REVIEW ARTICLE

Radial Glia and Neural Progenitors in theAdult Zebrafish Central Nervous System

Emmanuel Than-Trong and Laure Bally-Cuif

The adult central nervous system (CNS) of the zebrafish, owing to its enrichment in constitutive neurogenic niches, is becom-ing an increasingly used model to address fundamental questions pertaining to adult neural stem cell (NSC) biology, adultneurogenesis and neuronal repair. Studies conducted in several CNS territories (notably the telencephalon, retina, midbrain,cerebellum and spinal cord) highlighted the presence, in these niches, of progenitor cells displaying NSC-like characters.While pointing to radial glial cells (RG) as major long-lasting, constitutively active and/or activatable progenitors in mostdomains, these studies also revealed a high heterogeneity in the progenitor subtypes used at the top of neurogenic hierar-chies, including the persistence of neuroepithelial (NE) progenitors in some areas. Likewise, dissecting the molecular pathwaysunderlying RG maintenance and recruitment under physiological conditions and upon repair in the zebrafish model revealedshared processes but also specific cascades triggering or sustaining reparative NSC recruitment. Together, the zebrafish adultbrain reveals an extensive complexity of adult NSC niches, properties and control pathways, which extends existing under-standing of adult NSC biology and gives access to novel mechanisms of efficient NSC maintenance and recruitment in anadult vertebrate brain.

GLIA 2015;63:1406-1428.Key words: radial glia, neuroepithelial progenitors, zebrafish, adult neurogenesis

Introduction

Aquatic vertebrates (fish, amphibians) have long been rec-

ognized for their sustained adult neurogenesis. Among

them, the zebrafish is becoming increasingly popular to

address issues of adult neural stem cell (NSC) biology and

regeneration. In most subdivisions of the zebrafish adult cen-

tral nervous system (notably the telencephalon, retina and spi-

nal cord), radial glia (RG) were identified as a prominent

progenitor cell type, whether under physiological or regenera-

tive conditions. More recently, nonglial progenitors were also

found to constitute persistent niches contributing to adult

neurogenesis and, in some instance, to the generation of

long-lasting, self-renewing RG pools. This highlights the

complexity of adult neurogenesis and NSC processes in zebra-

fish, and the power of this model, unfairly qualified as

“simple”, to gain insight into the biology of NSCs/adult pro-

genitors. What is the exact hierarchical relationship of these

different progenitor types? How do they compare with char-

acterized mammalian NSCs? Do they truly exhibit characters

of adult NSCs, in spite of a morphology resembling that of

embryonic progenitors and the constant growth of zebrafish

individuals through life? What can we learn from zebrafish

adult neurogenesis that can be translated to other adult verte-

brates and in particular to an adult mammal?

To bring insight into these questions, we will first sum-

marize the characteristics and properties of mammalian RG,

the “standard” in the field. We will then review the status of

progenitor subtypes in the adult zebrafish, their physiological

function under normal and reparative conditions, and what is

known of their underlying molecular and cellular control

pathways.

The Standard: Radial Glia in MouseRadial glia (RG) arise in the developing mouse CNS around

the onset of neurogenesis and are classically defined by the

combination of several features: (i) the expression of glial

markers typical of astroglial cells, (ii) an elongated (radial)

morphology bridging the apical and basal surfaces of the

View this article online at wileyonlinelibrary.com. DOI: 10.1002/glia.22856

Published online May 14, 2015 in Wiley Online Library (wileyonlinelibrary.com). Received Feb 1, 2015, Accepted for publication Apr 22, 2015.

Address correspondence to L. Bally-Cuif, Paris-Saclay University, Paris-Sud University, CNRS, UMR 9197, Paris-Saclay Institute for Neuroscience (NeuroPSI), Team

Zebrafish Neurogenetics, Avenue de la Terrasse, bldg 5, Gif-sur-Yvette F-91190, France. E-mail: bally-cuif@inaf.cnrs-gif.fr

From the Team Zebrafisdh Neurogenetics, Paris-Saclay University, Paris-Sud University, CNRS, UMR 9197, Paris-Saclay Institute for Neuroscience (NeuroPSI), Avenue

De La Terrasse, Bldg 5, Gif-sur-Yvette, F-91190, France

ETT is recipient of a PhD fellowship from the Ministery of Research and Education attributed by University Paris 11.

1406 VC 2015 Wiley Periodicals, Inc.

neural tube, and (iii) the maintenance of an apico-basal polar-

ity. These features are a RG signature together but not indi-

vidually, as many astroglial markers are shared with astrocytes

and/or ependymal cells, while the morphological and polarity

criteria are also found in neuroepithelial (NE) cells, the first

progenitors of the neural plate and tube from which RG

emerge during development.

RG markers include expression of intermediate filaments

such as Vimentin and glial fibrillary acidic protein (GFAP),

expression of the glutamate transporters GLAST and GLT-1

and of glutamine synthase (GS), S100b, tenascin-C, meta-

bolic enzymes such as the aldolase Aldhlh1 and the brain

lipid binding protein (BLBP), and the oestrogen biosynthesis

enzyme aromatase B (Dimou and G€otz, 2014; G€otz, 2013).

They also express various neurotransmitter receptors and

voltage-dependent ion channels, and can release chemical

messengers (G€otz, 2013). These proteins attest of functional

astroglial properties deeply involved in brain homeostasis,

such as transmitter uptake and release, the buffering of extrac-

ellular K1 concentration, the provision of metabolic support

to neurons, the generation of calcium waves, and oestrogen

synthesis.

Apico-basal polarity is manifested by an apical mem-

brane compartment limited by tight and adherens junctions,

and organized in microdomains. The latter are enriched in

proteins of the Par complex (Par3, Par6 and APKc), cdc42,

cadherins and the glycoprotein Prominin-1. The apical

domain also harbours a primary cilium protruding into the

cerebrospinal fluid (Peyre and Morin, 2012). In contrast, the

basal process is elongated and connected to the basement

membrane of the pial surface of the neural tube. In addition,

RG share with NE cells their expression of the intermediate

filament Nestin and the transcription factor Sox2, directly or

indirectly implicated in the neural progenitor properties of

both NE and RG cells (Park et al., 2010). Finally, both cell

types undergo interkinetic nuclear migration (INM), a process

by which their nucleus translocates from apical to more basal

positions between the S and M phases of the cell cycle.

RG are mostly transient in mammals: they act as pro-

genitors, functional and structural components of the devel-

oping CNS, while they largely differentiate into astrocytes

and ependymal cells at postnatal stages. However, glial cells

with radial features can be found in the adult brain, some

being located in specialized niches where they act as constitu-

tive adult progenitor cells: the so-called “B” cells of the sub-

ependymal zone (SEZ) of the lateral ventricle, “type 1” cells

of the subgranular zone (SGZ) of the dentate gyrus of the

hippocampus, and, most likely, tanycytes bordering the hypo-

thalamic ventricle (Dimou and G€otz, 2014; Morrens et al.,

2012). Of these, “B” and “type 1” cells share all markers with

RG, except GS and S100b. They extend a parenchymal pro-

cess contacting the basement membrane surrounding blood

vessels, and, in addition, B cells contact the brain ventricle

through a specialized apical membrane domain comparable to

embryonic RG. Figure 1 illustrates the shared and divergent

properties of RG and related constituent cell types of the

developing and mature CNS.

Radial Glia Localization and PhysiologicalProperties in the Adult Zebrafish CentralNervous System

Progenitor cells with an overt radial morphology, apico-basal

polarity and displaying INM can be identified around mid-

embryogenesis in the developing zebrafish telencephalon

(Dong et al., 2012), retina (Baye and Link, 2007; Das et al.,

2003) and hindbrain (Alexandre et al., 2010; Coolen et al.,

2012; Lyons et al., 2003), and are likely broadly distributed

at this stage in the CNS. However, the expression of some

“RG markers” is initiated as early as the neural plate stage

(e.g., late epiboly for blbp (fabp7a), 1-somite stage for gfap(Thisse et al., 2001), making it perhaps more difficult than

in rodents to precisely define a transition from NE to RG.

Other markers are turned on during later embryogenesis and

FIGURE 1: Markers expression in the four related cell types“neuroepithelial cell”, “radial glia”, “mature astroglia” and“ependymal cell”. The markers currently identified and used inzebrafish are highlighted in yellow (markers listed in black areused in mammals, but their expression in the zebrafish adultbrain was not studied yet). BLBP is illustrated here as labelingboth RG and NE progenitors, reflecting the early expression ofblbp (fabp7) at neural plate stages in the embryo, and the defini-tion of radial progenitors of the ventral subpallium as NE cellsby some authors (Ganz et al., 2010). It is to note however thatthe NE pools maintained in the adult lateral pallium or optic tec-tum do not express BLBP (Dirian et al., 2014, and unpub.).

Than-Trong and Bally-Cuif: Adult Radial Glia in Zebrafish

August 2015 1407

juvenile stages, to reach a mature RG state very similar to

mammals (Fig. 1).

Contrasting with the situation in mammals, however,

RG are widely maintained in the zebrafish adult CNS. At

least in part, this is likely in functional relation with two spe-

cific features of the adult CNS in aquatic vertebrates: its

maintenance of widespread niches of constitutive neurogenesis

- where adult RG can, like in the embryo, serve as progenitor

cells (see below)-, and its relatively small size, which may alle-

viate the need to “delocalize” astroglial cells into the paren-

chyma to ensure proper astroglial function (Chapouton et al.,

2007). With the exception of the optic nerve (Koke et al.,

2010), the zebrafish brain is devoid of bona fide astrocytes

(Grupp et al., 2010), and only some ventricular domains

(e.g., the diencephalon or the spinal cord) contain

ependymal-like cells, characterized by the absence of expres-

sion of markers such as GFAP or Vimentin and the presence

of multiple beating cilia extending in the cerebrospinal fluid

(Kishimoto et al., 2011). Correlatively, in the adult zebrafish

brain, RG express the water channel Aquaporin 4 (a property

of astrocytes in mammals; Grupp et al., 2010) and, in the

subpallium, harbour single cilia with a 9 1 2 organization

suggesting that they can beat (Kishimoto et al., 2011). Little

is known of other functional properties of adult zebrafish RG

(neuronal migration, synaptic plasticity, glia-neuron commu-

nication, brain homeostasis. . .) except for an extensive series

of studies demonstrating their strong expression and activity

of steroidogenic enzymes, suggesting their production of neu-

rosteroids (Pellegrini et al., 2005). This property is shared

with mammalian astrocytes and neurons. In addition, RG in

the trout optic tectum exhibit voltage-gated Na1 inward cur-

rents and K1 outward currents (Rabe et al., 1999).

The following sections highlight the location and cellu-

lar characteristics of RG in the zebrafish adult brain.

RG in the Adult TelencephalonThe telencephalon of teleosts, like that of mammals, is com-

posed of a dorsally located pallium and a ventrally located

subpallium. Due to a peculiar morphogenesis and growth

process during development, dorsal structures are displaced

outward. This process, called “eversion”, (i) generates a pal-

lium with inverted medio-lateral organization compared with

other vertebrates and (ii) exposes the pallial ventricular zone

(VZ) dorsally, covered by an extended choroid plexus (Dirian

et al., 2014; Folgueira et al., 2012). Developmental, hodolog-

ical, functional and gene expression information support a

regional organization of the pallium in a lateral domain (Dl)

containing hippocampal-like structures, a medial domain

(Dm) containing amygdala-like nuclei (overlapping the

pallial-subpallial junction) and a dorsal/central domain (Dd/

Dc) corresponding to the neocortical area in mammals

(Mueller and Wullimann, 2009). Homologies in the subpal-

lium remain more uncertain (Ganz et al., 2012).

Pallium. Cells with RG characteristics are aligned along the

pallial ventricle, covering the large majority of the VZ (Fig.

2A). RG cell bodies organize into a monolayer in contact

with the CSF, while their long processes extend into the pal-

lial parenchyma to reach the pial surface of the brain in con-

tact with large blood vessels. Various RG morphologies have

been described in different pallial areas; they differ in cell

body shape and the length and degree of branching of the

radial process (M€arz et al., 2010a). Expression of markers,

e.g., Nestin, GFAP, BLBP, Cyp19b (Aromatase B), S100b,

Vimentin, GS and Fezf2 also shows some degree of heteroge-

neity (M€arz et al., 2010a; Berberoglu et al., 2014). These het-

erogeneities remain to be fully characterized and their

functional significance determined.

Subpallium. The VZ of the subpallium is bordered with a

pseudo-stratified monolayer of radial, actively proliferating

cells with ventricular (apical) and pial (basal) contacts. These

cells display elongated nuclei, apicobasal polarity, express glial

markers (e.g., GFAP, Vimentin, Cyp19b, BLBP, S100b) at

very weak levels, and generally divide close to the ventricular

surface while completing their S phase at a distance, suggest-

ing that they may undergo INM (M€arz et al., 2010a). A sub-

population of these cells expresses the transcription factor

gene olig2 (M€arz et al., 2010b), indicating some degree of

heterogeneity within this population.

RG in the Adult RetinaM€uller glia constitute the main glial population of the verte-

brate retina and display a radial shape, with cell bodies

located in the inner nuclear layer and extending processes

throughout the retina’s radial depth, from the apical epithelial

surface to the basal lamina. They express classical glial

markers such as GFAP and GS, and also stain positive for

carbonic anhydrase (Linser et al., 1985; Peterson et al.,

2001). Following the constant growth of the retina driven

from the ciliary marginal zone (CMZ) in teleosts, the least

mature M€uller glia are located in proximity to the CMZ;

these cells also express BLBP (Raymond et al., 2006). It is to

note that nonradial astrocytes from the optic nerve

partly extend into the retina, along the inner limiting mem-

brane; these are GFAP-negative but express Cytokeratin

(Koke et al., 2010).

RG in the Adult DiencephalonThe diencephalon of the zebrafish includes the preoptic area,

epithalamus, dorsal and ventral thalamus, posterior tubercu-

lum, hypothalamus (consisting of dorsal, ventral, and caudal

zones), synencephalon, and pretectum. RG cells expressing

1408 Volume 63, No. 8

GFAP, BLBP or AroB have been observed lining the ventricle

of most subdivisions, although their density varies (Menuet

et al., 2005). In the periventricular region of the hypothalamus

(periventricular organ, PVO), a region particularly enriched in

RG cells, the nuclei of BLBP- or AroB-positive RGs are located

in a subventricular location and RG extend a short cytoplasmic

FIGURE 2

Than-Trong and Bally-Cuif: Adult Radial Glia in Zebrafish

August 2015 1409

process toward the ventricle (Fig. 2B). In this territory, cerebro-

spinal fluid-contacting neurons intervene between the RG cell

bodies and the ventricular surface (Pellegrini et al., 2007).

RG in the Adult MesencephalonThe mesencephalon comprises the alar tectum opticum (TeO),

bordered medially by the torus longitudinalis and laterally by

the torus semicircularis, and the basal tegmentum. A promi-

nent layer of RG cells has been described in the TeO, with

their cell bodies aligned along the ventricle of the deepest tectal

layer (periventricular gray zone) and extending a basal process

reaching into upper TeO layers (Fig. 2C). The end feet of

these RG cells form a glial limitans in the more superficial

marginal zone, and also contact blood vessels (Corbo et al.,

2012). This RG arrangement however excludes the marginal

proliferation zone of the TeO, where no RG cells are present.

Tectal RG cells express markers such as GFAP, BLBP, S100b,

and AroB (Menuet et al., 2005; Ito et al., 2010). RG markers-

expressing cells are also observed along the ventricle of the teg-

mentum and torus semicircularis, although their processes are

shorter and/or not coursing radially (Menuet et al., 2005).

