
49. Oddi, S., Scipioni, L. & Maccarrone, M. Endocannabinoid system and adult
neurogenesis: a focused review. Curr. Opin. Pharmacol. 50, 25–32 (2020).
50. Hodge, R. D. et al. Conserved cell types with divergent features in human
versus mouse cortex. Nature 573, 61–68 (2019).
51. Gayoso, A. et al. A Python library for probabilistic analysis of single-cell
omics data. Nat. Biotechnol. 40, 163–166 (2022).
52. Korsunsky, I. et al. Fast, sensitive and accurate integration of single-cell data
with Harmony. Nat. Methods 16, 1289–1296 (2019).
53. Guo, Z. et al. RGCC balances self-renewal and neuronal dierentiation of
neural stem cells in the developing mammalian neocortex. EMBO Rep. 22,
e51781 (2021).
54. Adusumilli, V. S. et al. ROS dynamics delineate functional states of
hippocampal neural stem cells and link to their activity-dependent exit from
quiescence. Cell Stem Cell 28, 300–314 (2021).
55. Sun, J., Sun, J., Ming, G.-L. & Song, H. Epigenetic regulation of neurogenesis
in the adult mammalian brain. Eur. J. Neurosci. 33, 1087–1093 (2011).
56. Kimura, A., Matsuda, T., Sakai, A., Murao, N. & Nakashima, K. HMGB2
expression is associated with transition from a quiescent to an activated state
of adult neural stem cells. Dev. Dyn. 247, 229–238 (2018).
57. Augusto-Oliveira, M., Arrifano, G. P. F., Malva, J. O. & Crespo-Lopez, M. E.
Adult hippocampal neurogenesis in dierent taxonomic groups: possible
functional similarities and striking controversies. Cells 8, 125 (2019).
58. Gould, E., Vail, N., Wagers, M. & Gross, C. G. Adult-generated hippocampal
and neocortical neurons in macaques have a transient existence. Proc. Natl
Acad. Sci. USA 98, 10910–10917 (2001).
59. Urbán, N. et al. Return to quiescence of mouse neural stem cells by
degradation of a proactivation protein. Science 353, 292–295 (2016).
60. Bottes, S. et al. Long-term self-renewing stem cells in the adult mouse
hippocampus identied by intravital imaging. Nat. Neurosci. 24, 225–233
(2021).
61. Jia, J., Cui, Y., Tan, Z., Liu, M. & Jiang, Y. Transcriptional factor
FoxM1-activated microRNA-335-3p maintains the self-renewal of neural stem
cells by inhibiting p53 signaling pathway via Fmr1. Stem Cell Res. er. 12,
169 (2021).
62. DeCarolis, N. A. et al. In vivo contribution of nestin- and GLAST-lineage
cells to adult hippocampal neurogenesis. Hippocampus 23, 708–719 (2013).
63. Toda, T., Parylak, S. L., Linker, S. B. & Gage, F. H. e role of adult
hippocampal neurogenesis in brain health and disease. Mol. Psychiatry 24,
67–87 (2019).
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in
published maps and institutional affiliations.
© The Author(s), under exclusive licence to Springer Nature America, Inc. 2022
NATURE NEUROSCIENCE | www.nature.com/natureneuroscience
ResouRce NATuRE NEuROSCIENCE
Methods
Ethical compliance. All experimental procedures were approved by the Animal
Care and Use Committee of Zhongshan Ophthalmic Center, Sun Yat-sen
University. e study was performed in accordance with the Principles for the
Ethical Treatment of Non-Human Primates.
Macaques. Adult macaque monkeys were obtained from Blooming-Spring
Biotechnology Co., Ltd., in Guangdong, China, or were a generous gift from nearby
laboratories for terminal experiments.
Hippocampal tissue preparation. After 2 weeks of adaptive housing in the
institutional animal facility, animals were deeply sedated with isoflurane and then
euthanized with an overdose of pentobarbital. Animals were then transcardially
perfused with artificial cerebrospinal fluid (ACSF, in mM: 125 NaCl, 25 NaHCO3, 1
NaH2PO4, 2 KCl, 25 D-glucose, 2 CaCl2, 1 MgCl2, 5 sodium ʟ-ascorbate, 3 sodium
pyruvate, 0.01 taurine, 2 thiourea, 1 kynurenic acid, 0.1 DL-AP5 and 1 × 10−3
tetrodotoxin). The brain was removed from the skull, and the hippocampus was
obtained64. The tissue block was kept in the ACSF and sectioned on a vibratome
(VT1200S, Leica) in 300-μm-thick slices. The DG and attached hippocampus areas
were dissected and kept on ice for dissociation.
Single whole-cell dissociation. In total, eight macaque monkeys (4–15 years
old, six males and two females) were used. The dissociation solution including
40 U ml−1 of papain (Worthington Biochemical), 20 mg ml−1 of collagenase type II
(Worthington Biochemical), 0.25% trypsin-EDTA (Gibco) and 5 mg ml−1 of DNase
I (Roche) in ACSF warmed at 34 °C for 10 minutes. The tissues were transferred to
dissociation solution and incubated for 60 minutes at 34 °C. After incubation, the
tube was briefly spun, and the supernatant was gently removed. The pieces were
resuspended with 10% FBS (Gibco) in ACSF and filtered through a 70-μm cell
strainer. The debris and dead cells were removed using Myelin Removal Beads II
(Miltenyi) and Dead Cell Removal Kit (Miltenyi) and then washed two times with
0.04% PBS-BSA. Cells were counted (Bio-Rad, TC20) after visual inspection. Using
the optimized dissociation protocol, we obtained high-quality cell suspensions with
the rate of survival >90%.
Generation of scRNA-seq library (10x Genomics). scRNA-seq was performed
using 10x Genomics Chromium platform with Chromium Controller Readiness
Test. The samples were carried out with 10x Genomics Chromium Single Cell
Kit version 3, with a minority of samples done in version 2. The single-cell
suspensions were adjusted to 1,000 cells per μl in 0.04% PBS-BSA and added to
the Chromium RT mix. During the experiments, samples were loaded following
the 10x guideline, taking consideration of the 10x version, cell vitality, cell stock
concentration, and the available cell stock volume to achieve a target capture
of 5,000–10,000 cells. Cell barcoding, cDNA synthesis (12–14 PCR cycles),
library preparation, and quality control were performed according to the
manufacturer’s instructions. Libraries were sequenced on the Illumina NovaSeq
6000 platform.
