RES EARCH

◥ neuron, ependymoglia, and neuroblast pop-

RESEARCH ARTICLE SUMMARY ulations and determined their conservation
with amniotes by using comparative analyses.
NEUROEVOLUTION We found that the axolotl telencephalon con-
tains glutamatergic neurons with transcrip-
Single-cell analyses of axolotl telencephalon tional similarities to neurons of the turtle
and mouse hippocampus, dorsal cortex, and
organization, neurogenesis, and regeneration olfactory cortex. Olfactory cortex–like neu-
rons also show conserved neuronal input
Katharina Lust*†, Ashley Maynard†, Tomás Gomes†, Jonas Simon Fleck, J. Gray Camp, projections from the olfactory bulb. Axolotl
Elly M. Tanaka*, Barbara Treutlein* g-aminobutyric acid–releasing (GABAergic) in-
hibitory neurons show signatures of different
INTRODUCTION: Salamanders, such as the axolotl ular trajectories underlying developmental and subregions of the ganglionic eminence and
(Ambystoma mexicanum), play a role in the adult neurogenesis. We molecularly character- resemble turtle and mouse GABAergic inhib-
study of tetrapod-conserved traits. Cell-type ized axolotl telencephalon cell types, neuro- itory neurons. We used trajectory inference
diversity in salamander brains and their relation genesis, and evolutionary conservation. We to construct differentiation trajectories of ho-
to other vertebrate brains has until now been applied single-nucleus genomic profiling to meostatic neurogenesis and found that epen-
studied mainly histologically. Axolotl brains the axolotl telencephalon in steady state and dymoglia largely progress through distinct
grow during postembryonic life, with new neu- during regeneration to investigate its cell-type intermediate neuroblast types and use spe-
rons generated by proliferating ependymoglia diversity and molecular dynamics of homeo- cific gene regulatory networks to form distinct
cells. Axolotl brains also regenerate after in- static neurogenesis. We compared molecular glutamatergic neuron types. We tracked cycling
jury; however, it is still unclear how stem cells profiles to understand injury-specific features cells upon injury of the telencephalon and

Downloaded from https://www.science.org on September 24, 2024

regenerate the brain and whether neuronal of regenerative neurogenesis. found an injury-specific ependymoglia tran-
connections are appropriately recovered. scriptional state characterized by up-regulation
RESULTS: We determined the cellular diver- of wound healing and cell migration genes at
RATIONALE: Single-cell and single-nucleus ge- sity of the axolotl telencephalon using single- the beginning of regenerative neurogenesis.
nomic profiling of the telencephalon has re- nucleus RNA sequencing (snRNA-seq) and Neurogenesis after injury progresses similarly
vealed diversity and evolutionary relationships single-nucleus assay for transposase-accessible to homeostatic neurogenesis and results in re-
of cell types and brain regions among several chromatin with high-throughput sequencing establishment of lost neurons and input pro-
amniotes, including reptiles, birds, and mam- (snATAC-seq), as well as spatial transcrip- jections from the olfactory bulb.
mals. These methods have also revealed molec- tomics. We identified regionally distributed
CONCLUSION: Our findings indicate that cell
types and gene expression patterns associated
with mammalian telencephalon regions are also
evident in the amphibian brain. The evolution-
ary history of cell types clarified the larger diver-
gence in glutamatergic compared with GABAergic
neurons that we observed the axolotl, as was also
seen in reptiles. We conclude that in the postem-
bryonic axolotl, telencephalon neurogenesis pro-
gresses through diverse neuroblast progenitors,
which are associated with specific neuron types
and dependent on shared as well as specific
regulatory programs. We found implementa-
tion of these same programs in regenerative
neurogenesis, which indicates that brain injury
activates neurogenesis through existing path-
ways after inducing an injury-specific ependy-
moglia state. Regenerated neurons reestablish
their previous connections to distant brain re-
gions, suggesting potential functional recovery.
Our insight into how the axolotl brain regener-
ates may inform studies of brain regeneration
in other organisms.
▪
The list of author affiliations is available in the full article online.
PHOTO AND ILLUSTRATION: K. LUST

*Corresponding author. Email: [email protected]

(K.L.); [email protected] (E.M.T.); barbara.treutlein@
bsse.ethz.ch (B.T.)
†These authors contributed equally to this work.
Axolotl telencephalon organization, conservation, and neurogenesis. Single-nucleus multiome sequencing
Cite this article as K. Lust et al., Science 377, eabp9262
provides a comprehensive overview of neuronal and non-neuronal cell types in the axolotl telencephalon (2022). DOI: 10.1126/science.abp9262
and their specific genes and regulatory networks. Cross-species comparison of single-cell data assigned cell
populations to reptilian and mammalian brain regions and progenitor types. A snRNA-seq time course READ THE FULL ARTICLE AT
after brain injury revealed differences and parallels between regenerative and homeostatic neurogenesis. https://doi.org/10.1126/science.abp9262

Lust et al., Science 377, 1061 (2022) 2 September 2022 1 of 1

RES EARCH

◥ throughput sequencing (snATAC-seq) from

RESEARCH ARTICLE the same single nucleus] (Fig. 1A and fig. S1A).
We computationally integrated 48,136 nuclei
NEUROEVOLUTION and identified 95 molecularly distinct clusters
of neuronal and non-neuronal cells (Fig. 1, B
Single-cell analyses of axolotl telencephalon to E, and fig. S1, B to E). We annotated non-
neuronal clusters as endothelial cells [Collagen
organization, neurogenesis, and regeneration alpha-1(IV) chain+ (Col4a1+)] and glial cells,
including oligodendrocytes [Muscle-associated
Katharina Lust1*†, Ashley Maynard2†, Tomás Gomes2†, Jonas Simon Fleck2, J. Gray Camp3,4, receptor tyrosine kinase+ (Musk+)], microglia
Elly M. Tanaka1*, Barbara Treutlein2* [Colony-stimulating factor-1+ (Csf1r+)], and
ependymoglia [GLI family zinc finger 2+ (Gli2+)].
Salamanders are tetrapod models to study brain organization and regeneration; however, the Neuronal clusters included glutamatergic excit-
identity and evolutionary conservation of brain cell types are largely unknown. We delineated the cell atory neurons [Solute carrier family 17 member
populations in the axolotl telencephalon during homeostasis and regeneration using single-cell genomic 6/7+ (Slc17a6/7+)], g-aminobutyric acid–releasing
profiling. We identified glutamatergic neurons with similarities to amniote neurons of hippocampus, (GABAergic) inhibitory interneurons [Glutamate
dorsal and lateral cortex, and conserved g-aminobutyric acid–releasing (GABAergic) neuron classes. We decarboxylase1+ and 2+ (Gad1+ and Gad2+)],
inferred transcriptional dynamics and gene regulatory relationships of postembryonic, region-specific and a Mex-3 RNA binding family member A+
neurogenesis and unraveled conserved differentiation signatures. After brain injury, ependymoglia (Mex3a+)/Gli2– population of cells we termed
activate an injury-specific state before reestablishing lost neuron populations and axonal connections. neuroblasts (fig. S1E). Each cell type was pres-
Together, our analyses yield insights into the organization, evolution, and regeneration of a tetrapod ent in each microdissected region (Fig. 1, B
nervous system. and C, and fig. S1A), and we provide a set of

Downloaded from https://www.science.org on September 24, 2024

marker genes for each cluster identified at this

C
resolution (Fig. 1, D and E, and table S1). We
omparing brains between animals has neurogenesis throughout life; however, after performed immunofluorescence stainings and
been a means to analyze the evolution- brain injury neurogenesis is almost absent in situ RNA hybridization chain reaction (HCR)
ary origin and diversification of this (10, 11). The molecular relationship between to localize major cell types in the tissue accord-
structure. Comprehensive single-cell RNA neurogenesis seen in salamanders and mam- ing to marker expression (Fig. 1F). Glial fibrillary
sequencing (scRNA-seq) and single- mals has not been explored. Furthermore, acidic protein+ (GFAP+) ependymoglia line
nucleus RNA sequencing (snRNA-seq) have similarities and differences between homeo- the ventricle of every region, whereas Mex3a+
increased the resolution of cell identity and static and regenerative neurogenesis in the neuroblasts are sparsely distributed along the
development of the vertebrate brain. Cell types salamander brain are unclear. ventricle. GABAergic neurons were sparsely
from the dorsal region of the telencephalon— To understand the organizational features distributed along the medial, dorsal, and lateral
which in mammals includes the hippocampus, of the axolotl telencephalon, we used single- region and clustered together densely in the
amygdala, claustrum, olfactory (piriform) cortex, nuclei genomic methods and spatial profiling. striatum, whereas glutamatergic neurons were
and elaborate neocortex—have been resolved We analyzed cell types present in different re- located in the medial, dorsal, lateral, and ven-
and compared between amniotes, including gions and defined their similarities to amniote tral pallium. Small populations of myelin basic
turtle, lizards, birds, mouse, and human (1–4). telencephalic cell types. To delineate the cellu- protein+ (MBP+) oligodendrocytes and ion-
These studies have revealed evolutionary rela- lar and molecular dynamics of homeostatic ized calcium binding adaptor molecule 1+
tionships in cell types and brain regions among neurogenesis in the axolotl and its relation to (IBA1+) microglia cells were dispersed through-
amniotes; however, little is known about their adult neurogenesis in mice, we used clonal out all regions.
conservation in other vertebrates. tracing, trajectory analysis, and multiomic We next analyzed the abundance of each
The axolotl (Ambystoma mexicanum), as an sequencing. Analyzing regenerative neuro- cell cluster in the sampled pallial regions (Fig.
amphibian, represents one of the closest living genesis, we determined the similarities and 1G and fig. S1F). Oligodendrocytes, microglia,
relatives to amniotes and is therefore suited differences to homeostatic neurogenesis and and endothelial cell clusters originated from
for comparative studies of brain cell types, neu- found that regenerated neurons reestablish each region in similar proportions. By contrast,
ronal connectivity, and function. The axolotl is neuronal input from other regions of the ependymoglia, neuroblast, glutamatergic, and
also able to regenerate the telencephalon after telencephalon. Together, our comprehensive GABAergic neuron clusters showed variable
removal of the dorsal region by activating analyses of the axolotl telencephalon yielded region contributions, with some cell clusters
neurogenesis (5), which is also present during insights into the organization, evolution, and being completely region-restricted, whereas
post-embryonic life (6). Neurogenesis can be regeneration of a tetrapod nervous system. others could be found throughout the sampled
found in all metazoans with a nervous system pallium regions. These data provide an overview
(7–9), and in the mouse, cells of the adult sub- snRNA-seq of the axolotl telencephalon of cell populations in the axolotl telencephalon
ventricular zone (SVZ) undergo continuous We used snRNA-seq to generate a comprehen- and suggest substantial regional specificity in
sive dataset of the cell types and states from the neurogenic programs.
1
Research Institute of Molecular Pathology, Vienna Biocenter
axolotl telencephalon. We microdissected the
telencephalon into medial (containing medial Regional conservation of axolotl
(VBC), Campus Vienna Biocenter, 1030 Vienna, Austria.
2
Department of Biosystems Science and Engineering, ETH pallium and septum), dorsal (containing dorsal glutamatergic neurons
Zürich, 4058 Basel, Switzerland. 3Roche Institute for
pallium), and lateral (containing lateral pallium, Glutamatergic neurons of the amniote telen-
Translational Bioengineering (ITB), Roche Pharma Research and
Early Development, Roche Innovation Center Basel, Basel, ventral pallium, and dorsal striatum) regions. cephalon show a high degree of transcrip-
Switzerland. 4University of Basel, 4001 Basel, Switzerland. Additionally, we profiled all of these regions as a tomic diversification (1, 2). Our data revealed
*Corresponding author. Email: [email protected] (K.L.); whole with single-nucleus multiome sequenc- 29 glutamatergic neuron clusters differentially
[email protected] (E.M.T.); [email protected]
(B.T.) ing [snRNA-seq and single-nucleus assay for distributed in medial, dorsal, and lateral re-
†These authors contributed equally to this work. transposase-accessible chromatin with high- gions (Fig. 2, A to C, fig. S2A, and table S1).

