Neurogenin1 is a determinant of zebrafish basalforebrain dopaminergic neurons and is regulatedby the conserved zinc finger protein Tof�FezlJae-Yeon Jeong*, Zev Einhorn*†, Sara Mercurio†‡, Susie Lee*, Billy Lau*, Marina Mione‡, Stephen W. Wilson‡,and Su Guo*§

*Department of Biopharmaceutical Sciences, Programs in Human Genetics and Biological Sciences, University of California, San Francisco, CA 94143;and ‡Department of Anatomy and Developmental Biology, University College London, Gower Street, London WC1E 6BT, United Kingdom

Edited by Yuh Nung Jan, University of California School of Medicine, San Francisco, CA, and approved February 9, 2006 (received for reviewJanuary 12, 2006)

The development of vertebrate basal forebrain dopaminergic (DA)neurons requires the conserved zinc finger protein Too Few (Tof�Fezl) in zebrafish. However, how Tof�Fezl regulates the commit-ment and differentiation of these DA neurons is not known.Proneural genes encoding basic helix–loop–helix transcription fac-tors regulate the development of multiple neuronal lineages, buttheir involvement in vertebrate DA neuron determination is un-clear. Here we show that neurogenin 1 (ngn1), a vertebrateproneural gene related to the Drosophila atonal, is expressed inand required for specification of DA progenitor cells, and whenoverexpressed leads to supernumerary DA neurons in the fore-brain of zebrafish. Overexpression of ngn1 is also sufficient toinduce tyrosine hydroxylase expression in addition to the pan-neuronal marker Hu in nonneural ectoderm. We further show thatTof�Fezl is required to establish basal forebrain ngn1-expressingDA progenitor domains. These findings identify Ngn1 as a deter-minant of brain DA neurons and provide insights into how Tof�Fezlregulates the development of these clinically important neuronaltypes.

neurogenin 1 � pluripotent neural stem cell � neurotransmitterphenotype � commitment and differentiation

The determination of neurotransmitter phenotype is an im-portant aspect of neuronal differentiation, and in this regard,

dopaminergic (DA) neurons have attracted considerable atten-tion because of their functional and medical importance (1).Degeneration of substantia nigra DA neurons in humans is ahallmark of Parkinson’s disease, and the malfunction of DAneurons in other brain regions is implicated in psychiatricdisorders and neuroendocrine dysregulation. Therefore, under-standing the determination of DA phenotype and the specifi-cation of DA neuronal circuitry may provide mechanistic andtherapeutic insights into these disorders. To date, only limitednumber of known or putative transcriptional regulators, includ-ing Pax6, Dlx, Nurr1, Lmx1a, Lmx1b, Msx1, Foggy, and Too Few(Tof�Fez1), have been implicated in the specification of DAphenotype in vertebrates (2–7, 35). Despite this knowledge, themechanisms leading to the early commitment of pluripotentneural stem cells to DA lineage remain elusive.

The earliest DA neurons in zebrafish are detected at �24 hpostfertilization (hpf) in the basal forebrain (8). They expresstyrosine hydroxylase (TH), the rate-limiting enzyme in dopa-mine synthesis, and dopamine transporter (DAT), a proteininvolved in dopamine reuptake (9). Later during development,these DA neurons have both ascending and descending projec-tions, and are believed to be homologous to mammalian DAneurons of both the basal forebrain and midbrain (10, 11).

Through forward genetic analysis, an adult viable zebrafishmutant named too few (tofm808) has been isolated that displaysselective deficits of basal forebrain DA as well as adjacentserotonergic (5HT) neurons (8). Molecular characterizations

have revealed that the too few mutant carries a point mutationthat changes Cys-287 to Ser in the second of the six zinc fingermotifs of the conserved zinc finger protein Fezl (Tof�Fezl) (5).Whereas these studies establish an important role of Tof�Fezl inDA neuron development, the mechanism by which Tof�Fezl actsin DA neuron specification is not clear.

The role of basic helix–loop–helix (bHLH) proteins in neuraldevelopment has been initially discovered in Drosophila (12) andlater studied in multiple vertebrate neuronal lineages (13, 14).Neurogenins are a family of bHLH proteins related to DrosophilaAtonal (15). Multiple members of the Neurogenin family have beendiscovered in mice (16, 17), and they play partially overlapping rolesin regulating neuronal specification. For example, Neurogenin1 andNeurogenin2 are involved in the development of dorsal root ganglia(18, 19), and Neurogenin2 is a determination factor for placodalsensory neuron development (20). In zebrafish, only one neuroge-nin-like gene, named neurogenin 1 (ngn1), has been identified so far(21). Ngn1 is required for the development of spinal, cranialsensory, and epiphysial neurons (22–24).

