PAX6/Pax6 gene encodes a transcription factor that is crucially required for eye development. Pax6 heterozygous mutant mouse (Pax6Sey/+) shows various ocular defects, especially in the anterior segment. It has been well known that the induction of the lens and development of the cornea and retina are dependent on PAX6/Pax6 in a cell-autonomous fashion, although the influence of PAX6/Pax6 on the other tissues derived from the ocular mesenchyme is largely unknown. Using transgenic mouse lines in which neural crest cells are genetically marked by LacZ or EGFP, we revealed the extensive contribution of neural crest derived cells (NCDCs) to the ocular tissues. Furthermore, various eye defects in Pax6Sey/+ mouse were accompanied by abnormal distribution of NCDCs from early developmental stages to the adult. In Pax6Sey/+ mouse mice, neural crest cells abnormally migrated into the developing eye in a cell nonautonomous manner at early embryonic stages. These results indicate that normal distribution and integration of NCDCs in ocular tissues depend on a proper dosage of Pax6, and that Pax6Sey/+ eye anomalies are caused by cell autonomous and nonautonomous defects due to Pax6 haploinsufficiency.
In vertebrates, the eye originates from three different tissues, the head ectoderm, the neuroepithelium and the mesenchyme. The surface ectoderm later forms the lens and corneal epithelium, while the neuroepithelium gives rise to the neural retina and the retinal pigment epithelium. All other tissues in the eye are derived from the ocular mesenchyme. The ocular mesenchyme includes cells that originate from both the mesoderm and the neural crest, as demonstrated in the quail-chick chimera system (Le Douarin & Kalcheim 1999). For example, the neural crest contributes to the formation of the corneal endothelium and stromal cells (Hay et al. 1979; Johnston et al. 1979; Meier 1982; Nakamura 1982). More recently, extensive research in avian embryos (Creuzet et al. 2005) has shown that cranial neural crest cells (NCCs) give rise not only to the cornea but also to other intra- and extra-ocular structures such as the iris, ciliary body, choroid, sclera and part of the extra-ocular muscles.
In mice, vital dye analyses have indicated that NCCs from the midbrain and forebrain normally migrate to the craniofacial region including the prospective ocular mesenchyme (Serbedzija et al. 1992; Osumi-Yamashita et al. 1994; Trainor & Tam 1995). The ocular mesenchyme is thought to play an important role in the morphogenesis of the anterior eye segment in mammals (Cvekl & Tamm 2004). However, these studies have not revealed the precise contribution of the neural crest to eye development due to the lack of suitable markers for murine NCCs and the methodological limitation of tracing labeled cells through the later stages. Recently, genetic marking using Cre-recombinase has been applied to the long-term tracing of NCCs in Wnt1-Cre, protein zero (P0)-Cre and HTPA-Cre mice (Danielian et al. 1998; Yamauchi et al. 1999; Jiang et al. 2000; Pietri et al. 2003). Specifically, Matt et al. (2005) utilized Wnt1-Cre mice, reporting involvement of NCDCs in the ocular mesenchyme. This suggests that NCCs could be controlling their migration and distribution during ocular morphogenesis by various factors.
Homozygous Pax6 mutant mice/rats are eyeless due to failure to respond to induction signals from the optic vesicle (Fujiwara et al. 1994). Interestingly, these Pax6 mutants show not only anophthalmia but also various craniofacial defects including the absence of nasal structures and duplication of the upper teeth (Hogan et al. 1988; Quinn et al. 1997). Our previous studies demonstrated that the craniofacial defects were due to impaired migration of midbrain crest cells (Matsuo et al. 1993; Osumi-Yamashita et al. 1997). However, the mechanism for ocular defects in Pax6Sey/+ has not been studied with reference to NCCs.
Here we report the results of comprehensive phenotype analyses of Pax6Sey/+ eyes with reference to human eye diseases. To analyze the precise contribution of NCCs to eye development, we used P0-Cre transgenic mouse in which Cre recombination specifically occurred in migrating NCCs when it crosses with CAG-CAT-Z or CAG-CAT-EGFP line (Yamauchi et al. 1999; Kawamoto et al. 2000). By genetically marking neural crest-derived cells (NCDCs) in transgenic mice, we were able to identify the contribution of these cells to the development of intra- and peri-ocular tissues in both wild-type and Pax6Sey/+. We also showed that impaired migration of NCDCs nonautonomously leads to eye anomalies in Pax6Sey/+ mutant mice, suggesting that defects in the anterior and posterior segments of the eye are due to abnormal distribution and integration of NCDCs.
