Three Novel Pax6 Alleles in the Mouse Leading to the Same Small-Eye Phenotype Caused by Different Consequences at Target Promoters

Abstract

purpose. To characterize three new mouse small-eye mutants detected during ethylnitrosourea mutagenesis programs.

methods. Three new mouse small-eye mutants were morphologically characterized, particularly by in situ hybridization. The mutations were mapped, and the candidate gene was sequenced. The relative amount of Pax6-specific mRNA was determined by real-time PCR. Reporter gene analysis used Crygf and Six3 promoter fragments in front of a luciferase gene and HEK293 cells as recipients.

results. The new mutations—ADD4802, Aey11, and Aey18—were mapped to chromosome 2; causative mutations have been characterized in Pax6 (Aey11: C→T substitution in exon 8, creating a stop codon just in front of the homeobox; ADD4802: G→A substitution at the beginning of intron 8 changes splicing and leads to an altered open reading frame and then to a premature stop codon; Aey18: G→A exchange in the last base of intron 5a leads also to a splice defect, skipping exons 5a and 6). Real-time PCR indicated nonsense-mediated decay in Pax6 ^Aey11 and Pax6 ^Aey18 mutants but not in Pax6 ^ADD4802 . This result is supported by the functional analysis of corresponding expression constructs in cell culture, where the Aey11 and Aey18 alleles did not show a stimulation of the Six3 promotor or an inhibition of the Crygf promoter (as wild-type constructs do). However, the Pax6 ^ADD4802 allele stimulated both promoters.

conclusions. Together with functional analysis in a reporter gene assay and immunohistochemistry using Pax6 antibodies, it is suggested that the Pax6 ^Aey11 and Pax6 ^Aey18 alleles act through a loss of function, whereas ADD4802 represents a gain-of-function allele.

Since the pioneering work of Walter Gehring’s group in 1995, ¹ Pax6 has been considered the master control gene of eye development in the animal kingdom. Mutations in its orthologous genes cause severe defects in eye formation in Drosophila, mice, and humans. In heterozygotes, mutations in Pax6 are responsible for microphthalmia, anterior polar opacity with corneal adhesions, and iris anomalies in mice ^{2 3} and rats. ⁴ In humans, PAX6 mutations lead mainly to aniridia, cataracts, and Peter’s anomaly characterized by corneal opacity and anterior polar cataracts. ^{5 6 7} Moreover, mutations in PAX6 are also considered to be causative for hereditary foveal hypoplasia, ⁸ and a susceptibility locus for myopia is linked to the PAX6 gene region on human chromosome 11. ⁹ In addition to the eye, Pax6 is mainly involved in brain development ^{10 11 12 13} and pancreas function. ¹⁴ Homozygous mutants die shortly after birth. ^{15 16 17}  

The Pax6 gene belongs to the family of paired-box and homeobox encoding genes. In the protein, the two highly conserved domains are separated by a linker segment and are followed by a C-terminal proline-, serine-, and threonine-rich region (PST domain ¹⁸ ). Alternative splice products of Pax6 are differentially expressed in several tissues. Most prominent is the presence or absence of exon 5a, which codes for an additional in-frame, 14-amino-acid peptide. ^{19 20}  

The exon 5a–encoded peptide affects the DNA-binding properties of Pax6; the balance between the canonical Pax6 and the splice product containing exon 5a [Pax6(5a)] is essential for normal lens physiology. Overexpression of Pax6(5a) in lens fibers results in the formation of cataracts and the upregulation of α5β1 integrins. ²¹ Specific loss of exon 5a by targeted mutation results in a mild phenotype in the mouse. The corresponding homozygous mutants are viable and experience only iris hypoplasia and defects in the cornea, lens, and retina. ²²  

Here we report three novel Pax6 alleles: one Pax6 mutation results in a severe phenotype caused by a mutation in intron 5a, which leads to a loss of exons 5a and 6 in the final transcript. Two other mutations lead to stop codons at positions in front or at the beginning of the homeodomain. Detailed morphologic analysis of the ocular phenotypes is combined with the genetic and molecular characterization of all three mutants and with functional analysis of the mutant proteins in cotransfection assays in cell culture. 

Materials and Methods

Mice

Mice were maintained under specific pathogen-free conditions at the National Research Center for Environment and Health (GSF) and were monitored within the ethylnitrosourea (ENU) mouse mutagenesis screen project. ^{23 24} The use of animals was in accordance with the German Law on Animal Protection, the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research, and the tenets of the Declaration of Helsinki. 

Male C3HeB/FeJ mice were treated with ENU (3 × 100 mg/kg body weight) at the age of 10 weeks according to Ehling et al. ²⁵ and were mated to untreated female C3HeB/FeJ mice. The offspring of the ENU-treated mice were screened at the age of 3 weeks for abnormalities of the eye (Aey). Mice with phenotypic deviations were tested for a dominant mode of inheritance; lines were kept by crossing heterozygotes inter se or to wild-type partners of the C3HeB/FeJ strain. 

The ADD4802 mutant was recovered at ORNL in the offspring of an ENU-treated male and was brought to the GSF in 1997. Male C3Hf/Sl mice at least 8 weeks of age were treated with ENU (4 × 77.5 mg/kg body weight) and were mated to untreated inbred T-stock/Sl females. ²⁶ Offspring derived from the treatment of stem cell spermatogonia were examined as soon as possible after 11 weeks of age for abnormalities of the eye. Heterozygotes showed small dense white opacities in the cornea and the lens with posterior synechiae; some mice appeared to have smaller eyes. The phenotype had an incomplete (but high) penetrance (Selby P, unpublished observation, September 1997); the line was kept by crossing heterozygotes to strain C3Hf/Sl (at ORNL) or C3HeB/FeJ (at GSF). 

