Search tips
Search criteria 


Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
Dev Biol. Author manuscript; available in PMC 2011 April 15.
Published in final edited form as:
PMCID: PMC2854206

Neurog2 controls the leading edge of neurogenesis in the mammalian retina


In the mammalian retina, neuronal differentiation begins in the dorso-central optic cup and sweeps peripherally and ventrally. While certain extrinsic factors have been implicated, little is known about the intrinsic factors that direct this process. In this study, we evaluate the expression and function of proneural bHLH transcription factors during the onset of mouse retinal neurogenesis. Dorso-central retinal progenitor cells that give rise to the first postmitotic neurons express Neurog2/Ngn2 and Atoh7/Math5. In the absence of Neurog2, the spread of neurogenesis stalls, along with Atoh7 expression and RGC differentiation. However, neurogenesis is eventually restored, and at birth Neurog2 mutant retinas are reduced in size, with only a slight increase in the retinal ganglion cell population. We find that the re-establishment of neurogenesis coincides with the onset of Ascl1 expression, and that Ascl1 can rescue the early arrest of neural development in the absence of Neurog2. Together, this study supports the hypothesis that the intrinsic factors Neurog2 and Ascl1 regulate the temporal progression of retinal neurogenesis by directing overlapping waves of neuron formation.

Keywords: Neurog2, Atoh7, Ascl1, retinal ganglion cell, retina, neurogenesis


Visual processing in the retina depends on proper functioning of multiple neural classes. Thus, determining how this neuronal diversity arises is critical for understanding retinal function. Seven major retinal cell classes are generated between embryonic day (E) 11 and postnatal day (P) 10 in the mouse, in a conserved temporal order (Sidman, 1961; Young, 1985). In vertebrates, retinal ganglion cells (RGCs) differentiate first, as a wave front across the neuroepithelium of the optic cup (Easter, 2000; Holt et al., 1988; Masai et al., 2000; McCabe et al., 1999). In zebrafish, this wave begins near the optic stalk and radiates outward (Hu and Easter, 1999). In avians, the first RGCs appear in the dorsal-central retina, and neurogenesis simultaneously spreads peripherally and ventrally (Prada et al., 1991). Multiple extrinsic signals, including FGFs and sonic hedgehog, are required for the spatiotemporal progression of retinal neurogenesis (Jensen and Wallace, 1997; Macdonald et al., 1995; Martinez-Morales et al., 2005; McCabe et al., 1999; Neumann and Nuesslein-Volhard, 2000; Perron et al., 2003; Picker and Brand, 2005). However, little is known about the intrinsic factors that regulate this process.

The basic-loop-helix (bHLH) transcription factors, including Atoh7/Ath5, Ascl1/Ash1, Neurog2/Ngn2, and Neurod1, regulate multiple facets of neurogenesis, including cell cycle exit, neural versus glial determination, subtype specification, and survival (Ohsawa and Kageyama, 2008). Among the first proneural bHLHs expressed in the vertebrate retina, Atoh7 (atonal homologue 7) appears at the onset of retinal neurogenesis in the dorso-central mouse retina, and loss-of-function mutations result in the reduced differentiation of early progenitor cells and nearly complete loss of RGCs (Brown et al., 1998; Brown et al., 2001; Kanekar et al., 1997; Kay et al., 2001; Matter-Sadzinski et al., 2001; Wang et al., 2001). The vertebrate bHLH factor, Neurog2 (also an atonal homologue) is expressed during early retinogenesis (Brown et al., 1998; Ma and Wang, 2006). In the chick eye, Neurog2 can genetically activate Atoh7 and transdifferentiate cultured RPE cells into immature RGCs and photoreceptors (Matter-Sadzinski et al., 2005; Yan et al., 2001). By contrast, X-ngnr-1, a Xenopus Neurog2 homologue, promotes photoreceptor but not RGC formation (Perron et al., 1999). Recently, Neurog2/Ngn2 was demonstrated to bind to 5′ regulatory DNA and activate Atoh7/Ath5 transcription using distinct species-specific mechanisms in the mouse versus chick retina (Skowronska-Krawczyk et al., 2009). However, no individual role for Neurog2 has been uncovered, particularly in the mammalian retina (Akagi et al., 2004; Skowronska-Krawczyk et al., 2009).

In this report, we investigate intrinsic elements controlling the spatial and temporal onset of retinal neurogenesis, and define a novel role for Neurog2 during the outward expansion of retinal neurogenesis. Neurog2 and Atoh7 are simultaneously activated in cells that give rise to the first RGCs. Neurog2 is required for the spatial and temporal progression of both the expanding wave front and Atoh7 expression, but the resulting delay of neurogenesis is transient. The onset of Ascl1, a later-expressed bHLH factor, coincides with the restoration of retinal neurogenesis, and rescues neural differentiation in the absence of Neurog2. Together, these data demonstrate a critical role for bHLH factors in both propagating and maintaining the spatial and temporal progression of mammalian retinogenesis.

Materials and Methods


Neurog2GFP mice (Seibt et al., 2003) were maintained on an ICR background, and Atoh7LacZ (Brown et al., 2001), Ascl1KO/+ (Tomita et al., 1996) and Neurog2Ascl1KI mice (Fode et al., 2000) on a CD-1 background. For double-mutant studies, mice were bred together for a minimum of two generations. PCR genotyping was performed as described (Brown et al., 2001; Fode et al., 2000; Seibt et al., 2003; Tomita et al., 1996).

For embryonic studies, gestational age was determined by timed matings, with the date of the vaginal plug as E0.5. For somite counted embryos, 4-6 hour timed matings were carried out to precisely correlate somite number with gestational age. BrdU pulse labeling was performed by injecting pregnant dams with BrdU (0.1mg/g body weight of 10mg/mL BrdU in 0.9M NaCl) and harvesting embryos after 1.5 hours. P0.5 pups were collected on the morning after birth.

Immunohistochemistry and in situ hybridization

Immunohistochemistry was performed as described (Hufnagel et al., 2007). Antibodies used were rabbit anti-βIII-tubulin (Tubb3) (1:1000, Covance), rabbit anti-βgal (1:10000, Cappel), rat anti-βgal (1:1000, gift from Tom Glaser), rat anti-BrdU (1:100, AbD Serotec), goat anti-Pou4f2/Brn3b (1:50, Santa Cruz), rabbit anti-activated Caspase 3 (1:100, Cell Signaling), sheep anti-Chx10 (1:1000, Exalpha Biologicals), rabbit anti-GFP (1:1000, Molecular Probes), rabbit anti-Ascl1 (1:1000; Horton et al., 1999), rabbit anti-Neurog1 (1:1000, Gowan et al., 2001), rabbit anti-Neurog2 (1:1000), mouse anti-Neurog2 (1:10, Lo et al., 2002), mouse anti-Neurog3 (1:100, DSHB), mouse anti-p27 (1:200, Thermo scientific), rabbit anti-Pax6 (1:1000, Covance), rabbit anti-Pax2 (1:1000, Covance), mouse anti-AP2α (1:500, DSHB), rabbit anti-Prox1 (1:1000, Covance), rabbit anti-RXRγ (1:200, Santa Cruz), and rabbit anti-Sox2 (1:1000, Chemicon). Direct-conjugate secondary antibodies (Molecular Probes) or sequential biotinylated secondary (Jackson Immunoresearch) and streptavidin-conjugated Alexafluor tertiary antibodies (Molecular Probes) were used to visualize primary antibody labeling.

In situ hybridization was performed as described (Wallace and Raff, 1999). Briefly, embryos were collected and fixed in 4% PFA/PBS overnight, then cryoprotected in 30% sucrose overnight, embedded in 50:50 OCT:30% sucrose, and sectioned at a thickness of 10μm. DIG-labeled antisense Atoh7, Ascl1, and Neurod1 probes were hybridized to retinal sections overnight, detected with sheep anti-DIG antibody (1:2000; Roche), and developed with NBT and BCIP.

Measurements and cell counting

Microscopy was performed with a Zeiss fluorescent microscope, Zeiss camera and Apotome deconvolution device. For all retinal measurements or cell counts, a minimum of 3 embryos or postnatal pups per genotype from ≥2 independent litters were analyzed, matched for somite number across genotypes. Cell counts and measurements were performed using the Zeiss Axiovision software (v5.0), using the interactive events and curve spline tools. The circumference of the Tubb3 expression was compared to the circumference of the Neurog2-GFP domain and the total outer circumference from 4 images per animal, representing both eyes, containing the optic nerve or within 50μm dorsal to the optic nerve. The percentages of BrdU+/DAPI, act Caspase+/DAPI, RXRγ+/DAPI, Pou4f2+/DAPI, AP2α+/DAPI, Prox1+/DAPI nuclei were determined in 200X fields within in the central retina. Either a paired Student’s T test with Welch posthoc test or ANOVA with Tukey-Kramer posthoc test was used to determine p values (Instat Software, v3.0). Photoshop (v7.0) was used to adjust equally the brightness and contrast of images among different genotypes.


Comparison of Ngn2 and Atoh7 expression during early retinogenesis

During the initiation of retinal neurogenesis, progenitor cells exit the cell cycle, express general neuronal markers, and commit to a single cell fate. Among vertebrates, RGCs appear first (Altshuler et al., 1991), initially in the dorso-central retina of avians and mammals. In the chick retina, neurogenesis spreads in simultaneous central-peripheral and dorsal-ventral gradients, and is regulated partly by FGF signaling (McCabe et al., 1999; Prada et al., 1991). An analogous wave front in the mammalian retina has not been described, so we sought to understand the spatiotemporal kinetics of this process in the mouse eye, and test the hypothesis that bHLH factors Neurog2/Ngn2 and/or Atoh7/Math5 regulate the initial neurogenic wave.

