Effects of Elevated Pax6 Expression and Genetic Background on Mouse Eye Development

Abstract

purpose. To analyze the effects of Pax6 overexpression and its interaction with genetic background on eye development.

methods. Histologic features of eyes from hemizygous PAX77 ^+/− transgenic (high Pax6 gene dose) and wild-type mice were compared on different genetic backgrounds. Experimental PAX77 ^+/−↔wild-type and control wild-type↔wild-type chimeras were analyzed to investigate the causes of abnormal eye development in PAX77 ^+/− mice.

results. PAX77 ^+/− mice showed an overlapping but distinct spectrum of eye abnormalities to Pax6 ^+/− heterozygotes (low Pax6 dose). Some previously reported PAX77 ^+/− eye abnormalities did not occur on all three genetic backgrounds examined. Several types of eye abnormalities occurred in the experimental PAX77 ^+/−↔wild-type chimeras, and they occurred more frequently in chimeras with higher contributions of PAX77 ^+/− cells. Groups of RPE cells intruded into the optic nerve sheath, indicating that the boundary between the retina and optic nerve may be displaced. Both PAX77 ^+/− and wild-type cells were involved in this ingression and in retinal folds, suggesting that neither effect was cell-autonomous. Cell-autonomous effects included failure of PAX77 ^+/− and wild-type cells to mix normally and overrepresentation of PAX77 ^+/− in the lens epithelium and RPE.

conclusions. The extent of PAX77 ^+/− eye abnormalities depended on PAX77 ^+/− genotype, genetic background, and stochastic variation. Chimera analysis identified two types of cell-autonomous effects of the PAX77 ^+/− genotype. Abnormal cell mixing between PAX77 ^+/− and wild-type cells suggests altered expression of cell surface adhesion molecules. Some phenotypic differences between PAX77 ^+/−↔wild-type and Pax6 ^+/−↔wild-type chimeras may reflect differences in the levels of PAX77 ^+/− and Pax6 ^+/− contributions to chimeric lenses.

The eye develops from three major sources of cells. The surface ectoderm of the head produces the lens and the corneal, limbal, and conjunctival epithelia. The neuroectoderm produces the neural retina and retinal pigment epithelium (RPE), and extensions of both these tissues cover the ciliary body, ciliary process, and iris. The periocular mesenchyme, comprising mesoderm and neural crest–derived mesectoderm cells, produces the choroid, sclera, corneal stroma, corneal endoderm, and stroma of the ciliary body, ciliary process, and iris. ¹  

The transcription factor Pax6 is essential for normal eye development in vertebrates. ^{2 3 4 5 6} During development Pax6 is expressed in the developing lens; conjunctival, limbal and corneal epithelia; neural retina; RPE; ciliary body; and iris. ^{4 5} After birth, Pax6 is downregulated in many eye tissues, but expression continues in the amacrine cells of the retina and the lens and in the conjunctival, limbal, and corneal epithelia. ^{5 7 8}  

Homozygous Pax6 ^−/− mice, with two nonfunctional alleles, die at birth with no eyes or nose and with brain abnormalities. ^{5 9 10 11} In some cases they have abnormal dentition but the penetrance of this effect is highly dependent on the genetic background. ¹² Human PAX6 ^−/− homozygotes or compound heterozygotes are rare and the condition is lethal, causing anophthalmia with severe craniofacial and central nervous system defects. ¹³  

Heterozygous Pax6 ^+/− mice and PAX6 ^+/− humans produce low levels of Pax6 and are viable and fertile but have a range of eye abnormalities. ² In Pax6 ^+/− mice, abnormal eye development commonly results in small eyes, iris hypoplasia, cataracts, a thin corneal epithelium with fewer cell layers, corneal opacity, failure of the lens to separate completely from the corneal epithelium, and glaucoma. Other developmental abnormalities, including retinal dysplasia, coloboma, abnormal cell accumulation in the vitreous, and adhesions between the lens and cornea (keratolenticular strands) or between the iris and cornea (iridocorneal synechia), may also occur. ^{10 14 15 16 17 18} Adult Pax6 ^+/− mice also show progressive corneal deterioration. The corneal epithelium is thin and fragile and goblet cells accumulate; the stroma becomes vascularized from the periphery and infiltrated with inflammatory cells. ^{16 19}  

Human PAX6 ^+/− heterozygotes usually have normal-sized eyes but otherwise show a range of abnormalities similar to those in Pax6 ^+/− mice. Clinical conditions associated with PAX6 ^+/− heterozygosity include aniridia (absent or hypoplastic iris), Peters’ anomaly (keratolenticular adhesions with loss of posterior cornea), iridolenticular adhesions, iridocorneal adhesions, cataract, corneal opacity, congenital nystagmus, and foveal hypoplasia. ^{2 3 13 20 21} Postnatal changes include earlyonset glaucoma, corneal vascularization, corneal infiltration by inflammatory cells (autosomal dominant keratitis), and accumulation of goblet cells in the corneal epithelium. The corneal epithelium is not maintained adequately, and this is thought to involve a deficiency of limbal stem cells. ²²  

Eye development seems to be unusually sensitive to Pax6 dosage and both PAX6 gene duplication in humans ²³ and experimentally induced high expression levels in the mouse ²⁴ can also cause eye abnormalities. PAX77 ^+/− transgenic mice are hemizygous for five to six copies of a human PAX6 yeast artificial chromosome (YAC) inserted at a single locus. ²⁴ The YAC contains the human PAX6 gene with all the enhancer and promoter elements. The amino acid sequence of human PAX6 is identical with mouse Pax6 and the human PAX6 regulatory elements are fully functional in mouse, so that the transgene is expressed in an appropriate tissue-specific manner and fully functional. ²⁴  

The extra copies of PAX6 can compensate for the Pax6 deficiencies in both Pax6 ^−/− and Pax6 ^+/− mice. On a wild-type Pax6 ^+/+ genetic background, however, this overexpression of Pax6 in PAX77 ^+/− mice results in eye abnormalities, which overlap with those seen in Pax6 ^+/− mice. These hemizygous PAX77 ^+/− Pax6 ^+/+ mice often had microphthalmia, small corneas; flat irides; abnormal, small or absent ciliary bodies; small abnormal lenses; and abnormal photoreceptors, but the phenotype was variable. 

Pax6 has multiple pleiotropic roles in cell proliferation, migration, adhesion, and signaling ²⁵ and may act cell-autonomously or nonautonomously. Experimental studies have identified several primary cell-autonomous roles for Pax6 that underlie the eye abnormalities in homozygous Pax6 ^−/− mice. ^{6 26 27 28 29 30} Some of the eye abnormalities in heterozygous Pax6 ^+/− mice are caused by nonautonomous effects and may be secondary to primary affects of low Pax6 in the surface ectoderm and lens. ^{14 15} Nonautonomous effects probably also underlie some of the defects caused by abnormal migration of neural crest cells ¹⁸ but cell-autonomous effects may also be involved because Pax6 appears to be expressed transiently in the corneal stroma, ¹⁷ and chimera experiments indicate that Pax6 acts cell-autonomously in this tissue. ²⁸  

Analysis of mouse chimeras has been used successfully to distinguish between the cell-autonomous and nonautonomous effects of Pax6 ^−/− and Pax6 ^+/− genotypes on eye development. ^{14 26 27 28} One important observation to emerge from studies of Pax6 ^+/−↔wild-type chimeras was that Pax6 ^+/− cells were excluded from the lens epithelium of Pax6 ^+/−↔wild-type chimeras by embryonic day (E)16.5. Although Pax6 ^+/− cells contributed to tissues of the anterior segment, the chimeras did not show iris hypoplasia or other anterior segment abnormalities that occur in nonchimeric Pax6 ^+/− mice. This exclusion of Pax6 ^+/− cells from the lens epithelium is interpreted as a cell-autonomous effect, and it produces an almost entirely wild-type lens. It was further proposed that the largely wild-type lens cells exerted a nonautonomous effect on the anterior segment and prevented the formation of anterior segment abnormalities. ¹⁴ One of the purposes of this study was to determine whether PAX77 ^+/− cells are similarly excluded from the lens. 

In the present study, we first examined the range of eye abnormalities caused by the PAX77 ^+/− genotype on different genetic backgrounds. We then analyzed fetal PAX77 ^+/−↔ wild-type chimeras to investigate the underlying developmental causes of these eye abnormalities and distinguish between cell-autonomous and nonautonomous effects of elevated Pax6 expression in the eye. The use of these chimeras to demonstrate a cell-autonomous effect of increased Pax6 expression in the neurocortex has been reported elsewhere. ³¹ Both low levels of Pax6 in Pax6 ^+/− heterozygotes and high levels in PAX77 ^+/− transgenics cause lens abnormalities, and so we specifically wanted to use chimeras to test the hypothesis that PAX77 ^+/− cells would be depleted or excluded from the lens epithelium, like Pax6 ^+/− cells in our Pax6 ^+/−↔wild-type chimeras. 