Eye defects in the adult Pax6Sey/+ mouse
In human, haploinsufficiency in PAX6 gene causes various ocular anomalies (Prosser & van Heyningen 1998; Vincent et al. 2003). The Small eye mutant mouse with a mutation in the orthologous Pax6 gene is thus considered a suitable animal model for aniridia based on the phenotypic similarities (Hogan et al. 1988; Baulmann et al. 2002; Ramaesh et al. 2003). However, previous analyses have not shown the comparison of ocular anomalies, such as aniridia and Peters anomaly (Kenyon 1975), between human and Small eye mutant mouse. Pax6Sey/+ eyes showed various phenotypes not only in the size of the eye and structures of the lens and retina but also in other tissues of the ocular anterior segments and vitreous body (Table 1, Fig. 1A). The phenotype shows significant variation, some of it stochastic, with differences between the right and left eyes in the same mouse (see no. 2 vs. no. 10 and no. 20 vs. no. 28, Table 1). Corneal opacity was a typical phenotype observed in all Pax6Sey/+ eyes. Some of the smaller eyes showed sclerocornea, scleralization of the anterior surface of the eye globe, associated with corneal flattening and opacity (4/28 cases) (Fig. 1A; Pax6Sey/+ no. 4). Severe opacity was associated with peripheral vascularization of the cornea (Fig. 1A; Pax6Sey/+ no. 6). Complete lack of the iris was observed only in one (Table 1, no. 1) of 28 cases, but partial defects and reduction of the iris were frequently seen in the phenotype corresponding to aniridia. Keratolenticular strands, which are tissues bridging between the cornea and lens (arrow in Fig. 1A; Pax6Sey/+ no. 19), were also seen, leading to reduction of the anterior chamber. Irido-corneal synechia, i.e. adhesion of the iris to the cornea, was often detected (20/28 cases), and in severe cases, the anterior chamber and the angle were narrow or disappeared totally (Fig. 1A; Pax6Sey/+ no. 4, Pax6Sey/+ no. 6). Peters anomaly, which includes a central corneal opacity with underlying defects in the posterior stroma, Descemet membrane and corneal endothelium, and sometimes accompanied by cornea–lens adhesion (6/28 cases with asterisks in Table 1). In the posterior part of the eye, loss of a part of the eye tissues (coloboma) was noted in the chroid in association with a partial defect of the iris in 9/28 cases. Abnormal cell accumulation in the vitreous cavity was detected in 10/28 cases, and a vessel-like structure was noted in severe cases (arrowheads in Fig. 1A; Pax6Sey/+ no. 4, Pax6Sey/+ no. 6). The above-mentioned anomalies were never seen in wild-type eyes (n = 20; Fig. 1A; wild-type). Although the size of individual Pax6Sey/+ eyes varied widely from less than half of the normal eye to nearly the same as the normal, the weight and the axial length of Pax6Sey/+ eyes were significantly smaller than those of the wild-type (Fig. 1B,C). Notably, the severity of abnormalities correlated well with the eye size (Table 1, Fig. 1A). These results indicate that individual Pax6Sey/+ eyes show a wide spectrum of phenotypes reflecting not only defects directly due to loss of Pax6 function but also because of altered behavior of the ocular mesenchyme that is originally negative for Pax6 expression (see below).
Table 1. A wide spectrum of ocular abnormalities in Pax6Sey/+
Axial length (mm)
Peripheral vasculari- zation
Kerato- lenticular strands
Vitreous cell accumulation
NI: No Iris; NA: No Angle; NL: No Lens.
Peters anomaly: phenotype showing corneal endothelial defects with or without cornea-lens adhesion. Coloboma: defects of the ocular tissue; Sclerocornea: cornea presents scleral characteristics; Keratolenticular strands: tissues bridging between the cornea and lens; Irido-corneal synechia: adhesions of iris to the cornea.
Identification of NCDCs in P0-Cre transgenic mouse
Some lines of transgenics, such as Wnt1-Cre, P0-Cre, and HTPA-Cre mice, are now available for genetically marking NCCs (Danielian et al. 1998; Yamauchi et al. 1999; Jiang et al. 2000; Pietri et al. 2003). In this study, it was favorable for us to use albino mice to avoid pigmentation in the retinal pigment epithelium and iris. Therefore, we used the P0-Cre transgenic line (Yamauchi et al. 1999) because the expression patterns of reporter genes were unchanged, unlike those in the HTPA-Cre line, even though P0-Cre transgenic mice were crossed with CD-1 mice (albino color) for 5–6 generations.
