Regulation of Neurogenesisby Interkinetic Nuclear Migrationthrough an Apical-Basal Notch GradientFilippo Del Bene,1 Ann M. Wehman,1 Brian A. Link,2,* and Herwig Baier1,*1Department of Physiology, Programs in Neuroscience, Genetics, and Developmental Biology, University of California,

San Francisco, 1550 4th Street, San Francisco, CA 94158-2722, USA2Department of Cell Biology, Neurobiology, and Anatomy, Medical College of Wisconsin, 8701 Watertown Plank Road, Milwaukee,

WI 53226, USA*Correspondence: blink@mcw.edu (B.A.L.), herwig.baier@ucsf.edu (H.B.)

DOI 10.1016/j.cell.2008.07.017

SUMMARY

The different cell types in the central nervous systemdevelop from a common pool of progenitor cells. Thenuclei of progenitors move between the apical andbasal surfaces of the neuroepithelium in phase withtheir cell cycle, a process termed interkinetic nuclearmigration (INM). In the retina of zebrafish mikre oko(mok) mutants, in which the motor protein Dynac-tin-1 is disrupted, interkinetic nuclei migrate morerapidly and deeply to the basal side and more slowlyto the apical side. We found that Notch signaling ispredominantly activated on the apical side in bothmutants and wild-type. Mutant progenitors are,thus, less exposed to Notch and exit the cell cycleprematurely. This leads to an overproduction ofearly-born retinal ganglion cells (RGCs) at the ex-pense of later-born interneurons and glia. Our dataindicate that the function of INM is to balance theexposure of progenitor nuclei to neurogenic versusproliferative signals.

INTRODUCTION

The enormous neuronal diversity present in the vertebrate cen-

tral nervous system (CNS) arises from an apparently homoge-

neous progenitor population (Edlund and Jessell, 1999; Pearson

and Doe, 2004). The mechanisms that generate the correct num-

ber and relative proportion of each cell type are incompletely un-

derstood. Which factors influence, at any given time, how many

progenitors are undergoing a final neurogenic mitosis, producing

two neurons; how many are dividing asymmetrically, producing

one neuron and one progenitor; and how many are staying in

a proliferative state, producing two new progenitors?

Neurons arise during development from a pseudostratified co-

lumnar epithelium composed of mitotically active neuroepithelial

cells. These cells have an elongated shape, with cytoplasmic

connections to both the apical (ventricular) and basal surfaces.

Their nuclei occupy different levels within the epithelium depend-

ing on the phase of the cell cycle. Mitotic (M phase) nuclei are

located in close proximity to the apical surface, while nuclei un-

dergoing DNA synthesis (S phase) are displaced more basally.

This characteristic movement of nuclei within CNS neuroepithe-

lia was first predicted more than 70 years ago based on histolog-

ical observations (Sauer, 1935) and was termed interkinetic

nuclear migration (INM) (reviewed by Baye and Link, 2008). Re-

cently, the dynamics of INM have been shown to correlate with

neurogenic cell divisions within the retina, such that neuroepithe-

lia with more basal nuclear movements are biased to generate

postmitotic daughters (Baye and Link, 2007a). The precise

developmental function of INM, however, has remained elusive.

The retina consists of six neuronal and one glial cell types,

which differentiate in a stereotyped yet overlapping birth order

(Cepko et al., 1996; Marquardt and Gruss, 2002; Poggi et al.,

2005b). An extensive number of cell lineage studies have de-

scribed that retinal progenitors are multipotent, giving rise to all

of the major neuronal and glial cell types (Holt et al., 1988; Turner

and Cepko, 1987; Turner et al., 1990; Wetts and Fraser, 1988).

Together, these observations have led to the proposal that reti-

nal progenitors transit through a series of competence states

in a fixed order, and, during each one, they are intrinsically

able to generate one or a small subset of cell types (Livesey

and Cepko, 2001).

In Drosophila, Notch signaling, through lateral inhibition, main-

tains neighboring cells in a multipotent, proliferative state,

whereas downregulation of Notch is the prerequisite of neuronal

differentiation (Chitnis, 1995). This mechanism appears to be

largely conserved in the vertebrate retina (Jadhav et al., 2006;

Nelson et al., 2006; Perron and Harris, 2000). Thus, as the pro-

genitors produce the various cell types according to their com-

petence state, Notch ensures at each step that a subset of the

progenitors is retained for consecutive waves of neurogenesis.

A missing piece in the puzzle is the mechanism that controls

the dosage of Notch signaling such that a fraction of progenitors

exit the cell cycle and generate the appropriate number of each

cell type.

Here, we report a zebrafish mutant in which retinal progenitors

exit the cell cycle prematurely and retinal neurogenesis is accel-

erated. The progenitor population is quickly depleted and gives

rise to an imbalanced ratio of early cell fates versus late ones,

Cell 134, 1055–1065, September 19, 2008 ª2008 Elsevier Inc. 1055

reminiscent of Notch loss of function or a neurogenic gain of

function. To our initial surprise, we discovered that it is caused

by disruption of the motor protein Dynactin-1. We show that

this mutation leads to defects in specific aspects of INM without

affecting the intrinsic competence of the progenitors or their re-

sponsiveness to extrinsic cues. Furthermore, we demonstrate

the existence of a Notch gradient within the neuroepithelium,

with high levels in the apical domain. These results suggest

that INM regulates the duration and level of exposure of progen-

itor nuclei to neurogenic signals.

RESULTS

moks309 Mutant Retinas Have an Excess of RetinalGanglion CellsWe identified a new mutation in the zebrafish mikre oko (mok)

locus (Doerre and Malicki, 2001; Wehman et al., 2005). The

moks309 mutant (formerly named bugs309) has smaller eyes and

a protruding lens and is first recognizable at 4 days postfertiliza-

tion (4 dpf; Figures 1A and 1B). As reported for other mok mutant

alleles, photoreceptors are absent at 5 dpf (Figures 1C and 1D)

(Doerre and Malicki, 2001; Tsujikawa et al., 2007). Analysis of

embryonic stages revealed that photoreceptors are initially

produced in normal numbers but subsequently die by apoptosis

between 2.5 and 3 dpf (Figures 1E, 1F, and S1 available online).

In addition, we noticed a marked increase in the number of cells

located in the ganglion cell layer (GCL) in moks309 at 5 dpf (290 ±

11 cells/section, compared to 204 ± 11 in wild-type; n = 4 larvae

each; one central section per larva analyzed; p < 0.005; Figures

1C and 1D). These extra cells appear to be differentiated retinal

ganglion cells (RGCs), as they express the RGC-specific marker

Zn5 (neurolin/DM-GRASP) and form an optic nerve.

To test the possibility that the excess production of RGCs

might be related to the loss of photoreceptors, we crossed the

moks309 mutant to the Tg(Atoh7:GFP) transgenic line, in which

atoh7 regulatory sequence drives a green fluorescent protein

(GFP) reporter in RGC precursors and early differentiated RGC.

The atoh7 (ath5) gene encodes a basic helix-loop-helix transcrip-

tion factor and acts as a proneural gene for RGC fate (Kay et al.,

2001; Masai et al., 2000). We analyzed moks309 retinas at 48 hr

postfertilization (hpf), before the onset of photoreceptor differen-

tiation and when mutant embryos are morphologically undistin-

guishable from wild-type siblings. Already at this early stage of

retinogenesis, moks309 mutants show an increased number of

RGCs, as identified by their nuclear position in the innermost

layer and the expression of GFP (Figures 1G and 1H; 108.6 ±

3.8 cells/section in mutant retinas, compared to 82.4 ± 3.1 in

wild-type; n = 5 larvae each; p < 0.001). It is, therefore, unlikely

that RGC overproduction is a response to a defect in photore-

ceptor survival.

