Neuron, Vol. 40, 897–904, December 4, 2003, Copyright 2003 by Cell Press
Report Importance of Intrinsic Mechanisms in
Cell Fate Decisions in the Developing
nergic amacrine cells in the goldfish and frog retina, for
example, results in the overproduction of these cells by
RNECs, suggesting that negative feedback signals can
influence cell fate decisions (Negishi et al., 1982; Reh
and Tully, 1986). Similarly, ganglion cells in the chick
retina apparently secrete a factor that prevents cells
from choosing the ganglion cell fate (Waid and McLoon,
1998). Conversely, newborn rat retinal cells promote rod
photoreceptor development when cultured with embry-
onic retinal cells, suggesting that positive signals may
also contribute to cell fate choice (Watanabe and Raff,
1990, 1992). In these mixed-age experiments, however,
signals from older rat retinal cells cannot induce the
younger RNECs to make inappropriate cell fate choices
and produce late-born cell types prematurely, sug-
ent in the types of cells they can produce (Altshuler and
Cepko, 1992; Belliveau and Cepko, 1999; Lillien and
Cepko, 1992; Watanabe and Raff, 1990). These findings
indicate that extracellular signals are not the only deter-
minant of cell fate choice in the retina and that the
developmental potential of RNECs changes over time.
A current view of retinal development that tries to
reconcile all of these findings suggests that, as RNECs
proliferate and mature, they pass through a series of
competence states, progressively changing their re-
sponsiveness to instructive extracellular signals that in-
fluence what type of cell they become (Livesey and
Cepko, 2001; Watanabe and Raff, 1990). Although it is
now generally accepted that both cell-intrinsic mecha-
nisms and extrinsic signals influence cell fate choice
in the developing retina, their relative contributions to
particular cell fate decisions remain unclear.
nisms and extracellular signals in cell fate choice, we
have developed a serum-free and extract-free clonal-
density culture system, in which E16–17 rat RNECs
proliferate and differentiate. We have compared the de-
velopment of clones in these cultures, in which the envi-
ronment is relatively homogeneous and constant, to the
development of clones in serum-containing explants of
the same age retina, in which the environment is com-
plex and changes over time. Our findings suggest that
cell-intrinsic mechanisms play a major part in determin-
ing when E16–17 rat RNECs stop dividing and what
cell types they generate when they withdraw from the
Michel Cayouette,1,3,* Ben A. Barres,2
and Martin Raff1
1MRC Laboratory for Molecular Cell Biology and
Cell Biology Unit
University College London
London, WC1E 6BT
2Department of Neurobiology
299 Campus Drive
Stanford, California 94305
Cell diversification in the developing nervous system
is thought to involve both cell-intrinsic mechanisms
and extracellular signals, but their relative importance
in particular cell fate decisions remains uncertain. In
the mammalian retina, different cell types develop on
a predictable schedule from multipotent retinal neuro-
epithelial cells (RNECs). A current view is that RNECs
pass through a series of competence states, progres-
sively changingtheir responsiveness toinstructive ex-
tracellular cues, which also change over time. We
show here, however, that embryonic day 16–17 (E16–
17) rat RNECs develop similarly in serum-free clonal-
density cultures and in serum-containing retinal ex-
plants—in the number of times they divide, the cell
ate these cell types. These surprising results suggest
that extracellular signals may be less important than
currently believed in determining when RNECs stop
dividing and what cell types they generate when they
withdraw from the cell cycle, at least from E16–17
It is a major challenge to understand how the enormous
diversity of cell types develop in the mammalian central
nervous system (CNS). The retina is an attractive part
of the CNS to take on this challenge, as it contains a
manageable number of cell types, which develop on a
predictable schedule. Pioneering cell lineage studies in
the retina of vertebrates showed that photoreceptors,
neurons, and Mu ¨ller glial cells can all develop from a
single RNEC (Turner and Cepko, 1987; Holt et al., 1988;
Wetts and Fraser, 1988), raising the question of how
individual multipotent RNECs choose between differ-
A number of studies suggest a role for extracellular
signals in influencing cell fate choice in the developing
Embryonic Rat RNECs Proliferate
and Differentiate in Serum-Free
and Extract-Free Clonal-Density Cultures
One way to study what influences the fate of RNECs
is to remove them from their normal environment and
analyze their behavior in clonal-density culture, where
one can maintain a relatively homogeneous and con-
stant environment. We used E16–17 rat neural retinal
Fairchild Building, 299 Campus Drive, D231, Stanford, California
cells and plated them at clonal density in a defined
culture medium in a culture flask coated with poly-
L-lysine and laminin. At this stage, RNECs still have
the potential to generate rod photoreceptors, amacrine
cells, bipolar cells, and Mu ¨ller glia, but most cones,
ganglion cells, and horizontal cells have already been
generated. Two hours after plating, 65%–70% of the
cells were RNECs, as judged by their characteristic flat
and dark morphology, extensive lamellapodia, and ex-
pression of nestin (Figures 1A and 1B). Many of the
RNECs proliferated, and, after 7–10 days, had produced
clones that varied in size and contained only differenti-
ated cells, as judged by their morphology (Figure 1C),
lack of nestin staining (data not shown), and inability to
incorporate BrdU (data not shown).