Lust et al., Science 377, eabp9262 (2022) 2 September 2022 1 of 12

RES EARCH | R E S E A R C H A R T I C L E

A snRNA-seq analysis of E Marker gene sets for 95 molecularly distinct cell types G EPEN12
EPEN10
the axolotl telencephalon Dorsal EPEN2
EPEN6

Ependymoglia
EPEN14
EPEN1
EPEN7
EPEN0
EPEN3

Cell type marker genes

EPEN8
Medial Lateral EPEN4
EPEN13
EPEN9
EPEN11
EPEN5
NB1
NB13
B NB3
NB5
48,136 nuclei NB11

Neuroblast
95 clusters NB2
NB7
NB9
NB4

1
Expression
NB8
NB12
NB0
NB14
NB10
UMAP 2

NB6
0
GLUT14
GLUT7
UMAP 1 GLUT4
All Clusters GLUT18
GLUT19
GABA Endo F GLUT0
GLUT15
Glut MGs GFAP Mex3a GLUT6
GLUT17

Downloaded from https://www.science.org on September 24, 2024

NB Oligo GLUT5
Epen GLUT13
GLUT11

Glutamatergic
GLUT20
GLUT16
C Region
GLUT23
GLUT12
GLUT24
GLUT3
GLUT8
GLUT1
GLUT21
GLUT27
GLUT25
GLUT26
GLUT2
GLUT28
M D L M D L GLUT9
GLUT22
Gad2 Slc17a7 GLUT10
GABA24
GABA8
Medial GABA5
GABA0
Dorsal GABA21
Lateral GABA20
GABA16
Undetermined GABA6
GABA2
GABA23
D Gli2 G2M Score Mex3a
GABA27
GABA7
GABA18
GABAergic

GABA17
GABA12
GABA4
GABA29
GABA14
GABA10
M D L M D L GABA28
GABA22
MBP IBA1 GABA3
GABA9
GABA11
Snap25 Gad2 Slc17a6/7 GABA19
GABA15
GABA13
GABA26
GABA1
GABA25
OLIG10
OLIG15
Others

MG8
ENDO11
ENDO12
ENDO14
Expression
1 Max 100 % per region 0

Fig. 1. Cellular diversity of the axolotl telencephalon. (A) Schematic ependymoglia), Mex3a (neuroblasts), Snap25 (neurons), Gad2 (GABAergic
highlighting the regions of the axolotl telencephalon used for snRNA-seq. neurons), and Slc17a6/7 (glutamatergic neurons). (E) Heatmap illustrates the
(B) UMAP of all cell types across all regions colored by cell-type class. Subtypes expression of marker genes (table S1) for the 95 distinct cell types. (F) Antibody
are shown in different shades. GABA, GABAergic neuron; Glut, glutamatergic stainings and HCR for main cell types: GFAP (ependymoglia), Mex3a (neuroblasts),
neuron; NB, neuroblast; Epen, ependymoglia cell; Endo, endothelial cell; MGs, Gad2 (GABAergic neurons), Slc17a7 (glutamatergic neurons), MBP (oligodendro-
microglia; and Oligo, oligodendrocyte. (C) UMAP plot of regional distribution of cytes), and IBA1 (microglia cells). Scale bars, 100 mm. (G) Stacked barplots
all cell types. Shades indicate true region identity versus predicted regional illustrating the regional distribution of the populations of cells. GABA, GABAergic
identity. (D) UMAP plot of the expression of markers for ependymoglia, neuron; Glut, Glutamatergic neuron; NB, neuroblast; Epen, ependymoglia cell; Endo,
neuroblast, and neuronal cell types: Gli2 (ependymoglia), G2M score (active endothelial cell; MGs, microglia; Oligo, oligodendrocyte.

Lust et al., Science 377, eabp9262 (2022) 2 September 2022 2 of 12

RES EARCH | R E S E A R C H A R T I C L E

Fig. 2. Diversity of axolotl A 7

C 0
1
telencephalic glutamatergic 5
2
4 3
neurons. (A) UMAP plot 14 4
17
of 29 glutamatergic neuron 0 15 5

Axolotl Glutamatergic Clusters

12 6
16 18
types. (B) UMAP plots of the 23 24 28
7
8
regional distribution of 3 9
8 6 19 10
glutamatergic neurons types 13 11
20
12
and the expression of three 21 25
27 13
marker genes: Elmo1, Satb1, 11
UMAP 2
1 9 14
2 15
and Rbfox3. (C) Heatmap 26 16
UMAP 1 17
illustrating the expression of 22 Glutamatergic neurons 18
19
top markers for each gluta- 10 n= 25,432 20
matergic neuron cluster. 21

(D) Correlation analysis B 22

24
25
between expression profiles of Region 26

Max
27
axolotl glutamatergic neuron 28

Expression

Ndst4
Reln
Rbfox3
Nrxn3
Sema5a
Elmo1

Znf385b
Cdh11
Pik3cb
C1ql3

Sorcs1
Vwa5b1
Dnah3

Fstl4
Inpp5a
Spock1
Satb1

Arpp21

Crhr1
Rasa2
Nptx2

Pdzd2

Nfatc3
C1ql1
Tnfaip8l3

Rtn4rl1
Kcng1

Nxph1
Asic4
Scn5a
Grid1
Gria4
Htr5a
Tacr1
Adcy8

Cdh12

Cacng3
Elfn1
Cox1
Sim1
Shisa6

Tmem132c

Prkg2

Khdrbs2
Adcy2

Gpr22
Mchr2
Trpc6
Sox6

Syt6
Cntn5

Lhx6
Trpc7
Ptprt

Pex5l

Arhgap27
Nmbr

Cntfr
Bdnf
Medial
types and turtle glutamatergic Dorsal
Lateral
neuron types [data are from Undetermined

0
(2)]. Vertical lines indicate
highest correlated cluster to
D CA DG pDC aDC DVR PT LC
Medial
Dorsal
Lateral
E
Undeter. Cluster 11
Dorsal
turtle, and horizontal lines 22

|
* 25

|
indicate highest correlated

Downloaded from https://www.science.org on September 24, 2024

Ventral
Axolotl Glutamatergic Clusters

Elmo1 9

|
cluster to axolotl. (Right) 20

Cluster Score
10

0.10
|
*
Stacked barplots illustrating 24
|

the regional distribution of the 19

|

0.05
| | |
11 Cluster 1

|
cells from each cluster. 1 | | |
|
|
|
Asterisks indicate clusters 2

|
21

|
with agreeing assignments | |
8

|
* | | | | | |

Cluster Score
between all genes correlation 6
|

0.20
* |
Satb1 13

|
and integration analysis. CA, 28
| |

0.05
| | | |
cornu ammonis; DG, dentate 27 | Cluster 6
26

|
* | |
gyrus; pDC and aDC, posterior 16
|

* |
3
|

and anterior dorsal cortex; 0

|
| |

* | | |
DVR, dorsal ventricular ridge; Region * 5

Cluster Score

0.16
| |
12
|

PT, pallial thickening; and LC, aDC

17
|

aDVR
|
lateral cortex. (E) Spatial

0.04
18
|

|
Rbfox3 pDVR 4 Cluster 7
|

*
mapping of glutamatergic 7
| |
|

CA1 * |
neuron clusters 11, 1, 6, and 7. 14
|

CA3 |
15
|

*
(F) Heatmap illustrating the DG
e34
e33
e38
e37
e36
e35
e32
e31
e30
e29
e28
e27
e26
e24
e16
e15
e14
e13
e08
e07
e25
e23
e22
e21
e20
e19
e18
e02
e01
e06
e05
e04
e03
e17
e12
e11
e10
e09
0

100
% per

Cluster Score
pDC
expression of Satb1, Rorb,

3
region
aLC
Reln, Grik1, and Tbr1. pLC
Spearman’s rho Turtle Glutamatergic Clusters Highest Corr. | -log10 adj. p-value

1
|

(G) HCR for Satb1 and Rorb. Expression PT -0.13 0.00 0.17 Axolotl Turtle 4 8 12 16
1 Max
(H) Projection mapping of
input into glutamatergic F 1
G H Neurobiotin
tracing
1. Main OB 2. Acc. OB 3. Injection site 4. Caudal T. 5. Thalamus

3
cluster 1 by using Neurobiotin- 5
1
2
mediated anterograde and 7
9 3
retrograde tracing. Dots indi- 11
4
13 Satb1
cate locations of cell bodies, 15 5
Max

17
and lines indicate locations of 19
Expression

21
fibers. Main OB, main olfactory 24 Rorb
26 D
bulb; Acc. OB, accessory 28
Satb1
M L
0

DAPI
Satb1

Grik1
Rorb

Tbr1
Reln

olfactory bulb; and caudal T., V

caudal telencephalon. Anterior Posterior

Potential homologies of axolotl glutamatergic cordant similarities, using both gene sets as (12, 13). We found that the majority of medial
neurons to telencephalon neurons were probed indicators of conservation. In parallel, cluster glutamatergic clusters (Glut4, -5, -7, -15, and
with two separate comparisons, by using either similarities were also assessed through inte- -18) and two dorsal clusters (Glut12 and -16)
species-shared differentially expressed genes or gration of single-cell and single-nuclei datasets showed highest correlation to either turtle
species-shared differentially expressed tran- from axolotl, turtle pallium (2), and mouse cornu amonis (CA) or dentate gyrus (DG)
scription factors (TFs) to distinguish between telencephalon (3). clusters (Fig. 2D and fig. S2B). Correlations to
convergent evolution versus homology (Fig. 2D; Anatomical and functional evidence sug- the mouse dataset also revealed that medial
fig. S2, B to F; and tables S2 and S3). We focused gests that the amphibian medial pallium is glutamatergic clusters (Glut5, -7, -14, -17, -24, and
then on glutamatergic clusters that have con- homologous to the mammalian hippocampus -27) correlated most to hippocampus-related