Materials and Methods

Mice

Transgenic PAX77 mice on an outbred CD1 background were a kind gift from Veronica van Heyningen (MRC Human Genetics Unit, Edinburgh) and were produced in her laboratory. ²⁴ The PAX77 transgene was maintained on an outbred albino CD1A background (homozygous Gpi1 ^a/a, CD1 strain) and also crossed to inbred CBA/Ca mice. Outbred CD1 and inbred CBA/Ca, BALB/c, and A/J strain mice were obtained from Bantin & Kingman (Aldbrough, UK) and Harlan Olac (Bicester, UK). The mice were bred and maintained in the Medical Faculty Animal Area at the University of Edinburgh. Animal work was performed in accordance with U.K. institutional guidelines, Home Office regulations, and the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

Chimera Production

Two series of E16.5 chimeric fetuses (series SCA and SCB) were generated by aggregation of eight-cell–stage embryos, as described previously, ³² based on the original method of Tarkowski. ³³ One of each pair of aggregated embryos was hemizygous (Tg ^+/−) for the TgN(Hbb-b1) reiterated β-globin transgenic sequence that can be identified by DNA in situ hybridization on histologic sections and acts as a lineage marker. ^{34 35} The aggregated embryos also differed at the albino locus (Tyr) and for Gpi1, encoding glucose phosphate isomerase. Each series produced PAX77 ^+/−↔wild-type (PAX77 ^−/−) experimental chimeras and wild-type (PAX77 ^−/−)↔wild-type (PAX77 ^−/−) control chimeras. The genotype combinations for the experimental chimeras were PAX77 ^+/−, Gpi1 ^a/a , Tyr ^c/c , Tg ^−/−↔PAX77 ^−/−, Gpi1 ^b/b , Tyr ^+/+, and Tg ^+/− in series SCA and PAX77 ^+/−, Gpi1 ^b/b , Tyr ^+/+, Tg ^+/−↔PAX77 ^−/−, Gpi1 ^a/a , Tyr ^c/c , and Tg ^−/− in series SCB. 

For series SCA, crosses between CD1A females and CD1A-PAX77 ^+/− males produced PAX77 ^+/− and PAX77 ^−/− eight-cell stage embryos, all of which were Gpi1 ^a/a , Tyr ^c/c , and Tg ^−/−. These embryos were aggregated to PAX77 ^−/−, Gpi1 ^b/b , Tyr ^+/+, and Tg ^+/− eight-cell embryos produced from crosses between (C57BL/6 x CBA/Ca)F1 females (Gpi1 ^b/b , Tyr ^+/+, and Tg ^+/+) and males from stock BTC that were Gpi1 ^b/b , pigmented (Tyr ^+/+), and homozygous for the transgenic lineage marker (Tg ^+/+) on a largely (C57BL/6 x CBA/Ca)F1 genetic background. In PAX77 ^+/−↔wild-type experimental chimeras from series SCA, the transgenic lineage marker was carried by the wild-type embryo. 

The PAX77 ^+/− transgene was backcrossed for five generations onto the CBA/Ca inbred strain to produce series SCB chimeras. In this series, the embryos from the PAX77 ^+/− x PAX77 ^−/− cross were positively labeled with the β-globin marker transgene. For series SCB, eight-cell embryos were obtained from crosses between CBA-PAX77 ^+/− females and BTC males to produce PAX77 ^+/− and PAX77 ^−/− embryos, all of which were Gpi1 ^b/b , Tyr ^+/+, and Tg ^+/−. These embryos were aggregated to PAX77 ^−/−, Gpi1 ^+/+, Tyr ^c/c , and Tg ^−/− embryos produced from intercrosses between (BALB/c x A/J)F1 females and (BALB/c x A/J)F1 males. 

For each series, chimeras were produced as described previously, ³² except that embryos were cultured in either M16 ³⁶ or KSOM ^{37 38} culture medium and handled in either M2 ³⁹ or KSOM-H ⁴⁰ handling medium. Females were superovulated and mated to appropriate males. Eight-cell stage embryos were collected, denuded of their zonae pellucidae, and aggregated in the combinations described herein. After overnight culture, the E3.5 aggregated embryos were surgically transferred to the uteri of E2.5 pseudopregnant Gpi1 ^c/c females (hybrid stock CF1 ³² ) and recovered at E16.5 (which was timed with respect to the recipient female). 

Analysis of Chimeras

The recipient females were killed at E16.5, and the fetuses were dissected into cold PBS. GPI1 electrophoresis was performed on limb and tail tissue, and scanning densitometry was used to estimate the global chimeric composition as percentage of GPI1-A and percentage of GPI1-B, as described previously. ³² Any maternal contamination would have been represented as a GPI1-C band, which was excluded from the analysis. Heads were fixed and embedded in paraffin wax for analysis of the compositions of specific eye tissues by DNA-DNA in situ hybridization to detect the β-globin transgenic lineage marker. ³⁵ The fetal torsos were digested to obtain DNA for PCR genotyping to distinguish between control wild-type (PAX77 ^−/−)↔wild-type (PAX77 ^−/−) and experimental PAX77 ^+/−↔wild-type (PAX77 ^−/−) chimeras. 

For DNA-DNA in situ hybridization, fetal heads were sectioned coronally at 7 μm and mounted on slides coated with 3-aminopropyltriethoxysilane (Sigma-Aldrich, Poole, UK) and subjected to in situ hybridization with a digoxigenin labeled probe for the β-globin transgene, as described previously ³⁵ with the amendment that, after dewaxing, the slides were placed in 10 mM sodium citrate and microwaved at full power four times for 5 minutes each before being allowed to cool for 30 minutes. Hybridized probe was detected by diaminobenzidine (DAB) staining for peroxidase labeled antibody (polyclonal sheep anti-digoxigenin Fab fragments conjugated to horseradish peroxidase; Roche, Welwyn Garden City, UK). The slides were counterstained with hematoxylin, and the percentage of transgene-positive cells in each of the selected tissues was estimated in the chimeric embryo heads by counting nuclei with and without a hybridization signal under phase contrast light microscopy. Typically between 200 and 500 nuclei were scored as Tg-positive or -negative. Not all Tg ^+/− nuclei produce the expected hybridization signal, because only part of the nucleus may be included in the section. ³⁵ Consequently, the percentage Tg-positive cells was corrected with a tissue-specific correction factor determined by the percentage of Tg-positive cells present in eye tissues of E16.5 Tg ^+/−-positive control fetuses from (C57BL/6 x CBA/Ca) x BTC crosses. 

For PAX77 ^+/−↔wild-type experimental chimeras in series SCA the global PAX77 ^+/− contribution was estimated as the percentage GPI1-A, and the PAX77 ^+/− contributions to specific eye tissues were estimated from the corrected percentage of Tg-negative cells (or uncorrected percentage of albino cells for the RPE). Contributions to specific tissues were evaluated by comparing the corrected percentage Tg-negative/GPI1-A ratios (observed/expected ratios) for experimental and control chimeras. Underrepresentation of PAX77 ^+/− cells in specific tissue would produce a significantly lower observed/expected ratio in experimental chimeras than control chimeras. For PAX77 ^+/−↔wild-type experimental chimeras in series SCB the global PAX77 ^+/− contribution was estimated as percentage of GPI1-B and the PAX77 ^+/− contributions to specific eye tissues were estimated from the corrected percentage of Tg-positive cells (or uncorrected percentage of pigmented cells in the RPE). For this series of chimeras, underrepresentation of PAX77 ^+/− cells in a specific tissue would produce a significantly lower corrected percentage Tg-positive/GPI1-B ratio (observed/expected ratio) in experimental chimeras than control chimeras. 

Histology

Twelve-week-old mice were killed by cervical dislocation and the eyes were removed, weighed, and fixed in Bouin’s solution overnight at 4°C. After fixation, the lens was removed from most eyes via a hole cut in the posterior to facilitate sectioning. (If left intact, the brittle lenses often shatter during sectioning, and fragments can obscure other tissues.) The eyes were dehydrated in alcohols and embedded in paraffin wax and 7-μm microtome sections were cut along the anterior–posterior axis. Fetuses were decapitated, and heads were fixed in 3:1 ethanol/acetic acid overnight at 4°C, dehydrated, and embedded in paraffin wax, and 7-μm coronal sections were cut. Sections were mounted onto slides and stained with hematoxylin and eosin according to standard histologic procedures. For PAS staining, the slides were treated with periodic acid and then stained with Schiff’s reagent. 