Crossing P0-Cre line with CAG-CAT-Z Tg line led to lacZ expression by Cre-mediated recombination at stages after E9.0 in tissues considered as neural crest derivatives in the chick (Fig. 2A). To confirm directly the specificity of marking NCCs by this system, we crossed P0-Cre Tg line with CAG-CAT-EGFP Tg line (Kawamoto et al. 2000), and labeled migrating NCCs with DiI at E8.25. The embryos were cultured in a whole embryo culture system (Osumi-Yamashita et al. 1994) (Fig. 2B,C). DiI-labeled cells from the anterior hindbrain migrated ventrally to the first pharyngeal arch and to the trigeminal ganglion after 36 h in culture (corresponding to E9.5; Fig. 2D–F). GFP expression was clearly observed in the cytoplasm of cells labeled by DiI (Fig. 2G–J). The proportion of GFP-positive cells among DiI-labeled cells was 48.3 ± 3.9% (n = 3) in the area corresponding to the pathway of migration (arrowhead in Fig. 2F), and 95.7 ± 1.0% (n = 3) in the area corresponding to the migration terminal (arrow in Fig. 2F). These results suggest that P0-Cre activity is specific for migrating NCCs and NCDCs.
Altered distribution of NCDCs from early embryonic stages in Pax6Sey/+ mice
As described above, Pax6Sey/+ adult eyes showed variable phenotypes in tissues derived from the ocular mesenchyme. We focused on the onset of ocular defects in Pax6Sey/+ from the viewpoint of developmental association with NCCs. For this purpose, we analyzed the distribution of NCDCs by using P0-Cre; lacZ and P0-Cre; EGFP double Tg embryos in the wild-type and in Pax6Sey/+ backgrounds.
During E9.5–E10.5, lacZ expression was observed in various craniofacial regions including the frontonasal mass, maxillary and mandibular prominences in both the wild-type and Pax6Sey/+ embryos (Fig. 3A, B,D,E). A slight difference in lacZ expression patterns in the two genotypes became evident at E11.5. In Pax6Sey/+ embryos, which often showed smaller lenses, lacZ expression was more intense at the temporal part adjacent to the eye primordium (black arrow in Fig. 3F), while reduction or loss of lacZ expression was observed in the lateral nasal prominence (white arrow in Fig. 3F compared with Fig. 3C). Immunohistochemical staining for Pax6 protein showed that the expression was observed in the surface ectoderm, lens vesicle and optic cup, but not in the GFP-labeled cells (Fig. 3G–J). These results indicate that NCDCs, which are marked with lacZ or GFP, are negative for Pax6 expression, populate in the neighboring areas of the eye primordium by E11.5. In addition, more NCDCs seem to accumulate in the temporal side of the eye primordium rather than in the nasal side in Pax6Sey/+ embryos.
We investigated which population of the NCDCs contributes predominantly to the ocular structure. The periocular mesenchyme was labeled with DiI at either the temporal or nasal side of the eye primordium in wild-type embryos at E10.0 (Fig. 4A,D), and the labeled embryos were cultured for 36 h. DiI-labeled temporal-side cells were observed in the eye structure (Fig. 4B,C). Furthermore, these cells were seen beneath the surface ectoderm and the prospective vitreous in the section corresponding to the migration of midbrain NCCs (Fig. 4G). In contrast, the cells labeled at the nasal periocular mesenchyme migrated to the nasal prominence, and they were not observed in the eye (Fig. 4E,F). Therefore, it is likely that NCCs originating from the midbrain contribute to the eye tissues.
Next we examined histologically the development of the eye of wild-type and Pax6Sey/+ embryos during E9.5–11.5. At E9.5 and E10.5, morphological difference was not observed in Pax6Sey/+ eye primordium (data not shown). At E11.5, the lens vesicle was completely separated from the surface ectoderm in the wild-type, and most of the mesenchyme around the eye primordium expressed GFP (green cells in Fig. 5B). Notably, some GFP-labeled cells were identified in the space between the lens and optic cup (arrow in Fig. 5B). Compared with the wild-type, Pax6Sey/+ eyes frequently showed smaller lens vesicle and optic cup and incomplete lens separation from the surface ectoderm (arrowhead in Fig. 5C). Moreover, a large number of GFP-labeled cells accumulated within the vitreous cavity (arrow in Fig. 5C). Thus we showed that morphological abnormalities of Pax6Sey/+ eyes become apparent around E11.5 when separation of the lens is finished.