The first Atoh7-expressing cells are found in the dorso-central retina at E11.0, preceding the appearance of RGCs that critically require this factor (Brown et al., 1998; Brown et al., 2001; Wang et al., 2001). Neurog2 expression has also been reported to appear around the time of neurogenesis initiation in the early chick and mouse retina (Ma and Wang, 2006; Matter-Sadzinski et al., 2005). First, we compared the expression pattern of Neurog2/Ngn2 to Atoh7/Math5 and the initial spread of retinal neuron differentiation. To correlate Neurog2 and Atoh7 expression directly, we assessed the onset of Atoh7LacZ with that of Neurog2GFP and Neurog2 protein expression (Brown et al., 2001; Seibt et al., 2003), by antibody double labeling of retina sections from double-heterozygous animals (Neurog2GFP/+;Atoh7LacZ/+ mice; Fig. 1A,B), which are identical to wild types (not shown). Brief timed matings (4-6 hours) were used to precisely correlate gestational ages of somite-counted E10.75-E12.0 embryonic litters. Prior to E11.0, Neurog2 protein and Neurog2GFP expression were localized to the ventral thalamus and presumptive optic stalk, but excluded from the retina (Fig. 1C and data not shown). The earliest retinal Neurog2+ and GFP+ cells were found at E11.0 in the dorso-central retina (43 somites, Fig 1E,E’). From E11.0-E11.5 (43-50 somites), 6 of 11 embryos contained both Neurog2+ and Neurog2GFP+ cells, indicating that retinal onset of Neurog2 does not precisely correlate with somite number. All embryonic retinas at E11.75 (51-60 somites) contained Neurog2+/GFP+ cells.

Figure 1
Onset of Neurog2 protein, Neurog2GFP and Atoh7LacZ expression in the mouse retina. A-B) Whole mount micrographs of Atoh7LacZ/+ and Neurog2GFP/+ embryos at E11.5, demonstrated retinal expression for each reporter (arrows). C-D) Immunolabeling for GFP and ...

Next, we asked if the onset of Neurog2 or Atoh7 expression precedes the other. We performed antibody labeling in double-heterozygous mice (Neurog2GFP/+;Atoh7LacZ/+) and Atoh7 mutants (Neurog2GFP/+;Atoh7LacZ/LacZ), since bi-allelic expression of Atoh7LacZ enhanced the detection of βgal+ cells. We do not observe either Neurog2GFP+ or βgal+ retinal cells prior to 43 somites (Fig. 1D), although co-labeled GFP+ and βgal+ cells were noted in the diencephalon (arrow, Fig. 1D). Neurog2GFP and Atoh7LacZ were extensively coexpressed in the mouse dorso-central retina at E11.0 (arrows, Fig. 1F,G). At E11.75 and E12.5, both Neurog2 and Atoh7 expression had expanded peripherally, with a bias towards the temporal/caudal retina (Fig. 1H,I). The Neurog2GFP domain always extended more peripherally and encompassed more cells than the Atoh7LacZ domain (GFP+/βgal− region in brackets, Fig. 1I). At all ages examined, virtually all βgal+ cells were also GFP+ (arrows, Fig. 1F-I), indicating Atoh7LacZ was expressed in a subset of Neurog2GFP+ cells.

To further examine the coincidence between Neurog2 and Atoh7, we compared the pattern of Neurog2 protein with Atoh7LacZ and Neurog2GFP. Neurog2 is largely present in S-phase progenitor cells (Fig. 1K; Ma and Wang, 2006; Yan et al., 2001). Atoh7/Ath5 is not expressed during S-phase (Fig. 1J) (Le et al., 2006; Poggi et al., 2005), and has been extensively reported to be expressed by late G2/M phase and postmitotic retinal cells (Brown et al., 1998; Le et al., 2006, Brzezinski, 2005; Yang et al, 2003). This implies that, in mitotically active retinal progenitor cells, Neurog2 expression in S-phase precedes that of Atoh7. Consistent with this difference, very few cells co-labeled with Neurog2 and βgal proteins (fuchsia and white arrows, Fig. 1L). The extensive overlap of Neurog2GFP and Atoh7LacZ likely occurs because Neurog2GFP persists longer than Neurog2 protein, thereby acting as a short-term lineage tracer (Britz et al., 2006). Indeed, while all Neurog2+ cells co-express Neurog2GFP (fuchsia and white arrows, Fig. 1M), many GFP+/Neurog2− cells are present (yellow and white arrowheads, Fig. 1M). We conclude that Neurog2 and Atoh7 simultaneously initiate expression in dorsal-central retinal progenitor cells at E11.0, but at distinct phases of the mitotic cell cycle.

Neurog2GFP precedes the expansion of neurogenesis and RGC specification

Prior to retinogenesis, the optic vesicle becomes compartmentalized into the neural retina, RPE, and optic stalk. Optic vesicle cells initially co-express the paired-homeobox transcription factors Pax6 and Pax2 (Baumer et al., 2003; Schwarz et al., 2000). Pax2 is subsequently downregulated in the neural retina, but not Pax6 (Baumer et al., 2003). Importantly, Pax6 directly activates Atoh7 and Neurog2 (Marquardt et al., 2001; Riesenberg et al., 2009; Willardsen et al., 2009). Before the onset of Neurog2GFP expression from E11.0-11.5, Pax2+ cells were detected throughout the optic cup and stalk (Fig. 2A). By E11.75, after Neurog2GFP onset in the retina, Pax2 protein was restricted to the optic stalk and central-nasal optic cup, in GFP-negative cells (Fig. 2B,C). The Neurog2GFP domain bordered that of Pax2, and very few GFP+/Pax2+ cells were noted (arrow, Fig. 2C). Therefore, Pax2 downregulation precedes the initiation of Neurog2 expression in the presumptive neural retina. This pattern of Pax2 expression was unchanged in Neurog2 mutants (not shown), indicating that Neurog2 does not suppress Pax2 retinal expression. Pax6 protein was co-expressed with all GFP+ cells at this age (Fig. 2D). Neurog2-GFP also co-localized with Sox2 and Chx10/Vsx2 proteins (not shown), two other transcription factors required for normal retinal progenitor differentiation (Burmeister et al., 1996; Taranova et al., 2006).

Figure 2
Initiation of both retinal neurogenesis and Neurog2GFP expression are coincident in the mouse eye. (A-C) Pax2 and Neurog2GFP co-labeling. At E11.0-E11.5, Pax2+ cells are evident in the optic cup (A). When Neurog2GFP+ cells first appeared in the central ...

Next, we directly compared Neurog2-GFP expression with the onset and expansion of retinal neuron differentiation. Co-labeling for GFP and Tubb3 (βIII-Tubulin), a neural-specific marker (Brittis et al., 1995; Lee et al., 1990), revealed no differentiating retinal neurons prior to Neurog2GFP onset (Fig. 2E). Pou4f2/Brn3b, a marker of specified RGCs (Xiang et al., 1993), and p27/Kip1, a cyclin-dependent kinase inhibitor that promotes cell cycle exit of retinal progenitor cells (Dyer and Cepko, 2001; Levine et al., 2000), were also absent prior to Neurog2 onset (not shown). From E11.0-11.5, the first Tubb3+ and p27+ cells were detected in Neurog2-GFP+ retinal cells (Fig. 2F and not shown). By E11.75, the Tubb3 domain extended from the dorsal to central retina (Fig. 2I,J) but was not present ventral to the forming optic nerve (Fig. 2K). From E12.0-E13.5, the Tubb3+ region expanded peripherally and ventrally, with bias towards the temporal retina (Fig. 2G,H). The spread of the Neurog2 domain preceded that of neural differentiation, indicated by the peripheral subdomain of GFP+/Tubb3− cells (brackets, Fig. 2G,J), which likely represents proliferating cells that subsequently differentiate into retinal neurons. Differentiating neurons highly coexpressed Tubb3 and p27 (Fig. 2L), verifying the concurrence of cell cycle exit and neural differentiation in the earliest retinal neurons. Essentially all Tubb3+ and p27+ cells co-labeled with Neurog2GFP (Fig. 2F-J and not shown). At E11.75, the first GFP+/Pou4f2+ RGCs were detected in the dorso-central retina, proximal to the leading edges of the Neurog2GFP and Tubb3 domains (Fig. 2M and not shown). No Pou4f2+ cells were detected prior to E11.75 (not shown). We conclude that the onset and peripheral expansion of Neurog2 expression precedes the initiation of neurogenesis and subsequent differentiation of the first RGCs.

From E11.0-E13.5, neurogenesis spreads outward across the neural retina, excluding the optic nerve head and peripheral retina that give rise to the ciliary body and iris (Rodieck, 1998). Since Neurog2 expression correlates with the onset of neural differentiation, we predicted that Neurog2 expression would only be present in cells undergoing retinal neurogenesis. From E13.5 to birth, Neurog2GFP co-localizes with Tubb3+ cells in the neuroblastic layer (NBL) and the inner forming ganglion cell layer (GCL), excluding the optic nerve head and presumptive ciliary body (Fig. (Fig.2H,2H, 4B-C and not shown). Previous analysis of the Neurog2-lineage revealed that Neurog2-expressing cells are capable of adopting all the retinal fates (Ma and Wang, 2006). While that study found RGCs arise from the Neurog2-lineage starting at E14, here we found GFP+/Pou4f2+ RGCs much earlier, at E11.75 (Fig. 2M), suggesting that Neurog2GFP acts as a short-term lineage-tracer without the delay of Cre-mediated reporter activation by Neurog2CreER (Ma and Wang, 2006). We then compared GFP expression with markers of other embryonic retinal cell types: cones (RXRγ), horizontals (Prox1), and amacrines (AP2α) (Dyer et al., 2003; Mori et al., 2001; Yan and Wang, 2004). RXRγ+/GFP+ cone photoreceptors were detected in the outer retina at E13.5 (arrows, Fig. 2N), also with a bias for the temporal retina. GFP+ amacrine (Ap2α+) and horizontal (Prox1+) interneurons were also noted in the prenatal retina (Fig. 2O,P and data not shown).