RGCs are the first neurons born in the retina between 26 and

36 hpf. RGC precursors express atoh7 for a short time before

their last mitosis (Poggi et al., 2005a) and downregulate it as

they differentiate. RNA in situ hybridization revealed that initia-

tion of atoh7 is unaffected in mok mutants. However, atoh7 ex-

pression is retained in the mutant progenitor population after

36 hpf, when RGC production normally subsides (Figure S2)

1056 Cell 134, 1055–1065, September 19, 2008 ª2008 Elsevier Inc

(Masai et al., 2000). Thus, the wave of neurogenesis that pro-

duces RGCs is extended in the mok mutant.

In wild-type retina, RGCs are located exclusively in the

GCL, as evidenced by their expression of the POU domain

Figure 1. moks309 Mutants Have a Complex Retina Phenotype

(A and B) Whole-mount lateral views of live 4.5 dpf zebrafish larvae. In moks309

mutants (B), the eyes are smaller than in the wild-type (A), but external

morphology is otherwise indistinguishable.

(C and D) Horizontal sections of 5 dpf retinas stained with DAPI (blue) and Zpr1

antibody (green). moks309 retinas (D) have an expanded GCL (dashed white

lines), compared to wild-type (C), and no photoreceptors, as revealed by

Zpr1 staining.

(E and F) Coronal sections of 3 dpf retinas stained by TUNEL assay. moks309

retinas (F) have increased apoptosis, particularly in the photoreceptor layer

(asterisks), compared to wild-type (E).

(G and H) Horizontal sections of 48 hpf retinas expressing GFP driven by the

atoh7 promoter (green) and stained with DAPI (blue) and HuC/D antibody

(red). moks309 retinas (H) have an increased number of GFP-expressing

RGCs compared to wild-type (G).

Scale bars, 500 mm (A and B) and 100 mm (C–H).

.

transcription factors Pou4f2 (Brn3b) and Pou4f3 (Brn3c) (Xiao

et al., 2005). In mok mutants, however, some of the extra

RGCs are found in the inner nuclear layer (INL) (Figures 2B, 2F,

and S3). In Tg(atoh7:GFP) fish, mature RGCs, as well as cells

that were competent earlier to produce RGCs (but followed a

different cell fate), remain fluorescently labeled for several days

after the atoh7 promoter is turned off, due to perdurance of

GFP (Masai et al., 2003). In wild-type retina, GFP-labeled cells

accumulate predominantly in the GCL. In moks309 mutants, by

contrast, cells in the enlarged GCL, as well as cells located out-

side the GCL (Figures 2A and 2E), are strongly GFP labeled. This

finding indicates that a greater number of progenitors in mok

mutants become competent to produce RGCs.

Bipolar and Muller Glia Cells Are Absent or SeverelyReduced in moks309 RetinasIn the retina, all neuronal cell types and Muller glia cells arise from

a common progenitor pool (Holt et al., 1988; Turner and Cepko,

1987; Wetts and Fraser, 1988). Therefore, we speculated that

overproduction of RGCs could lead to depletion of the progenitor

pool available for the genesis of later-born cells. To test this

hypothesis, we investigated the presence of markers for other

cell types. Muller glia are absent or strongly reduced in number

in mutant retinas at 5 dpf, as revealed by immunohistochemical

staining for glutamine synthetase (GS) (Figures 2C and 2G). The

Figure 2. moks309 Retinas Have Increased

RGCs and Decreased Bipolar and Muller

Glia Cells Due to Premature Neurogenesis

(A–H) Horizontal 5 dpf retina sections. GFP ex-

pression under the control of atoh7 (A and E) and

Brn3c (Pou4f3) (B and F) promoters reveals an in-

crease in RGCs in moks309 (E and F) as compared

to the wild-type (A and B), with some RGCs ectop-

ically located outside the GCL (yellow arrowheads

in [E] and [F]). The Muller glia marker GS (C and G)

and the bipolar cell marker PKCa (D and H) show

fewer immunoreactive cell bodies (yellow arrow-

heads) in moks309 (G and H) as compared to

wild-type (C and D).

(I–P) Horizontal 50 hpf retina sections stained for

IdU (green, injected at 26 hpf), BrdU (red, injected

at 38 hpf), and DAPI (blue). moks309 retinas (M–P)

show more IdU-positive and BrdU-negative cells

than wild-type, indicating that a larger number of

progenitors had exited the cell cycle between 26

and 38 hpf.

Scale bars, 100 mm.

same pattern was observed for bipolar

cells, as shown by decreased PKCa (Fig-

ures 2D and 2H). We confirmed this result

by in situ hybridization for vsx1, which

encodes a transcription factor marking

bipolar cells at early stages of differentia-

tion (Passini et al., 1997). vsx1 RNA is

decreased in moks309 mutant retinas (Fig-

ure S3). In contrast, no differences were

observed in numbers of horizontal or

amacrine cells, based on expression ofGAD65/67, parvalbumin, and an amacrine-specific transgenic

reporter (Kay et al., 2001) (Figure S4). As reported above, photo-

receptors are eliminated by apoptosis in the moks309 mutants,

but their specification and number seem unaffected at earlier

stages (Tsujikawa et al., 2007). Together, these results show

that the moks309 mutation causes a severe depletion of bipolar

cells and Muller glia.

A Greater Number of Progenitors Exit the CellCycle in mok MutantsFactors that delay cell-cycle exit in the retina give rise to an ex-

cess of later-born cell types, while prematurely forcing progeni-

tor cells to become postmitotic increases the generation of

RGCs (Ohnuma et al., 2002). Therefore, we suspected that the

cell-fate switch observed in moks309 would be accompanied by

abnormalities in the timing of cell-cycle exit. To confirm this,

we performed double-labeling experiments with two thymidine

analogs, IdU and BrdU. We allowed IdU and BrdU incorporation

for periods longer than a full cell cycle (12 hr) in order to label the

total population of proliferative cells. We injected IdU into the

developing embryos at 26 hpf, followed by BrdU at 38 hpf, and

finally fixed the embryos at 50 hpf. Cells that are positively

labeled for IdU, but not BrdU, had undergone the last S phase

between 26 and 38 hpf, when RGCs are generated in wild-

type. moks309 retinas show an �25% increase in IdU-positive

Cell 134, 1055–1065, September 19, 2008 ª2008 Elsevier Inc. 1057

and BrdU-negative cells compared to wild-type retinas (Figures

2I–2P; 33.5 ± 1.9 cells/section in mutant, compared to 25.2 ± 1.1

in wild-type; n > 8 larvae each; p < 0.005). This result was con-

firmed by similar double-labeling experiments using BrdU, in-

jected at 28 hpf, and the mitotically active cell marker PCNA (pro-

liferating cell nuclear antigen). Embryos were fixed at 40 hpf and

stained for PCNA (Figure S5). Taken together, our data demon-

strate that more cells leave the cell cycle during the first wave

of neurogenesis in moks309, thus biasing the neurons to adopt

the RGC fate.

