Vitamin A (retinol) is required for vertebrate eye development as demonstrated by the observation of severe ocular defects during gestational vitamin A deficiency, including microphthalmia and coloboma of the retina (Wilson et al., 1953; Dickman et al., 1997). Retinoic acid (RA), an active metabolite of retinol, can rescue eye defects observed during gestational vitamin A deficiency (Dickman et al., 1997), indicating that RA mediates the developmental actions of vitamin A. The developing eye expresses three nuclear retinoic acid receptors (RARα, RARβ, and RARγ) whose transcriptional activities are altered when bound to RA (Mori et al., 2001). Gene knockout studies in mice have demonstrated that these three RARs play important, overlapping roles in eye development. Embryos carrying a null mutation of only one RAR isoform display relatively minor eye defects, but knockout of two RARs results in several embryonic eye defects, including microphthalmia and coloboma (Lohnes et al., 1994). Thus, it is clear that RARs are very important for eye development, but it is less clear how generation of the ligand RA occurs spatially and temporally to regulate RAR activity. Recent studies, however, suggest that the RA isomers all-trans-RA and 9-cis-RA are not both required, as all-trans-RA was found to be sufficient for early mouse organogenesis including eye development (Mic et al., 2003).
Several enzymes have been implicated in RA synthesis for the developing mouse eye. The metabolic pathway for conversion of retinol to RA involves two sequential oxidation steps (Duester, 2000). Retinol is first oxidized to retinaldehyde in a reaction that is likely to be a housekeeping function in all cells at all stages of development due to the involvement of ubiquitously expressed class III alcohol dehydrogenase (ADH3) as well as several other tissue-specific enzymes with overlapping tissue specificities (Molotkov et al., 2002). In the next step, retinaldehyde is tissue-specifically oxidized to RA by retinaldehyde dehydrogenase (RALDH) enzymes, which are members of the aldehyde dehydrogenase (ALDH) family, i.e., RALDH1 (ALDH1A1), RALDH2 (ALDH1A2), and RALDH3 (ALDH1A3; Duester, 2000; Sophos and Vasiliou, 2003). The genes for these three enzymes have distinct tissue-specific expression patterns in the eye as development proceeds from the optic vesicle to the optic cup. In mouse, Raldh1 is expressed in the dorsal neural retina at the optic cup stage (McCaffery et al., 1991, 1992), Raldh2 is expressed transiently in the optic vesicle but not the optic cup (Niederreither et al., 1997; Haselbeck et al., 1999), and Raldh3 is expressed in the surface ectoderm (prospective lens) at the optic vesicle stage then in the lens placode and retinal pigment epithelium (RPE) at the early optic cup stage and finally in the ventral neural retina at the late optic cup stage (Grün et al., 2000; Li et al., 2000; Mic et al., 2000; Suzuki et al., 2000).
RA signaling activity can be followed spatiotemporally in mouse embryos carrying a RARE-lacZ RA-reporter transgene (Rossant et al., 1991). RARE-lacZ activity is first observed in the optic vesicle neuroepithelium at embryonic day (E) 8.5 in response to Raldh2 expression, and although Raldh2 expression is short-lived, RARE-lacZ activity persists in the developing retina throughout its development and is associated with expression of Raldh1 in dorsal neural retina and Raldh3 in RPE and ventral neural retina (Wagner et al., 2000; Mic et al., 2002). By E13.5, Raldh1 and Raldh3 are expressed in distinct dorsal and ventral neural retinal domains, with RARE-lacZ expressed in each of these domains separated by a central zone lacking RA activity (Wagner et al., 2000). This dorsoventral pattern of RA activity was originally thought to be important for neural retina development, but Raldh1-/- embryos display normal dorsoventral polarity in the eye and normal neural retina lamination and retinal ganglion cell axonal outgrowth, despite a loss of dorsal RA signaling (Fan et al., 2003). Also, studies on Raldh3-/- embryos have demonstrated an apparently normal retina phenotype, despite loss of RA activity in the ventral retina, but the eye exhibits persistence of the primary vitreous body (retrolenticular membrane) and loss of the Harderian glands (Dupé et al., 2003). These findings suggest that the requirement of RA for retina development may be limited to early stages before axonal outgrowth and lamination, perhaps at the optic vesicle stage when Raldh2 is expressed. Molecular aspects of optic vesicle and optic cup formation have been previously reviewed (Zhang et al., 2002).