Phenotypic Characterization

The eyes of the mutants were examined at the age of 3 weeks with a slit lamp. For histologic analysis, the embryos or their heads were fixed for 3 hours in Carnoy’s solution and embedded in JB-4 plastic medium (Polysciences Inc., Eppelheim, Germany) according to the manufacturer’s protocol. Sectioning was performed with an ultramicrotome (OMU3; Reichert-Jung, Walldorf, Germany). Serial transverse 3-μm sections were cut with a glass knife and stained with methylene blue and basic fuchsin. 

In situ hybridization of sections from embryonic days (E)12.5 to E17.5 was performed essentially as described by Grimm et al. ²⁷ Briefly, the embryos were fixed in paraformaldehyde and embedded in Jung Histowax (Cambridge Instruments, Nussloch, Germany). Sections (7–10 μm) were cut with the RM-2065 microtome (Leica, Nussloch, Germany) and mounted onto slides. The probe for Six3 was kindly provided by Peter Gruss (Max-Planck-Institute of Biophysical Chemistry, Göttingen, Germany); the Crygd and Cryaa probes were published previously. ²⁸  

For immunohistochemical analysis, embryonic heads (E15.5) were fixed in 4% paraformaldehyde in PBS (PFA), and sections were cut with a cryostat after cryoprotection. Sections were pretreated by briefly boiling in 0.01 M citrate buffer (pH 6). We used the primary antibodies against Pax6 (mouse IgG1, 1:50; Developmental Hybridoma Bank, Iowa City, IA) and Pax2 (rabbit, 1:100; Covance, Princeton, NJ). Cell nuclei were visualized by DAPI treatment. Secondary antibodies used were from Jackson Immunoresearch (West Grove, PA) and Southern Biotechnology Associates (Birmingham, AL). Specimens were mounted in Aqua Poly/Mount (Polysciences, Northampton, UK) and analyzed by using a confocal microscope (FV1000; Olympus, Tokyo, Japan). 

The sections were evaluated with a light microscope (Axioplan; Zeiss, Oberkochen, Germany). Images were acquired by means of a scanning camera (Progress 3008; Jenoptik, Jena, Germany) and imported into an image-processing program (Photoshop 7.0, Adobe Illustrator 10.0; Adobe, Unterschleissheim, Germany; PowerPoint, Windows, Unterschleissheim, Germany). 

Mapping

For linkage analysis, heterozygous Aey11 or Aey18 mutants on a C3HeB/FeJ background or ADD4802 mutants on a C3Hf/Sl background (generation 1, G1) were crossed to wild-type C57BL/6J mice; heterozygous carriers (G2) were backcrossed to C57BL/6J mice. DNA was prepared from tail tips of affected offspring from the backcross generation (G3) according to standard procedures. For a genome-wide linkage analysis of the Aey11 and Aey18 mutations, several microsatellite markers spaced by 20 cM on average were used for each autosome. ²⁹  

PCR and Sequencing

For molecular analysis, RNA was isolated from embryonic tissues or from different tissues of newborn ADD4802, Aey11, Aey18, or wild-type mice (C3HeB/FeJ). RNA samples were reverse transcribed to cDNA using a kit (Ready-to-Go; Pharmacia Biotech, Freiburg, Germany). Genomic DNA was isolated from tail tips or spleen of wild-type C3HeB/FeJ and C57BL/6J mice or heterozygous mutants according to standard procedures. Genomic DNA from homozygous mutants was prepared from embryonic tissues. For RT-PCR amplification, primers were selected from the EMBL-GenBank databases; for primers designed to amplify genomic DNA, the ENSEMBL mouse genomic database was used (Table 1) . 

PCR was performed using a thermocycler (Clonentech, Heidelberg, Germany; or PE-Biosystems, Weiterstadt, Germany). Annealing temperatures are shown in Table 1 . PCR products were analyzed on a 1% agarose gel. Sequencing was performed commercially (SequiServe, Vaterstetten, Germany) after isolation of the DNA from the gel using an extraction kit (Qia-Quick; Qiagen, Hilden, Germany) or cloning into the pCR-2.1 cloning vector (Invitrogen, Karlsruhe, Germany). To analyze restriction sites, PCR products were digested by restriction endonucleases, and the reaction products were analyzed on a 10% acrylamide or a 1% agarose gel. 

Real-time PCR was performed (LightCycler; Roche Diagnostics, Mannheim, Germany) according to instructions for users. For specific amplification of Pax6 cDNA (prepared from E17.5 embryos), the primers Pax6RT1-f and Pax6RT1-r (Table 1)were used; the Hprt gene was taken for standardization according to Vandesompele et al. ³⁰ The relative concentration of the Pax6 cDNA was calculated according to Pfaffl et al. ³¹ using the REST program (settings: primer efficiency = 80%; ΔCt = n(hprt) –n(mutant) for n = cycle numbers at the crossing point; ratio = eΔ^Ct(mutant)/mean eΔ^Ct(+/+)). 

Functional test of the Mutated Proteins

The effects of wild-type Pax6 and its mutated alleles were tested in a functional assay after cotransfection with a luciferase reporter gene system. The promoters for Six3 (-703/-349) and Crygf (-219/+37) driving the Luc reporter gene and the wild-type form of Pax6 cloned into the eukaryotic expression vector pcDNA3.1 have been described previously. ^{32 33} Previous experiments showed that HEK293 cells do not express Pax6. ³⁴  

For the cotransfection of the mutant Pax6 alleles, cDNA was amplified from homozygous embryos of all three mutant lines using the primer pairs pax6-L1 and pax6-R2 (Table 1)and was cloned into the pCR-TOPO vector (Invitrogen). Depending on its original orientation, the fragments were linearized by BamH1/Not1 or BstX1/XbaI digest, and the corresponding fragments were subcloned into pcDNA3.1 or pcDNA3.1D (Invitrogen). The correct cloning was confirmed by sequencing. 