To investigate the cell autonomy of the mutation, we analyzed

mutant/wild-type chimeras following blastomere transplanta-

tions at the 1000 cell stage. If mok acted nonautonomously,

then mutant clones in a wild-type environment should generate

retinal cell types in normal proportions. Conversely, if mok func-

tioned intrinsically in progenitor cells to promote neurogenesis,

then mutant clones should produce an excess of RGCs and

fewer INL cells regardless of their genetic environment. Clonal

analysis showed that the latter possibility is correct. While

wild-type clones in a wild-type retina give rise to neurons located

in the GCL, INL, and ONL in a 3:5:2 ratio, respectively, mutant

clones produce these cells in a 5:3:2 ratio (Figures 3A, 3B, and

3E; n = 1447 wild-type and 649 mutant cells counted in total).

We conclude that mok acts cell autonomously.

We further observed that wild-type clones in moks309 mutant

retinas tend to be excluded from the GCL and produce fewer

RGCs (2:6:2 ratio; n = 786 cells counted; Figures 3C and 3E)

than when they develop in a wild-type environment (2:6:2 ratio

for mutant cells in wild-type versus 3:5:2 for wild-type cells in

wild-type). Mutant clones in moks309 retinas show an increase

of RGCs (4:5:1 ratio; n = 223 cells counted; Figures 3D and 3E)

compared to wild-type clones in wild-type. These results sug-

gest that cell-fate switches of genotypically mutant progenitors

are even more dramatically biased toward the RGC fate when

they are in a wild-type environment than when they are sur-

rounded by other mutant cells (5:3:2 versus 4:5:1). This is prob-

ably due to intercellular feedback regulation within the develop-

ing retina. For instance, newborn RGCs limit the production of

additional RGCs by secreting the GDF11, which suppresses ex-

pression of atoh7 in uncommitted progenitors (Kim et al., 2005).

Consistent with this view, wild-type progenitors are more likely to

generate RGCs when transplanted into an environment in which

RGCs are absent, as in lakritz zebrafish mutants, which carry

a null mutation in atoh7 (Poggi et al., 2005a). Our transplantation

data suggest that such negative feedback attenuates the fate

switch in moks309 retinas.

The mok Gene Encodes Dynactin-1To gain insight into the molecular basis of the cell-fate regulation

defect, we positionally cloned the mok gene. The moks309 phe-

notype is perfectly linked to a nonsense point mutation in Dynac-

tin-1 (Dnct1, p150Glued), introducing a stop codon at amino acid

867 and a complete deletion of the C-terminal third of this pro-

tein. This region contains important protein-protein interaction

domains responsible for the binding of Dnct1 to other dynactin

subunits like Arp1 and p25 (Schroer, 2004). In Drosophila, the

glued allele that carries a similar truncation in the dnct1 ortholog

is not incorporated in the dynactin complex (McGrail et al., 1995).

1058 Cell 134, 1055–1065, September 19, 2008 ª2008 Elsevier Inc.

Western blot of whole-embryo extracts showed that Dnct1

protein is weakly detected in both mutants and wild-type imme-

diately after fertilization and for the first 3 days of development,

indicating that it is maternally supplied (Figure 4B). In moks309

mutants, Dnct1 protein is no longer detectable at 4 dpf (Fig-

ure 4A). Consistent with these data, the expression of dnct1

Figure 3. Cell Transplantation Analysis Reveals that mok Acts Cell

Autonomously

(A–D) Representative sections of 5 dpf chimeric retinas. The transplanted cells

are clearly identified by the expression of H2A-GFP marker (green) in their

nuclei. Cell transplantation shows that moks309 mutant clones in wild-type

host retinas (B) have a higher propensity to generate neurons located in the

GCL compared to control (A). Conversely, wild-type clones in moks309 host ret-

inas preferentially generate INL neurons (C). moks309 mutant clones in moks309

host retinas are shown for comparison (D). Dashed lines indicate the outer limit

of the GCL. Scale bars, 100 mm.

(E) Quantification of the transplantation results showing the distribution of

clones in the three retinal nuclear layers. ***p < 0.001; *p < 0.01. Error bar

indicates SEM.

RNA is ubiquitous and weak at early stages (data not shown) and

is enriched at 3 dpf in most of the larval CNS, including the GCL

and part of the INL (Figures 4C and 4D). To demonstrate that

dnct1 is the gene affected in the moks309 mutation, we designed

a splicing morpholino oligonucleotide (MO) to disrupt dnct1

Figure 4. Expression of Dnct1 and Its Function in INM

(A) Western blotting of extracts from 4 dpf embryos shows that Dnct1 is unde-

tectable in moks309.

(B) Time course analysis of Dnct1 expression by western blot in wild-type

zebrafish (numbers on top indicate hours after fertilization).

(C and D) Whole-mount in situ hybridization shows dnct1 enriched in the head

and eye region (C) and in the notochord (D). Scale bars, 100 mm.

(E and F) Coronal sections of 2 dpf retinas stained for the mitotic marker PH3.

In moks309 (F), a number of mitotic cells are sparsely located throughout the

retina, while, in the wild-type (E), they are confined within 2–3 cell diameters

from the ventricular surface. Dashed lines demarcate the apical (right) and

basal (left) domains. Asterisks indicate mitotic cells residing in the developing

lens.

(G) Scatter plot of wild-type (gray triangles) and moks309 mutant (black

squares) circles, showing the maximum basal distance of nuclei during INM

from 30–48 hpf.

(H) Histogram showing that these populations are statistically different

(p = 0.001, Wilcoxon two-sample test). Error bar indicates SEM.

C

function in wild-type embryos (Draper et al., 2001). MO-injected

embryos phenocopied moks309 mutants, including thickening of

the GCL and absence of differentiated photoreceptors (Fig-

ure S6). In conclusion, we predict that moks309 is a null or strong

hypomorph of dnct1.

INM Is Perturbed in moks309 Mutant EmbryosDynactin mediates the interaction of the dynein motor with many,

if not all, of its cargoes and allows the motor to traverse the mi-

crotubule lattice over long distances by increasing its processiv-

ity (King and Schroer, 2000). Loss-of-function analysis during

Drosophila eye development revealed that dynactin is required

for correct nuclear migration and maintenance of nuclear posi-

tion within postmitotic photoreceptors (Fan and Ready, 1997;

Whited et al., 2004). A similar nuclear positioning defect is ob-

served in zebrafish photoreceptor cells that carry a different

mok mutation (Tsujikawa et al., 2007). Therefore, we speculated

that, in moks309 mutants, the INM of neuroepithelial cells could

be altered and that this could, in turn, be responsible for the

effects on cell-cycle exit. To test this hypothesis, we analyzed

the positions of neuroepithelial nuclei during late G2/M phase us-

ing phosphorylated histone H3 (PH3) as a marker. PH3-positive

nuclei in wild-type are located close to the apical (ventricular)

surface of the neural retina at all time points analyzed (24, 36,

and 48 hpf; Figure 4E and data not shown; 47/47 cells in two ret-

inas at 48 hpf). In contrast, about 40% of PH3-positive nuclei are

positioned ectopically toward the basal side in moks309 mutants

(Figure 4F; 25/63 cells in three retinas at 48 hpf).

Time-lapse analysis of single-cell INM revealed a significant

increase in the maximal-basal position and a faster than normal

apical-to-basal movement of the nuclei in mutant cells (Movies

S1 and S2; Figures 4G and 4H; Table S1). Similar results were

obtained in dnct1-MO-injected embryos. The basal-to-apical

migration velocity was also reduced in the absence of dnct1 func-

tion, although the effect in this direction was less pronounced.

The net result is that interkinetic nuclei in mok mutants migrate

faster to the basal surface and further basally, take longer to re-

turn, and often enter mitosis before they have reached the apical

domain. In the wild-type retina, the depth of INM is an approxi-

mate predictor of a neurogenic cell division upon return to the

apical side (Baye and Link, 2007a). Therefore, we hypothesized

that this net change in INM promotes production of neurons.