Lust et al., Science 377, eabp9262 (2022) 2 September 2022 3 of 12

RES EARCH | R E S E A R C H A R T I C L E

clusters (fig. S2, D to F). Data integration into the region containing Glut1 (Fig. 2H). level of effector genes, suggesting evolution-
additionally showed co-clustering of a large Labeling of cell bodies in the main and ac- ary divergence of these cells. Together, these
number of medial glutamatergic clusters to cessory olfactory bulb, caudal pallium (lateral data show that the axolotl telencephalon con-
either turtle or mouse hippocampus clusters amygdala) (19), and the thalamus indicate tains putative LGE-derived, CGE-derived, MGE-
(fig S3, A to E). To reveal the locations of that neurons in these regions project into the derived, and septal GABAergic neurons and
hippocampus-correlated glutamatergic clusters Glut1-containing domain. These data show a that LGE-like striatum and olfactory bulb
and unravel a potential subdivision into CA strong correspondence between presence of classes have strong transcriptional similarities
or DG, we performed spatial transcriptomics neurons with transcriptional similarity to between axolotl, turtle, and mouse (tables S2
using Visium (10x Genomics), resulting in spa- amniote olfactory cortex neurons and presence and S3).
tial resolution of approximately 1 to 30 cells of projections consistent with a role in olfac- In mammals and turtles, MGE-derived and
(Fig. 2E and figs. S4A and S5) (14). We found tory processing. CGE-derived interneurons are differentially dis-
that all aforementioned clusters except Glut18 tributed across cortical layers (2, 21). In axolotl,
showed exclusive localization in the medial Amniote-conserved signatures of axolotl we found MGE-like neurons [Somatostatin+
pallium; however, clear subdivisions were not GABAergic neurons (Sst+) and Satb1+] equally distributed in the
detectable, which was further confirmed by We identified 30 clusters (n = 15,665 cells) of medial pallium and enriched in the outer re-
means of HCR for ETS translocation variant 1 GABAergic (Gad1+/Gad2+) neurons (Fig. 3, A to gions of the dorsal and lateral pallium (Fig. 3E
(Etv1) (CA), Prospero homeobox 1 (Prox1), and B, and table S1) in the axolotl. In many ver- and fig. S8A). CGE-like neurons [Zinc finger
LIM domain only 3 (Lmo3) (DG) (fig. S4B). tebrates, GABAergic interneurons are born and BTB domain containing 16+ (Zbtb16+)]
These data show that neurons of the axolotl in the lateral, caudal, and medial ganglionic were equally distributed in the medial pallium
medial pallium have transcriptional similarities eminences (LGE, CGE, and MGE, respectively) but closer to the ventricle in the dorsal pallium.
to amniote hippocampal neurons, but a clear and migrate to the pallium during develop- LGE-like striatal GABAergic neurons [Forkhead
distinction into CA1, CA3, and DG cannot be ment (20). Comparative scRNA-seq analyses box P1+ (Foxp1+)] were found exclusively in the

Downloaded from https://www.science.org on September 24, 2024

observed. in mammals, reptiles, and songbirds have re- striatal region (fig. S8A). Last, we spatially
One cluster from the medial fraction (Glut6) vealed strong conservation of GABAergic inter- mapped LGE-like, CGE-like, and MGE-like
showed highest correlations to clusters of neurons (1, 2). To gain an understanding of the clusters and detected the majority of CGE-like
turtle anterior dorsal cortex (aDC) and pallial conservation of axolotl GABAergic clusters, and MGE-like clusters located in all regions of
thickening (PT). The aDC contains the most we performed cross-species comparisons using the pallium as well as the striatum (fig. S8B).
cell types homologous to mammalian cortical correlation and integration analysis as for the These data strongly suggest cell migration
cell types (2). The correlation to mouse showed glutamatergic neurons. from putative CGE and MGE regions, whereas
highest similarity to clusters belonging to Expression of conserved TFs allowed us to LGE clusters are predominantly located in the
cortical cells (fig. S2D), which was also sup- identify putative LGE-derived, CGE-derived, striatum, as in amniotes.
ported by data integration in which Glut6 was and MGE-derived (hereafter called LGE-like,
also co-clustered with cortical cell types. We CGE-like, and MGE-like, respectively) clusters Ependymoglia diversity in the axolotl
found that this cluster was predicted to be in the axolotl dataset (Fig. 3C). Correlation telencephalon
located at the border between the medial and analysis revealed that 13 out of 14 LGE-like The predominant glial cells in the salamander
dorsal pallium (Fig. 2E). clusters showed high correlations to turtle central nervous system are ependymoglia, which
One function of the amphibian pallium is LGE-derived clusters. Moreover, two out of four generate neurons in development, growth, and
the processing of olfactory input (15). One CGE-like clusters and five out of seven MGE- regeneration (5, 15, 22). In the adult red-spotted
axolotl glutamatergic cluster (Glut1) showed like clusters showed similarities to turtle CGE- newt telencephalon, two ependymoglia types
co-clustering with a cortical cluster and cor- derived and MGE-derived clusters, respectively. have been identified: quiescent type-1 cells that
relations to both cortical clusters and the Our analyses additionally identified five axolotl act as long-term stem cells and proliferative
turtle anterior LC (aLC), the main olfactory- clusters composed of medial cells likely de- type-2 cells that are progenitor-like (22).
input recipient region (Fig. 2D and fig. S2B) rived from the septum, which showed strong We examined the diversity of axolotl telen-
(16). Neurons in the aLC are homologous to correlations to the turtle GABAergic neurons cephalic ependymoglia (3590 cells) and iden-
neurons in the mammalian olfactory (piriform) assigned to septum. These results were addi- tified 15 transcriptionally distinct cell clusters
cortex, and Glut1 also correlated to a mouse tionally supported by data integration in which (Fig. 4A, fig. S9A, and table S1). These clusters
piriform cortex cluster (fig. S2, D and E). The the majority of axolotl LGE-, CGE-, and MGE- grouped into three types of ependymoglia that
piriform cortex contains semilunar cells ex- like clusters co-clustered with the respective were present in all dissected regions: quiescent,
pressing RAR-related orphan receptor beta turtle clusters (fig. S7, A to E). active, and a type that we termed pro-neuro
(Rorb), Reelin (Reln), Glutamate ionotropic Turtle LGE-, CGE-, and MGE-derived GABAergic ependymoglia (Fig. 4A). Quiescent ependymo-
receptor kainate type subunit 1 (Grik1), and neurons could be further subdivided into dif- glia were noncycling and expressed Endothelin
T-Box brain transcription factor 1 (Tbr1) (17), ferent GABAergic classes (2), but their existence 3 (Edn3), active ependymoglia were character-
and these same markers were expressed in in the axolotl was unknown. We found that 11 ized by Notch1 expression and a high cell cycle
Glut1 (Fig. 2F). Moreover, Glut1 expressed out of 13 axolotl LGE-like clusters correlated score, and pro-neuro ependymoglia showed
Special AT-rich sequence-binding protein-1 with either turtle striatum or olfactory bulb expression of neuron-related genes, such as
(Satb1), which we used in combination with GABAergic cells (Fig. 3D and fig. S6, A and B), Glutamate receptor ionotropic 1 (Grin1) (Fig. 4,
Rorb expression to define its location in the and the majority of these (eight clusters) also B and C, and fig. S9, B to D).
dorso-lateral region (Fig. 2G), which is con- correlated to mouse striatum or olfactory bulb We found a clear transcriptional distinction
sistent with the location in the spatial map- GABAergic cells (fig. S6, C to E). By contrast, between medial-, dorsal-, and lateral-derived
ping (Fig. 2E). The mammalian piriform cortex only one out of four CGE-like clusters (GABA 16) ependymoglia, which was most prominent for
receives input from the olfactory bulb, ento- correlated to turtle HTR3A VIP-like neurons. All signaling pathway components (Fig. 4, C and
rhinal cortex, orbitofrontal cortex, and the seven axolotl MGE-like neuron clusters matched D, and fig. S9E). A subset of medial quiescent
amygdala (18). We analyzed axolotl projections to turtle PV-like neurons by TF expression but ependymoglia showed strong expression of
by injecting the bidirectional tracer Neurobiotin had similarities to turtle SST neurons at the Wnt2b, Wnt3a, and Wnt8b—genes that are

Lust et al., Science 377, eabp9262 (2022) 2 September 2022 4 of 12

RES EARCH | R E S E A R C H A R T I C L E

Fig. 3. Conversation A C LGE CGE MGE D LGE CGE MGE Sep.

of axolotl GABAergic GABAergic Neurons 22
* 22

| |
n=15,665
neuron types. (A) UMAP 2 26 * 29
11 * 3

| | | | |
plots of 30 GABAergic 19 * 10
20 18 15 13 * 1 |
14

Axolotl GABAergic Clusters

neuron types (top) and 6 17 9
* 12
23 * 25 |
regional distribution of 4 * 0 |

| | | |
* 8 |
GABAergic neurons types, 16 7 27 * 5
1 * 24
as well as two marker 12 25 * 9

| |
UMAP 2
genes, Lhx6 and Meis2. 21 * 11 | |
10 29 4

|
(B) Heatmap illustrating 3 * 13

|
UMAP 1 8 * 19

|
the expression of differen- 5 28 * 26

|
0 * 15

| |
tial expressed markers for 24 Expression
* 14
* 2 | | |

|
each GABAergic neuron 1 Max
* 17 |

| |
Region Lhx6 Meis2 * 6 | |
cluster. (C) Heatmap illus- * 20 | |

|
* 7 |

|
trating the expression of 23

|
* 16 | | |

|
TFs known to define 18

|
GABAergic subtypes (LGE-, 21

|
* 27

|
CGE-, and MGE-derived) * 28

i16 |
Foxp1
Tshz1
Pbx3

Prox1
Nr2e1

Satb1
Bcl11a

Lmo3
Etv1
Foxp2

Meis2

Nr2f2
Id2
Z eb 2
Sp8

Lhx6
Sox6
Zbtb16

A rx

100

i04
i05
i06
i14
i15

i17
i18
i07
i08
i09
i10
i11
i12
i13
i02
i03
Lateral

0
Dorsal

i01
Medial Undetermined
for each axolotl GABAergic % per
B region
neuron type. (Right) Nxph1 Highest Match
Turtle GABAergic Clusters
St6galnac5 Expression Spearman’s rho | -log10 adj. p-value
Stacked barplots illustrat- 0 Max Medial Lateral

|
Arhgap6 10 20 Dorsal Undetermined
Kcnab1 0.4 0 -0.2 Axolotl Turtle
ing the regional distribution

Downloaded from https://www.science.org on September 24, 2024

of the cells from each
Grm1
Pbx1 E Medial Dorsal Lateral snRNA-seq
Lhx6
Satb1
cluster. (D) Correlation Elfn1 3

Gad2 DAPI
Unc13b
analysis between expres- Synpr
Adarb2 2
sion profiles of axolotl Ncam2
Arpp21
GABAergic neuron types Sulf2 1
Scube2
and turtle GABAergic neu- Shisa6
Adamts6 0
ron types [data are from Pnoc
Ptprd
(2)]. Vertical lines indicate Cntfr 4
DAPI

Baiap3
highest correlated cluster Kl
Dach1
to turtle, and horizontal Dach2 2
Sst

Cpne6
lines indicate highest Syt6
Prkd1
correlated cluster to turtle. Foxp1 0
Bcl11b
Asterisks indicate Snca 4
Ebf1
clusters with agreeing
DAPI