For luciferase (Luc) reporter assays, HEK293 cells were cultivated in 96-well plates for 24 hours and transfected by PolyFect Transfection Reagent (Qiagen) using 0.7 μg plasmid DNA. The DNA mix in the transfection reactions contained 1 μg reporter vector, a DNA fragment representing part of the Crygf or Six3 promoter fused to the Luc reporter gene, 0.01 to 0.50 μg effector (either wild-type or mutated Pax6-pcDNA3, or the parental plasmid pcDNA3.1 as negative control), and 30 ng pRL-SV40 for internal transfection control (Promega, Mannheim, Germany). It contains a cDNA encoding Renilla luciferase, which was originally cloned from the marine organism Renilla reniformis (Rluc). The SV40 early enhancer/promoter region provides strong, constitutive expression of Rluc in a variety of mammalian cell types. Cells were harvested 48 hours after transfection, and cellular extracts were assayed with a reporter system (Dual-Luciferase Reporter Assay System; Promega) two times in sets of three, and the SEM was calculated (n = 6). The effects of the highest concentrations of transfected Pax6-containing expression plasmids were compared statistically using the t-test and software (SigmaPlot 8.02; SPSS Inc., Chicago, IL). 

General

Chemicals were from Merck (Darmstadt, Germany) or Sigma Chemicals (Deisenhofen, Germany). The enzymes used for cloning and reverse transcription were from Roche Diagnostics. Restriction enzymes were purchased from MBI Fermentas (St. Leon-Rot, Germany), Pharmacia (Freiburg, Germany), or New England Biolabs (Frankfurt, Germany). 

Results

Morphologic Description

The Aey11 and Aey18 mutants were identified at weaning because of their small eyes. The cornea at the anterior part is opaque and has a central depression caused by a corneal-lens adhesion, which is also opaque. The opacity in the Aey11 mutants is a bit stronger, whereas it is broader in the Aey18 mutants. In general, heterozygous carriers of the ADD4802 mutation express a similar phenotype (Fig. 1a) . During intensive breeding of the mutants, the phenotype appeared to be variable, including corneal opacities, cataracts, and small eyes with vacuolated lenses; frequently, this phenotype was only unilateral. In brother-sister mating, no viable homozygous mutants could be obtained. However, in embryos resulting from inter-se crosses of heterozygotes, a third phenotype without eyes became apparent in addition to the small-eye and the wild-type phenotypes. The number of embryos expressing these phenotypes did not deviate from a classical 1:2:1 ratio. 

Eyes of heterozygotes were investigated histologically at E12.5 and E17.5 (Fig. 1b) . At E12.5, the lens vesicle in the wild type was filled with the primary lens fibers, but in the heterozygous embryos the lens vesicle was not separated from the presumptive corneal epithelium; the lens stalk persisted and was present even at the end of embryonic development at E17.5 in all three mutant lines investigated. The lenses of the mutants showed swollen secondary fiber cells; cell nuclei were not yet degraded at this stage and were present in wild types and heterozygotes. 

In Situ Hybridization and Immunostaining

To determine whether the new mutants altered the expression pattern of genes known to be important for eye development, we checked for the expression of Pax2 and Pax6 by immunohistochemistry (Fig. 2)and for the expression of Six3, Cryaa, and Crygd by in situ hybridization (Fig. 3a 3b 3c) . The expression of Pax2 (Fig. 2a)was usually restricted to the area of the optic nerve head and the optic nerve itself. A similar feature was seen in all mutants at embryonic day 15.5; there were no major differences between wild type and mutants (Fig. 2b 2c 2d) . Because antibodies can be labeled with different fluorophores, it was possible to check the Pax6 expression in the same sections. In these posterior areas of the developing eye, Pax6 was less intensively expressed in the three mutants compared with the wild type. Particularly, in the Aey11 mutants, Pax6 expression was restricted to the inner segments of the retina. 

From the in situ hybridization data (Fig. 3a 3b 3c) , it became obvious that Six3 expression at E12.5 was stronger in Aey18 and ADD4802 mutant eyes, both in the retina and in the lens. In contrast, in Aey11 mutants, it was much weaker and was almost restricted to the anterior part of the lens. At later stages, there was weaker Six3 expression in all mutants in the lens, but the results in the retina were different: the Aey11 and Aey18 mutants had broader expression patterns in the retina, whereas in the ADD4802 mutant mice the Six3 expression was restricted to the innermost cell layer, as in the wild type (Fig. 3a) . 

The crystallin expression tested in these mutants (Cryaa, Fig. 3b ; Crygd, Fig. 3c ) was different among the mutants and compared with the wild type. In particular, Cryaa was expressed at early stages in the heterozygous Aey11 and Aey18 mutants, as in the wild type; in the lenses of the younger ADD4802 embryos, expression was weaker than in the wild type. However, at E17.5, Cryaa was expressed in the wild-type lens mainly in the anterior part of the fiber cells (only weak expression in the lens epithelial cells), whereas it was shifted in the Aey11 lenses to the equator and in the Aey18 lenses to their posterior pole; in the ADD4802 lenses, it was present throughout the lens. Finally, Cryaa was present in the innermost layer of the developing wild-type retina at E17.5. This feature was not observed in the Aey11 and Aey18 mutants; however, in the ADD4802 mutants, Cryaa was present at the anterior part of the retina even at E15.5, and at E17.5 the entire retina was colored, indicating the presence of Cryaa. 

The expression pattern of Crygd was more intense in all the mutants. Whereas Crygd expression in wild type was more pronounced at the anterior portion of the lens, its expression in the mutant lenses extended posteriorly. Finally, in the ADD4802 mutants, Crygd was expressed in the retina at E17.5; at this stage, it could even be detected in the cornea (Fig. 3c) . 