RG in the Adult Rhombencephalon and Spinal Cord

Cerebellum. Cell types in the adult zebrafish cerebellum

have been extensively characterized, in link with the promi-

nent neurogenesis occurring in this territory. GFAP, BLBP,

S100b and Vimentin-positive RG cells can be found at the

cerebellar midline in the molecular layer, close to the main

neurogenic site generating granule neurons (Fig. 2D; Kaslin

et al., 2009). Their processes extend into the molecular layer

and are postulated to guide neuronal migration. An addi-

tional set of RG cells, so called Bergmann glia, can be

observed sparsely distributed in the inner molecular layer

throughout the two anterior cerebellar lobes (Fig. 2E). Com-

pared with RG cells in other brain subdivisions, they harbour

a process with multiple main branches.

Medulla Oblongata and Spinal Cord. GFAP-positive radial

cells are present bordering the medulla and spinal cord ventri-

cle in the adult zebrafish (Fig. 2F). Cells processes reach the

spinal cord surface, and some have also been observed con-

tacting small vessels (Kawai et al., 2001; Tomizawa et al.,

2000). Some radial cells positive for BLBP expression also

line the spinal ventricle. Heterogeneity within the spinal

radial population is notably suggested by the nonubiquitous

expression of molecular markers such as olig2, expressed in a

discrete bilateral population of BLBP-positive but GFAP-

negative radial cells in the basal plate (Park et al., 2007). In

the juvenile, this population is distinct from another radial

subpopulation expressing dbx1a (Briona and Dorsky, 2014).

A population of radial cells of similar morphology and

expressing Shh is also found lining the most ventral aspect of

the central canal (Reimer et al., 2009).

Zebrafish Adult Radial Glia as NeuralProgenitors

The discovery that RG cells are neural progenitors during

development in the mouse (Anthony et al., 2004; Malatesta

et al., 2000, 2003; Noctor et al., 2002), and that neural stem

cell potential is maintained by glial cells in the adult rodent

brain, strongly suggested that adult RG may be, at least in

part, at the origin of the extensive neurogenesis observed in

adult teleost fish. Recent work exploring this possibility in

detail in the adult zebrafish confirmed this hypothesis but also

revealed a more complex situation, where distinct RG pools

may be recruited in constitutive versus reactive neurogenesis.

Active neurogenesis persists in most brain subdivisions

of the zebrafish adult central nervous system, albeit at differ-

ent rates, being notably low in the spinal cord. Overall, 16

constitutively neurogenic domains have been described

(Grandel et al., 2006). RG or radial cells were directly impli-

cated in constitutive neurogenesis in some of these domains:

the pallium, the subpallium, and the hypothalamus. The pal-

lial population can be further boosted in reaction to lesion.

In addition, reactive neurogenesis recruits normally silent RG

populations such as M€uller glia in the retina, or ventricular

ependymoglial cells in the spinal cord. The neuronal progeni-

tor status of other RG or radial populations remains to be

assessed, notably in the optic tectum.

Although this process was much less studied to date,

glial cells are also generated lifelong in the adult zebrafish

brain. The generation of new RG from RG progenitors was

FIGURE 2: RG in the zebrafish adult central nervous system, revealed by gfap:egfp expression. Cross sections of the telencephalon (A),hypothalamus (B), optic tectum (C), cerebellum (D and E), and spinal cord (F) from gfap:egfp adult transgenic zebrafish (Bernardos and Ray-mond, 2006), immunostained for GFP (photographed under confocal microscopy). Arrows point to RG cell bodies (Bergman glia in E). Theidentification of GFAP-positive cells as RG originate from the following publications: Ito et al., 2010; Kaslin et al., 2009; Marz et al., 2010a;Menuet et al., 2005; Park et al., 2007. Abbreviations are from Wullimann et al., 1996: C: central canal, CCe: corpus cerebellaris, CCer: com-missura cerebelli, CTec: commissura tecti, D: dorsal telencephalon (pallium), Dc: central subdivision of D, Dd: dorsal subdivision of D, Dm:medial subdivision of D, Dl: lateral subdivision of D, DH: dorsal horn, DiV: diencephalic ventricle, EG: eminentia granularis, Hc: central zoneof periventricular hypothalamus, Hd: dorsal zone of periventricular hypothalamus, LR: lateral recess of the hypothalamic ventricle, MA:Mauthner axon, MLF: medial longitudinal fascicle, PGZ: periventricular gray zone, Pit: pituitary, PPa: posterior preoptic area, RV: rhomben-cephalic ventricle, SRF: superior reticular formation, SY: sulcus ypsiloniformis, TeO: tectum opticum, TeV: tectal ventricle, TL: torus longitu-dinalis, TS: torus semicircularis, VH: ventral horn, VOT: ventrolateral optic tract, Vs: ventral subdivision of the subpallium.

1410 Volume 63, No. 8

formally documented in the pallium, and can be indirectly

inferred from the growth of ventricular zones coupled with

RG proliferation in other areas. The generation of oligoden-

drocytes is still prominent at adult stages, from source

cells that seem to vary in nature between CNS territories

(Chapouton et al., 2006).

These findings are detailed below.

RG or Radial Progenitors in ConstitutiveNeurogenesis

Pallial RG are Life-Long Constitutive Neural Progenitors

and a Main Amplification Step of the Neurogenic Line-

age.

Neuronal generation. BrdU labeling, and expression of

positive proliferation markers such as Proliferating Cell

Nuclear Antigen (PCNA) or Mini-Chromosome Maintenance

proteins (e.g., MCM5) indicated that dividing cell types

along the pallial ventricle include RG (Adolf et al., 2006a).

Following BrdU administration, neurons (expressing the pro-

teins HuC/D) start appearing in the pallium parenchyma

after a few days of chase, and their number increases with

time until ca. 2–3 weeks (Adolf et al., 2006a). Lentivirus-

mediated cell transduction likewise indicated that pallial RG

generate neurons that reach a mature neuronal phenotype

(with expression of synaptic proteins and the firing of axon

potential) 4–8 weeks after birth (Rothenaigner et al., 2011).

Cre-lox genetic tracing provided the direct demonstration of

a RG origin for pallial neurons (Fig. 3A). Driver transgenic

lines expressing an inducible form of the Cre recombinase

under control of regulatory elements of the e(spl) her4 gene,

selectively expressed in pallial RG, were crossed into reporter

lines expressing a fluorescent protein upon recombination.

FIGURE 3: Stem-like progenitor cells in the adult zebrafish cen-tral nervous system and their fate under normal and regenera-tive conditions. RG (green) and NE (blue) progenitors areschematized on cross sections of the pallium (A), optic tectum(B), retina (C), cerebellum (D), and spinal cord (E). In the pallium,optic tectum and retina, NE stem-like progenitors are associatedwith a highly proliferating growth zone (pink). The activation sta-tus (“2“, “1”) and fate (arrows) of these progenitors under phys-iological conditions (“norm.”) and under conditions of neuronalregeneration (“reg.”) are also indicated on each section and inthe tables (yellow, orange, pink, red: neurons of different identi-ties, including rods (yellow ovals); gray: oligodendrocytes). (A)RG in the pallium generate RG and neurons. They likely canincrease their repertoire of generated neuronal subtypes underregenerative conditions. NE progenitors in the lateral palliumgenerate RG NSCs. (B) RG in the optic tectum are endogenouslymostly silent, but their fate, notably in freshly generated RGclose to the growth zone, has not been directly assessed. NEprogenitors at the posterior edge of the PGZ generate RG andneurons, most likely via a direct mode that does not involveRGs. Regenerative contexts have not been studied yet in theadult TeO. (C) M€uller glia in the central retina are engaged in aslow rate generation of rods (yellow ovals), but can generate allneuronal subtypes upon repair activation. NE progenitors of theCMZ generate M€uller glia and all other neuronal types; theirrecruitment and fate upon repair has not been analyzed in detail.(D) NE progenitors generate granule neurons, while mediallylocated RG, and Bergman glia, are silent. The fate of these cellsupon repair has not been analyzed. (E) RG lining the centralcanal are mostly silent and engaged in a slow rate generation ofoligodendrocytes. Following spinal transection, RG activate andcan generate motoneurons.

Than-Trong and Bally-Cuif: Adult Radial Glia in Zebrafish

August 2015 1411

Activation of Cre at late juvenile stage (1.5 month) or at

adult stage resulted in labeled neurons in the adult pallium

(Kroehne et al., 2011).

We still know very little of the neuronal subtypes gener-

ated by RG in the pallium during juvenile and adult life.

Parvalbumin-positive and calbindin-positive interneurons are

present in high numbers, respectively concentrated in the cen-

tral/lateral and medial pallium (Mueller et al., 2011; von Tro-

tha et al., 2014). In addition, some GABAergic and

glutamatergic neurons are scattered throughout the pallium

(Mueller et al., 2011; von Trotha et al., 2014). The entire lat-

eral pallium, including all its constituent neurons, is gener-

ated from juvenile stages onward; in contrast, neurons of the

dorso-medial pallium are generated from both embryonic and

postembryonic stages (Dirian et al., 2014). As a first step, a

recent study demonstrated that neurons born in the ventral-

most area of the medial pallium at around 1mpf were

recruited in goal-directed behavior in adults (von Trotha

et al., 2014). It remains a key point to analyze how the gener-

ation of distinct neuronal subtypes is orchestrated by juvenile

or adult RG—at the single RG or the RG population level—

and how they integrate into the distinct pallial areas and con-

tribute to their function. Adult neurogenesis in the mammalian

hippocampus is associated with a high rate of neuronal replace-

ment (Spalding et al., 2013). It is possible that distinct integra-

tion processes are at stake in the zebrafish adult pallium, where

very little cell death has been observed and adult neurogenesis

is associated with neuronal addition and pallial growth.

Division modes and rate. During a cell cycle, PCNA and

MCM5 concentrations increase during late G1 phase, peak in

S, and decrease thereafter. The average proportion of PCNA-

or MCM5-positive RG is of 8–10% in the medial and dorsal

pallium of a 3–6-month-old (3 cm-long) adult zebrafish

(Adolf et al., 2006a; M€arz et al., 2010a). Measurements of

cell cycle length based on the doubling time of activated RG

indicate that the G1-S-G2-M cycle is accomplished in ca. 2

days (Alunni et al., 2013). Following cumulative labeling and

several weeks of chase, the BrdU label is retained in 1–2% of

RG, some of which also stain positive for PCNA or MCM5

(Ganz et al., 2010; M€arz et al., 2010a). Clonal analyses 4

weeks after lentiviral transfection also indicate that most

(75%) transduced RG have remained as single cells, while the

remainder generated small clones of 2–4 RG cells (Rothe-

naigner et al., 2011). Together, these data attest of the overall

slow and to some extent heterogeneous proliferation rate of

adult pallial RG. Under physiological conditions, it remains

unknown whether this heterogeneity reflects the existence of

RG subtypes, a hierarchy in the RG recruitment cascade (as

postulated for mammalian neural and other stem cells), or

simply stochastic variations within a population of otherwise

equivalent cells. Abrogation of Notch signaling, a major path-

way gating RG proliferation in the adult zebrafish pallium,

can bring all RG into cycle, indicating that all have the

capacity to cycle (Alunni et al., 2013; Chapouton et al.,

2010). Finally, pulses of different thymidine analogues sepa-

rated by a few days demonstrate that a second cycle of the

same RG virtually never happens during this time frame but

rather that different RG activate over time (unpub.). These

data together suggest that the RG population is largely com-

posed of quiescent RG that divide asynchronously. Analysis of

the fucci zebrafish lines, as well as centrosome labeling, fur-

ther indicates that most nondividing RG are held in the G0

phase of the cell cycle (unpub).

Progenitor cells of the pallial germinal zone comprise,

in addition to RG, nonglial proliferating cells, interpreted as

intermediate neuronal progenitors (M€arz et al., 2010a), and

the current model postulates that (at least) some of these pro-

genitors are generated from RG. Short-term tracing of pallial

RG divisions following lentiviral transduction or BrdU incor-

poration indicated a predominance of symmetric gliogenic

divisions (ie. divisions generating two RG) over asymmetric

divisions (ie. divisions generating one RG and one intermedi-

ate progenitor; Fig. 4A; Rothenaigner et al., 2011). Whether

all individual RG are capable of both division modes, and

how the switch between these modes is controlled, remains

unknown. The high frequency of symmetric divisions con-

trasts with the behavior of adult mouse NSCs (Costa et al.,

2011; Suh et al., 2007). In an opposite manner, mouse inter-

mediate progenitors, at least in the SEZ, are considered

actively cycling to amplify neuronal production (Doetsch

et al., 1999), while nonglial progenitors in the adult zebrafish

pallium follow a low number of divisions before differentia-

tion (Rothenaigner et al., 2011). Thus, the amplification steps

that underlie pallial neurogenesis in mammals and zebrafish

likely differ, and it was proposed that the amplification

potential of zebrafish RG accounts for the persistence of a

large population of active and recruitable NSCs in adults.

Radial Progenitors of the Adult Subpallium Are Constitutive

Neuronal Progenitors.

Neuronal generation. BrdU administration to zebrafish

adults results in intense labeling of subpallial progenitors.

These can be chased to reveal the neogeneration of neurons

in two areas, the olfactory bulb and subpallium proper, over

a few weeks (Adolf et al., 2006a; Grandel et al., 2006). Many

of the labeled progenitors express PSA-NCAM, suggesting

that they are migrating (M€arz et al., 2010a). The horizontal

migration of neuronal progenitors along the subpallial ven-

tricular zone into the olfactory bulb could also be followed

over a few micrometers in time-lapse imaging on explants

(Kishimoto et al., 2011). This and the contribution to OB

1412 Volume 63, No. 8

neurogenesis led to the conclusion that progenitors found in

the subpallium include, likely in addition to the radial pro-

genitors (see above), a population in part equivalent to the

“rostral migratory stream” (RMS) of mammals (Adolf et al.,

2006a). The “zebrafish RMS” however differs from the mam-

malian RMS by the absence of glial migration guides. The

lineage relationship between subpallial radial progenitors, the

migrating PSA-NCAM-positive progenitors, and OB or sub-

pallial parenchymal neurons remains unclear and an impor-

tant point to assess. Supporting the hypothesis that radial

progenitors and PSA-NCAM-positive migrating progenitors

are intermingled progenitors of at least partially distinct line-

ages, it is to note that no radial progenitors have been

described intermingled with the population of PSA-NCAM-

positive progenitors in posterior subpallial locations.

In the olfactory bulb, adult-born neurons include

GABAergic interneurons in the internal layer and TH-

positive and GABAergic interneurons in the glomerular layer,

comparable to the situation in mammals (Adolf et al.,

2006b). Unlike in mammals however, adult-born neurons in

the zebrafish subpallium also settle radially into the subpallial

parenchyma proper. These include GABAergic and TH-

positive neurons, and likely other neuronal subtypes, yet to

be characterized (Adolf et al., 2006b). In addition, goal-

directed behavior (von Trotha et al., 2014) and light avoid-

ance behavior (Lau et al., 2011) involve subpallial neurons,

but the birthdate of these neurons has not been determined.

Division modes and rate. BrdU incorporation and the pro-

portion of ventricular progenitors expressing PCNA are

much higher in the subpallium than in the pallium, indicat-

ing overall a faster cell cycle. Detailed analyses of progenitors

properties in this region are however complicated by the

intermingling of radial and migrating progenitors, which

cannot be distinguished in BrdU tracing experiments. Over-

all, bona fide NSC properties have not been demonstrated

in the subpallium, and the division mode(s) of progenitors

remain to be characterized. The high proliferation rate of

these progenitors makes it possible that they correspond to

“transit amplifying”-like compartment. It also remains

unclear whether slow cycling progenitors exist in (or feed

into) this domain, where they are located, and what is their

nature. Label retention assays indicate the presence of some

slower-cycling progenitors in this location (Adolf et al.,

2006a; Lam et al., 2009; M€arz et al., 2010a), but whether

this is a transient property of all resident progenitors or

reflects a specific subpopulation is not known. Subpallial

progenitors, for their very low expression of glial markers,

have been considered by some authors as NE progenitors

(Ganz et al., 2010). We find the distinction difficult to

make at this point.

