Introduction

Analysis of genetic mutations has provided key insights in developmental genetics and an allelic series of different types of mutations provides a powerful way of studying gene function. Many new mouse alleles have been generated by conventional mutagenesis but some have also been produced as by-products when a neomycin-resistance selection cassette (neo cassette) is included in the construction of a floxed allele for Cre-loxP conditional knockout experiments. Retention of the neo cassette often reduces gene expression, sometimes causing hypomorphic phenotypes (Wassarman et al. 1997; Meyers et al. 1998; Nagy et al. 1998; Ashery-Padan et al. 2000; Levin and Meisler 2004; Walisser et al. 2004a, b; Moza et al. 2007; Kerr et al. 2008; Lin et al. 2008; Wang et al. 2011) but it may also increase expression of a reporter transgene (Scarff et al. 2003). Sometimes new alleles are deliberately generated in this way but an abnormal phenotype is usually an unwelcome side effect of constructing a floxed allele so the neo selection cassette is often excised.

Pax6 is a semi-dominant gene with important roles in development of the eye, brain, nose, craniofacial tissues, pancreas, pituitary gland and pineal gland (Simpson and Price 2002; Favor et al. 2008). The floxed Pax6 tm1Ued allele (abbreviated to Pax6 fl) was originally produced to analyse these roles using a conditional knockout approach (Simpson et al. 2009) but the possibility that it might produce an abnormal phenotype has not been investigated.

Homozygous Pax6 / and heterozygous Pax6 +/ phenotypes have been well characterised. The mouse small eye mutation, Sey (Roberts 1967) was identified as a Pax6 null allele and renamed Pax6 Sey (Hill et al. 1991). Homozygous Pax6 / null mice, such as Pax6 Sey/Sey and Pax6 Sey-Neu/Sey-Neu, die around the time of birth with no eyes, no nasal cavities and abnormal brain development (Hogan et al. 1986; Hill et al. 1991; Schmahl et al. 1993; Grindley et al. 1995).

In Pax6 +/ heterozygotes, the nasal cavities are unaffected, brain defects appear to be restricted to the subcommissural organ (Estivill-Torrus et al. 2001) but eyes show a wide range of abnormalities, including small size, iris hypoplasia, cataracts, a closed anterior chamber angle (irido-corneal angle) and corneal abnormalities that may include a persistent lens stalk, indicative of incomplete separation of the lens and corneal epithelium (Hogan et al. 1986; Hill et al. 1991; Grindley et al. 1995; van Raamsdonk and Tilghman 2000; Collinson et al. 2001; Baulmann et al. 2002). The adult mouse Pax6 +/ cornea shows progressive deterioration, leading to loss of transparency and this is probably caused partly by a stem cell deficiency. The corneal epithelium is thin and fragile with reduced keratin 12 expression, increased cell turnover, accumulation of goblet cells, abnormal cell movement and abnormal wound-healing (Ramaesh et al. 2003; Davis et al. 2003; Sivak et al. 2004; Collinson et al. 2004; Ramaesh et al. 2005, 2006; Leiper et al. 2006; Douvaras et al. 2013). The Pax6 +/ cornea is also invaded by sub-epithelial blood vessels with associated connective tissue (pannus) and inflammatory cells, which causes corneal opacities (Ramaesh et al. 2003).

The severity of Pax6 mutant phenotypes may be affected both by stochastic effects and the genetic background (Pritchard 1973; Quinn et al. 1997; Ramaesh et al. 2009). For example, in heterozygous Pax6 +/ mice, eye size may vary between left and right eyes as well as among different mice (Roberts 1967; Clayton and Campbell 1968). Also, selection for a more permissive combination of modifier genes, present in gene pools of outbred mouse colonies, might explain why eye abnormalities in outbred mice may be more severe than for the same genotypes on more defined genetic backgrounds (Ramaesh et al. 2009). Examples of unusually severe adult Pax6 +/ eye phenotypes include highly abnormal lenses on an undefined mixed genetic background (Fig. 18 in Clayton 1970; Clayton and Campbell 1968) and adhesions and other mesenchymal abnormalities in outbred CD-1 mice (Kanakubo et al. 2006).