Linkage Analysis

Segregation analysis revealed for all three mutant lines clear linkage of the small-eye phenotype to the agouti (A) locus on mouse chromosome 2. Based on >400 G3 mice each for Aey11 and Aey18 and 50 G3 mice for ADD4802, the genetic distances to A were calculated to be 24.1 ± 4.8 cM, 21.6 ± 2.0 cM, and 26.0 ± 6.2 cM, respectively. 

To confirm the results of linkage to mouse chromosome 2, 45 small-eye offspring from the backcross (G3) of the Aey11 and Aey18 mutant lines were used for segregation analysis with additional chromosome 2 markers. The results placed both mutations between the markers D2Mit38 and D2Mit206 on mouse chromosome 2 (Fig. 4a 4b) . For the ADD4802 mutation, 50 G3 carriers were tested for linkage with the additional marker D2Mit249. One recombination was observed, resulting in a genetic distance of 2.0 ± 2.0 cM (Fig. 4c)and indicating that ADD4802 also maps to the same chromosome 2 region as Aey11 and Aey18. Based on the mapping results and the similarity of phenotypes, we hypothesized the three mutations to be allelic. 

In the Aey18 mutant, the calculated genetic distance is 11.1 ± 4.7 cM to D2Mit38 and 4.4 ± 3.1 cM to D2Mit206. Based on the ENSEMBL physical map of the mouse genome (http://www.ensembl.org/Mus_musculus), this interval covers physically 32 Mb containing approximately 60 known genes, including Pax6. Because of the small-eye phenotype of the mutants, Pax6 was considered an excellent candidate gene for all three mutant lines. 

Sequence Analysis

From all three mutant lines, homozygous embryos expressing anophthalmia could be recognized at E17.5. RNA was isolated from the head and reverse transcribed to cDNA. For PCR, primers were selected covering the coding region, including parts of the 5′ and 3′ untranslated regions. The designation of the introns and exons of Pax6 follows the classical nomenclature as outlined by Hanson. ³⁵ The translation start site is located in exon 4, and the alternatively spliced exon is referred to as exon 5a. 

In Aey11, a silent mutation was detected at the beginning of exon 6 that changed its third codon from AAC→AAT. However, because both triplets code for Asn, this is considered a polymorphic site without pathologic consequences. The only functional difference between Aey11 mutant and wild type was observed in exon 8 in front of the homeobox: a C→T alteration changed the codon CAG (coding for Glu) to a premature stop codon (TAG). It is predicted to lead to a truncation of the protein, which in the mutants consists of the paired box and most of the linker region (Fig. 5a) . In addition to this consequence at the protein level, the mutation created a new HindIII restriction site. Using primers in the introns flanking exon 8, two small fragments of 115 and 118 bp could be observed after digestion in the mutants only, whereas in different wild-type strains just the undigested fragment of 233 bp could be observed. The presence and absence of this restriction site perfectly cosegregated with the phenotype (Fig. 5b) . 

In Aey18, the situation seemed more complex. Exon 6 was missing from the cDNA. Therefore, we tested exon 6 and its flanking regions in genomic DNA using primers starting in the flanking introns. Sequence analysis revealed a G→A exchange at the last base of intron 5 that affected the highly conserved intron-exon boundary (Fig. 6a) . Moreover, the mutation destroyed a restriction site for MspMI or EcoNI, which was present in all wild-type strains that we analyzed, indicating that this mutation is not a polymorphic site (Fig. 6b) . 

Analyzing the splice variations in the region between exons 4 to 7, we used the primer pairs pax6-L3/pax6-Ex7-R1. In the wild type, we always found both splicing variants, the canonical form (without the additional exon 5a) and the exon 5a isoform. However, in the homozygous mutants, we found only one small but intense band of 88 bp that lacked exons 5a and 6. In the P1 heterozygous embryos, we found the expected mixture of all 3 splice variants, but the 88-bp band was the most intense one (Fig. 6c) . Because the loss of exons 5a and 6 did not change the open reading frame (ORF), the rest of the Pax6 transcription factor remained unchanged, leading to a loss of 86 (with exon 5a) or 72 (without exon 5a) amino acids from the center of the paired box. 

The cDNA of the ADD4802 mutants was longer than that of the wild-type cDNA; sequence analysis revealed an insert of 41 bp immediately between exon 8 and exon 9. Alignment with genomic sequences identified the insert as part of the regular intron 9, but its first base was changed from G→A, which led to the destruction of the splice site and the use of a cryptic splice site. The 41-bp insert also affected the ORF; after 11 novel amino acids, the protein was predicted to be truncated because of a premature stop codon in the new ORF. The truncation affected the homeodomain of the Pax6 protein (Fig. 7a) . The mutation led to the loss of a Tsp45I restriction site. It was tested in 5 mutant mice and in 5 different strains of wild-type mice and was confirmed to be the causative mutation and not a polymorphism (Fig. 7b) . 

Relative Quantification of Pax6-Specific mRNA

We determined the relative amount of Pax6-specific mRNA in all three new Pax6 alleles at the age of E17.5 in head mRNA because mRNA containing premature stop codons may be degraded (referred to as “nonsense-mediated decay”). We used the housekeeping gene Hprt as internal standard. As demonstrated clearly in Figure 8 , the Aey11 and Aey18 alleles had a reduced amount of Pax6 mRNA that was more pronounced in the Aey11 allele. Accordingly, nonsense-mediated decay of the mRNA was very likely in the Aey11 and Aey18 alleles. However, there was great variation in the amount of heterozygous ADD4802 mRNA; nevertheless, the smaller variation in the homozygotes supported the interpretation that these particular mutants were not affected by degradation of their Pax6 mRNA. These results are in good agreement with the expression pattern of Pax6 in the corresponding mutants analyzed by immunohistochemistry (Fig. 2) . 