The Expected Differentiated Cell Types Develop
in Clonal-Density Cultures
In vivo, photoreceptor cells display a characteristic pat-
tern of heterochromatin in their nucleus, which is not
seen in the other cell types in the retina (Figure 1D), and
previous studies showed that postmitotic rods acquire
this chromatin pattern before they express rhodopsin
(Neophytou et al., 1997). A large proportion of cells in
our clonal cultures displayed this chromatin pattern,
along with a typical rod cell size and morphology (Fig-
ures 1E and 2); as expected in such cultures (Harris and
Messersmith, 1992; Neophytou et al., 1997), these cells
did not express rhodopsin. When we stained medium-
tor protein recoverin, however, we found that 97% ?
0.1% of the cells with this characteristic chromatin pat-
1E and 1F) and did not stain with any of the other cell
type-specific antibodies used in this study, as was the
case for all the rod-like cells in our clonal-density cul-
tures (Figure 2).
Neurons in the clones could be readily distinguished
from rods by their larger size, longer processes, and
more homogeneous chromatin (see Figures 1E, 2I, and
2M). To help identify the two neuronal cell types ex-
pected to develop from E16–17 RNECs, amacrine, and
bipolar cells, we used antibodies against syntaxin and
islet-1. As expected (Barnstable et al., 1985; Galli-Resta
et al., 1997), staining cryosections of postnatal day 10
(P10) retina revealed that syntaxin antibodies stained
amacrine cells, while islet-1 antibodies stained bipolars,
a subset of amacrines, and a subset of ganglion cells
(Figure 1G). Consistent with these results, we found two
populations of syntaxin?amacrine cells in our cultures,
one that was islet-1?(Figure 1H) and one that was islet-
1?(Figures 2E–2H and 2M–2P); both had a typical ama-
crine cell morphology (Figures 2E and 2M). Bipolar cells
in our cultures had a simpler morphology (Figures 2I
and 2M) and, as expected from staining cryosections
(Figure 1G), were islet-1?and syntaxin?(Figures 1H,
2I–2L, and 2M–2P); they did not, however, express late
ath et al., 1990) or 115A10 (Onoda and Fujita, 1987). The
islet-1?cells in our cultures were unlikely to be ganglion
cells, as they did not stain with Brn3b, a specific marker
for retinal ganglion cells (data not shown).
We identified Mu ¨ller glial cells by their characteristic
Figure 1. The Various Markers Used to Distinguish the Differenti-
ated Cell Types Produced in Our Dissociated-Cell Cultures
(A and B) Two E16–17 RNECs after 24 hr in clonal-density culture
cell marker nestin ([B], red) and Hoechst to visualize the nucleus
of differentiated cell types after 7 days in culture. (D) A cryosection
of a P14 rat retina stained with Hoechst. Note the characteristic
patternofheterochromatin inphotoreceptor(PR)cell nuclei(arrows)
comparedto themorehomogeneous patternincellsof theinterneu-
ron layer (INL). (E and F) Three rod cells (arrows) stained with
Hoechst (E) and anti-recoverin antibodies (F) after 10 days in vitro.
Note that the recoverin?photoreceptors display the same pattern
of heterochromatin as photoreceptors in vivo. (G) Fluorescence mi-
crograph of a cryosection of P10 rat retina. Syntaxin is expressed
in amacrine cells, as seen by the staining in the innermost layer of
the interneuron layer (INL), which contains amacrine cell bodies,
and of the inner plexiform layer (IPL), which contains amacrine ax-
ons. Islet-1 is expressed in bipolar cells, as seen by the nuclear
staining of the outermost layer of the INL (asterisk), which contains
bipolar cell bodies. It is also expressed by a subset of amacrine
(arrows) and ganglion cells (GCL). Nuclei are stained with Hoechst
(blue). (H) Fluorescence micrographs of a syntaxin?/islet-1?ama-
crine cell and an islet-1?/syntaxin?bipolar cell (arrow) in a clonal
culture after 10 days. (I) Fluorescence micrograph of a cryosection
of P10 rat retina. The nucleus of the Mu ¨ller cells in the INL express
cyclin D3 (green). (J and K) Phase-contrast (J) and fluorescence (K)
micrographs showing a cyclin-D3?Mu ¨ller cell (arrow), surrounded
by a number of cyclin-D3?cells after 10 days in clonal culture. Scale
bars, 10 ?m (D); 15 ?m (A, B, E, F, H, J, and K); 50 ?m (C); 60 ?m
(G and I).