Here we report that early-born basal forebrain DA neurons arederived from Ngn1-expressing progenitor cells. We demonstratethat Ngn1 is necessary for the development of these DA neurons,and overexpression of Ngn1 leads to supernumerary DA neuronsin the zebrafish forebrain, and induces TH� cells with an apparentneuronal morphology on the yolk surface ectoderm. Furthermore,we show that Tof�Fezl is expressed in an overlapping fashion withngn1, and is required to specify ngn1-expressing DA progenitordomains in the basal forebrain. Together, our data identify earlyregulatory steps that lead to the commitment of pluripotent neuralstem cells to dopaminergic phenotype.

ResultsNgn1 Is Expressed in Dopaminergic Progenitor Cells. While examin-ing genes that are expressed in the vicinity of DA neurons in thebasal forebrain of zebrafish, we noted that ngn1 is expressed inthe basal forebrain as distinct clusters before DA neuron ap-pearance (Fig. 1 A and B), and later in close proximity to theappearing nascent TH� DA neurons (Fig. 1C). At 28 hpf,between approximately three and six DA neurons were detectedin the basal forebrain on each side of the midline, and they werethe earliest born DA neurons in the entire zebrafish embryo (Fig.1E). To determine whether these early-born basal forebrain DAneurons are derived from ngn1-expressing progenitor cells, a

Conflict of interest statement: No conflicts declared.

This paper was submitted directly (Track II) to the PNAS office.

Abbreviations: DA, dopaminergic; TH, tyrosine hydroxylase; hpf, hours postfertilization;DAT, dopamine transporter; 5HT, serotonin; MO, morpholino.

†Z.E. and S.M. contributed equally to this work.

§To whom correspondence should be addressed. E-mail: suguo@itsa.ucsf.edu.

© 2006 by The National Academy of Sciences of the USA

www.pnas.org�cgi�doi�10.1073�pnas.0600337103 PNAS � March 28, 2006 � vol. 103 � no. 13 � 5143–5148

NEU

ROSC

IEN

CE

Dow

nloa

ded

by g

uest

on

Dec

embe

r 15

, 202

0

GFP transgene driven by the ngn1 promoter (25) was used tofollow the fates of ngn1� cells after they stop expressing ngn1,because GFP protein is more stable than ngn1 mRNA, andusually lasts �6 h after the cessation of gfp transcript synthesis(26). Because the first TH-immunoreactive (ir) DA neurons aredetected �24 hpf (8), we would be able to observe that DAneurons were GFP� at 28 hpf, should they be derived fromngn1-expressing progenitor cells. Therefore, double immuno-staining with GFP and TH antibodies were carried out in 28 hpfembryos, and the results revealed that all TH� DA neurons werepositive for GFP at 28 hpf (Fig. 1 D–I, and Fig. 7 A–F, which ispublished as supporting information on the PNAS web site).Among GFP�TH� neurons, some displayed strong GFP signalthat might correspond to newly born cells, whereas faint GFPsignals were detected in all other TH� DA neurons that werelikely born at earlier times (Fig. 1I).

A group of serotonin (5HT) neurons (usually about one to two

neurons at �28 hpf) develop in close proximity to DA neuronsin the basal forebrain (4). To determine whether ngn1-GFP isalso detectable in these neighboring 5HT neurons, we carried outdouble immunostaining with GFP and 5HT antibodies. Theseresults indicate that these neighboring serotonergic (5HT) neu-rons were always GFP negative (Figs. 1 J–O and 7 G–L). Thisobservation indicates that the early-born basal forebrain DAneurons are derived from ngn1-expressing progenitor cells, but5HT neurons are probably not.

Ngn1 Is Required for the Development of Dopaminergic Neurons in theBasal Forebrain. To determine whether ngn1 is required for DAneuron development, we injected a ngn1 morpholino (MO) that has

Fig. 2. ngn1 is required for DA neuron development in the basal forebrain.All images are lateral views of anterior brain regions except I–L. Anterior is tothe left, and dorsal is up. (A–L) Immunostaining with TH antibody (A, B, G, andL), in situ with TH (E and F), or DAT antisense probe (C and D), or immuno-staining with 5HT antibody (M and N), shows that DA neurons in the ventralforebrain is completely absent at 32 hpf, largely absent at 48 hpf, and remainsignificantly defective at 72 hpf in ngn1-morphants, whereas TH� DA neuronsappear normal in the retina, TH� gut sympathetic neurons are normal in thetrunk region of ngn1 morphants, and 5HT neurons are not affected. (Insets)Magnified views of basal forebrain 5HT neurons. DA, dopaminergic neurons;DAT, dopamine transporter; di, diencephalon; epi, epiphysis; hb, hindbrain;LC, locus coeruleus; mhb, midhindbrain boundary; t, telencephalon. (Scalebar, 32 �m.)