Figure 4
Arrested neurogenesis in the absence of Neurog2 is temporary. (A,A’) At E12.0, the peripheral extent of the Tubb3 domain is reduced in Neurog2GFP/GFP retinas compared to controls (brackets, A,A’). (B,B’) At E13.5, the peripheral ...

Neurog2 is required for the peripheral expansion of retinal neurogenesis

Neurog2 expression at the leading edge of retinal neurogenesis precedes the expansion of Atoh7, neural commitment, and RGC differentiation. Therefore, we asked if Neurog2 is required for the peripheral propagation of neural development. GFP is still expressed in the absence of Neurog2 (Neurog2GFP/GFP, Fig. 3B), and marks the lineage of Neurog2-mutant cells. Thus, Neurog2GFP allows for comparison of the peripheral extent of reporter-expressing cells (GFP+ domain) and nascent neurons (Tubb3+ domain) in heterozygous (Neurog2GFP/+) and mutant (Neurog2GFP/GFP) retinas. To confirm that the size of the GFP domain is not different for single or bi-allelic GFP expression, we compared the GFP domain relative to the total retinal circumference in Neurog2GFP/+ and Neurog2GFP/GFP and found no difference between genotypes (Fig. 3H; See Methods for description of domain measurements). To determine whether loss of Neurog2 and/or Atoh7 affects the ventral-peripheral expansion of retinal neurogenesis, we examined double-heterozygote controls (Neurog2GFP/+;Aoth7LacZ/+), Neurog2 single-mutant (Neurog2GFP/GFP;Atoh7LacZ/+), Atoh7 single-mutant (Neurog2GFP/+;Atoh7LacZ/LacZ), and double-mutant (Neurog2GFP/GFP;Atoh7LacZ/LacZ) retinas. The double-heterozygotes are appropriate controls since Atoh7 heterozygotes have no phenotypes compared to wild types (Brown et al., 2001; Le et al., 2006; Wang et al., 2001), and both Neurog2GFP/+ and Neurog2GFP/+;Atoh7LacZ/+ retinas exhibited no significant differences from wild type eyes (not shown).

Figure 3
Delay of early neurogenesis in Neurog2 mutants. (A-D) Tubb3 and Neurog2GFP labeling of double-heterozygote controls (A), Neurog2GFP/GFP;Atoh7LacZ/+ (Neurog2 mutants) (B), Neurog2GFP/+; Atoh7LacZ/LacZ (Atoh7 mutants) (C), and double mutant (D) embryos ...

First, we evaluated the expansion of Tubb3+ cells in relation to the Neurog2GFP domain in somite-matched embryos at E11.75 (54-60 somites). In control retinas, the Tubb3 domain was slightly smaller and included within the GFP domain (Fig. 3A, brackets 3A’). In Neurog2 mutants, the Tubb3 domain was decreased (Fig. 3B), with a greater separation between the leading edge of Tubb3+ cells and the peripheral extent of the Neurog2GFP domain (brackets, Fig. 3B’). To quantify the peripheral spread of neurogenesis, we measured the outer length of the Tubb3 and GFP domains in matched central retinal sections (Fig. 3E). At E11.75, the Tubb3 domain was significantly reduced in Neurog2 mutant retinas. In controls, the Tubb3 domain occupied 84.7±2.5% of the GFP domain (31.4±1.4% of total circumference), but in Neurog2 mutants, the Tubb3 domain was only 34.8±7.0% of the GFP domain (11.9±2.3% of total circumference; Fig. 3F and not shown). We also noted reduced neurogenesis in the nasal half of the retina and ventral to the optic nerve (not shown). In addition to Tubb3, we also observed reduced p27/Kip1 and Pou4f2/Brn3b expression domains in the absence of Neurog2 (Fig. 3I-J’). We conclude that Neurog2 mutants exhibit a reduction in retinal neurogenesis concomitant with reduced RGC specification and cell cycle exit.

Intriguingly, Atoh7 is not required to propagate the spread of neurogenesis, as the size of the Tubb3 domain was unaffected in the absence of Atoh7 (Fig. 3C,C’,F and not shown). Like Neurog2 mutants, mice lacking both Atoh7 and Neurog2 had diminished expansion of Tubb3 in relation to both the GFP domain and total circumference, though not different from Neurog2 single mutants (Fig. 3D,D’,F). Therefore, Atoh7 and Neurog2 do not work synergistically to promote the propagation of neurogenesis. To investigate further, we assessed the percentage of differentiating neurons within the Tubb3 domain and observed fewer Tubb3+ cells in Neurog2 mutants, Atoh7 mutants, and double mutants compared to controls, again in a non-synergistic manner (Fig. 3G). Previous studies indicate that at E11.5 p27+ postmitotic retinal cells and Pou4f2+ RGCs are significantly reduced in Atoh7 mutants (Le et al., 2006; Wang et al., 2001). Although Atoh7 mutants do exhibit fewer p27+ cells, the peripheral extent of the p27 domain was not reduced (not shown). As expected, the Pou4f2/Brn3b-expressing cells were virtually absent in Atoh7 mutants and Neurog2;Atoh7 double mutants (not shown). Thus, the expansion of neurogenesis requires Neurog2, but not Atoh7, although each is required to produce normal numbers of differentiating neurons.

The co-localization of Neurog2 and Atoh7 reporters and the reduced propagation of neurogenesis in Neurog2 mutants from E11.0-E11.75 suggested that the peripheral spread of endogenous Atoh7 might also be affected. Indeed, Neurog2 mutants had a reduction in the width of the Atoh7 mRNA expression domain at E11.75 (Fig. 3K,K’). Further, in the diencephalon, Atoh7-expressing cells were virtually absent in Neurog2 mutants (arrows, Fig. 3K,K’). Another early bHLH factor, Neurod1, is required for normal amacrine, S-cone, and rod photoreceptor development (Inoue et al., 2002; Liu et al., 2008; Morrow et al., 1999). Although the Neurod1 and Atoh7 domains were the same width in controls, we did not observe any appreciable changes in the Neurod1 expression domain in Neurog2 mutant retinas (Fig. 3L,L’), consistent with a previous study (Akagi et al., 2004). Together, the outward spread of Neurog2 specifically affects the expansion of Atoh7 but is not required for its initial activation.

Delayed neurogenesis in Neurog2 mutants is restored

Neurog2 is required for the propagation, but not the initiation of neurogenesis, as a cluster of neural precursor cells appears in Neurog2 mutants between E11.0-E11.75. Next, we analyzed retinal development in these mice from E12.0 to E15.5. From E12.0-E12.5, the Tubb3, p27/Kip1, and Pou4f2/Brn3b domains were truncated relative to the Neurog2GFP domain (brackets, Fig. 4A,A’ and not shown). We also noted reduced neurogenesis in the nasal half of the retina and ventral to the optic nerve (not shown), indicating that the progression of neurogenesis was affected in both central-peripheral and dorsal-ventral axes. However, by E13.5 the pattern of neurogenesis in Neurog2 mutants was very similar to that of controls. In the temporal retina of both genotypes, the neurogenic domain extended to the periphery, to the border of the Neurog2-GFP domain (Fig. 4B,B’). On the nasal side, however, the Tubb3 and Pou4f2/Brn3b domains were still reduced relative to the GFP domain in Neurog2 mutants (brackets, Fig. 4B,B’ and not shown). By E15.5, the central to peripheral distribution of Tubb3+ or Pou4f2/Brn3b+ cells throughout the retina had caught up to that of controls (Fig. 4C,C’ and not shown). Thus, neurogenesis and RGC specification are restored in Neurog2 mutants, largely between E12.5 and E15.5.

Neurog2 and Atoh7 coordinate normal retinal size, but control distinct aspects of fate determination

At the initiation of retinal neurogenesis, Neurog2 is required for the propagation of the Atoh7 expression domain. Therefore, to test for cross-regulation or synergistic activities between Neurog2 and Atoh7, we compared the four earliest retinal fates (RGCs, cone photoreceptors, amacrine and horizontal interneurons) in Neurog2 and Atoh7 single and double mutants. Since loss of Neurog2 results in neonatal lethality, mutant mice were analyzed at P0.5.