An alternative hypothesis posits that cell polarity may be dis-

rupted in moks309 mutants. Indeed, previous work established

that intrinsic cell polarity is essential for the relationship between

nuclear position and neurogenesis (Baye and Link, 2007a). How-

ever, analysis of a series of apical and basal markers (laminin,

ZO-1, aPKCz) showed a correct localization in the mutant ret-

inas. We also investigated the distribution of a GFP-tagged

Par3 protein (Pard3), which partially overlaps with the adherens

junction-associated actin bundles (Wei et al., 2004). Apical local-

ization of Pard3-GFP was indistinguishable from wild-type

(Figure S7). These findings are consistent with a recent study

showing that polarity of photoreceptors is intact in mok mutants

(Tsujikawa et al., 2007). Cumulatively, these observations do not

support the interpretation that apical-basal polarity defects

underlie the displaced mitosis and neurogenic phenotypes in

mok retina.

ell 134, 1055–1065, September 19, 2008 ª2008 Elsevier Inc. 1059

A Notch Gradient Creates Distinct Signaling Milieusalong the Apical-Basal AxisNotch activation is known to delay neurogenesis (Furukawa

et al., 2000; Gaiano et al., 2000; Morrison et al., 2000; Scheer

et al., 2001). We observed that notch1a RNA is enriched in the

apical domain of the neuroepithelium in zebrafish (Figure 5A),

extending earlier reports in chick (Murciano et al., 2002). In addi-

tion, the Notch receptor ligands DeltaB and DeltaC are mostly

found in the basal half of the developing neuroretina (Figures

5B and 5C). Reflecting the localization of its transcript, DeltaC

protein is restricted to the basal half of the developing neurore-

tina both in wild-type and mutant embryos (Figures 5E and 5F).

Thus, Notch receptor and Delta ligands appear to be preferen-

tially expressed on opposite sides of the neuroepithelium.

We asked whether the Notch localization gradient leads to an

apical-to-basal gradient in Notch signaling. Upon binding of

ligand, Notch receptor undergoes a proteolytic cleavage. The cy-

tosolic fragment (Notch intracellular domain [NICD]) translocates

to the nucleus, where it modulates the transcription of target

genes. A specific anti-NICD antibody has been used in mouse

to detect the pattern of Notch activation in the developing brain

(Tokunaga et al., 2004). Using this antibody in mouse retina sec-

tions, we detected Notch activation only in nuclei located at the

apical surface, consistent with data in zebrafish (Figure 5J).

her4 is one target gene of the Notch/Delta pathway that is upre-

gulated by NICD and involved in neurogenesis (Takke et al.,

1999). Using the Tg(her4.1:dRFP) transgenic line, in which

a short-lived form of red fluorescent protein (dRFP) is expressed

under the control of the her4 promoter (Yeo et al., 2007), we mon-

itored her4 expression in vivo by time-lapse analysis (Movie S3;

Figures 5D and S8). In 13/15 cells moving from basal to apical,

dRFP fluorescence increased during the 2 hr observation period,

while the fluorescence decreased in 11/15 cells moving apical to

basal. This suggests that Notch signaling is activated as the

nuclei move into the apical compartment and downregulated in

nuclei that move in the opposite direction.

Dynactin-Driven INM Serves to Expose ProgenitorNuclei to the Notch GradientThe apical distribution of Notch, together with the dynactin-

driven, basally directed migration of interkinetic nuclei, offers

a mechanism by which the rate of neurogenesis could be regu-

lated. Consistent with our model, the microtubule minus ends,

marked by centrioles, are located at the apical surface (Figures

5G, 5H, and 5K), and microtubules are primarily oriented parallel

to the AB axis (Figure 5I). Using anti-Dnct1 antibodies, we ob-

served cytoplasmic punctate staining with enrichment at the

apical surface both in zebrafish wild-type and mouse retinal neu-

roepithelia, but not in moks309 mutants (Figures 5G, 5H, and 5K).

BBS4 protein localization is dependent on dynein motor function

and disrupted in moks309 mutants (Tsujikawa et al., 2007) (Fig-

ures 5L and S9). Thus, Dnct1 likely exerts its influence on neuro-

genesis by controlling INM within this graded signaling environ-

ment. Nuclei moving further basally appear to downregulate

Notch, while nuclei that remain closer to the ventricular surface

during INM retain high Notch activity levels.

An alternative way by which microtubule-associated motors

could affect Notch/Delta signaling in the neuroepithelial cells is

Figure 5. A Gradient of Notch Signaling

along the Apical-Basal Axis of the Develop-

ing Retina

(A–C) Coronal sections of 26 hpf retinas showing

mRNA expression levels of components of the

Notch/Delta signaling pathway. In situ hybridiza-

tion shows higher levels of notch1a close to the

apical surface of the retina (A) and deltaB and

deltaC close to the basal surface (B and C).

(D) Optical section of a 33 hpf retina expressing

her4:dRFP and H2A-GFP transgenes in a mosaic

manner.

(E and F) Coronal sections of 26 hpf retinas stained

with anti-DeltaC antibody, showing punctate cyto-

plasmic staining distributed in the basal half of the

tissue both in wild-type and moks309 retinas.

(G and H) Coronal sections of 26 hpf retinas

stained with anti-Dnct1 antibody, showing an en-

richment at the apical surface in wild-type retinas

(G), which is virtually absent in mutants (H).

(I) Coronal section of 26 hpf retina; a-tubulin stain-

ing reveals the parallel orientation of microtubules

to the apical-basal axis.

(J–L) Sections of mouse retina. Activated Notch1

antibody labels a subset of nuclei at the apical sur-

face (J). Anti-Dnct1 and anti-BBS4 antibodies

show a cytoplasmic, punctated staining enriched

at the apical surface. g-tubulin staining (G, H, K, L)

reveals the apical localization of the centrioles in

retinal progenitors. In (D)–(L), DAPI (blue) stains

the nuclei. In all panels, apical surface is on the left.

Scale bars, 25 mm (A–I) and 50 mm (J–L).

1060 Cell 134, 1055–1065, September 19, 2008 ª2008 Elsevier Inc.

Figure 6. Atoh7 Loss of Function and Notch

Activation Can Rescue Late Cell Fates in

moks309

(A–D) Compared to wild-type (A), mok mutants (B)

have an excess of RGCs and few bipolar cells,

labeled by PKCa immunoreactivity. lak (atoh7)

mutants fail to generate RGCs and overproduce

bipolar neurons (C). lak/mok double mutants

have rescued bipolar cell production.

(E–H) Heat-shocked induction of an activated form

of Notch receptor NICD at 32 hpf drives progeni-

tors toward a Muller glia fate (GS positive) in

both wild-type (G) and moks309 retinas (H). The

presence of the transgenes (hsp:Gal4, UAS:NICD)

has no effect on Muller glia differentiation in the

absence of heat shock (E and F).

Scale bars, 100 mm.

by altering the distribution and, therefore, the function of the

endocytic compartment. In fact, several studies have linked

endosomal sorting and endocytosis to the regulation of Notch

signaling (Le Borgne, 2006; Nichols et al., 2007). In particular,

both Delta activity and Notch proteolytic activation require endo-

cytic internalization and recycling. The small GTPase Rab11 reg-

ulates formation of recycling endosomes, through which Delta

passes, and unequal distribution of this protein in daughter cells

modulates asymmetric cell divisions in Drosophila sensory organ

formation (Emery et al., 2005). Numb is another endocytic pro-

tein that regulates Notch activity through direct binding and by

targeting the receptor for endocytosis (Berdnik et al., 2002;

Guo et al., 1996; McGill and McGlade, 2003). Therefore, we

investigated whether loss of Dnct1 activity could alter the distri-

bution of GFP-tagged versions of both Numb and Rab11. We did

not detect any alteration of these endocytic components in

dnct1-MO-injected embryos compared to control embryos

(Figure S10). This result indicates that Dnct1 influences Notch

signaling primarily by regulating INM in progenitor cells.

