The core planar cell polarity gene, Vangl2, maintains apical‐basal organisation of the corneal epithelium

Abstract

The role of the core planar cell polarity (PCP) pathway protein, Vangl2, was investigated in the corneal epithelium of the mammalian eye, a paradigm anatomical model of planar cell migration. The gene was conditionally knocked out in vivo and knocked down by siRNA, followed by immunohistochemical, behavioural and morphological analysis of corneal epithelial cells. The primary defects observed in vivo were of apical‐basal organisation of the corneal epithelium, with abnormal stratification throughout life, mislocalisation of the cell membrane protein, Scribble, to the basal side of cells, and partial loss of the epithelial basement membrane. Planar defects in migration after wounding and in the presence of an applied electric field were noted. However, knockdown of Vangl2 also retarded cell migration in individual cells that had no contact with their neighbours, which precluded a classic PCP mechanism. It is concluded that some of the planar polarity phenotypes in PCP mutants may arise from disruption of apical‐basal polarity.

Introduction

The cornea is the main refractive constituent of the anterior segment of the vertebrate eye. Its tissue organisation is relatively simple: an outer stratified epithelium representing 10–30% of the corneal thickness (depending on species) sits on a hypocellular collagenous stroma, and an inner endothelial monolayer (Li et al. 1997; Forrester et al. 2002; Henriksson et al. 2009). The structure and transparency of the cornea must be maintained throughout adult life for normal sight.

Homeostatic maintenance of the corneal epithelium requires tight control of tissue organisation and cell migration. The corneal and conjunctival epithelia are contiguous, and stem cells that maintain the corneal epithelium are located at the boundary between the two – the limbus – around the corneal periphery (Cotsarelis et al. 1989). These limbal epithelial stem cells generate transit amplifying cells that migrate centripetally along the basal layer of the corneal epithelium (Collinson et al. 2002; Nagasaki & Zhao, 2003; Di Girolamo et al. 2015). Proliferative basal corneal epithelial cells may divide several times before losing contact with the basement membrane and differentiating as suprabasal epithelial ‘wing cells’ (Lehrer et al. 1998). Superficial squamous cells are continuously lost from the corneal epithelial surface, for example, by abrasion. The balance between cell migration into the corneal epithelium, basal cell division and apical cell loss from the corneal surface must be maintained such that the multilayered structure of the corneal epithelium is preserved (Thoft & Friend, 1983; reviewed in Mort et al. 2012).

The corneal epithelium exhibits a rigid apical‐basal polarity. Cell proliferation is restricted to the basal layer of the corneal epithelium, and apoptosis is normally restricted to desquamating superficial cells (reviewed in Mort et al. 2012). Tight junctions between cells prevent free diffusion across the epithelium and contribute to its protective function (Yi et al. 2000). In addition, desmosomes, gap junctions and immunoglobulin‐class cell adhesion molecules connect the cells of the adult corneal epithelium. Hemidesmosomes anchor the basal cells to the underlying basal lamina (Buck, 1982; Forrester et al. 2002; Smith et al. 2001). The corneal epithelium originates as a monolayer during development and in mice stratification occurs neonatally concomitant with eye opening, controlled at least in part by Wnt/β‐catenin signalling inhibiting action of bone morphogenetic protein 4 (Zhang et al. 2015a).

Studies investigating the control of apical‐basal polarity have highlighted three complexes whose subcellular localisation directs the internal polarity of epithelial cells in vertebrates or invertebrates: Crumbs and Stardust apically; the Par6/Bazooka/αPKC complex towards the apical side of the lateral edge of the cell; and Lethal Giant Larvae, Discs Large and Scribble homologues laterally (Xiao et al. 2011; Kumichel & Knust, 2014; Rodriguez‐Boulan & Macara, 2014; Whiteman et al. 2014).