Fig. 1. ngn1 is expressed in DA progenitors. All images are lateral views ofanterior brain regions. Anterior is to the left, and dorsal is up. (A and B) ngn1expression at 14- and 20-somite stages, respectively (arrows point to severalclusters in the ventral forebrain). (C) A 28-hpf embryo showing ngn1 expres-sion (purple) in close proximity to TH� DA neurons (red). (D–F) Confocalimages of 28 hpf ngn1-GFP transgenic embryos immunostained with GFPantibody (green, D), TH antibody (red, E), and the merged image (F), showingthat GFP is detected in TH� DA neurons. (G–I) High-magnification views of D–F.(J–L) Confocal images of 28-hpf ngn1-GFP transgenic embryos immunostainedwith GFP antibody (green, J), 5HT antibody (red, K), and the merged image (L),showing that GFP is not detected in 5HT neurons. (M–O) High-magnificationviews of J–L. di, diencephalon, t, telencephalon. (Scale bars, 64 �m in A and B,60 �m in C–F and J–L, and 3 �m in G–I and M–O.)

5144 � www.pnas.org�cgi�doi�10.1073�pnas.0600337103 Jeong et al.

Dow

nloa

ded

by g

uest

on

Dec

embe

r 15

, 202

0

been shown to effectively and specifically knockdown Ngn1 proteinexpression (22, 23). Abrogation of Ngn1 activity by MO injectionled to a severe loss of DA neurons in the ventral forebrain (95% ofembryos showing none or few TH or DAT-positive cells by 48 hpf,n � 54) (Fig. 2 A–D). Some residual TH staining was detected in48 hpf ngn1 MO embryos, but the intensity was much reduced ascompared to control embryos (Fig. 2 A and B). DAT expressionappeared to be completely abolished in these embryos (Fig. 2 C andD). This phenotype was confirmed in ngn1 mutants (see Fig. 6 Gand H) isolated from mutagenesis screens (24, 27). Examination ofan earlier developmental stage revealed that, at 32 hpf, DA neuronswere completely absent in Ngn1-defective animals (Fig. 2 E and F).Examination of a later postembryonic larval stage show that, at 72hpf, DA neurons appeared to recover somewhat but remainedsignificantly defective in the basal forebrain (estimated to be�80–90% deficiency compared to control) (Fig. 2 G and H). Theoverall brain patterning of ngn1 morphants was normal, and TH�

DA neurons in the retina (Fig. 2 I and J), TH� noradrenergicneurons of locus coeruleus (Fig. 2 A and B) and the sympatheticganglia (Fig. 2 K and L) appeared not significantly affected in thengn1 morphants. Moreover, the neighboring basal forebrain 5HTneurons were not obviously reduced (Fig. 2 M, N, and Insets). Thelack of 5HT defects in the absence of ngn1 activity is consistent withthe lineage study showing that these 5HT neurons were not derivedfrom ngn1-expressing progenitor cells (Fig. 1 J–O). These datasuggest that ngn1 is required for the development of basal forebrainDA neurons, but not for the development of neighboring 5HTneurons and several TH� neuronal groups in other regions of thenervous system.

To determine the state of DA progenitor domains in the absence

of ngn1 activity, we analyzed ngn1 expression in the ngn1 mor-phants. We observed strongly enhanced ngn1 RNA labeling in themorphant embryos, possibly due to the fact that the ngn1 MOinhibits the translation of ngn1 transcript, and somehow leads to anenhanced transcript stability, although a negative feedback regu-lation of ngn1 transcription by Ngn1 protein is also a possibility.Nevertheless, these analyses revealed that the basal forebrainngn1-expressing progenitor domains were intact in the ngn1 mor-phants at multiple developmental stages examined (Fig. 3), sug-gesting that, in the absence of ngn1 activity, progenitor domains areformed but fail to commit and differentiate into DA fate.

Overexpression of Ngn1 Leads to Supernumerary DA Neurons in theBasal Forebrain and TH� Cells with an Apparent Neuronal Morphologyon the Yolk Surface Ectoderm. Given that ngn1 is essential for DAneuron development, we next determined whether overexpress-ing ngn1 has an impact on DA phenotype by microinjecting ngn1

Fig. 3. ngn1 expression in ngn1 morphants. Anterior is to the left, and dorsalis up. ngn1 expression is higher in the ngn1 morphants, and Ngn1-expressingprogenitor domains are intact in the basal forebrain of ngn1 morphants atvarious developmental stages as indicated.