The Pax6 fl allele was produced by inserting a neo cassette, flanked by two loxP sites, into Pax6 intron 4 (in sense orientation with respect to the endogenous Pax6 locus) and another loxP site was inserted into intron 6. Thus, Pax6 exons 5, 5a and 6, which encode the paired domain, were flanked by loxP sites. Exposure of this conditional null Pax6 fl allele to Cre-recombinase causes recombination of the loxP sites and deletion of exons 5, 5a and 6, which are needed for Pax6 function. Deletion of these exons is expected to cause a nonsense frame-shift mutation that converts the floxed Pax6 fl allele to the deleted Pax6 Δ form, which is functionally a null allele (Simpson et al. 2009).

Unlike Pax6 / and Pax6 +/, the Pax6 fl/+, Pax6 fl/fl and Pax6 fl/ phenotypes have not been characterised. Despite the presence of the neo cassette in the Pax6 fl allele, homozygous Pax6 fl/fl mice were reported to be viable, fertile and phenotypically indistinguishable from wild-type mice (Simpson et al. 2009). Although the gross eye phenotype of Pax6 fl/fl homozygotes appeared to be normal, this was not investigated in detail and eye development is known to be particularly sensitive to both reduced and elevated Pax6 levels, resulting in a range of eye abnormalities (Hill et al. 1991; Schedl et al. 1996).

The aim of the present study was to use a more sensitive method to determine whether the Pax6 fl allele produced an abnormal eye phenotype. To compare the effects of a single Pax6 fl allele versus a single wild-type Pax6 + allele on eye phenotypes we crossed Pax6 fl/+ heterozygotes with Pax6 +/ mice, heterozygous for the Pax6 Sey-Neu (Pax6 ) null allele on the same genetically defined, inbred background. This revealed that eye phenotypes in Pax6 fl/ compound heterozygotes were more abnormal than those in Pax6 +/ heterozygous mice, implying that the Pax6 fl allele is not equivalent to the wild-type Pax6 + allele.

Materials and methods

Mice

Animal work was approved by the University of Edinburgh Ethical Review Committee and performed in accordance with UK Home Office regulations under project licenses PPL 60/3635 and PPL 60/4302. Heterozygous Pax6 +/Sey-Neu mice (abbreviated to Pax6 +/) were maintained by crossing to inbred CBA/Ca mice (N > 20 backcross generations) and genotyped by polymerase chain reaction (PCR) (Quinn et al. 1996). Heterozygous Pax6 tm1Ued/+ (abbreviated to Pax6 fl/+) mice (Simpson et al. 2009) on an outbred CD-1 genetic background were crossed to CBA/Ca (N = 8) and genotyped by PCR (Simpson et al. 2009). (Unlike CBA/J and some CD-1 mice, the CBA/Ca inbred strain does not carry the Pde6b rd1 retinal degeneration mutation, which can be confused with other retinal defects.) Adult and fetal mice were produced from Pax6 fl/+ × Pax6 +/ and Pax6 fl/+ × Pax6 fl/+ crosses. The morning that a vaginal plug was found was defined as embryonic day (E) 0.5 and E14.5 fetuses were chilled in ice cold PBS, decapitated and tail tips used for genotyping. Adult mice were killed at 12 weeks by cervical dislocation following inhalation of gaseous isoflurane anaesthetic.

Histology and morphometric measurements

Fetal heads were fixed in 4 % paraformaldehyde (PFA) overnight at 4 °C and processed to paraffin wax. For adults, left and right eyes were removed immediately, rinsed, dried and weighed to determine the mean eye mass. The mean eye and corneal diameters were both calculated from two orthogonal measurements of each eye using a Wild M5A dissecting microscope fitted with a calibrated measuring graticule in one eyepiece. Eyes were then fixed in 4 % PFA overnight at 4 °C and processed to paraffin wax. One adult eye was sectioned per mouse in an anterior–posterior plane to produce 7 µm sections through the cornea, lens and retina. To avoid adult lenses shattering the wax block was kept wet during sectioning. Fetal heads were sectioned anterior–posterior in a horizontal plane to produce 7 µm sections through the developing eye that included cornea, lens, and retina. Morphological features were compared in sections using standard haematoxylin and eosin (H & E) staining. Tissue sections of adult eyes were photographed and measured using a calibrated Zeiss Axiovision 4.8 digital camera system on a Zeiss Axioplan 2 compound microscope. The average thickness of the whole cornea, corneal stroma (plus endothelium) and corneal epithelium were determined from measurements at two places, near the centre of the cornea, in sections close to the mid-section in one eye per mouse and the number of corneal epithelial cell layers was counted.