Planar polarity, i.e. the polarity that epithelial cells exhibit in the plane of their basement membrane, is directed during embryogenesis by a non‐canonical branch of the Wnt signalling pathway – the planar cell polarity (PCP) pathway. In invertebrate systems at least this derives from interaction between transmembrane proteins frizzled and van gogh/strabismus, asymmetrically localised to opposite edges of cells, conferring patterns of epithelial cell directionality in the plane of their basement membrane (Taylor et al. 1998; Devenport, 2014). Vertebrates have multiple PCP gene homologues, and PCP pathways have multiple functions during embryogenesis (Lu et al. 2004; Gao et al. 2011; Mao et al. 2011; Wansleeben & Meijlink, 2011; Wallingford, 2012). Although extrapolation between vertebrate and invertebrate systems is not straightforward, mutation of the van gogh homologue, Vangl2, in vertebrate embryos ablates all PCP signalling, and produces heart, inner ear and neural tube defects consistent with failure of cell directionality and migration (Greene et al. 1998; Henderson et al. 2001; Kibar et al. 2001; Murdoch et al. 2001; Park & Moon, 2001; Goto & Keller, 2002; Montcouquiol et al. 2003; Phillips et al. 2005; Roszko et al. 2009; Lei et al. 2010; Yates et al. 2010; Ramsbottom et al. 2014).

There is evidence of interaction between apical‐basal and planar polarity pathway components. Apical localisation of PCP proteins is critical for their function (Das et al. 2004), and protein kinase C alpha (αPKC), part of the Par6/Bazooka/αPKC complex that characterises the apical‐lateral side of the cell, can inhibit Frizzled‐PCP signalling in Drosophila by phosphorylating Frizzled and thereby stopping it from signalling (Djiane et al. 2005). Par3 (Bazooka) and Scribble have been linked to defects in asymmetric cell division and cell polarity defects (Lin et al. 2000). Scribble protein is key to potential interactions between apical/basal and planar polarity. Within Drosophila, Scribble binds Van Gogh protein (also known as Strabismus) via its PDZ domain 3 and cooperates in establishing PCP (Courbard et al. 2009). In addition, heterozygous scribble mutations were found to exacerbate PCP defects of van gogh (vang) mutants, whereas Discs Large and Lethal Giant Larvae, the other proteins located with Scribble along the baso‐lateral edge of cells, did not, indicating a specific role for Scribble in Drosophila PCP (Courbard et al. 2009). The scribble ⁵ mutation, encoding a protein truncated before the 3rd PDZ domain, exhibited PCP defects in wing cell alignment, but not apical‐basal polarity defects (Courbard et al. 2009).

In vertebrates the 3rd and 4th PDZ domains of Scribble bind the C‐terminus of Vangl2, and this interaction is required for asymmetric localisation of Vangl2 in cochlear hair cells (Montcouquiol et al. 2006). Vangl2 ^Lp/+ Scribble ^Crc/+ double‐mutants show cochlea PCP defects akin to those observed in Vangl2 ^Lp/Lp mutants, indicating that Scribble and Vangl2 proteins work in the same genetic pathway during PCP cell alignment (Montcouquiol et al. 2003). Hence, Scribble, although having a canonical role in establishment of apical/basal polarity, also has a role in the establishment of PCP in vertebrates, regulating cell cohesion junctional complex maintenance (Yates et al. 2013).

Little work has been done on PCP gene function in adult vertebrate tissues. As part of a study investigating the control of patterning in the corneal epithelium, we conditionally knocked out the core PCP gene Vangl2 in the corneal epithelium, and were able to show classic planar polarity defects – Vangl2 acted through PCP intermediates Dishevelled, DAAM1 and ROCK1/2, modulating corneal epithelial cell cytoskeletal rearrangement, cell alignment and migration in vivo and in vitro; (Findlay et al. 2016). Here we describe also a severe disruption of apical‐basal polarity with failure of basal cells to maintain a normal basement membrane. The data show PCP genes interact with apical‐basal pathways in the adult vertebrate system and that Vangl2 activity is required for normal stratification of the corneal epithelium.