Fig. 4. Misexpression of ngn1 leads to supernumerary DA neurons in theforebrain and induces TH� cells with an apparent neuronal morphology onthe yolk surface ectoderm. (A–H) Immunostaining with TH antibody (A–F) andin situ with DAT antisense probe (G and H) shows that DA neurons aresignificantly increased in the ventral forebrain and ectopically induced in thetelencephalon (Di DA neurons are out of focal planes in E–H) in ngn1-mRNAinjected embryos. (I and J) Immunostaining with TH antibody showing anectopic TH� cell on the yolk surface ectoderm of ngn1-injected embryo. (K andL) A high magnification view of (I and J), and Inset showing an ectopic TH� cellwith an apparent neuronal morphology on the yolk surface ectoderm. DA,dopaminergic neurons; DAT, dopamine transporter; di, diencephalon; LC,locus coeruleus; mhb, midhindbrain boundary; t, telencephalon. (Scale bar, 15�m in K and L, 32 �m in A–H, and 64 �m in I and J.)

Jeong et al. PNAS � March 28, 2006 � vol. 103 � no. 13 � 5145

NEU

ROSC

IEN

CE

Dow

nloa

ded

by g

uest

on

Dec

embe

r 15

, 202

0

mRNA into one- to eight-cell-stage zebrafish embryos. A pro-portion of ngn1 mRNA injected embryos (usually �15%) dis-played morphological deformity including malformed eyes andbrain (data not shown), as has been reported (21). Therefore, inall mRNA injection experiments carried out in our study, onlythe injected embryos that display grossly normal brain morphol-ogy were used to determine the impact of ngn1 misexpression onDA neuron development. We found that misexpression of ngn1at 100 ng�ml led to a significant increase of DA neurons in thebasal forebrain (77% embryos showing a significant increase ofDA neurons, n � 64) (Fig. 4 A–H) and in some instances a fewectopic DA neurons were detected in the telencephalon (13%,n � 220, Fig. 4 E–H). Using the pan-neuronal marker HuC, weobserved that, as previously reported (21), the expression ofpan-neuronal marker HuC in the CNS was not grossly increasedby Ngn1 overexpression (data not shown). These observationssuggest that Ngn1’s ability in increasing DA neuronal subtypesis not a consequence of an overall increased neuronal productionin CNS. Interestingly, we found that misexpression of ngn1 alsoinduced TH� cells on the yolk surface ectoderm, albeit at a lownumber and a low frequency (�10% of injected embryos haveabout one to three TH� cells on the yolk surface ectoderm, n �50) (Fig. 4 J and L), despite this, such ectopic TH� cells werenever observed in control embryos (0% of control embryos haveTH� cells on the yolk surface ectoderm, n � 211) (Fig. 4 I andK). DAT-positive cells appeared not induced on the yolk surfaceectoderm by ngn1 misexpression at the concentration that wetested (data not shown). These TH� cells appeared to have aneuronal morphology (Fig. 4 L and Inset). This observation isconsistent with the fact that ngn1 is shown to be capable ofinducing the pan-neuronal marker Hu on the yolk surfaceectoderm (21). Taken together, we conclude that overexpressionof ngn1 is able to increase the production of DA neurons in theforebrain and induce TH� neurons in nonneural ectoderm. It is

worth noting that increased production of DA neurons waslargely detected in their endogenous location of the basalforebrain, suggesting that Ngn1 requires additional factors in thisregion to induce DA neuron production.

Tof�Fezl Expression Overlaps with ngn1-Expressing DA ProgenitorDomains in the Basal Forebrain. Our loss- and gain-of-functionanalyses indicate that Ngn1 is a crucial factor for basal forebrain DAneurons. To explore how Ngn1 might be regulated in the develop-ment of DA neurons, we investigated the relationship betweentof�fezl and ngn1. Tof�Fezl was previously identified through ze-brafish forward genetics as an essential factor for the developmentof basal forebrain DA neurons; however, its mechanism of actionis unknown. In 10-somite (�14 hpf) embryos, tof�fezl expressionwas detected broadly in the forebrain (Fig. 5A), preceding thedetection of ngn1 expression in this region (Fig. 5B). At �18 hpf,whereas tof�fezl expression remained broad in the forebrain (Fig.5C), ngn1 expression became detectable in distinct clusters ofprogenitor cells in the forebrain as well as other brain regions (Fig.5D). During the period when DA neurons are specified, tof�Fezlexpression became less broadly expressed and was detected intelencephalon, as well as, in diencephalon (the basal forebrain),close to but not within DA neurons (Fig. 5E). Double labelingexperiments showed that tof�fezl expression domains encompassedthe ngn1-expressing domains in the basal forebrain (Fig. 5F). Bothtof�fezl and ngn1 expression in these domains were down-regulatedby 48 hpf (Fig. 5 G and H). These analyses suggest that ngn1 andtof�fezl are coexpressed in progenitor domains that are in closeproximity to basal forebrain DA neurons.