Immunofluorescence staining. The hippocampal tissues were dissected and fixed
with 4% paraformaldehyde (PFA) for up to 24 hours and cryoprotected in 30%
sucrose at 4 °C for 72 hours. The tissue samples were frozen in optimal cutting
temperature (OCT) compound (Tissue-Tek) at −80 °C and sectioned at 30 μm on
a cryostat microtome (Leica CM1950). Sections were rinsed in PBS, incubated
for 30 minutes in 0.3% Triton X-100 (Sigma-Aldrich) and then for 2 hours in
5% donkey serum (Vector Laboratories). Subsequently, sections were incubated
overnight at 4 °C with the primary antibodies and for 2 hours at room temperature
with the secondary antibodies. The antibodies used included mouse anti-SOX-2,
1:1,000 dilution; chicken anti-glial fibrillary acidic protein, 1:2,000; rabbit
anti-GLAST, 1:1,500;
rabbit anti-Ki67, 1:200; mouse anti-HMGB2, 1:2,000; rabbit anti-Sox2, 1:200;
goat doublecortin (C-18), 1:1,000; mouse anti-prox1, 1:500; rabbit anti-vimentin,
1:1,000; rabbit anti-HOPX, 1:100; mouse anti-nestin, 1:100; and rabbit
anti-MASH1 (ASCL1), 1:100. Sections were mounted with DAPI (Abcam) and
coverslipped. Images were obtained with an LSM880 Zeiss confocal microscope.
Isolation and culture of NSCs from macaque hippocampus. The brain tissue
was removed and collected from 4–6-year-old male macaques (n = 4). The
hippocampi were quickly dissected into a 10-cm dish containing 10 ml of pre-cold
Hanks’ Balanced Salt Solution (HBSS) (HyClone). After washing five times with
cold HBSS, the hippocampi were put in a 10-cm cell-cultured plate containing
10 ml of NSC medium, constituted of N3 medium supplemented with 20 ng ml−1
of recombinant mouse EGF (R&D Systems), 20 ng ml−1 of recombinant human
bFGF (R&D Systems), 2 µg ml−1 of heparin (Sigma-Aldrich) and 1× primocin
(InvivoGen), and incubated in a 5% CO2 humidified cell incubator at 37 °C
for 30 minutes. Preparation of the N3 medium was carried out as previously
described65.
Meninges were removed from the hippocampi under a stereomicroscope. The
remaining tissues were cut into small pieces and enzymatically dissociated first
in 0.25% trypsin-EDTA (Gibco) at 37˚C for 20 minutes and then in 40 U ml−1 of
papain (Worthington Biochemical) and 7.5 mg ml−1 of DNase I (Roche) at 37 °C for
30 minutes. During dissociation, tissues were gently shaken every 10 minutes. After
dissociation, tissue suspension was gently mixed with 1 volume of NSC medium
and pelleted at 180g for 5 minutes. The supernatant was removed, and a fresh NSC
medium was added. Tissues were carefully triturated first with a Pasteur pipette
and then with a 1,000-µl pipette tip. Cells were then filtered through a 70-µm
strainer and plated in a six-well plate pre-coated with 1.25% Matrigel (Corning)
or Poly-L-ornithine and laminin (Sigma-Aldrich), containing NSC medium.
Half of the medium was changed every 2 days. After 2 months of culture, many
neurospheres can be observed using the microscope. When reached sufficient
volume, NSC clones were dissociated into single cells using Accumax (Millipore)
and replated in a 6-cm cell-cultured plate pre-coated with Matrigel to be further
expanded and characterized.
Immunocytochemistry for culture of NSCs. Cells grown on glass coverslips
pre-coated with Poly-L-ornithine and laminin were fixed with 4% PFA in PBS
(pH 7.4) for 15 minutes at room temperature, washed three times for 10 minutes
each with 0.3% Triton X-100 in PBS (PBST), blocked with 10% normal donkey
serum in PBS for 1 hour at room temperature and then incubated at 4 °C overnight
with primary antibodies diluted in PBS and 5% normal donkey serum. After
three rinses for 10 minutes each with PBST, the cells were incubated in secondary
antibodies diluted in 2% normal donkey serum for 1 hour at room temperature and
washed three times for 10 minutes each with PBST. The coverslips were mounted
on glass slides using Aqua-Poly/Mount medium. Images were captured by a laser
scanning confocal microscope (Zeiss, LSM880).
EdU labeling. Isolated cells from adult macaque hippocampi were plated on glass
coverslips pre-coated with 1.25% Matrigel or Poly-L-ornithine and laminin. After
incubation for 3 days with NSC medium, the medium was changed with NSC
medium supplemented with 10 μM of EdU (Thermo Fisher Scientific), and half
of the medium was changed every day for 42 days. Then, the cells were continued
to complete EdU staining according to the manufacturer’s instructions (Thermo
Fisher Scientific).
In vitro differentiation of macaque neural stem cells. To differentiate macaque
neural stem cells (mNSCs), mNSCs (passage 2 (P2)) were plated on glass coverslips
pre-coated with Poly-L-ornithine and laminin using NSC medium for 1 week. For
neuronal differentiation, mNSCs were cultured in neurobasal-A medium (Gibco)
supplemented with 5 μg ml−1 of insulin (Sigma-Aldrich), 20 ng ml−1 of BDNF
(PeproTech), 20 ng ml−1 of CNTF (PeproTech), 20 ng ml−1 of GDNF (PeproTech),
10 μM forskolin (Sigma-Aldrich), 50 µM cAMP (MedChemExpress), 25 mM
L-glutamic acid, 200 mM L-glutamine (Gibco), 1% B27 (Gibco), 1% N2 (Gibco)
and 1× penicillin–streptomycin (Gibco). For generation of astrocytes, mNSCs
were cultured in DMEM supplemented with 1% N2, 2 mM GlutaMAX (Gibco),
1% FBS and 1× penicillin–streptomycin. For generation of oligodendrocytes,
mNSCs were cultured in DMEM/F12 supplemented with 1% N2, 1% B27 without
vitamin A (Gibco), 2 mM GlutaMAX, 40 ng ml−1 of T3 (Sigma), 200 ng ml−1 of SHH
(PeproTech), 20 ng ml−1 of PDGF-AA (PeproTech), 10 ng ml−1 of NT3 (PeproTech)
and 1× penicillin–streptomycin. mNSCs was cultured in the three kinds of
differentiation medium for 21 days, and half of the medium was changed every
2 days.
Computational analysis of scRNA-seq data. Cell Ranger version 3.0.2 with the
Macaca fascicularis genome version 5.0 (ref. 66) was used to generate the output
count matrix. Cells with higher than 30% mitochondrial contents and fewer than
200 features, and features expressed in fewer than three cells, were excluded from
the analysis. We used Scrublet (version 0.2.1)67 and the gene expression patterns
to determine the doublets. The doublet score was calculated independently for
each sample with default settings and with the expected doublet score rate at 10%.
The clusters (see below for the clustering method) were determined as doublet
clusters if (1) the cluster showed obvious elevation of the Scrublet score and (2)
the marker gene expression of the cluster shared the expression pattern of two or
more clusters. The astrocyte–OPC doublets co-express AQP4 and CSPG4. The
microglia–OPC doublets co-express C1QA and CSPG4. The microglia–astrocyte
doublets co-express C1QA and AQP4. Subpopulations of cells with Scrublet scores
of >0.5 were also removed. Several clusters share the marker genes of microglia
and a particular cell type (for example, pyramidal neurons). There were putatively
phagocytosis microglia and technically also doublets. They were labeled as PhgMG
and kept.