Immunohistochemical staining

Antibodies and methods used for immunohistochemical staining of Pax6, keratin 5, keratin 12 and keratin 19 were the same as those reported previously for mouse eyes (Dorà et al. 2014). Sections were dewaxed in Histoclear (Fisher Scientific), rehydrated through graded alcohols to water, incubated in 3 % hydrogen peroxide in methanol for 15 min. then washed in 0.1 % phosphate buffered saline-Tween (PBS-T). To unmask antibodies, slides were incubated in 0.01 M citrate buffer (pH 6.0), in a 95 °C water bath for 35 min. then allowed to cool for 20 min. and washed in PBS. Sections were treated with 10 % blocking serum (species according to secondary antibody), 0.1 % bovine serum albumin (BSA) in PBS for 1 h. at room temperature then incubated with the primary antibody, overnight at 4 °C. Slides were washed in PBS, incubated in blocking serum (10 min.) then secondary antibody (45 min. at room temperature), washed in PBS and incubated with avidin-biotin reagent (ABC RTU Vectastain, Vector Labs PK-7100). Antibody was then visualised with 3,3′-diaminobenzidine (DAB) stain (5.9 ml 20 mM Tris pH 7.6, 100 μl 50 mg/ml DAB, 1 μl H2O2). Finally, sections were lightly counterstained with haematoxylin, dehydrated and coverslips were mounted with DPX mounting medium (BDH from VWR). Control slides were treated with blocking serum in place of primary antibody but otherwise treated identically.

For Pax6, the primary mouse anti-Pax6 antibody (Developmental Studies Hybridoma Bank, University of Iowa; DSHB Cat# pax6, RRID:AB_528427) was diluted 1:500 and the secondary antibody (Vector labs BA-9200 biotinylated goat anti-mouse) was diluted 1:200. For keratin 5 (K5), the primary rabbit anti-K5 antibody (Abcam ab53121) was diluted 1:100 and the secondary antibody (Vector Labs BA-1000, biotinylated goat anti-rabbit IgG) was diluted 1:200. For keratin 12 (K12), the primary goat anti-K12 antibody (Santa Cruz Biotechnology sc-17101) was diluted 1:500 and the secondary antibody (Vector Labs BA-5000) biotinylated rabbit anti-goat IgG was diluted 1:200. For keratin 19 (K19), the primary rabbit anti-K19 antibody (Lifespan Biosciences LS-C3372) was diluted 1:200 and the goat anti-rabbit secondary antibody was the same as that used for K5.

Statistical analysis

Minimum group sizes were guided by previous experience and power calculations, using standard deviations from comparisons of other genotypes (Ramaesh et al. 2009) to ensure sufficient power to detect, as significant (P < 0.05), mean eye mass differences of 5 mg. One-way analysis of variance (ANOVA), followed by Tukey’s multiple comparison tests, and non-parametric Kruskal–Wallis tests, followed by Dunn’s multiple comparison tests, were calculated using GraphPad Prism 5.0c (GraphPad Software Inc., San Diego, USA).

Results

Abnormal Pax6 fl/ eye phenotype

Pax6 fl/+ and Pax6 +/ mice were crossed to produce offspring of four genotypes: Pax6 +/+ (wild-type, WT), Pax6 fl/+, Pax6 +/ and Pax6 fl/. Homozygous Pax6 fl/fl mice, on a similar, predominantly CBA/Ca genetic background, were produced in a separate Pax6 fl/+ × Pax6 fl/+ cross. Phenotypes of Pax6 +/+ , Pax6 fl/+, Pax6 fl/fl, Pax6 +/ and Pax6 fl/ eyes are shown in Figs. 1 and 2. For both E14.5 fetuses and 12-week-old adults, heterozygous Pax6 fl/+ and homozygous Pax6 fl/fl eyes appeared to be grossly unaffected and qualitatively equivalent to Pax6 +/+ (wild-type). However, Pax6 fl/ compound heterozygous eyes were clearly more severely affected than Pax6 +/ heterozygotes. At E14.5 (Fig. 1), the most obvious abnormalities of heterozygous Pax6 +/ eyes were that the anterior chamber was not yet formed, as reported previously (Ramaesh et al. 2003), and some had an indentation in the middle of the cornea, indicative of a persistent lens stalk (lens-corneal plug). In contrast, lenses of E14.5 Pax6 fl/ compound heterozygotes eyes were small and usually did not protrude beyond the anterior of the optic cup. The anterior chamber had not formed and mesenchymal cells extended from the cornea into the anterior of the optic cup and coated the anterior of the lens. Persistent lens stalks, involving the surface ectoderm, were only seen in some of the eight E14.5 Pax6 fl/ eyes that were examined (e.g. Fig. 1f).