Intrinsic Mechanisms in Cell Fate Choice
Figure 2. Phase-Contrast and Fluorescence
Micrographs of Clones Produced by E16-17
The cells were stained as indicated. (A–D) A
a characteristic heterochromatin pattern and
are negative for islet-1 and syntaxin. (E–H) A
clone containing three rods (arrows) and a
syntaxin?/islet-1?amacrine cell. (I and J) A
clone containing three rods (arrows) and an
islet-1?/syntaxin?bipolar cell. (M–O) A clone
containing a rod (arrow), a syntaxin?/islet-1?
lar cell. (Q–S) A clone containing two rods
(arrows) and a cyclin D3?Mu ¨ller cell. Scale
bars, 8 ?m (A–D and I–L); 10 ?m (E–H); 15 ?m
(M–P); 20 ?m (Q–S).
glial morphology and by the expression of cyclin D3 in
of cyclin D3 was consistent with it being specific for
Mu ¨ller cells (Figure 1I), as previously reported (Dyer and
Cepko, 2000). As shown in Figures 1J, 1K, and 2Q–2S,
small numbers of cyclin-D3?Mu ¨ller cells developed in
our clonal-density cultures.
The clones that developed in our cultures were very
heterogeneous in both size and cell type composition,
much as observed in vivo (Turner and Cepko, 1987;
Turner et al., 1990). Rods were the major cell type that
type of clone (Figures 2A–2D and Table 1), as expected
from in vivo clonal analyses (Turner and Cepko, 1987;
Turner et al., 1990). We also found clones containing
mixtures of rods and amacrines (Figures 2E–2H), rods
and bipolars (Figures 2I–2L), rods, amacrines, and bipo-
lars (Figures 2M–2P), rods and Mu ¨ller cells (Figures 2Q–
2S), Mu ¨ller cells and neurons, with or without rods, and
all four cell types (Table 1). As expected from previous
in vivo studies (and from our own clonal analysis in
explants—see below), we did not find any clones con-
taining just bipolars and Mu ¨ller cells or just amacrine
and Mu ¨ller cells (Table 1). Although clonal development
in our cultures was remarkably similar to that expected
from previous studies in vivo, some of the cell types
tiation markers, suggesting that environmental cues
may be required for their phenotype maturation.
Clone Development in Dissociated-Cell Culture
Closely Resembles that in Explant Cultures
To assess more accurately how closely RNEC develop-
ment in our E16–17 clonal-density cultures resembled
that in normal developing retina, we compared it to
clonal development in explant cultures of E16–17 rat
retinas containing 10% FCS. In these conditions, retinal
explants develop remarkably similarly to retinas in vivo
(Sheedlo and Turner, 1996). We infected the explants
with retroviral vectors that encoded either enhanced
green fluorescentprotein (GFP), which wasused to ana-
lyze clone size (Figure 3A), or placental alkaline phos-
phatase (PLAP), which was used to analyze the cell type
composition of the clones (Figure 3B).
As shown in Figure 3C, clone sizes in explant and
clonal-density cultures were remarkably similar. Clones
Table 1. Clonal Analysis in Clonal-Density Cultures and Retinal Explants
Total Number of
Number of Clones with
Dead Cells (%)
Total Number of
Clones (%)Cell Combination
Photoreceptor ? Amacrine
Photoreceptor ? Bipolar
Photoreceptor ? Mu ¨ller
Photoreceptor ? Amacrine ? Bipolar
Photoreceptor ? Amacrine ? Bipolar ? Mu ¨ller
Photoreceptor ? Bipolar ? Mu ¨ller
Photoreceptor ? Amacrine ? Mu ¨ller
Mu ¨ller only
Bipolar ? Amacrine
Bipolar ? Amacrine ? Mu ¨ller
Bipolar ? Mu ¨ller
Amacrine ? Mu ¨ller
A total of ten dissociated-cell cultures of E16–17 neural retina from three separate experiments were analyzed after 7–10 days in culture. A
total of four separate retinal explants were analyzed after 10 days in culture. Numbers in parentheses represent the proportions.
in clonal-density cultures contained an average of 2.9
cells per clone, whereas clones in explants contained
an averageof 3.2cells perclone. Moreover,the distribu-
tion of clone sizes was very similar in the two types of
cultures (Figure 3C). By contrast, the distribution of
clone sizes that developed from E20–21 RNECs was
Figure 3. Comparison of Clones that De-
velop in Explant versus Clonal-Density Reti-
nal Cultures and the Sequence of Develop-
ment of Different Cell Types in Clonal-
(A) Example of a clone containing four rods,
10 days after an E17 retinal explant was in-
fected with a GFP-encoding retroviral vector.