We used SCANPY (version 1.4.6)68 for major data analysis and
visualization. The following steps were performed in order: data normalization,
log-transformation, highly variable gene selection, and PCA. Expression levels
were calculated as counts per 10,000 (NC). That is, the total mapped exonic reads
of a gene within one cell is scaled by the number of total mapped reads of that
cell and times 10,000. The log2 (NC + 1)-transformed values were used for further
analysis, unless indicated otherwise. The highly variable genes were selected based
on the log-transformed data. In total, 5,087 genes with a minimal mean expression
of 0.015 and a minimum dispersion of 0.05 were selected. The PCA was run with
the selected high variable genes.
NATURE NEUROSCIENCE | www.nature.com/natureneuroscience
ResouRce
NATuRE NEuROSCIENCE
We used SCENIC69 and pySCENIC70 (version 0.11.2) for the GRN analysis. Two
hundred cells per population were randomly selected. The co-expression modules
were inferred with a regression per-target approach (GRNBoost2) using a human
transcription factor list69. The indirect targets were pruned using cis-regulatory
motif discovery (cisTarget). The activity of these regulons was quantified based on
the enrichment scores for the regulon’s target genes (AUCell). The networks were
visualized by Cytoscape (version 3.9.0).
UMAP71 and Louvain72 clustering were performed on the Harmony-corrected
(version 0.0.5)52 PCA embedding. The neighborhood graphs were calculated using
scanpy.pp.neighbors, with 15 local neighborhoods and 50 Harmony-corrected
PCAs using mKNN graph. The connectivities were computed using the ‘umap’
method with Euclidean distance. Then, UMAP embedding was calculated using
scanpy.tl.umap with minimal effective distance of 0.5, spread of 1.0, initial learning
rate of 1.0, negative sample weighting of 1.0 and negative edge sample rate of
5. Louvain clustering was calculated using sc.tl.louvain on the neighborhood
calculated in the previous steps, with the resolution of 2, with the vtraag package
and with no weights from mKNN graph. Several clusters were further divided
using the resolution ranging from 1 to 5. SCCAF (version 0.0.10)73 was used
to determine the self-consistency of the clusters using the SCCAF_assessment
function. Two hundred cells per cluster were picked for five-fold cross-validation
training of the classifier, and the remaining cells were used to test for segregation
accuracy. Clustering with an accuracy on the hold-out set >84% was accepted.
The cell cycle scores were calculated using ‘scanpy.tl.score_genes_cell_cycle’
based on the published gene list74,75.
DEGs for each population were selected by SCANPY68 rank genes using
t-test_overestim_var test with Benjamini–Hochberg correction. The top 200–500
ranked genes for each cell population were selected for GO and KEGG enrichment
analysis using g:Profiler76 (version 1.0.0)76. Representative terms were selected
within the top 30 terms with the cumulative hypergeometric P values <0.05. The
heatmap plot of the gene express z-score was first averaged of all cells within the
cell population and then divided by the maximum z-score of each gene for a robust
display.
We performed the cross-correlation of mice gene expression (adult cells from
the dataset 1 of Hochgerner et al.) with that of macaques using neurogenic-related
genes. All genes within the selected GO terms (Supplementary Table 1) shared
between two species are used for the comparison. The z-scores for each gene
in each cell are calculated by normalizing, log-transforming and scaling of the
gene counts, following the standard SCANPY analysis pipeline68. To reduce the
sampling bias, we first calculated the mean z-score within each cell subclass and
then calculated the mean of the mean z-score. The Spearman’s rank correlation
coefficient77,78 (r) and P value were calculated using the scipy.stats.spearmanr
package.
DEGs between two selective cell populations or between two age (Y4–6 versus
Y13–15) or sex (female versus male) groups were selected by MAST incorporated
by Seurat79, which used the hurdle model80 to remove the technical effects. Genes
that were expressed in at least 0.1% of the cell populations were used for the test.
The significant DEGs were determined using the Wilcoxon rank-sum test with
FDR correction using all genes in the dataset. All significant DEGs for the cell
population of interest were selected for GO81,82 and KEGG enrichment analysis83–85
using g:Profiler76.
RNA velocity analysis. RNA velocyto43 (version 0.17.15) and scVelo42 (version
0.2.2) were used for RNA velocity analysis. RNA velocyto was run on the 10x Cell
Ranger result using the ‘run10x’ option. The resulted loom files were merged with
the AnnData in SCANPY and analyzed with scVelo.
RNA velocity was calculated using the standard process of scVelo. The
moments are computed for each cell across its nearest neighbors, where the
neighbor graph is obtained from Euclidean distances in PCA space using 30
principal components and 30 neighbors with the connectivities mode. Scvelo.
tl.recover_dynamics with default setting was used to recover the full splicing
kinetics of the genes. The velocities were estimated using scvelo.tl.velocity with
dynamical mode. The velocity graph was calculated with scvelo.tl.velocity_graph
with default settings.
Reference datasets. Hochgerner_1 dataset: a mouse hippocampus dataset30 from
both embryonic day (E) 16.5 and postnatal day (P) (P0–P132) mice downloaded
from the Gene Expression Omnibus (GEO)86,87 under accession code (GSE104323).
Hochgerner_2 dataset: a mouse hippocampus dataset30 from P12–P35 mice
(GSE95315).
Joglekar dataset: a mouse hippocampal dataset37 from developing mice (P7).
We downloaded the raw expression matrix from the GEO under accession code
GSE158450. Cell type annotation was downloaded from GitHub (https://raw.
githubusercontent.com/noush-joglekar/scisorseqr/master/inst/extdata/userInput/
bc_celltype_assignments).
Li dataset: a human brain dataset of 5–20-post-conceptional week (PCW)
postmortem human brains39. We downloaded the raw expression matrix with the
cell type annotation from http://psychencode.org.
Shin dataset: a mice hippocampal dataset from adult mice33. The Nestin-drived
CFP+ cells were enriched for the experiment under accession code GSE71485.
Zhong dataset: a human hippocampal dataset38 from gestational weeks 16–27
(GSE131258). Experimental batch information was inferred from the cell barcodes.
The dataset was annotated based on the marker gene expression pattern consistent
with accompanied literature.
Cross-species projection. scRNA-seq data from different species were merged
according to the uniquely mapped orthology genes from Ensmbl. The datasets
were first downsampled to 100–500 cells per population. The subpopulations
of a cell type (such as RGL_1 and RGL_2) are merged and treated as a single
population, unless otherwise noted. The data from the two species were integrated
using the scVI method51 followed by Harmony52 to account for the batch effect,
using categorical covariates of species and continuous covariates including n_genes
and n_counts. The model was trained for 600 times to ensure fully converged
training results. Harmony-corrected PCA was then calculated on the scVI latent
space. The UMAP was then calculated in the Harmony-corrected PCA space.