We further quantified accumulation of NCCs in the temporal and nasal sides of the eye primordium of the wild-type and Pax6Sey/+ embryos at E11.5. Sections were made parallel to the plane of the migration pathway of the midbrain neural crest cells (line in Fig. 5A) and GFP-labeled cells were counted in the nasal and temporal areas (Fig. 5B,C). In the temporal area, the number of GFP-labeled cells of Pax6Sey/+ embryos was significantly higher than that of the wild-type (Fig. 5D; P = 0.0024). On the other hand, there was no difference between the number of GFP-labeled cells in the nasal area of the wild-type and Pax6Sey/+ (P = 0.57). We next investigated whether the accumulation of NCDCs around Pax6Sey/+ eyes was due to increased cell proliferation. The percentage of BrdU-incorporated cells relative to GFP-labeled cells was similar in the two genotypes (wild-type: 28.3%; Pax6Sey/+: 25.4% in Fig. 5E), suggesting that the cell accumulation in Pax6Sey/+ does not represent increased cell proliferation. These results suggest that the altered distribution of NCDCs in the periocular region is due to the abnormal migration of midbrain crest cells in Pax6Sey/+.
Abnormal distribution of NCDCs in Pax6Sey/+ eyes at later stages
We further examined the contribution of NCDCs at later stages of eye development. At E18.5, GFP-labeled cells were distributed in the cornea, choroid and sclera of the wild-type (Fig. 6A,B). In addition, a small number of GFP-labeled cells were found in the vitreous cavity, and some of these were recognized along with PECAM-positive vascular endothelial cells (data not shown). It is thus assumed that NCDCs would contribute to the pericytes of the vitreous vessels, as shown in avian embryos (Etchevers et al. 2001; Korn et al. 2002). GFP-labeled cells accumulated in the angle between the cornea and the anterior edge of the optic cup, the latter of which subsequently forms the iris tissue (arrowheads in Fig. 6F). At this stage, Pax6 was expressed in the corneal epithelium, the lens epithelium, neural retina and retinal pigment epithelium (Fig. 6E,F). Notably, GFP-labeled NCDCs never expressed Pax6, as well as in the early stages (Fig. 3H,J).
In contrast to the wild-type, Pax6Sey/+ embryos exhibited various eye anomalies accompanied by different distribution of GFP-labeled cells both in the anterior and posterior eye segments. In a severe phenotype with a hypoplastic lens (Fig. 6C,D), abnormal accumulation of GFP-labeled cells was seen in the anterior chamber (arrowheads in Fig. 6H). Moreover, the most distal tip of the optic cup was irregularly folded, and the retinal pigment epithelium was obscure (Fig. 6D,G). Expression of Pax6 was reduced in the Pax6Sey/+ eye (Fig. 6H), especially in the distal part of the retina compared with the wild-type (Fig. 6F). In the posterior eye segment, numerous GFP-positive cells were still observed within the vitreous cavity (Fig. 6C). Compared with wild-type, GFP-positive cells in the vitreous never formed a normal vascular structure (Fig. 6A). This cell accumulation in the vitreous cavity could be traced back to E11.5, and these NCDCs never expressed Pax6 (data not shown). These results suggest that the abnormal distribution of NCDCs in the early stage may result in the subsequent abnormal eye phenotypes in Pax6Sey/+.
Identification of NCDCs in defective eye tissues of the adult Pax6Sey/+ mice
We next examined the distribution of NCDCs in adult eye tissues to determine the association of NCDCs in the ocular anomalies found in Pax6Sey/+mice. In the wild-type, GFP-labeled NCDCs were distributed in the sclera, the iris stroma and the ciliary body (Fig. 7A,B,I). These cells were also seen in the chamber angle tissue including the trabecular meshwork and Schlemm's canal, the latter of which is known to play important roles in humor outflow (Fig. 7J). Furthermore, NCDCs were also observed in the corneal stroma and endothelium but not in the corneal epithelium (Fig. 7K). These results are in agreement with the findings of a recent study in the chick (Creuzet et al. 2005).