Fig. 1
figure 1

Histology of E14.5 fetal mouse eyes of five genotypes. af H & E stained sections showing whole eye morphology of a wild-type (WT) Pax6 +/+ (+/+), b Pax6 fl/+(fl/+), c Pax6 fl/fl (fl/fl), d Pax6 +/ (+/−) and e, f Pax6 fl/ (fl/−) fetal eyes. The arrowheads in d and f indicate indentations in the corneas caused by persistent lens stalks (lens-corneal plugs). gl Histology of anterior of eye of g +/+, h fl/+, i fl/fl, j +/− and k, l fl/− fetal eyes. Figures show representative eyes from 4 fetuses that were sectioned for each genotype. Different fl/− eyes are shown in e, f, k and l. Scale bars a (for af) = 200 μm; g (for gl) = 50 µm. ac anterior chamber, co cornea, le lens, re retina

Fig. 2
figure 2

Histology of 12-week adult mouse eyes of five genotypes. af H & E stained sections showing whole eye morphology of a wild-type (WT) Pax6 +/+ (+/+), b Pax6 fl/+(fl/+), c Pax6 fl/fl (fl/fl), d Pax6 +/ (+/−) and e, f Pax6 fl/ (fl/−) eyes. Eyes in e and f are small and are shown at a greater magnification than eyes in ad. The arrowhead in e indicates an indentation in the cornea, caused by a persistent lens stalk. g, h Histology of anterior of fl/− eyes showing lens abnormalities. il Histology of central cornea of i +/+, j fl/+, k fl/fl and l +/− eyes. mp Histology of ciliary body and irido-corneal angle (anterior chamber angle) of m +/+, n fl/+, o fl/fl and p +/−, showing closed angle in p. qt Histology of anterior of fl/− eyes showing pigmented presumptive iris tissue adhering to the central cornea q, r and peripheral cornea s, t. Different views of the same fl/− eye are shown in e, g, r and s but they are not all from the same section. Different views of another fl/− eye are shown in h and t. Figures show representative eyes from 4 (Pax6 fl/fl) or 5 mice (other 4 genotypes) that were sectioned (1 eye per mouse). Scale bars a (for ad) and e (for e, f) = 500 μm; g (for g, h) = 100 µm. i (for it) = 50 µm; an irido-corneal angle, cb ciliary body, co cornea, ir iris, le lens, p pigmented presumptive iris tissue, re retina

At 12 weeks, Pax6 +/+, Pax6 fl/+ and Pax6 fl/fl eyes all appeared grossly normal (Fig. 2a–c, i–k, m–o) apart from some common processing artefacts, which included separation of the retinal pigment epithelium from the neural retina and some lens damage. As described elsewhere (Ramaesh et al. 2003), Pax6 +/ eyes (Fig. 2d, l, p) showed a range of abnormalities, including slightly reduced size, hypoplastic irises, thinner corneal epithelium and sometimes a lens-corneal plug (persistent lens stalk). Adult Pax6 +/ eyes had an anterior chamber but, in some cases, the irido-corneal angle (anterior chamber angle) was closed, as reported previously (Baulmann et al. 2002). However, Pax6 fl/ eyes were much smaller and more severely affected (Fig. 2e–h, q–t). They had no ciliary body, anterior chamber or irido-corneal angle and a layer of pigmented tissue, with no pupil opening, adhered to the corneal endothelium. This is presumed to be an abnormal iris but a persistent pupillary membrane could also be present, as described for Pax6 Leca4/+ heterozygotes (Ramaesh et al. 2009). It is also possible that this is at least partly derived from the retinal pigment epithelium but we did not use markers to attempt to identify the origin of the pigmented layer. The Pax6 fl/ lens was small and dysplastic with vacuoles and sometimes adhesions to pigmented, presumptive iris tissue anteriorly and/or the retina posteriorly. In some Pax6 fl/ eyes a lens-corneal plug was present. The retina was severely dysplastic with, irregular layers, folds and rosettes (Fig. 2e) but the retinal pigment epithelium, adjacent to the neural retina, appeared grossly normal.