To quantitatively assess the correlation of the clusters across species, the
matrices of taxonomies were calculated based on previous publications50 on
the clustering based on the Euclidean distance calculated from the scVI and
Harmony-corrected PCA space. The alignment score was defined as the sum of
the minimum proportion of samples in each cluster that overlapped within each
Louvain cluster. Cluster alignment scores were visualized as heat maps.
The cell population annotations were based on the vote of the integration
results of all datasets30,37. The data projection from in vivo data to ex vivo data
was performed with SCCAF where logistic regression models were trained on the
Harmony latent space and projected to the other species.
Statistics and reproducibility. No statistical method was used to predetermine
sample size. The numbers of animals and cells were determined to ensure that
the biological replicates for each cell population and the number of cells in each
population met or exceeded the comparable published single-cell datasets30.
Some cells were excluded based on high mitogene percentage (>30%) and low
unique molecular identifier count (<200). Clusters were identified as doublets
and excluded if they had an elevated doublet score and combined marker
gene expression profiles of more than one cell type. The experiments were not
randomized. The investigators were not blinded to allocation during experiments
and outcome assessment. Rank-sum-based statistical approaches, such as
the Mann–Whitney U-test88 and the rank-sum-based Spearman’s correlation
coefficient test77, were used to avoid the assumption of data distribution or equal
variances.
Reporting Summary. Further information on research design is available in the
Nature Research Reporting Summary linked to this article.
Data availability
The datasets generated in this paper are publicly available at ArrayExpress under
accession codes E-MTAB-10225 and E-MTAB-10236. Source data are provided
with this paper.
Code availability
The code for generating figures can be found at https://github.com/haozhaozhe/
FM_hippo.
References
64. Saleem, K. S. & Logothetis, N. K. A Combined MRI and Histology Atlas of the
Rhesus Monkey Brain in Stereotaxic Coordinates (Academic Press, 2012).
65. Xiao, D. et al. Direct reprogramming of broblasts into neural stem cells by
single non-neural progenitor transcription factor Ptf1a. Nat. Commun. 9,
2865 (2018).
66. Yates, A. D. et al. Ensembl 2020. Nucleic Acids Res. 48, D682–D688 (2019).
67. Wolock, S. L., Lopez, R. & Klein, A. M. Scrublet: computational identication
of cell doublets in single-cell transcriptomic data. Cell Syst. 8, 281–291 (2019).
68. Wolf, F. A., Angerer, P. & eis, F. J. SCANPY: large-scale single-cell gene
expression data analysis. Genome Biol. 19, 15 (2018).
69. Aibar, S. et al. SCENIC: single-cell regulatory network inference and
clustering. Nat. Methods 14, 1083–1086 (2017).
70. Van de Sande, B. et al. A scalable SCENIC workow for single-cell gene
regulatory network analysis. Nat. Protoc. 15, 2247–2276 (2020).
71. Becht, E. et al. Dimensionality reduction for visualizing single-cell data using
UMAP. Nat. Biotechnol. 37, 38–44 (2018).
72. Blondel, V. D., Guillaume, J.-L., Lambiotte, R. & Lefebvre, E. Fast unfolding of
communities in large networks. J. Stat. Mech. 2008, P10008 (2008).
73. Miao, Z. et al. Putative cell type discovery from single-cell gene expression
data. Nat. Methods 17, 621–628 (2020).
74. Satija, R., Farrell, J. A., Gennert, D., Schier, A. F. & Regev, A. Spatial
reconstruction of single-cell gene expression data. Nat. Biotechnol. 33,
495–502 (2015).
75. Tirosh, I. et al. Dissecting the multicellular ecosystem of metastatic
melanoma by single-cell RNA-seq. Science 352, 189–196 (2016).
NATURE NEUROSCIENCE | www.nature.com/natureneuroscience
ResouRce NATuRE NEuROSCIENCE
76. Raudvere, U. et al. g:Proler: a web server for functional enrichment analysis
and conversions of gene lists (2019 update). Nucleic Acids Res. 47,
W191–W198 (2019).
77. Spearman, C. e proof and measurement of association between two things.
Am J Psychol 15, 72 (1904).
78. Glasser, G. J. & Winter, R. F. Critical values of the coecient of rank
correlation for testing the hypothesis of independence. Biometrika 48,
444–448 (1961).
79. Hao, Y. et al. Integrated analysis of multimodal single-cell data. Cell 184,
3573–3587 (2021).
80. Finak, G. et al. MAST: a exible statistical framework for assessing
transcriptional changes and characterizing heterogeneity in single-cell RNA
sequencing data. Genome Biol. 16, 278 (2015).
81. Ashburner, M. et al. Gene ontology: tool for the unication of biology. e
Gene Ontology Consortium. Nat. Genet. 25, 25–29 (2000).
82. Gene Ontology Consortium. e Gene Ontology resource: enriching a GOld
mine. Nucleic Acids Res. 49, D325–D334 (2021).
83. Kanehisa, M. & Goto, S. KEGG: Kyoto Encyclopedia of Genes and Genomes.
Nucleic Acids Res. 28, 27–30 (2000).
84. Kanehisa, M. Toward understanding the origin and evolution of cellular
organisms. Protein Sci. 28, 1947–1951 (2019).
85. Kanehisa, M., Furumichi, M., Sato, Y., Ishiguro-Watanabe, M. & Tanabe, M.
KEGG: integrating viruses and cellular organisms. Nucleic Acids Res. 49,
D545–D551 (2021).
86. Edgar, R., Domrachev, M. & Lash, A. E. Gene Expression Omnibus: NCBI
gene expression and hybridization array data repository. Nucleic Acids Res.
30, 207–210 (2002).
87. Barrett, T. et al. NCBI GEO: archive for functional genomics data sets—
update. Nucleic Acids Res. 41, D991–D995 (2013).
88. Mann, H. B. & Whitney, D. R. On a test of whether one of two random
variables is stochastically larger than the other. Ann. Math. Stat. 18, 50–60
(1947).
Acknowledgements
This research is supported by research grants from the National Key R&D Program
of China (2018YFA0108300), the Natural Science Foundation of China (81870682
and 81961128021), the Guangdong Provincial Key R&D Programs (2018B030335001
and 2018B030337001) and the Science and Technology Program of Guangzhou
(202007030011, 202007030010 and 202007030001) to S.L.; the Natural Science
Foundation of China (81721003) and the Local Innovative and Research Teams Project
of Guangdong Pearl River Talents Program (2017BT01S138), CAMS Innovation Fund
for Medical Sciences (2019-I2M-5-005) to Y.L.; the Natural Science Foundation of China
(81970794), the Science and Technology Program of Guangzhou (201904020036) and the
Technology Innovation 2030-Major Project on Brain Science and Brain-Like Computing
of the Ministry of Science and Technology of China (2021ZD0202603) to M.X.; the
Wellcome BioResource for a ‘Single Cell Gene Expression Atlasʼ (WT 108437/Z/15/Z)
and the Open Targets grant (OTAR2067) to Z.M.; and the China Postdoctoral Science
Foundation (2019M663256) to Z.-Z.H. The funders had no role in study design, data
collection and analysis, decision to publish or preparation of the manuscript.