Immunohistochemistry

Embryos were fixed in 4% wt/vol paraformaldehyde, dehydrated, and embedded in paraffin wax. The 7-μm sections were dewaxed and rehydrated. Antigen retrieval and immunohistochemistry was performed simultaneously on control and transgenic sections as described in Collinson et al. ²⁸ Pax6/PAX6 staining levels in PAX77 ^+/− and PAX77 ^−/− (wild-type) eyes were quantified by a modification of the protocol described in Song et al. ⁴¹ To control for variability of immunostaining efficiency between sections and samples, AP2α staining was used as an internal control, and Pax6 levels were expressed as a ratio relative to AP2α labeling intensity. Wax sections of E14.5 eyes were rehydrated, and antigen retrieval performed by boiling in 0.01 M citrate buffer (pH 6) 20 minutes. The sections were simultaneously double-labeled for Pax6 and AP2α with the PAX6 (1:40 dilution) and 3b5 (1:40 dilution) monoclonal antibodies, respectively, obtained from the Developmental Studies Hybridoma Bank (developed under the auspices of the National Institute of Child Health and Human Development, Bethesda, MD, and maintained by the University of Iowa, Iowa City). Secondary antibodies were Alexa Fluor-488 labeled goat anti-mouse IgG1 for Pax6 and Alexa Fluor-594 labeled goat anti-mouse IgG2b for AP2α (1:300 dilution for both; Invitrogen, Paisley, UK). PAX77 ^+/− and wild-type samples were processed identically and immunostained side-by-side. After mounting, fluorescence photographs were taken in monochrome using a camera (QICAM Fast1394) with identical exposure settings within commercial software (Volocity 4; ImproVision, Coventry, UK). Both Pax6 and AP2α images were captured from each field, and underexposed to avoid burnout. Images were exported as unprocessed .tif files into image-analysis software (Photoshop 7; Adobe Systems, San Jose, CA) and the Pax6 and AP2α fluorescence intensity of sections of the lens epithelium was measured using the histogram tool (Luminosity; Adobe Systems). Pax6 and AP2α luminosities were measured at identical sites on each lens epithelium, and a mean Pax6/AP2α ratio determined for each lens. The increase in Pax6 protein levels in PAX77 ^+/− lenses was then calculated by dividing the mean Pax6/AP2α value for PAX77 ^+/− lenses by the mean Pax6/AP2α value for wild-type lenses, separately for each genetic background. 

PAX77 Genotyping by PCR Analysis of Genomic DNA

PCR was performed on genomic DNA from digested mouse tail tips or fetal torsos. Tissues were digested overnight at 55°C in 1× digestion buffer (10 mM Tris [pH 7.8], 50 mM KCl, 5 mM MgCl₂, 0.45% Igepal, 0.45% Tween-20, and 0.1 mg/mL gelatin) with 0.6 mg/mL freshly prepared proteinase K added. The following day, the proteinase K was inactivated by heating the tubes at 100°C for 5 to 10 minutes. The tubes were then briefly centrifuged at 12,000g to pellet precipitates, and the solution was removed to a fresh tube. Genomic DNA was stored at −20°C until required. 

Two separate pairs of primers were used to amplify a 282-bp region of transgenic human PAX6 and a 398-bp region of mouse Pax6 which can be resolved using agarose gel electrophoresis. The sequences of the primers used to amplify the mouse PAX6 region were 5′-GAGGGTTTCCTGGATCTGG-3′ and 5′-CGCAAATACACCTTTGCTCA-3′ for the forward and reverse primers, respectively. The sequences of the primers used to amplify the human PAX6 transgene were 5′-CACGGTTTACTGGGTCTGG-3′ and 5′-CCGTGTGCCTCAACCGTA-3′ for the forward and reverse primers, respectively. A volume of 1 μL of genomic DNA was added to a PCR reaction of 50-μL total volume containing 1× Taq buffer, 1.5 mM MgCl₂, 200 ng of each primer, 10 nM of each dNTP, and 2.5 units of Taq DNA polymerase (Roche). The reactions were subjected to 30 cycles of 95°C for 1 minute, 57.5°C for 45 seconds, and 72°C for 45 seconds. A volume of 2 μL 6× DNA loading buffer (0.25% bromophenol blue, 15% Ficoll 400; Roche) was added to 12 μL of each PCR reaction, and the DNA was subjected to gel electrophoresis in a 2% agarose gel containing 1× TAE (40 mM Tris [unbuffered], 0.2 mM EDTA, and 0.1% acetic acid) and 0.6 μg/mL ethidium bromide. Electrophoresis proceeded at 100 V for between 1 and 2 hours. The DNA was then viewed by placing the gel under UV light with a wavelength of 320 nm. Both human PAX6 and mouse Pax6 DNA fragments were amplified from the DNA of mice carrying the PAX77 transgene, whereas only the mouse Pax6 fragment was amplified from wild-type mice and control chimeras. 

Results

Reduced Eye Size Caused by Interactions between Genetic Background and PAX77 Genotype

Microphthalmia was the most noticeable external phenotype in our founder PAX77 ^+/− transgenic mice, and this characteristic was retained when PAX77 ^+/− was maintained on a similar outbred CD1 genetic background. Overall, 4/29 (14%) of these CD1-PAX77 ^+/− mice, whose genotype was confirmed by PCR, were anophthalmic and the others were all severely microphthalmic. On this genetic background, the wet mass of PAX77 ^+/− eyes ranged from 1.2 to 11.3 mg. Figure 1Ashows that the mean mass was reduced to only 22% of the wild-type mean at 12 weeks—a highly significant difference. 

CD1-PAX77 ^+/− males were crossed to inbred CBA/Ca females to produce (CBA/Ca x CD1)F1-PAX77 ^+/− (hereafter abbreviated to F1-PAX77 ^+/−) mice. On this F1 hybrid genetic background, there was no significant difference in eye size and the mean mass of PAX77 ^+/− eyes was 95% of the wild-type mean. The eye size was reduced after further backcrosses to inbred CBA/Ca mice to produce partially congenic CBA-PAX77 ^+/− mice, and by the fifth-generation (N5) the mean mass of PAX77 ^+/− eyes was significantly reduced to 60% of the wild-type mean (Fig. 1A) . Thus, both the penetrance and expressivity of the PAX77 ^+/− microphthalmia phenotype was dependent on genetic background; the expressivity differed on CBA and CD1 genetic backgrounds and on an F1 genetic background the phenotype was rarely penetrant. (The overlap between F1 and F1-PAX77 ^+/− eye sizes was demonstrated by the observation that at 12 weeks, 38 of 43 F1-PAX77 ^+/− eyes were heavier than the lightest wild-type F1 eye, and 19 of 43 F1-PAX77 ^+/− eyes exceeded the wild-type F1 median eye mass.) Analysis of variance showed a highly significant interaction between genotype and genetic background for the mean eye mass of each mouse (P < 0.001), confirming that the effect of the PAX77 ^+/− genotype depends on the genetic background. 

The wet mass of the CBA-PAX77 ^+/− eyes ranged from 5.0 to 31.4 mg and varied not only among mice, but also between eyes in individual mice (Fig. 1B) . The percentage coefficient of variation of eye mass was below 20% for all three groups of wild-type mice (7.2%, 10.5%, and 4.4% for CD1, F1, and CBA genetic backgrounds, respectively) but was higher in CBA-PAX77 ^+/− (48.2%) and CD1-PAX77 ^+/− (39.6%) than in F1-PAX77 ^+/− (18.28%) mice. The within-mouse differences in wet mass between left and right eyes (Fig. 1B)were consistently low in wild-type mice on all three genetic backgrounds and F1-PAX77 ^+/− mice but statistically significant for CD1- and CBA-PAX77 ^+/− mice. Neither the left nor the right eyes were consistently heavier. The differences between eyes were greater for CBA-PAX77 ^+/− than CD1-PAX77 ^+/−, but this is partly because CD1-PAX77 ^+/− eyes were much smaller and partly because mice with no eyes or eyes that were too small to weigh were excluded so the CD1-PAX77 ^+/− variation is underestimated. The percentage difference provides a relative measure of variation that is less affected by differences in overall mean eyemass (Fig. 1C)and this shows larger percentage differences between left and right eyes in CBA-PAX77 ^+/− and CD1-PAX77 ^+/− than in F1-PAX77 ^+/− mice. 

Histologic Evaluation of Genetic Background Effects on Adult PAX77^+/− Eye Abnormalities

Figures 2A 2B 2C 2Dshow representative normal histologic features in adult wild-type CBA mouse eyes. Histologic examination of the lens and the vitreous was not performed because the lenses were removed from all the wild-type eyes (see the Materials and Methods section). Histologic features were entirely normal in wild-type eyes on all three genetic backgrounds except that outbred CD1 mice are albino (Tyr ^c/c ) and some CD1 eyes had no outer nuclear layer in the neural retina. The latter abnormality is characteristic of mice homozygous for the common Pde6b ^rd1 allele, which causes retinal degeneration. This allele is absent from the inbred CBA/Ca strain that we used and the histology of the outer nuclear layer was entirely normal. The major effects of the genetic background on the range of PAX77 ^+/− morphologic abnormalities are summarized in Table 1and are described in the following sections. 

F1-PAX77 ^+/− mice, whose genotype was verified by PCR, displayed a less severe eye phenotype than CD1-PAX77 ^+/− or CBA-PAX77 ^+/− mice. Histologic analysis of 14 F1-PAX77 ^+/− eyes, mostly with lenses removed, revealed several consistent abnormalities that were absent from wild-type F1 littermates. F1-PAX77 ^+/− eyes had smaller corneas than normal (Figs. 2A 2E)but there were no other obvious corneal abnormalities (Fig. 2G) . The iris contained cystlike structures (11/14 cases), and the ciliary body was disorganized and small compared with that in their wild-type littermates (Fig. 2F) . The retinal pigmented epithelium (RPE) appeared normal, but the neural retina was sometimes disorganized, particularly in the outer nuclear layer (ONL); Figure 2Hshows a minor abnormality. 