Functional Analysis in Transfection Assays

We used an established assay system for two promoters, both coupled to Luc as the reporter gene and to HEK293 cells as recipients. ^{32 33 34} We chose two promoters—Crygf, which was demonstrated to be inhibited by functional Pax6 protein, ³⁶ and Six3, which was reported to be stimulated by Pax6. ³²  

Our results are summarized in Figure 9 . Using the Six3 promoter and the canonical, wild-type allele of Pax6, reporter gene activity was enhanced (Fig. 9a)as reported previously. ³³ However, the wild-type isoform containing the additional exon 5a does not have an effect on the Six3-promoter–dependent reporter gene activity. Therefore, it is also not surprising that the exon 5a isoforms of the Pax6 ^Aey11 and Pax6 ^ADD4802 mutations do not show any influence on the Six3 promoter activity. However, there is also no stimulatory effect for the canonical form of the Pax6 ^Aey11 and Pax6 ^Aey18 mutations. In contrast, the canonical form of the Pax6 ^ADD4802 allele has a significantly (P < 0.001) stronger stimulatory effect on the Six3-promoter dependent gene expression than the wild-type form. This is in line with the results of the real-time PCR demonstrating a loss of Pax6 mRNA for the Aey11 and Aey18 alleles but not for the ADD4802 allele. Moreover, the in situ hybridization data for the ADD4802 mutants at the early developmental stage (E12.5) showed increased Six3 expression compared with the other mutants and the C3H wild-type control. At later stages, this stimulatory effect could not be detected, most likely because of secondary effects. 

In contrast to the stimulatory action on the Six3 promoter, wild-type Pax6 represses the Crygf-mediated promoter activity ³⁷ ; this effect is stronger for the Pax6 exon 5a isoform (Fig. 9b) . For the Pax6 ^Aey11 mutation, an inhibitory effect can be seen only for the highest concentration of the exon 5a isoform. The Pax6 ^Aey18 allele and the Pax6 ^ADD4802 allele with exon 5a inhibit the Crygf promoter activity even at lower concentrations. In contrast, the canonical form of the Pax6 ^ADD4802 allele shows a stimulatory effect (instead of an inhibition as the wild-type form). Although the effect is not as large as was seen in the Six3 promoter assay, the difference was statistically significant (P < 0.001), when comparing Pax6 ^ADD4802 to the wild type or to the Pax6 ^Aey18 allele. No statistically significant difference was observed when the results for Pax6 ^ADD4802 were compared with those for Pax6 ^Aey11 . In conclusion, this demonstrates an altered function of the Pax6 ^ADD4802 allele compared with the other two Pax6 alleles, which should be considered “loss of function alleles” because of nonsense-mediated decay of the mRNA in these mutants. Moreover, these reporter gene assays correlate with the increased Crygd expression in the mutants (Fig. 3c) , presumably because of a loss of the inhibitory action of Pax6. 

Discussion

Pax6 is an essential gene for early eye development in a wide variety of different species, including fly, frog, fish, mice, rat, and humans. Mutations in the Pax6 gene are known to be causative for aniridia, Peter’s anomaly, and cataracts in humans and for “small eyes” in the mouse. Although the ocular phenotype is most obvious, mutations of Pax6 also affect the development of the nose, the pancreas and the brain (for reviews, see Prosser and van Heyningen ⁷ and Gupta et al. ³⁷ ). 

In the mouse, some Pax6 alleles have been described since the first molecular characterization of the small-eye phenotype in 1991. ² Favor et al. ³ reported nine independent Pax6 mutations of the mouse; seven of them resulted in premature termination of translation, and all exhibited phenotypes characteristic of null alleles. The other two alleles are characterized by a mutation in the Kozak sequence (leading to a reduced level of Pax6 translation product) or by a missense mutation in the homeodomain (leading to reduced DNA binding as shown by in vitro assays). They represent possible hypomorph alleles because their phenotypes are less severe, and the homozygotes develop rudimentary eyes and nasal processes. 

Just recently, four additional mouse alleles of Pax6 have been described by Thaung et al. ³⁸ In contrast to Favor et al., ³ they reported only one premature stop codon (in the PST domain) leading to the classical null phenotype. Interestingly, exactly the same molecular lesion was found in a human Aniridia patient. ⁵ The other three alleles represent missense mutations; two affect the paired domain and the third one the homeodomain. It is suggested that they all lead to an alteration of DNA binding either by changing the recognition site or by their general affinity to DNA. 

Here we described three mouse mutants leading to severe phenotypes: one (Aey18) caused by a loss of exons 5a and 6 (affecting the paired domain), the second (Aey11) by a premature stop codon in exon 8 (loss of the entire homeo- and PST-domain), and the third (ADD4802) by an insertion after exon 8, resulting in an altered ORF and a premature stop codon (loss of the C-terminal half of the homeodomain and the following PST domain). 

It is suggested that most mutant mRNA containing a premature stop codon is subject to nonsense-mediated decay. ³⁹ To test this suggestion we measured the relative Pax6 mRNA levels in all three truncated alleles by real-time PCR. Since the truncation occurs in all cases in the middle of the gene, it should be targeted by nonsense-mediated decay; its current mechanistic understanding predicts that transcripts containing a premature truncation codon 5′ to the last 50 nucleotides of the penultimate exon should be targeted by this process. ⁴⁰ However, this hypothesis could be confirmed only for two of the three alleles. The levels of Pax6 mRNA in the mutant Pax6 ^ADD4802 were similar to those in the wild type. 

Extensive phenotypical analysis of the three mutants shows similarities to the known mouse Pax6 null mutants—small eyes, corneo-lenticular adhesions, and opacities of the cornea and the lens. The homozygous mutants are anophthalmic and die perinatally. The similarity of the three phenotypes demonstrates that both loss of function (as indicated by the absence of most of the mutant PCR products) and the new function (as indicated by the presence of the Pax6^ADD4802 allele) are not compatible with proper eye development. 