Disruption of atoh7 or Activation of the Notch PathwayRescue Later Cell Fates in mok MutantsTo further test the hypothesis that INM is crucially involved in reg-

ulating cell-cycle exit and cell-fate determination, we examined

the moks309 phenotype in different genetic backgrounds. First,

we crossed moks309 with lakritz (lakth241) (Kay et al., 2001). The

lakth241 mutation disrupts Atoh7. In its absence, neuronal pro-

genitors stay longer in a proliferative state and give rise to an

excessive number of later-born neurons, including bipolar cells

(Kay et al., 2001). We hypothesized that, if the primary reason

for the cell-fate changes observed in moks309 is premature neu-

rogenesis, then keeping retinal progenitor cells proliferative for

a longer time, as in lakth241, would rescue late cell fates. Indeed,

moks309; lakth241 double mutants have an increased number of

bipolar cells compared to moks309 mutants (Figures 6A–6D).

Therefore, inhibition of early neurogenesis can rescue late cell

fates in moks309 mutants.

Second, we overexpressed an activated form of the Notch re-

ceptor (NICD) in a temporally controlled manner using a heat-

shocked promoter coupled to the Gal4/UAS system. If reduced

exposure of retinal neuroepithelial cells to Notch causes prema-

ture cell-cycle exit in moks309, then uniformly activating Notch

should overcome the effect caused by altered INM in moks309

mutants. As previously reported, NICD overexpression between

27 and 42 hpf forces central retina cells to adopt a Muller glia fate

(Scheer et al., 2001). We found that moks309 retinas respond sim-

ilarly to NICD overexpression, as indicated by staining for the

glial marker GS, which was otherwise essentially absent in

moks309 (Figures 6E–6H). This result shows that increased Notch

signaling promotes gliogenesis both in wild-type and moks309

mutants. Taken together, these data demonstrate that mutant

progenitor cells retain their competence to generate later-born

neuronal types and glia.

Nuclear Envelope Components of the KASH DomainFamily Are Required for Normal INM and RegulateNeurogenesisINM requires factors that anchor the nucleus to motor proteins.

Good candidates for this role are the KASH domain protein

family members. These proteins are localized to the outer

nuclear envelope and mediate the interaction between the

nucleus and the cytoskeleton acting in nuclear positioning in

a variety of eukaryotic cells from fungi to mammals (Starr and

Fischer, 2005). Therefore, we hypothesized that knockdown of

KASH proteins would disrupt the coupling of the nucleus to

the dynein motor, perturbing INM and affecting neurogenesis

similar to mok.

To test this hypothesis, we targeted the KASH-containing

protein Syne2a. Interfering with Syne2a function at late stages

of retina development results in photoreceptor nuclear displace-

ment (Tsujikawa et al., 2007). First, we globally reduced the levels

of syne2a using a splice-blocking MO. While high doses of

syne2a-MO resulted in grossly malformed embryos (79/85

larvae), low doses had milder effects that resulted in larvae with

only slightly smaller eyes with normal lamination by 5 dpf (67/98

larvae). We detected, nevertheless, a dramatic reduction in the

number of GS-positive Muller glia cells (37.7 ± 1.2 cells/section

in control retinas; 21.2 ± 1.2 cells/section in syne2a-MO-injected

Cell 134, 1055–1065, September 19, 2008 ª2008 Elsevier Inc. 1061

retinas; n = 4 in each group; p < 0.001; Figures 7A and 7B), similar

to the phenotype observed in moks309 retinas.

To separate the global effects of syne2a knockdown on embry-

onic development from specific effects in the developing retina,

we overexpressed a dominant-negative, GFP-tagged Syne2a

KASH domain in a mosaic fashion under control of a heat-

shocked promoter (Tsujikawa et al., 2007). Plasmid DNA was in-

jected at the 2–4 cell stage, and heat shock was applied at 23 hpf.

Figure 7. Disruption of the Nuclear Anchor to Dynactin Phenocopies

the moks309 Mutation

(A and B) Expression of the Muller glia marker GS is reduced in the retina of

larvae injected with a syne2a-MO (B) compared to control MO-injected larvae

(A). Scale bars, 100 mm.

(C and D) Representative examples of sections of 5 dpf retinas overexpressing

a control vector (C) or a dominant-negative Syne2a (KASH, [D]) under the con-

trol of a heat-shock promoter. The cells that express the constructs are iden-

tified by the expression of GFP marker (green). Clones of cells that express the

syne2a dominant-negative construct preferentially generate GCL neurons.

Scale bars, 50 mm.

(E) Quantification of the KASH overexpression results, showing the distribution

of clones in the three retinal nuclear layers. *p < 0.01, **p < 0.005. Error bar

indicates SEM.

(F) A model of the mechanism that couples INM with graded Notch activation.

1062 Cell 134, 1055–1065, September 19, 2008 ª2008 Elsevier Inc

KASH-expressing clones preferentially generated RGC neurons

at the expense of INL and ONL cells (Figures 7C–7E). These re-

sults are not the consequence of reduced INL or ONL cell survival,

because similar overexpression experiments after 48 hpf did not

cause immediate cell loss (Tsujikawa et al., 2007). These results

reinforce the notion that interfering with INM, either through the

motor protein complex or the nuclear anchor, perturbs cell-cycle

exit and neurogenesis.

DISCUSSION

We show here that a mutation in Dnct1, a microtubule-motor-as-

sociated protein, perturbs INM in a selective fashion and results

in an overproduction of early-born neurons in the zebrafish ret-

ina. The progenitor nuclei move more quickly and deeply in the

basal direction and more slowly apically. We further established

that antineurogenic Notch signals are enriched on the apical side

of the neuroepithelium in both mutants and wild-type. Combined

with previous observations that progenitors whose nuclei mi-

grate deep are more likely to produce postmitotic neuronal

daughters following their return to the apical side (Baye and

Link, 2007a), our studies suggest a mechanism by which INM

cooperates with an apical-basal Notch gradient to select pro-

genitors for cell-cycle exit and apportion cell fates (Figure 7F).

The phenotype of the mok mutant is reminiscent of manipula-

tions of cell-fate determinants, such as overexpression of bHLH

proneural genes. Therefore, we asked whether the mok mutation

altered the intrinsic competence of progenitors. The bHLH tran-

scription factor Atoh7 functions as a proneural factor to set neu-

rogenic competency of early retinal cell types and is essential for

RGC fate. Disruption of atoh7 eliminates RGCs and increases the

number of progenitors that remain in the cell cycle (Brown et al.,

2001; Kay et al., 2001; Wang et al., 2001). As a consequence of

this enlarged progenitor pool, bipolar cells and glia are increased

in number in the zebrafish lak (atoh7) mutant. Conversely, over-

expression of atoh7 results in an excess of RGCs, resembling

the mok phenotype (Hatakeyama and Kageyama, 2004; Vetter

and Brown, 2001; Yan et al., 2005). In lak; mok double mutants,

glia and bipolar neurons, which are severely reduced in mok

single mutants, develop similarly to lak single mutants, demon-

strating that mok mutant progenitors retain their potential to

generate later-born neurons. Photoreceptors, which develop

normally in lak single mutants, however, are absent in the double

mutants, as they are in mok single mutants, due to a related but

independent function of Dnct1 in photoreceptor nuclear posi-

tioning (Tsujikawa et al., 2007). In conclusion, Dnct1 does not

affect the neurogenic competence of retinal progenitor cells.