Histologic analysis of eight CBA-PAX77 ^+/− eyes revealed a variable phenotype that was intermediate between that of CD1-PAX77 ^+/− and F1-PAX77 ^+/− mice (Table 1) . The corneal diameter of the CBA-PAX77 ^+/− eyes was usually disproportionately reduced, relative to the eye diameter, compared with that of wild-type CBA eyes (Figs. 2I-2E) , but the corneal epithelium thickness was not significantly affected (Figs. 2C 2K) . In contrast to Pax6 ^+/− heterozygotes, ¹⁶ no goblet cells were identified in PAX77 ^+/− corneal epithelia stained with hematoxylin and eosin (Figs. 2G 2K 2N)or PAS (Fig. 2Rand data not shown). The iris was thickened and cystic (Fig. 2J)and the ciliary bodies were either absent or disorganized and small (Fig. 2J) . The retina appeared thinner (6/8 eyes) and the inner and outer nuclear layers were disorganized (4/8 eyes) and often contained rosettes (Fig. 2L) . Lenses were left in situ in some (3/8 eyes) CBA-PAX77 ^+/− eyes to allow histologic examination. Lenses were small relative to the size of the eyes, and all contained cataracts. The anterior chamber and the vitreous usually contained eosinophilic material (7/8 eyes; Fig. 2L ). 

Inspection of the intact CD1-PAX77 ^+/− eyes with a dissecting microscope showed that they all had lens opacities (cataracts). Histologic analysis of 13 CD1-PAX77 ^+/− eyes revealed a more severe phenotype than either CBA-PAX77 ^+/− or F1-PAX77 ^+/− eyes (Table 1) , which included severe microphthalmia and significant morphologic abnormalities in several eye tissues (Figs. 2M 2N 2O 2P 2Q 2R 2S 2T) . Corneal diameters were small but not disproportionately so in the smallest CD1-PAX77 ^+/− eyes. In all CD1-PAX77 ^+/− eyes, the iris was malformed, thicker than both wild-type and CBA-PAX77 ^+/− irides, and displayed adhesions and sometimes cystlike cavities (7/13; Fig. 2N ). The ciliary bodies were disorganized in all eyes, and some had a persistent pupillary membrane (Figs. 2O 2Q 2R) . CD1-PAX77 ^+/− lenses, left in situ for histology, were small and often vacuolated (Fig. 2S) , and the lens capsule, demonstrated by PAS staining, was thickened, wrinkled (6/13), and adherent to the adjacent tissues (Fig. 2R) . Densely packed cells with epithelial morphology formed multiple layers beneath the capsule in all CD1-PAX77 ^+/− eyes examined (implying subcapsular epithelial cell proliferation; Fig. 2Q ) and were also sometimes seen between differentiated lens fibers. Retinas showed several abnormalities (Figs. 2P 2S 2T) ; most were dysplastic (10/13) and contained rosettes (7/13), and some were folded (4/13) and/or adherent to the lens capsule (10/13), but they were still arranged in recognizable layers. Occasional absence of the outer nuclear layer was attributed to homozygosity for a retinal degeneration allele (e.g., Pde6b ^rd1 ) rather the PAX77 genotype, because it also occurred in some control, wild-type CD1 animals. Sometimes the vitreous cavity was not detectable (3/13), as the retina adhered to the entire posterior lens capsule (Fig. 2M) . In one of these eyes, the anterior chamber was also absent, because the iris was adherent to the cornea and anterior lens capsule. Less frequently, some CD1-PAX77 ^+/− eyes contained an ectopic ball of retinal cells, possibly retinal ganglion cells, between the retina and lens (Fig. 2T) . The anterior chamber and the vitreous often contained eosinophilic material (Figs. 2N 2O)similar to that seen in CBA-PAX77 ^+/− eyes. 

Genetic Background Effects on Fetal PAX77^+/− Eye Abnormalities

Eye histology was examined in coronal sections of the heads of E16.5 fetuses on the two most extreme genetic backgrounds (CD1 and F1) to compare wild-type CD1 (n = 6; Figs. 3A 3B ), wild-type F1 (n = 6; Fig. 3C ), CD1-PAX77 ^+/− (n= 8; Figs. 3D 3E 3F ), and F1-PAX77 ^+/− (n= 6; Figs. 3G 3H 3I ) eyes. Although the anterior segment appeared similar in wild-type embryos on both genetic backgrounds, the vitreous showed some differences. In the E16.5 wild-type F1 mice, most of the hyaloid system had disappeared from the vitreous, and the blood vessels were mainly located on the surface of the lens and on the inner surface of the retina, but persistent small clumps of cells were detected in some wild-type CD1 eyes (4/6 eyes; Fig. 3B ). 

Although the size of the PAX77 ^+/− eyes and lens was not markedly affected at E16.5, comparison of CD1-PAX77 ^+/− and F1-PAX77 ^+/− fetal eyes showed that genetic background differences were already apparent by this stage of development. Lens vacuoles occurred in some of the eyes on both backgrounds (8/14). In the F1-PAX77 ^+/− eyes, the vitreous appeared normal, but the neuroblastic layers were slightly disorganized in two of six eyes, and ingression of RPE cells was observed in the optic nerve sheath at the optic disc in some (3/6), whereas the other eyes showed a sharp boundary between the RPE and the optic nerve at the disc. Pigment cells were detected further proximally in the optic nerve sheath in one F1-PAX77 ^+/− eye (Figs. 3H 3I) . Neural retinal and vitreous abnormalities were more prominent in the CD1-PAX77 ^+/− eyes but RPE abnormalities were difficult to see on an albino CD1 background. Most strikingly, CD1-PAX77 ^+/− eyes showed various degrees of retinal folding (4/8). These folds were not restricted to any portion of the retina and could be observed along the dorsal, ventral, and nasal axes (Figs. 3D 3E) . Sometimes a retinal fold was in contact with the posterior surface of the lens. The posterior lens cells were epithelial-like, suggesting that they may have been displaced from their normal anterior location. Furthermore, the CD1-PAX77 ^+/− eyes sometimes contained a clump of retinal cells posterior to the lens (3/8 cases; Fig. 3F ) similar to that seen in some adult PAX77 ^+/− eyes. In some (3/8) of the CD1-PAX77 ^+/− eyes, the anterior chamber angle was obstructed with cells, but this was not seen in F1-PAX77 ^+/− eyes. 

Pax6 Expression in PAX77^+/− Eyes

Previous studies have shown that Pax6 expression is increased in PAX77 ^+/− transgenic mice on a CD1 genetic background. ^{24 31} Immunohistochemical staining suggested that Pax6 protein levels were also elevated in PAX77 ^+/− fetal eye tissues on a CBA/Ca inbred background (Figs. 4A 4B) . For example, Pax6 staining was stronger in the proximal RPE of E14.5 PAX77 ^+/− eyes than in wild-type eyes, where it is downregulated. ⁵  

It was possible that the variation in severity of the PAX77 eye phenotype may reflect differences in levels of PAX6 overexpression in ocular tissues among the mice on different genetic backgrounds. Initial investigation by Western blot on fetal eyes did not reveal clear differences in Pax6 expression between PAX77 ^+/− mice and wild-types on any genetic background (data not shown). This is presumably because the transgenic eyes have proportionately less Pax6-expressing tissue (smaller lens and smaller thinner retina) and this would offset any increase in Pax6 level per cell in those tissues. Pax6 immunostaining levels in the E14.5 lens epithelium were therefore compared by using a modification of the semiquantitative method described by Song et al. ⁴¹ (see the Materials and Methods section), with AP2α expression used as an internal control. This analysis is presented in Figures 4C and 4Dand more fully in Supplementary Table S1, and shows that for all three genetic backgrounds Pax6/PAX6 levels in the PAX77 ^+/− mice were slightly (20%–29%) but significantly higher than in the wild-types. Although the severities of the abnormal PAX77 phenotypes on different genetic backgrounds increased in the rank order F1 < CBA/Ca < CD1, the rank order for normalized Pax6 staining levels in PAX77 lenses was F1 < CD1 < CBA/Ca. There was no significant difference in Pax6 protein levels between CD1-PAX77 ^+/− and F1-PAX77 ^+/− lenses (t-test: P = 0.43, n = 8), which represent the two extremes of the PAX77 eye phenotypes. Analysis of variance confirmed that the normalized Pax6/PAX6 levels were significantly higher in PAX77 than wild-type lenses (P < 0.0001) but, in contrast to the results for eye mass (discussed earlier), there was no significant interaction between genotype and genetic background for Pax6 staining (P = 0.53). These results do not support the hypothesis that the more severe eye abnormalities in CD1-PAX77 ^+/− eyes reflect higher Pax6 protein levels than in F1-PAX77 ^+/− or CBA-PAX77 ^+/− eyes. 