First, we examined retinal thickness of single and double mutants (Fig. 5A-D,M). Adult Atoh7 mutants have reduced laminar thickness (Brown et al., 2001; Brzezinski et al., 2005), already present at P0.5 (Fig. 5C,M). Compared to controls (Fig. 5A), Neurog2 mutant mice also had significantly thinner retinas (Fig. 5B,M), similar to Atoh7 mutants (Fig. 5C,M). Furthermore, Neurog2;Atoh7 double mutant retinas were significantly reduced in thickness compared to both wild types and single mutants (Fig. 5D,M). This indicates that the loss of both Neurog2 and Atoh7 has an additive effect on retinal size, presumably representing synergistic or parallel roles in proliferation and/or survival during embryonic retinogenesis. To understand if reduced proliferation or increased cell death are responsible for the smaller retinas, we analyzed proliferating S-phase retinal progenitors by BrdU pulse-labeling cells, and apoptotic cells by activated Caspase-3 expression at several embryonic ages. Atoh7 single mutants had no defect in proliferation or apoptosis at E15.5 (Le et al., 2006). At both E11.5 and E15.5, there was no difference in BrdU+ cells between wild type, Neurog2 mutant, and Neurog2;Atoh7 double mutant retinas (Fig. 6A-D and data not shown). The percentage of Caspase-3+ cells was normal in E15.5 Neurog2−/− eyes (Fig. 6E-H), as well as at E12.0 during the delay in neurogenesis (not shown). However, the number of apoptotic cells was significantly increased in Neurog2;Atoh7 double mutants (Fig. 6H), suggesting an overlapping function for these bHLH factors in regulating some aspect of cell survival. Therefore, the increased apoptosis and enhanced reduction of retinal thickness were consistent with one another in double mutants.

Figure 5
Comparison prenatal retinal cell types in Neurog2, Atoh7 and Neurog2;Atoh7 double mutants. (A-D) Retinal thickness was measured as the vitreal-scleral width of DAPI+ nuclei in the NBL and GCL at P0.5. (E-H) Cone precursors were assessed by RXRγ ...
Figure 6
Cell -proliferation and apoptosis in the absence of Neurog2. (A-D) The percentage of BrdU+ S-phase cells at E15.5 was normal in Neurog2 mutants (Neurog2GFP/GFP;Atoh7LacZ/+) and double-mutants (Neurog2GFP/GFP;Atoh7LacZ/LacZ). (E-H) The percentage of activated ...

To understand the extent by which the four early cell types might be altered in Neurog2 mutants, we quantified RGCs, cones, horizontal and amacrine interneurons in P0.5 retinas. Although a loss of RGCs might be expected since their progression was delayed from E11.5-E13.5, we instead found a 2% ± 0.2% increase in Pou4f2+ RGCs within P0.5 Neurog2 mutants (Fig. 5K). As expected, Atoh7 and Atoh7;Neurog2 double mutants had essentially no Pou4f2/Brn3b+ RGCs at this age (not shown). We conclude that although the percentages of RGCs in P0.5 Neurog2 mutant eyes are significantly elevated, this phenotype cannot overcome the agenesis of RGCs in the absence of Atoh7.

Cone photoreceptors and Neurog2+ progenitor cells are significantly increased in Atoh7 mutants (Brown et al., 2001; Brzezinski et al., 2005; Le et al., 2006), suggesting that cone photoreceptor genesis might normally be blocked by Atoh7 indirect suppression of Neurog2 expression. Analysis of single and double-mutants (Neurog2GFP/GFP; Atoh7LacZ/LacZ) showed the trend of increased RXRγ+ cone precursor cells in the outer retinas of Atoh7LacZ/LacZ and double mutant mice (Fig. 5G,N). However, loss of Neurog2 alone (Neurog2GFP/GFP) had no significant effect on the percentage of cone photoreceptors (Fig. 5F,N), nor did it enhance or suppress the percentages of cones in Atoh7;Neurog2 double mutants (Fig. 5H,N). This suggests that although there is a simultaneously nonautonomous increase in cone photoreceptors and Neurog2+ cells in Atoh7 mutants (Le et al., 2006), these are independent events that are likely to occur in separate populations of retinal progenitor cells.

Characterization of Neurog2;Ascl1;Atoh3 and Neurog2;Neurod1;Atoh3 triple mutant mice suggested a partial requirement for Neurog2 during horizontal and amacrine interneuron differentiation (Akagi et al., 2004). To determine if loss of Neurog2 alone affects these cell types, we quantified the percentages of AP2α+ amacrine cells (arrow, Fig. 2O)(Yan and Wang, 2004) and Prox1+ cells, which give rise to a mixed population of horizontal and amacrine neurons (arrow, Fig. 2P)(Dyer et al., 2003). The AP2α protein (Figs. 5I-L,P-R) is expressed by both displaced amacrines in the GCL and amacrines that reside in the INL. We found normal distributions and percentages of AP2α+ amacrines in Neurog2 single mutants (Figs 5J,P-R). However, Atoh7 single mutants, and the double mutants had significant increases in amacrines (Figs 5K,L,P-R), consistent with a previous analysis of amacrines in Atoh7 mutants (Wang et al., 2001). Finally, we compared the percentages of Prox1+ horizontals and amacrines (Fig 5S-U). Here we observed only a significant increase in Prox1+ displaced amacrines in Atoh7;Neurog2 double mutants (Figs 5S-U). The different outcomes between Prox1+ and AP2α amacrines in Atoh7 single and double mutants probably resulted because the Prox1+ population (0.8%) is such a small subset of AP2α amacrines (29%). Regardless, we conclude that Neurog2 alone is not required for the specification of prenatal cone, amacrine and horizontal interneurons.

Ascl1 can compensate for the loss of Neurog2

Removal of Neurog2 during embryonic retinal development results in a temporal delay of early retinal neurogenesis, which then returns to normal between E12.5-E15.5 (Fig. 4). Therefore, it is plausible that other factors, for example another bHLH proneural factor, compensate for loss of Neurog2 in the early retina. We tested several such candidates here. First, at E15.5 the patterns of Atoh7 and Neurod1 mRNA were indistinguishable in control and Neurog2 mutants (not shown). If one of these factors compensates for the loss of Neurog2, we should have observed overexpression of Atoh7 or Neurod1. Next, other neurogenin gene family members, Neurog1 and Neurog3, are expressed in the chick retina, but not that of frog (Ma et al., 2009; Nieber et al., 2009). So we asked whether either paralogue might be ectopically upregulated in Neurog2 mutant eyes, but neither Neurog1 nor Neurog3 protein were detectable in E11.75-E15.5 control and Neurog2−/− retinas (not shown). Finally, we evaluated the onset of Ascl1 expression in wild type retinas. In the E11.5 optic cup, Ascl1 mRNA and protein are not expressed (Fig 7I and data not shown), but beginning at early E12.5, a small population of Ascl1+ cells is detectable in the dorso-central retina (arrow, Fig. 7A). By E13.5, Ascl1 expression has spread outward to the peripheral and ventral poles of the retina (not shown). Therefore, the normal onset and progression of Ascl1 expression coincides both spatially and temporally with the recovery of neurogenesis observed in Neurog2 mutants from E12.5-E13.5.

Figure 7
Ascl1 rescue of delayed neurogenesis in Neurog2 mutants. (A,A’) At E12.0, only rare Ascl1+ cells are present in Neurog2GFP/+ eyes. The Ascl1+ cell in A coexpresses GFP, thus is in the Neurog2 lineage (arrow in A’. and higher magnification ...

The timing of these events suggested that Ascl1 might be capable of restoring the delayed neurogenesis of Neurog2 mutants. If so, then misexpression of Ascl1 within the Neurog2-lineage should restore the peripheral expansion of retinal neurogenesis. To test this directly, we took advantage of the Neurog2Ascl1KI allele, a homologous recombination of an IRES-Ascl1 cassette into the endogenous Neurog2 gene locus, thereby functionally replacing Neurog2 with Ascl1 (Fode et al., 2000). By mating Neurog2Ascl1KI/+ and Neurog2GFP/+ heterozygotes, Neurog2GFP/Ascl1KI embryos were generated, in which Neurog2 function was removed and replaced by that of Ascl1 within the Neurog2-lineage (Fode et al., 2000). At E12.0, both Neurog2GFP/+ and Neurog2GFP/GFP retinas exhibited only rare Ascl1+ cells by immunofluorescence (Fig. 7A,A’,A” and not shown), indicating that Ascl1 is not precociously expressed in the absence of Neurog2. By contrast, Neurog2GFP/Ascl1KI retinas had abundant numbers of ectopic Ascl1+ cells (Fig. 7B,B’,B”), most of which were also GFP+, indicating a substitution of Ascl1 in cells that normally express Neurog2. We then compared the width of the Tubb3 and GFP domains in Neurog2GFP/+, Neurog2GFP/GFP, and Neurog2GFP/Ascl1KI embryonic retinas at E12.0. Strikingly, upon Ascl1 replacement of Neurog2, the width of the Tubb3+ domain was now the same as in controls (Fig. 7C-E). To determine the effects of Ascl1 on RGC differentiation, we similarly evaluated the Pou4f2/Brn3b expression domain in these three genotypes. Indeed, the width of the Pou4f2/Brn3b domain in Neurog2GFP/Ascl1KI retinas was identical to controls (Fig. 7F-H). Thus, although Ascl1 normally activates in the retina after Neurog2, it is sufficient to rescue the block in the progression of early neurogenesis and RGC differentiation found in Neurog2 mutant eyes.