Author contributions
S.L., Z.M., Y.L. and M.X. conceived and supervised the project. Z.-Z.H., Z.M., L.T., M.H.,
Y.S. and X.L. analyzed the data. J.-R.W., Z.-Z.H., R.L. and W.H. collected the tissues
and performed the scRNA-seq experiment. D.X. performed the neurosphere culture
experiment. J.-R.W., N.X., C.X. and Z.-Z.H. performed the immunofluorescence staining.
S.L., Z.M., Y.L., M.X. and Z.-Z.H. wrote the manuscript, with input from all authors.
Competing interests
The authors declare no competing interests.
Additional information
Extended data is available for this paper at https://doi.org/10.1038/s41593-022-01073-x.
Supplementary information The online version contains supplementary material
available at https://doi.org/10.1038/s41593-022-01073-x.
Correspondence and requests for materials should be addressed to
Mengqing Xiang, Yizhi Liu, Zhichao Miao or Sheng Liu.
Peer review information Nature Neuroscience thanks Gerd Kempermann, Orly Lazarov
and the other, anonymous, reviewer(s) for their contribution to the peer review of this
work.
Reprints and permissions information is available at www.nature.com/reprints.
NATURE NEUROSCIENCE | www.nature.com/natureneuroscience
ResouRce
NATuRE NEuROSCIENCE
Extended Data Fig. 1 | Quality control of the dataset. a) Distribution of the UMI counts detected in each cell in the whole dataset. b) Distribution of the
unique genes detected in each cell in the whole dataset. c) UMAP colored by the 10x version and the animal ID. d) Violin plots of the UMI counts for each
sample. The 10x version used for each sample is indicated by the colors. e) Violin plots of the unique genes detected for each sample. The 10x version used
for each sample is indicated by the colors. f) The distribution of sample sizes for each animal. g) Proportion of different cell types present in individual
animals. Cell types were colored using the same color scheme in Fig. 1d, and labeled at the bottom of panel.
NATURE NEUROSCIENCE | www.nature.com/natureneuroscience
ResouRce
NATuRE NEuROSCIENCE
Extended Data Fig. 2 | The enriched gene regulatory networks (GRNs, termed regulons) in the hippocampal cell population. a) tSNE plot derived from
the regulon score, showing the neurogenic-related populations, colored by the cell types or the Gene Regulatory Network enrichment score derived from
SCENIC AUcell. b) The dot plot shows the Gene Regulatory Network enrichment score derived from SCENIC AUcell. The color shows the normalized
mean enrichment score, while the size of the dots shows the fraction of cells in each group that express the genes. The x-axis shows the name of the
enriched regulons (Gene Regulatory Network), while the y-axis shows the cell populations in the dataset. c-f) TF regulatory networks predicted by SCENIC
for the RGLs, IPCs and NBs.
NATURE NEUROSCIENCE | www.nature.com/natureneuroscience
ResouRce
NATuRE NEuROSCIENCE
Extended Data Fig. 3 | The age-related shift in population distribution and gene expression. a) Sample distribution grouped by age. *, p < 0.05; using
two-tailed Mann-Whitney U test for each category between the young adults (age 4–6 year-old, n = 17 biologically independent samples) and the middle-
aged adults (age 13–15-year-old, n = 8 biologically independent samples). Error bars, standard error of mean. The p-value for the comparison between
young and middle-aged astrocytes is 0.039; for NB, 0.0039. All other comparisons are not significantly different. b) 1,435 DEGs between the Y4–6
and Y13–15 neuroblasts. These genes are visualized using the volcano plot. Genes that are significantly enriched (blue dots, determined by a two-part
generalized linear model implemented by MAST) and exhibit > 0.5 log2 fold-change are marked in red. Selective genes are marked with text. c) A strip-
chart showing the logarithmic-fold change (Log2 FC) of all genes between young adult (Y4–6) and middle-aged adult (Y13–15) for neurogenic-related cell
populations. The dashed line at the top and bottom of the plot indicate the 2-fold change threshold. The dashed line in the middle indicates no differential
expression. Genes that are upregulated in young adults have positive FCs. Genes that are differentially expressed (Wilcoxon rank-sum test with FDR
correction for multi-comparison) and >30% max FC are colored using the same color scheme in Fig. 1. d) Pair-wise comparison of the neurogenic-related
gene (n = 2,081 genes) enrichment between young and middle-aged adults for selective cell populations. Solid line, equal expression; dashed lines: 2-fold
enrichment. Gray dots, the mean expression of each neurogenic-related gene of all cells of each population. Blue dots, genes that show >0.5 log2 fold-
change enrichment in young adults. Red dots, genes that show >0.5 log2 fold-change enrichment in middle-aged adults. All correlation with a rank-sum
based Spearman’s correlation coefficient (r)>0.95 with p < 0.0001. e) Heatmaps showing the selective genes that are upregulated in young adults (upper
panel) or middle-aged adults (lower panel) exclusively for selective cell populations. Nomalized FC, normalized fold-change.
NATURE NEUROSCIENCE | www.nature.com/natureneuroscience
ResouRce NATuRE NEuROSCIENCE
Extended Data Fig.4 | The sex-related shift in population distribution and gene expression. a) Sample distribution grouped by sex. *, p < 0.05; using
two-tailed Mann-Whitney U test for each category between the females (n = 7 biologically independent samples) and the males (n = 18 biologically
independent samples). Error bars, standard error of mean. The p-value for the comparison between female and male IPC is 0.0015; for pre-OPC, 0.049;
for PVM, 0.008. All other comparisons are not significantly different. b) Strip-chart showing the logarithmic-fold change (Log2FC) for all genes between
females and males for neurogenic-related cell populations. The dashed line at the top and bottom of the plot indicate the 2-fold change threshold. The
dashed line in the middle indicates not differentially expressed. Genes that are upregulated in females have potived FCs. Genes that are significantly
expressed (Wilcoxon rank-sum test with FDR correction for multi-comparison) and >30% max FC are colored using the same color scheme in Fig. 1. c)
Pair-wise comparison of the neurogenic-related gene (n = 2,081 genes) enrichment between females and males for selected cell populations. Solid line,
equal expression; dashed lines: 2-fold enrichment. Gray dots, mean expression of each gene of all cells of each population. Blue dots, genes that show
>0.5-fold enrichment in females. Red dots, genes that show >0.5-fold enrichment in males. All correlation with a rank-sum based Spearman’s correlation
coefficient (r)>0.95 with p < 0.0001. d) Heatmaps showing the selective genes that are upregulated in female adults (upper panel) or male adults (lower
panel) exclusively for selective cell populations. Normalized FC, normalized fold-change.