Quantitative comparisons of eye mass confirmed that heterozygous Pax6 +/ eyes were significantly smaller than wild-type Pax6 +/+ eyes (P < 0.001) but compound heterozygous Pax6 fl/ eyes were even smaller. Pax6 fl/ eye mass and diameter were significantly smaller than the other four genotypes (P < 0.001 in each case) but Pax6 fl/+ and Pax6 fl/fl genotypes did not differ significantly from Pax6 +/+ (Fig. 3a, b). The corneal diameter of Pax6 fl/ compound heterozygotes was smaller than the other four genotypes (P < 0.001 in each case) but it was also significantly smaller in Pax6 fl/fl than Pax6 +/+ (P < 0.01), Pax6 fl/+ (P < 0.001) and Pax6 +/ (P < 0.05), as shown in Fig. 3c. Furthermore, the relative corneal diameter (expressed as a percentage of eye diameter) was significantly smaller for Pax6 fl/ than Pax6 +/+ (P < 0.05), Pax6 fl/+ (P < 0.01) and Pax6 +/ (P < 0.05) but not Pax6 fl/fl (Fig. 3d). The Pax6 fl/fl relative corneal diameter was also significantly less than Pax6 +/+, Pax6 fl/+ and Pax6 +/ (P < 0.001 in each case).

Fig. 3
figure 3

Comparisons of eye and corneal sizes for 12-week adult mice of five genotypes. Each point is the mean of left and right eyes of one mouse and the horizontal bars are group means for a eye mass, b eye diameter, c corneal diameter and d relative corneal diameter (corneal diameter as a percentage of eye diameter). The numbers of mice per group were 5 +/+, 7fl/+, 4fl/fl, 5 +/− and 6fl/−. Groups were compared by 1-way analysis of variance (ANOVA; results shown on figure) and Tukey’s multiple comparison tests. Genotypes, with only different letters above the scatter plots, differ significantly (P < 0.05). Genotypes, with any shared letters above the scatter plots, do not differ significantly. Sexes are coloured differently (red female, blue male) but were not analysed separately. (Color figure online)

Morphometric measurements were also used to compare corneal thickness but variation within groups was relatively high and, as group sizes were small, these results are considered preliminary and are presented in the graphs included as Supplementary Fig. S1. Despite the small group sizes (thus low statistical power), there was also a clear trend for Pax6 fl/fl eyes to have thicker corneas than Pax6 +/+. Differences between these two genotypes were significant for the thickness of the corneal epithelium (P < 0.05) and whole cornea (P < 0.01) but not for the thickness of the corneal stroma (plus the thin corneal endothelium) or the number of corneal epithelial layers. Pax6 +/ mice are known to have thin and fragile corneal epithelia (Ramaesh et al. 2003) and the mean number of corneal epithelial cell layers was lowest for Pax6 +/ and Pax6 fl/. However, for the small samples analysed here, only Pax6 fl/ and Pax6 fl/fl differed significantly (P < 0.05).

Immunohistochemical staining for Pax6 was positive in the lens (not shown), corneal epithelium and retina of all 5 genotypes (Fig. 4). Although nuclear staining appeared to be stronger in Pax6 +/+, Pax6 fl/+ and Pax6 fl/fl than Pax6 +/ and Pax6 fl/ sections, Pax6 levels were not compared quantitatively.