In contrast to the similarity of the ocular phenotypes, the brain phenotypes of the Pax6 ^Aey18 mutants (lacking the major part of the paired domain) and of a mutant lacking a functional homeodomain (Pax6 ^4Neu ) are surprisingly different: in the Pax6 ^Aey18 mutant, the neurogenic, antiproliferative, and patterning effects are severely affected, but only subtle defects of the forebrain development could be found in the Pax6 ^4Neu mutants, ⁴¹ which shows the need for the analysis of an allelic series to determine the functional aspects of a given gene. A more detailed analysis of the distinct function of the different domains of the Pax6 protein could be obtained by investigation of various alleles of a gene even in different species. For human PAX6 mutations, a database (http://pax6.hgu.mrc.ac.uk/) contains all available information on genotypes and phenotypes. At present, there are 309 records in the database; the largest subgroup affects exons 4 and 6 and their connecting intron region including exon 5a (approximately 38%). 

Moreover, one major aspect in mutation research is the analysis of splice-site mutations. From human mutations it is often impossible to analyze the corresponding mRNA/cDNA, which is a great advantage of the mouse system. One of the major questions is the effect of the additional exon 5a compared with the canonical form of Pax6. Exon 5a codes for 14 additional amino acids in the paired domain and is thought to be important for tissue-specific modifications of the selection of the target sequence for binding and activation. ¹⁹ A T→C mutation at position –3 of the alternative splice acceptor site causes a distinct human ocular syndrome (bilateral juvenile cataracts, peripheral corneal opacification, glaucoma, pendular nystagmus, and unsteady oscillatory gaze). Using an in vivo splicing assay, the authors could show that this particular mutation caused increased splicing of exon 5a, and they attributed the phenotype to the shift of the ratio of the canonical (and predominant) form of PAX6 to the exon 5a isoform. 

This interpretation is supported by observations by Duncan et al., ²¹ who demonstrated overexpression of the Pax6(5a) isoform to result in cataracts using transgenic mice expressing Pax6(5a) under the control of the mouse Cryaa promoter, specifically in the lens. In contrast, the loss of exon 5a was demonstrated recently by a Pax6-exon5 knock-out mouse to be less severe in the eye than the entire loss of Pax6. ²²  

We conducted in our functional assays a parallel study of the alleles with and without the exon 5a isoforms. Even if the final phenotypes of the mutant eyes were similar, the underlying mechanisms might be different. The Pax6 ^ADD4802 allele obviously led to gain of function because it not only overstimulated its target gene Six3, it lost repressor activity to the Cryg genes as seen by in situ hybridization (Crygd) and promoter analysis (Crygf). Since the Pax6 ^ADD4802 allele is a hybrid protein consisting of the Pax6 paired domain and 11 additional amino acids instead of the homeodomain and the PST domain, it remains an open question whether this effect was caused by the presence of these 11 amino acids or by the truncation and loss of the homeodomain and PST domains. The results of the Pax6 ^Aey11 allele argue for a loss-of-function hypothesis because the Pax6 ^Aey11 allele behaves mainly like a null mutant; this is similar to, but not as pronounced as in, the Pax6 ^Aey18 mutants. 

In conclusion, we demonstrated here the finding of three novel, but phenotypically similar, small-eye phenotypes that were caused by different mutations in the Pax6 gene. The underlying mechanisms by which these mutations act are different: in two cases, action might be attributed to a loss-of-function effect caused by nonsense-mediated decay (Aey11 and Aey18), but it might be attributed to a gain-of-function event in the third case because, in these Pax6 ^ADD4802 mutants, the mutated Pax6 protein is most likely present and active. 

 Supported in part by Grant 01KW9610/I from the German Human Genome Project (DHGP) (EW, MHdA).

 Submitted for publication December 2, 2004; revised June 10, 2005; accepted October 19, 2005.

 Disclosure: J. Graw, None; J. Löster, None; O. Puk, None; D. Münster, None; N. Haubst, None; D. Soewarto, None; H. Fuchs, None; B. Meyer, None; P. Nürnberg, None; W. Pretsch, None; P. Selby, None; J. Favor, None; E. Wolf, None; M. Hrabé de Angelis, 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.

 

Table 1.

 

Primers Used for PCR

Designation	Laboratory No.	Sequence (5′→3′)	Accession No.	Location*	Annealing Temperature; Fragment Size
pax6-L1	38964	AAAGGCAGAAGACTTTAACCAAGGGC	X63963	5′-UTR	51°–62°C; 860 bp
pax6-R1	38965	CTCTTTCTCCAGAGCCTCAATCTGCTG	X63963	Exon 8
pax6-L2	38966	GATGAAGCTCAGATGCGACTTCAGC	X63963	Exon 8	51°–62°C; 752 bp
pax6-R2	38967	ACTGCTGTGTCCACATAGTCATTGGC	X63963	3′-UTR
pax6-L3	40104	GAGCTAGCTCACAGCGGGGCC	X63963	Exon 4	61°C; 161 bp including exon 5
pax6-R3	40105	TGCCCTGGGTCTGATGGAGCC	X63963	Exon 6
pax6-L4	MWG	CCGGCTTTGTGCAAGATTCTCC	BC011272	Upstream of exon 1	51°–62°C; 463 bp
pax6-R4	MWG	GGCTCTGAGAACTGGGATATACGTCC	BC011272 (X63963)	Exon 2/exon 3
pax6-Ex7-R1	MWG	AGGTTGCGAAGAACTCTGTTTATTGATGAC	X63963	Exon 7	60°C; with pax6-L3: 340 bp (including exon 5)
pax6-intron5-L1	42026	GTAAGCTTGTCATTGTTTAATGCATAC	AL512589	Intron 5	47°–54°C; 352 bp
pax6-intron6-R1	42027	GCCTTGGGGCAAGGTAGAC	AL512589	Intron 6
pax6-ex8-L1	MWG	GCCGTTCCGATTTCTCTTATTGTCC	AL512589	Intron 7	53°–64°C; 233 bp
pax6-ex8-R1	MWG	AGGTTTCTCTGCTTTGGAAATTTCCC	AL512589	Intron 8
Pax6RT1_f	MWG	AAGCAACAGATGGGCG	NM_013627	Exon 7	56°C; 200 bp
Pax6RT1_r	MWG	GCTTCATCCGAGTCTTC	NM_013627	Exon 8

*  Designation of exons follows the ENSEMBL database (ENSMUST00000016435).