Radial Progenitors and Constitutive Neurogenesis in the

Hypothalamus. BrdU incorporation and neuronal genera-

tion was observed in several diencephalic areas, including the

preoptic area, the thalamic ventricle, and the hypothalamus.

Adult-born TH-positive neurons were documented in all

these domains, and serotonin (5HT)-positive neurons in the

PVO in the hypothalamus (Grandel et al., 2006). However,

the exact origin of these different neurons remains uncertain,

as BrdU incorporating cells include RG but also nonradial

progenitors in their vicinity (Pellegrini et al., 2007). One

exception may be the PVO, where the large majority of pro-

liferating cells is comprised within the RG population, lead-

ing to the conclusion that these RG are at least a source of

the PVO 5HT neurons (Perez et al., 2013).

FIGURE 4: Division modes and progeny clones of RG and NEprogenitors in the pallium. (A) Short term tracing of RG (trian-gles) in the adult pallium indicates that they are capable of bothsymmetric gliogenic (left) and asymmetric (right) divisions. Thelatter generate an intermediate, nonglial progenitor (brown oval)that undergoes limited divisions to generate neurons (pink,orange). (B) The division modes(s) of adult NE progenitors (blue)of the lateral pallium are not known. As a population, this poolgenerates neurogenic RG. (C) Clones generated from singleembryonic pallial progenitors traced using the brainbow strategyfrom 24-h postfertilization until adulthood, schematized on a tel-encephalic cross section (RG—triangles—and their generatedneuronal progeny—neuron shape, same color; dashed line to thepallial-subpallial boundary). The symmetric and asymmetric divi-sions of RG give rise to clones containing several RG (adjacentand aligned triangles along the ventricular surface) and severalneurons (reaching to the deepest parenchymal areas in the pal-lium). Whether each lateral pallial clones retain a NE progenitorat the origin of their “mother” lateral RG is not known (indicatedby “?”). After Dirian et al. (2014).

Than-Trong and Bally-Cuif: Adult Radial Glia in Zebrafish

August 2015 1413

M€uller Glia and Constitutive Neurogenesis in the Reti-

na. Constitutive neurogenesis in the teleost retina originates

from two sources, the ciliary marginal zone (CMZ, also called

circumferential germinal zone, which persists until adulthood)

and M€uller glia (Fig. 3C). While the CMZ generates all retinal

cell subtypes, M€uller glia are specialized in the generation of a

specific lineage, rod photoreceptors. Proliferating at low fre-

quency, M€uller glia generate a small population of amplifying

“rod precursors” that migrate into the ONL and are the pri-

mary source of new rods in the central retina (Bernardos et al.,

2007). In goldfish, it has been proposed that this constitutive

low-rate generation of rods responds to the lowering in rods

density that results from the retinal stretching accompanying

eye growth (Raymond et al., 1988). It may also result from a

low rate of rod death (Morris et al., 2008).

RG or Radial Progenitors in ConstitutiveGliogenesis

Amplification of RG Populations During Adult Life. Brain

growth is observed lifelong in zebrafish, accompanying body

growth, and is paralleled with increased brain ventricular

surfaces (unpub.). Ventricular zones generally display a con-

tinuous arrangement of RG, suggesting that most of these

cells are generated during adulthood. This was directly dem-

onstrated by tracing symmetric gliogenic divisions of RG in

the adult pallium (Rothenaigner et al., 2011). The progenitor

cells at the origin of adult-born RG in other brain territories

remains uncertain.

RG at the Origin of Oligodendrocytes in the Spinal

Cord. The proliferative capacity of parenchymal

Olig2;Sox10-double positive cells in the telencephalon sug-

gests that these cells, rather than ventricular RG, are at the

origin of adult-born oligodendrocytes in this territory. The

availability of Cre-expressing lines driven by RG-specific pro-

moters (Boniface et al., 2009) will however make it possible

to directly assess the contribution of RG to adult oligoden-

drogenesis in many brain areas.

In the spinal cord, RG cells expressing Olig2 can be found

to divide, incorporating BrdU or expressing PCNA. Tracing

using GFP stability of the olig2:gfp transgene further suggests

that, under physiological conditions, these progenitors most likely

do not generate neurons. The increased number of oligodendro-

cytes over time in the adult spinal cord, and their expression of a

common set of markers with ventricular RG progenitors (Olig2,

Sox10), suggest that the latter cells may be involved in oligoden-

drocyte generation (Fig. 3E; Park et al., 2007).

RG or Radial Progenitors in Reactive NeurogenesisReactive neurogenesis has been induced in the zebrafish adult

central nervous system by various means: stab wounds or

toxic compounds in the telencephalon, retina and cerebellum,

phototoxic lesions in the retina, or trans-sections in the spinal

cord. In striking contrast with the situation in mammals,

these lesions in zebrafish are followed by an efficient regenera-

tive process generally leading to full functional recovery

within a few weeks. In all cases studied, RG are involved in

the regenerative response, as attested by a boost in their pro-

liferation. Hence, such paradigms are instrumental both in

enhancing the activation potential of constitutively active RG

(e.g., in the pallium) and in revealing the progenitor potential

of RG in domains where they are normally largely silent,

such as the retina and spinal cord. In many cases however, it

remains to be precisely demonstrated where activated RG sit

in the cellular hierarchy(ies) leading to neuronal repair, and

whether they are also involved in replenishing glial pools.

Finally, it will be important to determine to which extent the

molecular cascades that trigger RG activation and the division

mode they adopt in response to lesion recapitulate an endoge-

nous activation process.

Radial Glia and Reactive Neurogenesis in the Pallium:

Boosting Existing Neurogenesis. Stab wounds of various

sizes have been inflicted to the telencephalon using sticks that

were inserted into the brain either through the nostrils, perfo-

rating the subpallium and pallium longitudinally (Ayari et al.,

2010; Baumgart et al., 2012; Kroehne et al., 2011), or radi-

ally or obliquely through the dorsal pallial surface, lesioning

concomitantly the choroid plexus and germinal zone itself

(Diotel et al., 2013; Kishimoto et al., 2012; M€arz et al.,

2011). RG activation (ie. cell cycle re-entry) is a shared

response of all these paradigms (Kizil et al., 2012a; Schmidt

et al., 2013) between approximately 2–3 days postlesion (dpl)

and 7 dpl. In one important study, RG fate was genetically

traced using her4:CreERT-mediated recombination, the

expression of which is specific of quiescent RG prior to

lesion. At 21 dpl, genetically labeled Hu-positive neurons

were found at the site of lesion, unambiguously demonstrat-

ing that initially quiescent RG were recruited to generate neu-

rons in the repair process (Fig. 3A; Kroehne et al., 2011).

Different RG markers nonetheless highlight different propor-

tions of RG re-entering division upon lesion (e.g., an impor-

tant reactivation of her4-positive cells, some reactivation of

S100b2 or BLBP-positive cells, no reactivation of AroB-

positive cells; Baumgart et al., 2012; Diotel et al., 2013;

M€arz et al., 2011; Kroehne et al., 2011), suggesting the exis-

tence of subpopulations of RG with different functions and/

or sensitivity to the repair process. Further, some RG remain-

ing silent were shown to upregulate a quiescence-promoting

factor, Id1, suggesting an active refractory response to reacti-

vation that may be used to preserve in part the NSC pool

(Schmidt et al., 2014). In addition to RG, proliferation is

1414 Volume 63, No. 8

boosted in pallial nonglial “intermediate” (“transit

amplifying”-like) progenitors upon lesion (Baumgart et al.,

2012; M€arz et al., 2011), suggesting that progenitor recruit-

ment is possible directly at a level downstream of NSCs.

Whether this differential recruitment is a general phenom-

enon or is dictated by the type or extent of the lesion, for

example, remains to be assessed. Overall, which exact cellular

hierarchies are involved in the neuronal regeneration process,

and possibly also in the GZ replenishment process following

its recruitment for repair, is unknown.

RG properties other than activation rate may be modi-

fied upon parenchymal lesion. Interestingly, in all paradigms

studied, the precursors of regenerated neurons migrate to the

lesioned area, which is located far deeper into the parenchyma

than the area normally populated by adult-born neurons. In

addition, the germinal zone areas activated during the repair

process can be from a different neuroanatomical domain (this

is obviously more evident in the case of small wounds, for

example, where RG in the medial and dorsal pallium—Dm

and Dd domains—are reactivated upon a lesion in the central

pallium—Dc; Baumgart et al., 2012; Kishimoto et al., 2012).

Although this has not been formally demonstrated, these

observations suggest that the neuronal subtypes generated

from stab wound-activated RG may be different from those

they would generate under constitutive neurogenesis. A

recently developed telencephalic lesion model using quinolinic

acid further supports this conclusion: lesion-activated RG,

traced with her4:CreERT, could regenerate long-distance pro-

jection neurons located deep into the parenchyma at the site

of lesion and extending axons contra-laterally through the

anterior commissure (Skaggs et al., 2014). Thus, lesion-

activated RG participated in re-establishing a circuitry nor-

mally not generated at this stage by RG NSCs during consti-

tutive neurogenesis.

Radial Glia and Reactive Neurogenesis in the Retina: Acti-

vating M€uller Glia. The best-studied repair process to date

involving RG is certainly retinal regeneration. Over the years,

several lesion paradigms have been developed (chemical, ther-

mal, light, mechanical, cytotoxic, or genetic), which differ in

the extent of the lesion generated and the cell types affected.

Specifically, it was shown that most injury modes stimulate a

proliferative response from M€uller glial cells, and are followed

by the regeneration of the lesioned neuronal subtypes (Fig.

3C; reviewed in Goldman, 2014; Gorsuch and Hyde, 2014;

Lenkowski and Raymond, 2014). M€uller glia (followed by

the gfap:gfp or alpha1tubulin:gfp transgenes or GS expression)

were suspected to be at the origin of neuronal regeneration in

a number of studies, since they generally constitute the main

proliferating cell type of the regenerating retina (after a tran-

sient microglial reaction; Fausett and Goldman, 2006), and

BrdU tracing indicates that proliferating progenitors then

migrate and give rise to bipolar and amacrine cells in the

INL and to cone photoreceptors in the ONL (Fausett and

Goldman, 2006). Genetic tracing, conducted using a alpha1-

tubulin:CreERT2 transgene following a small mechanical

lesion, confirms that reactivated M€uller glial cells are recruited

to give rise to cone photoreceptors, bipolar and amacrine

cells, and M€uller glia, under these conditions (Ramachandran

et al., 2010b). Thus, reactive M€uller glia can bias their nor-

mal fate (rods) in response to lesion to regenerate the missing

neuronal subtypes. Not all M€uller glia however reinitiate divi-

sion, highlighting heterogeneity within this population, and

single M€uller glial cells (as opposed to the population, e.g.,

using sparse recombination) were not traced to assess

multipotency.

Proliferation of M€uller glia is (directly or indirectly)

necessary for the regenerative response, since blocking prolif-

eration with PCNA morpholinos results in M€uller glia death

and a failure to generate new neurons (Thummel et al.,

2008). Like in embryonic NE cells and RG, INM accompa-

nies the reactive proliferation of M€uller glia (Nagashima,

2004). Although the exact regeneration cascade from M€uller

glia has not been traced in all cases, their proliferation

response to regenerate rods following light-induced damage,

and INL neurons following ouabain treatment, involves asym-

metrical self-renewing divisions that, like constitutive rod

neurogenesis, generate intermediate amplifying progenitors in

the ONL (Nagashima, 2004). These observations argue that

reactive M€uller glia behave as bona fide, self-renewing NSCs.

Interestingly, within this sequence of events, the extent of the

lesion within the rod population, and the specificity of the

lesion, may determine the step at which progenitors are

recruited: large rod lesions and lesions affecting other retinal

neuronal types recruit M€uller glia cells, while small rod

lesions primarily recruit ONL-located precursors (Montgom-

ery et al., 2010). How this specificity is controlled, as well as

which exact cellular hierarchies are involved in reparative neu-

rogenesis, remain important questions that will benefit from

generating targeted genetic lesions and tracing.

Radial Glia and Reactive Neurogenesis in the Optic Tectu-

m. In terms of progenitor cells and growth, the organization

of the teleost TeO is very similar to that of the retina, with

the maintenance of a marginal nonglial proliferation zone

responsible for life-long neuronal addition (see below), while

tectal RG appear nonproliferating (Fig. 3B; Alunni et al.,

2010; Ito et al., 2010). Thus, it is tempting to speculate that

tectal RG could, like M€uller glial cells, serve as an additional

progenitor source activated for repair upon lesion (and per-

haps for low-rate neuronal production under physiological

conditions). To date, no studies directly address this question.

Than-Trong and Bally-Cuif: Adult Radial Glia in Zebrafish

August 2015 1415

Radial Glia and Reactive Neurogenesis in the Spinal

Cord. Lesions in the spinal cord are generally inflicted

mechanically, by transection. Proliferation, starting 3 days

after lesion and peaking at 2 weeks, is induced around the

central canal in ependymoglial cells (Reimer et al., 2008).

Using GFP stability as a tracer in olig2:gfp transgenics

together with BrdU labeling, the generation of motorneurons

(HB9- or Islet1/2-positive cells) from activated Olig2-positive

progenitors could be documented (Fig. 3E; Reimer et al.,

2008). Thus, the normal fate of these progenitors, ie. oligo-

dendrocytes generation, is modified toward producing motor-

neurons in response to lesion. More dorsally located

ependymoglial cells, expressing the transcription factors Pax6

and Nkx6.1 and activated as well upon lesion, may be at the

origin of regenerated V2 interneurons (Kuscha et al., 2012b).

The fate of other activated ependymoglial cells, e.g., those

expressing shh ventrally or those located in the alar plate, has

not been traced to date (Reimer et al., 2009). Other neuronal

subtypes, including Pax2-positive interneurons and spinal

serotonergic neurons, are regenerated as well (Kuscha et al.,

2012a, 2012b). To date, the regenerating spinal cord is the

only case in the zebrafish central nervous system where some

hints on the factors modulating the neuronal identities pro-

duced have been proposed: the neurotransmitter dopamine,

originating from descending projection neurons, biases toward

motorneuron production at the expense of V2 interneurons

(Reimer et al., 2013).

Single cell fate has not been determined within the spi-

nal ventricular ependymoglial population upon lesion. The

increased number of such cells in reaction to lesion indicates

the occurrence of symmetric gliogenic divisions (Reimer

et al., 2009). Which division type is involved in the genera-

tion of neurons, and whether division mode is controlled at

the single cell or population level, is not known.

Activating Adult Radial Glia

Work conducted in mouse in a number of organs, including

the brain, suggests that the frequency of SC activation events

conditions, in the long term, SC maintenance (for example,

see Calzolari et al., 2015; de Haan et al., 1997; Encinas

et al., 2011; Vaziri et al., 1994). This may be linked with

some senescence or lack of self-renewal taking place over

time. This makes it crucial to understand which molecular

and cellular mechanisms gate SC activation from quiescence.

In most areas discussed above (pallium, hypothalamus,

retina, spinal cord), RG progenitors involved in constitutive

neurogenesis are characterized by a strong quiescence.

Whether different RG subtypes exist with respect to this

parameter remains unclear, but the presence of label-retaining

cells in most of these domains was demonstrated: RG having

incorporated and maintained BrdU can be found in division

(ie. expressing proliferation markers such as PCNA or MCM

proteins) after a few weeks of chase (M€arz et al., 2010a).