Nova1 3
Gpr149
assignments between all Tac1
Asic4 2
genes correlation and
Satb1

Oprk1
Cdh13 1
integration analysis. Sep., Sorcs1
Tacr3
Septum. (E) HCR and Cntn5 0
Npy
snRNA-seq quantifications Tafa1
Slc17a6 3
Zbtb16 DAPI

for LGE-, CGE-, and MGE- Spock1

Plch2
derived GABAergic neurons Nfib 2
Gabra3
types (Gad2, pan; Sst and Pcbp3 1
Thsd7b
Satb1, MGE; Zbtb16, CGE; Eps8
Chst8 0
and Foxp1, LGE). Scale Tmem88b
Inpp5a
bars, 25 mm. Gipr 3
Foxp1 DAPI

Col19a1
Nmur1
Crtac1 2
Rxfp1
Kctd8
Shisa9 1
Cdh23
Brinp1
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29

0
Expression
Axolotl GABAergic Clusters 0 Max

known to be expressed in the developing me- enriched for expression of Epidermal growth tains three main ependymoglia types (quiescent,
dial pallium in human, mouse, and chicken factor receptor kinase substrate 8-like protein active, and pro-neuro) that divide into regionally
(23). HCR confirmed expression of Wnt2b, 2 (Eps8l2). Because Wnt and Sfrp genes have distinct subtypes and, barring pro-neuro epen-
Wnt3a, and b-catenin target gene Axin2 in a been implicated in patterning of the develop- dymoglia, continue to express pallial patterning
restricted domain of the medial pallium at ing pallium, we performed HCR in embryonic genes in the postembryonic brain.
the border to the septum. Lateral and ventral axolotl brains (stage 44). We detected Wnt3a
ependymoglia showed strong expression of in medial ventricular cells and Sfrp1 lateral Identification of glutamatergic and
Secreted frizzled-related protein (Sfrp1), a and ventral ventricular cells; however, Eps8l2 GABAergic neuroblasts
gene known to be expressed in the antihem was not expressed (Fig. 4E and fig. S9F). These In addition to ependymoglia, we identified
in amniotes (23). Dorsal ependymoglia were data show that the axolotl telencephalon con- cells that we termed neuroblasts because of

Lust et al., Science 377, eabp9262 (2022) 2 September 2022 5 of 12

RES EARCH | R E S E A R C H A R T I C L E

A Ependymoglia 11 C Act. Quiescent Pro-Neu. D Medial Dorsal Lateral

Notch2

Wnt3a
n=3,590 3 Wnt7b
Active 5 Irs1

Etv1 DAPI
4 Cdk14
14 Ephb2
Gli2

Sfrp1 Wnt2b
9 Wnt5b
7 Nox4
13 Meis2 8
Chk1
1 10 6 Fgfr3
0 Ski
Pro-Neuro Shisa6
St6galnac5
Grin1
UMAP 2

Eps8l2
Gria3
12 2 Grin2b
Quiescent Kif1a
UMAP 1 Fgf14
Hip1
B Region

Axin2 DAPI
Notch2 Edn3 Grm3
Eps8l2
Grm8
Lrrtm4
Edn3
Wnt8b
Wnt2b
Wnt3a
Rspo2 Wnt3a D Sfrp1
Medial Lateral Expression Fgfrl1 E Stage 44
D
DAPI DAPI
Dorsal Undetermined 1 Max Ar M M
Sfrp1 P
OT
HB L L
Notch1
OT

G2M Score Notch1 Grin1 Pask

Vax2

0
0

Downloaded from https://www.science.org on September 24, 2024

region
% per

100
4
3
6
5
11
10
12
2
0
9
8
13
1
7
14
F
Expression
0 Max Ependymoglia clusters
Neuroblasts
9
3 1
n=1,501 Slc17a6/7 I Ependymal B cells A cells C Mitosis

2 * 2

|
Late NB * 12

|
11 VGLUT+ * 13

|
13 14 3
*

| | | |
4 * 0
0

Neuroblasts
* 14
7
Axolotl neuroblaset and ependymoglia clusters

Gad2 * 1
* 9

|
Early 12 G * 7
|
1.00

|
8 10 * 6

||
NBs 5
G2M Score

* 5
* 10
|

|
6 GABA+ * 11
|||
0.25
UMAP 2

* 8 | | | | | |
* 4
* 1 |
|| |

UMAP 1 12
E.a E.q E.pn NB.e NB *
|

* 2 | |
H * 6 | | |

Ependymoglia
10
|

11 | | |
Medial

| | | | |

9
5
* 0
Slc17a7 8
Slc17a7 Mex3a
4 |
|

Slc17a7 Mex3a Mex3a DAPI

* 3
| |
|

Gad2 Mex3a Gad2 Gad2 14

|

*
Mex3a Mex3a 13
|

DAPI 7
|
Dorsal

2
23

32
34
5
14
13
22
6
1
4
0
15
12
10
8
16
17
30

Adult Mouse SVZ Clusters

-log10 adj. p-value Highest Corr. | Spearman’s rho
|

100 200 Axolotl Mouse −0.34 0.00 0.42

Fig. 4. Axolotl telencephalic ependymoglia and neuroblasts. (A) (Left) telencephalon. Scale bars, 50 mm. (F) UMAP plots of the regional distribution
Morphology of an axolotl ependymoglia cell and (right) UMAP plot of 15 neuroblast clusters. Black boxes outline the four main cell types: early
15 ependymoglia clusters. Boxes outline the three main cell types: quiescent neuroblasts, VGLUT+ and GABA+ neuroblasts, as well as late neuroblasts.
ependymoglia, active ependymoglia, and pro-neuro ependymoglia. (B) UMAP plot The red box outlines the early neuroblast population. (G) Boxplot of G2M score
of the regional distribution of ependymoglia and UMAP plots colored by for quiescent, active, and pro-neuro ependymoglia as well as neuroblasts.
expression for Notch2 (quiescent and active ependymoglia), Edn3 (quiescent (H) HCR for Slc17a7, Gad2, and Mex3a. Arrows indicate coexpressing cells. Scale
ependymoglia), G2M score (active ependymoglia), Notch1 (active ependymoglia), bars, 10 mm. (I) Correlation analysis between expression profiles of axolotl
and Grin1 (pro-neuro ependymoglia). (C) Heatmap illustrating the expression neuroblasts, ependymoglia, and adult mouse SVZ cell types [scRNA-seq data are
of differentially expressed cell-to-cell communication–related genes in quiescent, from (25)]. Vertical lines indicate highest correlated cluster to mouse, and
active, and pro-neuro ependymoglia. (D) HCR and snRNA-seq quantifications horizontal lines indicate highest correlated cluster to axolotl. Asterisks indicate
for Wnt2b, Wnt3a, Eps8l2, Sfrp1, and Axin2 in medial, dorsal, and lateral regions. clusters with agreeing assignments between all genes correlation and integration
Scale bars, 25 mm. (E) HCR for Wnt3a and Sfrp1 in the stage 44 developing analysis. C, C cells.

Lust et al., Science 377, eabp9262 (2022) 2 September 2022 6 of 12

RES EARCH | R E S E A R C H A R T I C L E

the expression of Mex3a, an RNA-binding pro- these two clusters showed strongest correla- intermediates. We identified many genes with
tein expressed in proliferating neuroblasts in tion to A cell clusters 6 and 15, which were trajectory-specific varying expression along
the Xenopus laevis central nervous system (24). defined previously as dividing neuroblasts and pseudotime, but also some genes with pseudo-
Additionally, neuroblasts were characterized early A cells, as well as to IP cell clusters in the temporal expression consistent across trajecto-
by the absence of ependymoglia markers such developmental dataset (Fig. 4I and fig. S10, A ries (Fig. 5D).
as Gli2, Aquaporin-4 (Aqp4), and Potassium and B). Together, these results show that the To resolve the gene regulatory relationships
inwardly rectifying channel subfamily J axolotl telencephalon contains neuroblast pop- underlying glutamatergic neurogenesis, we
member 10 (Kcnj10). We detected 15 neuro- ulations that already express neurotransmitter leveraged our single-nucleus multiome se-
blast clusters, which formed two distinct groups signatures of downstream neurons. Moreover, quencing of the axolotl whole pallium. We
that expressed either Slc17a6/7 (VGLUT+) or neuroblasts are most similar to mouse progen- identified proximal and distal candidate
Gad1/2 (GABA+) (Fig. 4F; fig. S9, G and H; itor cells and neuroblasts, whereas ependymoglia regulatory regions differentially accessible in
and table S1). In contrast to intermediate pro- harbor transcriptional similarities to mouse each of the terminal glutamatergic neuron
genitors (IPs) found in other vertebrate brains, ependymal cells as well as neural stem cells. clusters and assessed at which stage in the
Mex3a+ neuroblasts were largely nonprolifer- respective trajectory they become accessible
ative (Fig. 4G and fig. S9I). Two neuroblast Transcriptional dynamics of postembryonic (Fig. 5, E and F; fig. S13, A and B; and table
clusters (7 and 8) showed an increased G2M glutamatergic neurogenesis S4). Most elements identified as specific for a
score when compared with the rest and were We labeled ependymoglia using Cre-loxP– given terminal glutamatergic cluster already
therefore termed early neuroblasts. VGLUT+ mediated tracing to investigate their self- obtained accessibility in the corresponding
neuroblasts (except cluster 4) were enriched in renewing properties and determine the clonal neuroblast clusters earlier in the trajectory.
medial and dorsal datasets; GABA+ neuroblasts patterns they generate during post-embryonic We inferred a gene regulatory network (GRN)
(except cluster 5) were mostly found in the neurogenesis (Fig. 5A). This uncovered dis- using Pando (31) by combining gene expres-
lateral dataset. HCR for Scl17a6 or Gad2 in tinct neurogenesis patterns in medial, dorsal, sion, chromatin accessibility, and TF binding