Because progression of the Atoh7 domain is initially delayed in Neurog2 mutants from E11.75-E12.5 (Fig. 3K,K’), we asked whether Ascl1 rescues the neurogenic wave via activation of Atoh7. The expression of Atoh7 was compared among E11.5-E12.5 Neurog2+/+, Neurog2GFP/+, Neurog2GFP/GFP and Neurog2GFP/Ascl1KI litters (Figs. 7I-P and not shown). To verify the presence of ectopic Ascl1, its expression was monitored on adjacent sections from each embryo. At E11.5, Neurog2GFP/Ascl1KI optic cups had a reduced domain of Atoh7 mRNA (compare Figs 7K,L), but, ectopic Ascl1 was not yet present (Figs 7I,J). A day later at E12.5, when Ascl1 is normally expressed by a few retinal cells, we found abundant ectopic expression in Neurog2GFP/Ascl1KI retinas (Figs 7M,N), along with a normal pattern of Atoh7 mRNA (Figs 7O,P). Although ectopic Ascl1 and Atoh7 mRNA both appeared at E12.5, we do not think that rescue occurred at the level of Atoh7 transcriptional regulation. In support of this idea, loss of Ascl1 has no effect on Atoh7 mRNA expression from E11.5 and E15.5 (Suppl Figs 1A-D), the Ascl1 protein does not bind to Atoh7 5′ regulatory DNA (Skowronska-Krawczyk et al., 2009), and Drosophila Scute and Atonal proteins have different E Box binding site consensus sequences (Powell et al., 2004). Somewhat paradoxically, at E17.5 Ascl1 was proposed to suppress Atoh7 (Akagi et al., 2004), although Atoh7 upregulation was only found in the retinas of two bHLH triple mutant combinations that included Asc1l mutants. Furthermore, Ascl1 and Neurog2 mutually suppress each other’s mRNA expression in the E17.5 retina (Akagi et al., 2004), which is somewhat at odds with normal expression of Ascl1 in E11.5-E15.5 Neurog2 mutants (Suppl1 Figs 1E,F), and of Neurog2 GFP or protein in E11.5-E15.5 Ascl1-/- retinas (not shown). Although particular bHLH factors can suppress one another’s expression at older stages of retinal formation, there is no evidence that these regulatory interactions are direct (Akagi et al., 2004). We hypothesize that late embryonic retinal bHLH cross suppression involves intermediate genes and/or occurs nonautonomously, particularly since these factors do not encode transcriptional repressors.


Here, we investigated bHLH transcription factor expression and function during the initiation of retinogenesis in mouse, and identify Neurog2 as one intrinsic regulator of the leading edge of neurogenesis. Onset and expansion of Neurog2 and Atoh7 expression predicts the initial wave front, concomitant with the compartmentalization of the neural retina and optic stalk by Pax6 and Pax2, respectively. The first RGCs are subsequently specified in the dorso-central retina, and differentiation spreads ventrally and peripherally, similar to that found in fish and chick (Hu and Easter, 1999; Prada et al., 1991). Neurog2 is required for the propagation of neurogenesis, and though its loss initially causes a dramatic phenotype, retinal neurogenesis becomes corrected in a few days. At P0.5, mutant retinas exhibited only a minor increase in RGCs, with no defect in cone, amacrine or horizontal neuron genesis. Interestingly, this recovery occurred during the onset and expansion of Ascl1 expression, which was sufficient to correct the initial delay in RGC genesis.

Conservation of proneural bHLH function during initiation of murine retinal neurogenesis

The initial wave of retinal neurogenesis in mouse closely resembles the same process in non-mammalian vertebrate and Drosophila eyes. In fruit flies, a morphogenetic furrow sweeps across the eye imaginal disc from posterior to anterior ahead of retinal neurogenesis (Ready et al., 1976). At the anterior edge of the morphogenetic furrow, the bHLH protein atonal specifies the first ommatidial photoreceptor (R8) and promotes the progression of the morphogenetic furrow (Brown et al., 1995; Jarman et al., 1994; Jarman et al., 1995). Like in Drosophila, the progression of neurogenesis in the vertebrate retina exhibits wave-like properties. In zebrafish, cells cease proliferation and adopt an RGC fate in a nasal-to-temporal sequence, determined by the atonal-orthologue Ath5/lakritz (Hu and Easter, 1999; Kay et al., 2001). In chick, RGC differentiation proceeds outward from the optic stalk, with a bias for the temporal half of the retina (McCabe et al., 1999; Prada et al., 1991). In chicken, neurogenin2 and Ath5 expression are present in the central retina at the onset of neurogenesis, and microarray profiling of mouse retinal progenitor cells identified a subpopulation with Neurog2 and Atoh7 mRNA coexpression (Trimarchi et al., 2008b). More recently, chick and mouse Ngn2/Neurog2 were shown to activate directly the Ath5/Atoh7 promoter, although the number of binding sites utilized differs between these two species (Matter-Sadzinski et al., 2001; Matter-Sadzinski et al., 2005; Skowronska-Krawczyk et al., 2009).

Here, we show that these atonal family members have distinct functions in mouse where Neurog2 controls the propagation of neurogenesis, and Atoh7 regulates RGC specification. Not surprisingly, together Atoh7 and Neurog2 reconstitute the orthologous roles of atonal in the Drosophila eye. Subdivisions of atonal functions during vertebrate development were already known, since the semiorthologues Atoh7 and Atoh1 are present in mutually exclusive regions of the nervous system, thereby parsing Drosophila atonal functions within the mouse visual, auditory, and proprioceptive systems, respectively (Helms et al., 2000; Hufnagel et al., 2007; Saul et al., 2008).

Another example of functional subdivision relates to the ability of Drosophila atonal to autoregulate its own expression, which does not occur for the Xenopus Ath5 or mouse Atoh7 genes (Hutcheson et al., 2005; Riesenberg et al., 2009). Previously, Atoh7 was reported to suppress Neurog2 expression nonautonomously in the E13-15 retina (Le et al., 2006). Here, we found that the earliest Atoh7LacZ+ cells are also in the Neurog2GFP lineage, and that Neurog2 is present in S-phase cells, slightly preceding Atoh7 expression in these cells as they become newly postmitotic. We conclude that Neurog2 is a positive regulator of Atoh7 expression, since the peripheral expansion of Atoh7 was delayed in Neurog2 mutants. Thus, in mouse these two genes cross-regulate one another, but at different stages of retinal neurogenesis. During the initial propagation of neurogenesis, Neurog2 directly activates Atoh7 expression (this paper and Skowronska-Krawczyk et al., 2009), but several days later Atoh7 nonautonomously suppresses Neurog2 expression (Le et al., 2006). Importantly, like atonal autoregulation within committed R8 cells in the morphogenetic furrow, Neurog2 cross-regulation of Atoh7 is an integral part of wave front progression during the initiation of mammalian retinal neurogenesis.

Directing the wavefront of mammalian retinal neurogenesis

Neurog2 expression expands peripherally ahead of multiple markers of retinal neurogenesis. This small Neurog2GFP+/Tubb3-negative domain likely contains Neurog2+ cells in S-phase. As these cells progress through the terminal mitosis, a subset of Neurog2GFP+ cells express Atoh7, p27/Kip1 and Tubb3. Therefore, the spatial difference between the GFP+/Tubb3+ and peripheral GFP+/Tubb3-negative domains likely reflects the temporal difference in cell cycle status between differentiating neurons and proliferating progenitors poised to differentiate, respectively. This is also supported by Pou4f2/Brn3b onset more centrally in newly postmitotic RGCs. Hence, the outward spread of Neurog2 expression demarcates the leading edge of neurogenesis, in which progenitor cells exit the cell cycle and become specified as retinal neurons, most of which differentiate as RGCs.

We predicted that BrdU+ S phase progenitors would be increased in E11.5 Neurog2-/- eyes, since there was an obvious reduction in p27/Kip1+ postmitotic cells. The correlation of these outcomes would indicate that Neurog2 regulates retinal cell cycle progression, however this was not the case. It remains plausible that E11.5-E13.5 Neurog2 −/− cells inappropriately accumulate in G2 phase. However, we currently favor a different possibility in which Neurog2 mutant cells undergo transient changes in cell cycle length. Determining percentages of individual cell cycle markers at single time points would not uncover this defect. Instead, window labeling should be employed in the future to measure the cell cycle length of GFP+ retinal progenitors in Neurog2GFP/+, Neurog2GFP/GFP and Neurog2GFP/Ascl1KI retinas. In this regard, Ascl1 may uniquely rescue the Neurog2 phenotype, since mitotically active retinal progenitors appear to only express these two bHLH factors during embryonic retinal neurogenesis. Moreover, Neurod1 only partially rescues the Atoh7 RGC phenotype, and Atoh3 not at all (Mao et al., 2008), while Ascl1 cannot rescue the Atoh7 RGC phenotype (Hufnagel et al, in prep).

Extrinsic signal pathways, like FGF and sonic hedgehog (shh), direct key aspects of retinal patterning and neurogenesis (Martinez-Morales et al., 2005; McCabe et al., 1999; Neumann and Nuesslein-Volhard, 2000; Picker and Brand, 2005). A decade ago, shh was shown to propagate a retinal wave in the zebrafish retina (Neumann and Nuesslein-Volhard, 2000), but the mechanism for this subsequently underwent modification. Ath5 expression and RGC genesis were subsequently shown to initiate normally in sonic you (syu) mutants (Kay et al., 2005; Masai et al., 2005). However, the period for the retinal wave to progress from nasal to temporal becomes extended when postmitotic retinal neurons are unable to secrete Shh. Therefore, retinal shh maintains progression but cannot initiate retinal neurogenesis. Instead, shh in the midline appears to trigger initiation of retinogenesis and Ath5 expression. It is unknown if midline shh activates Neurog2 in the zebrafish optic cup. In the mouse retina, activation and expansion of Neurog2 and Atoh7 expression precedes the appearance of retinal derived shh at E12.5 (Jensen and Wallace, 1997). In the future, it will be important to correlate the onset of midline and retinal shh with a) the time course of Neurog2 expression, b) Neurog2 regulation of early neurogenesis, c) the period when the delay is overcome in Neurog2 mutants and d) the ability of Ascl1 to rescue the Neurog2 phenotype.

There are other signaling pathways that should be considered as well. For example, thyroid hormone signaling, which is important for photoreceptor differentiation, is deployed in multiple coordinated waves, at different phases of progenitor proliferation (Trimarchi et al., 2008a). Yet another example of extrinsic signaling is the Notch pathway, which also controls the timing of RGC differentiation and bHLH expression (Austin et al., 1995; Bao and Cepko, 1997; Nelson et al., 2006; Nelson and Reh, 2008).