Fig. 4
figure 4

Pax6 immunohistochemistry in corneal epithelia and retinas from 12-week adult mouse eyes of five genotypes. Pax6 immunostaining is shown by the brown DAB endpoint and sections were counterstained with haematoxylin. aj Corneal epithelium; kt retina. The lens epithelium is also visible in a and d. a, f, k, p Wild-type (WT) Pax6 +/+ (+/+), b, g, l, q Pax6 fl/+(fl/+), c, h, m, r Pax6 fl/fl (fl/fl), d, i, n, s Pax6 +/ (+/−) and e, j, o, t Pax6 fl/ (fl/−) mice. Scale bars for columns ae and ko, 50 μm; fj and pt, 20 µm. co cornea, en endothelium, ep epithelium, GCL ganglion cell layer, INL inner nuclear layer, ONL outer nuclear layer

Abnormal expression of corneal epithelial markers in Pax6 fl/ mice

Keratin 5 (K5) is normally present in the mouse corneal epithelium, conjunctiva and the limbal epithelium, which is a junctional zone between these two tissues (Byrne and Fuchs 1993; Douvaras et al. 2012). Strong K5 immunostaining was present in these three ocular surface epithelia in all five genotypes and the central corneal epithelium is illustrated in Fig. 5a–e.

Fig. 5
figure 5

Keratin 5, 19 and 12 immunohistochemistry of corneal epithelia from 12-week adult mouse eyes of five genotypes. Immunostaining is shown by the brown DAB endpoint and sections were counterstained with haematoxylin. ae Keratin 5 immunohistochemistry in the central cornea. The three cellular layers of the cornea are labelled in a. fo Keratin 19 immunohistochemistry in the limbus and peripheral cornea fj and central cornea ko. pt Keratin 12 immunohistochemistry in the central corneas. Genotypes: a, f, k, p Pax6 +/+ (+/+), b, g, l, q Pax6 fl/+(fl/+), c, h, m, r Pax6 fl/fl (fl/fl), d, i, n, s Pax6 +/ (+/−) and e, j, o, t Pax6 fl/ (fl/−). Scale bars for each column, 50 μm

Keratin 19 (K19) is a marker of the mouse limbal epithelium and conjunctiva but not the central corneal epithelium (Yoshida et al. 2006). The limbal and peripheral corneal epithelia of all five genotypes stained positive for K19 (Fig. 5f–j). In the central corneal epithelia, K19 immunostaining was weak or absent in Pax6 +/+ , Pax6 fl/+ and Pax6 fl/fl genotypes (Fig. 5k–m), patchy in Pax6 +/ (Fig. 5n) but stronger and clearly above background levels in Pax6 fl/ compound heterozygotes (Fig. 5o).

Keratin 12 (K12) is a specific corneal epithelial marker (Liu et al. 1993) and K12 staining was present in the central corneal epithelia of Pax6 +/+ , Pax6 fl/+ and Pax6 fl/fl eyes (Fig. 5p–r). Some weak and patchy staining was present in the Pax6 +/ corneal epithelium (Fig. 5s) but K12 staining was absent from the Pax6 fl/ compound heterozygous corneal epithelium (Fig. 5t).

Discussion

Compound heterozygous Pax6 fl/ mice had much more severe eye abnormalities than Pax6 +/ heterozygotes and expression of K12 and K19 were also more severely affected in Pax6 fl/ corneas. K12 is normally only expressed in the corneal epithelium and is regulated by Pax6 (Liu et al. 1993; Shiraishi et al. 1998). Conversely, K19 is normally expressed in the mouse limbal epithelium and conjunctiva but not the corneal epithelium (Yoshida et al. 2006). Consistent with previous reports, the Pax6 +/ corneal epithelium showed weak, patchy and sparse K12 staining (Ramaesh et al. 2003, 2005) and patchy K19 staining (Douvaras et al. 2013). Although K12 and K19 immunostaining was normal in the Pax6 fl/+ and Pax6 fl/fl ocular surface, it was even more abnormal in Pax6 fl/ than in Pax6 +/. The more strongly positive K19 staining and absence of K12 in the Pax6 fl/ central corneal epithelium suggests that it is abnormally differentiated. However, we did not determine whether cells in the Pax6 fl/ corneal epithelium expressed other specific limbal or conjunctival markers or test for hallmarks of conjunctivilisation. Together, the morphology and corneal immunohistochemistry results for Pax6 +/ and Pax6 fl/ eyes imply that the floxed Pax6 fl allele is not equivalent to the wild-type Pax6 + allele.