Downloaded from https://www.science.org on September 24, 2024

combination with Mex3a revealed VGLUT+ and lateral regions. Medial and dorsal clones motifs. A uniform manifold approximation and
neuroblasts at the ventricle in all pallial regions, were continuous, spanning from the ventricle projection (UMAP) embedding of this GRN re-
whereas GABA+ neuroblasts were predomi- to the neuronal layers, indicating a stacking vealed distinct groups of TFs, corresponding to
nantly present at the striatum ventricle, with growth mode reminiscent of zebrafish pallial the transition from ependymoglia to glutama-
the exception of a few cells in the pallium (Fig. post-embryonic neurogenesis (27). By contrast, tergic neurons (Fig. 5G and table S5). To bet-
4H and fig. S9J). This pattern was also verified lateral clones were dispersed with labeled neu- ter understand how gene regulation differs
when mapping the neuroblast clusters to our rons separated from ependymoglia, indicating between neuronal trajectories, we performed
Visium data (fig. S9K). neuronal migration. differential accessibility analysis and identified
To define the transcriptional similarities of We used RNA velocity and URD-based tra- regulatory regions enriched within each tra-
axolotl ependymoglia and neuroblasts to mouse jectory inference (28–30) to explore the cellular jectory. On the basis of these regions, we
neural stem and progenitor cells, we performed and molecular dynamics of post-embryonic constructed subnetworks of the GRN that re-
cluster correlation analysis and cross-species neurogenesis (fig. S12, F to T). We focused our flect trajectory-specific regulatory features
data integration with an adult mouse SVZ analysis on glutamatergic neurons that are (Fig. 5H, fig. S13C, and table S5). This allowed
dataset (Fig. 4I and figs. S10, A and B, and S11, known to be generated locally, in contrast to us to identify the TFs with high centrality in a
A to C) (25) and a mouse developmental cor- GABAergic neurons that migrate across the given trajectory, such as Nuclear receptor
tex dataset (figs. S10, C to E, and S11, D to F, pallium from the striatum and for which we subfamily 3 group C member 2 (Nr3c2) in the
and tables S2 and S3) (26). This analysis re- likely miss corresponding progenitor popula- hippocampus or Rorb, Forkhead box O3 (Foxo3),
vealed correlation of quiescent ependymoglia tions in our dataset (20). We focused on tran- and Myocyte enhancer factor 2A (Mef2a) in the
with mouse ependymal cells (adult and devel- sitions from active ependymoglia to the most lateral cortex (Fig. 5I and fig. S13D). Genes dif-
opment) and B cells (adult), whereas active differentiated glutamatergic neurons. To con- ferentially expressed between glutamatergic
ependymoglia (clusters 3 and 4) showed strong struct trajectories, we first identified the key clusters were linked to trajectory-specific TFs
correlation to mitotic cells (adult), including VGLUT+ neuroblast populations that have with high centrality (fig. S13, E to F). Nuclear
dividing A cells (cluster 16) and developmental highest transcriptional similarity to the re- factor 1 X (Nfix) was one of the most central
apical radial glia and IPs. Among the pro- spective glutamatergic neuronal clusters (fig. TFs in all subnetworks; however, the regulomes
neuro ependymoglia, only cluster 1 showed S12A) and are therefore able to connect active controlled by Nfix were distinct for each re-
strong correlation to ependymal cells (adult ependymoglia to neuronal clusters in a trajec- spective trajectory (fig. S13, G and H, and table
and development), whereas clusters 7 and 13 tory. Using these groups of clusters, we then S5). Together, these data highlight the regu-
weakly correlated to ependymal cells (adult) constructed five trajectories that represent dif- latory relationships that shape neuronal diver-
or migrating neurons (development), and ferent region-specific neurogenesis (Fig. 5, B sification in the axolotl telencephalon.
cluster 14 weakly correlated to A cells. Data and C, and fig. S12, B to E). Although all tra-
integration revealed co-clustering of epen- jectories were rooted at the active ependymoglia, Molecular dynamics of axolotl
dymoglia with adult ependymal cells or B cells not all trajectories contained neuroblast in- telencephalon regeneration
and with apical or IPs in the developmental termediates. Hippocampal neuronal clusters, To study the cellular and molecular dynamics
dataset. lateral cortex clusters, and laterally derived during axolotl telencephalon regeneration, we
Comparison of axolotl neuroblasts to the clusters, including the lateral pallium group implemented Div-Seq (32), which combines
adult mouse SVZ dataset revealed strong cor- and the Eomesodermin (Eomes) group, all snRNA-seq with EdU labeling of S-phase cells.
relation and co-clustering with A cells, sup- used neuroblast intermediates. By contrast, We injured the dorso-lateral region of the
ported by strong correlation and co-clustering the dorso-medial neuronal clusters formed a telencephalon (including the Satb1+, Rorb+
with migrating neurons in the developmental group with lower correlations with the least domain) by excising a 1- by 1- by 1-mm region
dataset (fig. S10, C to E). We found that early differentiated neuroblast clusters and were and applied Div-Seq throughout regeneration
neuroblast clusters 7 and 8 correlated also to thus inferred to originate through a direct by labeling cells with EdU at 2, 5, 12, 19, and
either C cells or mitotic cells. Furthermore, trajectory using pro-neuro ependymoglia as 26 days post injury and collecting EdU+ cells

Lust et al., Science 377, eabp9262 (2022) 2 September 2022 7 of 12

RES EARCH | R E S E A R C H A R T I C L E

Fig. 5. Gene regulatory A Stacking Medial Dorsal

Migration
Lateral
C Hippocampal trajectory Lateral cortex trajectory Dorso-medial trajectory
programs underlying

proportions
Cell state
post-embryonic neuro-
genesis. (A) Schematic
D Glut 0
Pseudotime Pseudotime Pseudotime
describing the out- Glut 1 Glut 6
vs 1
comes of Cre-loxP fate

Exp.
mapping performed to
assess clonal dynamics 0
Pseudotime Pseudotime Pseudotime
and potential clone Elmo1
Etv1
Gria3
Msi2
Nfix
Sez6l
Elmo1
Foxo3
Mef2c Rorb
Nfix Sema5a
Foxp1 Msi2 Sez6
Htr5a Nfix Slc2a13
shapes during homeo- Glut 7 Glut 11 Glut 8
static neurogenesis of B Hippocampal Lateral cortex Dorso-medial 1

Exp.
trajectory trajectory trajectory
the axolotl telencephalon Glut Glut 0 Glut 11 Glut 6
adjacent to measured NB 13 0
NBs
NBs 4, 7 Pseudotime Pseudotime Pseudotime
Epen
clonal patterns in medial, Etv1 Meis2 Sez6l Bcl11a Mex3a Rbfox3 Foxo3 Msi2 Slc2a13
Glra2 Nfix Shisa6 Mef2c Nfix Rorb Htr5a Nfix Sox6
NBs 3,9
dorsal, and lateral pal- Glut 7
E

Normalized accessibility

Normalized accessibility
lium. (B) Glutamatergic NB 2 Glut 0 Glut 0
NBs 7,11
trajectories reflecting Glut 7 Glut 7

(range 0 − 1.9)

(range 0 − 2.1)
NB 1
postembryonic neuro- Glut 1
Glut 1 Glut 1
Glut 8
genesis of axolotl neurons Glut 11 Glut 11

matched to amniote hip- Glut 6 Glut 6

pocampus, lateral cortex, Glut 8 Glut 8
and the dorso-medial

Downloaded from https://www.science.org on September 24, 2024

0 4 0 exp.3
Genes Kcnq5 exp. Genes Zfpm2
pallium. UMAPs are
Pseudotime
colored by cell type (top) 0 1
Peaks Peaks
623167449 623169449 909809000 909811000
and pseudotime (bottom). F Proximal and distal linked DA peaks grouped by class chr4p position (bp) chr5q position (bp)
Pie charts represent 6
Trajectoreis to terminal

the regional composition 8

Glut. clusters

of neuron clusters. 7
(C) Pseudotemporal cell 0
type progression from
11
ependymoglia to gluta-
matergic neurons during 1
Ndst4
Trarg1
Shisal1
Reln
Grik3
Klhdc8b

Mdga1
Plpp4
Satb1
Nmb

Grik3
Tmem88b
Syndig1
Rapgef4

Mmel1
Opn3
Slitrk1
Mras
Elmo1
Megf11

Shank3
Slit3
Daam1
Epha4
Grin2a
Atp2b1

Eml4
Nrp1
Wnt4
Tp53i11

Trpm4
Sorbs1
Btbd3
Tmcc1
Rims4
Negr1
Unc13b
Garem1
Rtn4rl1
C1ql1
Brs3
Grik1
Fam131a

Oprm1
C1ql3
Pde1b

Fibcd1
Garem1
Fam131b
Tiam1
Rtn4rl1
C1ql1
Kank2

Atp2b2

Cpne2
Msi2

Plxdc2
Brinp2

Meis2

Pdzd2

Rhbdf2
Flrt2

Nr3c2

C1ql2

Kank2
Npffr2

Slc35g2
Unc13c

Unc13c

Shisa6
Sstr5

St6galnac5

Sstr5
Kcnq5

Kcnq5
Kcnh7

Cabp7

Cpne7
neurogenesis in the three

Ptprt
Penk

Nsmf
trajectory groups.
(D) Pseudotemporal gene Glut NBs Epen acc. scale

G H I 100
−2.5 5.0
expression changes dur- Nr3c2 Etv1

% specific
Global Gene Zfpm2 Aff3 2.0 0.4
ing neurogenesis for each Hippocampal trajectory
exp.

Mitf
Zeb2
Neuronal % exp. 0.3
Regulatory Network St18 Hivep2 0.25 0.5
terminal branch from the -log10(padj) Neurod4 0.50 50 0.2
Lhx9 0.75
5 Nfix
trajectories of each 10 # of outgoing Nfib Nfic Cux2
0
Number of Connected Genes

15 connections Znf469
group. (E) Representa-
Nfib

Nfia
Nfatc3
Stat5b
Etv1

Mef2a
Tcf4
Eomes
Rorb

Foxo3
Gli3
Foxp1
Nr3c2

Pou2f2
Meis2
Nfic

Pknox2

Tcf7l2
Gli2

% specific Bach2
Mef2c

Sox6
Nfix

Ahr
Mitf
20 0
100
tive example peaks Interactions Hippocampal
Positive 200
associated with (left) Negative 300 Tox2 1.6 40 Lateral cortex trajectory 0.2
exp.

Bcl11b Rorb % exp.

Kcnq5 and (right) Foxp1 Pknox2 Runx2 0.2 0.4 20 0.1
Satb1 0.4
Zfpm2. (F) Heatmap of Ependymoglia Nr2f2 Lin28b
Znf831 Mef2c 0.6
Lhx1 0.8
Arid5b 0
chromatin accessibility Zbtb20 Nfea Tshz2 Ebf3 Mlxipl
Mkx Bcl11a
Mef2a

Nfib
Rorb
Nfia
Gli3

Rreb1
Eomes

Tcf7l1
Foxo3
Tcf4
Zeb1
Etv1
Tead4
Rfx4
Nfatc3
Pbx3
Meis2
Pknox2
Gli2
Tcf7l2

Nr3c2
Mef2c
Nfic
Nfix

Mef2d

Tead1 Etv4 Nfil3

changes in distal and Rreb1 Hes5
Ahr Tshz3 Mef2a
Zfhx3
Paxbp1 Meox2 Tfap2e Lhx5 Eomes Tshz1 Lateral cortex
proximal elements for Sall2 Gli1 Ar
Rfx4
Plagl1 Esr1 Fosl2 Meis2 80
Gli3 Lef1 Foxa1 Tfec Foxp2 1.6 0.5 % specific
Tcf7l1 Zfp36l1 Npas4
glutamatergic clusters
exp.