(B) Example of a clone containing three rods
and one bipolar, 10 days after an E17 retinal
explant was infected with a PLAP-encoding
retroviral vector. (C) The distribution of clone
sizes in explants and clonal-density cultures
from E16–17 or E20–21 rat retina, after 7–10
of clones in explants and clonal-density cul-
tures, as a percent of either the total number
of cells in clones analyzed (D) or the total
clones analyzed (E). Results are shown as
mean ? SD of four separate cultures. (F) Se-
quence of development of different cell types
in cultures of E16–17 retinal cells. Results are
shown as mean ? SD of three separate cul-
tures per time point. (G) BrdU was added to
the cultures at 2 hr, 18 hr, and 48 hr post-
plating, and the clones were then allowed to
develop for 7 days. Results are shown as
mean ? SD of three cultures per time point
Intrinsic Mechanisms in Cell Fate Choice
much more restricted than observed with E16–17 cells
(Figure 3C), consistent with previous evidence that the
proliferative potential of RNECs progressively decreases
as the RNECs mature (Lillien and Cepko, 1992; Wata-
nabe and Raff, 1990).
The cell type compositions of the clones generated
from E16–17 RNECs in clonal-density cultures and in
explants were also remarkably similar (Figures 3D and
3E). When analyzed as a proportion of the total number
of cells assessed in clones, the proportions of rods,
ilar in the two cases (Figure 3D). The proportions of
clones containing only rods, or at least one bipolar,
one amacrine, or one Mu ¨ller cell, were also remarkably
similar in the two cases (Figure 3E).
As RNECs change their developmental potential over
Watanabe and Raff, 1990), one would expect that
E20–21 RNECs would produce fewer early-born cell
types, such as amacrines, and more single-cell-type
clones than would E16–17 RNECs. This was the case.
Only 7% ? 3% of clones produced by E20–21 RNECs
contained at least one amacrine cell, compared to
17% ? 7% of clones produced by E16–17 RNECs (p ?
0.05, Student’s t test), and 75% ? 3% of clones pro-
to 64% ? 7% of clones produced by E16–17 RNECs
(p ? 0.05, Student’s t test). Even if we only compared
clones containing four cells in E16–17 and E20–21 cul-
tures, 21% of the E20–21 clones contained amacrine
cells, whereas 33% of E16–17 clones contained ama-
RNECs to give rise to early-born cell types was not
caused only by their reduced proliferative capacity.
Does Instructive Intraclonal Signaling Play
a Part in Cell Fate Choice?
Detailed analysis of the clones that developed in clonal-
signals from previously differentiated cells might be re-
quired for other cell types to develop within the same
clone. As shown in Table 1, we found almost all possible
combinations of the four expected cell types within
clones. Clones containing only rods, for example, were
the most common, suggesting that signals from other
cell types were not required for E16–17 RNECs to de-
velop into rods in these cultures. Similarly, we found
clones containing only amacrine cells and rods, only
bipolar cells and rods, and only Mu ¨ller cells and rods,
suggesting that the development of RNECs into ama-
crine, bipolar, orMu ¨ller cells did notrequire signals from
other cell types, with the possible exception of signals
from rods. We also found, however, a small number of
clones containing amacrine, bipolar, and/or Mu ¨ller cells
that did not contain photoreceptors, suggesting that
to develop into these cell types. Although cell death
could have confounded this analysis (Voyvodic et al.,
1995), we found that little cell death occurred in our
clones (Table 1). The 905 clones analyzed contained a
total of 2625 cells, and only 106 dead cells (4%) were
observed among them, and only 74 clones contained
one or more dead cells (Table 1). Most importantly, no
dead cells were seen in 507 out of 545 clones that con-
tained only rods, in 26 out of 27 clones that did not
contain rods, or in 15 clones that contained only ama-
crines or Mu ¨ller cells. Thus, cell death is unlikely to have
significantly confounded our analyses.
RNECs in Dissociated-Cell Culture Can Reorient
Their Mitotic Spindle before Division
We recently reported evidence that the plane of RNEC
division in the newborn rat retina can influence the fate
of the daughter cells (Cayouette and Raff, 2003). How
tion of cell division for cell fate choice with our present
density cultures? To study the orientation of division of
RNECs in such cultures, we used time-lapse video mi-
with their mitotic spindle oriented perpendicular to their
long (putative apical-basal) axis (Figure 4A). In some
cases, however, they rotatedtheir spindle at metaphase
to divide with their spindle oriented parallel to this axis
(Figure 4B). Very few cells divided with their spindle in
an intermediate orientation. Thus, it seems that, even
in dissociated-cell cultures, RNECs can position their
spindle so as to divide in either of two orientations,
suggesting that they could segregate asymmetrically
localized proteins to only one of the two daughter cells
during cell division.
Different Cell Types Are Born and Differentiate
in the Predicted Sequence
in Clonal-Density Cultures
To help determine if the different cell types develop in
the normal sequence in our clonal-density cultures, we
either stained the cultures with cell type-specific mark-
ers at different times or added bromodeoxyuridine
(BrdU) to the culture at different times and analyzed the
cultures after 7 days. As shown in Figure 3F, we found
that amacrine cells and rods were the first differentiated
cells detectable, followed by bipolar and Mu ¨ller cells.