Morphologic Abnormalities in Fetal PAX77^+/−↔ Wild-Type Chimeras

Two series of E16.5 experimental PAX77 ^+/−↔wild-type chimeras were produced, each with their own set of PAX77 ^−/−↔ wild-type control chimeras, as described in the Materials and Methods section. CD1A x CD1-PAX77 ^+/− embryos were incorporated into series SCA chimeras (genotype combination of SCA experimental chimeras: PAX77 ^+/−, Gpi1 ^a/a , Tyr ^c/c , Tg ^−/−↔PAX77 ^−/−, Gpi1 ^b/b , Tyr ^+/+, and Tg ^+/−). CBA-PAX77 ^+/− x BTC embryos were used for series SCB (genotype combination of SCB experimental chimeras: PAX77 ^+/−, Gpi1 ^b/b , Tyr ^+/+, Tg ^+/−↔PAX77 ^−/−, Gpi1 ^a/a , Tyr ^c/c , and Tg ^−/−). Eight experimental chimeras (seven SCA plus one SCB) and seven control chimeras (five SCA plus two SCB) were analyzed. The global PAX77 ^+/− contribution (or PAX77 ^−/− contribution in control chimeras) of cells from the PAX77 ^+/− x wild-type embryo used to produce the chimera was estimated as the mean percentage of GPI1-A in several nonocular tissues for series SCA and the percentage of GPI1-B for series SCB (see the Materials and Methods section). The contributions to the lens, corneal epithelium and corneal stroma were estimated as a corrected percentage of Tg-negative cells for SCA chimeras and corrected percentage of Tg-positive cells for SCB chimeras. The contribution to the RPE was estimated as the percentage of albino cells for series SCA and the percentage of pigmented cells for SCB chimeras. 

Fetal experimental PAX77 ^+/−↔wild-type chimeric eyes displayed a few morphologic abnormalities, some of which were similar to those seen in E16.5 PAX77 ^+/− transgenic fetuses. No gross abnormalities were seen in the cornea, lens, or nasal epithelium but retinal abnormalities occurred relatively frequently (Table 2) . The frequency of abnormalities correlated significantly with the global composition of the chimeras (P = 0.032 by Kendall’s correlation), such that chimeras with high overall proportions of PAX77 ^+/− cells tended to have more abnormalities (Table 2) . 

Retinal Dysgenesis in Fetal PAX77^+/−↔ Wild-Type Chimeras

Nine of the 16 eyes from the experimental chimeras displayed retinal dysgenesis in the form of large retinal folds similar to those in E16.5 PAX77 ^+/− nonchimeric fetuses. These retinal overgrowths occurred only in chimeras with >40% PAX77 ^+/− cells overall (Table 2)and were restricted to the nasal side of the retina. Spatial analysis of the distribution of wild-type and PAX77 ^+/− cells in the dysmorphic retinas suggests both cell-autonomous effects and nonautonomous effects are involved. In some cases, the folds were composed either entirely of PAX77 ^+/− or entirely of wild-type cells but in other cases the folds contained patches of both PAX77 ^+/− and wild-type cells (Fig. 5A 5B 5C) . The presence of both cell types in the folds suggests that this is a non–cell-autonomous defect resulting from Pax6 overexpression or a secondary defect caused by some other affected tissue. 

In some cases clusters of wild-type or PAX77 ^+/− cells were organized into small vesicles or nests of cells of the same genotype. Figure 5Dshows a small vesicle of wild-type (Tg ^+/−) cells within a PAX77 ^+/− domain in the retina. One possibility is that overexpression of Pax6 in PAX77 ^+/− cells causes a cell-autonomous effect on expression of cell surface adhesion molecules. This could cause reduced cell mixing, resulting in the formation of coherent clones of daughter cells and/or sorting-out of neighboring wild-type and PAX77 ^+/− cells. 

A small ectopic ball of retinal cells occurred between the lens and nasal side of the retina in one eye of PAX77 ^+/−↔ wild-type experimental chimera SCA12 (Fig. 5E) . This chimera had the highest global proportion of PAX77 ^+/− cells (87%) and the ectopic ball of cells was also mostly PAX77 ^+/−, but it did contain a few Tg-positive wild-type cells. The presence of both cell types suggests that this is a non–cell-autonomous defect caused by overexpression of Pax6 in PAX77 ^+/− tissues but further examples are needed to determine whether PAX77 ^+/− cells are overrepresented. Similar ectopic balls of cells occurred in some nonchimeric, adult, and fetal CD1-PAX77 ^+/− eyes (Figs. 2T 3F)but not in the wild-type eyes. 

Sorting-Out of PAX77^+/− and Wild-Type Cells in Other Ocular Tissues in Chimeras

In the anterior segment of control chimeras, Tg-positive and -negative cells contributed freely to all tissues and were fairly evenly distributed (Fig. 5F) . In contrast, cells in the developing ciliary bodies were segregated into patches of like genotype in 6 of 16 eyes from experimental PAX77 ^+/−↔wild-type chimeras (Figs. 5G 5H 5I ; Table 2 ). In these cases, the near-distal domain of the neural retina that is fated to become the ciliary body was largely composed of Tg-negative PAX77 ^+/− cells, but had balls or buds of Tg-positive wild-type cells projecting from the inner edge, facing the lens (Figs. 5H 5I) . This observation shows that PAX77 ^+/− cells can contribute to the early stages of ciliary body specification and suggests that Pax6 dosage could modulate cell surface adhesion properties in a cell-autonomous way that leads to reduced cell mixing and sorting-out of the cells in chimeras. 

Cell segregation in the iris pigment epithelium (IPE) occurred in 6 of 16 eyes from the experimental chimeras (Table 2)with PAX77 ^+/− cells ectopically bulging from a largely wild-type iris pigment epithelium (Fig. 5G) . Balls of PAX77 ^+/− cells were also occasionally observed in the trabecular meshwork of a few of these eyes, which may indicate segregation on the basis of Pax6 dosage in this tissue too (Fig. 5J) . However, it is possible that these were balls of dislodged PAX77 ^+/− IPE cells, for which the link to the rest of the IPE had been lost or was not visible. 

Segregation of PAX77^+/− and Wild-Type Cells in Nasal Tissues of Chimeras

A clump of serous gland tubules, on each side of the nasal region, expresses Pax6 (Fig. 5K)and is connected to and forms part of the nasal epithelia. Frequently, in the PAX77 ^+/−↔ wild-type chimeras, PAX77 ^−/− cells were segregated to the smaller peripheral tubules, whereas the larger central tubules were very predominantly PAX77 ^+/− (Fig 5L) . This segregation is incomplete and could be partly a consequence of clonal growth of the tubules. However, it occurred much more frequently in experimental PAX77 ^+/−↔wild-type chimeras (11/16) than control chimeras (1/14), and this difference was highly significant by Fisher’s exact test (P = 0.0008). If the two genetically distinct cell populations showed a greater tendency to clump together in larger patches during development of experimental chimeras than in wild-type chimeras, it may cause tubules to develop from cells of a single genotype more frequently. 

Abnormal Intrusion of PAX77^+/− Cells into the Optic Nerve Sheath in Chimeras

Coherent groups of RPE cells intruded into the sheath of the optic nerve in most (14/16) of the eyes of experimental PAX77 ^−/−↔wild-type chimeras (Figs. 5M 5N 5O ; Table 2 ). This involved both PAX77 ^+/− and PAX77 ^−/− RPE cells, implying that it was not a cell autonomous effect. Although single RPE cells also sometimes occurred in the distal optic nerve sheath in control PAX77 ^−/−↔wild-type chimeras, groups of RPE cells were never seen. This ingression of RPE cells into the optic nerve sheath was also noted in nonchimeric fetal PAX77 ^+/− eyes (Figs. 3H 3I) . 

Elevated Contributions of PAX77^+/− Cells to the Lens Epithelium and RPE in Fetal Chimeras

The analysis of chimeras discussed earlier indicates that PAX77 ^+/− cells contributed to all ocular and nasal tissues that were examined in the experimental PAX77 ^+/−↔wild-type chimeras in series SCA and SCB. The qualitative observations showed that, unlike Pax6 ^+/− cells, ¹⁴ PAX77 ^+/− cells were not excluded from the lens epithelium of chimeras by E16.5. Quantitative analysis of four eye tissues was undertaken to test whether PAX77 ^+/− cells were at least depleted from the lens epithelium of chimeras. 