Figure 1.

 

Morphologic analysis of three novel small-eye phenotypes. (a) An adult wild-type (Wt; top left) eye is compared with the three novel heterozygous mutants, Aey11 (top right), Aey18 (bottom left), and ADD4802 (bottom right). Phenotypes were typical for Pax6 mutants; the cornea is frequently opaque with a central dimple. (b) Wild-type and heterozygous mutants are compared at two stages during embryonic development (E12.5 and E17.5). At E12.5, the lens vesicle of the wild-type mouse is filled by the primary lens fiber cells; the neural retina (NR) and the presumptive cornea (C) are present. In the heterozygous mutants, the lens stalk, the connection between the lens and the cornea, is still present; the lens is much smaller than in the control. At E17.5 of wild-type embryonic development, the cornea and the lens are well separated, and an anterior chamber is formed. However, in the heterozygous mutants, the lens stalk connecting lens and cornea persists (arrow), and the anterior chamber is very small.

Figure 2.

 

Immunostaining of Pax6 and Pax2. To analyze the effects of the three mutations on the translational level, immunostaining of Pax6 and Pax2 were undertaken at E15.5 in heterozygous mutants. In the wild type (a), Pax2 (green fluorescence) is transcribed in the area of the optic nerve head, and immunostaining of Pax6 (red fluorescence) was detected in the developing retina. In the Aey11 (b), Aey18 (c), and ADD4802 (d) mutants, the transcription patterns of Pax2 are similar to those of wild type. Pax6 is less intensively expressed in the three mutants than in the wild type. In Aey11, the Pax6 transcription is restricted to the inner segments of the retina. (blue fluorescence) Counterstaining with DAPI (4′,6-diamidino-2-phenylindole) indicating cell nuclei.

Figure 3.

 

In situ hybridization with Six3, Cryaa, and Crygd. All three mutants were tested as heterozygotes for differences in the expression patterns of some of the downstream targets of Pax6, Six3 (a), Cryaa (b), and Crygd (c) during eye development. Scale bars, 100 μm. (a) Six3 is expressed at E12.5 in the anterior part of the lens vesicle; later, it is concentrated only in the central part of the anterior lens epithelium, and at the end of embryonic development it can be observed mainly in the inner nuclear layer. In the Aey11 mutants, Six3 is present in the inner nuclear layer at E15.5, and in the outer nuclear layer at E17.5. A similar expression pattern is observed in the Aey18 mutant. However, the ADD4802 mutant shows an expression pattern similar to that of the wild type, but it is stronger at E12.5 and at E17.5; Six3 still can be observed in the anterior parts of the lens fiber cells. (b) The expression pattern of Cryaa in the lens is similar in the wild type and in the Aey11 and Aey18 mutants. Its expression is restricted to the lens (the expression in the wild type at the very inner surface of the retina at E17.5 cannot be observed in the mutants); however, it is obviously weaker in the ADD4802 mutant at early stages but is present throughout the lens at E17.5. At this stage, low expression can be seen in the retina of the ADD4802 mutants. (c) Crygd expression is restricted in the wild type and, in most cases, in the mutants to the lens and its primary and secondary fiber cells. In the Aey18 mutant at E12.5 and in the ADD4802 mutant at E17.5, expression can be observed in the cornea (and in the innermost part of the retina; ADD4802 only).

Figure 4.

 

Haplotype analysis. Haplotype analysis is given for G3 offspring expressing the heterozygous phenotype. Black squares: C3H markers (the original background of all mutations); white squares: C57BL/6 strain markers. Total numbers of each haplotype (N) are indicated, as is the frequency of recombinations between the loci (n); the calculated genetic distance is given in cM ± 95% confidential interval.

Figure 5.

 

Molecular analysis of Aey11. Pax6 was tested as candidate gene. The structure of its transcript and the exons, nucleotides, and amino acids are designated according to Hanson. ³⁵ Numbering of the mRNA starts at the initial ATG, as indicated by the reference sequence NM_013627. Counting of the amino acids corresponds to the protein sequence NP_038655. PB, paired box; LNK, linker region; HB, homeobox; PST, PST-domain; CT, C-terminal peptide. (a) Sequence analysis. The mutation site is indicated in bold; sequence analysis indicates a C→T substitution in exon 8 changing the codon CAG (encoding Glu) into a stop codon; the HindIII site created in the mutant is boxed. The corresponding amino acid sequence is given above or below the cDNA. X indicates the premature stop codon in the mutant sequence. (b) Restriction analysis. A 233-bp fragment of exon 8, including its flanking regions, was amplified from genomic DNA of different wild-type and Aey11 mutant mice. The PCR fragment was analyzed by agarose electrophoresis with (+) or without (–) digest by HindIII. The fragment from the wild-type mice could not be digested, whereas the fragments from the 3 homozygous Aey11 mutants led to the expected digestion pattern; the 2 heterozygotes show both patterns. C3H, C3HeB/FeJ; JF-1, Japanese fancy mouse; DBA, DBA/2J; M, marker.

Figure 6.