While this, per se, does not attest of NSC character, it shows

that at least some RG divide slowly and recurrently. Zebrafish

pallial RG also display a decreased activation frequency with

aging, and can even get lost in the medial pallium in some

individuals, with a fusion of the two pallial hemispheres

(Edelmann et al., 2013). Perhaps in link with the regionalized

occurrence of this phenomenon, the medial pallium belongs

to the germinal zone domain where telencephalic RG cells are

the earliest formed in embryos and maintained as neurogenic

RG until adulthood (Dirian et al., 2014), and harbors the

highest activation rate along the dorso-medial ventricular

zone in young adults (Alunni et al., 2013). Together, these

points suggest that adult zebrafish RG progenitors can help

provide information on the control of the quiescence/activa-

tion balance of adult NSCs and its impact on NSC

maintenance.

Significant contributions on this front were made

recently, which both stress striking similarities in the control

of adult NSC physiology between zebrafish and mammals

and allowed moving beyond current understanding. Various

signaling pathways, morphogens, neurotransmitter and envi-

ronmental stimuli were identified to promote or restrict NSC

activation. An important question remains to determine the

extent to which these same pathways are involved in activat-

ing RG in regenerative contexts.

Notch Signaling as a Gatekeeper of EndogenousNSC ActivationRG in most subdivisions of the zebrafish adult central nervous

system express high levels of notch receptor genes (Chapouton

et al., 2010). Although, to our knowledge, it has proven very

difficult to date to provide convincing immunohistochemical

staining for Notch intracellular fragments in this system (but

see Berberoglu et al., 2014), several Notch signaling target

genes are also expressed in RG, among which the basic helix-

loop-helix (bHLH) protein-encoding genes of the hairy/e(spl)family such as her4 and her15 (related to mammalian Hes5),

and her6 and her9 (Hes1; Chapouton et al., 2010, 2011; Ganz

et al., 2010), attesting of active Notch signaling in RG. In one

recent report, the NICD fragment was also detected in the

nuclei of RG (Berberoglu et al., 2014), at higher level in quies-

cent than in activated RG, in a manner reminiscent of her4(Berberoglu et al., 2014).

The function of Notch in RG was directly assessed in

the pallium, retina, and spinal cord. Lowering Notch signal-

ing systemically with gamma-secretase inhibitors leads to the

progressive activation of all pallial RG, which turn on expres-

sion of PCNA and MCM proteins and complete a first divi-

sion within 2 days and further divisions at this rate if the

1416 Volume 63, No. 8

treatment is prolonged (Alunni et al., 2013; Chapouton

et al., 2010). In line with the predominance of symmetric

self-renewing divisions occurring in pallial RG under physio-

logical conditions, RG activation upon Notch blockade leads

to a massive amplification of the pallial germinal zone, with

the production of several RG layers. Because of toxicity of

the gamma-secretase compounds, however, the effect of treat-

ments extending beyond one week could not be tested. These

results have several implications: (i) they identify Notch sig-

naling as a major gatekeeper of adult NSC quiescence, (ii)

they demonstrate that all pallial RG are capable of entering

into the cell cycle (thus have potentially maintained NSC

properties), and (iii) they highlight the fact that RG exhibit

different propensities to respond to Notch blockade, some

activating immediately whereas others take a few days.

Whether the latter observation reflects different RG subtypes,

or is in link with the position of the cell within the quies-

cence phase, is not known. In line with the second interpreta-

tion, we demonstrated that RG found in S phase at a given

time are very rarely found in S phase again a few days later,

suggesting the existence of a refractory phase in RG activation

following a first division (Alunni et al., 2013). Although

some discrepancies exist, possibly due to the efficiency of

Notch blockade, comparable findings were obtained in the

retina, where lowering Notch signaling induces cell cycle

entry in some M€uller glia cells (Conner et al., 2014). In the

spinal cord, a comparable quiescence-promoting effect of

Notch signaling was evidenced in the context of regeneration

(Dias et al., 2012).

In the adult zebrafish pallium, notch3 is the sole Notch

receptor gene with detectable expression in quiescent RG

(Alunni et al., 2013). Accordingly, morpholino-mediated

knock-down of notch3 in vivo leads to RG activation, while

notch1b, solely expressed in activated RG, is necessary for self-

renewal (Alunni et al., 2013). These results demonstrate that

distinct Notch receptors are involved along the successive

steps of RG recruitment: Notch3 signaling prevents RG acti-

vation, while Notch1 maintains activated RG. A similar con-

clusion was attained from conditional knock-out experiments

in germinal niches of the adult mouse telencephalon,

although which Notch receptors promote NSC quiescence

remains to be identified (Giachino and Taylor, 2014).

Remaining key questions include whether Notch3 and

Notch1 intrinsically differ in their downstream targets or

whether their different effects on RG fate is contextual, and

which downstream components of Notch signaling promote

RG quiescence versus self-renewal. The transcription factor

Id1, expressed in quiescent RG, was recently identified as a

quiescence-promoting factor (Rodriguez Viales et al., 2014).

Although not directly dependent on Notch for its expression,

Id1 may promote Notch activity through its inhibition of the

Hes proteins negative feedback loop, demonstrated in mouse

(Bai et al., 2007). In the retina, M€uller glia activation upon

Notch blockade necessitates the two transcription factors

Stat3 and Ascl1a, respectively, induced in all M€uller glia or

activated M€uller glia by gamma-secretase inhibitors (Conner

et al., 2014). Thus, these factors are likely involved in RG

activation, and repressed by Notch signaling, under physio-

logical conditions. A potential difference between territories

concerns RG fate upon activation: while Notch blockade is

sufficient to enhance neuronal production in the pallium

(Alunni et al., 2013), neurogenesis fails to complete in the

retina where the induced neuronal progenitors do not survive

(Conner et al., 2014). Different levels of Notch signaling

remaining in the two contexts may however account for this

discrepancy. These observations together suggest the existence

of a common RG quiescence pathway driven by Notch

signaling in the adult central nervous system, although

whether upstream and downstream regulators are shared

remains to be verified.

The mechanisms leading to Notch activation in RG

(and spatio-temporally controlled inhibition to control RG

activation events) largely remain to be identified. In the pal-

lium, the zinc-finger transcription factor Fezf2, expressed at

high levels in quiescent RG, is necessary for the maintenance

of quiescence and correlates with a high level of Notch signal-

ing (Berberoglu et al., 2014), suggesting that it may act

upstream of Notch (or indirectly affect Notch signaling via its

regulation of RG state). The Notch ligand DeltaA is strongly

expressed in activated RG and nonglial proliferating progeni-

tors along the pallial VZ, suggesting that these cells may pro-

mote quiescence in neighboring RG, in a negative feedback-

like mechanism limiting proliferation and recruitment within

the pool of progenitor cells (Chapouton et al., 2010). Other

ligands, such as Jagged, are also expressed in this location.

Other Pathways Controlling RG Quiescence/Activation: Learning from RegenerationNot surprisingly, most pathways promoting RG activation

have been identified in the context of reparative neurogenesis.

In the retina, Ascl1a and Stat3 are necessary for the pro-

liferative response and neurogenesis upon mechanical lesion

(Conner et al., 2014; Fausett et al., 2008). Ascl1a upregulates

expression of Pax6 and Wnt signaling, and directly binds the

promoters of the alpha1-tubulin and lin-28 genes, which are

both activated during the repair process in reactive M€uller

glia (Fausett et al., 2008). The RNA-binding protein Lin-28,

through repression of the miRNA let-7, in turn permits the

lesion-dependent increased expression of hspd1, klf4, oct4(pou5f3) and c-myca and induced expression of pax6, ascl1a

and c-mycb (Ramachandran et al., 2010a). Wnt signaling is

also sufficient to elicit a similar gene expression response and

Than-Trong and Bally-Cuif: Adult Radial Glia in Zebrafish

August 2015 1417

M€uller glia activation (Ramachandran et al., 2011). Several

pathways cross-talking and synergizing with Wnt/bCatenin

were recently identified, including Insulin, Insulin-like growth

factor 1 (IGF-1), FGF2, Heparin-binding EGF-like growth

factor (HB-EGF; Wan et al., 2014). Except for IGF-1 and

FGF2, which are potent together but have little effect alone,

activation of these pathways in isolation is sufficient to induce

M€uller glia activation and the partial reprogramming neces-

sary to confer them, at the population level, a multipotent

progenitor status capable of generating the different retinal

neuronal subtypes. These pathways converge onto the activa-

tion of MAPK and PI3K, and, downstream, Stat3 and

bCatenin (Wan et al., 2014, reviewed in Goldman, 2014). In

parallel, expression of the transcriptional co-repressors Tgif1

and Six3b, induced upon photoreceptors lesion and likely

antagonizing the activity of Smad2/3, limit TGFb signaling

to permit the reactive proliferation of M€uller glia and lower

the gliosis response (Lenkowski et al., 2013; Lenkowski and

Raymond, 2014). These factors may also promote activity of

Ascl1a, Pax6 and Wnt (Lenkowski et al., 2013).

In the pallium, mechanical lesions induce activation of

the transcription factor Gata3 in RG, which is necessary to

promote RG activation. Expression of Gata3 in this context

depends on Fgf signaling. Fgf also promotes progenitor cell

proliferation under physiological conditions (Ganz et al.,

2010) but, both in the na€ıve and lesioned contexts, whether

this regulation directly takes place in RG is not clear. In con-

trast to the factors above, however, Gata3 alone is not suffi-

cient to trigger the re-entry of RG into cycle, indicating that

its activity depends on a regenerative context, or at least on a

context prone to NSC activation (Kizil et al., 2012b).

Finally, in the spinal cord, one of the earliest factors over-

expressed in ventricular ependymoglial cells after injury is Sox2,

and blocking Sox2 function lowers the proliferative response

(Ogai et al., 2014). Sox2 may act redundantly with Sox11

(Guo et al., 2011). Activation of ependymoglial cells along the

central canal upon lesion also depends, at least in part, on Shh

signaling, which is enhanced during repair in ventral ependy-

moglial cells (Reimer et al., 2009). The upregulation of Shh

itself is promoted by the neurotransmitter Dopamine, released

from transsected descending axons (Reimer et al., 2013).

Activation of these pathways/factors in RG could be

triggered by cytokines produced during the inflammatory

response, or by factors released by injured neurons. In the

pallium, the leukotriene C4 (LTC4) inflammation pathway,

mediated by the cysteinyl leukotriene receptor 1 (CysLT1) is

activated in RG upon lesion, is necessary for RG reactive pro-

liferation, and is by itself sufficient to induce Gata3 expres-

sion in RG and cell cycle re-entry (Kyritsis et al., 2012).

TNFa released by dying neurons activates M€uller glia in the

lesioned retina (Nelson et al., 2013; Wan et al., 2014).

Importantly, while activation of several of these path-

ways/factors is sufficient to trigger RG activation in the unin-

jured germinal zones, they are not expressed at detectable

levels under physiological conditions. This is the case for

Wnt/Insulin/IGF-1/FGF2/HB-EGF components, as well as

for Lin-28/Sox2/Nanog/c-Mycb in the retina, and for Gata3

in the pallium. Because activated RG are generally rare under

physiological conditions, they may be difficult to sample in

expression analyses. Thus, one may wonder whether activa-

tion of these pathways truly reflects the use of unique cas-

cades during regenerative RG recruitment, or may be shared

with the “normal” activated state of RG under physiological

conditions. Addressing this question directly remains key for

the future.

In link with this, how to interpret RG activation needs

to be discussed. In the retina, M€uller glia activation during

regeneration is generally interpreted as a partial

“dedifferentiation” process, permitting cells to re-enter the

cell cycle and act as progenitor cells. Arguments in favor of

this interpretation are the concomitant downregulation of

glial marker genes such as GFAP, and the re-expression of

pluripotency factors such as Lin-28, Sox2, Nanog, and c-

Mycb. To which extent this corresponds to the “normal” acti-

vated state of M€uller glia under physiological conditions is

not clear. Further, the low level expression of another subset

of pluripotency factors in the uninjured retina (Klf4, Oct4,

and c-Myca), and the hypomethylation status of several pluri-

potency genes in quiescent M€uller glia (ascl1a, insm1a, hbegfa,

lin28, sox2, oct4, nanog, c-mycb, and klf4), suggest that M€uller

glia constantly retain, albeit partially, some progenitor signa-

ture (Powell et al., 2013). RG activation in the pallium is a

more frequent event and, to our knowledge, is not accompa-

nied by downregulated expression of glial attributes. Pluripo-

tency factors such as Sox2 for example are steadily expressed

in both quiescent and activated RG. Thus, it is possible that

the physiological differentiation status of pallial and retinal

RG are different. Finally, it will be important to determine

whether “dedifferentiation” and cell cycle re-entry can be

uncoupled. In the retina for example, whether Notch block-

ade alone has the full potency to re-induce pluripotency

markers remains to be shown.

Preserving NSCs, Limiting the RegenerativeResponse, and Limiting Gliosis in Response toLesionAmong the factors discussed above, Notch3 signaling and Id1

appear upregulated upon lesion in the pallium, and antagonize

RG proliferation in this context (de Oliveira-Carlos et al.,

2013). Thus, a quiescence-promoting, negative response to

lesion is induced, which involves an enhancement of the nor-

mal quiescence-maintenance cascade. This reaction could be

1418 Volume 63, No. 8

necessary to limit RG recruitment for repair and avoid exhaus-

tion of the SC pool. Along similar lines, expression of the tran-

scription factor Insm1a is initially induced by Ascl1a to

potentiate its proliferation/dedifferentiation effects in reactive

M€uller glia, but displays a later expression phase a few days

postlesion to limit the reactive zone. This is achieved by inhibi-

ting expression of HB-EGFa, and induces the differentiation

of newly formed neuronal precursors. Insm1a acts in the latter

process, in part, by upregulating expression of p57kip (Rama-

chandran et al., 2012).

Other levels of control must also take place, for instance

to adjust the identity of regenerated neurons to those needing

replacement, or to prevent the overproduction of neurons in

response to lesion. On the latter point, we note that this con-

trol can be bypassed in the case of extensive retinal lesions,

where an overproduction of neurons can be observed, or

when a reactive response is initiated by stimulating inflamma-

tion without injury (hence without neurons to be replaced) in

the pallium (Kyritsis et al., 2012; Sherpa et al., 2014).

Finally, a striking difference between zebrafish and

mammals is the lack of glial scaring in the former species,

which likely plays an important role in the efficiency of repar-

ative neurogenesis. Like in mammals, an early upregulation in

the expression of GFAP and BLBP by M€uller glia follows

lesion in the retina, but this response is transient. In the case

of large wounds in the pallium, RG also upregulate their

expression of intermediate filament proteins such as GFAP,

Vimentin or Nestin and of the calcium-binding protein

S100b, and exhibit hypertrophic processes. This resembles a

reactive gliosis response, although the displacement of RG

cells toward the lesion, and scar formation, appear minimal

and transient (Baumgart et al., 2012; M€arz et al., 2011).

Given the deleterious effects of reactive gliosis on repair in

mammals, it is of high interest to understand the bases for

this attenuated glial response in zebrafish. In the retina,

endogenous limitation of the TGFb response plays a key role

in this process (Lenkowski et al., 2013).

Physiological Status and Environmental StimuliVariations in the physiological status of the animal, outside of

regenerative conditions, are important players controlling the

activation state of germinal zones in the mammalian fore-

brain. These aspects remain understudied in the zebrafish.

The lack of standardized animal maintenance conditions

(including the lack of inbred lines) introduces a large degree

of variability between animals that complicates the reading of

subtle effects. Nevertheless, several neurotransmitters and hor-

mones were recently shown to impact RG activation. In the

PVO for example, serotonine (5HT) is necessary for the pro-

liferation of radial progenitors (Perez et al., 2013), and the

projection of 5HT processes toward pallial germinal zones

suggests that it may be the case in this location as well (Lille-

saar et al., 2009).