Importantly in the chick eye, McCabe et al (1999) demonstrated that proximity to the wave front is not required for the progression of RGC genesis, indicating that this process depends more strongly on intrinsic components than extrinsic signals. Here, Neurog2 retinal expression was correlated with and identified as required for the spatiotemporal progression of the wave of neurogenesis in the mouse eye. Potentially, Neurog2 may act as a temporal integrator, interpreting combinations of extrinsic signals and multiple intrinsic inputs, from transcription factors such as Pax6 and Sox2 (Marquardt et al., 2001; Taranova et al., 2006), resulting in the activation and expansion of Neurog2, followed by neurogenic wave initiation. There is evidence for other intrinsic factor regulation of spatiotemporal progression of neurogenesis. In the orJ mouse, loss of Vsx2/Chx10, which is critical for maintaining retinal progenitor proliferation, results in severe microphthalmia, lack of peripheral neurogenesis, and a delay in RGC-derived shh signaling (Bone-Larson et al., 2000; Burmeister et al., 1996; Sigulinsky et al., 2008). Although Vsx2/Chx10 is ubiquitously expressed in retinal progenitors prior to the initiation of neurogenesis, it likely acts in concert with Neurog2 and other factors to control the wave of neurogenesis in the retina. Another spatiotemporal process, cell migration is tightly coordinated for normal laminar patterning in the neocortex and retina – as cells exit the cell cycle and adopt a neural fate, they must migrate out of the ventricular zone to reach the proper layer. Recently, Neurog2 and other proneural genes have been shown to regulate cortical migration, in part through regulation of Rnd2, a small GTP binding protein (Ge et al., 2006; Heng et al., 2008). Thus, coordinating spatiotemporal aspects of retinal development seems to require the tight coupling of multiple facets of neurogenesis by proneural bHLH and homeodomain transcription factors.

While Neurog2 is necessary for the propagation of neurogenesis and Atoh7 expression, it is not required for their initiation, clearly indicating that other factors are required. The initiation is neurogenesis is highly dependent on Pax6, critical for the expression of multiple bHLH factors (Brown et al., 1998; Marquardt et al., 2001; Riesenberg et al., 2009). The onset of proneural bHLH gene expression and retinal neurogenesis closely coincides with the downregulation of Pax2 in the nascent neural retina, a known regulator of Pax6 (Schwarz et al., 2000). It stands to reason, then, that the timing of bHLH initiation may be controlled indirectly by Pax2 regulation of Pax6 function or directly by Pax2 repression of bHLH gene expression.

A bHLH network controlling retinal neurogenesis

In different contexts of the developing nervous system, Neurog2 controls proliferation, cell cycle exit, cell fate identity, neurotransmitter specification, cell migration, axon guidance, and survival (Aaker et al., 2009; Britz et al., 2006; Cai et al., 2000; Fode et al., 1998; Fode et al., 2000; Seibt et al., 2003). However, previous to this study no phenotype was attributable solely to Neurog2 function during vertebrate retinal development. Here, we uncovered a key role for Neurog2 in regulating the initial progression of early retinal neurogenesis and RGC specification, which can be compensated for by substitution of Ascl1 for Neurog2. Throughout the CNS, Ascl1 and Neurog2 are intricately linked in a context-dependent manner. In the forebrain, Neurog2 represses Ascl1 to maintain dorsal projection neuron identity, while in the dorsal neural tube Neurog2 appears to function temporally downstream of Ascl1 to influence the timing of cell cycle exit (Fode et al., 2000; Helms et al., 2005). Other Neurog2Ascl1KI replacement experiments demonstrate that Ascl1 cannot rescue the Neurog2 phenotype in the dorsal forebrain or dorsal root ganglia (Fode et al., 2000; Parras et al., 2002), but can partially compensate for ventral spinal cord and midbrain dopaminergic neuron phenotypes (Kele et al., 2006; Parras et al., 2002).

In the retina, Neurog2 and Ascl1 both appear to promote cell cycle exit and neuronal determination analogously, such that Ascl1 expressed from the Neurog2 locus can rescue the temporal delay of RGC genesis. This was unexpected, since RGCs are unaffected in Ascl1 mutants, and Ascl1 is thought to function primarily in specification of later-born retinal fates, particularly rod photoreceptors and bipolar interneurons (Hatakeyama et al., 2001; Tomita et al., 1996). Here we propose that the normal onset of endogenous Ascl1 expression activates a subsequent wave of neurogenesis. In Neurog2 mutants, retinal second wave cells could either autonomously produce first wave and second wave neurons, or nonautonomously jumpstart the stalled first wave cells. The absence of increased retinal cell proliferation in Neurog2 mutants suggests the first scenario as the least likely. At present there is no hard evidence for Ascl1 regulation of a subsequent wave, although Ascl1 impressively rescues the Neurog2 phenotype. To settle this question, the Ascl1 retinal lineage and mutant phenotypes (ideally with a conditional allele) will need careful examination during prenatal retinogenesis.

Conversely, Ascl1 may compensate for the loss of Neurog2 by an unknown mechanism. Interestingly, Ascl1 performs a critical function during zebrafish retinal regeneration (Fausett et al., 2008). Both Neurog2 and Ascl1 are present in proliferating neural progenitor cells (Jasoni and Reh, 1996; Yan et al., 2001), implying that they share a common set of downstream target genes critical for controlling cell cycle progression versus exit for neural differentiation. The expression of Neurog2 and Ascl1 at different times during retinogenesis seems integral with their context-specific functions. Intriguingly, the removal of both Neurog2 and Ascl1 did not result in the total loss of neurogenesis or Atoh7 expression (Akagi et al., 2004), suggesting that further levels of compensation exist. In postnatal Ascl1 mutant retinas, horizontal interneuron and rod photoreceptor differentiation is temporarily reduced (Tomita et al., 1996), potentially restored by yet another compensatory factor. Overall, we conclude that the spatial and temporal progression of mammalian retinal neurogenesis is regulated by the bHLH factor Neurog2, and that a remarkable compensatory potential exists in the developing retina, potentially through a secondary wave of neurogenesis directed by Ascl1.

Supplementary Material


Supplemental Figure 1. Absence of bHLH factor cross-regulation at early stages of retinogenesis. (A-D) Atoh7 mRNA expression is normal in E12.5 (B) and E15.5 (D) Ascl1 mutant eyes. (E-F) Similarly, Ascl1 mRNA expression in E15.5 Neurog2GFP/GFP retinas is indistinguishable from wild type. This is in contrast to E17.5 where the loss of Neurog2 derepresses Ascl1 mRNA expression (Akagi et al., 2004). Scale bar = 50 microns and in A,E; n = 3 embryos per genotype.


The authors thank François Guillemot for Neurog2GFP and Neurog2Ascl1KI mice; Kenny Campbell for Ascl1 embryonic litters; David Anderson and Masato Nakafuku for Neurog2 antibodies; Jane Johnson for Ascl1 antibody; Lev Prasov and Tom Glaser for helpful discussions, and Kenny Campbell, Masato Nakafuku, Brian Gebelein, Steve Woods, and Noah Shroyer for critical comments. This work supported by the Edith J. Crawley Memorial Scholars Program, University of Cincinnati Department of Ophthalmology (RBH), and NIH grants EY13612 and EY18097 (NLB).


Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.