Inhibition of Notch (Austin et al., 1995) or disruption of GDF11

(Kim et al., 2005) mimic some aspects of the mok phenotype.

Notch prevents progenitors from leaving the cell cycle and differ-

entiating prematurely. Its activation during the time progenitors

that are competent to produce RGCs leads to a depletion of

this cell type. GDF11, on the other hand, is secreted by differen-

tiated RGCs to negatively regulate the number of new RGCs

produced by suppressing atoh7. As a consequence, progenitors

are more likely to generate RGCs when they are transplanted into

an environment in which RGCs are absent, as in lak mutants

(Poggi et al., 2005a). We made a complementary observation

.

by transplanting wild-type cells into mok mutants. These clones

tend to form fewer RGCs, suggesting that the negative feedback

signals are intact in the mutant environment. Moreover, we show

that constitutive activation of Notch in mok mutants blocks RGC

production and leads to an overproduction of glia, as it does in

wild-type. Together, these results demonstrate that mok mutant

cells are still able to produce and respond to extrinsic regulators

of cell fate.

A Notch gradient along the apical-basal axis of the neuroepi-

thelium is likely to play a key role in neurogenesis. Notch

mRNA is increased on the apical side, whereas Delta mRNA

and protein are enriched basally. This results in a gradient of

Notch transcriptional activity, as demonstrated by the higher

concentration of Notch ICD in the apical domains and the in-

creased her4 expression in cells whose nuclei move into the

high-Notch environment. We have found no evidence that Notch

signaling, neuroepithelial polarity, or the Notch gradient itself are

altered in mok mutants. Distribution of the endocytic pathway

components Rab11 and Numb, which, in other contexts, regu-

late Notch, is unaltered. Moreover, an independent manipulation

of the dynein/dynactin-dependent component of INM (disruption

of the nuclear anchor protein Syne2a) perturbs cell-fate deci-

sions similarly to the mok mutation.

Between mitoses, a progenitor nucleus moves twice through

a Notch spatial gradient. If the nucleus stays close to the apical

side, it will encounter high Notch levels throughout the cell cycle,

and both of its daughters are likely to remain proliferative. On the

other hand, if the nucleus is translocated more basally, Notch

activity is reduced, predisposing the progenitor to produce one

or two daughter neurons during its subsequent mitosis. It follows

that, in mok mutants, the balance of neurogenic versus prolifer-

ative divisions is shifted by decreasing exposure to Notch across

the progenitor population. Given that both INM and Notch signal-

ing compartments are ubiquitous features of CNS neuroepithelia

(Frade, 2002) as well as nonneuronal epithelia (Bort et al., 2006),

it seems likely that this mechanism is widely employed during

embryonic development, growth, and regeneration.

EXPERIMENTAL PROCEDURES

Immunohistochemistry and In Situ Hybridization

Retinal sections were stained using standard protocols (Kay et al., 2001). E12

CD1 mice embryos (gift from Florence Lee) were fixed in 4% PFA and pro-

cessed for cryosectioning according to standard protocols. The full list of pri-

mary and secondary antibodies is given in the Supplemental Data. Staining for

g-tubulin required the treatment of the samples for 5 min in acetone at �20�C

prior to blocking. Fixation and staining for a-tubulin were performed as de-

scribed (Wehman et al., 2007). Apoptosis was detected by whole-mount

TUNEL assay using the ApoTag kit (Chemicon). Whole-mount in situ hybridiza-

tion was performed according to standard protocols, and a list of antisense

probes is reported in the Supplemetal Data.

BrdU/IdU Incorporation

Cell proliferation was assayed by BrdU/IdU incorporation as previously de-

scribed (Burns and Kuan, 2005; Kay et al., 2001) with the following modifica-

tion. Embryos were injected with 10 mM BrdU or IdU solutions into the yolk

and grown until fixation.

Cell Transplantation Analysis

Chimeric embryos were generated using standard methods (Ho and Kane,

1990). Donor embryos were Tg(h2afv:GFP)kca6 to easily identify transplanted

clones. Both donor embryos and chimeras were allowed to develop until 5

dpf to identify mutants. Chimeric larvae were then fixed and processed for im-

munohistochemistry.

DNA and MO Injections

All injections were performed at the 1 cell stage unless specified otherwise.

dnct1-MO were injected at a concentration of 20 mM. Sequence: 50-ctgagg

gacggccggtctgtggagg. syne2a-MO were injected at 500 mM (high dose) or

50 mM (low dose). The sequence has been described previously (Tsujikawa

et al., 2007). Expression of Syne2a-KASH domain Pard3-GFP, EGFP-Numb,

and EGFP-Rab11 was carried out as described (Geldmacher-Voss et al.,

2003; Muto et al., 2006; Reugels et al., 2006; Tsujikawa et al., 2007). The

pard3-EGFP construct was injected at 2–4 cell stages to achieve sparse ex-

pression. EGFP-Numb and EGFP-rab11 were subcloned in the Tol2-based

vector and injected with transposase RNA to achieve uniform expression

(Kawakami, 2004).

Nuclear Labeling to Record INM

To label nuclei of retinal progenitor cells, plasmid DNA encoding the histone

H2B-GFP fusion protein was microinjected into 1–4 cell-stage embryos to

label nuclei in a mosaic fashion throughout the embryo (Koster and Fraser,

2001; Meng et al., 1999). Embryos were derived from incrosses of heterozy-

gous moks309 pairs. All embryos were grown in 0.003% 1-phenyl-2-thiourea

(PTU) to block pigmentation. At 22 hpf, small pieces of tail tissue were used

to PCR genotype either wild-type or mutant embryos prior to imaging. Follow-

ing imaging, all embryos were allowed to develop and were scored for their

phenotypes in order to validate genotyping results.

Confocal Time-Lapse Microscopy

At 26 hpf, labeled embryos were anesthetized with 0.05% Tricane in 0.003%

PTU and embedded in 1.0% low-temperature-melting agarose. Embryos

were placed in a glass bottom culture dish and oriented so that the eye was

facing up. GFP-labeled nuclei and dRFP-labeled cells were imaged on a Nikon

C1 confocal microscope. Transmitted light images were also collected during

the time-lapse to enable accurate measurement of apical and basal surfaces

during nuclear movements. Optical z sections were collected at 2 mm steps

every 12 min for 30–48 hr. These parameters were sufficient to capture M

phase for each cell while reducing photobleaching during the extended time

course. Temperature was maintained throughout all experiments at 28.5�C

using a stage incubator. Nuclear migration velocities were measured as

described (Baye and Link, 2007b). For details, see the Supplemental Data.

SUPPLEMENTAL DATA

The Supplemental Data include Supplemental Experimental Procedures, ten

figures, one table, and three movies and can be found with this article online

at http://www.cell.com/cgi/content/full/134/6/1055/DC1/.

ACKNOWLEDGMENTS

We thank D.Y. Stainier, S. Guo, J. Malicki, J. Lewis, I. Masai, P.A. Raymond,

B.D. Perkins, A.B. Chitnis, N. Katsanis, K.T. Vaughan, R.B. Vallee, and H.A.

Ingraham for fish strains, antibodies, in situ probes, and mouse embryos.