AromataseB (cyp19a1b), which converts androgens to

estrogen, is strongly expressed in zebrafish RG (like in mam-

malian astrocytes under physiological conditions), and is itself

expressed in response to oestrogens, indicating that RG both

produce and receive this signal (Menuet et al., 2005; Mouriec

et al., 2009). Blocking Aromatase activity in adult zebrafish

led to a trend increase in the number of proliferating progeni-

tors in several brain areas, notably the pallium, suggesting

that oestrogens lower proliferation (Diotel et al., 2013).

Whether they primarily target RG or nonglial proliferating

progenitors remains however to be assessed. In line with a

quiescence-promoting effect on RG, RG maintaining AroB

expression after injury are not activated and presumably do

not participate in the repair process (Diotel et al., 2013).

The roles of other hormones in zebrafish have not been

assessed yet.

Finally, recent works addressed the effect of behavioral

challenges or sensory stimuli on zebrafish adult neurogenesis.

Most interestingly, it was found that sensory stimulation (che-

mosensory or visual stimulations) selectively affected the neu-

rogenic niches in brain areas involved in processing these

stimulations, and this in different ways: for example, enriched

odors increased the number of adult-born neurons in the OB

and vagal lobe, while modifying the spectrum of light or

brightness decreased the size of the progenitor niche in the

TeO area (Lindsey et al., 2014). Changes in the social con-

text, such as social isolation or social novelty, also affect

(decrease) the number of proliferating progenitors in sensory

niches and increase the number of newborn neurons, in a

manner independent of the levels of cortisol, the steroid hor-

mone released in response to stress (Lindsey and Tropepe,

2014). The number and complexity of neurogenic niches in

teleosts, and the complex repertoire of behaviors that a zebra-

fish can exhibit, suggest that much is to be learned from this

model in terms of the environmental modulation of adult

stem cell pools and neurogenesis.

Neurogenesis Cascades From Adult RGAdult neurogenesis cascades have not been extensively studied

using the zebrafish. The expression of proneural factors

(Ascl1, Neurog1) in proliferating, non-RG progenitors, as

well as the involvement of Notch1 in progenitor mainte-

nance, suggests a recapitulation of embryonic cascades starting

from the amplifying neurogenic progenitors step, after these

are produced from activated RG (Alunni et al., 2013; Cha-

pouton et al., 2010).

The factors driving RG toward neurogenic divisions

and/or the generation of neuronal progenitors remain also

understudied.

Than-Trong and Bally-Cuif: Adult Radial Glia in Zebrafish

August 2015 1419

Non-RG Adult NSCs

RG are not however the only NSC type in the zebrafish adult

brain. Indeed, an interesting finding of recent years is the fact

that cells with neuroepithelial (NE) characteristics, rather

than RG, are retained in some territories of the adult zebra-

fish brain and may serve as adult neural progenitors under

physiological and/or regenerative conditions. In most instan-

ces, key questions remain to define the cellular hierarchies

involving these NE progenitors during adult neurogenesis,

and to determine whether they are also at the origin of the

RG considered as NSCs. The long-lasting maintenance of

these NE pools is also an issue given that, in several instances,

they happen to be highly proliferative. Finally, whether such

pools could have equivalent(s) in mammals remains a provoc-

ative but interesting issue.

Neuroepithelial Progenitors in the Retina and OpticTectumThe first territory where a long-lasting neurogenic NE pool

was described is certainly the adult zebrafish retina, where

highly proliferating cells negative for mature glial markers

(e.g., GFAP) but positive for progenitor and NSC markers

(Nestin, BLBP, Sox2) are present at the CMZ, at the interface

between the neural retina and the ciliary epithelium (Fig. 3C;

Raymond et al., 2006). Tracing individual CMZ cells from

the embryonic and juvenile Medaka demonstrates that the

CMZ contains multipotent retinal stem cells that generate all

the neuronal subtypes, as well as M€uller glia, found in the

adult eye (with the exception, under physiological conditions,

of the central retina; Centanin et al., 2011, 2014). The per-

sistence of such multipotent stem cells into adulthood is sug-

gested by the fact that progeny clones from one embryonic or

juvenile CMZ progenitor retain long-lasting contact with the

CMZ. In agreement with such multilineage capacity, progeni-

tors located at the adult CMZ tip maintain expression of

early markers of retinal identity, such as rx1, vsx2/Chx10 or

pax6a. Starting from this territory (peripheral CMZ), the

CMZ is organized as concentric rings (middle and central

CMZ) of progressively increasing commitment. Expression of

the hairy/e(spl) gene her6 is restricted to the peripheral CMZ.

Like NE progenitors, CMZ progenitors also strongly express

Cadherin2 (N-Cad), contact both the apical surface and basal

lamina of the neural retina. Proneural genes such as ascl1a,

and notch/delta pathway genes, are most prominently

expressed in the middle CMZ, and genes activated at differ-

entiation stages, such as neuroD, in the central CMZ. In Xen-

opus at postembryonic stages, it was shown that the

transition from tip CMZ stem cells to committed progenitors

was controlled by the antagonistic actions of Wnt (active at

the tip) and Shh (active further centrally) signaling (Borday

et al., 2012).

Together, these results suggest that, while adult CMZ

progenitors are at the origin of M€uller glia, which can them-

selves act as neural progenitors, their primary contribution to

adult neurogenesis does not involve a transition through the

RG state. From the genes expression analyzed, retinal adult

neurogenesis also largely recapitulates an embryonic neuro-

genesis sequence. Many questions remain open pertaining to

the cellular properties of adult CMZ cells as potential NSCs.

The first is to determine which cellular hierarchies, notably

within the CMZ tip, sustain persistent neurogenesis in the

adult: do quiescent CMZ tip cells exist that generate the bulk

of the CMZ proliferating pool, and which relevant markers

do these cells express? Are these cells inherited from selected

embryonic long-lasting progenitors that maintain the same

properties and role through life? Also, are they multipotent at

the single cell level, as was shown in the embryo, or do they

exhibit some degree of fate specialization? Finally, what are

the relevant factors of their niche and, in that respect, is their

localization “at the tip” a meaningful characteristic?

A seemingly similar organization is found at the poste-

rior edge of the periventricular gray zone (PGZ) of the adult

TeO, where a persistent growth zone has been described

through life in goldfish, Medaka and zebrafish (Fig. 3B; Ray-

mond and Easter, 1983). Most cells of the adult posterior tec-

tal zone display a high proliferative activity, express

progenitor markers (e.g., Sox2, Musashi, Bmi1), show polar-

ized expression of apical markers (ZO-1, gamma-tubulin,

APKc), but do not express glial markers, hence exhibiting

characteristics of NE progenitors (Alunni et al., 2010). Their

tracing with BrdU indicates that they contribute, as a popula-

tion, to glutamatergic or GABAergic neurons in the PGZ

layer, as well as to oligodendrocytes, and also generate the

RG lining the ventral side of the PGZ. Successive pulses with

different thymidine analogues, followed by a chase, also indi-

cate that this zone, like the retina CMZ, is ordered by pro-

gressive stages of commitment: the most proliferating and

least committed progenitors are found close to the “tip” (ie.

in the most posterior location) and commitment is increasing

with cell’s “age” as one moves toward anterior. Interestingly,

in its most posterior domain, the TeO growth zone ends in a

narrow column of aligned cells with NE characteristics, bridg-

ing the TeO (along the torus longitudinalis), the tegmentum

and the cerebellar valvula (Alunni et al., 2010; Chapouton

et al., 2006; Ito et al., 2010; Lindsey et al., 2014). This pop-

ulation is possibly inherited from an equivalent cell pool

described at embryonic and early postembryonic stages and

referred to as the “posterior marginal layer” (PML; Recher

et al., 2013). Both the embryonic and adult PMLs display a

similar columnar arrangement and express the hairy/e(spl)gene her5 (Chapouton et al., 2006). Long-term BrdU tracing

indicates that the tip of the TeO growth zone contains a

1420 Volume 63, No. 8

discrete population of label-retaining, slow cycling progenitors

(Alunni et al., 2010). Within the PML pool, her5-positive

cells tend to show a low proliferation rate and contain some

of the label-retaining cells. They also co-express classical stem

cell markers such as Musashi (Chapouton et al., 2006). Par-

tial and transient tracing of this population using GFP stabil-

ity in her5:egfp transgenic adult zebrafish indicates that it

contributes to generating neurons and oligodendrocytes in the

tegmentum. Thus, the adult her5-positive pool may contrib-

ute to the stem population.

However, the exact location and nature of the adult pro-

genitor cells driving tectal (and tegmental) growth remains

uncertain. Whether these are a specific subpopulation of cells,

or rather a temporary cell state, which are their specific

markers, whether they exhibit NSC properties at the single

cell level, and whether they are inherited from long-lasting

progenitors maintaining such properties through life, remain

open questions. The exact cellular hierarchies driving neuro-

genesis in this domain are also unclear. However, like for the

retina CMZ, the fact that RG generated from the growth

zone rapidly become quiescent makes it very likely that the

nonglial progenitors of the growth zone do directly act as a

major source of adult-born neurons, ie. without an obligatory

transition step through a RG identity.

Neuroepithelial Progenitors in the Adult PalliumUnexpectedly, NE progenitors were also recently discovered

in the adult pallium (Dirian et al., 2014; Fig. 3A). As

described above, the large majority of the adult pallial ventric-

ular zone is occupied by RG that act as constitutive NSCs.

At the lateral edge of the pallium, however, a restricted pool

of nonglial, PCNA-positive progenitors is maintained through

life. Tracing this pool at late juvenile stages indicates that it

contributes to the generation of RG and neurons in the pos-

terolateral pallial domain (Dl, area homologous to the hippo-

campus), and to the progressive growth of this territory

during adulthood, in a manner reminiscent of a growth zone

(Fig. 4B,C). In contrast to the retina and TeO growth zones,

however, the generated RG remain actively neurogenic, and

the genetic tracing of their fate indicates that they likely con-

tribute most of the neurons produced in the lateral pallium

(Dirian et al., 2014). Thus, the NE pool of the pallial edge is

probably neurogenic in a large part via its generation of neu-

rogenic RG NSCs, and may be considered itself as a NSC

population located “hierarchically upstream” of RG NSCs of

the lateral pallium. Genetic tracing indicates that this NE

pool is long-lasting from embryonic stages onward, originat-

ing from a few NE cells located along the roof plate at the

telencephalic-diencephalic junction in the 24hpf embryo

(Dirian et al., 2014). There again, however, the high prolifer-

ative activity of this NE pool raises the question of its main-

tenance through life, and whether a more quiescent

population fuelling this pool exists remains to be determined.

The area of interest, at the junction between the edge of the

pallial ventricular zone and the telencephalic tela choroida,

expresses different signaling factors (Wnt3a in the adult,

Wnt8b, Wnt3a, Fgf8, and Bmp6 in the embryo) and tran-

scription factors (Sox2, Her9) and could show some microre-

gionalization, although this has not been analyzed in detail

with respect to proliferation activity.

Neuroepithelial Progenitors in the AdultCerebellumNeurons are continuously added to the adult cerebellum,

albeit at a decreased rate in adults, from a ventricular progen-

itor zone located dorso-medially along the cerebellar recess, a

remnant of the IVth ventricle, in the corpus and valvula cere-

belli (Kaslin et al., 2009). Progenitors in this cap-like struc-

ture do not express RG markers but are positive for the

progenitor markers Nestin, Sox2, Meis, and Musashi, and

exhibit apico-basal polarity, illustrated by the ventricular local-

ization of ZO-1. They generate proliferating granule cell pre-

cursors that eventually differentiate into granule cell

interneurons, the main neuronal subtype generated in the

adult zebrafish cerebellum (Fig. 3D). These neurons settle

individually into the cerebellar granule zone to complete the

existing circuitry. This adult progenitor domain is inherited

from embryonic cerebellar upper rhombic lip (URL) progeni-

tors, the URL thus being involved in generating granule

interneurons through life (Kizil et al., 2012a; Kaslin et al.,

2013). BrdU pulse-chase assays demonstrate that the cap-like

progenitor pool contains label-retaining cells, although, again,

whether this identifies a specific subpool of progenitors

remains unclear. In contrast, the embryonic and juvenile ven-

tricular zone progenitors, involved in generating other cere-

bellar cell types, give rise to quiescent RG located below the

cap structure in the adult cerebellum and that are mostly

silent in adult. Thus, because of the dual origin of cerebellar

neurons and the differential maintenance of the correspond-

ing progenitors through life, neurogenesis from NE progeni-

tors in the adult zebrafish cerebellum does not generate all

neural subtypes of the adult structure (Kaslin et al., 2013).

This contrasts with the situation in the retina, TeO and pal-

lium. Furthermore, adult cerebellar neurogenesis does not

involve a local growth zone. Whether adult NE progenitors

of the cap-like structure would be capable of increasing their

neurogenic repertoire under regenerative conditions remains

to be determined.

Stem or Progenitor Cells?

The definition of a stem cell is based on functional criteria

(self-renewal and multipotency, at the single cell level) that

Than-Trong and Bally-Cuif: Adult Radial Glia in Zebrafish

August 2015 1421

are often difficult to rigorously assess experimentally, in par-

ticular through the life of an individual.

Of the zebrafish RG progenitors discussed above, the

only one tested, and demonstrated, to get close to the rigor-

ous definition of a NSC are RG of the pallium. Indeed,

brainbow-mediated tracing from embryonic stages until adult-

hood revealed a large number of clones consisting of ventricu-

lar RG and parenchymal neurons produced at all stages of

developmental juvenile and adult life (until the time of analy-

sis; Dirian et al., 2014; Fig. 4C). The presence of several RG

per clone, together with the known progenitor status of adult

RG, demonstrates the occurrence of at least some (symmetric)

self-renewing RG divisions from the embryonic progenitor

that was uniquely labeled, and the additional presence of neu-

rons attests of bipotency of this original progenitor. The indi-

vidual properties of adult pallial RG over successive divisions

during adulthood, however, have not been tested using such a

clonal method yet, and the potential heterogeneity of the

adult RG population in terms of the individual potency and

division frequency of each cell is not known. Accordingly, it

remains unclear whether the pallial germinal zone uses single

cell- or population-based rules to maintain its homeostasis,

and how this compares with the clonal rules used in other

NSC systems ( Calzolari et al., 2015; Klein and Simons,

2011; Simons and Clevers, 2011).

The NSC status of other RG in the zebrafish adult cen-

tral nervous has been tested, at best, over a limited number

of cell divisions. Self-renewing and multipotent cell divisions

were observed in clonal analyses of M€uller glia under regener-

ative conditions. This interesting study demonstrated the

maintenance, in each clone, of one cell expressing RG

markers following M€uller glia reactivation (Nagashima,

2004). Whether the M€uller glia generated through this reac-

tive asymmetrical division is however capable of itself being

reactivated and self-renewing if challenged in additional

lesional events, hence whether it harbors bona fide NSC

properties, remains an important point to assess.

The discovery of maintained NE progenitors in the

adult brain, and, in several cases (TeO, retina, pallium), their

generation of RG that could act as progenitors, poses the

question of their NSC character and their position relative to

RG in the NSC hierarchy. In the lateral pallium, which is

generated from the NE progenitor pool, brainbow tracing

revealed that individual embryonic NE progenitors generate

clones containing ventricular RG and lateral pallial neurons,

demonstrating multipotency. Whether each of these clones

retains a/some adult NE progenitor(s), or whether new NE

progenitors populate the late NE pool from another source to

accommodate RG generation and neurogenesis in the lateral

pallium, has not been rigorously assessed (Fig. 4B,C). As

mentioned, a surprising feature of the NE pool is its overall

high proliferating activity, a feature that cannot be immedi-

ately reconciled with the long-lasting maintenance of stem

cells. One could imagine either the existence of a few steadily

quiescent cells within this pool, of quiescence phases within

the life of each cell of this pool, or of a regular reconstruction

of the pool from another, quiescent source. On the latter

point, it is difficult to imagine that the neighboring RG could

play such a role, since the genetic tracing of pallial RG never

gave rise to cells in the position of the NE pool. Thus, the

upstream quiescent source would be located elsewhere. The

same series of questions remains to be asked for all other

adult NE pools described above.