Raldh2-/- mouse embryos display a growth defect from E8.75 to E10.5 followed by lethality at E11.5 (Niederreither et al., 1999; Mic et al., 2002). We have now examined retina development in Raldh2-/- embryos as well as Raldh1-/-:Raldh2-/- double-mutant embryos and find that RA is required on E8 for the transition from optic vesicle to optic cup. We also demonstrate that loss of both Raldh1 and Raldh2 function does not create a more severe retinal defect than that observed by loss of Raldh2 alone and that RA synthesized in the lens placode (presumably by RALDH3) cannot support retina invagination needed for optic cup formation.
Generation of Raldh1-/-:Raldh2-/- Double-Knockout Embryos
Matings of Raldh1-/- and Raldh2-/+ mice resulted in the generation of Raldh1-/+:Raldh2-/+ double-heterozygous adult mice at close to the normal Mendelian ratio (∼55%). Back-cross matings of Raldh1-/+:Raldh2-/+ and Raldh1-/- mice resulted in the production of Raldh1-/-:Raldh2-/+ mice at close to the normal Mendelian ratio (∼22%). Matings of Raldh1-/-:Raldh2-/+ mice resulted in no detection of Raldh1-/-:Raldh2-/- double-homozygous mice at weaning (0 of 64). Further analysis determined that Raldh1-/-:Raldh2-/- embryos were detected from E8.75 to E10.25 (Fig. 1E–G) but not at E11.5 similar to what we have described previously for Raldh2-/- embryos, which undergo resorption by E11.5 (Mic et al., 2002). The external morphological defects observed in Raldh1-/-:Raldh2-/- embryos (Fig. 1E–G) were essentially the same as those found in Raldh2-/- embryos (Mic et al., 2002), i.e., lack of axial turning, large distended heart, and reduction in overall size. Thus, at the gross morphological level, the loss of both RALDH1 and RALDH2 does not appear to create a more severe embryonic defect than that observed by loss of RALDH2 alone.
Eye Field RA Signaling Activity in Raldh1-/-:Raldh2-/- Embryos
We have reported previously that RA signaling activity in Raldh2-/- embryos is eliminated in the optic vesicle but that some RA activity still remains in the eye field surface ectoderm (Mic et al., 2002). We also previously have found that Raldh1-/- embryos do not exhibit a significant reduction in RA signaling activity in the eye field at the optic cup stage (E10.5), but that at later stages (E16.5) the dorsal retina lacks RA activity (Fan et al., 2003). This finding was examined previously by monitoring expression of the RARE-lacZ RA-reporter transgene (Rossant et al., 1991), which was introduced into the Raldh2-/- and Raldh1-/- backgrounds. Here, RARE-lacZ expression was compared in wild-type (Fig. 1A–C) and Raldh1-/-:Raldh2-/- double-mutant (Fig. 1E–G) embryos carrying RARE-lacZ. Raldh1-/-:Raldh2-/- embryos from E8.75 to E10.25 displayed a pattern of RARE-lacZ expression essentially identical to that described for Raldh2-/- embryos (Mic et al., 2002). This indicates that Raldh1 is not responsible for the RA synthesis remaining in the eye field of Raldh2-/- embryos.
At E10.25, Raldh3 mRNA was detected in the eye field of wild-type embryos (Fig. 1D), and expression in the eye field was maintained in Raldh1-/-:Raldh2-/- embryos (Fig. 1H). This finding provides strong evidence that RALDH3 is responsible for RA synthesis in the eye field of Raldh2-/- and Raldh1-/-:Raldh2-/- embryos. However, we cannot yet rule out whether another enzyme may also contribute to this RA activity.
Raldh1-/-:Raldh2-/- embryos were viable between E8.75 and E10.25, as evidenced by an increase in heart size and up-regulation of RARE-lacZ expression in the eye field, neural tube, and heart by enzymes other than RALDH1 or RALDH2 (Fig. 1E–G). The obvious increase in RARE-lacZ expression in the eye field of Raldh1-/-:Raldh2-/- embryos between E8.75 and E9.25 (Fig. 1E,F) suggests that some aspect of eye development is continuing without RALDH1 and RALDH2.