The Basal Forebrain Expression of Ngn1 Is Transiently Reduced in thetoo few Mutant and Is Abrogated in the tof�fezl Morphant. Becausetof�fezl expression is detected earlier than ngn1 in the forebrain

Fig. 5. tof�fezl is expressed earlier than ngn1 and later in overlappingdomains with ngn1 in the basal forebrain All images are lateral views ofanterior brain regions. (A–D) In situ hybridization with tof�fezl cRNA probe (Aand C) and ngn1 cRNA probe (B and D). (E–H) Double in situ hybridization oftof�fezl (red)�TH (purple) (E and G), tof�fezl (red) � ngn1 (purple) (F), and TH(red) � ngn1 (pruple) (H). DA, dopaminergic neurons; di, diencephalon; t,telencephalon. (Scale bar, 32 �m in E–H and 64 �m in A and D.)

Fig. 6. The requirement of tof�fezl in establishing ngn1-expressing DAprogenitor domains. (A–D) ngn1 expression in 20-somite wild-type sibling andthe tof mutant embryos shows that ngn1 expression in the basal forebrain(arrows) is reduced at 20 somites (A and B) but appears normal at 28 hpf (C andD). (E and F) ngn1 expression is largely absent in the basal forebrain of tof�fezlmorphant. (G and H) TH and tof�fezl in situ (both purple) shows that, althoughTH� DA neurons are largely absent, the fezl expression appears normal in thengn1 mutant. di, diencephalon; t, telencephalon. (Scale bars, 32 �m in G andH and 64 �m in A–F.)

5146 � www.pnas.org�cgi�doi�10.1073�pnas.0600337103 Jeong et al.

Dow

nloa

ded

by g

uest

on

Dec

embe

r 15

, 202

0

region and later overlaps with basal forebrain ngn1-expressingprogenitor domains, we determined whether tof�fezl acts upstreamof ngn1 in regulating DA neuron development. We first examinedngn1 expression in the tofm808 mutant, which results from a singleamino acid change in the Tof�Fezl protein (5). The results showedthat ngn1 expression was reduced in the basal forebrain at �20-somite stage (Fig. 6 A and B), but was later found comparable tothe wild-type siblings at 28 hpf (Fig. 6 C and D). Because it ispossible that tofm808 mutation might be a hypomorphic allele, wesought to determine whether the basal forebrain ngn1 expressiondomains would be more severely affected in a case of stronger lossof Tof�Fezl activity. Therefore, we designed two splicing morpho-linos that target different exon�intron boundaries to knockdownTof�Fezl activity (Fig. 8, which is published as supporting infor-mation on the PNAS web site). Knockdown of Tof�Fezl with eithermorpholino led to a severe defect of ngn1-expressing domains in thebasal forebrain, whereas ngn1 expressing domains in other brainregions were largely unperturbed (Fig. 6 E and F). These analysessuggest that tof�fezl is required to establish ngn1-expressing pro-genitor domains in the basal forebrain. tof�fezl expression in thengn1 mutant was largely normal as compared to the wild-typesiblings (Fig. 6 G and H), suggesting that ngn1 is not required toregulate tof�fezl expression.

DiscussionCompared to some other neuronal types, our knowledge on thedevelopment of vertebrate DA neurons is rather limited. Theinvolvement of these neurons in multiple human neurologicaldisorders including Parkinson’s disease, addiction, and schizophre-nia makes it a worthwhile endeavor to understand the mechanismsunderlying their development. Here we provide in vivo evidence inzebrafish for an important role of Ngn1 in basal forebrain DAneuron development: Ngn1 is required for the development ofDA neurons, and when overexpressed, can lead to supernumeraryDA neurons in the forebrain and TH� neurons on the yolk surfaceectoderm. Furthermore, we show that Tof�Fezl, an evolutionarilyconserved zinc finger protein, is required to establish basal fore-brain ngn1-expressing progenitor domains. These analyses identifycrucial early regulatory steps in the commitment of multipotentneural stem cells to a dopaminergic lineage.

The Requirement of Ngn1 in DA Neuron Development. Expressionanalysis, lineage study, as well as loss-of-function genetic analysisestablishes that ngn1 is expressed in the basal forebrain DAneuronal progenitor cells and is required for the specification ofDA neurons. Because ngn1-expressing progenitor domains ap-pear intact in the absence of Ngn1 activity, the cellular role ofNgn1 is likely required within the progenitor cells. Because theother neuronal types that the basal forebrain ngn1-expressingprogenitor cells may give rise to (we showed that the adjacent5HT neurons are unlikely to be derived from these ngn1-expressing progenitor cells and are not obviously increased inngn1-defective embryos) are unlikely to be known, the fate ofthese basal forebrain ngn1-expressing progenitors in the absenceof Ngn1 activity remains to be determined. In addition, althoughbasal forebrain DA neuron defects remain in the postembryoniclarval brain, they appear to be less severe than earlier stages,suggesting that additional mechanisms likely exist to regulate thegeneration of DA neurons at later stages.