In Pax6Sey/+ eyes, the defects shown in Table 1 correlated with abnormal distribution of NCDCs. For example, coloboma was observed as an area containing a small number of GFP-labeled cells (Fig. 7C; arrows in Fig. 7D). The defects of the anterior eye segment were frequently associated with aberrant distribution of NCDCs (Fig. 7E–H,L–N). For instance, keratolenticular strands were detected with corneal opacity in the central cornea, where abundant GFP-labeled cells were noted (arrow in Fig. 7G,H). Although the cornea in Pax6Sey/+ developed into three layers like in the wild-type eyes, the structure of each layer was abnormal; the corneal stroma was thicker in the central region with an irregular lamellar alignment, and many GFP-positive cells were present in the thickened corneal endothelium (Fig. 7H,N). Furthermore, the hypoplastic iris stroma and ciliary body in Pax6Sey/+ eyes contained fewer GFP-labeled cells (Fig. 7L,M) than in the wild-type (Fig. 7I). In contrast, the anterior iris stroma adherent to the cornea showed intense GFP expression (Fig. 7L). In severe cases of Pax6Sey/+, the chamber angle completely vanished owing to the adhesion of the iris (Fig. 7M), and the presence of NCDCs was associated with cell accumulation and/or vessel-like structures in the vitreous body (data not shown).
NCDCs contribute to the ocular and periocular tissues
One of the aims of the present study was to clarify the contribution of cranial NCCs and their derivatives in the ocular structures. It has been well known that Wnt1-Cre mouse can provide as a tool for marking neural crest-derived components (Danielian et al. 1998; Chai et al. 2000; Matt et al. 2005). It is also reported that in Wnt1-Cre; R26R embryos, lacZ expression is observed ectopically in the dorsal neural tube and cardiac tissue (Jiang et al. 2000). Meanwhile, P0-Cre can label migrating NCCs at stages later than 9.0 dpc after detaching from the neuroepithelium (Yamauchi et al. 1999). The activity of P0 promoter was previously reported in various tissues such as dorsal root ganglia, sympathetic ganglia, melanocytes, and craniofacial mesenchyme (Hasegawa et al. 2002; Tomita et al. 2005; Yamazaki et al. 2005). We were able to analyze the expression pattern of reporter genes driven by Cre in CD-1 background (albino color) by virtue of eliminated iris pigmentation. NCDCs were identified as the same pattern as in the C57/B6J background (black color) when the P0-Cre line crossed the CAG-CAT-Z or CAT-CAT-EGFP line, although the expression pattern was altered in HTPA-Cre mice, showing an unexpectedly leaky expression pattern (data not shown). We proved for the first time the specificity of P0-Cre as a genetic marker for NCC through direct DiI labeling in P0-Cre; GFP transgenic embryos in whole embryo culture. Thus, the P0-Cre reporter system seems to be adequate for genetic marking of the NCCs.
Matt et al. (2005) reported distribution of Wnt1-Cre-labeled NCDCs in the periocular mesenchyme. In the present study, we found a broad contribution of NCCs to the tissues in the anterior eye segment, such as the endothelium and stroma of the cornea and also the stroma of the iris, the wall of Schlemm's canal and ciliary body (Fig. 7O,P). In the posterior segment, NCCs also contributed to the sclera and choroidal tissues in both embryonic and adult eyes. These findings are consistent with the results shown by quail-chick chimeras (Creuzet et al. 2005). We also demonstrated existence of NCDCs in the vitreous cavity (Fig. 6A). Considering that cephalic neural crest provides pericytes to all blood vessels of the face and forebrain (Etchevers et al. 2001; Korn et al. 2002), these NCDCs in the vitreous cavity are likely to give rise to pericytes of the vitreous vessels, which has not been reported in the chick eye since it lacks vessel structures in the vitreous cavity.
Previous studies by DiI labeling have revealed the contribution of both midbrain and forebrain-derived NCCs to the periocular mesenchyme (Serbedzija et al. 1992; Osumi-Yamashita et al. 1994; Trainor & Tam 1995). In this study, we further found that the midbrain NCCs predominantly penetrated into the prospective vitreous cavity and the space between the lens and surface ectoderm by labeling periocular mesenchyme at the temporal side at E10.0 (Fig. 4C,G). Thus, it is reasonably assumed that midbrain crest cells are the major contributors in ocular morphogenesis.
Ocular anomalies in Small eye mouse in comparison to human eye diseases
In humans, heterozygous mutations in the PAX6 gene have been detected in various ocular anomalies, including aniridia, Peters anomaly, autosomal dominant keratitis, isolated foveal hypoplasia and optic nerve malformations (Hanson et al. 1994; Prosser & van Heyningen 1998; Azuma et al. 2003). The present study also demonstrates a variety of comparable eye defects in Pax6Sey/+ mice, especially in the anterior eye segment as compared to the retina (Fig. 7O,P). Individual eyes showed large variation in size with smaller ones showing more severe phenotypes, whereas corneal defects were quite common. All of Pax6Sey/+ eyes showed corneal opacity, which may correspond to the human aniridia-related keratopathy (Mirzayans et al. 1995; Ramaesh et al. 2003). Thus, the ocular tissues derived from the mesenchyme seem to be more vulnerable to the loss of PAX6/Pax6 function than the lens, neural retina and retinal pigment epithelium.