The observations that corneas in Pax6 fl/fl homozygotes were thicker and had a smaller diameter than normal indicates that an abnormal Pax6 fl phenotype is detectable in Pax6 fl/fl homozygotes as well as Pax6 fl/ compound heterozygotes but the Pax6 fl/fl phenotype is much less severe. Although Pax6 fl/fl homozygotes and Pax6 +/+ wild-type mice were produced by different genetic crosses, they were on very similar genetic backgrounds. Nevertheless, as sample sizes were small and corneal stromal thickness can be affected by histological processing, this evidence for an abnormal Pax6 fl/fl corneal phenotype should be considered preliminary. Apart from the subtle corneal differences revealed by morphometrics, there were no major qualitative morphological ocular abnormalities or aberrant K12 or K19 immunostaining in the corneal epithelium of Pax6 fl/+ or Pax6 fl/fl mouse eyes. The phenotypic evidence suggests that Pax6 fl is an abnormal allele with a minor effect on corneas of Pax6 fl/fl homozygotes but no effect on Pax6 fl/+ heterozygotes.

Most of the phenotypic abnormalities can be ordered as: −/− (anophthalmic; not shown) >fl/− > +/− > fl/fl > fl/+ = +/+ (normal). Although we have not investigated the molecular mechanisms, it is tempting to speculate how the different genotype-phenotype correlations could arise. For example, if this rank order of decreasing phenotype severity were caused by an equivalent rank order of increasing Pax6 levels, this would predict that the floxed Pax6 fl allele causes a relatively small quantitative reduction in Pax6 level, such that Pax6 fl/fl homozygotes have more Pax6 than Pax6 +/.

The above phenotype severity rank order is consistent with the differences in eye size but it does not fit so well with the results for corneal diameter or corneal epithelial thickness. However, the floxed Pax6 fl allele might affect different tissues in different ways. Although the Pax6 null allele only affects Pax6 quantitatively, Pax6 fl might cause both a quantitative effect and a qualitative difference in Pax6 protein or a difference in the ratio of Pax6 isoforms. Any such effects would be expected to be most pronounced in Pax6 fl/fl and Pax6 fl/ genotypes, where there was no wild-type allele, and might affect Pax6 fl/+ to a lesser extent but would not affect Pax6 +/ or Pax6 +/+. This is consistent with the smaller relative corneal diameter seen in Pax6 fl/fl and Pax6 fl/ genotypes. In principle, the results for corneal epithelial thickness could be explained by an interaction between two opposing effects. Low Pax6 levels produce a thin corneal epithelium in Pax6 +/ heterozygotes (Davis et al. 2003; Ramaesh et al. 2003) and this may also be the predominant effect in Pax6 fl/ heterozygotes. However, if Pax6 levels are higher, an altered Pax6 protein (or altered isoform ratio) might have the opposite effect in Pax6 fl/fl homozygotes but not in Pax6 fl/+ heterozygotes where wild-type protein is also present. Although such speculations, based on phenotypes, are useful for generating hypotheses, molecular investigations will be required to test these and other possibilities.

In contrast to the mildly abnormal Pax6 fl/fl and wild-type Pax6 fl/+ phenotypes reported here, a more overtly abnormal phenotype was reported for mice carrying a different floxed Pax6 allele (with the neo cassette in the intron between exons 6 and 7) and this was corrected by removing the neo cassette (cited in Ashery-Padan et al. 2000). Abnormal phenotypes have been reported for floxed alleles of other genes with retained neo cassettes (see Introduction) and some aberrant phenotypes have been attributed to altered gene expression caused by cryptic splice sites in the neo cassette (Nagy et al. 1998; Meyers et al. 1998). The retained neo cassette may also affect Pax6 expression in the floxed Pax6 fl allele, at least in the eye. K12 is regulated by Pax6 (Shiraishi et al. 1998; Liu et al. 1999) and, in humans, the Pax6(5a) isoform is known to be critical (Sasamoto et al. 2016). In principle, therefore, both the abnormal Pax6 fl/ eye morphology and the absence of K12 immunostaining in the Pax6 fl/ corneal epithelium could be mediated via direct effects on Pax6, such as reduced levels or altered ratios of different isoforms. However, we have not investigated the mechanisms that underlie the abnormal Pax6 fl/ phenotype.