Materials and methods

Mice

Hemizygous Tg(Pax6‐cre, GFP)1Pgr transgenic mice (henceforth ‘Le‐Cre ^Tg/−') drive Cre recombinase expression in the lens and corneal epithelia (Williams et al. 1998; Ashery‐Padan et al. 2000). Le‐Cre ^Tg/−, Vangl2 ^Lp/+ and Vangl2 ^fl/fl floxed mice (Ramsbottom et al. 2014) were maintained on a congenic CBA/Ca genetic background. All animal procedures were carried out according to Animals (Scientific Procedures) Act 1986 and were passed by University of Aberdeen Ethical Review Board.

Due to lethality of the Vangl2 ^Lp/Lp homozygous mutants (Strong & Hollander, 1949), Cre‐loxP technology (Gu et al. 1994) was used to delete Vangl2 conditionally in the corneal epithelium of Le‐Cre ^Tg/−; Vangl2 ^fl/fl animals. Vangl2 ^fl/fl animals were mated with Le‐Cre ^Tg/− mice and their genotypes were confirmed by polymerase chain reaction (PCR) using primers and conditions described in Findlay et al. (2016). The Le‐Cre transgene is active from E8.75 in the lens placode, and is expressed continuously throughout the lens and corneal epithelium (Ashery‐Padan et al. 2000). Le‐Cre ^Tg/−; Vangl2 ^fl/+ from F1 were backcrossed with Vangl2 ^fl/fl mice to obtain Le‐Cre ^Tg/−; Vangl2 ^fl/fl mice and Le‐Cre ^Tg/−; Vangl2 ^fl/+ littermate controls. Le‐Cre ^−/− Vangl2 ^fl/fl animals were normal and exhibited no looptail defects, indicating the floxed allele was neutral and the Le‐Cre transgene was not showing leaky expression in the germline. Mice were killed by cervical dislocation, their eyes enucleated and fixed for processing.

Cell culture

In vitro experiments were carried out by using an immortalised human corneal epithelium cell line (HCE‐S) donated by Julie Daniels, Institute of Ophthalmology (Notara & Daniels, 2010). Cells were maintained in Dulbecco's modified Eagle's medium (DMEM)/F‐12 media (Life Technologies), 10% fetal calf serum (Invitrogen), 1% penicillin streptomycin solution (10 000 units penicillin and 10 mg streptomycin mL⁻¹; Sigma) at 37 °C in a humidified atmosphere of 5% CO₂. Culture medium was replenished every 72 h.

A robust knockdown of VANGL2 to ~30% of normal levels was obtained in transfected HCE cells using 10 nm high purity Vangl2_5 siRNA 5′‐UAGAAUUAGGAAGUACCCAUA‐3′ as described in Findlay et al. (2016). In brief, 60 000 cells were seeded in each well of a 24‐well plate in 0.1 mL of culture media. Cells were incubated under their normal growth condition, while 75 ng siRNA was diluted down in 100 μL of serum‐free culture medium to a final concentration of 10 nm. The diluted siRNA was mixed with 3 μL of HiPerFect transfection reagent and incubated for 10 min at room temperature before adding drop‐wise to HCE cells. After 3 h incubation of the transfection complexes, 400 μL of culture medium was added in each well. Cells were passaged 24 h later for experiments.

Cellular migration studies upon application of electric fields

Following gene silencing with siRNAs, HCE cells were exposed to physiological electric field stimulation to measure forward migration index (FMIX) and directionality of electrotactic movement. Control and knockdown HCE cells were harvested from 24‐well plates by addition of trypsin‐EDTA, spun and re‐suspended in 500 μL of culture media. Cells were seeded into ‘ibidi’ 15 μm chamber‐slides (Thistle Scientific) and time‐lapse video recording performed, using a Nikon Diaphot inverted phase contrast microscope with a temperature‐controlled environment chamber in a direct current 200 mV mm⁻¹ electric field by adopting a set‐up previously described by Rajnicek et al. (2006).