For PAX77 ^+/−↔wild-type experimental chimeras in series SCA the PAX77 ^+/− contributions to specific eye tissues were estimated from the observed corrected percentage of Tg-negative cells (or observed percentage of albino cells for the RPE). This was divided by the contribution expected from the global estimate of the percentage of GPI1-A. Low observed/expected (corrected percentage of Tg-negative/percentage of GPI1-A) ratios suggest underrepresentation of PAX77 ^+/− cells in the specific tissue. Observed/expected ratios for four eye tissues from experimental chimeras in series SCA were compared to the equivalent ratios for the control chimeras. The results (Fig. 6)showed that PAX77 ^+/− cells were not underrepresented in the lens epithelium but were slightly but significantly overrepresented. The mean observed/expected contribution ratio for the lens epithelia in the 14 eyes from experimental SCA chimeras was 1.41 ± 0.14 versus 0.98 ± 0.08 for the 10 control chimeric eyes (P = 0.03). PAX77 ^+/− cells also showed a small but significant overrepresentation in the RPE (1.52 ± 0.19 vs. 1.03 ± 0.05; P = 0.04). PAX77 ^+/− cells were neither significantly overrepresented nor underrepresented in the corneal epithelium and corneal stroma. Addition of results for the SCB chimeras made no difference in the conclusions (Fig. 6B) . For chimeras in series SCB, observed contributions were estimated as the corrected percentage of Tg-positive cells or observed percentage of pigmented RPE cells and the expected contribution was estimated as the percentage of GPI1-B. These quantitative results suggest that elevated Pax6 expression in PAX77 ^+/− cells has a cell-autonomous effect in both the lens epithelium and RPE although, in each case, the effect is small. 

Discussion

Genetic Background and Stochastic Effects on the PAX77^+/− Phenotype

The original report of elevated Pax6 expression in PAX77 ^+/− transgenic mice on a wild-type (Pax6 ^+/+) background ²⁴ described a variable phenotype that included microphthalmia and other eye abnormalities. We found a similar range of abnormalities but showed that the PAX77 ^+/− phenotype is critically dependent on the genetic background. Penetrance and expressivity of different phenotypes increased in the order F1-PAX77 ^+/− < CBA-PAX77 ^+/− < CD1-PAX77 ^+/−. Eye mass correlated well with the severity of the morphologic abnormalities and provided a sensitive quantitative measure of both the severity and the variability of the phenotype. Semiquantitative analysis of Pax6 protein levels provided no evidence that the differences in PAX77 eye phenotypes on different genetic backgrounds were caused by differences in Pax6 protein levels in PAX77 ^+/− eyes. 

PAX77 ^+/− eye size was originally described as varying from almost normal to severe microphthalmia. ²⁴ but when we obtained CD1-PAX77 ^+/− mice in 2001, the phenotype was very severe, and no eyes were nearly normal in size. This apparent increase in severity with time may reflect selection for a permissive genetic background, that produces a more consistently severe phenotype, in the outbred CD1 strain as a byproduct of choosing mice with an unambiguous phenotype for breeding. Genetic background affects the penetrance of some homozygous Pax6 ^−/− phenotypes ¹² and may also explain the greater range of abnormal eye phenotypes reported for heterozygous Pax6 ^Sey/+ mice on a CD1 genetic background ¹⁸ than reported previously for Pax6 ^Sey/+ heterozygotes. This suggests that the gene pool of the outbred CD1 strain contains alleles of modifier genes that influence the expressivity and penetrance of various abnormal ocular phenotypes in Pax6 ^Sey/+ or PAX77 ^+/− mice. Selection for a severe phenotype would then select for combinations of modifier genes that were permissive for many abnormal PAX77 ^+/− phenotypes. 

Within-mouse variation in eye mass was greater on inbred CBA/Ca and outbred CD1 genetic backgrounds than the F1 hybrid background. Genetic differences do not cause this variation directly but could affect the occurrence of stochastic variation. Eye size is, therefore, likely to be affected by three factors: PAX77 ^+/− genotype, genetic background, and stochastic variation. This finding highlights the importance of standardizing the genetic background of mice when comparing the phenotypic effects of different mutants or transgenes. 

Differences between Phenotypes Caused by High and Low Pax6 Expression

Although both Pax6 ^+/− mice with low Pax6 levels and PAX77 ^+/− mice with elevated Pax6 expression cause microphthalmia and an overlapping spectrum of other eye abnormalities, there are significant differences in the abnormalities produced by the two genotypes (Table 3) . Abnormalities of CBA-PAX77 ^+/− corneas are reported in more detail elsewhere. ⁴²  

Possible Mechanisms Underlying Effects of Pax6 Overexpression in PAX77^+/− Tissues

Absence of Pax6 may either reduce or enhance proliferation, depending on the tissue, ²⁵ and the same may be true of overexpression. Indeed, we found PAX77 ^+/− cells to be overrepresented in the RPE and lens epithelium, whereas PAX77 ^+/− cells were depleted in some regions of the brain. ³¹ Pax6 overexpression slowed proliferation in rabbit corneal epithelial cells ^{45 46} but increased retinal cell proliferation. ⁴⁷ Reduced cell proliferation may affect eye size and/or corneal diameter. Increased cell proliferation could contribute to the retinal dysplasia in PAX77 ^+/− eyes and provide a competitive advantage for PAX77 ^+/− cells in the lens epithelium and/or RPE of PAX77 ^+/−↔wild-type chimeras, where PAX77 ^+/− cells were enriched. 

Some PAX77 ^+/− eye abnormalities may be due to a disruption of the ratio between Pax6 and other factors, with which it interacts to control gene expression. ⁴⁸ However, the extent to which Pax6 protein levels are elevated in each eye tissue is difficult to predict. According to Schedl et al., ²⁴ PAX77 ^+/− mice have five to six copies of the human PAX6 gene and express the equivalent of 5 to 8 copies of human PAX6 mRNA in both severely and mildly affected eyes. The increase in Pax6 levels in fetal PAX77 ^+/− lenses was less than predicted from the gene copy numbers. This finding is consistent with those in studies of fetal brain ³¹ and adult cornea ⁴² and probably reflects negative autoregulation of Pax6. ³¹  

Alternative splicing ^{49 50 51 52} and promoter usage ^{53 54} to produce different isoforms causes another complication. Pax6(5a) contains 14 more amino acids than the predominant Pax6 isoform, which disrupts its ability to bind to DNA. Overexpression of either the Pax6 or Pax6(5a) isoform in different systems has highlighted effects of specific isoforms on gene expression, cell biology, and eye development. ^{55 56 57 58 59 60 61} The ratio of Pax6 to Pax6(5a) protein isoforms differs in wild-type and PAX77 ^+/− brains. ³¹ If the isoform ratio varies with Pax6 genotype and among different eye tissues, the consequences for eye development would be difficult to predict. 

Cell-Autonomous Effect of Pax6 Overexpression in the PAX77^+/− Lens Epithelium

PAX77 ^+/− cells showed a small but significant overrepresentation in the lens epithelium, implying that the PAX77 ^+/− genotype has a cell-autonomous effect. The type of effect differs from that seen in Pax6 ^+/−↔wild-type chimeras where Pax6 ^+/− cells were excluded from the lens epithelium by E16.5. ¹⁴ The reciprocal cell-autonomous effects of Pax6 ^+/− and PAX77 ^+/− in chimeras suggest that the probability of lens epithelial cell survival depends on the level of Pax6 expression. For example, an effect of Pax6 level on differential cell adhesion may explain the chimera observations if cell adhesiveness increased in the order Pax6 ^+/− < wild-type < PAX77 ^+/−. If the two populations of lens epithelial cells in chimeras tended to sort out so that less adhesive cells surrounded more adhesive cells, ^{62 63} the less adhesive cells would preferentially move to the periphery and differentiate into fiber cells. 

A more direct effect of Pax6 expression on epithelial cell proliferation and/or lens fiber cell differentiation is probably also involved. Fiber cell differentiation is characterized by a decline in Pax6 levels and upregulation of the transcription factor cMaf, which regulates expression of lens crystallins. ^{6 64 65 66 67} Overexpression of Pax6 in the lenses of transgenic mice causes low levels of cMaf, ⁵⁸ incomplete fiber elongation, and cataracts. Thus, high Pax6 levels in PAX77 ^+/− lens epithelial cells could impair or delay their differentiation into fiber cells and explain both the occurrence of apparently undifferentiated lens epithelial cells among lens fibers in PAX77 ^+/− lenses (Fig. 2Q)and the enrichment for PAX77 ^+/− cells in the PAX77 ^+/−↔wild-type lens epithelium. The depletion of Pax6 ^+/− cells from the lens epithelium of Pax6 ^+/−↔wild-type chimeras ¹⁴ could similarly be explained by low Pax6 levels causing early upregulation of cMaf and precocious differentiation of Pax6 ^+/− lens epithelial cells into fiber cells. 

Cell-Autonomous and Nonautonomous Effects of Pax6 in Other PAX77^+/− Eye Tissues

Cell-autonomous effects were implicated by the overrepresentation of PAX77 ^+/− cells in the RPE and abnormal cell mixing in several other tissues. Normally, Pax6 expression is downregulated in the RPE during fetal development ⁵ but downregulation appears to be delayed in PAX77 ^+/− eyes (Figs. 4A 4B) . This delay might increase proliferation of PAX77 ^+/− RPE cells, resulting in their overrepresentation in chimeras. PAX77 ^+/− and wild-type cells failed to mix normally in several tissues of PAX77 ^+/−↔wild-type chimeras and tended to sort out or form vesicles or buds of cells of a single genotype, which were not seen in control chimeras. This observation implies that differences in Pax6 levels in PAX77 ^+/− and wild-type cells act cell autonomously to reduce cell mixing between cells of different genotypes, probably by altering expression of cell surface adhesion molecules. 