  • Aaker JD, Patineau AL, Yang HJ, Ewart DT, Gong W, Li T, Nakagawa Y, McLoon SC, Koyano-Nakagawa N. Feedback regulation of NEUROG2 activity by MTGR1 is required for progression of neurogenesis. Mol Cell Neurosci. 2009 [PMC free article] [PubMed]
  • Akagi T, Inoue T, Miyoshi G, Bessho Y, Takahashi M, Lee JE, Guillemot F, Kageyama R. Requirement of multiple basic helix-loop-helix genes for retinal neuronal subtype specification. J Biol Chem. 2004;279:28492–8. [PubMed]
  • Altshuler DM, Turner DL, Cepko CL. Specification of Cell Type in the Vertebrate Retina. In: Lam DM-K SC, editor. Cell Lineage and Cell Fate in Visual System Development. MIT Press; Cambridge, MA: 1991. pp. 37–58.
  • Austin CP, Feldman DE, Ida JA, Jr., Cepko CL. Vertebrate retinal ganglion cells are selected from competent progenitors by the action of Notch. Development. 1995;121:3637–50. [PubMed]
  • Bao ZZ, Cepko CL. The expression and function of Notch pathway genes in the developing rat eye. J Neurosci. 1997;17:1425–34. [PubMed]
  • Baumer N, Marquardt T, Stoykova A, Spieler D, Treichel D, Ashery-Padan R, Gruss P. Retinal pigmented epithelium determination requires the redundant activities of Pax2 and Pax6. Development. 2003;130:2903–15. [PubMed]
  • Bone-Larson C, Basu S, Radel JD, Liang M, Perozek T, Kapousta-Bruneau N, Green DG, Burmeister M, Hankin MH. Partial rescue of the ocular retardation phenotype by genetic modifiers. J Neurobiol. 2000;42:232–47. [PubMed]
  • Brittis PA, Meiri K, Dent E, Silver J. The earliest patterns of neuronal differentiation and migration in the mammalian central nervous system. Exp Neurol. 1995;134:1–12. [PubMed]
  • Britz O, Mattar P, Nguyen L, Langevin LM, Zimmer C, Alam S, Guillemot F, Schuurmans C. A role for proneural genes in the maturation of cortical progenitor cells. Cereb Cortex. 2006;16(Suppl 1):i138–51. [PubMed]
  • Brown NL, Kanekar S, Vetter ML, Tucker PK, Gemza DL, Glaser T. Math5 encodes a murine basic helix-loop-helix transcription factor expressed during early stages of retinal neurogenesis. Development. 1998;125:4821–33. [PubMed]
  • Brown NL, Patel S, Brzezinski J, Glaser T. Math5 is required for retinal ganglion cell and optic nerve formation. Development. 2001;128:2497–508. [PMC free article] [PubMed]
  • Brown NL, Sattler CA, Paddock SW, Carroll SB. Hairy and emc negatively regulate morphogenetic furrow progression in the Drosophila eye. Cell. 1995;80:879–87. [PubMed]
  • Brzezinski J. A. t., Brown NL, Tanikawa A, Bush RA, Sieving PA, Vitaterna MH, Takahashi JS, Glaser T. Loss of circadian photoentrainment and abnormal retinal electrophysiology in Math5 mutant mice. Invest Ophthalmol Vis Sci. 2005;46:2540–51. [PMC free article] [PubMed]
  • Burmeister M, Novak J, Liang MY, Basu S, Ploder L, Hawes NL, Vidgen D, Hoover F, Goldman D, Kalnins VI, Roderick TH, Taylor BA, Hankin MH, McInnes RR. Ocular retardation mouse caused by Chx10 homeobox null allele: impaired retinal progenitor proliferation and bipolar cell differentiation. Nat Genet. 1996;12:376–84. [PubMed]
  • Cai L, Morrow EM, Cepko CL. Misexpression of basic helix-loop-helix genes in the murine cerebral cortex affects cell fate choices and neuronal survival. Development. 2000;127:3021–30. [PubMed]
  • Dyer MA, Cepko CL. p27Kip1 and p57Kip2 regulate proliferation in distinct retinal progenitor cell populations. J Neurosci. 2001;21:4259–71. [PubMed]
  • Dyer MA, Livesey FJ, Cepko CL, Oliver G. Prox1 function controls progenitor cell proliferation and horizontal cell genesis in the mammalian retina. Nat Genet. 2003;34:53–8. [PubMed]
  • Easter SS., Jr. Let there be sight. Neuron. 2000;27:193–5. [PubMed]
  • Fausett BV, Gumerson JD, Goldman D. The proneural basic helix-loop-helix gene ascl1a is required for retina regeneration. J Neurosci. 2008;28:1109–17. [PMC free article] [PubMed]
  • Fode C, Gradwohl G, Morin X, Dierich A, LeMeur M, Goridis C, Guillemot F. The bHLH protein NEUROGENIN 2 is a determination factor for epibranchial placode-derived sensory neurons. Neuron. 1998;20:483–94. [PubMed]
  • Fode C, Ma Q, Casarosa S, Ang SL, Anderson DJ, Guillemot F. A role for neural determination genes in specifying the dorsoventral identity of telencephalic neurons. Genes Dev. 2000;14:67–80. [PubMed]
  • Ge W, He F, Kim KJ, Blanchi B, Coskun V, Nguyen L, Wu X, Zhao J, Heng JI, Martinowich K, Tao J, Wu H, Castro D, Sobeih MM, Corfas G, Gleeson JG, Greenberg ME, Guillemot F, Sun YE. Coupling of cell migration with neurogenesis by proneural bHLH factors. Proc Natl Acad Sci U S A. 2006;103:1319–24. [PubMed]
  • Gowan K, Helms AW, Hunsaker TL, Collisson T, Ebert PJ, Odom R, Johnson JE. Crossinhibitory activities of Ngn1 and Math1 allow specification of distinct dorsal interneurons. Neuron. 2001;31:219–32. [PubMed]
  • Hatakeyama J, Tomita K, Inoue T, Kageyama R. Roles of homeobox and bHLH genes in specification of a retinal cell type. Development. 2001;128:1313–22. [PubMed]
  • Helms AW, Abney AL, Ben-Arie N, Zoghbi HY, Johnson JE. Autoregulation and multiple enhancers control Math1 expression in the developing nervous system. Development. 2000;127:1185–96. [PubMed]
  • Helms AW, Battiste J, Henke RM, Nakada Y, Simplicio N, Guillemot F, Johnson JE. Sequential roles for Mash1 and Ngn2 in the generation of dorsal spinal cord interneurons. Development. 2005;132:2709–19. [PMC free article] [PubMed]
  • Heng JI, Nguyen L, Castro DS, Zimmer C, Wildner H, Armant O, Skowronska-Krawczyk D, Bedogni F, Matter JM, Hevner R, Guillemot F. Neurogenin 2 controls cortical neuron migration through regulation of Rnd2. Nature. 2008;455:114–8. [PubMed]
  • Holt CE, Bertsch TW, Ellis HM, Harris WA. Cellular determination in the Xenopus retina is independent of lineage and birth date. Neuron. 1988;1:15–26. [PubMed]
  • Horton S, Meredith A, Richardson JA, Johnson JE. Correct coordination of neuronal differentiation events in ventral forebrain requires the bHLH factor MASH1. Mol Cell Neurosci. 1999;14:355–69. [PubMed]
  • Hu M, Easter SS. Retinal neurogenesis: the formation of the initial central patch of postmitotic cells. Dev Biol. 1999;207:309–21. [PubMed]
  • Hufnagel RB, Riesenberg AN, Saul SM, Brown NL. Conserved regulation of Math5 and Math1 revealed by Math5-GFP transgenes. Mol Cell Neurosci. 2007;36:435–48. [PMC free article] [PubMed]
  • Hutcheson DA, Hanson MI, Moore KB, Le TT, Brown NL, Vetter ML. bHLH-dependent and -independent modes of Ath5 gene regulation during retinal development. Development. 2005;132:829–39. [PubMed]
  • Inoue T, Hojo M, Bessho Y, Tano Y, Lee JE, Kageyama R. Math3 and NeuroD regulate amacrine cell fate specification in the retina. Development. 2002;129:831–42. [PubMed]
  • Jarman AP, Grell EH, Ackerman L, Jan LY, Jan YN. Atonal is the proneural gene for Drosophila photoreceptors. Nature. 1994;369:398–400. [PubMed]
  • Jarman AP, Sun Y, Jan LY, Jan YN. Role of the proneural gene, atonal, in formation of Drosophila chordotonal organs and photoreceptors. Development. 1995;121:2019–30. [PubMed]
  • Jasoni CL, Reh TA. Temporal and spatial pattern of MASH-1 expression in the developing rat retina demonstrates progenitor cell heterogeneity. J Comp Neurol. 1996;369:319–27. [PubMed]
  • Jensen AM, Wallace VA. Expression of Sonic hedgehog and its putative role as a precursor cell mitogen in the developing mouse retina. Development. 1997;124:363–71. [PubMed]
  • Kanekar S, Perron M, Dorsky R, Harris WA, Jan LY, Jan YN, Vetter ML. Xath5 participates in a network of bHLH genes in the developing Xenopus retina. Neuron. 1997;19:981–94. [PubMed]
  • Kay JN, Finger-Baier KC, Roeser T, Staub W, Baier H. Retinal ganglion cell genesis requires lakritz, a Zebrafish atonal Homolog. Neuron. 2001;30:725–36. [PubMed]
  • Kay JN, Link BA, Baier H. Staggered cell-intrinsic timing of ath5 expression underlies the wave of ganglion cell neurogenesis in the zebrafish retina. Development. 2005;132:2573–85. [PubMed]
  • Kele J, Simplicio N, Ferri AL, Mira H, Guillemot F, Arenas E, Ang SL. Neurogenin 2 is required for the development of ventral midbrain dopaminergic neurons. Development. 2006;133:495–505. [PubMed]
  • Le TT, Wroblewski E, Patel S, Riesenberg AN, Brown NL. Math5 is required for both early retinal neuron differentiation and cell cycle progression. Dev Biol. 2006 [PubMed]
  • Lee MK, Tuttle JB, Rebhun LI, Cleveland DW, Frankfurter A. The expression and posttranslational modification of a neuron-specific beta-tubulin isotype during chick embryogenesis. Cell Motil Cytoskeleton. 1990;17:118–32. [PubMed]
  • Levine EM, Close J, Fero M, Ostrovsky A, Reh TA. p27(Kip1) regulates cell cycle withdrawal of late multipotent progenitor cells in the mammalian retina. Dev Biol. 2000;219:299–314. [PubMed]