Cryptic hypomorphic phenotypes have been unmasked for floxed alleles (fl) of Nodal and Smad2 with retained neo or β-geo selection cassettes, using compound heterozygotes with null alleles (Lowe et al. 2001; Liu et al. 2004). Like Pax6 fl, these were identified as abnormal alleles because fl/null compound heterozygotes had a more severe phenotype than +/null heterozygotes, even though fl/fl homozygotes appeared to be normal. The unmasking of cryptic hypomorphic phenotypes produced by floxed alleles of Nodal, Smad2 and now Pax6, using compound heterozygotes with null alleles, demonstrates that this approach is likely to be useful for identifying other floxed alleles with weak effects that are otherwise difficult to detect.

The availability of different Pax6 alleles provides a useful resource for the analysis of the roles of Pax6 in various tissues because Pax6 null alleles alone are not always sufficiently informative. The anophthalmia and early lethality of Pax6 / homozygous mean that this genotype is only useful for early developmental stages unless used in mouse chimaeras (Quinn et al. 1996) and the Pax6 +/ genotype has only minor effects in tissues other than the eye. However, by using combinations of different mouse Pax6 alleles, different phenotypes can be produced to suit the experimental requirements. Early studies showed that the phenotypes of heterozygous Pax6 +/ mice were similar to some human inherited conditions caused by PAX6 mutants, including aniridia and Peters’ anomaly (Hill et al. 1991; Ton et al. 1991; Glaser et al. 1994; Hanson et al. 1994; Nishida et al. 1995; Davis et al. 2003; Ramaesh et al. 2003). The availability of a wider array of different mouse Pax6 alleles may also provide models for some atypical human phenotypes that are caused by PAX6 missense mutations (Hanson et al. 1999; Hingorani et al. 2012).

A large allelic series of mouse Pax6 mutants has been assembled, mainly from conventional mutagenesis screens (Roberts 1967; Hogan et al. 1986; Hill et al. 1991; Lyon et al. 2000; Favor and Neuhauser-Klaus 2000; Favor et al. 2001; Thaung et al. 2002; Favor et al. 2008, 2009; Puk et al. 2013). As already noted, Pax6 / null homozygotes are anophthalmic and die around birth and heterozygotes have significant ocular defects including microphthalmia (Hill et al. 1991; Grindley et al. 1995). Four hypomorphic alleles (Pax6 coop, Pax6 4Neu, Pax6 7Neu and Pax6 132-14Neu) with less severe phenotypes have already been produced in mutagenesis experiments (Lyon et al. 2000; Favor et al. 2001, 2008). However, none of these is equivalent to the Pax6 fl allele reported here. Pax6 coop, Pax6 4Neu and Pax6 7Neu have more severe phenotypes. Homozygotes are anophthalmic and die perinatally while heterozygotes have significant overt eye abnormalities. Pax6 132-14Neu has a weaker effect than Pax6 coop, Pax6 4Neu or Pax6 7Neu but a stronger effect than Pax6 fl. Homozygous Pax6 132-14Neu/132-14Neu mice are viable and fertile but eye defects are more severe than in Pax6 +/ heterozygotes and, although some Pax6 132-14Neu/+ heterozygotes have no overt eye defects, others have a range of relatively minor abnormalities.

Two other alleles affect specific Pax6 isoforms. Mice heterozygous for a targeted deletion of the Pax6(5a) isoform develop iris hypoplasia and homozygotes have additional minor defects of the iris, retina, lens and corneal stroma but eye size is not reduced (Singh et al. 2002). Unlike the Pax6 fl/ phenotype, however, no ocular defects developed until after birth. The T(2;14)1GSo reciprocal translocation disrupts Pax6 regulation, causing elevated expression of the Pax6 P1’ isoform that results in thickening of the corneal epithelium and stroma (Elso et al. 2013). Further work is required to determine whether the floxed Pax6 fl allele shares features with T(2;14)1GSo translocation mice, which might predict the involvement of altered ratios of the Pax6 P1’ isoform.

Existing mouse hypomorphic alleles have already proved useful for analysis of Pax6 gene function in eye development. Combinations of Pax6 4Neu, Pax6 7Neu, Pax6 132-14Neu and other Pax6 alleles in various compound heterozygotes have produced a range of ocular phenotypes that have been ranked and grouped into three phenotype classes (Favor et al. 2008). Pax6 fl now provides a novel addition to the existing Pax6 allelic series that, in combination with other alleles, should be useful for refining these phenotype classifications.