As expected, cells expressing nestin decreased over
time and were mostly gone by 8 days. The proportion
of BrdU-positive amacrine cells was greater when BrdU
was added at 2 hr than when it was added at 48 hr
(Figure 3G), indicating that most amacrine cells were
born early. In contrast, the proportion of BrdU-positive
Mu ¨ller cells was greater when BrdU was added at 48 hr
than when it was added at 2 hr (Figure 3G), indicating
that most of these cells were born later than amacrine
cells. Large numbers of BrdU-positive rods were seen
at all time points. All of these results are consistent with
the birthdating experiments previously reported in vivo
(Rapaport et al., 1996; Young, 1985), suggesting that the
various cell types are born and differentiate in the same
general order as in vivo, although it will be important to
confirm this by following individual clones over time.
We are interested in how RNECs choose between alter-
native fates to generate the different cell types in the
developing mammalian retina. To help address this
Figure 4. RNECs in Dissociated-Cell Culture
Divide Either Parallel or Perpendicular to the
Long Axis of the Cell
showing an RNEC dividing with its mitotic
spindle oriented perpendicular to the long
axis of the cell. Note that the nucleus mi-
grated to one pole of the cell to divide. (B)
Snapshots of a time-lapse video recording
showing an RNEC dividing with its mitotic
spindle oriented parallel to the long axis of
the cell. Its nucleus also migrated to one pole
of the cell before the cell divided. Scale bar,
problem, we have developed a serum-free and extract-
free clonal-density culture system in which E16–17 rat
RNECs proliferate and differentiate in the presence of
a cocktail of growth factors and neurotrophins. We find
that the RNECs form clones of various sizes and cellular
compositions, even though they are in the same flask.
Unexpectedly, the clones that develop are remarkably
similar in size and composition to the clones that devel-
op in retinal explants of the same age cultured in serum,
and the various cell types are born and differentiate in
a similar general sequence to that observed in vivo.
Finally, we find that some RNECs in dissociated cell
cultures reorient their mitotic spindle through 90? before
they divide. We discuss the possible implications of
these surprising findings below.
gene has been inactivated, but all other major retinal
cell types develop, suggesting that signals from rods
may not be required for the other retinal cell types to
develop in vivo (Mears et al., 2001). None of these find-
ings exclude the possibility that negative feedback sig-
nals operate to inhibit differentiation. Indeed, there is
strong evidence for such signals, as mentioned in the
Introduction. Moreover, our findings do not exclude the
possibility that positive instructive signals play a crucial
part before E16–17 in the rat retina, either to influence
cell fate directly when RNECs withdraw from the cell
cycle or to program cycling RNECs to influence their
Extracellular signals are almost certainly required for
cell survival and cell proliferation in the developing ret-
ina, as well as for the phenotype maturation of some cell
types after they initially differentiate. We rarely detect
differentiated cells expressing late cell type-specific
markers in our clonal-density cultures, suggesting that
cell-cell interactions are required for the expression of
these proteins. There is strong previous evidence that
opsin expression in photoreceptors requires cell-cell
interactions (Harris and Messersmith, 1992; Morrow et
ofsome differentiationmarkersinother retinalcelltypes
(Akagawa and Barnstable, 1986; Akagawa et al., 1987;
Intrinsic versus Extrinsic Mechanisms
of Cell Diversification
The finding that E16–17 RNECs produce clones of simi-
lar size and cellular composition in our clonal-density
in which the RNECs develop in the two culture systems
are very different. It suggests that cell-intrinsic mecha-
nisms may be more important than previously sus-
pected in determining both when RNECs stop dividing
and differentiate and what cell types they produce when
exiting the cell cycle, at least from E16–17 onward in
resembles that in vivo (Sheedlo and Turner, 1996), it
seems likely that cell-intrinsic mechanisms are also im-
portant in vivo, at least after E16–17 in rats, but this
remains to be shown directly.
We also find in our clonal-density cultures that the
presence of a particular cell type is not required for
the development of any other cell type within a clone,
suggesting that positive instructive signals from differ-
entiatedretinal cellsareunlikelyto determinethechoice
that a RNEC makes when it withdraws from the cell
cycle. There is previous genetic evidence that this may
also be true in vivo. Inactivation of the pax6 gene in the
developing mouse retina, for example, results in the
exclusive development of amacrine cells, suggesting
that amacrine cells can develop without signals from
other cells (Marquardt et al., 2001). Remarkably, various
subtypes of amacrine cells also develop in these mice,
indicating that amacrine cells do not need to make syn-
aptic contact with, or receive signals from, other cell
types in the retina to diversify into subtypes. Similarly,
rods are not found in the retina of mice in which the nrl
Orientation of Cell Division
in Dissociated-Cell Culture
We find that most RNECs proliferating in dissociated-
cell culture divide with their mitotic spindle oriented
perpendicular to their long axis. Remarkably, however,
some rotate their spindle through 90? so that it aligns
along the long axis. We previously showed that an anti-
gen recognized by a monoclonal anti-m-Numb antibody
is asymmetrically localized at one pole in many RNECs
in dissociated-cell cultures and can segregate to only
one of the daughter cells when the RNEC divides (Cay-
ouette et al., 2001). Thus, our current finding raises the
possibility that spindle rotation and asymmetric segre-
gation of cell fate determinants may contribute to RNEC
fate choice, even in dissociated-cell cultures.
asymmetrically segregate cell fate determinants, even
though the cells are not within a polarized epithelium.