Ngn1 may play one or multiple of the following roles in DAprogenitor cells: it may regulate their proliferation or cell cycle exit,regulate their acquisition of general neuronal properties, and�orregulate their acquisition of subtype identity, as proposed based onstudies of proneural genes in multiple other neuronal lineages (14).Understanding the precise role of Ngn1 in DA neuron developmentrequires the understanding of Ngn1’s function at the molecularlevel, for instance, the identification of Ngn1 target genes in the DAneuronal lineage. The well known direct target of Ngn1, NeuroD,

is surprisingly not detected in the basal forebrain region in zebrafish(J.-Y.J. and S.G., unpublished data), although the possibility existsthat perhaps other yet unidentified NeuroD-like genes are ex-pressed in this region in zebrafish. Nevertheless, the requirementand sufficiency of Ngn1 for the appearance of TH indicates thatNgn1 may directly regulate DA neuron subtype-related genes. Atpresent, few genes are known to be involved in DA neurondevelopment in zebrafish. Although the differentiation of basalforebrain DA neurons in mice is shown to require Pax6 and Dlx (6,7), and the differentiation of midbrain DA neurons in mice is shownto require the nuclear receptor Nurr1 (2, 28) and the LIM home-odomain transcription factor Lmx1b (3), the involvement of thesefactors in the development of basal forebrain DA neurons inzebrafish remains to be determined. Taken together, it will be ofgreat interest to test whether Ngn1 regulate these factors and�orother unidentified pathways during DA neuron determination.

The Role of Ngn1 in Inducing DA Neurons. Our gain-of-functionanalysis demonstrates that Ngn1 is capable of inducing super-numerary DA neurons and a DA subtype-related gene innonneural ectodermal cells. These data strengthen the idea thatNgn1 may be involved not only in specifying general neuronalproperty, but also in specifying subtype identity in the DAneuronal lineage. However, a definitive role of Ngn1 in DAneuron subtype specification awaits future analysis assessingwhether Ngn1 can directly regulate DA subtype-related genes.

Although a few ectopic DA neurons are detected in the telen-cephalon, the most striking increase of DA neurons is restrictedwithin its endogenous location of the basal forebrain. In our view,this observation has two implications. First, because more Ngn1leads to more DA neurons in the basal forebrain, it suggests thatNgn1 is a limiting factor in the development of basal forebrain DAneurons. The nature of this limiting effect remains to be under-stood: Ngn1 may be a limiting factor in determining how manyprogenitors are initially specified toward DA lineage, or it may bea limiting factor in determining how many rounds of proliferationcan occur before DA progenitors can exit cell cycle and embark ona differentiation pathway. Second, because DA neuron induction byNgn1 is largely restricted to the basal forebrain, it suggests thatNgn1 requires additional factors that are present in the basalforebrain to determine DA neurons, or its pro-DA activity isinhibited in other regions of the nervous system.

Regulation of Ngn1 by the Conserved Zinc Finger Protein Tof�Fezl.Previous molecular genetic analysis using the too few mutantzebrafish has revealed the importance of Tof�Fezl, a conservedzinc finger protein, in DA neuron specification (5, 8), but itsmechanism of action is unknown. The analyses carried out herereveal that Tof�Fezl controls DA neuron development in part byestablishing ngn1-expressing DA progenitor domains in the basalforebrain. How does Tof�Fezl specify ngn1-expressing progen-itor domains? The fact that Tof�Fezl protein contains six C2H2zinc fingers suggests that it is a putative transcription regulatorand may directly or indirectly regulate ngn1 expression. There-fore, the identification of downstream target genes of Tof�Fezlwill provide crucial insights into understanding its role inspecifying ngn1-expressing progenitor domains. In addition tothe zinc finger domain, Tof�Fezl possesses a Groucho-TLE-likerepressor domain, which is found in transcription factors such asTCF that can serve as both transcriptional activators and re-pressors. Thus, Tof�Fezl may repress a factor that normallyinhibits the formation of ngn1-expressing progenitor domains;alternatively, Tof�Fezl may be involved in promoting progenitordomains by activating the expression of ngn1.

A Homologous Role of Neurogenins in the Development of ZebrafishBasal Forebrain DA and Mouse Mesencephalic DA Neurons. While ourwork was being prepared for publication, it was reported that loss

Jeong et al. PNAS � March 28, 2006 � vol. 103 � no. 13 � 5147

NEU

ROSC

IEN

CE

Dow

nloa

ded

by g

uest

on

Dec

embe

r 15

, 202

0

of neurogenin 2 (ngn2) function in mice impairs the developmentof mesencephalic DA neurons, whereas non-DA neurons in themidbrain are unaffected (29). The mouse study and our workpresented here have two important implications: first, theyindicate that the role of neurogenins in DA neuron developmentis evolutionarily conserved. Second, they show that basal fore-brain DA neurons in zebrafish share common developmentalmechanisms with the mammalian midbrain DA neurons. To-gether with the dye-tracing experiments (10) and a functionalstudy (30), our work suggests that some basal forebrain DAneurons in zebrafish are developmentally and functionally ho-mologous to mammalian midbrain DA neurons.