A measure of planar‐oriented migration, the FMIX, was measured by using the ‘Manual Tracking’ and ‘Chemotaxis’ plugins (available at http://rsbweb.nih.gov/ij/plugins/index.html) for ImageJ. FMIX is one of the most indicative measures of migration and is determined by measuring, for each cell, the movement along the x‐axis (x) as a proportion of the total migration distance (d), evaluating the magnitude of the directional cell movement towards the cathode. Conventionally, the cathode is to the left, so the closer the value of FMIX to −1 the more directed is migration to the cathode on average. FMIX was measured separately for individual cells in culture and cells as part of confluent sheets.

Histology

Eyes were enucleated from adult animals killed by a Schedule 1 procedure and fixed in paraformaldehyde (4% in phosphate‐buffered solution, PBS) for 4–6 h at 4 °C prior to processing for paraffin embedding.

Eyes were washed three times in PBS, 20 min, then for 15 min in saline solution (0.9% NaCl). Eyes were then dehydrated through a series of 15 min ethanol changes (70, 85, 95 and 100%) before being cleared with xylene (2 × 5 min washes at room temperature and incubation overnight in fresh xylene) and embedding in paraffin wax. Sections (7 μm thick) were cut in the transverse plane.

Haematoxylin and eosin (H&E) staining

Wax sections were deparaffinised in Histoclear (HS‐200, National Diagnostics), 2 × 10 min and washed 2 × 5 min in 100% ethanol. Rehydration in a serial change of ethanol washes for 5 min each (95, 85 and 70% ethanol) followed. Slides were then washed with PBS, incubated in Gill's haematoxylin solution for 1 min and washed in tap water. Slides were washed in Scott's tap water (20 g L⁻¹ MgSO₄.7H₂O, 20 g L⁻¹ NaHCO₃), rinsed with tap water and incubated for 30 s in eosin solution (1% eosin in 50% ethanol–5 mm acetic acid). Slides were then dehydrated in ethanol series, cleared in xylene and mounted with di‐n‐butylphthalate in xylene. Imaging was performed using a Nikon E400 Eclipse light microscope in bright field.

Immunohistochemistry

Detection of PCP proteins in the corneal tissue and HCE cells was carried out by immunohistochemistry using material and reagents as described in Findlay et al. (2016). Primary antibodies were: bromodeoxyuridine (BrdU), ab181664 (Abcam); Scribble sc‐28737 (Santa Cruz), phosphatidylinositol‐4,5‐bisphosphate 3‐kinase (PI3K), #4252 (Cell Signalling Technology); laminin, ab11575 (abcam); PAX6 (Developmental Studies Hybridoma Bank), mouse monoclonal ‘4A4’ recognising ΔN‐P63 (Santa Cruz); cytokeratin‐12 (Santa Cruz). Secondary antibodies were: (all Molecular Probes) Alexa 488‐conjugated goat anti‐mouse IgG1 A21121; Alexa 568‐conjugated donkey anti‐mouse IgG A10037; Alexa 488‐conjugated rabbit anti‐goat IgG A110178; Alexa 594‐conjugated donkey anti‐rabbit IgG A21207. Confocal LSM700‐Zeiss Imager M2 Upright and Nikon 400 Eclipse Microscopes were used to image fluorescent sections.

Immunofluorescence on whole‐mount corneas

Corneas were dissected from fixed eyeballs of adult animals. Dissected corneas were washed in PBS, 3 × 10 min, permeabilised in methanol for 20 min at −20 °C and washed 3 × 10 min with PBS. Corneas were incubated with 1% pepsin in 10 mm HCl for 15 min at 37 °C, neutralised in 0.1 m sodium borate buffer, pH 8, for 10 min and washed 2 × 10 min with PBS. Antigen retrieval was achieved by treating the corneas with 1% sodium dodecyl sulphate in PBS for 5 min and washing for 3 × 5 min with PBS. Primary and secondary antibodies incubations followed as described previously, and then corneas were flattened by four scalpel incisions and mounted in fluorescence mounting medium with the corneal epithelium facing upwards. Confocal LSM700‐Zeiss Imager M2 Upright and Nikon 400 Eclipse Microscopes were used for imaging.