Previous studies suggested that the corneal lesion in human aniridia is initially marked by the ingrowth of vessels from the limbal region into the peripheral cornea (Collinson et al. 2004; Ramaesh et al. 2005). Our results showed that the severity of corneal opacity was proportional to the degree of peripheral vascularization (Fig. 1A), as often observed in patients (Mackman et al. 1979; Nishida et al. 1995). Some Pax6Sey/+ eyes showed corneal endothelial defect with incomplete separation of the lens, a feature that characterizes Peters anomaly in human (Kenyon 1975; Hanson et al. 1994). The defects in the chamber angle of Pax6Sey/+ eyes were associated with the hypoplastic structure of the ciliary body and the iris, the latter often broadly adhered to the cornea. These anterior anomalies could cause glaucoma due to increased intraocular pressure because of defective Schlemm's canal that forms the major drainage pathway of aqueous humor (Grant & Walton 1974). Smaller Pax6Sey/+ eyes exhibited a stalk-like cell aggregation linking the optic nerve to the posterior lens (Fig. 1A, Pax6Sey/+ no. 4). This is a novel finding in Pax6Sey/+ phenotypes and is considered equivalent to the persistent hyperplastic primary vitreous in human patients with mutations in PAX6 (Azuma et al. 2003). These phenotypic similarities between patients with mutations of PAX6 and Pax6Sey/+ mutant mice further support the importance of Pax6 in eye development.
Abnormal distribution of NCCs causes various small eye phenotypes
In Pax6Sey/+ embryos, morphological differences were observed as early as E11.5, at the time of incomplete lens separation from the surface ectoderm. At the same time, numerous NCDCs were observed between the surface ectoderm and the optic cup as well as in the vitreous cavity of the Pax6Sey/+eye. Accumulated NCDCs in the vitreous cavity transiently contribute to the vitreous vessel in normal development. However, NCDCs further remained in a third of Pax6Sey/+ adult eye, which is regarded to correspond to the hyperplastic primary vitreous of human. In Pax6Sey/+, the lens is often smaller and incompletely separated from the surface ectoderm, thereby causing a wider space between the lens and the optic cup. This may eventually cause the massive influx of the NCCs from the temporal side of the periocular region into these intraocular areas. That is, some of the ocular phenotypes in Pax6Sey/+ may be caused by the abnormal accumulation and distribution of NCDCs in early developmental stages.
A recent study by Matt et al. (2005) also reports involvement of NCDCs in eye development. They found that mice deficient for RALDH1/3 genes encoding enzymes for retinoic acid (RA) synthesis show ocular abnormalities such as accumulation of NCDCs in the vitreous. The RA-synthesizing enzymes are expressed in the retina and corneal ectoderm but not in the neural crest-derived periocular mesenchyme. RA produced in the retina and ectoderm is released into the ocular mesenchyme, thereby regulating eye morphogenesis. We have previously shown that expression of RADH1 and RADH3 are down-regulated in the Pax6 homozygous mutant rat embryo (Suzuki et al. 2000). Thus, a part of Pax6Sey/+ phenotypes can also be explained by abnormal RA-signaling in the periocular mesenchyme containing NCCs. However, the defect of RALDH1/3 null mice is milder compared with Pax6Sey/+ phenotypes, implying that more complicate mechanisms underlie hypoplasia of the Pax6Sey/+ eye.
The anterior structures of the eye is extremely complicated, and develop particularly from the late embryonic to the postnatal stages in mice (Baulmann et al. 2002; Cvekl & Tamm 2004). We have shown here that the iris and corneal tissues contain a large number of NCDCs in the wild-type eyes, while in the adult Pax6Sey/+ eyes, these tissues were undifferentiated or hypoplastic, sometimes leading to a total loss of the anterior chamber angle. Baulmann et al. (2002) reported a transient expression of lacZ in the iris stroma of the Pax6lacZ/+ eye. However, we did observe Pax6 protein expression in the iris pigment epithelium and the corneal epithelium, but not in the iris stroma nor in the corneal endothelium and stroma, those tissues containing a large amount of NCDCs (Fig. 6H, and data not shown). This is quite reasonable because NCDCs never express Pax6 throughout development (Fig. 3H,J). A recent study indicates that a proper Pax6 dosage in the distal optic cup at the mid-embryonic stage is required for subsequent iris development (Davis-Silberman et al. 2005). We also found that Pax6 expression decreased significantly in the anterior eye structures of postnatal Pax6Sey/+ mouse compared to that of the wild-type one (data not shown). Taken together, these results suggest that precise regulation of Pax6 protein amount in the iris pigment epithelium and the corneal epithelium is critical for proper development of the anterior structures of the eye containing the neural crest-derived mesenchyme.