What Did/Can Zebrafish Adult RG Teach Us ofNSC Biology?

The shared molecular and cellular attributes of mammalian

and zebrafish NSCs in the adult pallium was an initial justifi-

cation for using the zebrafish model to gain relevant and evo-

lutionary insight into adult NSC biology. Much beyond this,

the maintenance of germinal niches containing long-lasting

progenitors in many areas of the zebrafish adult central nerv-

ous system, the recently revealed huge diversity of adult NSCs

in this model, the varieties of their potential and the diversity

of their fates, make the zebrafish an invaluable source of infor-

mation to appreciate the different strategies that can sustain

adult neurogenesis and NSC maintenance in an adult verte-

brate. One hope is to be able to discriminate the obligatory

versus diverged components of the adult NSC state and fate.

Another is to make use of the high neurogenic activity and

reactivation potential of zebrafish germinal pools, both under

physiological and regenerative conditions, to gain insight into

the relevant regulatory cascades involved and to evaluate the

strategies leading to efficient adult neurogenesis. Finally, the

discovery of distinct NSC subtypes in the zebrafish pallium,

the territory homologous to that hosting mammalian adult

NSCs, raises the question of the evolutionary conservation of

these different progenitor subtypes and cascades in this

domain.

NSCs and Neurogenesis in the Adult Zebrafish AreNot a Mere Continuation of Embryogenesis andTeach Us About an Adult ProcessUnder proper nutritional conditions, the zebrafish grows life-

long. In addition, little cell death has been observed among

adult-born neurons, suggesting that most of these do add to

the existing circuitries (Rothenaigner et al., 2011). This has

been interpreted to mean that adult neurogenesis is primarily

taking place to accommodate growth, in a manner more akin

to embryonic neurogenesis than to the adult process of interest

in mammals. Several strong arguments, pertaining to NSCs or

to neurogenesis, stand however against this interpretation.

1422 Volume 63, No. 8

As discussed, zebrafish adult RG are largely quiescent. In

the pallium, where RG are involved in constitutive neurogene-

sis, the average time between two divisions has been estimated

to one month (Rothenaigner et al., 2011), which is of the range

postulated in mouse for adult NSCs of the SEZ and SGZ.

Adult pallial RG require a primary step of activation before

contributing to the neurogenesis, and this step is gated by

Notch3 signaling (Alunni et al., 2013). This is unlike embry-

onic neural progenitors, which do not enter quiescence, and do

not express notch3. In the adult CNS, Notch3 maintains

NSCs, but we postulate that this is rather indirect through lim-

iting the number of NSC divisions and NSC exhaustion, given

that Notch3 blockade primarily triggers NSC amplification. In

contrast, the maintenance of embryonic neural progenitors

requires Notch1 activity, thus resembles the process involved in

both the adult zebrafish and mouse to maintain the progenitor

status of transit amplifying progenitors (Ables et al., 2010;

Ehm et al., 2010; Imayoshi et al., 2010). The role of Notch3

in gating NSC activation in the adult zebrafish pallium is also

reminiscent of the molecular process controlling activation of

satellite cells in the adult muscle, where Notch3 also limits sat-

ellite cell activation (Bjornson et al., 2012; Kitamoto and

Hanaoka, 2010; Mourikis et al., 2012; Mourikis and Taj-

bakhsh, 2014). Thus, the cell cycle characteristics of zebrafish

adult pallial RG, and its molecular control, is similar to the

processes controlling SC activation in mature organs such as

the brain and muscle in the adult mouse.

A second important argument lies in the selectivity of

adult neurogenesis in zebrafish. Unlike an embryonic process,

which would continuously generate all neuronal subtypes,

adult neurogenesis is limited to a subset of neurons in some

of the neurogenic niches (such as the adult cerebellum, for

example, where adult progenitors solely produce granule neu-

rons under physiological conditions (Kaslin et al., 2009). In

addition, recent studies were able to link the rate of adult

neurogenesis with ongoing processing demand of the external

environment. Hence, for example, new neurons are more

actively produced in the visual processing centers upon chal-

lenge of visual stimulation. These observations together argue

against neurogenesis being an additive process for the sole

purpose of growth (Lindsey and Tropepe, 2014; Lindsey

et al., 2014). The zebrafish is capable of complex behaviors,

in particular regarding social behaviors. Thus, it will also help

unravel new links between neurogenesis control and physio-

logical status. In fact, juvenile/adult neurogenesis may also be

particularly involved in processing reward (Webb et al.,

2009).

The NSC NicheThe “niche”, the cellular or chemical microenvironment neces-

sary to maintain stemness, is an important concept of the SC

field, but is experimentally more difficult to define, as it is

often not easy to disentangle between direct and indirect

effects on SC fate. In the adult mammalian SEZ, the mor-

phology of NSCs, contacting a tight arrangement of ependy-

mal cells and the cerebro-spinal fluid at the apical side, as well

as blood vessels and/or pericytes by their basal process, sug-

gested that these components were involved in maintaining

NSC stemness, hence were part of the niche (recently reviewed

in Lim and Alvarez-Buylla, 2014; Silva-Vargas et al., 2013;

Urban and Guillemot, 2014). Experimental manipulations

confirmed this hypothesis. In the embryonic mouse cortex, a

further signal driving RG maintenance was identified as Notch

signaling emanating from neuronal precursors, in a negative

feedback mechanism (Yoon et al., 2008). The data obtained

so far in the adult zebrafish stress the importance of secreted

signaling factors (for example, Fgf signaling in the subpallium

or cerebellum; Kaslin et al., 2009; Ganz et al., 2010) in con-

trolling the maintenance of RG or NE progenitors. The

source of these factors however remains to be studied in detail;

they could emanate from neighboring cells, the CSF or blood

vessels, with which RG cells also establish tight contacts.

An interesting aspect of most ventricular surfaces in the

zebrafish adult CNS, and in particular the telencephalon, is

the quasi-absence of ependymal cells. Ependymal cells in the

forebrain have only been described in the diencephalon

(Lindsey et al., 2012), and the apical membranes of RG

themselves constitute virtually all of the pallial and subpallial

ventricular surface. This excludes the ependymal layer as a

niche factor in these domains. In fact, analysis of the pallial

RG stressed the importance of local interactions taking place

within the progenitor population itself as major homeostasis

factors. Activation of the major quiescence pathway identified

to date, Notch signaling, is triggered by contact with cells

expressing the Notch ligands Delta or Jagged. Expression

analyses showed that these ligands (DeltaA, Jagged1b) are

essentially expressed by activated RG and proliferating neuro-

nal progenitors along the ventricular zone (Alunni et al.,

2013; Chapouton et al., 2010; de Oliveira-Carlos et al.,

2013). In addition, following a short pulse of Notch block-

ade, RG activation primarily takes place close to cells having

recently divided (Chapouton et al., 2010). These observations

suggest that dividing progenitors are major activators of

Notch signaling in RG, contributing directly (Notch1) or

indirectly (Notch3) to their maintenance and/or their silenc-

ing, in a feedback mechanism. Together, these results are

pointing to the germinal pool of progenitors itself as a major,

self-regulating niche, capable of adjusting the demand in

NSC recruitment to ongoing neurogenesis.

In addition to feedback control from activated progeni-

tors, analysis of the zebrafish adult pallial ventricular zone

highlights a striking widespread pattern of activation events

Than-Trong and Bally-Cuif: Adult Radial Glia in Zebrafish

August 2015 1423

across the RG sheet (unpub.). Thus, some additional large

scale regulation must take place, at any given time point and

independently of proliferating neuronal progenitors (which

are uniformly distributed just below the RG layer) to spread

out RG activation events across the RG pool. Whether such a

mechanism is also at play within the mouse SEZ and/or SGZ

pools was not analyzed to date, in part due to the complex

geometry of germinal zones, their relative paucity in NSCs

and the presence of ependymal cells, in this model. Under-

standing the large-scale cues that control the spatio-temporal

pattern of NSC activation, recruitment and properties will

bring important insight on the spatio-temporal organization

of the niche. Thus, for the first time, the zebrafish will pro-

vide invaluable access to the organization of NSC niches at a

large scale, and to the homeostasis of NSC pools at the popu-

lation level with spatial resolution.

Finally, the joint analysis of adult NE pools in the

zebrafish adult brain points to common properties in their

location: in most instances, these are positioned in marginal

aspects of the neural plate/tube, at the junction between neu-

rogenic and choroid plexus tissues. In the lateral pallium, the

NE pool is juxtaposed to the tela choroida bridging the first

ventricle, this from the earliest developmental stages following

formation of the neural rod until adulthood (Dirian et al.,

2014). In the cerebellum, progenitors of the URL, at least at

embryonic and juvenile stages, are bordered by the tela cho-

roida covering the fourth ventricle (Kaslin et al., 2009). At

the posterior edge of the TeO, NE progenitors are associated/

partly overlapping with a small tela bridging the PGZ and

the cerebellar valvula, possibly deriving from the anterior tip

of the fourth ventricle tela (Chapouton et al., 2006). Finally,

in the retina, the CMZ is located at the junction between the

neural retinal tissue and the nonpigmented ciliary epithelium,

a cellular monolayer with secretory properties (Kasper et al.,

1987). In addition to the secretory activity of choroid plex-

uses, which may provide NE cells with immediate access to

secreted factors, these junctional territories themselves are

often also signaling centers. This is well known in the

embryo, where the neural tube roof plate is bordered by

Wnt1-expressing cells, for example, and where the juxtaposi-

tion of territories of different positional identities can be suf-

ficient to trigger the generation of signaling centers. But

signaling activity is also present next to zebrafish adult NE

pools. Indeed, a sustained expression of the signaling factor

Wnt3 is apparent at the lateral pallial edge (Dirian et al.,

2014), or of Wnt factors at the CMZ tip (Gorsuch and

Hyde, 2014; Lenkowski and Raymond, 2014). Expression of

Wnt at the posterior edge of the TeO, to our knowledge, has

not been analyzed in the adult, but is present in the embryo

and may be maintained. Finally, there is an Fgf signaling

source close to the cap-like progenitor pool in the cerebellum

(Kaslin et al., 2009). Another important common molecular

signature of these adult domains is the expression of “Notch-

independent” e(spl) genes, such as her6 in the retinal CMZ

(Raymond et al., 2006), her5 in the posterior TeO (Chapou-

ton et al., 2006), and her9 in the NE pool of the adult pal-

lium (Dirian et al., 2014). Similar expression is found in the

corresponding NE pools at embryonic stages. Given that such

junctional locations exist between neurogenic tissue and cho-

roid plexuses in all vertebrates, it would be highly interesting

to examine the status of neural cells in their proximity. The

pallium is an especially promising territory, given that it har-

bours neurogenic domains in rodents. Most interestingly, a

late source for SGZ NSCs was recently identified as a Gli1-

positive territory at the edge of the tel- and di-encephalon (Li

et al., 2013), in a location that could be holomogous to the

zebrafish lateral pallial edge. Analyzing these source cells in

more detail will reveal they have the characteristics of a long-

lasting NE pool.

NSC HeterogeneityAn arising debate in the adult mammalian NSC field is the

degree of heterogeneity of NSCs within germinal niches. In

fact, genetic tracing using various “NSC promoters” (e.g.,

GFAP, BLBP, GLAST, Nestin. . .) to drive conditional Cre

recombination revealed a variety of fates, suggesting that sup-

posedly “generic” molecular markers in fact distinguish NSC

subtypes or substates (Giachino and Taylor, 2014). In addi-

tion to the basic difference between NE and RG pools high-

lighted above, zebrafish adult germinal zones also display a

high degree of heterogeneity between RG located within an

individual pool. In the pallium for example, some degree of

heterogeneity was revealed when comparing the markers her4,

GFAP, BLBP, AroB, S100b, and Nestin on individual sec-

tions. Our unpublished data further document large-scale het-

erogeneities between pallial areas, which can be more or less

enriched in the proportion of RG expressing a given marker,

associated with different properties (different birth dates for

example, or different growth rates). Together, it is likely that

the cellular compositions of zebrafish neurogenic RG niches

are similar to that of the mammalian SEZ and SGZ in both

heterogeneity and richness of cellular states (Lindsey et al.,

2012; Marz et al., 2010a), and that the zebrafish will be an

invaluable model to approach the meaning of these heteroge-

neities. Several key questions may be related to these hetero-

geneities in RG, including the degree of stemness, the degree

of quiescence, or neurogenic competence. Importantly, the

zebrafish adult pallial germinal zone encompasses the territo-

ries homologous to the mammalian SEZ and SGZ, which

will permit direct comparative and cross-feeding studies

between these species to understand the dynamics of NSC

properties.

1424 Volume 63, No. 8

Concluding Remarks

In conclusion, the zebrafish adult CNS does not simply retain

a high number of progenitor cells with activatable neurogenic

potential. It offers a number of niches containing bona fide

NSCs of adult characteristics, sharing cellular attributes and

molecular recruitment pathways with mammalian adult NSCs,

and organized in pools of rich complexity and precise spatial

regionalization. Like in mammals, these pools respond differen-

tially to environmental demand, notably producing neurons

dedicated to distinct functions. Through these properties, this

model will provide key insight into the molecular mechanisms

controlling the adult NSC state(s) and fate(s), in their hetero-

geneity, their dynamic transitions, and their regulation at the

population scale. The variety of locations (niches) and NSC

subtypes (RG, NE progenitors) present in the zebrafish adult

CNS will further expand our understanding of the possible

strategies sustaining efficient NSC maintenance and neurogene-

sis. Finally, the efficiency of zebrafish neuronal repair pathways

includes both the boosting of constitutively neurogenic NSCs,

the partial dedifferentiation/reactivation of silent progenitors,

and the induction of NSC protection mechanisms limiting

reactive NSC recruitment, a diversity that will be instrumental

to design the best strategies when trying to stimulate adult

neurogenesis for repair in mammals.

Acknowledgment

Grant sponsor: EU project ZF-Health (agreement no

242048); Grant number: FP7/2010-2015; Grant sponsor: the

ANR; Grant number: ANR-2012-BSV4-0004-01; Grant

sponsor: European Research Council; Grant number: AdG

322936.

We are grateful to all members of the ZEN team for their

constant critical insight on the topics discussed in this review,

and to M. Coolen for her critical reading of the manuscript.

ReferencesAbles JL, Decarolis NA, Johnson MA, Rivera PD, Gao Z, Cooper DC, RadtkeF, Hsieh J, Eisch AJ. 2010. notch1 is required for maintenance of the reser-voir of adult hippocampal stem cells. J Neurosci 30:10484–10492.

Adolf B, Chapouton P, Lam CS, Topp S, Tannhauser B, Strahle U, Gotz M,Bally-Cuif L. 2006a. Conserved and acquired features of adult neurogenesisin the zebrafish telencephalon. Dev Biol 295:278–293.

Adolf B, Chapouton P, Lam CS, Topp S, Tannhauser B, Strahle U, Gotz M,Bally-Cuif L. 2006b. Conserved and acquired features of adult neurogenesisin the zebrafish telencephalon. Dev Biol 295:278–293.

Alexandre P, Reugels AM, Barker D, Blanc E, Clarke JD. 2010. Neuronsderive from the more apical daughter in asymmetric divisions in the zebrafishneural tube. Nat Neurosci 13:673–679.