Loss of Optic Vesicle RA Signaling Prevents Formation of Optic Cup
At E8.5, Raldh2 mRNA is normally found at high levels in the optic vesicle neuroepithelium but not the adjacent surface ectoderm where the lens placode develops (Fig. 2A,B). By E9.0, Raldh2 mRNA is significantly reduced (Fig. 2C). By E9.5, it has been reported that Raldh2 mRNA is no longer detected in the eye field but that Raldh1 and Raldh3 are now expressed in the optic cup (Mic et al., 2002). Thus, RALDH2 is responsible for much of the RA synthesis in the wild-type eye at the optic vesicle stage (E8.5–E9.0), but other enzymes are responsible for RA synthesis at the optic cup stage (E10.25).
To investigate the contributions of RALDH1 and RALDH2 for eye field RA synthesis, RARE-lacZ expression was compared in horizontal sections across the eye field of wild-type, Raldh2-/-, Raldh1-/-, and Raldh1-/-:Raldh2-/- embryos. At E8.75, RARE-lacZ expression was observed throughout the optic vesicle and overlying surface ectoderm in both wild-type (Fig. 2D) and Raldh1-/- embryos (Fig. 2F). Optic vesicles were still observed in Raldh2-/- (Fig. 2E) and Raldh1-/-:Raldh2-/- embryos (Fig. 2G) at E8.75, but in each case, RARE-lacZ expression was absent from the optic vesicle while still being apparent in the surface ectoderm (including prospective lens placode). This finding suggests that RA synthesis independent of RALDH1 and RALDH2 is occurring in the surface ectoderm.
At E10.25, an optic cup structure surrounding a lens vesicle was now apparent in wild-type (Fig. 2H) and Raldh1-/- embryos (Fig. 2J), which exhibited RARE-lacZ expression throughout the eye field. However, there was a failure of optic cup formation in Raldh2-/- (Fig. 2I) and Raldh1-/-:Raldh2-/- (Fig. 2K) embryos, which retained an optic vesicle-like structure that failed to invaginate, and a surface ectoderm that failed to invaginate to form a lens vesicle. The persistent optic vesicle-like structure observed at E10.25 in Raldh2-/- and Raldh1-/-:Raldh2-/- embryos still lacked RARE-lacZ expression, but the surface ectoderm now had increased RARE-lacZ expression compared with that observed at E8.75, and some underlying mesoderm now expressed RARE-lacZ (Fig. 2I,K). Analysis of retinal cell proliferation using anti-phosphohistone 3 antibodies indicated that E9.5 Raldh2-/- optic vesicles had approximately the same number of cells undergoing mitosis as wild-type (data not shown). Thus, the lack of retinal invagination in Raldh2-/- embryos does not appear to be the result of reduced cell proliferation. These findings suggest that optic vesicle RA signaling controlled by Raldh2 is necessary for cellular differentiation within the retinal neuroepithelium leading to invagination.
Lens RA Signaling Is Insufficient to Induce Optic Cup Formation
As RARE-lacZ expression was still observed in the prospective lens ectoderm of E10.25 Raldh2-/- and Raldh1-/-:Raldh2-/- embryos, we examined horizontal sections at this stage for expression of Raldh3. At E10.25, Raldh3 expression was observed in the RPE and lens of both wild-type (Fig. 2L) and Raldh1-/- embryos (Fig. 2N). In contrast, Raldh3 expression was limited to the prospective lens surface ectoderm in E10.25 Raldh2-/- (Fig. 2M) and Raldh1-/-:Raldh2-/- embryos (Fig. 2O), which had failed to undergo optic cup formation. These findings indicate that, in the absence of RALDH2 activity in the optic vesicle, Raldh3 is not expressed in the retinal neuroepithelium but is still expressed in the prospective lens surface ectoderm, which displays an increase in RA signaling activity between E8.75 and E10.25. However, RA signaling activity in the prospective lens is insufficient to induce invagination of the optic vesicle to form an optic cup.
Rescue of Optic Cup Formation by Maternal RA Administration
The above findings suggest that RA synthesized by RALDH2 in the optic vesicle during E8 may be necessary to induce optic cup formation. To test this, we examined E10.25 Raldh1-/-:Raldh2-/- embryos subjected to maternal dietary RA administration from E7.25 to E8.75, which has been demonstrated previously to significantly rescue Raldh2-/- embryonic development (Niederreither et al., 2001; Mic et al., 2003). This limited RA treatment resulted in nearly normal external morphology (including eye development) of Raldh1-/-:Raldh2-/- embryos at E10.25, with the exception that the forelimb buds displayed a significant growth retardation (Fig. 3A), similar to that previously reported for RA-rescued Raldh2-/- embryos. Raldh1-/-:Raldh2-/- embryos lacked much of the RARE-lacZ expression normally observed in the trunk (due to loss of RALDH2 activity), but had a normal pattern of RARE-lacZ expression in the eye region (Fig. 3A; compare with wild-type in Fig. 1C). Because Raldh1-/-:Raldh2-/- embryos behaved similarly to Raldh2-/- embryos during the rescue, this finding provides evidence that RALDH1 activity is not required for the rescue to be successful. RA treatment also restored a normal pattern of Raldh3 expression in the eye (RPE and ventral retina) of E10.25 Raldh1-/-:Raldh2-/- embryos (Fig. 3C; compare with wild-type in Fig. 1D).