One difference between the mouse study and our work is that,although ngn2 is not sufficient to induce DA neurons in mice (29),we show that ngn1 is capable of inducing DA neurons in zebrafish.Thus, whereas ngn2 appears to have a permissive role in mousemidbrain DA neuron development, ngn1 may have an instructiverole in basal forebrain DA neuron development in zebrafish. Thesedifferences may be gene- and�or species-dependent.

In conclusion, our study demonstrates an important role ofngn1 in basal forebrain DA neuron development in zebrafish,and moreover, reveals that the establishment of ngn1-expressingDA progenitor domains requires the conserved zinc fingerprotein Tof�Fezl. Future analyses of these transcription regula-tors promise to unravel further mechanisms governing thecommitment and differentiation of multipotent neural stem cellsto a dopaminergic fate.

Materials and MethodsFish Stocks and Maintenance. Fish breeding and maintenance wereperformed as described (31). Embryos were raised at 28.5°C andstaged according to Kimmel et al. (32). Fish heterozygous for thengn1hi1059 and tofm808 mutations were bred to obtain homozygousembryos for analysis: ngn1 homozygous embryos were identifiedby applying the ratio of 25% to a population (�50) of stainedembryos; tof mutant embryos were identified by genotyping forthe missense mutation (5).

Analysis of Neuronal Phenotypes in Transgenic ngn1:GFP Embryos.Twenty-eight-hpf Tg(-8.4ngn1:GFP) embryos were fixed over-night in 4% PFA in 0.1 M phosphate buffer (pH 7.4) and storedin 100% methanol at �20°C. Whole mount immunohistochem-istry was performed as described (33). The following primaryantibodies were used: anti-GFP (Chemicon, monoclonal,1:1,000), anti-TH (Chemicon, rabbit polyclonal, 1:1,000) andanti-5HT (DiaSorin, rabbit polyclonal, 1:4,000). The followingsecondary antibodies were used: anti-rabbit Alexa 633 (Molec-ular Probes, 1:200) and anti-mouse Alexa 488 (MolecularProbes, 1:200). Fluorescent labeling was analyzed by laser scan-ning confocal microscopy.

MO and mRNA Injections and Analysis. The ngn1 MO was synthe-sized and injected as described (22, 23). For misexpressionexperiments, capped RNAs from pCS2-�-gal and pCS2-ngn1plasmids were synthesized and injected at 100–800 ng��l with2–3 nl into the yolk of one- to eight-cell-stage embryos asdescribed (34). For details, see Supporting Text, which is pub-lished as supporting information on the PNAS web site.

In Situ Hybridization and Immunohistochemistry. RNA In situ hy-bridization and immunohistochemistry were performed asdescribed (8).

We thank Drs. Frances Brodsky, Yuh Nung Jan, Bingwei Lu, andJohn Rubenstein for their helpful comments on the manuscript;Drs. Adam Amsterdam and Nancy Hopkins (Massachusetts Institute ofTechnology, Cambridge) for the ngn1 mutant; Uwe Strahle and PatrickBlader (University of Heidelberg, Heidelberg) for ngn1:GFP transgenicfish; and Drs. Say-Yeob Yeo (National Institutes of Health, Bethesda),Ajay Chitnis (National Institutes of Health, Bethesda), and MasahikoHibi (RIKEN, Tokyo) for plasmids. This work was supported by SearleScholars Award, Burroughs Wellcome Fund, and National Institutes ofHealth grants (to S.G.), and by grants from the Wellcome Trust,Biotechnology and Biological Sciences Research Council and EuropeanCommunity (to S.W.W.). The financial support of Telethon–Italy Fel-lowship GFP03011 (to S.M.) is gratefully acknowledged.

1. Goridis, C. & Rohrer, H. (2002) Nat. Rev. Neurosci. 3, 531–541.2. Zetterstrom, R. H., Solomin, L., Jansson, L., Hoffer, B. J., Olson, L. &

Perlmann, T. (1997) Science 276, 248–250.3. Smidt, M. P., Asbreuk, C. H. J., Cox, J. J., Chen, H., Johnson, R. L. & Burbach,

J. P. H. (2000) Nat. Neurosci. 3, 337–341.4. Guo, S., Yamaguchi, Y., Schilbach, S., Wada, T., Goddard, A., Lee, J., French,

D., Handa, H. & Rosenthal, A. (2000) Nature 408, 366–369.5. Levkowitz, G., Zeller, J., Sirotkin, H. I., French, D., Schilbach, S., Hashimoto,

H., Hibi, M., Talbot, W. S. & Rosenthal, A. (2003) Nat. Neurosci. 6, 28–33.6. Vitalis, T., Cases, O., Engelkamp, D., Verney, C. & Price, D. J. (2000)

J. Neurosci. 20, 6501–6516.7. Andrews, G. L., Yun, K., Rubenstein, J. & Mastick, G. S. (2003) Mol. Cell.

Neurosci. 23, 107–120.8. Guo, S., Wilson, S. W., Cooke, S., Chitnis, A. B., Driever, W. & Rosenthal, A.