 

Molecular analysis of Aey18. Pax6 was tested as candidate gene. The structure of its transcript and the exons, nucleotides, and amino acids are designated according to Hanson. ³⁵ Numbering of the mRNA starts at the initial ATG, as indicated by the reference sequence NM_013627. Counting of the amino acids corresponds to the protein sequence NP_038655. PB, paired box; LNK, linker region; HB, homeobox; PST, PST-domain; CT, C-terminal peptide. (a) Sequence analysis. The mutation site is indicated in bold in the genomic DNA; sequence analysis of the mutants indicates a G→A substitution in the last base of intron 5a (the intronic sequence is underlined); the BspMI site in the wild type (lost in the Aey18 mutants) is boxed. At the cDNA level, exons 5a and 6 are lost, as indicated by a deletion of 258 bp; the predicted amino acid sequence (missing 86 amino acids) is given above the cDNA sequence of the mutant. The asterisk in the genomic sequence marks a polymorphic site, which was found both in the C3H wild-type control and in the Aey18 mutants (and in the GenBank sequence AJ307468, but not in the entries X63963, NM_013627, and AF44223). (b) Restriction analysis. A 352-bp fragment of exon 6, including its flanking regions, was amplified from genomic DNA of different wild-type and Aey18 mutant mice. The PCR fragment was analyzed by agarose gel electrophoresis with (+) or without (−) digest by BspMI. The fragment from 5 wild-type mice of different genetic background led to the expected digest pattern, whereas the fragment from the homozygous Aey18 mutants contained only one restriction site; the 2 heterozygotes show the expected mixture of both patterns. C3H, C3HeB/FeJ; C57, C57BL/6J; T, test stock; JF-1, Japanese fancy mouse; DBA, DBA/2J; M, marker. Asterisk: BspMI site missing in the mutants. (c) Genomic organization of the regions of exons 5 to 7 is given schematically, including the sequences at the start and the end of the exons and at the end of intron 5a. The mutation site at the last base of intron 5a is given in bold (green, wild type; red, mutant; green A is responsible for the presence of a BspMI restriction site at this particular position). Dotted lines: splicing in the wild type; continuous line: skipping of exons 5a and 6. The interpretation is based on the analysis of the PCR products in wild-type C3H mice, heterozygous (+/−) and homozygous (−/−) Aey18 mutants. M, marker.

Figure 7.

 

Molecular analysis of ADD4802. Pax6 was tested as candidate gene. The structure of its transcript and the exons, nucleotides, and amino acids are designated according to Hanson. ³⁵ Numbering of the mRNA starts at the initial ATG, as indicated by the reference sequence NM_013627. Counting of the amino acids corresponds to the protein sequence NP_038655. PB, paired box; LNK, linker region; HB, homeobox; PST, PST-domain; CT, C-terminal peptide. (a) Sequence analysis. The mutation site is indicated in bold; sequence analysis indicates a G→A substitution in the first base of intron 8 (intronic sequence is underlined). The Tsp45I site present only in the wild types is boxed. The deduced protein sequence in the mutant is given in bold above the corresponding cDNA; the premature stop codon is double underlined. (b) Restriction analysis. A 259-bp fragment, including parts of exon 8 and intron 8, was amplified from genomic DNA of different wild-type and ADD4802 mutant mice. The PCR fragment was analyzed by polyacrylamide gel electrophoresis with (+) or without (−) digest by Tsp45I. The fragment from the wild-type mice could be digested, whereas the fragment from the 5 homozygous ADD4802 mutants remained undigested. BALB, BALB/c; C57, C57BL/6J; JF-1, Japanese fancy mouse; T, test stock; C3H, C3HeB/FeJ; M, marker.

Figure 8.

 

Relative quantification of Pax6-specific mRNA. Real-time PCR was performed using E17.5 mRNA from the head region of embryos recovered from inter-se crosses of heterozygous mutants of the three different Pax6 alleles. The relative amount of Pax6-specific mRNA is normalized to 1 for each wild type; scale bars indicate SD.

Figure 9.

 

Functional assays in cell culture cotransfection studies. For functional analysis, various wild-type and mutant Pax6 expression constructs (without and with exon 5a, as indicated by the suffix “+5a”) were cotransfected in increasing amounts (0.01–0.5 μg DNA) either with a Six3 promoter fragment or a Crygf promoter fragment, both cloned in front of a Luc reporter gene. All experiments were performed two times in sets of three. “Blank ” represents the luciferase activity from the empty vector; all values are given ± SD. Statistical significance (P < 0.001; t-test) is indicated by asterisks. (a) With the Six3 promoter, a slight increase was observed using wild-type Pax6 without exon 5a (with exon 5a, no stimulation was observed in wild type or in mutant forms). The stimulatory action of the form without exon 5a is significantly increased in the ADD4802 allele (indicating the gain-of-function character of this allele). In contrast, the Pax6 ^Aey11 and Pax6 ^Aey18 alleles do not show any effects indicating their loss-of-function character. (b) With the Crygf promoter, a slight decrease was observed using wild-type Pax6 without exon 5a, which is more pronounced, if exon 5a is present. The Pax6 ^Aey11 construct did not show an effect on the Crygf promoter, whereas the Pax6 ^Aey18 construct reduced the promoter activity to the background level. However, using the Pax6 ^ADD4802 expression construct, a slight stimulation of the luciferase reporter gene activity was observed; the corresponding construct with exon 5a showed a slight repression of the reporter gene activity similar to the effect of the Pax6 ^Aey18 construct.

The authors thank Erika Bürkle, Sabrina Hauser, Michaela List, and Monika Stadler for expert technical assistance. Part of the ADD4802 mutant analysis was performed by Daniela Woide during her practical course as an undergraduate student of the Technical University Munich. Oligonucleotides were provided by Utz Linzner (GSF-Institute for Experimental Genetics) or purchased from MWG Biotech (Ebersberg, Germany).