Intrusion of RPE cells into the optic nerve sheath and folding of the neural retina seem to be nonautonomous effects because both wild-type and PAX77 ^+/− cells were involved. Intrusion of RPE cells into the sheath of the optic nerve has been reported previously in Pax2 ^−/− embryos ⁶⁸ and the position of the optic cup/optic stalk boundary depends on expression of Pax2 and Pax6, which negatively regulate each other. Overexpression of Pax6 or a delay in its downregulation could cause the optic cup/optic stalk boundary to be displaced proximally. Convoluted overgrowth of retinal tissue occurs in the absence of a lens ^{6 26 30 69 70 71} and in the presence of a small lens in the aphakia mutant. ⁷² Thus, the retinal overgrowth in fetal PAX77 ^+/−↔wild-type chimeras could be secondary to a lens abnormality, but analysis of later stage chimeras may be necessary to demonstrate any correlation. 

Differences between Pax6^+/−↔Wild-Type and PAX77^+/−↔Wild-Type Chimeras

Overall, PAX77 ^+/−↔wild-type chimeras were less severely affected than homozygous Pax6 ^−/−↔wild-type chimeras ^{26 27 28} but had more abnormalities than Pax6 ^+/−↔wild-type chimeras ¹⁴ (Table 4) . The only effect seen in our Pax6 ^+/−↔wild-type chimeras was a cell-autonomous exclusion of Pax6 ^+/− cells in the lens epithelium by E16.5, producing an almost entirely wild-type lens. As suggested previously, ¹⁴ normalization of the lens genotype may have secondary effects elsewhere in the eye so that no morphologic abnormalities or nonautonomous effects of the Pax6 ^+/− genotype were apparent. In contrast, PAX77 ^+/− cells were overrepresented in the lens epithelia of PAX77 ^+/−↔wild-type chimeras, and these chimeras showed several morphologic abnormalities as well as other cell-autonomous and nonautonomous effects (Table 4) . 

 Supported by Wellcome Trust Grant 065035 (JDW).

 Submitted for publication December 19, 2007; revised January 30, 2009; accepted June 22, 2009.

 Disclosure: S.A. Chanas, None; J.M. Collinson, None; T. Ramaesh, None; N. Dorà, None; D.A. Kleinjan, None; R.E. Hill, None; J.D. West, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

Effect of elevated Pax6 expression and genetic background on eye size and variation in eye size within mice. (A) Mean mass (±SEM) of wild-type and PAX77 ^+/− eyes from 12-week-old mice on three different genetic backgrounds. (B) Mean difference in mass between left and right eyes (±SEM) for the six groups. (C) The percentage eye mass difference, calculated as [1 − (smaller eye mass/larger eye mass)] ×100%, for the six groups. The numbers above each bar in (A) are the number of eyes and in (B, C) the number of pairs of eyes. (Only mice with two undamaged eyes were included in B, C.) *P < 0.01; **P < 0.001 ; ***P < 0.0001 for comparison between PAX77 ^+/− and wild-type eyes on the same genetic background (t-tests).

Figure 2.

 

Histologic comparison of adult PAX77 ^+/− eye abnormalities (12 weeks) on three different genetic backgrounds. (A–D) Wild-type CBA eyes showing normal features: (A) whole eye with lens removed, (B) ciliary body (arrow) and iris, (C) cornea, (D) retina. (E–H) F1-PAX77 ^+/− eyes: (E) normal-sized eye with reduced corneal diameter and flattened cystic irides (lens removed); (F) abnormal ciliary body and cystic iris (arrow; lens present); (G) normal cornea (H) minor abnormality in ONL of neural retina (arrow) (neural retina has separated from RPE during processing). (I–L) CBA-PAX77 ^+/− eyes: (I) small eye with proportionately reduced corneal diameter (lens removed), (J) absent ciliary body and thickened, cystic iris (arrow), (K) cornea with normal epithelium, (L) eosinophilic material in the vitreous and thin retina with rosettes (RO, arrow). (M–T) Albino CD1-PAX77 ^+/− eyes: (M) severely microphthalmic and disorganized eye (lens intact), (N) absent ciliary body and thickened, cystic iris (arrow) and eosinophilic staining in the anterior chamber (O) persistent pupillary membrane (PM, arrow) and eosinophilic staining in the anterior chamber, (P) abnormal and folded retina, (Q) lens epithelium is multi-layered (three arrowheads on left) and there appears to be undifferentiated lens epithelial cells among the lens fibers (arrowhead on right), (R) extensively folded lens capsule (arrows; stained with PAS), (S) vacuolated lens and abnormal retina with rosettes (RO, arrow), (T) ectopic ball of retinal cells (RC, arrow) between retina and lens. D-ONL, disorganized outer nuclear layer; LC, lens capsule; PM, pupillary membrane; RC, retinal cells; RO, rosette.

Table 1.

 

Major Effects of the Genetic Background on PAX77 ^+/− Eye Abnormalities

Eye Abnormality	Genetic Background
		CD1	CBA	F1
Adult
Mean percentage of wild-type eye mass	22%	60%	95%
Microphthalmia	+	+	−
Absent or disorganized ciliary bodies	+	+	+
Cystic iris	+	+	+
Abnormal retina	+	+	+
Small cornea diameter	±*	+	+
Small lens	+	+	ND
Cataract	+	+	ND
Thickened iris	+	+	−
Eosinophilic staining in anterior chamber	+	+	−
Ectopic retinal cells posterior to lens	+	ND	ND
Persistent pupillary membrane	+	−	−
Fetus (E16.5)
Ingression of RPE cells into optic nerve sheath	ND, †	ND	+
Retinal folding	+	ND	−
Ectopic retinal cells posterior to lens	+	ND	−

 ND, not done (e.g. because lenses were removed from most F1-PAX77 ^+/− and CBA-PAX77 ^+/− eyes before embedding for histology).

*  Corneal diameter was not obviously disproportionately reduced in the smallest CD1-PAX77 ^+/− eyes.

†  CD1-PAX77 ^+/− mice are albino so any RPE cells in the optic nerve sheath would be less obvious.

Figure 3.

 

Histologic analysis of E16.5 fetal mouse eyes on two genetic backgrounds. (A, B) Wild-type, albino CD1 eyes showing normal histology but with persistent clumps of cells (arrow) in the vitreous in (B). (C) Wild-type pigmented F1 eye showing normal histology. (D–F) CD1-PAX77 ^+/− eyes with abnormal retinal folding (D, E) and ectopic ball (arrow) of cells (F). (D, arrow) A retinal fold in contact with the posterior surface of the lens in a region that contains cells with anterior lens epithelial morphology. (G–I) F1-PAX77 ^+/− eye showing normal morphology (G) but ingression of pigmented RPE cells into the optic nerve sheath (arrow in H; enlarged in I, arrow). C, clump of cells; PC, pigmented cells; RC, ectopic ball of retinal cells. Scale bar, 100 μm.

Figure 4.

 

Pax6 expression in PAX77 ^+/− eyes. (A, B) Pax6 immunohistochemistry of (A) an E14.5 wild-type CBA eye and (B) an eye of a hemizygous CBA-PAX77 ^+/− littermate. (C, D) Semiquantitative immunohistochemistry of Pax6 protein levels, normalized using AP2α immunohistochemistry as an internal control. (C) Normalized mean Pax6 fluorescence intensity (arbitrary units determined as described in the Methods section) for wild-type lens epithelia and PAX77 ^+/− lens epithelia, for different genetic backgrounds. *P = 0.03; **P = 0.001; ***P = 0.0002 (t-tests; n = 4–8 per group). (D) The proportionate increase in Pax6 expression in PAX77 ^+/− lens epithelia, compared to wild-type, for the different genetic backgrounds. Scale bar, 200 μm.

Table 2.

 

Eye and Nasal Abnormalities and Global Composition of Chimeras (Ranked by Composition)

Chimera	Abnormality Score					Cell Segregation Score*					Total Score	% Global Composition
		Neural Retina Dysgenesis	RPE in Optic Nerve Sheath	Ectopic Group of Retinal Cells	IPE (or Tr Mesh)	Iris or Ciliary Body (NR ep Layer)	Nasal (Serous) Epithelium		GPI1-A	GPI1-B
Experimental PAX77 +/−↔ wild-type chimeras
Series SCA
SCA13	0	1	0	1	0	0	2	20.0
SCA21	0	2	0	2	0	2	6	40.0
SCA11	1	1	0	1	1	1	5	55.1
SCA17	2	2	0	2	1	2	9	56.9
SCA18	2	2	0	0	1	1	6	68.9
SCA15	2	2	0	0	2	1	7	76.9
SCA12	2	2	1	0	1	2	8	87.2
Series SCB
SCB9	0	2	0	0	0	2	4		50.1
Frequency	9/16	14/16	1/16	6/16	6/16	11/16
Percentage	56	88	6	38	38	69
Control PAX77 −/−↔ wild-type chimeras
Series SCA
SCA19	0	(2), †	0	0	0	0	(2)	28.6
SCA20	0	0	0	0	0	0	0	62.1
SCA16	0	0	0	0	0	0	0	62.2
SCA14	0	0	0	0	0	0	0	71.1
SCA23	0	0	0	0	0	0	0	84.7
Series SCB
SCB11	0	0	0	0	0	1	1		12.7
SCB17	0	(1), †	0	0	0	0	(1)		52.0
Frequency	0/14	(3)/14	0/14	0/14	0/14	1/14
Percentage	0	(21)	0	0	0	7

 The total abnormality/segregation score in PAX77 ^+/−↔wild-type chimeras showed a significant correlation (Kendall τ = 0.618; P = 0.032) with the global composition (% GPI1-A or %GPI1-B depending which is produced by the PAX77 ^+/− cells). IPE, iris pigment epithelium; RGCs, retinal ganglion cells RPE, retinal pigment epithelium; Tr mesh, trabecular meshwork; NR ep layer, epithelial layer of iris and ciliary body that extends from the neural retina.