(1999) Dev. Biol. 208, 473–487.9. Holzschuh, J., Ryu, S., Aberger, F. & Driever, W. (2001) Mech. Dev. 101,

237–243.10. Rink, E. & Wullimann, M. F. (2001) Brain Res. 889, 316–330.11. Kapsimali, M., Bourrat, F. & Vernier, P. (2001) J. Comp. Neurol. 431, 276–292.12. Jan, Y. N. & Jan, L. Y. (1993) Cell 75, 827–830.13. Chitnis, A. B. (1999) Curr. Opin. Neurobiol. 9, 18–25.14. Bertrand, N., Castro, D. S. & Guillemot, F. (2002) Nat. Rev. Neurosci. 3, 517–530.15. Jarman, A. P., Grau, Y., Jan, L. Y. & Jan, Y. N. (1993) Cell 73, 1307–1321.16. Ma, Q., Kintner, C. & Anderson, D. J. (1996) Cell 87, 43–52.17. Sommer, L., Ma, Q. & Anderson, D. J. (1996) Mol. Cell Neurosci. 8, 221–241.18. Ma, Q., Fode, C., Guillemot, F. & Anderson, D. J. (1999) Genes Dev. 13,

1717–1728.19. Ma, Q., Chen, Z., Barrantes, I., Pompa, J. & Anderson, D. J. (1998) Neuron 20,

469–482.20. Fode, C., Gradwohl, G., Morin, X., Dierich, A., LeMeur, M., Goridis, C. &

Guillemot, F. (1998) Neuron 20, 483–494.

21. Blader, P., Fischer, N., Gradwohl, G., Guillemot, F. & Strahle, U. (1997)Development (Cambridge, U.K.) 124, 4557–4569.

22. Andermann, P., Ungos, J. & Raible, D. W. (2002) Dev. Biol. 251, 45–58.23. Cornell, R. A. & Eisen, J. S. (2002) Development (Cambridge, U.K.) 129,

2639–2648.24. Cau, E. & Wilson, S. W. (2003) Development (Cambridge, U.K.) 130, 2455–2466.25. Blader, P., Plessy, C. & Strahle, U. (2003) Mech. Dev. 120, 211–218.26. Halloran, M. C., Sato-Maeda, M., Warren, J. T., Su, F., Lele, Z., Krone, P. H.,

Kuwada, J. Y. & Shoji, W. (2000) Development (Cambridge, U.K.) 127,1953–1960.

27. Amsterdam, A., Nissen, R. M., Sun, Z., Swindell, E. C., Farrington, S. &Hopkins, N. (2004) Proc. Natl. Acad. Sci. USA 101, 12792–12797.

28. Sakurada, K., Ohshima-Sakurada, M., Palmer, T. D. & Gage, F. H. (1999)Development (Cambridge, U.K.) 126, 4017–4026.

29. Andersson, E., Jensen, J. B., Parmar, M., Guillemot, F. & Bjorklund, A. (2006)Development (Cambridge, U.K.) 133, 507–516.

30. Lau, B., Bretaud, S., Huang, Y., Lin, E. & Guo, S. (2005) Genes Brain Behav.,in press.

31. Guo, S., Driever, W. & Rosenthal, A. (1999) Handbook of Molecular-GeneticTechniques for Brain and Behavior Research (Elsevier, Amsterdam), pp. 166–176.

32. Kimmel, C. B., Ballard, W. W., Kimmel, S. R., Ullmann, B. & Schilling, T. F.(1995) Dev. Dyn. 203, 253–310.

33. Concha, M. L., Russell, C., Regan, J. C., Tawk, M., Sidi, S., Gilmour, D. T.,Kapsimali, M., Sumoy, L., Goldstone, K., Amaya, E., et al. (2003) Neuron 39,423–438.

34. Guo, S., Brush, J., Teraoka, H., Goddard, A., Wilson, S. W., Mullins, M. C. &Rosenthal, A. (1999) Neuron 24, 555–566.

35. Andersson, E., Tryggvason, U., Deng, Q., Friling, S., Alekseenko, Z., Robert,B., Perlmann, T. & Ericson, J. (2006) Cell 124, 393–405.

5148 � www.pnas.org�cgi�doi�10.1073�pnas.0600337103 Jeong et al.

Dow

nloa

ded

by g

uest

on

Dec

embe

r 15

, 202

0