Drosophila neuroblasts can asymmetrically localize the
proteins Inscuteable, Prospero, and Staufen at mito-
sis in dissociated-cell culture, just as they do in vivo
Intrinsic Mechanisms in Cell Fate Choice
(Broadus and Doe, 1997). Similarly, mouse cortical pro-
genitors segregate m-Numb asymmetrically in dissoci-
ated-cell culture, and this segregation seems to influ-
ence cell fate choice (Shen et al., 2002).
generated in explants with those generated in clonal-
density cultures, using long-term time-lapse video-
by stereotyped patterns of cell division in both types of
cultures, it would strongly suggest that intrinsic pro-
grams are at work.
Preprogramming versus Stochastic Models
Our findings strongly suggest that intrinsic mechanisms
play an important part in determining when E16–17
RNECs stop dividing and differentiate and what cell
types they produce. The alternative explanation for our
findings—that it is a coincidence that E16–17 RNECs
behave so similarly when in the retinal neuroepithelium
and when isolated in clonal-density culture—seems
There are at least two types of cell-intrinsic mecha-
nisms that could explain our results. One is that RNECs
make decisions stochastically, with the probabilities of
cell fate choices weighted toward certain cell types and
changing over time. Late RNECs, for example, would
would be biased to produce cones and ganglion cells.
poiesis (Till et al., 1964). An alternative possibility is
that individual RNECs become differently programmed
before E16–17 and then step through their specific de-
velopmental program independently of instructive sig-
nals from the environment.
We prefer the second possibility for several reasons.
(1) It could more easily explain how a clone containing
exclusively 33 rods could develop in vivo from an E14
mouse RNEC (Turner et al., 1990). As pointed out by
Williams and Goldowitz (Williams and Goldowitz, 1992),
the chance that such a clonewould develop if all RNECs
although weighted probabilities were not considered
in this calculation. (2) There are other examples where
neural precursor cells seem to be specified early and
thenstep throughaprescribed developmentalprogram.
Neuroblasts in the Drosophila CNS are a particularly
impressive example. They go through a series of asym-
in dissociated-cell culture (Furst and Mahowald, 1985;
Luer and Technau, 1992), and they sequentially express
different sets of transcription factors with each cell divi-
sion (Isshiki et al., 2001). Similarly, dissociated cortical
progenitor cells can undergo stereotyped patterns of
cell division and differentiation to produce neurons and
glial cells in a normal sequence in clonal cultures (Qian
et al., 1998, 2000). (3) Some vertebrate RNECs have
been shown to be biased to produce amacrine cells:
this is the case for embryonic rat RNECs that express
and for some Xenopus RNECs (Moody et al., 2000) and
even some Xenopus blastomeres (Huang and Moody,
1995, 1997). Taken together, these findings and ours
seem most consistent with the possibility that individual
RNECs become preprogrammed in various ways to di-
and to give rise to one or more specific cell types. If so,
then some RNECs are apparently programmed to rotate
their mitotic spindle through 90? after a certain time or
number of divisions, which would be remarkable.
To test further this model of RNEC preprogramming,
Clonal-Density Culture of RNECs
Dawley rats were dissected and dissociated following a previously
published method (Jensen and Raff, 1997). The cells were resus-
pended in serum-free medium consisting of a 1:1 mixture of DMEM-
F12 medium with N2 supplement and of Neurobasal medium with
B27 supplement. This medium also contained 8-(4-chlorophe-
nylthio) adenosine 3?5?-cyclic monophosphate (cpt-cAMP, 0.1 mM),
forskolin (25 ?M), N-acetyl-L-cystein (6.3 mg/ml), insulin (20 ?g/ml),
FGF-2 (10 ng/ml), EGF (50 ng/ml), BDNF (10 ng/ml), NT-3 (10 ng/ml),
and penicillin/streptomycin. The cell suspension was filtered twice
through an 8 ?m nylon mesh to obtain a single-cell suspension.
Between 3000 and 5000 cells were plated in T-25 Falcon flasks
coated with poly-L-lysine (10 ?m/ml) and laminin (10 ?g/ml). After
a few hours, clumps of more than one cell were ringed and excluded
from the study. Such clumps represented less than 1% of the
Retinal Explant Cultures and Retroviral Infection
E16–17 rat retinal explants were prepared as previously described
(Cayouette et al., 2001). The explants were allowed to settle for a
few hours in a CO2incubator at 37?C before they were infected
with a retroviral vector encoding either enhanced green fluorescent
protein (GFP) or placental alkaline phosphatase (PLAP). Retroviral
vectors were prepared and used to infect explants as described
previously (Cayouette and Raff, 2003).