Horizontal sections through the eye of E10.25 RA-treated Raldh1-/-:Raldh2-/- embryos revealed that optic cup formation was rescued as we could now detect a neural retina, RPE, optic stalk, and lens (Fig. 3B). A normal pattern of RARE-lacZ expression was restored to the eye of RA-rescued Raldh1-/-:Raldh2-/- embryos (Fig. 3B), similar to untreated wild-type embryos (Fig. 2H). In such embryos, Raldh3 expression was observed in the RPE, and a lens vesicle expressing Raldh3 was now observed (Fig. 3D) similar to that observed in an untreated wild-type embryo (Fig. 2L). This was in stark contrast to untreated Raldh1-/-:Raldh2-/- embryos in which the optic cup was absent and Raldh3 expression was observed only in the prospective lens surface ectoderm (Fig. 2O). These findings provide evidence that RA synthesized by RALDH2 in the retinal neuroepithelium at the optic vesicle stage (∼E8.5) is required for optic cup formation and is needed to activate expression of Raldh3 in the retina at the early optic cup stage, which continues RA synthesis in the retinal neuroepithelium after Raldh2 expression has ended.
Expression of Early Retinal Determinants Occurs Without RA Signaling
We examined the effect of a loss of RA signaling on expression of several important transcription factors required for early eye development. These factors included Pax6 and Six3, both expressed throughout the optic vesicle (Oliver et al., 1995; Stoykova et al., 2000), Rx expressed in early optic vesicle tissue fated to become neural retina (Mathers et al., 1997), and Mitf expressed in optic vesicle tissue fated to become RPE (Bumsted and Barnstable, 2000). Examination of E8.75–E9.25 wild-type and Raldh2-/- embryos revealed that mRNAs for Pax6, Six3, Rx, and Mitf were still detected in the optic vesicle in the absence of Raldh2 function (Fig. 4A–H). These findings suggest that RA signaling functions downstream of these early determinants of retina development during progression of a predetermined optic vesicle to an optic cup.
Previous studies using either vitamin A deficiency or RA receptor mutations have demonstrated marked defects in eye development resulting in microphthalmia (Wilson et al., 1953; Lohnes et al., 1994; Dickman et al., 1997). RALDH1, RALDH2, and RALDH3 have been identified as RA-generating enzymes expressed in unique spatiotemporal patterns during eye development, and genetic inactivation of these enzymes has been performed to identify the critical stage(s) when RA is needed for eye development. Previously, we have shown that Raldh1-/- mice do not display eye defects, but this finding may be due to redundancy with either Raldh2 or Raldh3 (Fan et al., 2003). However, the Raldh1-/- studies revealed that RA is unlikely to be required for the later stages of retina lamination and retinal ganglion cell axonal outgrowth, as the dorsal retina lacked RA signaling activity from E13 onward when these events occur, but still underwent normal lamination and axonal outgrowth (Fan et al., 2003). Of interest, studies on Raldh3-/- embryos have shown that a lack of RA signaling in the ventral retina at the optic cup stage also does not result in defective retina development (Dupé et al., 2003). Here, we have shown that inactivation of Raldh2 prevents retina development beyond the optic vesicle stage, and we have further demonstrated that this effect can be rescued by maternal RA treatment during that stage.