Tead4
Znf536
Egr4 Foxp4 % exp. Dorso-medial trajectory 0.3
Gli2 Npas3 Prdm16 Fosb Nr4a1 Meis1 Pbx3 0.25 0.4 40
8, 6, 7, 0, 11, and 1. Tcf7l2 Creb5 Sox6
Sox2
Foxm1
Pbx1 0.50 0.1
Dach1 0.75
Zeb1
(G) UMAP embedding of Znf516
Bnc2 E2f1
E2f2 0
the inferred gene mod- Dach2 Nsd2 Sall3
Gli3

Rreb1
Rfx4
Tcf7l1
Nfib
Nfia
Tead4

Mef2a
Eomes

Etv1
Nfatc3
Foxo3
Tcf4
Lef1
Rorb
Foxp1
Gli2
Tcf7l2

Nr3c2
Pknox2

Meis2

Foxp2
Nfic

Sox6
Nfix

Hivep1 Rxra
ules based on coex- Dorso-medial
pression and inferred Transcription Factors

interaction strength
between TFs. Size represents the number of connections for each TF. Color scale indicates expression, and size indicates the percent of cells
(H) Trimmed GRN UMAP embedding of the inferred gene modules based expressing. (I) Barplot of the top 25 TFs ranked by number of connections for
on coexpression and inferred interaction strength between TFs for (top) each TF and colored by fraction of trajectory-specific peaks out of total
hippocampal, (middle) lateral cortex, and (bottom) dorsal-medial trajectories. number of peaks in the global module.

at 1, 2, 4, 6, 8, and 12 weeks post injury (Fig. 6A). erating brains (Fig. 6B and fig. S14, A and B). At 2 weeks post injury, the injury site was
To visualize the location of EdU+ cells during At 1 week post injury, the injury site was still starting to close from the accumulation of
the regeneration process, we fluorescently open, and EdU+ ependymoglia were present EdU+ cells. Throughout the following time
labeled EdU using click chemistry on regen- in medial and lateral wound-adjacent regions. points, EdU+ cells remained accumulated at

Lust et al., Science 377, eabp9262 (2022) 2 September 2022 8 of 12

RES EARCH | R E S E A R C H A R T I C L E

A ∗Div-Seq / Fixation EdU B 1 wpi 2 wpi 4 wpi

Injury ∗ ∗ ∗ ∗ ∗ ∗

EdU DAPI
D0 D5 1 wpi D12 2 wpi D19 D26 4 wpi 6 wpi 8 wpi 12 wpi

C Endo D Cell type changes over regeneration

0.5

proportion
Epen MG 0.4
6 wpi 8 wpi 12 wpi
0.3

EdU DAPI
Div-seq NB 0.2
n=19300 0.1
0.0
1 2 4 6 8 12
time (wpi)
UMAP 2

Glut
Epen a Epen q Glut Oligo MG
Fractional abundance of ependymoglia clusters per time point
Epen n NB GABA Endo
UMAP 1 GABA
F
Active Quiescent Pro-neuro

2e-04 4e-04
norm. % occurr.
Ependymoglia 1 wpi
Steady-State 3 Clusters 2 wpi
n=3590 5 4 wpi
4 6 wpi
8 wpi
Div-Seq 8 10 19 12 wpi

0
n=3753 13 SS
12
21
G
11 16 1 172213 4 14 6 15 9 1211 0 2 8 20 2118 7 19 5 1016 3
20 22 % Exp.
18 Plat 20
17 Anxa2 40
2 0
1 wpi 8 wpi 7 14 Kazald1 60
UMAP 2

2 wpi 12 wpi Erbb3 80

4 wpi SS 1 Col7a1 Exp.
6 wpi 15 Runx1 2

Downloaded from https://www.science.org on September 24, 2024

UMAP 1 6
9 1
Ependymoglia Cluster 0
-1
GO Term expression over time
H I 2 dpi 1 wpi

Kazald1 Runx1 DAPI

Wound Collagen Positive reg.of Chromosome Neurotrans. Uninjured
healing fibrill org. cell adhesion segregation transport
1 wpi
2 wpi
4 wpi
6 wpi
8 wpi
12 wpi
SS
K
Erbb3

Anxa1
Fn1

Col7a1
Col3a1

Runx1

Akt1
Cbfb

Ncaph

Smc4
Bub1
Nrxxn1
Syp
Smc2
Col1a2
Anxa2
Plod2

S100a10

Slc6a12
Slc1a2
Ccdc80
Tnc

Dlgap5

Snap25
Plat

2 wpi 4 wpi 8 wpi

Satb1 Rorb DAPI

J 1 wpi 2 wpi 4 wpi 6 wpi 8 wpi 12 wpi

Glut neurons: 1
3
Predicited Glut. Cluster

Div-seq 0
projected
to SS
9
5
L GlutGlut
1 1
M Neurogenesis
NB 7,13
8 Glut 1 Glut 11
Correlation (Div-seq)

2
10
0.5

0.4
11 Glut 11
12
NB 4, 2
UMAP 2

4
0.0

0. 0
24
13
1 wpi 8 wpi 6
UMAP 2

−0.5

UMAP 1 −0.4
2 wpi 12 wpi 15
4 wpi SS −0.5 0.0 0.5 −0.5 0.0 0.5
0

600
0

600
0

125
0

150
0

500
0

100

6 wpi Pseudotime
UMAP 1 0 1 Correlation (Steady-state)
N 1 Main OB 2 Acc. OB NB 1 Main OB 2 Acc. OB
NB
8 weeks post injury

3
3
Non-injured

1
1
4
2 3 Injec. site 4 Caudal pal. 4 3 Injec. site 4 Caudal pal.
2

D D
A P A P
V V

Fig. 6. Axolotl telencephalon regeneration after injury. (A) Schematic cluster occurrence per time point. (G) Dotplot of selected ependymoglia
describing the Div-Seq protocol used during axolotl telencephalon regener- cluster 21 differentially expressed genes. (H) Gene average expression
ation. (B) EdU stainings of all regeneration time points. Scale bars, 50 mm. change per time point, grouped by GO terms. (I) HCR for Kazald1 and Runx1 in
(C) UMAP plot of all EdU+ cells across all regeneration time points, colored by steady state, 2 days post injury (dpi) and 1 week post injury (wpi). Scale
cell-type class. Predicted cell clusters are shown in different shades. GABA, bars, 50 mm. (J) (Left) Correlation projection of all Div-Seq glutamatergic
GABAergic neuron; Glut, Glutamatergic neuron; NB, neuroblast; Epen, neurons (pink) to steady-state glutamatergic neurons (gray). (Right) Barplots
ependymoglia cell; Endo, endothelial cell; and MGs, microglia. (D) Change of largest glutamatergic neuron clusters recovered throughout regeneration
in cell type relative abundance along regeneration time points. (E) (Left) time points. (K) HCR for Satb1 and Rorb throughout regeneration. Scale
UMAP plot of all Div-Seq ependymoglia (shades of pink indicate different time bars, 50 mm. (L) Trajectories reflecting regenerative neurogenesis of
points) and noninjured brain steady-state (SS) ependymoglia (gray). (Right) glutamatergic neuron clusters 1 and 11. UMAPs are colored by (left) cell type
UMAP plot of clustering of all ependymoglia. (F) Heatmap of normalized and (right) pseudotime. (M) Correlations of lineage driver genes with the

Lust et al., Science 377, eabp9262 (2022) 2 September 2022 9 of 12

RES EARCH | R E S E A R C H A R T I C L E

assignment probability for (top) Glut1 and (bottom) Glut11 trajectories, in domain. (Insets) Labeled cell bodies in the olfactory bulb, accessory olfactory
(y axis) regenerative and (x axis) steady-state neurogenesis. (N) Whole- bulb, injection site, and caudal telencephalon. Scale bars, (overviews)
mount cleared Neurobiotin stainings on (left) noninjured brains and (right) 100 mm; (zooms) 50 mm. Main OB, main olfactory bulb; Acc. OB, accessory
brains 8 weeks post injury. Neurobiotin was injected in the Satb1+, Rorb+ olfactory bulb; and caudal T., caudal telencephalon.

the regeneration site until tissue architecture state (Fig. 6F). Cluster 21 cells differentially reestablish afferent and efferent projections.
was largely reestablished. expressed genes such as Kazal type serine We injected Neurobiotin into the Satb1+, Rorb+
Next, we investigated the transcriptomes of peptidase inhibitor domain 1 (Kazald1) and domain in noninjured as well as regenerated
EdU+ cells during the regeneration time course. RUNX family transcription factor 1 (Runx1) brains at 8 weeks post injury and performed
We used our steady-state data as a reference and were enriched in Gene Ontology (GO) whole-mount immunohistochemistry to iden-
and identified all major cell types, including terms relating to wound healing and cell tify cell bodies and projections. Similarly to
ependymoglia, neuroblasts, glutamatergic and adhesion, indicating early response programs noninjured brains, stained cell bodies were
GABAergic neurons, as well as endothelial to injury (Fig. 6, F to H). Staining for Kazald1 located in the olfactory bulb, accessory olfactory
cells and microglia (Fig. 6C). Each cell type and Runx1 confirmed absence of expression bulb, and the caudal telencephalon (amygdala),
was represented in different proportions in the uninjured telencephalon and strong indicating that the input from these regions is
throughout regeneration (Fig. 6D and fig. S14C), expression in a subpopulation of cells at reestablished in the regenerated telencephalon
reflecting a wave of neurogenesis induced by 1 week post injury, with an induction of ex- at 8 weeks post injury (Fig. 6N).
the injury. Whereas at 1 week post injury, active pression as early as 2 days post injury (Fig. 6I).
ependymoglia constituted the majority of EdU+ Cluster 8 was enriched for Lc27 and Inositol Discussion
cells, neuroblasts were the most abundant cell polyphosphate-5-phosphatase (Inpp5d), whereas Using snRNA-seq, multiomic sequencing, and
type at 2 and 4 weeks post injury. Starting from cluster 21 was enriched for cilia-related genes, Div-Seq along with spatial transcriptomics,