  • Liu H, Etter P, Hayes S, Jones I, Nelson B, Hartman B, Forrest D, Reh TA. NeuroD1 regulates expression of thyroid hormone receptor 2 and cone opsins in the developing mouse retina. J Neurosci. 2008;28:749–56. [PubMed]
  • Lo L, Dormand E, Greenwood A, Anderson DJ. Comparison of the generic neuronal differentiation and neuron subtype specification functions of mammalian achaete-scute and atonal homologs in cultured neural progenitor cells. Development. 2002;129:1553–67. [PubMed]
  • Ma W, Wang SZ. The final fates of neurogenin2-expressing cells include all major neuron types in the mouse retina. Mol Cell Neurosci. 2006;31:463–9. [PMC free article] [PubMed]
  • Ma W, Yan RT, Mao W, Wang SZ. Neurogenin3 promotes early retinal neurogenesis. Mol Cell Neurosci. 2009;40:187–98. [PMC free article] [PubMed]
  • Macdonald R, Barth KA, Xu Q, Holder N, Mikkola I, Wilson SW. Midline signalling is required for Pax gene regulation and patterning of the eyes. Development. 1995;121:3267–78. [PubMed]
  • Mao CA, Wang SW, Pan P, Klein WH. Rewiring the retinal ganglion cell gene regulatory network: Neurod1 promotes retinal ganglion cell fate in the absence of Math5. Development. 2008;135:3379–88. [PubMed]
  • Marquardt T, Ashery-Padan R, Andrejewski N, Scardigli R, Guillemot F, Gruss P. Pax6 is required for the multipotent state of retinal progenitor cells. Cell. 2001;105:43–55. [PubMed]
  • Martinez-Morales JR, Del Bene F, Nica G, Hammerschmidt M, Bovolenta P, Wittbrodt J. Differentiation of the vertebrate retina is coordinated by an FGF signaling center. Dev Cell. 2005;8:565–74. [PubMed]
  • Masai I, Stemple DL, Okamoto H, Wilson SW. Midline signals regulate retinal neurogenesis in zebrafish. Neuron. 2000;27:251–63. [PubMed]
  • Masai I, Yamaguchi M, Tonou-Fujimori N, Komori A, Okamoto H. The hedgehog-PKA pathway regulates two distinct steps of the differentiation of retinal ganglion cells: the cell-cycle exit of retinoblasts and their neuronal maturation. Development. 2005;132:1539–53. [PubMed]
  • Matter-Sadzinski L, Matter JM, Ong MT, Hernandez J, Ballivet M. Specification of neurotransmitter receptor identity in developing retina: the chick ATH5 promoter integrates the positive and negative effects of several bHLH proteins. Development. 2001;128:217–31. [PubMed]
  • Matter-Sadzinski L, Puzianowska-Kuznicka M, Hernandez J, Ballivet M, Matter JM. A bHLH transcriptional network regulating the specification of retinal ganglion cells. Development. 2005;132:3907–21. [PubMed]
  • McCabe KL, Gunther EC, Reh TA. The development of the pattern of retinal ganglion cells in the chick retina: mechanisms that control differentiation. Development. 1999;126:5713–24. [PubMed]
  • Mori M, Ghyselinck NB, Chambon P, Mark M. Systematic immunolocalization of retinoid receptors in developing and adult mouse eyes. Invest Ophthalmol Vis Sci. 2001;42:1312–8. [PubMed]
  • Morrow EM, Furukawa T, Lee JE, Cepko CL. NeuroD regulates multiple functions in the developing neural retina in rodent. Development. 1999;126:23–36. [PubMed]
  • Nelson BR, Gumuscu B, Hartman BH, Reh TA. Notch activity is downregulated just prior to retinal ganglion cell differentiation. Dev Neurosci. 2006;28:128–41. [PubMed]
  • Nelson BR, Reh TA. Relationship between Delta-like and proneural bHLH genes during chick retinal development. Dev Dyn. 2008;237:1565–80. [PubMed]
  • Neumann CJ, Nuesslein-Volhard C. Patterning of the zebrafish retina by a wave of sonic hedgehog activity. Science. 2000;289:2137–9. [PubMed]
  • Nieber F, Pieler T, Henningfeld KA. Comparative expression analysis of the neurogenins in Xenopus tropicalis and Xenopus laevis. Dev Dyn. 2009;238:451–8. [PubMed]
  • Ohsawa R, Kageyama R. Regulation of retinal cell fate specification by multiple transcription factors. Brain Res. 2008;1192:90–8. [PubMed]
  • Parras CM, Schuurmans C, Scardigli R, Kim J, Anderson DJ, Guillemot F. Divergent functions of the proneural genes Mash1 and Ngn2 in the specification of neuronal subtype identity. Genes Dev. 2002;16:324–38. [PubMed]
  • Perron M, Boy S, Amato MA, Viczian A, Koebernick K, Pieler T, Harris WA. A novel function for Hedgehog signalling in retinal pigment epithelium differentiation. Development. 2003;130:1565–77. [PubMed]
  • Perron M, Opdecamp K, Butler K, Harris WA, Bellefroid EJ. X-ngnr-1 and Xath3 promote ectopic expression of sensory neuron markers in the neurula ectoderm and have distinct inducing properties in the retina. Proc Natl Acad Sci U S A. 1999;96:14996–5001. [PubMed]
  • Picker A, Brand M. Fgf signals from a novel signaling center determine axial patterning of the prospective neural retina. Development. 2005;132:4951–62. [PubMed]
  • Poggi L, Vitorino M, Masai I, Harris WA. Influences on neural lineage and mode of division in the zebrafish retina in vivo. J Cell Biol. 2005;171:991–9. [PMC free article] [PubMed]
  • Powell LM, Zur Lage PI, Prentice DR, Senthinathan B, Jarman AP. The proneural proteins Atonal and Scute regulate neural target genes through different E-box binding sites. Mol Cell Biol. 2004;24:9517–26. [PMC free article] [PubMed]
  • Prada C, Puga J, Perez-Mendez L, Lopez And R, Ramirez G. Spatial and Temporal Patterns of Neurogenesis in the Chick Retina. Eur J Neurosci. 1991;3:1187. [PubMed]
  • Ready DF, Hanson TE, Benzer S. Development of the Drosophila retina, a neurocrystalline lattice. Dev Biol. 1976;53:217–40. [PubMed]
  • Riesenberg AN, Le TT, Willardsen MI, Blackburn DC, Vetter ML, Brown NL. Pax6 regulation of Math5 during mouse retinal neurogenesis. Genesis. 2009;47:175–87. [PMC free article] [PubMed]
  • Rodieck RW. The First Steps in Seeing. Sinauer; Sunderland, MA: 1998.
  • Saul SM, Brzezinski J. A. t., Altschuler RA, Shore SE, Rudolph DD, Kabara LL, Halsey KE, Hufnagel RB, Zhou J, Dolan DF, Glaser T. Math5 expression and function in the central auditory system. Mol Cell Neurosci. 2008;37:153–69. [PMC free article] [PubMed]
  • Schwarz M, Cecconi F, Bernier G, Andrejewski N, Kammandel B, Wagner M, Gruss P. Spatial specification of mammalian eye territories by reciprocal transcriptional repression of Pax2 and Pax6. Development. 2000;127:4325–34. [PubMed]
  • Seibt J, Schuurmans C, Gradwhol G, Dehay C, Vanderhaeghen P, Guillemot F, Polleux F. Neurogenin2 specifies the connectivity of thalamic neurons by controlling axon responsiveness to intermediate target cues. Neuron. 2003;39:439–52. [PubMed]
  • Sidman R. Histogenesis of mouse retina studied with thymidine-H3. In: Smelser G, editor. Structure of the Eye. Academic Press; New York: 1961. pp. 487–506.
  • Sigulinsky CL, Green ES, Clark AM, Levine EM. Vsx2/Chx10 ensures the correct timing and magnitude of Hedgehog signaling in the mouse retina. Dev Biol. 2008;317:560–75. [PMC free article] [PubMed]
  • Skowronska-Krawczyk D, Chiodini F, Ebeling M, Alliod C, Kundzewicz A, Castro D, Ballivet M, Guillemot F, Matter-Sadzinski L, Matter JM. Conserved regulatory sequences in Atoh7 mediate non-conserved regulatory responses in retina ontogenesis. Development. 2009;136:3767–77. [PubMed]
  • Taranova OV, Magness ST, Fagan BM, Wu Y, Surzenko N, Hutton SR, Pevny LH. SOX2 is a dose-dependent regulator of retinal neural progenitor competence. Genes Dev. 2006;20:1187–202. [PubMed]
  • Tomita K, Nakanishi S, Guillemot F, Kageyama R. Mash1 promotes neuronal differentiation in the retina. Genes Cells. 1996;1:765–74. [PubMed]
  • Trimarchi JM, Harpavat S, Billings NA, Cepko CL. Thyroid hormone components are expressed in three sequential waves during development of the chick retina. BMC Dev Biol. 2008a;8:101. [PMC free article] [PubMed]
  • Trimarchi JM, Stadler MB, Cepko CL. Individual retinal progenitor cells display extensive heterogeneity of gene expression. PLoS One. 2008b;3:e1588. [PMC free article] [PubMed]
  • Wallace VA, Raff MC. A role for Sonic hedgehog in axon-to-astrocyte signalling in the rodent optic nerve. Development. 1999;126:2901–9. [PubMed]
  • Wang SW, Kim BS, Ding K, Wang H, Sun D, Johnson RL, Klein WH, Gan L. Requirement for math5 in the development of retinal ganglion cells. Genes Dev. 2001;15:24–9. [PubMed]
  • Willardsen MI, Suli A, Pan Y, Marsh-Armstrong N, Chien CB, El-Hodiri H, Brown NL, Moore KB, Vetter ML. Temporal regulation of Ath5 gene expression during eye development. Dev Biol. 2009;326:471–81. [PMC free article] [PubMed]
  • Xiang M, Zhou L, Peng YW, Eddy RL, Shows TB, Nathans J. Brn-3b: a POU domain gene expressed in a subset of retinal ganglion cells. Neuron. 1993;11:689–701. [PubMed]
  • Yan RT, Ma WX, Wang SZ. neurogenin2 elicits the genesis of retinal neurons from cultures of nonneural cells. Proc Natl Acad Sci U S A. 2001;98:15014–9. [PubMed]
  • Yan RT, Wang SZ. Requirement of neuroD for photoreceptor formation in the chick retina. Invest Ophthalmol Vis Sci. 2004;45:48–58. [PMC free article] [PubMed]
  • Young RW. Cell differentiation in the retina of the mouse. Anat Rec. 1985;212:199–205. [PubMed]