NATURE NEUROSCIENCE | www.nature.com/natureneuroscience
ResouRce NATuRE NEuROSCIENCE
Extended Data Fig. 5 | Immunostaining of canonical neural markers validated the existence of neural progenitor cells in SGZ. a) NESTIN + cells in SGZ
with apical processes (white arrowheads) touch the capillaries (yellow arrows), Scale bar, 10 μm. n = 4 brains. b) NESTIN + MKI67 + cells in SGZ with
apical processes crossing the GCL into the ML, the processes touch capillaries of tubular morphology (yellow arrows). Scale bar, 10 μm. n = 4 brains. c)
HOPX protein (gray), progenitor marker SOX2 (green), and neural stem cell marker GFAP (red) are colocalized in RGLs of macaques. Scale bar, 10 μm.
n = 4 brains. Dashed lines indicate the outline of the cell. d) Immunolabeling for ASCL1, SOX2 and GFAP in the subgranular zone (SGZ) of the dentate
gyrus (DG) of macaque. White arrowheads indicate ASCL1 + SOX2 + GFAP- IPC; yellow arrowheads indicate ASCL1 + SOX2 + GFAP + RGL. Scale bars, 25
μm (low magnification), 10 μm (high magnification). n = 4 brains.
NATURE NEUROSCIENCE | www.nature.com/natureneuroscience
ResouRce NATuRE NEuROSCIENCE
Extended Data Fig. 6 | Macaque adult hippocampal transcriptomic cell types are aligned with well-established mouse datasets. a) UMAP visualization
of macaque hippocampal transcriptomic cell types aligned with the Hochgerner_2 dataset. UMAP visualization of macaque cells (n = 15,878) and mouse
cells (n = 3,132), stained with the species. b) UMAP visualization of macaque cell annotations using the same UMAP coordinates in panel a, stained
according to the color scheme as in Fig. 1d. Astro, Astro_1–4; Astro_im, Astro_im1, Astro_im2; GC, GC_1–3; Microglia, Microglia_1 and 2; OPC, OPC_1 and
2; Pyr, Pyr_1 and 2; RGL, RGL_1, RGL_2. c) UMAP visualization of mouse cell annotations using the same UMAP coordinates in panel a, stained according
to the color scheme as in Fig. 1d. d) Cell-type homologous heatmap between macaque (in rows) and mouse (in columns) for the Hochgerner_2 dataset.
Gray shades correspond to the minimum proportion of co-clustering between mouse and macaque cells. Rows show macaque populations and columns
show mouse populations. Colorbar indicates the alignment score. e) Macaque hippocampal transcriptomic cell types aligned with the Joglekar dataset,
visualization of macaque cells (blue, n = 15,878) and mouse cells (orange, n = 5,305) f) UMAP visualization of macaque cells, stained according to the
color scheme in Fig. 1d using the same coordinates as in panel d. g) UMAP visualization of mouse cells, stained according to the color scheme in Fig. 1d
using the same coordinates as in panel d. h) Cell-type homologous heatmap between macaque (in rows) and mouse (in columns) for the Joglekar dataset.
Gray shades correspond to the minimum proportion of co-clustering between mouse and macaque cells. Rows show macaque populations and columns
show mouse populations. Colorbar indicates the alignment score. i) Macaque hippocampal transcriptomic cell types aligned with the Shin dataset,
visualization of macaque cells (blue, n = 800) and mouse cells (orange, n = 132). j) UMAP visualization of macaque cells, stained according to the color
scheme in Fig. 1d using the same coordinates as in panel i. k) UMAP visualization of mouse cells, stained according to the color scheme in Fig. 1d using the
same coordinates as in panel i. Gray dots are macaque cells to facilitate the visualization of the coordinates.
NATURE NEUROSCIENCE | www.nature.com/natureneuroscience
ResouRce NATuRE NEuROSCIENCE
Extended Data Fig. 7 | Macaque adult hippocampal transcriptomic cell types are aligned with well-established human datasets. a) UMAP visualization
of macaque hippocampal transcriptomic cell types aligned with the Li dataset. UMAP visualization of macaque cells (n = 2,400) and human embryonic/
fetal cells (n = 323), stained with the species. b) UMAP visualization of macaque cell annotations using the same UMAP coordinates in panel a, stained
according to the color scheme as in Fig. 1d. Astro, Astro_1–4; GC, GC_1–3; MG, Microglia_1 and 2; OPC, OPC_1 and 2; Pyr, Pyr_1 and 2; RGL, RGL_1 and
2. c) UMAP visualization of human cell annotations using the same UMAP coordinates in panel a, stained according to the color scheme as in Fig. 1d. d)
Macaque hippocampal transcriptomic cell types aligned with the Zhong dataset, visualization of macaque cells (blue, n = 15,878) and human cells (orange,
n = 6,383) e) UMAP visualization of macaque cells, stained according to the color scheme in Fig. 1d using the same coordinates as in panel d. f) UMAP
visualization of human cells, stained according to the color scheme in Fig. 1d using the same coordinates as in panel d. g) Cell-type homologous heatmap
between macaque (in rows) and human (in columns) for the Zhong et al. dataset. The color of each cell represents the alignment score between human
and macaque cells. The larger value (darker in the heatmap) indicates a better alignment. Rows show macaque populations and columns show human
populations. Colorbar indicates the alignment score.
NATURE NEUROSCIENCE | www.nature.com/natureneuroscience
ResouRce
NATuRE NEuROSCIENCE
Extended Data Fig. 8 | The lineage of the neurogenic-related populations. a) The PCA visualization of the neurogenic-related population stained by cell
populations, following the same color scheme as in Fig. 1. b) The PCA visualization of the neurogenic populations stained by canonical neurogenic marker
genes. Colorbar indicates log-normalized gene-expression. c) the PCA visualization with the RNA velocities visualized as arrows. d) The heatmap shows
the expression of the pseudotime related genes as Viridis (light yellow as a high expression), while the colorbar on top shows the cells in the neurogenic
stages using the same color scheme in a. Selective genes are labeled to the left of the heatmap. Colorbar indicates gene expression level. e) The
neurogenic lineage inferred by Slingshot. f) Pseudotime analysis derived from Slingshot suggests trajectories from RGL, via IPC to NB. Colorbar indicates
the derived pseudotime.