Based on the aforementioned findings, we can classify the ocular defects of Pax6Sey/+ mutant into three categories taking into consideration the expression of Pax6 and distribution of NCDCs: Pax6-autonomous defects (the lens and retina), nonautonomous defects (vitreous body), and their combinations (the cornea, iris, angle, and ciliary body). This complex mechanism of anterior eye development would explain the wide spectrum of abnormalities seen in humans and mice with PAX6/Pax6 mutations.
Possible involvement of Pax6 downstream molecules in migration of NCCs
The Pax6 transcription factor regulates diverse developmental processes in craniofacial and ocular morphogenesis. As for the migration of NCCs, we previously reported that homozygous Pax6 mutant rats show craniofacial defects due to impaired migration of midbrain crest cells into the nasal region (Matsuo et al. 1993; Osumi-Yamashita et al. 1997). In the present study, Pax6Sey/+ embryos showed altered distribution of NCDCs in the eye and lateral nasal prominence; GFP-labeled NCDCs in Pax6Sey/+ were significantly increased in the temporal side of the eye, the site of midbrain crest cell contribution. The measurement of BrdU-incorporated NCDCs indicated that it was not due to the cell proliferation but could be caused by the different migration pattern (Fig. 5D,E). This suggests that the unbalanced distribution of NCDCs in Pax6Sey/+ mice may be led by abnormal migration as observed in Pax6 mutant rats. Since migrating NCDCs never expressed Pax6, it is regarded that the defective migration of NCCs in Pax6Sey/+ mouse is a non-cell autonomous effect of Pax6 haploinsufficiency.
In the Pax6 homozygous mutant rat, midbrain NCCs accumulate in the dorsal side of eye primordium due to ectopically localized HNK-1 epitope in the frontnasal ectoderm under which NCCs migrate (Nagase et al. 2001). The GlcAT-P gene encoding an enzyme for the synthesis of HNK-1 carbohydrate epitope is ectopically up-regulated in the frontonasal ectoderm (i.e. Pax6-expressing tissue) including the future lens and corneal epithelium. Taking account of in vitro experiments, HNK-1 carbohydrate seems to be functional in repelling NCCs to migrate into the nasal region. In the mouse, however, we could not observe altered expression of HNK-1 epitope in the head ectoderm of the wild-type and Pax6Sey/+ embryos (data not shown). Such a species-specific regulation of downstream gene expression by Pax6 was previously reported (Arai et al. 2005). The HNK-1 epitope is often found on various cell adhesion molecules (CAM) and extracellular matrix, and it has been reported that Pax6 controls expression of genes related with cell-cell and cell-substrate interactions (Chauhan et al. 2002). We checked the expression of several candidate molecules including NCAM, L1 and β-integrin both in the wild-type and Pax6Sey/+, but no difference was observed (data not shown).
With regard to other candidates that control the migration of NCCs, previous studies have demonstrated that Eph-ephrin and Semaphorin/Neuropilin signaling influences migration of craniofacial NCCs in the mouse as well as in Xenopus, chick and zebrafish (Smith et al. 1997; Osborne et al. 2005; Yu & Moens 2005). Moreover, in human, mutations of Semaphorin 3E and ephrin B1 were found in patients with congenital anomalies associated with neural crest derived tissues (Lalani et al. 2004; Twigg et al. 2004).
Thus, these genes could be involved in ocular morphogenesis, especially in the NCDC-containing anterior and posterior eye regions. Future studies focusing on molecular and cellular mechanisms of distribution and differentiation of NCCs in the ocular region would provide a clearer answer to the pathogenesis of various ocular malformations in human.