Histology and Immunostaining
The retinal explants were fixed and cryosectioned after 10 days in
culture as previously described (Cayouette et al., 2001). The follow-
ing antibodies were used for immunofluorescence: monoclonal
mouse anti-syntaxin (1:1000; Sigma); monoclonal mouse anti-islet-1
(produced by T. Jessell and obtained from the Developmental Stud-
and monoclonal mouse anti-cyclin D3 (1:100, Santa Cruz Biotech.).
Primary antibodies were detected using the appropriate Alexa Fluor
488 or 594 goat antibodies (Molecular Probes). In all cases, we
added to the culture at a concentration of 10 ?M. After 7 days
in culture, cells were fixed and BrdU incorporation detected as
described (Neophytou et al., 1997).
We are grateful to Gord Fishell for the CLE retroviral vector; Bill
Harris for insightful comments; and members of the Raff and Barres
labs for stimulating discussions and support. This work was funded
by a Long-Term Fellowship from the Human Frontier Science Pro-
gram Organization and a senior postdoctoral fellowship from the
Canadian Institute of Health Research (M.C.), the NIH-NEI (B.A.B,
grant R01 EY11310), the March of Dimes Foundation (B.A.B, grant
1FY01-352), and the Medical Research Council UK (M.R.).
Received: October 1, 2002
Revised: June 18, 2003
Accepted: November 12, 2003
Published: December 3, 2003
Akagawa, K., and Barnstable, C.J. (1986). Identification and charac-
terization of cell types in monolayer cultures of rat retina using
monoclonal antibodies. Brain Res. 383, 110–120.
nization and cell differentiation in rat retinal reaggregate cultures.
Brain Res. 437, 298–308.
Alexiades, M.R., and Cepko, C.L. (1997). Subsets of retinal progeni-
tors display temporally regulated and distinct biases in the fates of
their progeny. Development 124, 1119–1131.
activity is required for rod photoreceptor development in vitro. De-
velopment 114, 947–957.
Barnstable, C.J., Hofstein, R., and Akagawa, K. (1985). A marker of
early amacrine cell development in rat retina. Brain Res. 352,
control the genesis of amacrine and cone cells in the rat retina.
Development 126, 555–566.
Broadus, J., and Doe, C.Q. (1997). Extrinsic cues, intrinsic cues and
microfilaments regulateasymmetric proteinlocalization inDrosoph-
ila neuroblasts. Curr. Biol. 7, 827–835.
Cayouette, M., and Raff, M. (2003). The orientation of cell division
influences cell-fate choice in the developing mammalian retina. De-
velopment 130, 2329–2339.
of the pigmented epithelium. J. Neurosci. 21, 5643–5651.
Dyer, M.A., and Cepko, C.L. (2000). Control of Muller glial cell prolif-
eration and activation following retinal injury. Nat. Neurosci. 3,
Furst, A., and Mahowald, A.P. (1985). Cell division cycle of cultured
neural precursor cells from Drosophila. Dev. Biol. 112, 467–476.
Galli-Resta, L., Resta, G., Tan, S.S., and Reese, B.E. (1997). Mosaics
of islet-1-expressing amacrine cells assembled by short-range cel-
lular interactions. J. Neurosci. 17, 7831–7838.
Greferath, U., Grunert, U., and Wassle, H. (1990). Rod bipolar cells
in the mammalian retina show protein kinase C-like immunoreactiv-
ity. J. Comp. Neurol. 301, 433–442.
Harris, W.A., and Messersmith, S.L. (1992). Two cellular inductions
involved in photoreceptor determinationin the Xenopus retina. Neu-
ron 9, 357–372.
Holt, C.E., Bertsch, T.W., Ellis, H.M., and Harris, W.A. (1988). Cellular
determination in the Xenopus retina is independent of lineage and
birth date. Neuron 1, 15–26.
Huang, S., and Moody, S.A. (1995). Asymmetrical blastomere origin
and spatial domains of dopamine and neuropeptide Y amacrine
subtypes in Xenopus tadpole retina. J. Comp. Neurol. 360, 442–453.
Huang, S., and Moody, S.A. (1997). Three types of serotonin-con-
taining amacrine cells in tadpole retina have distinct clonal origins.
J. Comp. Neurol. 387, 42–52.
Isshiki, T.,Pearson, B.,Holbrook, S., andDoe, C.Q.(2001). Drosoph-
ila neuroblasts sequentially express transcription factors which
specify the temporal identity of their neuronal progeny. Cell 106,
Jensen, A.M., and Raff, M.C. (1997). Continuous observation of
multipotential retinal progenitor cells in clonal density culture. Dev.
Biol. 188, 267–279.
Lillien, L., and Cepko, C. (1992). Control of proliferation in the retina:
temporal changes in responsiveness to FGF and TGF alpha. Devel-
opment 115, 253–266.