NATURE NEUROSCIENCE | www.nature.com/natureneuroscience
ResouRce
NATuRE NEuROSCIENCE
Extended Data Fig. 9 | Comparison of the gene enrichment between macaques and mice. a) Neurogenic markers that are conserved between mice and
macaques. b) Species-specific marker genes that are only enriched in macaques or in mice. c) Comparison of the gene enrichment between macaques
and mice for the RGLs, IPCs, neuroblasts, and GCs, together with astrocytes and mature oligodendrocytes. Solid line, equal expression; dashed lines:
2-fold enrichment. Blue dots, high-diversity genes that are >2-fold enrichment in mice. Top five genes labeled by text. Gray dots, mean expression of each
neurogenic-related gene (n = 2,081 genes) of all cells of each population. Red dots, high-diversity genes that are >2-fold enrichment in macaques with the
top five genes labeled by text. The “r” value, the Spearman’s correlation coefficient for each comparison. All comparisons are significant with p < 0.001. d)
Neurogenic-related cell composition for macaques (n = 39,555) and mice (n = 3,711) of the Hochgerner datasets.
NATURE NEUROSCIENCE | www.nature.com/natureneuroscience
ResouRce NATuRE NEuROSCIENCE
Extended Data Fig. 10 | Process of neurosphere formation and gene expression of cultured hippocampal neurospheres. a-d) In vitro recording process
of a cultured neurosphere clone from the adult macaque hippocampi. n = 4 clones. Scale bars, 40 μm. D, day. e) UMAP visualization of the cultured
sphere cells (n = 6,988 cells), stained by the expression of selective marker genes. VIM, RGL and astrocyte marker; NNAT, novel neuroblast marker; SOX4,
neuroblast marker; TUJ1, neuronal marker. Colorbar indicates log-normalized gene expression level. f-h) The cultured neurosphere cells from macaque
hippocampus were immunoreactive for RGL and astrocyte marker VIMENTIN in f, neuroblast marker NNAT and SOX4 in g and h. n = 4 clones. Scale bars,
40 μm. i) Immunostaining analysis of EdU long-term labeling of the cultured neurospheres from the adult macaque hippocampus showed a small number
of cells in the neurosphere were labeled by fluorescence for both TUJ1 and EdU in i (indicated by arrows). Scale bars, 40 μm. n = 4 clones for each staining.
j) UMAP visualization of transcriptomic cell types projected from the Hochgerner_1 dataset. Red dots represent the distribution of the predicted cell type.
IPC, intermediate progenitor cell; NB, neuroblast; GC, granule cell; Astro, astrocyte.
NATURE NEUROSCIENCE | www.nature.com/natureneuroscience
1
nature portfolio | reporting summary
March 2021
Corresponding author(s):
Mengqing Xiang, Yizhi Liu, Zhichao Miao,
Sheng Liu
Last updated by author(s):
Mar 29, 2022
Reporting Summary
Nature Portfolio wishes to improve the reproducibility of the work that we publish. This form provides structure for consistency and transparency
in reporting. For further information on Nature Portfolio policies, see our Editorial Policies and the Editorial Policy Checklist.
Statistics
For all statistical analyses, confirm that the following items are present in the figure legend, table legend, main text, or Methods section.
n/a
Confirmed
The exact sample size (n) for each experimental group/condition, given as a discrete number and unit of measurement
A statement on whether measurements were taken from distinct samples or whether the same sample was measured repeatedly
The statistical test(s) used AND whether they are one- or two-sided
Only common tests should be described solely by name; describe more complex techniques in the Methods section.
A description of all covariates tested
A description of any assumptions or corrections, such as tests of normality and adjustment for multiple comparisons
A full description of the statistical parameters including central tendency (e.g. means) or other basic estimates (e.g. regression coefficient)
AND variation (e.g. standard deviation) or associated estimates of uncertainty (e.g. confidence intervals)
For null hypothesis testing, the test statistic (e.g. F, t, r) with confidence intervals, effect sizes, degrees of freedom and P value noted
Give P values as exact values whenever suitable.
For Bayesian analysis, information on the choice of priors and Markov chain Monte Carlo settings
For hierarchical and complex designs, identification of the appropriate level for tests and full reporting of outcomes
Estimates of effect sizes (e.g. Cohen's d, Pearson's r), indicating how they were calculated
Our web collection on statistics for biologists contains articles on many of the points above.
Software and code
Policy information about availability of computer code
Data collection
Illumina and 10x Genomics with Chromium Controller Readiness Test were used for data collection.
Data analysis
We used Cellranger v3 (10x Genomics), Scrublet 0.2.1, Scanpy 1.4.6, pySCENIC 0.11.2, GRNBoost2, Cytoscape 3.9.0, Harmony 0.0.5, gProfiler
1.0.0, RNA velocyto 0.17.15, and scVelo 0.2.2, SCCAF 0.0.10 for primary analysis of single-cell sequencing data. Customized source code is
available at https://github.com/haozhaozhe/FM_hippo
For manuscripts utilizing custom algorithms or software that are central to the research but not yet described in published literature, software must be made available to editors and
reviewers. We strongly encourage code deposition in a community repository (e.g. GitHub). See the Nature Portfolio guidelines for submitting code & software for further information.
Data
Policy information about availability of data
All manuscripts must include a data availability statement. This statement should provide the following information, where applicable:
- Accession codes, unique identifiers, or web links for publicly available datasets
- A description of any restrictions on data availability
- For clinical datasets or third party data, please ensure that the statement adheres to our policy
The datasets generated in this paper are publicly available at the arrayExpress under the accession codes E-MTAB-10225 (scRNA-seq for hippocampus) and E-
MTAB-10236 (scRNA-seq for sphere).
2
nature portfolio | reporting summary
March 2021
Field-specific reporting
Please select the one below that is the best fit for your research. If you are not sure, read the appropriate sections before making your selection.
Life sciences
Behavioural & social sciences
Ecological, evolutionary & environmental sciences
For a reference copy of the document with all sections, see nature.com/documents/nr-reporting-summary-flat.pdf
Life sciences study design
All studies must disclose on these points even when the disclosure is negative.
Sample size
No statistical method was used to predetermine sample sizes. The number of animals and cells were determined to ensure the biological
replicates for each cell population and the number of cells in each population met or exceed the comparable published single-cell datasets
(e.g. Hochgerner et al. 2018, Nat. Neurosci. 2018).
Data exclusions
During data analysis, some cells within each sample were excluded for the following reasons: 1) high mitogene percentage (>30%) and 2) low
UMI count (< 200), and 3) clusters were identified as doublets and excluded if they had elevated doublet score and the combined marker
gene expression profiles of more than one cell type. The criteria was determined based on the distribution of the datasets and applied equally
to all samples.
Replication
Full replication of the observations was not feasible due to the size and cost of the experiment. The observations were validated by previous
work or other methodology including neurosphere culture and immunofluoscent staining. The immunostaining results were replicated with at
least four animals and indicated in the figure legend. The experiments were performed based on the availability of the animals without a set
frequency.
Randomization
Randomization was not applicable for this study. There is no treatment or intervention to the samples. Therefore there is no need for
randomization.
Blinding
Blinding was not applicable for this study. To avoid bias, all samples were treated equally with the same rigorous criteria.