*  Abnormality or cell segregation score: 0, absent; 1, present unilaterally; 2, present bilaterally.

†  Only individual RPE cells were present in the optic nerve sheath of control chimeras.

Figure 5.

 

Abnormalities in E16.5 PAX77 ^+/−↔wild-type chimeras. Retinal abnormalities in E16.5 PAX77 ^+/−↔wild-type SCA chimeras: Retinal folds (arrows) could be composed largely of PAX77 ^+/− (Tg-negative) cells (A) or almost entirely of wild-type (Tg ^+/−) cells (B), or mixtures of cells of both genotypes (C). (D) Example of a small vesicle of wild-type (Tg ^+/−) cells (arrow) within PAX77 ^+/− domain of retina. (E) Detached group of retinal cells (arrow) between the lens and retina of chimera SCA12. Several wild-type (Tg-positive) cells were present in the ectopic clump of predominantly PAX77 ^+/− (Tg-negative) cells. Composition of iris, ciliary body, and lens in E16.5 fetal chimeras: (F) SCA control chimera with apparently random distribution of pigmented and Tg ⁺ cells in different eye tissues. SCA experimental chimeras: (G) contribution of PAX77 ^+/−, Tg ^−/− and PAX77 ^−/−, Tg ^+/− cells to all tissues (including the lens) and a bud of PAX77 ^+/−, Tyr ^c/c , and Tg ^−/− cells (arrow) projecting from the otherwise pigmented iris pigment epithelium. (H, I) Buds of Tg-positive wild-type (PAX77 ^−/−, Tg ^+/−) cells (arrows) in developing ciliary bodies that are otherwise PAX77 ^+/−, and Tg ^−/−. (J) Putative segregation of Tg-negative, PAX77 ^+/− cells (arrow) within the trabecular meshwork of an experimental SCA chimera. Serous gland: (K) Pax6 immunohistochemical staining of nonchimeric serous gland. (L) Partial segregation of Tg-negative, PAX77 ^+/− cells to larger central serous gland tubules (nasal region) and Tg-positive, wild-type cells to smaller tubules (arrows). Ingression of RPE cells into the optic nerve heads of experimental chimeras: (M) Control SCA chimera showing a sharp boundary between pigmented cells of the RPE and the optic nerve. (N, O) Ingression of pigmented cells (arrow) into the peripheral sheath of the optic nerve in experimental SCA and SCB chimeras. Because of the composition of the chimeras, the pigmented cells were PAX77 ^+/− in the SCB chimera shown in (N) but wild-type in the SCA chimera shown in (O). Scale bar: (A, B, C, L) 100 μm; (D, F–K, M–O) 50 μm; (E) 30 μm.

Figure 6.

 

Contribution of PAX77 ^+/− cells to four eye tissues in E16.5 fetal chimeras. (A) Observed tissue-specific contribution to the lens epithelium, corneal epithelium, corneal stroma (corrected percentage of Tg-negative cells) and RPE (percentage of albino cells) in SCA chimeras relative to the expected contribution from the global contribution (percentage of GPI1-A analysis) for each chimera. Comparison of the mean observed/expected ratios in experimental and control chimeras shows a small but significant overrepresentation of PAX77 ^+/− cells (high observed/expected ratio) in the lens epithelium and RPE in the 14 experimental chimeric eyes compared with the 10 eyes from control chimeras. (B) Histogram showing the overall observed/expected ratios for four eye tissues from 16 experimental chimeric eyes and 12 control chimeric eyes in series SCA and SCB combined. *P < 0.05; **P < 0.01 by Student’s t-test. ep, epithelium; str, stroma; RPE, retinal pigment epithelium.

Table 3.

 

Comparison of Eye Abnormalities Reported for Pax6 ^+/− (e.g. Pax6 ^Sey/+ or Pax6 ^Sey-New/+) and PAX77 ^+/− Genotypes

Eye Abnormality*	Pax6 +/−	PAX77 +/−
Similarities between Pax6 +/− and PAX77 +/−
Microphthalmia	Yes, †	Yes 24 42 , ‡
Small abnormal lens with cataracts	Yes 10 43	Yes 24 , ‡
Folded lens capsule	Yes 43	Yes, ‡
Multilayered lens epithelium	Yes 43	Yes 24 , ‡
Retinal dysplasia	Yes 18	Yes 24 , ‡
Adhesions in the anterior chamber (±obstruction of the angle)	Yes 16 18	Yes, ‡
Abnormal cell accumulation in the vitreous	Yes 18	Yes, ‡
Differences between Pax6 +/− and PAX77 +/−
Absent or reduced irides (±coloboma)	Yes 2	No 24 42 , ‡
Persistent bridge of cells between cornea epithelium and lens	Yes 10 14 15 43	No, ‡
Fusion of lens and corneal stroma (central corneal endothelium lost)	Yes 18	No, ‡
Thin corneal epithelium, §	Yes 16 19	No 42 , ‡
Corneal opacity with stromal vascularisation and cellular infiltrates	Yes 16 18	No 42 , ‡
Presence of goblet cells in the corneal epithelium	Yes 16	No 42 , ‡
Cystic and thickened irides	No	Yes 24 42 , ‡
Persistent pupillary membrane	No	Yes, §
Small, abnormal or absent ciliary bodies	No	Yes 24 42 , ‡
Disproportionately reduced corneal diameter	No	Yes 24 42 , ‡
Ingression of RPE cells into optic nerve sheath	No/ND	Yes, ‡

*  Penetrance of some abnormal eye phenotypes depends on genetic background (see text). Yes = feature reported for at least one genetic background; No = feature not reported; No/ND = feature not reported (but probably not done).

†  References ^{2 10 14 15 16 18 19 43 44} .

‡  Present study.

§  See Dorà et al. ⁴² for more detailed comparison of Pax6 ^+/− and PAX77 ^+/− corneas.

Table 4.

 

Comparison of Effects on E16.5 Fetal Eyes in Pax6 ^+/−↔Wild-Type and PAX77 ^+/−↔Wild-Type Chimeras

Tissue	Pax6 +/−↔Wild-Type Chimeras (Collinson et al.14)	PAX77 +/−↔Wild-Type Fetal Chimeras (Present Study)
Morphological abnormalities
Optic nerve/retina boundary	None	RPE cells in optic nerve sheath
Neural retina	None	Retinal dysgenesis and ectopic balls of RGCs
Probably nonautonomous effects
Optic nerve/retina boundary	None	PAX77 +/− and wild-type RPE cells in ONS
Neural retina	None	PAX77 +/− and wild-type cells in retinal folds (also wild-type cells among ectopic balls of PAX77 +/− RGCs* )
Cell-autonomous effects
Lens epithelium	Pax6 +/− cells excluded, †	PAX77 +/− cells overrepresented
RPE	None	PAX77 +/− cells overrepresented
Neural retina, iris pigment epithelium, ciliary body and iris NR ep layer	None	Some segregation of PAX77 +/− and wild-type cells

 NR ep layer, epithelial layer of iris and ciliary body that extends from the neural retina; ONS, optic nerve sheath; RGCs, retinal ganglion cells; RPE, retinal pigment epithelium.

*  Further work is needed to determine whether the effect is cell-autonomous or nonautonomous.

†   Pax6 ^+/− cells were excluded from the lens epithelium in the first study of Pax6 ^+/−↔wild-type chimeras.¹⁴ A more recent study showed that Pax6 ^+/− cells could contribute to the lens epithelium of Pax6 ^+/−↔wild-type chimeras,³⁰ but it was unclear whether they were depleted because the contribution of Pax6 ^+/− cells was not quantified.

Supplementary Materials

Supplementary Table S1 - (PDF) 

The authors thank Veronica van Heyningen for providing a founder stock of PAX77 mice; Jean Flockhart, Ailsa Travers, Julie Thomson, and Matthew Sharp for assistance with chimera production; Jane Quinn and Michael Molinek for technical advice; Denis Doogan, Maureen Ross and Mark Fisken for expert mouse husbandry; Rob Elton for statistical advice; and David Price for helpful discussion.