Livesey, F.J., and Cepko, C.L. (2001). Vertebrate neural cell-fate
determination: lessons from the retina. Nat. Rev. Neurosci. 2,
Luer, K., and Technau, G.M. (1992). Primary culture of single ecto-
dermal precursors of Drosophila reveals a dorsoventral prepattern
of intrinsic neurogenic and epidermogenic capabilities at the early
gastrula stage. Development 116, 377–385.
lemot, F., and Gruss, P. (2001). Pax6 is required for the multipotent
state of retinal progenitor cells. Cell 105, 43–55.
Mears, A.J., Kondo, M., Swain, P.K., Takada, Y., Bush, R.A., Saun-
ders, T.L., Sieving, P.A., and Swaroop, A. (2001). Nrl is required for
rod photoreceptor development. Nat. Genet. 29, 447–452.
Moody, S.A., Chow, I., and Huang, S. (2000). Intrinsic bias and lin-
eage restriction in the phenotype determination of dopamine and
neuropeptide Y amacrine cells. J. Neurosci. 20, 3244–3253.
Morrow, E.M., Belliveau, M.J., and Cepko, C.L. (1998). Two phases
of rod photoreceptor differentiation during rat retinal development.
J. Neurosci. 18, 3738–3748.
Negishi, K., Teranishi, T., and Kato, S. (1982). New dopaminergic
and indoleamine-accumulating cells in the growth zone of goldfish
retinas after neurotoxic destruction. Science 216, 747–749.
Neophytou, C., Vernallis, A.B., Smith, A., and Raff, M.C. (1997).
Muller-cell-derived leukaemia inhibitory factor arrests rod photo-
receptor differentiation at a postmitotic pre-rod stage of develop-
ment. Development 124, 2345–2354.
Onoda, N., and Fujita, S.C. (1987). A monoclonal antibody specific
for a subpopulation of retinal bipolar cells in the frog and other
vertebrates. Brain Res. 416, 359–363.
Qian, X., Goderie, S.K., Shen, Q., Stern, J.H., and Temple, S. (1998).
Intrinsic programs of patterned cell lineages in isolated vertebrate
CNS ventricular zone cells. Development 125, 3143–3152.
Qian, X., Shen, Q., Goderie, S.K., He, W., Capela, A., Davis, A.A.,
sequence of neuron and glial cell production from isolated murine
cortical stem cells. Neuron 28, 69–80.
Rapaport, D.H., Rakic, P., and LaVail, M.M. (1996). Spatiotemporal
gradients ofcell genesisin theprimate retina.Perspect. Dev.Neuro-
biol. 3, 147–159.
Reh, T.A. (1992). Cellular interactions determine neuronal pheno-
types in rodent retinal cultures. J. Neurobiol. 23, 1067–1083.
Reh, T.A., and Tully, T. (1986). Regulation of tyrosine hydroxylase-
containing amacrine cell number in larval frog retina. Dev. Biol.
Sheedlo, H.J., and Turner, J.E. (1996). Effects of retinal pigment
epithelial cell-secreted factors on neonatal rat retinal explant pro-
genitor cells. J. Neurosci. Res. 44, 519–531.
Shen, Q., Zhong, W., Jan, Y.N., and Temple, S. (2002). Asymmetric
Numb distribution is critical for asymmetric cell division of mouse
cerebral cortical stem cells and neuroblasts. Development 129,
Till, J.E., McCulloch, E.A., and Siminovitch, L. (1964). A stochastic
model of stem cell proliferation based on the growth of spleen
colony-forming cells. Proc. Natl. Acad. Sci. USA 51, 29–36.
Turner, D.L., and Cepko, C.L. (1987). A common progenitor for neu-
rons and glia persists in rat retina late in development. Nature
Turner, D.L., Snyder, E.Y.,and Cepko, C.L. (1990). Lineage-indepen-
dent determination of cell type in the embryonic mouse retina. Neu-
ron 4, 833–845.
Voyvodic, J.T., Burne, J.F., and Raff, M.C. (1995). Quantification of
in cell lineage analysis. Eur. J. Neurosci. 7, 2469–2478.
Waid, D.K., and McLoon, S.C. (1998). Ganglion cells influence the
fate of dividing retinal cells in culture. Development 125, 1059–1066.
Watanabe, T., and Raff, M.C. (1990). Rod photoreceptor development
in vitro: intrinsic properties of proliferating neuroepithelial cells change
as development proceeds in the rat retina. Neuron 4, 461–467.
Watanabe, T., and Raff, M.C. (1992). Diffusible rod-promoting sig-
nals in the developing rat retina. Development 114, 899–906.
Wetts, R., and Fraser, S.E. (1988). Multipotent precursors can give
rise toall majorcell typesof the frogretina. Science239, 1142–1145.
Williams, R.W., and Goldowitz, D. (1992). Lineage versus environ-
Young, R.W. (1985). Cell differentiation in the retina of the mouse.
Anat. Rec. 212, 199–205.