Downloaded from https://www.science.org on September 24, 2024

6 weeks post injury, most of the EdU+ cells which could relate to reestablishment of the Cre-loxP tracing, HCR, and antibody staining,
were glutamatergic and GABAergic neurons. ependymoglia layer (table S6). we have generated a comprehensive single-
In line with the results from the Div-Seq data, Projection and classification of Div-Seq neuro- cell atlas of the axolotl telencephalon during
HCR and antibody staining for ependymoglia blasts and neurons based on steady-state data homeostasis and regeneration. Comparative
(GFAP+, Eps8l2+, and Sfrp1+) and neuroblast showed that a majority of steady-state popula- analysis with turtle and mouse datasets al-
(Mex3a+) populations demonstrated that tions had been reestablished during regener- lowed us to reveal transcriptional similarities
ependymoglia were recovering radial mor- ation (Fig. 6J and fig. S14, C, I, and J). Among of axolotl telencephalon cell types and their
phology and regional identity, whereas neuro- glutamatergic neurons, all but one steady- conservation between tetrapods.
blasts were accumulating at the wound site state cluster was captured in the Div-Seq data Axolotl glutamatergic neurons showed lower
between 2 and 4 weeks post injury (fig. S14, D (fig. S14C), with the most expanded clusters pairwise correlations between species com-
and E). We next inferred regional identities predicted to be of dorso-medial origin (fig. S14, pared with those of other cell populations,
of Div-Seq cells by transferring region labels C, E, and F). HCR staining demonstrated the indicating their evolutionary diversification.
from our steady-state data and found that recovery of Satb1+, Rorb+ glutamatergic neurons Nonetheless, we found glutamatergic neurons
early cell types (ependymoglia and neuroblasts) between 4 and 8 weeks post injury (Fig. 6K). To similar to amniote hippocampus, turtle aDC,
largely consisted of dorsal regional transcrip- explore the dynamics of regenerative glutama- and olfactory cortex. Glut1 cells exhibited tran-
tional identities. This predominance of dorsal- tergic neurogenesis, we first determined the scriptional similarities to the amniote olfactory
like cells was propagated to glutamatergic cells correlation of regenerating neuroblasts to re- cortex, and consistently, these neurons also
appearing in weeks 6 to 12, whereas GABAergic generating glutamatergic neurons and found showed olfactory bulb input, indicating a
neuronal populations were dominated by a a similar correlation pattern as that in ho- conserved role in olfactory processing. Ad-
lateral identity (fig. S14, F to G). This data as meostatic neurogenesis (fig. S15A). We then dressing the functional properties and input-
well as EdU staining and comparison of EdU+ constructed trajectories and identified high output connectivity will be critical to gain a
cells between uninjured and regenerating similarity between regenerative and steady- better understanding of conservation. Our
telencephalon (fig. S14H) suggests that our state neurogenesis trajectories regarding pseudo- multiomic sequencing analysis has revealed
Div-Seq data largely captured cells in the acute temporal ordering of cells and correlation of differentially accessible regions in Glut1, which
injury area of the dorso-lateral pallium but lineage driver genes (Fig. 6, L and M, and fig. could be used in the future to achieve targeted
likely also contains a minority of cells derived S15, B to G). Most TFs were similarly detected expression of connectivity and optogenetic and
from homeostatic neurogenesis in areas not in regenerative and steady-state neurogenesis chemogenetic tools.
associated with the injury site. trajectories, with highly correlated timings and We identified LGE-like, CGE-like, MGE-
Previous studies on axolotl spinal cord re- high centrality in trajectory-specific GRNs like, and septal GABAergic neurons in the
generation showed that injury-induced epen- (fig. S16). Despite these similarities, some axolotl and found conservation of LGE-like
dymoglia activate a transcriptional signature gene expression differences could also be striatal and olfactory bulb classes between
similar to that of embryonic neuroepithelial observed with steady state–specific genes axolotl and other tetrapods. The LGE and MGE
cells (33). To understand whether telencephalon enriched in GO terms related to cell cycle have been found in all studied vertebrates—
injury induces injury-specific transcriptomic and cell adhesion and regeneration-specific including anamniotes such as lamprey, fish,
changes in ependymoglia, we compared Div- genes enriched in neurogenesis and neu- and amphibians (34–36)—but the existence of
Seq and steady-state ependymoglia by integrat- ronal activity (fig. S17). Part of the differ- the CGE in anamniotes is unclear. Our identi-
ing and clustering both datasets and assessing ential expression might have technical origins fication of putative CGE-derived neurons in
differential abundance and expression in each because of differences in transcriptomic this work hints at the existence of a CGE in
cluster across time points (Fig. 6E). We found coverage between the samples (fig. S14, A axolotl. GABAergic neuron migration has not
three clusters (clusters 8, 20, and 21) strongly and B). been studied in salamanders, and it will be im-
enriched at 1 and 2 weeks post injury that were Last, we set out to determine whether the portant to determine the developmental ori-
rare or absent at later time points and steady regenerated Satb1+, Rorb+ neuron domain would gin and timing of GABAergic neurogenesis in

Lust et al., Science 377, eabp9262 (2022) 2 September 2022 10 of 12

RES EARCH | R E S E A R C H A R T I C L E

the putative ganglionic eminences. In axolotl, facilities, and each animal is kept individually. Brain regeneration with Div-Seq
GABAergic neurogenesis likely continues in All handling and surgical procedures were Axolotls were injected intraperitoneally with
the postembryonic brain because we detected carried out in accordance with the local ethics EdU. After the desired pulse-chase period,
GABA+ neuroblasts and found one cluster (NB 5) committee guidelines. Animal experiments were nuclei were isolated, EdU staining was per-
that contains cells from medial and dorsal performed as approved by the Magistrate formed immediately by using Click-iT EdU
pallium, which suggests local pallial GABAergic of Vienna (Genetically Modified Organism Flow Cytometry assay Kit (Thermo Fisher
neurogenesis, a phenomenon observed in the Office and MA58, City of Vienna, Austria, Scientific, #C10424), and EdU+ nuclei were
developing primate brain (37). license GZ51072/2019/16 and license GZ665226/ sorted by means of fluorescence-activated cell
We found that axolotl ependymoglia show 2019/21). sorting (FACS). Nuclei from the Div-Seq dataset
transcriptomic signatures of both mouse SVZ were classified into cell types, clusters, and
ependymal cells and B cells and function as snRNA-seq library preparation and sequencing
brain regions by using the steady-state data
stem cells during homeostatic neurogenesis. We used snRNA-seq (10x Genomics) to profile as a reference. Subsequently, ependymoglia
Postembryonic ependymoglia in different pal- medial, dorsal, and lateral regions of the were subset from the steady-state and Div-Seq
lial regions maintained expression of develop- telencephalon. We additionally profiled all datasets and integrated by using harmony,
mental patterning genes thought to regulate these regions as a whole with single-nucleus with the library chemistry as a covariate.
dorsal and ventral pallial subdomain size multiome sequencing (10x Genomics). Libra- Neurogenesis trajectories in regeneration were
during development in amniotes (23). It is ries were sequenced and then aligned to the inferred similarly to steady-state trajectories.
possible that expression of these factors main- axolotl genome. Control and regenerated brains were injected
tains pallial domain proportions during con- with neuronal tracer Neurobiotin, whole-mount
tinuous neurogenesis. The mammalian SVZ Data integration and clustering
stained, and cleared to determine projection
also contains IPs and migratory neuroblasts Datasets from each chemistry were integrated patterns.
that express GABAergic or glutamatergic fate by using harmony integration. Identification

Downloaded from https://www.science.org on September 24, 2024

markers (38, 39). The axolotl contains differ- of major cell types followed an iterative clus-
REFERENCES AND NOTES
ent types of neuroblasts: early, potentially pro- tering approach by using the Louvain algorithm.
1. B. M. Colquitt, D. P. Merullo, G. Konopka, T. F. Roberts,
liferating neuroblasts with high transcriptional Identities of the clusters were assigned on the M. S. Brainard, Cellular transcriptomics reveals evolutionary
similarity to mammalian IPs, and late VGLUT+ basis of the expression of established marker identities of songbird vocal circuits. Science 371, eabd9704
or GABA+ neuroblasts with transcriptional genes. For each major cell type (glutamatergic (2021). doi: 10.1126/science.abd9704; pmid: 33574185
2. M. A. Tosches et al., Evolution of pallium, hippocampus, and
similarities to mammalian neuroblasts. The neurons, GABAergic neurons, ependymoglia, cortical cell types revealed by single-cell transcriptomics in
majority of axolotl VGLUT+ neuroblasts could and neuroblasts), the clustering procedure reptiles. Science 360, 881–888 (2018). doi: 10.1126/
be subdivided depending on their similarity repeated. science.aar4237; pmid: 29724907
3. A. Zeisel et al., Molecular architecture of the mouse nervous
to different glutamatergic neurons, impli-
Comparison with turtle and mouse datasets system. Cell 174, 999–1014.e22 (2018). doi: 10.1016/
cating lineage restriction that should be in- j.cell.2018.06.021; pmid: 30096314
vestigated in the future by using genetic Four single-cell datasets were downloaded for 4. R. D. Hodge et al., Conserved cell types with divergent features
barcoding (40). interspecies comparison: mouse brain, mouse in human versus mouse cortex. Nature 573, 61–68 (2019).
doi: 10.1038/s41586-019-1506-7; pmid: 31435019
In contrast to mammalian glial cells, sal- development, mouse SVZ, and turtle forebrain. 5. R. Amamoto et al., Adult axolotls can regenerate original
amander ependymoglia show neurogenic We used two approaches to compare the ex- neuronal diversity in response to brain injury. eLife 5, e13998
activity after injury. We found that axolotl pression profiles cells between species: (i) pair- (2016). doi: 10.7554/eLife.13998; pmid: 27156560
6. M. Maden, L. A. Manwell, B. K. Ormerod, Proliferation zones in
ependymoglia go through an injury-specific wise correlations between clusters by using all the axolotl brain and regeneration of the telencephalon. Neural
transcriptional state defined by the up- differentially expressed genes or differentially Dev. 8, 1 (2013). doi: 10.1186/1749-8104-8-1; pmid: 23327114
regulation of genes involved in wound healing expressed TFs and (ii) integration of single-cell 7. F. Doetsch, J. M. García-Verdugo, A. Alvarez-Buylla, Cellular
composition and three-dimensional organization of the
and cell migration. The early up-regulated and single-nuclei datasets. subventricular germinal zone in the adult mammalian brain.
gene Runx1 is also expressed in the mesen- J. Neurosci. 17, 5046–5061 (1997). doi: 10.1523/
chymal limb blastema of axolotl and X. laevis Region-specific neurogenesis JNEUROSCI.17-13-05046.1997; pmid: 9185542
(41, 42) and planarian neoblasts after ampu- (pseudotime analysis) 8. J. Kaslin, J. Ganz, M. Brand, Proliferation, neurogenesis and
regeneration in the non-mammalian vertebrate brain. Philos.
tation (43), suggesting its involvement in a Glutamatergic NB progenitors were associated Trans. R. Soc. Lond. B Biol. Sci. 363, 101–122 (2008).
generic injury response regardless of cell type with their corresponding differentiated pop- doi: 10.1098/rstb.2006.2015; pmid: 17282988
and species. Apart from the early wound re- ulations through correlation. For each of the 9. V. Hartenstein, A. Stollewerk, The evolution of early
neurogenesis. Dev. Cell 32, 390–407 (2015). doi: 10.1016/
sponse, regenerative neurogenesis is very sim- five groups determined, a pseudotemporal tra- j.devcel.2015.02.004; pmid: 25710527
ilar to homeostatic neurogenesis. We found jectory based on RNA velocity was inferred by 10. M. Faiz et al., Adult neural stem cells from the subventricular
that Satb1+/Rorb+ glutamatergic neurons re- use of scVelo and CellRank. The resulting tra- zone give rise to reactive astrocytes in the cortex after
stroke. Cell Stem Cell 17, 624–634 (2015). doi: 10.1016/
generate after injury and reestablish input jectories were used to determine lineage driver j.stem.2015.08.002; pmid: 26456685
connectivity from the olfactory bulb. The genes, and gene expression was modeled by 11. A. Arvidsson, T. Collin, D. Kirik, Z. Kokaia, O. Lindvall, Neuronal
conservation of transcriptional profiles, con- using a spline with five degrees of freedom. replacement from endogenous precursors in the adult brain
nectivity, and potential function of these after stroke. Nat. Med. 8, 963–970 (2002). doi: 10.1038/
GRN construction nm747; pmid: 12161747
neurons makes them suited to study recovery 12. G. Roth, G. Westhoff, Cytoarchitecture and connectivity of the
of functional neuronal circuits. The R package Pando was used to infer a GRN amphibian medial pallium. Eur. J. Morphol. 37, 166–171 (1999).
from multiome data of cells from all gluta- doi: 10.1076/ejom.37.2.166.4759; pmid: 10342450
Methods summary matergic trajectories. To select candidate re- 13. M. I. Sotelo, M. F. Daneri, V. P. Bingman, R. N. Muzio,
Animal use Telencephalic neuronal activation associated with spatial
gions, peaks were linked to genes according memory in the terrestrial toad Rhinella arenarum: Participation
White (d/d) axolotls were used for all experi- to correlation with a distance threshold of of the medial pallium during navigation by geometry. Brain
ments. Animals of a size of 10 to 11 cm nose to 10 megabase. Vertebrate motif annotations Behav. Evol. 88, 149–160 (2016). doi: 10.1159/000447441;
pmid: 27889766
tail were used for all sequencing experiments. were obtained from JASPAR2020. Modules
14. E. Zhao et al., Spatial transcriptomics at subspot resolution
All lines were bred and maintained in Re- were identified with a significance thresh- with BayesSpace. Nat. Biotechnol. 39, 1375–1384 (2021).
search Institute of Molecular Pathology (IMP) old of false discovery rate (FDR) < 0.05. doi: 10.1038/s41587-021-00935-2; pmid: 34083791