Alunni A, Hermel JM, Heuze A, Bourrat F, Jamen F, Joly JS. 2010. Evidencefor neural stem cells in the medaka optic tectum proliferation zones. DevNeurobiol 70:693–713.

Alunni A, Krecsmarik M, Bosco A, Galant S, Pan L, Moens CB, Bally-Cuif L.2013. notch3 signaling gates cell cycle entry and limits neural stem cellamplification in the adult pallium. Development 140:3335–3347.

Anthony TE, Klein C, Fishell G, Heintz N. 2004. Radial glia serve as neuronalprogenitors in all regions of the central nervous system. Neuron 41:881–890.

Ayari B, El Hachimi KH, Yanicostas C, Landoulsi A, Soussi-Yanicostas N.2010. Prokineticin 2 expression is associated with neural repair of injuredadult zebrafish telencephalon. J Neurotrauma 27:959–972.

Bai G, Sheng N, Xie Z, Bian W, Yokota Y, Benezra R, Kageyama R, GuillemotF, Jing N. 2007. Id sustains hes1 expression to inhibit precocious neurogene-sis by releasing negative autoregulation of hes1. Dev Cell 13:283–297.

Baumgart EV, Barbosa JS, Bally-Cuif L, Gotz M, Ninkovic J. 2012. Stabwound injury of the zebrafish telencephalon: A model for comparative analy-sis of reactive gliosis. Glia 60:343–357.

Baye LM, Link BA. 2007. Interkinetic nuclear migration and the selection ofneurogenic cell divisions during vertebrate retinogenesis. J Neurosci 27:10143–10152.

Berberoglu MA, Dong Z, Li G, Zheng J, Trejo Martinez Ldel C, Peng J,Wagle M, Reichholf B, Petritsch C, Li H, Pleasure SJ, Guo S. 2014. Heteroge-neously expressed fezf2 patterns gradient notch activity in balancing the qui-escence, proliferation, and differentiation of adult neural stem cells.J Neurosci 34:13911–13923.

Bernardos RL, Barthel LK, Meyers JR, Raymond PA. 2007. Late-stage neuro-nal progenitors in the retina are radial muller glia that function as retinal stemcells. J Neurosci 27:7028–7040.

Bernardos RL, Raymond PA. 2006. GFAP transgenic zebrafish. Gene Expr Pat-terns 6:1007–1013.

Bjornson CR, Cheung TH, Liu L, Tripathi PV, Steeper KM, Rando TA. 2012.Notch signaling is necessary to maintain quiescence in adult muscle stemcells. Stem Cells 30:232–242.

Boniface EJ, Lu J, Victoroff T, Zhu M, Chen W. 2009. FlEx-based transgenicreporter lines for visualization of cre and flp activity in live zebrafish. Genesis47:484–491.

Borday C, Cabochette P, Parain K, Mazurier N, Janssens S, Tran HT, SekkaliB, Bronchain O, Vleminckx K, Locker M, Perron M. 2012. Antagonistic cross-regulation between wnt and hedgehog signalling pathways controls post-embryonic retinal proliferation. Development 139:3499–3509.

Briona LK, Dorsky RI. 2014. Radial glial progenitors repair the zebrafish spinalcord following transection. Exp Neurol 256:81–92.

Calzolari F, Michel J, Baumgart V, Theis E, G€otz F, Ninkovic MJ. 2015. Fastclonal expansion and limited neural stem cell self-renewal in the adult sub-ependymal zone. Nat Neurosci 4:490–494.

Centanin L, Ander JJ, Hoeckendorf B, Lust K, Kellner T, Kraemer I, Urbany C,Hasel E, Harris WA, Simons BD, Wittbrodt J. 2014. Exclusive multipotencyand preferential asymmetric divisions in post-embryonic neural stem cells ofthe fish retina. Development 141:3472–3482.

Centanin L, Hoeckendorf B, Wittbrodt J. 2011. Fate restriction and multipo-tency in retinal stem cells. Cell Stem Cell 9:553–562.

Chapouton P, Adolf B, Leucht C, Tannhauser B, Ryu S, Driever W, Bally-CuifL. 2006. her5 expression reveals a pool of neural stem cells in the adultzebrafish midbrain. Development 133:4293–4303.

Chapouton P, Jagasia R, Bally-Cuif L. 2007. Adult neurogenesis in non-mammalian vertebrates. Bioessays 29:745–757.

Chapouton P, Skupien P, Hesl B, Coolen M, Moore JC, Madelaine R,Kremmer E, Faus-Kessler T, Blader P, Lawson ND, Bally-Cuif L. 2010. Notchactivity levels control the balance between quiescence and recruitment ofadult neural stem cells. J Neurosci 30:7961–7974.

Chapouton P, Webb KJ, Stigloher C, Alunni A, Adolf B, Hesl B, Topp S,Kremmer E, Bally-Cuif L. 2011. Expression of hairy/enhancer of split genes inneural progenitors and neurogenesis domains of the adult zebrafish brain.J Comp Neurol 519:1748–1769.

Conner C, Ackerman KM, Lahne M, Hobgood JS, Hyde DR. 2014. Repressingnotch signaling and expressing TNFalpha are sufficient to mimic retinalregeneration by inducing muller glial proliferation to generate committedprogenitor cells. J Neurosci 34:14403–14419.

Than-Trong and Bally-Cuif: Adult Radial Glia in Zebrafish

August 2015 1425

Coolen M, Thieffry D, Drivenes O, Becker TS, Bally-Cuif L. 2012. miR-9 con-trols the timing of neurogenesis through the direct inhibition of antagonisticfactors. Dev Cell 22:1052–1064.

Corbo CP, Othman NA, Gutkin MC, Alonso Adel C, Fulop ZL. 2012. Use ofdifferent morphological techniques to analyze the cellular composition of theadult zebrafish optic tectum. Microsc Res Tech 75:325–333.

Costa MR, Ortega F, Brill MS, Beckervordersandforth R, Petrone C,Schroeder T, Gotz M, Berninger B. 2011. Continuous live imaging of adultneural stem cell division and lineage progression in vitro. Development 138:1057–1068.

Das T, Payer B, Cayouette M, Harris WA. 2003. In vivo time-lapse imaging ofcell divisions during neurogenesis in the developing zebrafish retina. Neuron37:597–609.

de Haan G, Nijhof W, Van Zant G. 1997. Mouse strain-dependent changes infrequency and proliferation of hematopoietic stem cells during aging: Corre-lation between lifespan and cycling activity. Blood 89:1543–1550.

de Oliveira-Carlos V, Ganz J, Hans S, Kaslin J, Brand M. 2013. Notch recep-tor expression in neurogenic regions of the adult zebrafish brain. PLoS One8:e73384.

Dias TB, Yang YJ, Ogai K, Becker T, Becker CG. 2012. Notch signaling con-trols generation of motor neurons in the lesioned spinal cord of adult zebra-fish. J Neurosci 32:3245–3252.

Dimou L, G€otz M. 2014. Glial cells as progenitors and stem cells: New rolesin the healthy and diseased brain. Physiol Rev 94:709–737.

Diotel N, Vaillant C, Gabbero C, Mironov S, Fostier A, Gueguen MM,Anglade I, Kah O, Pellegrini E. 2013. Effects of estradiol in adult neurogene-sis and brain repair in zebrafish. Horm Behav 63:193–207.

Dirian L, Galant S, Coolen M, Chen W, Bedu S, Houart C, Bally-Cuif L,Foucher I. 2014. Spatial regionalization and heterochrony in the formation ofadult pallial neural stem cells. Dev Cell 30:123–136.

Doetsch F, Garcia-Verdugo JM, Alvarez-Buylla A. 1999. Regeneration of agerminal layer in the adult mammalian brain. Proc Natl Acad Sci USA 96:11619–11624.

Dong Z, Yang N, Yeo SY, Chitnis A, Guo S. 2012. Intralineage directionalnotch signaling regulates self-renewal and differentiation of asymmetricallydividing radial glia. Neuron 74:65–78.

Edelmann K, Glashauser L, Sprungala S, Hesl B, Fritschle M, Ninkovic J,Godinho L, Chapouton P. 2013. Increased radial glia quiescence, decreasedreactivation upon injury and unaltered neuroblast behavior underliedecreased neurogenesis in the aging zebrafish telencephalon. J Comp Neu-rol 521:3099–3115.

Ehm O, Goritz C, Covic M, Schaffner I, Schwarz TJ, Karaca E, Kempkes B,Kremmer E, Pfrieger FW, Espinosa L, Bigas A, Giachino C, Taylor V, Frisen J, LieDC. 2010. RBPJkappa-dependent signaling is essential for long-term maintenanceof neural stem cells in the adult hippocampus. J Neurosci 30:13794–13807.

Encinas JM, Michurina TV, Peunova N, Park JH, Tordo J, Peterson DA, FishellG, Koulakov A, Enikolopov G. 2011. Division-coupled astrocytic differentia-tion and age-related depletion of neural stem cells in the adult hippocampus.Cell Stem Cell 8:566–579.

Fausett BV, Goldman D. 2006. A role for alpha1 tubulin-expressing mullerglia in regeneration of the injured zebrafish retina. J Neurosci 26:6303–6313.

Fausett BV, Gumerson JD, Goldman D. 2008. The proneural basic helix-loop-helix gene ascl1a is required for retina regeneration. J Neurosci 28:1109–1117.

Folgueira M, Bayley P, Navratilova P, Becker TS, Wilson SW, Clarke JD. 2012.Morphogenesis underlying the development of the everted teleost telen-cephalon. Neural Dev 7:32.

Ganz J, Kaslin J, Freudenreich D, Machate A, Geffarth M, Brand M. 2012.Subdivisions of the adult zebrafish subpallium by molecular marker analysis.J Comp Neurol 520:633–655.

Ganz J, Kaslin J, Hochmann S, Freudenreich D, Brand M. 2010. Heterogene-ity and fgf dependence of adult neural progenitors in the zebrafish telen-cephalon. Glia 58:1345–1363.

Giachino C, Taylor V. 2014. Notching up neural stem cell homogeneity inhomeostasis and disease. Front Neurosci 8:32.

Goldman D. 2014. Muller glial cell reprogramming and retina regeneration.Nat Rev Neurosci 15:431–442.

Gorsuch RA, Hyde DR. 2014. Regulation of muller glial dependent neuronalregeneration in the damaged adult zebrafish retina. Exp Eye Res 123:131–140.

G€otz M. 2013. Radial glial cells. In: Neuroglia, 3rd, edited by Helmut Ketten-mann and Bruce R. Ransom, Oxford University Press.

Grandel H, Kaslin J, Ganz J, Wenzel I, Brand M. 2006. Neural stem cells andneurogenesis in the adult zebrafish brain: Origin, proliferation dynamics,migration and cell fate. Dev Biol 295:263–277.

Grupp L, Wolburg H, Mack AF. 2010. Astroglial structures in the zebrafishbrain. J Comp Neurol 518:4277–4287.

Guo Y, Ma L, Cristofanilli M, Hart RP, Hao A, Schachner M. 2011. Transcrip-tion factor Sox11b is involved in spinal cord regeneration in adult zebrafish.Neuroscience 172:329–341.

Imayoshi I, Sakamoto M, Yamaguchi M, Mori K, Kageyama R. 2010. Essentialroles of notch signaling in maintenance of neural stem cells in developingand adult brains. J Neurosci 30:3489–3498.

Ito Y, Tanaka H, Okamoto H, Ohshima T. 2010. Characterization of neuralstem cells and their progeny in the adult zebrafish optic tectum. Dev Biol342:26–38.

Kaslin J, Ganz J, Geffarth M, Grandel H, Hans S, Brand M. 2009. Stem cellsin the adult zebrafish cerebellum: Initiation and maintenance of a novel stemcell niche. J Neurosci 29:6142–6153.

Kaslin J, Kroehne V, Benato F, Argenton F, Brand M. 2013. Developmentand specification of cerebellar stem and progenitor cells in zebrafish: Fromembryo to adult. Neural Dev 8:9.

Kasper M, Munchow T, Stosiek P, Karsten U. 1987. Similarities in the compositionof intermediate filaments between human ciliary and choroid plexus epithelia-correlation with similar secretory functions. Acta Histochem 82:134–136.

Kawai H, Arata N, Nakayasu H. 2001. Three-dimensional distribution of astro-cytes in zebrafish spinal cord. Glia 36:406–413.

Kishimoto N, Alfaro-Cervello C, Shimizu K, Asakawa K, Urasaki A, Nonaka S,Kawakami K, Garcia-Verdugo JM, Sawamoto K. 2011. Migration of neuronalprecursors from the telencephalic ventricular zone into the olfactory bulb inadult zebrafish. J Comp Neurol 519:3549–3565.

Kishimoto N, Shimizu K, Sawamoto K. 2012. Neuronal regeneration in azebrafish model of adult brain injury. Dis Model Mech 5:200–209.

Kitamoto T, Hanaoka K. 2010. notch3 null mutation in mice causes musclehyperplasia by repetitive muscle regeneration. Stem Cells 28:2205–2216.

Kizil C, Kaslin J, Kroehne V, Brand M. 2012a. Adult neurogenesis and brainregeneration in zebrafish. Dev Neurobiol 72:429–461.

Kizil C, Kyritsis N, Dudczig S, Kroehne V, Freudenreich D, Kaslin J, Brand M.2012b. Regenerative neurogenesis from neural progenitor cells requiresinjury-induced expression of gata3. Dev Cell 23:1230–1237.

Klein AM, Simons BD. 2011. Universal patterns of stem cell fate in cyclingadult tissues. Development 138:3103–3111.

Koke JR, Mosier AL, Garcia DM. 2010. Intermediate filaments of zebrafish ret-inal and optic nerve astrocytes and M€uller glia: Differential distribution ofcytokeratin and GFAP. BMC Research Notes 3:50–56.

Kroehne V, Freudenreich D, Hans S, Kaslin J, Brand M. 2011. Regenerationof the adult zebrafish brain from neurogenic radial glia-type progenitors.Development 138:4831–4841.

Kuscha V, Barreiro-Iglesias A, Becker CG, Becker T. 2012a. Plasticity of tyro-sine hydroxylase and serotonergic systems in the regenerating spinal cord ofadult zebrafish. J Comp Neurol 520:933–951.

Kuscha V, Frazer SL, Dias TB, Hibi M, Becker T, Becker CG. 2012b. Lesion-induced generation of interneuron cell types in specific dorsoventral domainsin the spinal cord of adult zebrafish. J Comp Neurol 520:3604–3616.

1426 Volume 63, No. 8

Kyritsis N, Kizil C, Zocher S, Kroehne V, Kaslin J, Freudenreich D, Iltzsche A,Brand M. 2012. Acute inflammation initiates the regenerative response in theadult zebrafish brain. Science 338:1353–1356.

Lam CS, Marz M, Strahle U. 2009. gfap and nestin reporter lines reveal char-acteristics of neural progenitors in the adult zebrafish brain. Dev Dyn 238:475–486.

Lau BY, Mathur P, Gould GG, Guo S. 2011. Identification of a brain centerwhose activity discriminates a choice behavior in zebrafish. Proc Natl AcadSci USA 108:2581–2586.

Lenkowski JR, Qin Z, Sifuentes CJ, Thummel R, Soto CM, Moens CB,Raymond PA. 2013. Retinal regeneration in adult zebrafish requires regulationof TGFbeta signaling. Glia 61:1687–1697.

Lenkowski JR, Raymond PA. 2014. Muller glia: Stem cells for generation andregeneration of retinal neurons in teleost fish. Prog Retin Eye Res 40:94–123.

Li G, Fang L, Fernandez G, Pleasure SJ. 2013. The ventral hippocampus isthe embryonic origin for adult neural stem cells in the dentate gyrus. Neuron78:658–672.