Apoptosis assay

The terminal deoxynucleotidyl transferase dUTP nick‐end labelling (TUNEL) In Situ Cell Death Detection kit‐(fluorescein) (Roche, UK) was used to label apoptotic cells in the corneal epithelium of tissue sections. Deparaffinised sections were incubated with proteinase K (80 μg mL⁻¹) for 5 min at room temperature and then washed in PBS. A positive control slide was treated with DNAase1 (50 units mL⁻¹ DNAase 1 in 50 mm Tris–HCl pH 7.5, 1 mm MgCl₂, 1 mg mL⁻¹ bovine serum albumen) for 30 min at 37 °C. TUNEL labelling was performed by incubating slides with coverslips with the TUNEL reaction mixture according to manufacturer's instructions for 60 min at 37 °C in the dark. A negative control was included in each experimental set‐up by incubating fixed and permeabilised tissue in 50 μL Label solution without enzyme. Slides were then washed for 3 × 5 min in PBS, mounted in Vectashield (Vector Laboratories, UK) and viewed under a Nikon Eclipse E400 fluorescence microscope.

Morphometric measurements of the cornea

Images of sagittal adult eye sections of Le‐Cre ^Tg/−; Vangl2 ^fl/fl and Le‐Cre ^Tg/−; Vangl2 ^fl/+ animals following H&E staining were captured by a digital camera (Qimaging, QICAM Fast1394) at 400 × magnification, and morphometric measurements of the corneal epithelium thickness were made with ImageJ (available online at http://imagej.nih.gov/ij/). The method of Ramaesh et al. (2003) was used to measure the thickness of the corneal epithelium and the whole cornea. In brief, measurements were made in three different areas (two peripheral and one central) in the five central serial sections of each adult eye. Mean thicknesses were calculated for the peripheral regions of each section and used to calculate the mean thickness in each eye.

Corneal diameters were measured using a stereomicroscope. An image of the eye was taken alongside a calibrated ruler and measurements of the corneal diameter were made.

Transmission electron microscopy (TEM)

For TEM, eyes were collected and fixed in 2% gluteraldehyde for 4 h. The corneas were dissected and the tissue post‐fixed in osmium, dehydrated through increasing concentrations of ethanol and propylene oxide, embedded in plastic and semi‐thin sections cut tetroxide at the Histology Facility (Institute of Medical Sciences, University of Aberdeen, UK). The sections were mounted on copper grids, stained with lead citrate and uranyl acetate before imaging on a JEOL 1400 plus TEM.

Cell proliferation study

Le‐Cre ^Tg/−; Vangl2 ^fl/fl and Le‐Cre ^Tg/−; Vangl2 ^fl/+ adult littermates were given a single intraperitoneal injection of 10 mg mL⁻¹ BrdU in sterile PBS, and killed 2 h later by cervical dislocation. Eyes were enucleated and processed for immunohistochemistry as above.

Statistical analysis

For normally distributed data, a two‐tailed unpaired t‐test was used in most cases to determine statistical significance when comparing results obtained from Le‐Cre ^Tg/−; Vangl2 ^fl/fl and Le‐Cre ^Tg/−; Vangl2 ^fl/+ littermates or V2_KD and NT control cells. Mann–Whitney U‐test was used for non‐parametric data to compare between genotypes. One‐way analysis of variance (anova) was used when three or more groups of data were compared, with post hoc Tukey HSD test to identify the significant difference between pairs of groups.