Transgenic mice expressing Cre enzyme driven by myelin protein zero (P0) promoter were crossed with another transgenic lines CAG-CAT-Z (Yamauchi et al. 1999), or CAG-CAT-EGFP (Kawamoto et al. 2000). In each double transgenic mouse, NCDCs could be identified by the expression of lacZ or GFP after the P0-Cre mediated DNA recombination. These double transgenic lines were crossed with Pax6-heterozygous mice (Pax6Sey/+) (Hill et al. 1991), kindly provided by Dr van Heyningen. To eliminate pigmentation in the iris tissue of the double transgenic lines, these mice originally of C57/B6J background (black color) were crossed with CD-1 background (albino color) (Japan Charles River, Tokyo, Japan) for 5-6 generations. The mid-day of identifying a vaginal plug was considered as embryonic day (E) 0.5. Genomic DNA was prepared from the yolk sac of embryos or tails of pups. To examine the genotypes, polymerase chain reaction (PCR) analyses for P0-Cre, CAG-CAT-Z and CAG-CAT-EGFP were performed as previously described (Yamauchi et al. 1999; Kawamoto et al. 2000), in which the enzymatic digestion of specific PCR products was carried out for Pax6Sey/+ mice (Grindley et al. 1995). All experimental procedures described in the present study were approved by the Ethics Committee for Animal Experiment of Tohoku University Graduate School of Medicine, and animals were treated according to the National Institute of Health guidance for the care and use of laboratory animals.
Detection of β-galactosidase (LacZ) activity in embryos
Embryos were fixed for 45 min at 4 °C with solution containing 1% formaldehyde, 0.2% glutaraldehyde, and 0.02% Norident P-40 in phosphate buffered saline (PBS) as previously described (Yamauchi et al. 1999). To detect LacZ activity, embryos were incubated for 30–60 min at 37 °C in a staining solution containing 3 mm potassium ferricyanide, 3 mm potassium ferrocyanide, 1 mm MgCl2, 0.1% TritonX-100 and 2% X-gal in PBS (Suemori et al. 1990). Samples were then rinsed 3 times in PBS with 2 mm ethylenediaminetetracetic acid (EDTA) to stop the reaction. Whole-mount staining samples were examined under a stereomicroscope (LEICA CLS-150XD) equipped with a digital camera system (FINGGAL LINK).
All procedures were performed as previously described (Osumi et al. 1997). Detailed information will be provided on request. The following antibodies were used: anti-GFP (rabbit polyclonal, Chemicon); anti-Pax6 (rabbit polyclonal, Inoue et al. 2000); anti-BrdU (mouse monoclonal, Becton Dickinson); anti-α smooth muscle actin (mouse monoclonal, Sigma); and anti-PECAM (mouse monoclonal, Becton Dickinson). Then the sections were incubated with Cy3-conjugated goat-anti-mouse IgG (Jackson) or FITC-conjugated goat anti-rabbit IgG (Jackson), and examined with fluorescent microscope (AXIO PLAN-2, ZEISS) equipped with a cooled CCD camera (Roper). To perform enzymatic detection of antibodies, the sections were incubated with Streptavidin-horseradish peroxidase (HRP) complex (VECTASTAIN ABC kit, Vector Laboratories) after secondary antibody incubation. Diamonobenzidine enhance kit (PIERCE) was used to detect HRP activity.
Whole embryo culture and DiI labeling of cranial neural crest cells
Embryos at E 8.25 (1–4 somite stage) were obtained from CD-1 background female mice that were crossed with P0-Cre; CAG-CAT-GFP double transgenic male mice. The procedures of whole embryo culture and DiI labeling of NCC were based on the methods previously described (Osumi-Yamashita et al. 1994). Similarly, the embryos at E10.0 were labeled with DiI in the temporal or nasal side of the periocular mesenchyme. After 36 h culture, the sections were made parallel to the plane along the migration pathway of the midbrain crest cells, and directly observed in the dried condition to see the DiI.
5-bromodeoxyuridine (BrdU) labeling
To analyze cell proliferation, 10 mg/kg BrdU (Boehringer Mannheim, Germany) was injected intraperitoneally into pregnant mice 45 min prior to extraction of embryos. For detection of BrdU-incorporated cells, sections were treated with 2 m HCl solution at 37 °C for 10 min before incubation with anti-BrdU antibody.
Data are expressed as mean ± SD. The two-sample t-test was used to compare the data in different groups. A P < 0.05 was considered to denote statistical significance.
We thank Drs Veronica van Heyningen and Noriyuki Azuma for many valuable comments on the manuscript. We are also grateful to Mss. Michi Otonari, Hisako Yusa, and Ayumi Ogasawara for maintenance of transgenic mouse lines and technical assistance, and all other members of our laboratories for helpful advises. This work was supported by Grants-in Aid for Scientific Research from the Japanese Ministry of Education, Culture, Sports, Science and Technology (No. 10220209, 16015205) (to N.O.). S.K. was supported by the 21st Century COE Program “Future Medical Engineering Based on Bio-nanotechnology.”