RA Synthesis by RALDH2 in the Optic Vesicle Is Required for Optic Cup Formation
Our findings have revealed that a loss of Raldh2 function eliminates RA signaling activity in the optic vesicle. As Raldh2 is expressed only up to E9.0 in the optic vesicle, this is consistent with an early function in retina development. An optic vesicle still forms in the absence of RA signaling, and important early determinants of eye development such as the transcription factors encoded by Pax6, Six3, Rx, and Mitf are also normally expressed in the optic vesicle in the absence of RA signaling. However, we find that Raldh2-/- and Raldh1-/-:Raldh2-/- embryos do not develop an optic cup by E10, indicating that RA signaling is needed in order for the predetermined retinal neuroepithelium of the optic vesicle to invaginate into an optic cup. This defect in retina development is consistent with a previous report on vitamin A–deficient rat embryos describing a failure in retinal invagination at the optic cup stage resulting in microphthalmia (Dickman et al., 1997). In that study, it was shown that vitamin A–deficient rat embryos supplemented with maternal RA had relatively normal eye development. Therefore, our studies have identified an important role for RALDH2 in providing RA at the optic vesicle stage to stimulate optic cup formation.
RA Synthesis in the Lens Ectoderm Is Insufficient for Optic Cup Formation
Previous studies have demonstrated that optic cup formation requires a signal from pre-lens surface ectoderm that acts on the optic vesicle (Hyer et al., 2003). We found that Raldh2-/- and Raldh1-/-:Raldh2-/- embryos maintain RA signaling activity in the prospective lens ectoderm and adjacent mesenchyme, but the RA signal does not reach the optic vesicle. This is most likely due to expression of Raldh3 in the lens ectoderm, which supplies RA locally for the lens and possibly through diffusion to the mesenchyme. Our studies demonstrate that RA signaling in the lens ectoderm is insufficient to induce optic cup formation, suggesting that another signal is provided to the optic vesicle by the pre-lens ectoderm. The RA requirement during optic cup formation evidently requires the presence of RA within the retina, which is normally supplied by local expression of Raldh2 in the optic vesicle retinal neuroepithelium. This function was able to be restored by maternal administration of RA to Raldh2-/- embryos. However, as the amount of RA normally present in the maternal circulation is insufficient to provide this function, local RA synthesis by RALDH2 is the normal mechanism used to generate the required RA.
Raldh1 Is Not Required for Optic Cup Formation in Rescued Raldh2-/- Embryos
In mouse embryos, from the optic vesicle stage to the optic cup stage, the entire retina is exposed to RA starting with RALDH2 expressed throughout the optic vesicle from E8.5 to E9.0, then continuing with RALDH1 expressed in the developing dorsal retina from E9 onward and RALDH3 expressed in the developing RPE and ventral retina from E9 onward (Grün et al., 2000; Li et al., 2000; Mic et al., 2000; Suzuki et al., 2000). Results obtained from Raldh1-/- embryos demonstrate that optic cup formation proceeds normally in the absence of RALDH1 and that there is no significant reduction in RA signaling activity in the optic cup of E10.25 Raldh1-/- embryos compared with wild-type (Fan et al., 2003). However, we also examined here whether RALDH1 might still be needed for optic cup formation during maternal RA rescue of Raldh2-/- embryos. The timing of the administered RA (E7.25–E8.75) mimics the period when RALDH2 would normally synthesize RA in the optic vesicle (before the stage when Raldh1 expression initiates), but we tested whether induction of Raldh1 during optic cup formation was required to provide endogenous RA synthesis needed to complete optic cup rescue. As we obtained essentially identical optic cup rescue using either Raldh2-/- or Raldh1-/-:Raldh2-/- embryos subjected to maternal RA, we conclude that Raldh1 is not essential for optic cup formation during the RA rescue procedure.
We previously reported that Raldh1 is not expressed in unrescued Raldh2-/- embryos at E10.25 but that RA-rescued Raldh2-/- embryos do express Raldh1 in the dorsal neural retina (Mic et al., 2002). However, because Raldh1-/-:Raldh2-/- embryos lack RALDH1 activity and are still rescued efficiently, this indicates that any RA that may have been synthesized by RALDH1 in RA-rescued Raldh2-/- embryos was unnecessary for optic cup formation. As RA-rescued Raldh1-/-:Raldh2-/- embryos still have RA signaling activity in the dorsal neural retina, this may be due to the existence of RALDH3 activity in the dorsal RPE, which could supply RA by diffusion to the adjacent dorsal neural retina in the absence of RALDH1. Our findings suggest that RA generated by RALDH3 in the mouse may sufficiently overlap that of RALDH1 to make the latter enzyme unnecessary for eye development.