Reporting for specific materials, systems and methods
We require information from authors about some types of materials, experimental systems and methods used in many studies. Here, indicate whether each material,
system or method listed is relevant to your study. If you are not sure if a list item applies to your research, read the appropriate section before selecting a response.
Materials & experimental systems
n/a
Involved in the study
Antibodies
Eukaryotic cell lines
Palaeontology and archaeology
Animals and other organisms
Human research participants
Clinical data
Dual use research of concern
Methods
n/a
Involved in the study
ChIP-seq
Flow cytometry
MRI-based neuroimaging
Antibodies
Antibodies used
Mouse anti-SOX-2, clone 10H9.1 -2523218, Millipore, Cat# MAB4423, 1:1000 dilution
Chicken anti-Glial Fibrillary Acidic Protein, Millipore, Cat# AB5541, 1:2000 dilution
Rabbit GLAST Polyclonal, Proteintech, Cat# 20785-1-AP-50ul, 1:1500 dilution
Rabbit anti-Ki67, clone SP6, Abcam, Cat# ab16667, 1: 200 dilution
Mouse monoclonal anti-HMGB2, clone 3E5, Sigma, Cat# WH0003148M5, 1: 2000 dilution
Rabbit Sox2 (D6D9) XP® Rabbit mAb, clone D6D9, Cell Signaling Technology, Cat# 3579, 1: 200 dilution
Goat doublecortin (C-18), Santa Cruz, Cat# sc-8066, 1: 1000 dilution
Mouse anti-Prox1, clone 4G10, Millipore, Cat# MAB5654, 1: 500 dilution
Rabbit recombinant anti-Vimentin, clone EPR3776, Abcam, Cat# ab92547, 1: 1000 dilution
Rabbit polyclonal anti-HOPX, Proteintech, Cat# 11419-1-AP, 1/100 dilution
Mouse monoclonal anti-Nestin, R&D, clone 196908, Cat# MAB1259-SP, 1:100 dilution
Rabbit monoclonal anti-MASH1 (ASCL1), Abcam: clone EPR19840, Cat# ab211327, 1: 100 dilution
Validation
Well characterized commercial antibodies were used.
3
nature portfolio | reporting summary
March 2021
SOX-2 Millipore: Validated by the vendor with western blotting in mouse embryonic stem cell lysate.
https://www.merckmillipore.com/INTERSHOP/web/WFS/Merck-HU-Site/hu_HU/-/USD/ShowDocument-File?ProductSKU=MM_NF-
MAB4423&DocumentId=null&DocumentType=COA&Language=EN&Country=US&ProductBatchNo=2006962&Origin=PDP
Glial Fibrillary Acidic Protein Millipore: This Anti-Glial Fibrillary Acidic Protein Antibody is validated by the vendor for use in IC, IH,
IH(P), WB for the detection of GFAP.
https://www.merckmillipore.com/HK/en/product/Anti-Glial-Fibrillary-Acidic-Protein-Antibody,MM_NF-AB5541
GLAST Proteintech: Validated by the vendor. Positive IHC Detected In Human Brain Tissue And Mouse Brain Tissue. Positive WB
Detected In Neuro-2a Cells, C6 Cells, Mouse Brain Tissue.
https://www.ptglab.com/products/GLAST-Antibody-20785-1-AP.htm
Ki67 Abcam: Validated by the vendor with the detection of protein depletion in gene knockout cells.
https://www.abcam.com/Ki67-antibody-SP6-ab16667.html
HMGB2 Sigma: Validated by the vendor with IHC staining, western blotting, and ELISA.
https://www.sigmaaldrich.com/HK/zh/product/sigma/wh0003148m5
Sox2 (D6D9) Cell Signaling: Validated in the references provided by the vendor.
DOI: 10.1016/j.omtn.2021.05.002, DOI: 10.1111/jcmm.16701. https://www.cellsignal.com/products/primary-antibodies/sox2-d6d9-
xp-rabbit-mab/3579
Doublecortin Santa Cruz: Validated in the references provided by the vendor.
DOI: 10.1002/glia.22906; DOI: 10.1007/s12035-016-0151-5; DOI: 10.1523/JNEUROSCI.0343-15.2015. https://www.scbt.com/p/
doublecortin-antibody-c-18
Prox1 Millipore: Validated by the vendor with western blot, in which the antibody recognizes a ~83 kDa protein in fetal mouse brain
lysate.
https://www.merckmillipore.com/HK/en/product/Anti-Prox1-Antibody-clone-4G10,MM_NF-MAB5654
Vimentin, Abcam: Validated by the vendor. WB: HeLa, HEK293, Jurkat, A549, NIH3T3, PC12, HUVEC, Daudi, Caco-2 and COS-1 cell
lysates; mouse and rat brain tissue lysates. IHC-P: Human kidney, colon, breast adenocarcin
oma, cervical carcinoma and ovarian cancer tissues, mouse brain and kidney, E17 rat cheek and rat skin tissue sections; Rhesus
monkey retina tissue. IHC-Fr: Mouse testis tissue. ICC/IF: HeLa, human adenocarcinoma, human schlemms canal endothelium and
wild-type HAP1 cells. Flow Cyt (intra): HeLa cells.
https://www.abcam.com/vimentin-antibody-epr3776-cytoskeleton-marker-ab92547.html
HOPX, Proteintech: Validated by the vendor using western blotting.: The antibody is also validated by immunohistochemistry staining
in human brains from fetuses/neonates with Down Syndrome. DOI: 10.1186/s40478-020-01015-3 https://www.ptglab.com/
products/HOPX-Antibody-11419-1-AP.htm
Nestin, R&D: Validated by the vendor using immunofluorescent staining in immersion fixed human fetal neural progenitor cells.
https://www.rndsystems.com/cn/products/human-nestin-antibody-196908_mab1259
MASH1 (ASCL1) Abcam: Validated by immunohistochemistry staining in mouse lungs as a pulmonary neuroendocrine cell. DOI:
10.1002/stem.2744 https://www.abcam.com/mash1achaete-scute-homolog-1-antibody-epr19840-ab211327.html
Animals and other organisms
Policy information about studies involving animals; ARRIVE guidelines recommended for reporting animal research
Laboratory animals
We used adult macaque monkeys (Macaca fascicularis) of both sex (6 males and 2 females) at the age of 4-15 YO.
Wild animals
The study did not involve wild animals.
Field-collected samples
The study did not involve samples collected from the field.
Ethics oversight
All experimental procedures were approved by and in accordance with the Animal Care and Use Committee of Zhongshan
Ophthalmic Center, Sun Yat-sen University. The study was performed in accordance with the Public Health Service Guide to the Care
and Use of Laboratory Animals.
Note that full information on the approval of the study protocol must also be provided in the manuscript.
A preview of this full-text is provided by Springer Nature.
Content available from Nature Neuroscience
This content is subject to copyright. Terms and conditions apply.