We thank Jeremy Reiter for critically reading the manuscript. F.D.B. was

supported by a Human Frontier Science Program long-term fellowship.

A.M.W. was supported by an American Association of University Women

dissertation fellowship. This work was supported by NIH grants EY013855

(H.B.), EY012406 (H.B.), EY01467 (B.A.L.), and the March of Dimes Foundation

(H.B.).

Received: September 12, 2007

Revised: April 25, 2008

Accepted: July 11, 2008

Published: September 18, 2008

Cell 134, 1055–1065, September 19, 2008 ª2008 Elsevier Inc. 1063

REFERENCES

Austin, C.P., Feldman, D.E., Ida, J.A., Jr., and Cepko, C.L. (1995). Vertebrate

retinal ganglion cells are selected from competent progenitors by the action

of Notch. Development 121, 3637–3650.

Baye, L.M., and Link, B.A. (2007a). Interkinetic nuclear migration and the

selection of neurogenic cell divisions during vertebrate retinogenesis.

J. Neurosci. 27, 10143–10152.

Baye, L.M., and Link, B.A. (2007b). The disarrayed mutation results in cell cycle

and neurogenesis defects during retinal development in zebrafish. BMC Dev.

Biol. 7, 28.

Baye, L.M., and Link, B.A. (2008). Nuclear migration during retinal develop-

ment. Brain Res. 1192, 29–36.

Berdnik, D., Torok, T., Gonzalez-Gaitan, M., and Knoblich, J.A. (2002). The

endocytic protein alpha-Adaptin is required for numb-mediated asymmetric

cell division in Drosophila. Dev. Cell 3, 221–231.

Bort, R., Signore, M., Tremblay, K., Martinez Barbera, J.P., and Zaret, K.S.

(2006). Hex homeobox gene controls the transition of the endoderm to a pseu-

dostratified, cell emergent epithelium for liver bud development. Dev. Biol.

290, 44–56.

Brown, N.L., Patel, S., Brzezinski, J., and Glaser, T. (2001). Math5 is required

for retinal ganglion cell and optic nerve formation. Development 128, 2497–

2508.

Burns, K.A., and Kuan, C.Y. (2005). Low doses of bromo- and iododeoxyuri-

dine produce near-saturation labeling of adult proliferative populations in the

dentate gyrus. Eur. J. Neurosci. 21, 803–807.

Cepko, C.L., Austin, C.P., Yang, X., Alexiades, M., and Ezzeddine, D. (1996).

Cell fate determination in the vertebrate retina. Proc. Natl. Acad. Sci. USA

93, 589–595.

Chitnis, A.B. (1995). The role of Notch in lateral inhibition and cell fate specifi-

cation. Mol. Cell. Neurosci. 6, 311–321.

Doerre, G., and Malicki, J. (2001). A mutation of early photoreceptor develop-

ment, mikre oko, reveals cell-cell interactions involved in the survival and

differentiation of zebrafish photoreceptors. J. Neurosci. 21, 6745–6757.

Draper, B.W., Morcos, P.A., and Kimmel, C.B. (2001). Inhibition of zebrafish

fgf8 pre-mRNA splicing with morpholino oligos: a quantifiable method for

gene knockdown. Genesis 30, 154–156.

Edlund, T., and Jessell, T.M. (1999). Progression from extrinsic to intrinsic

signaling in cell fate specification: a view from the nervous system. Cell 96,

211–224.

Emery, G., Hutterer, A., Berdnik, D., Mayer, B., Wirtz-Peitz, F., Gaitan, M.G.,

and Knoblich, J.A. (2005). Asymmetric Rab 11 endosomes regulate delta recy-

cling and specify cell fate in the Drosophila nervous system. Cell 122, 763–773.

Fan, S.S., and Ready, D.F. (1997). Glued participates in distinct microtubule-

based activities in Drosophila eye development. Development 124, 1497–

1507.

Frade, J.M. (2002). Interkinetic nuclear movement in the vertebrate neuroepi-

thelium: encounters with an old acquaintance. Prog. Brain Res. 136, 67–71.

Furukawa, T., Mukherjee, S., Bao, Z.Z., Morrow, E.M., and Cepko, C.L. (2000).

rax, Hes1, and notch1 promote the formation of Muller glia by postnatal retinal

progenitor cells. Neuron 26, 383–394.

Gaiano, N., Nye, J.S., and Fishell, G. (2000). Radial glial identity is promoted by

Notch1 signaling in the murine forebrain. Neuron 26, 395–404.

Geldmacher-Voss, B., Reugels, A.M., Pauls, S., and Campos-Ortega, J.A.

(2003). A 90-degree rotation of the mitotic spindle changes the orientation of

mitoses of zebrafish neuroepithelial cells. Development 130, 3767–3780.

Guo, M., Jan, L.Y., and Jan, Y.N. (1996). Control of daughter cell fates during

asymmetric division: interaction of Numb and Notch. Neuron 17, 27–41.

Hatakeyama, J., and Kageyama, R. (2004). Retinal cell fate determination and

bHLH factors. Semin. Cell Dev. Biol. 15, 83–89.

Ho, R.K., and Kane, D.A. (1990). Cell-autonomous action of zebrafish spt-1

mutation in specific mesodermal precursors. Nature 348, 728–730.

1064 Cell 134, 1055–1065, September 19, 2008 ª2008 Elsevier Inc.

Holt, C.E., Bertsch, T.W., Ellis, H.M., and Harris, W.A. (1988). Cellular determi-

nation in the Xenopus retina is independent of lineage and birth date. Neuron 1,

15–26.

Jadhav, A.P., Cho, S.H., and Cepko, C.L. (2006). Notch activity permits retinal

cells to progress through multiple progenitor states and acquire a stem cell

property. Proc. Natl. Acad. Sci. USA 103, 18998–19003.

Kawakami, K. (2004). Transgenesis and gene trap methods in zebrafish by

using the Tol2 transposable element. Methods Cell Biol. 77, 201–222.

Kay, J.N., Finger-Baier, K.C., Roeser, T., Staub, W., and Baier, H. (2001). Ret-

inal ganglion cell genesis requires lakritz, a Zebrafish atonal Homolog. Neuron

30, 725–736.

Kim, J., Wu, H.H., Lander, A.D., Lyons, K.M., Matzuk, M.M., and Calof, A.L.

(2005). GDF11 controls the timing of progenitor cell competence in developing

retina. Science 308, 1927–1930.

King, S.J., and Schroer, T.A. (2000). Dynactin increases the processivity of the

cytoplasmic dynein motor. Nat. Cell Biol. 2, 20–24.

Koster, R.W., and Fraser, S.E. (2001). Tracing transgene expression in living

zebrafish embryos. Dev. Biol. 233, 329–346.

Le Borgne, R. (2006). Regulation of Notch signalling by endocytosis and endo-

somal sorting. Curr. Opin. Cell Biol. 18, 213–222.

Livesey, F.J., and Cepko, C.L. (2001). Vertebrate neural cell-fate determina-

tion: lessons from the retina. Nat. Rev. Neurosci. 2, 109–118.

Marquardt, T., and Gruss, P. (2002). Generating neuronal diversity in the retina:

one for nearly all. Trends Neurosci. 25, 32–38.

Masai, I., Stemple, D.L., Okamoto, H., and Wilson, S.W. (2000). Midline signals

regulate retinal neurogenesis in zebrafish. Neuron 27, 251–263.

Masai, I., Lele, Z., Yamaguchi, M., Komori, A., Nakata, A., Nishiwaki, Y., Wada,

H., Tanaka, H., Nojima, Y., Hammerschmidt, M., et al. (2003). N-cadherin me-

diates retinal lamination, maintenance of forebrain compartments and pattern-

ing of retinal neurites. Development 130, 2479–2494.