RA Synthesis by Raldh2 May Be Sufficient for Retina Formation
The function of Raldh3 in eye development is still unclear. We have shown that Raldh2 function is needed in order for Raldh3 to be expressed in the retina, whereas Raldh3 expression in the lens placode is independent of Raldh2 function. Thus, Raldh3 could play roles in both retina and lens development, but this is not supported by recent gene knockout studies. Optic cup and lens vesicle formation as well as retina growth are normal in Raldh3-/- embryos, suggesting that RA synthesized by RALDH3 is not needed for retina and lens development (Dupé et al., 2003). However, later aspects of eye development such as vitreous body formation and Harderian gland development were affected by loss of Raldh3 function indicating a later role for RALDH3 in RA synthesis for the eye. (Dupé et al., 2003). Thus, RA synthesized by RALDH2 may be sufficient for retina invagination during the transition from optic vesicle to optic cup, and further genetic studies can be performed to test this hypothesis.
Generation of Raldh1:Raldh2 Double-Null Mutant Mice
We previously have described production and genotyping of Raldh1-/- mice, which are viable and fertile (Fan et al., 2003), and Raldh2-/- mice, which survive only until E10.5 (Mic et al., 2002). Raldh1-/- and Raldh2-/+ mice were mated to generate Raldh1-/+:Raldh2-/+ double-heterozygous adult mice, and then further matings generated Raldh1-/-:Raldh2-/+ adult mice identified by Southern blot and polymerase chain reaction (PCR) analysis of tail DNA. Raldh1-/-:Raldh2-/+ adult mice were mated to produce Raldh1-/-:Raldh2-/- double-homozygous embryos.
Detection of RA Activity in Mouse Embryos
RA activity was detected in mouse embryos by using the RAREhsplacZ (RARE-lacZ) RA-reporter transgene in which lacZ (encoding β-galactosidase) is controlled by a retinoic acid response element (RARE; Rossant et al., 1991). Raldh1-/- mice, Raldh2-/+ mice, and Raldh1-/-:Raldh2-/+ mice were mated to RARE-lacZ mice to generate strains carrying one copy of RARE-lacZ identified by PCR analysis of tail DNA. Embryos used for analysis of RA activity were derived from matings of males carrying RARE-lacZ and females lacking RARE-lacZ, to ensure that all embryos with RARE-lacZ expression contained no more than one copy of RARE-lacZ. Embryos were stained for 18 hr with X-gal (5-bromo-4-chloro-3-indolyl-β-D-galactopyranoside) as the substrate for β-galactosidase activity (Hogan et al., 1994). After staining, embryos were embedded in 3% agarose and sectioned at 50 μm with a Vibratome.
Whole-Mount In Situ Hybridization
Detection of mRNA was performed by whole-mount in situ hybridization of mouse embryos using digoxigenin-labeled antisense RNA probes as described previously (Wilkinson, 1992). Antisense RNA probes were prepared from cDNAs encoding mouse Raldh1, Raldh2, and Raldh3 (Mic et al., 2002), Pax6 (Stoykova et al., 2000), Six3 (Oliver et al., 1995), Rx (Mathers et al., 1997), and Mitf (Bumsted and Barnstable, 2000). Stained embryos were embedded in 3% agarose and sectioned at 50 μm with a Vibratome.
Rescue of Mutants With Maternal Dietary RA Administration
Dietary RA rescue of Raldh2-/- embryos was performed similar to previous descriptions (Niederreither et al., 2001; Mic et al., 2003). Briefly, all-trans-RA (Sigma Chemical Co.) was dissolved in corn oil and mixed with powdered mouse chow to achieve an RA concentration in the food of 0.1 mg/g for treatment from E7.25 to E8.25 and 0.2 mg/g for treatment from E8.25 to E8.75. Such food was prepared fresh at E7.25, E7.75, and E8.25. At E8.75, mice were returned to normal mouse chow until the point of analysis.
Cell Proliferation Assay
Embryos (E9.5) were treated with Dent's fixative (methanol:dimethylsulfoxide 4:1), endogenous peroxidase activity was quenched by incubation with 5% hydrogen peroxide, and paraffin embedding was performed. Tissue sections (7 μm) were incubated with anti-phosphohistone 3 antibodies (1:200; Upstate, Lake Placid, NY), then with peroxidase-linked anti-rabbit secondary antibodies, and finally with diaminobenzidine substrate.
We thank J. Rossant for RARE-lacZ mice and the following for mouse cDNAs: P. Gruss (Pax6, Six3), P. Mathers (Rx), and C. Barnstable (Mitf). G.D. was funded by the National Institutes of Health.