Lust et al., Science 377, eabp9262 (2022) 2 September 2022 11 of 12

RES EARCH | R E S E A R C H A R T I C L E

15. A. Joven et al., Cellular basis of brain maturation and dynamical modeling. Nat. Biotechnol. 38, 1408–1414 (2020). associated with regeneration initiation in planarians. Genes
acquisition of complex behaviors in salamanders. doi: 10.1038/s41587-020-0591-3; pmid: 32747759 Dev. 26, 988–1002 (2012). doi: 10.1101/gad.187377.112;
Development 145, dev.160051 (2018). doi: 10.1242/dev.160051; 29. M. Lange et al., CellRank for directed single-cell fate mapping. pmid: 22549959
pmid: 29217751 Nat. Methods 19, 159–170 (2020). 44. A. Maynard, tomasgomes/pallium_evo: Post-revision 1 release.
16. P. S. Ulinski, W. T. Rainey, Intrinsic organization of snake 30. J. A. Farrell et al., Single-cell reconstruction of developmental Zenodo (2022); doi: 10.5281/zenodo.6627781
lateral cortex. J. Morphol. 165, 85–116 (1980). doi: 10.1002/ trajectories during zebrafish embryogenesis. Science 360, 45. K. Lust et al., Single-cell analyses of axolotl forebrain
jmor.1051650108; pmid: 30180381 eaar3131 (2018). doi: 10.1126/science.aar3131; organization, neurogenesis, and regeneration. Zenodo (2022);
17. A. Diodato et al., Molecular signatures of neural connectivity in pmid: 29700225 doi: 10.5281/zenodo.6390083
the olfactory cortex. Nat. Commun. 7, 12238 (2016). 31. J. S. Fleck et al., Inferring and perturbing cell fate regulomes in
doi: 10.1038/ncomms12238; pmid: 27426965 human cerebral organoids. bioRxiv 457460 [Preprint] (2021). AC KNOWLED GME NTS
18. D. M. G. Johnson, K. R. Illig, M. Behan, L. B. Haberly, New doi: 10.1101/2021.08.24.457460 We thank the Treutlein, Camp, and Tanaka laboratories for their
features of connectivity in piriform cortex visualized by 32. N. Habib et al., Div-Seq: Single-nucleus RNA-seq reveals discussions, input, and support. FACS support was provided by the
intracellular injection of pyramidal cells suggest that “primary” dynamics of rare adult newborn neurons. Science 353, single-cell facility at D-BSSE, ETH Zurich. Illumina sequencing
olfactory cortex functions like “association” cortex in other 925–928 (2016). doi: 10.1126/science.aad7038; was done in the Genomics Facility at D-BSSE, ETH Zurich. We
sensory systems. J. Neurosci. 20, 6974–6982 (2000). pmid: 27471252 thank the IMP Bio-Optics facility and the animal caretaker team
doi: 10.1523/JNEUROSCI.20-18-06974.2000; pmid: 10995842 33. A. Rodrigo Albors et al., Planar cell polarity-mediated induction (Vienna Biocenter) for outstanding service. Funding: This work
19. N. Moreno, A. González, Evolution of the amygdaloid of neural stem cell expansion during axolotl spinal cord was supported by a Long-Term Fellowship from the Human
complex in vertebrates, with special reference to the anamnio- regeneration. eLife 4, e10230 (2015). doi: 10.7554/ Frontier Science Program LT000605/2018-L (K.L.); EMBO Long-
amniotic transition. J. Anat. 211, 151–163 (2007). eLife.10230; pmid: 26568310 Term Fellowship ALTF 738-2019 (T.G.); European Research Council
doi: 10.1111/j.1469-7580.2007.00780.x; pmid: 17634058 34. F. Sugahara et al., Evidence from cyclostomes for RegGeneMems 742046 (E.M.T.); Special research program of the
20. O. Marín, J. L. Rubenstein, A long, remarkable journey: complex regionalization of the ancestral vertebrate brain. Austrian Science Fund project F78 (E.M.T.); European Research
Tangential migration in the telencephalon. Nat. Rev. Neurosci. Nature 531, 97–100 (2016). doi: 10.1038/nature16518; Council Organomics 758877 (A.M. and B.T.); European Research
2, 780–790 (2001). doi: 10.1038/35097509; pmid: 11715055 pmid: 26878236 Council Braintime 874606 (B.T.); and Swiss National Science
21. Z. Almási, C. Dávid, M. Witte, J. F. Staiger, Distribution patterns 35. I. Carrera, S. Ferreiro-Galve, C. Sueiro, R. Anadón, Foundation Project Grant 310030_192604 (B.T.). Author
of three molecularly defined classes of GABAergic neurons I. Rodríguez-Moldes, Tangentially migrating GABAergic cells of contributions: Designed study: K.L., A.M., T.G., E.M.T., and B.T.
across columnar compartments in mouse barrel cortex. Front. subpallial origin invade massively the pallium in developing Prepared and collected the samples and performed lineage tracing
Neuroanat. 13, 45 (2019). doi: 10.3389/fnana.2019.00045; sharks. Brain Res. Bull. 75, 405–409 (2008). doi: 10.1016/ and staining: K.L. Performed single-nuclei experiments: A.M.
pmid: 31114486 j.brainresbull.2007.10.013; pmid: 18331906 Analyzed the sequencing data: T.G., A.M., and J.S.F. Prepared the

Downloaded from https://www.science.org on September 24, 2024

22. M. Kirkham, L. S. Hameed, D. A. Berg, H. Wang, A. Simon, 36. I. Bachy, J. Berthon, S. Rétaux, Defining pallial and subpallial figures and wrote and approved the manuscript: K.L., A.M., T.G.,
Progenitor cell dynamics in the Newt Telencephalon during divisions in the developing Xenopus forebrain. Mech. Dev. 117, J.S.F., J.G.C., E.M.T., and B.T. Competing interests: The authors
homeostasis and neuronal regeneration. Stem Cell Reports 2, 163–172 (2002). doi: 10.1016/S0925-4773(02)00199-5; declare that they have no competing interests. Data and
507–519 (2014). doi: 10.1016/j.stemcr.2014.01.018; pmid: 12204256 materials availability: Code is available at Github (https://github.
pmid: 24749074 37. Z. Petanjek, B. Berger, M. Esclapez, Origins of cortical com/tomasgomes/pallium_evo) (44). Count tables are deposited
23. F. García-Moreno, Z. Molnár, Variations of telencephalic GABAergic neurons in the cynomolgus monkey. Cereb. Cortex at (45). Fastq files have been deposited in ArrayExpress under
development that paved the way for neocortical evolution. 19, 249–262 (2009). doi: 10.1093/cercor/bhn078; accession nos. E-MTAB-11638, E-MTAB-11662, E-MTAB-11665,
Prog. Neurobiol. 194, 101865 (2020). doi: 10.1016/ pmid: 18477686 and E-MTAB-11666. All other data are in the main paper or
j.pneurobio.2020.101865; pmid: 32526253 38. M. S. Brill et al., A dlx2- and pax6-dependent transcriptional supplementary materials. License information: Copyright © 2022
24. V. Naef et al., The stemness gene Mex3A is a key regulator of code for periglomerular neuron specification in the adult the authors, some rights reserved; exclusive licensee American
neuroblast proliferation during neurogenesis. Front. Cell Dev. olfactory bulb. J. Neurosci. 28, 6439–6452 (2008). Association for the Advancement of Science. No claim to original
Biol. 8, 549533 (2020). doi: 10.3389/fcell.2020.549533; doi: 10.1523/JNEUROSCI.0700-08.2008; pmid: 18562615 US government works. https://www.science.org/about/science-
pmid: 33072742 39. M. S. Brill et al., Adult generation of glutamatergic olfactory licenses-journal-article-reuse
25. A. Cebrian-Silla et al., Single-cell analysis of the ventricular- bulb interneurons. Nat. Neurosci. 12, 1524–1533 (2009).
subventricular zone reveals signatures of dorsal and ventral doi: 10.1038/nn.2416; pmid: 19881504
adult neurogenesis. eLife 10, e67436 (2021). doi: 10.7554/ 40. R. C. Bandler et al., Single-cell delineation of lineage and SUPPLEMENTARY MATERIALS
eLife.67436; pmid: 34259628 genetic identity in the mouse brain. Nature 601, 404–409
science.org/doi/10.1126/science.abp9262
26. D. J. Di Bella et al., Molecular logic of cellular diversification in (2022). doi: 10.1038/s41586-021-04237-0; pmid: 34912118
Materials and Methods
the mouse cerebral cortex. Nature 595, 554–559 (2021). 41. T.-Y. Lin et al., Fibroblast dedifferentiation as a determinant of
Figs. S1 to S17
doi: 10.1038/s41586-021-03670-5; pmid: 34163074 successful regeneration. Dev. Cell 56, 1541–1551.e6 (2021).
Tables S1 to S7
27. G. Furlan et al., Life-long neurogenic activity of individual doi: 10.1016/j.devcel.2021.04.016; pmid: 34004152
References (46–71)
neural stem cells and continuous growth establish an outside- 42. T. Gerber et al., Single-cell analysis uncovers convergence of
MDAR Reproducibility Checklist
in architecture in the teleost pallium. Curr. Biol. 27, cell identities during axolotl limb regeneration. Science 362,
3288–3301.e3 (2017). doi: 10.1016/j.cub.2017.09.052; eaaq0681 (2018). doi: 10.1126/science.aaq0681; View/request a protocol for this paper from Bio-protocol.
pmid: 29107546 pmid: 30262634
28. V. Bergen, M. Lange, S. Peidli, F. A. Wolf, F. J. Theis, 43. D. Wenemoser, S. W. Lapan, A. W. Wilkinson, G. W. Bell, Submitted 7 March 2022; accepted 29 June 2022
Generalizing RNA velocity to transient cell states through P. W. Reddien, A molecular wound response program 10.1126/science.abp9262

Lust et al., Science 377, eabp9262 (2022) 2 September 2022 12 of 12