McGill, M.A., and McGlade, C.J. (2003). Mammalian numb proteins promote

Notch1 receptor ubiquitination and degradation of the Notch1 intracellular

domain. J. Biol. Chem. 278, 23196–23203.

McGrail, M., Gepner, J., Silvanovich, A., Ludmann, S., Serr, M., and Hays, T.S.

(1995). Regulation of cytoplasmic dynein function in vivo by the Drosophila

Glued complex. J. Cell Biol. 131, 411–425.

Meng, A., Jessen, J.R., and Lin, S. (1999). Transgenesis. Methods Cell Biol. 60,

133–148.

Morrison, S.J., Perez, S.E., Qiao, Z., Verdi, J.M., Hicks, C., Weinmaster, G.,

and Anderson, D.J. (2000). Transient Notch activation initiates an irreversible

switch from neurogenesis to gliogenesis by neural crest stem cells. Cell 101,

499–510.

Murciano, A., Zamora, J., Lopez-Sanchez, J., and Frade, J.M. (2002). Interki-

netic nuclear movement may provide spatial clues to the regulation of neuro-

genesis. Mol. Cell. Neurosci. 21, 285–300.

Muto, A., Arai, K., and Watanabe, S. (2006). Rab11-FIP4 is predominantly

expressed in neural tissues and involved in proliferation as well as in differen-

tiation during zebrafish retinal development. Dev. Biol. 292, 90–102.

Nelson, B.R., Gumuscu, B., Hartman, B.H., and Reh, T.A. (2006). Notch activity

is downregulated just prior to retinal ganglion cell differentiation. Dev. Neuro-

sci. 28, 128–141.

Nichols, J.T., Miyamoto, A., and Weinmaster, G. (2007). Notch signaling–con-

stantly on the move. Traffic 8, 959–969.

Ohnuma, S., Hopper, S., Wang, K.C., Philpott, A., and Harris, W.A. (2002).

Co-ordinating retinal histogenesis: early cell cycle exit enhances early cell

fate determination in the Xenopus retina. Development 129, 2435–2446.

Passini, M.A., Levine, E.M., Canger, A.K., Raymond, P.A., and Schechter, N.

(1997). Vsx-1 and Vsx-2: differential expression of two paired-like homeobox

genes during zebrafish and goldfish retinogenesis. J. Comp. Neurol. 388,

495–505.

Pearson, B.J., and Doe, C.Q. (2004). Specification of temporal identity in the

developing nervous system. Annu. Rev. Cell Dev. Biol. 20, 619–647.

Perron, M., and Harris, W.A. (2000). Determination of vertebrate retinal progen-

itor cell fate by the Notch pathway and basic helix-loop-helix transcription fac-

tors. Cell. Mol. Life Sci. 57, 215–223.

Poggi, L., Vitorino, M., Masai, I., and Harris, W.A. (2005a). Influences on neural

lineage and mode of division in the zebrafish retina in vivo. J. Cell Biol. 171,

991–999.

Poggi, L., Zolessi, F.R., and Harris, W.A. (2005b). Time-lapse analysis of retinal

differentiation. Curr. Opin. Cell Biol. 17, 676–681.

Reugels, A.M., Boggetti, B., Scheer, N., and Campos-Ortega, J.A. (2006).

Asymmetric localization of Numb:EGFP in dividing neuroepithelial cells during

neurulation in Danio rerio. Dev. Dyn. 235, 934–948.

Sauer, F.C. (1935). Mitosis in the neural tube. J. Comp. Neurol. 62, 377–405.

Scheer, N., Groth, A., Hans, S., and Campos-Ortega, J.A. (2001). An instruc-

tive function for Notch in promoting gliogenesis in the zebrafish retina. Devel-

opment 128, 1099–1107.

Schroer, T.A. (2004). Dynactin. Annu. Rev. Cell Dev. Biol. 20, 759–779.

Starr, D.A., and Fischer, J.A. (2005). KASH ’n Karry: the KASH domain family of

cargo-specific cytoskeletal adaptor proteins. Bioessays 27, 1136–1146.

Takke, C., Dornseifer, P., v. Weizsacker, E., and Campos-Ortega, J.A. (1999).

her4, a zebrafish homologue of the Drosophila neurogenic gene E(spl), is a tar-

get of NOTCH signalling. Development 126, 1811–1821.

Tokunaga, A., Kohyama, J., Yoshida, T., Nakao, K., Sawamoto, K., and Okano,

H. (2004). Mapping spatio-temporal activation of Notch signaling during neu-

rogenesis and gliogenesis in the developing mouse brain. J. Neurochem. 90,

142–154.

Tsujikawa, M., Omori, Y., Biyanwila, J., and Malicki, J. (2007). Mechanism of

positioning the cell nucleus in vertebrate photoreceptors. Proc. Natl. Acad.

Sci. USA 104, 14819–14824.

Turner, D.L., and Cepko, C.L. (1987). A common progenitor for neurons and

glia persists in rat retina late in development. Nature 328, 131–136.

Turner, D.L., Snyder, E.Y., and Cepko, C.L. (1990). Lineage-independent de-

termination of cell type in the embryonic mouse retina. Neuron 4, 833–845.

Vetter, M.L., and Brown, N.L. (2001). The role of basic helix-loop-helix genes in

vertebrate retinogenesis. Semin. Cell Dev. Biol. 12, 491–498.

Wang, S.W., Kim, B.S., Ding, K., Wang, H., Sun, D., Johnson, R.L., Klein, W.H.,

and Gan, L. (2001). Requirement for math5 in the development of retinal gan-

glion cells. Genes Dev. 15, 24–29.

Wehman, A.M., Staub, W., Meyers, J.R., Raymond, P.A., and Baier, H. (2005).

Genetic dissection of the zebrafish retinal stem-cell compartment. Dev. Biol.

281, 53–65.

Wehman, A.M., Staub, W., and Baier, H. (2007). The anaphase-promoting

complex is required in both dividing and quiescent cells during zebrafish

development. Dev. Biol. 303, 144–156.

Wei, X., Cheng, Y., Luo, Y., Shi, X., Nelson, S., and Hyde, D.R. (2004). The

zebrafish Pard3 ortholog is required for separation of the eye fields and retinal

lamination. Dev. Biol. 269, 286–301.

Wetts, R., and Fraser, S.E. (1988). Multipotent precursors can give rise to all

major cell types of the frog retina. Science 239, 1142–1145.

Whited, J.L., Cassell, A., Brouillette, M., and Garrity, P.A. (2004). Dynactin is

required to maintain nuclear position within postmitotic Drosophila photore-

ceptor neurons. Development 131, 4677–4686.

Xiao, T., Roeser, T., Staub, W., and Baier, H. (2005). A GFP-based genetic

screen reveals mutations that disrupt the architecture of the zebrafish retino-

tectal projection. Development 132, 2955–2967.

Yan, R.T., Ma, W., Liang, L., and Wang, S.Z. (2005). bHLH genes and retinal

cell fate specification. Mol. Neurobiol. 32, 157–171.

Yeo, S.Y., Kim, M., Kim, H.S., Huh, T.L., and Chitnis, A.B. (2007). Fluorescent

protein expression driven by her4 regulatory elements reveals the spatiotem-

poral pattern of Notch signaling in the nervous system of zebrafish embryos.

Dev. Biol. 301, 555–567.

Cell 134, 1055–1065, September 19, 2008 ª2008 Elsevier Inc. 1065