Lillesaar C, Stigloher C, Tannhauser B, Wullimann MF, Bally-Cuif L. 2009.Axonal projections originating from raphe serotonergic neurons in the devel-oping and adult zebrafish, danio rerio, using transgenics to visualize raphe-specific pet1 expression. J Comp Neurol 512:158–182.

Lim DA, Alvarez-Buylla A. 2014. Adult neural stem cells stake their ground.Trends Neurosci 37:563–571.

Lindsey BW, Darabie A, Tropepe V. 2012. The cellular composition of neuro-genic periventricular zones in the adult zebrafish forebrain. J Comp Neurol520:2275–2316.

Lindsey BW, Di Donato S, Kaslin J, Tropepe V. 2014. Sensory-specific modu-lation of adult neurogenesis in sensory structures is associated with the typeof stem cell present in the neurogenic niche of the zebrafish brain. Eur J Neu-rosci 40:3591–3607.

Lindsey BW, Tropepe V. 2014. Changes in the social environment induce neu-rogenic plasticity predominantly in niches residing in sensory structures of thezebrafish brain independently of cortisol levels. Dev Neurobiol 74:1053–1077.

Linser PJ, Smith K, Angelides K. 1985. A comparative analysis of glial andneuronal markers in the retina of fish: Variable character of horizontal cells.J Comp Neurol 237:264–272.

Lyons DA, Guy AT, Clarke JD. 2003. Monitoring neural progenitor fatethrough multiple rounds of division in an intact vertebrate brain. Develop-ment 130:3427–3436.

Malatesta P, Hack MA, Hartfuss E, Kettenmann H, Klinkert W, Kirchhoff F,Gotz M. 2003. Neuronal or glial progeny: Regional differences in radial gliafate. Neuron 37:751–764.

Malatesta P, Hartfuss E, Gotz M. 2000. Isolation of radial glial cells byfluorescent-activated cell sorting reveals a neuronal lineage. Development127:5253–5263.

M€arz M, Chapouton P, Diotel N, Vaillant C, Hesl B, Takamiya M, Lam CS, KahO, Bally-Cuif L, Strahle U. 2010a. Heterogeneity in progenitor cell subtypes inthe ventricular zone of the zebrafish adult telencephalon. Glia 58:870–888.

M€arz M, Schmidt R, Rastegar S, Strahle U. 2010b. Expression of the transcrip-tion factor olig2 in proliferating cells in the adult zebrafish telencephalon.Dev Dyn 239:3336–3349.

M€arz M, Schmidt R, Rastegar S, Strahle U. 2011. Regenerative response follow-ing stab injury in the adult zebrafish telencephalon. Dev Dyn 240:2221–2231.

Menuet A, Pellegrini E, Brion F, Gueguen MM, Anglade I, Pakdel F, Kah O.2005. Expression and estrogen-dependent regulation of the zebrafish brainaromatase gene. J Comp Neurol 485:304–320.

Montgomery JE, Parsons MJ, Hyde DR. 2010. A novel model of retinal abla-tion demonstrates that the extent of rod cell death regulates the origin ofthe regenerated zebrafish rod photoreceptors. J Comp Neurol 518:800–814.

Morrens J, Van Den Broeck W, Kempermann G. 2012. Glial cells in adult neu-rogenesis. Glia 60:159–174.

Morris AC, Scholz TL, Brockerhoff SE, Fadool JM. 2008. Genetic dissectionreveals two separate pathways for rod and cone regeneration in the teleostretina. Dev Neurobiol 68:605–619.

Mouriec K, Gueguen MM, Manuel C, Percevault F, Thieulant ML, Pakdel F,Kah O. 2009. Androgens upregulate cyp19a1b (aromatase B) gene expres-sion in the brain of zebrafish (danio rerio) through estrogen receptors. BiolReprod 80:889–896.

Mourikis P, Sambasivan R, Castel D, Rocheteau P, Bizzarro V, Tajbakhsh S.2012. A critical requirement for notch signaling in maintenance of the quies-cent skeletal muscle stem cell state. Stem Cells 30:243–252.

Mourikis P, Tajbakhsh S. 2014. Distinct contextual roles for notch signalling inskeletal muscle stem cells. BMC Dev Biol 14:2.

Mueller T, Dong Z, Berberoglu MA, Guo S. 2011. The dorsal pallium inzebrafish, danio rerio (cyprinidae, teleostei). Brain Res 1381:95–105.

Mueller T, Wullimann MF. 2009. An evolutionary interpretation of teleosteanforebrain anatomy. Brain Behav Evol 74:30–42.

Nagashima RJ. 2004. Decreased incidence of capsule complications and vit-reous loss during phacoemulsification in eyes with pseudoexfoliation syn-drome. J Cataract Refract Surg 30:127–131.

Nelson CM, Ackerman KM, O’Hayer P, Bailey TJ, Gorsuch RA, Hyde DR.2013. Tumor necrosis factor-alpha is produced by dying retinal neurons andis required for muller glia proliferation during zebrafish retinal regeneration.J Neurosci 33:6524–6539.

Noctor SC, Flint AC, Weissman TA, Wong WS, Clinton BK, Kriegstein AR.2002. Dividing precursor cells of the embryonic cortical ventricular zone havemorphological and molecular characteristics of radial glia. J Neurosci 22:3161–3173.

Ogai K, Nakatani K, Hisano S, Sugitani K, Koriyama Y, Kato S. 2014. Functionof sox2 in ependymal cells of lesioned spinal cords in adult zebrafish. Neuro-sci Res 88:84–87.

Park D, Xiang AP, Mao FF, Zhang L, Di CG, Liu XM, Shao Y, Ma BF, Lee JH,Ha KS, Walton N, Lahn BT. 2010. Nestin is required for the proper self-renewal of neural stem cells. Stem Cells 28:2162–2171.

Park HC, Shin J, Roberts RK, Appel B. 2007. An olig2 reporter gene marksoligodendrocyte precursors in the postembryonic spinal cord of zebrafish.Dev Dyn 236:3402–3407.

Pellegrini E, Menuet A, Lethimonier C, Adrio F, Gueguen MM, Tascon C,Anglade I, Pakdel F, Kah O. 2005. Relationships between aromatase and estro-gen receptors in the brain of teleost fish. Gen Comp Endocrinol 142:60–66.

Pellegrini E, Mouriec K, Anglade I, Menuet A, Le Page Y, Gueguen MM,Marmignon MH, Brion F, Pakdel F, Kah O. 2007. Identification of aromatase-positive radial glial cells as progenitor cells in the ventricular layer of the fore-brain in zebrafish. J Comp Neurol 501:150–167.

Perez MR, Pellegrini E, Cano-Nicolau J, Gueguen MM, Menouer-Le GuillouD, Merot Y, Vaillant C, Somoza GM, Kah O. 2013. Relationships betweenradial glial progenitors and 5-HT neurons in the paraventricular organ ofadult zebrafish—potential effects of serotonin on adult neurogenesis. Eur JNeurosci 38:3292–3301.

Peterson RE, Fadool JM, McClintock J, Linser PJ. 2001. Muller cell differentia-tion in the zebrafish neural retina: Evidence of distinct early and late stagesin cell maturation. J Comp Neurol 429:530–540.

Peyre E, Morin X. 2012. An oblique view on the role of spindle orientation invertebrate neurogenesis. Dev Growth Differ 54:287–305.

Powell C, Grant AR, Cornblath E, Goldman D. 2013. Analysis of DNA methyl-ation reveals a partial reprogramming of the muller glia genome during ret-ina regeneration. Proc Natl Acad Sci USA 110:19814–19819.

Rabe H, Koschorek E, Nona SN, Ritz HJ, Jeserich G. 1999. Voltage-gatedsodium and potassium channels in radial glial cells of trout optic tectum stud-ied by patch clamp analysis and single cell RT-PCR. Glia 26:221–232.

Ramachandran R, Fausett BV, Goldman D. 2010a. Ascl1a regulates mullerglia dedifferentiation and retinal regeneration through a Lin-28-dependent,let-7 microRNA signalling pathway. Nat Cell Biol 12:1101–1107.

Than-Trong and Bally-Cuif: Adult Radial Glia in Zebrafish

August 2015 1427

Ramachandran R, Reifler A, Parent JM, Goldman D. 2010b. Conditional geneexpression and lineage tracing of tuba1a expressing cells during zebrafishdevelopment and retina regeneration. J Comp Neurol 518:4196–4212.

Ramachandran R, Zhao XF, Goldman D. 2011. Ascl1a/dkk/beta-catenin sig-naling pathway is necessary and glycogen synthase kinase-3beta inhibition issufficient for zebrafish retina regeneration. Proc Natl Acad Sci USA 108:15858–15863.

Ramachandran R, Zhao XF, Goldman D. 2012. Insm1a-mediated gene repres-sion is essential for the formation and differentiation of muller glia-derivedprogenitors in the injured retina. Nat Cell Biol 14:1013–1023.

Raymond PA, Barthel LK, Bernardos RL, Perkowski JJ. 2006. Molecular char-acterization of retinal stem cells and their niches in adult zebrafish. BMC DevBiol 6:36.

Raymond PA, Easter SSJ. 1983. Postembryonic growth of the optic tectum ingoldfish. I. Location of germinal cells and numbers of neurons produced.J Neurosci 3:1077–1091.

Raymond PA, Hitchcock PF, Palopoli MF. 1988. Neuronal cell proliferationand ocular enlargement in black moor goldfish. J Comp Neurol 276:231–238.

Recher G, Jouralet J, Brombin A, Heuze A, Mugniery E, Hermel JM,Desnoulez S, Savy T, Herbomel P, Bourrat F, Peyrieras N, Jamen F, Joly JS.2013. Zebrafish midbrain slow-amplifying progenitors exhibit high levels oftranscripts for nucleotide and ribosome biogenesis. Development 140:4860–4869.

Reimer MM, Kuscha V, Wyatt C, Sorensen I, Frank RE, Knuwer M, Becker T,Becker CG. 2009. Sonic hedgehog is a polarized signal for motor neuronregeneration in adult zebrafish. J Neurosci 29:15073–15082.

Reimer MM, Norris A, Ohnmacht J, Patani R, Zhong Z, Dias TB, Kuscha V,Scott AL, Chen YC, Rozov S, Frazer SL, Wyatt C, Higashijima S, Patton EE,Panula P, Chandran S, Becker T, Becker CG. 2013. Dopamine from the brainpromotes spinal motor neuron generation during development and adultregeneration. Dev Cell 25:478–491.

Reimer MM, Sorensen I, Kuscha V, Frank RE, Liu C, Becker CG, Becker T.2008. Motor neuron regeneration in adult zebrafish. J Neurosci 28:8510–8516.

Rodriguez Viales, R, Diotel, N, Ferg, M, Armant, O, Eich, J, Alunni, A, Marz,M, Bally-Cuif, L, Rastegar, S, Strahle, U. 2014. The helix-loop-helix proteinid1 controls stem cell proliferation during regenerative neurogenesis in theadult zebrafish telencephalon. Stem Cells 33: 892–903.

Rothenaigner I, Krecsmarik M, Hayes JA, Bahn B, Lepier A, Fortin G, Gotz M,Jagasia R, Bally-Cuif L. 2011. Clonal analysis by distinct viral vectors identifiesbona fide neural stem cells in the adult zebrafish telencephalon and charac-terizes their division properties and fate. Development 138:1459–1469.

Schmidt, R, Beil, T, Strahle, U, Rastegar, S. 2014. Stab wound injury of thezebrafish adult telencephalon: A method to investigate vertebrate brain neu-rogenesis and regeneration. J Vis Exp e51753.

Schmidt R, Strahle U, Scholpp S. 2013. Neurogenesis in zebrafish—Fromembryo to adult. Neural Dev 8:3.

Sherpa T, Lankford T, McGinn TE, Hunter SS, Frey RA, Sun C, Ryan M,Robison BD, Stenkamp DL. 2014. Retinal regeneration is facilitated by thepresence of surviving neurons. Dev Neurobiol 74:851–876.

Silva-Vargas V, Crouch EE, Doetsch F. 2013. Adult neural stem cells and theirniche: A dynamic duo during homeostasis, regeneration, and aging. CurrOpin Neurobiol 23:935–942.

Simons BD, Clevers H. 2011. Strategies for homeostatic stem cell self-renewal in adult tissues. Cell 145:851–862.

Skaggs K, Goldman D, Parent JM. 2014. Excitotoxic brain injury in adultzebrafish stimulates neurogenesis and long-distance neuronal integration.Glia 62:2061–2079.

Spalding KL, Bergmann O, Alkass K, Bernard S, Salehpour M, Huttner HB,Bostrom E, Westerlund I, Vial C, Buchholz BA, Possnert G, Mash DC, DruidH, Frisen J. 2013. Dynamics of hippocampal neurogenesis in adult humans.Cell 153:1219–1227.

Suh H, Consiglio A, Ray J, Sawai T, D’Amour KA, Gage FH. 2007. In vivo fateanalysis reveals the multipotent and self-renewal capacities of sox21 neuralstem cells in the adult hippocampus. Cell Stem Cell 1:515–528.

Thisse, B, Pflumio, S, F€urthauer, M, Loppin, B, Heyer, V, Degrave, A, Woehl,R, Lux, A, Steffan, T, Charbonnier, XQ, Thisse, C. Expression of the zebrafishgenome during embryogenesis. ZFIN direct Data 2001. Submission. http://zfin.org.

Thummel R, Kassen SC, Montgomery JE, Enright JM, Hyde DR. 2008. Inhibi-tion of muller glial cell division blocks regeneration of the light-damagedzebrafish retina. Dev Neurobiol 68:392–408.

Tomizawa K, Inoue Y, Nakayasu H. 2000. A monoclonal antibody stains radialglia in the adult zebrafish (danio rerio) CNS. J Neurocytol 29:119–128.

Urban N, Guillemot F. 2014. Neurogenesis in the embryonic and adult brain:Same regulators, different roles. Front Cell Neurosci 8:396.

Vaziri H, Dragowska W, Allsopp RC, Thomas TE, Harley CB, Lansdorp PM.1994. Evidence for a mitotic clock in human hematopoietic stem cells: Lossof telomeric DNA with age. Proc Natl Acad Sci USA 91:9857–9860.

von Trotha JW, Vernier P, Bally-Cuif L. 2014. Emotions and motivated behav-ior converge on an amygdala-like structure in the zebrafish. Eur J Neurosci40: 3302–3315.

Wan J, Zhao XF, Vojtek A, Goldman D. 2014. Retinal injury, growth factors,and cytokines converge on beta-catenin and pStat3 signaling to stimulate ret-ina regeneration. Cell Rep 9:285–297.

Webb KJ, Norton WH, Trumbach D, Meijer AH, Ninkovic J, Topp S, Heck D,Marr C, Wurst W, Theis FJ, Spaink HP, Bally-Cuif L. 2009. Zebrafish rewardmutants reveal novel transcripts mediating the behavioral effects of ampheta-mine. Genome Biol 10:R81.

Wullimann MF, Rupp B, Reichert H. 1996. Neuroanatomy of the zebrafishbrain. A topological atlas. Basel: Birkh€auser Verlag. pp 160.

Yoon KJ, Koo BK, Im SK, Jeong HW, Ghim J, Kwon MC, Moon JS, Miyata T,Kong YY. 2008. Mind bomb 1-expressing intermediate progenitors generatenotch signaling to maintain radial glial cells. Neuron 58:519–531.

1428 Volume 63, No. 8

本文献由“学霸图书馆-文献云下载”收集自网络，仅供学习交流使用。

学霸图书馆（www.xuebalib.com）是一个“整合众多图书馆数据库资源，

提供一站式文献检索和下载服务”的24 小时在线不限IP

图书馆。

图书馆致力于便利、促进学习与科研，提供最强文献下载服务。

图书馆导航：

图书馆首页 文献云下载 图书馆入口 外文数据库大全 疑难文献辅助工具