Results

Ablation of Vangl2 results in abnormal stratification of the corneal epithelium

Multiple PCP genes including Vangl2 have been shown by reverse transcriptase (RT)‐PCR, immunohistochemistry and Western blot to be expressed in the murine corneal epithelium (Findlay et al. 2016). The eyes of adult ‘loop tail’ mice that are heterozygous for an inactivating mutation in Vangl2 (Vangl2 ^Lp/+) were grossly normal, but tissue sectioning revealed mild corneal defects, with disruption of epithelial stratification and irregularities of basal cell nuclei seen in all corneas that were not observed in their wild‐type littermates (Fig. 1A,B; n = 8). Vangl2 ^Lp is a semi‐dominant allele, and Vangl2 ^Lp/Lp mice die during embryogenesis or shortly after birth with severe neural tube defects (Strong & Hollander, 1949; Yin et al. 2012; Chen et al. 2013). In order to study the role of Vangl2 in the adult corneal epithelium, a conditional knockout was made by breeding Vangl2‐floxed mice (Vangl2 ^fl/fl) to Le‐Cre ^Tg/− mice expressing Cre in the lens and corneal epithelia (Ashery‐Padan et al. 2000; Ramsbottom et al. 2014). The progeny of these matings were backcrossed onto the Vangl2 ^fl/fl line to generate Le‐Cre ^Tg/− Vangl2 ^fl/+ and Le‐Cre ^Tg/− Vangl2 ^fl/fl littermates (as well as Le‐Cre ^−/− Vangl2 ^fl/+ and Le‐Cre ^−/− Vangl2 ^fl/fl control mice). H&E staining of adult Le‐Cre ^Tg/− Vangl2 ^fl/fl eyes revealed a highly disrupted stratification of the corneal epithelium that was not observed in Le‐Cre ^Tg/− Vangl2 ^fl/+ controls (n > 24; Fig. 1C–E). Sections revealed patches of both abnormally thin (Fig. 1D) and abnormally thick corneal epithelium (Fig. 1E). Thin regions of the epithelium were composed of as little as one or two cellular layers with no consistent cellular morphology, and were sharply juxtaposed with epithelium of normal thickness (5–7 cells) or hypertrophic epithelium up to 10 cells thick. In Le‐Cre ^Tg/− Vangl2 ^fl/+ controls (n = 24) and all Le‐Cre ^−/− corneas (n = 40) the epithelium was uniform and smooth, 5–7 cells thick.

Figure 1.

Corneal epithelial abnormalities in adult Vangl2‐mutant mice. Haematoxylin and eosin (H&E) staining of tissue sections of adult mice corneas. (A) Wild‐type cornea showing stratified epithelium sitting on hypocellular collagenous corneal stroma. (B) Cornea from Vangl2 ^Lp/+ mouse (littermate of A) with mild disruption to apical‐basal organisation of epithelium (arrowhead). (C–E) Cornea from Le‐Cre ^Tg/− Vangl2 ^fl/fl mouse (conditional knockout, cKO) showing irregularity of corneal epithelium, disruption to normal stratification, and projection of basal cells into the corneal stroma. Corneal epithelia could be abnormally thin (D) or thick (E) with sharp transitions between the thinner and thicker domains, phenomena never seen in Le‐Cre ^Tg/− Vangl2 ^fl/+ or Le‐Cre ^−/− Vangl2 ^fl/fl (n > 100) controls. Scale bar: 50 μm (A–C); 20 μm (D, E).

Stratification of the corneal epithelium is a postnatal event (Zieske, 2004). While the corneal epithelia of both Vangl2‐null and control mice were monolayers during embryogenesis, and the epithelia were morphologically normal to birth, examination of corneas from neonatal mice showed that stratification of the epithelium was delayed in the mutants. At postnatal day 5, in contrast to Le‐Cre ^−/− Vangl2 ^fl/+, Le‐Cre ^−/− Vangl2 ^fl/fl and Le‐Cre ^Tg/− Vangl2 ^fl/+ control mice whose epithelium was typically 2–3 cells thick, the epithelium of Le‐Cre ^Tg/− Vangl2 ^fl/fl mice was a monolayer, and in some areas only a thin cytoplasmic covering was visible (n = 9; Fig. 2).

Figure 2.

Corneal epithelial abnormalities in neonatal Vangl2‐mutant mice. (A) At postnatal day 5, the corneal epithelium of Le‐Cre ^Tg/− Vangl2 ^fl/+ control mice is a uniform 2–3 cells thick (n = 6). (B) In contrast, all corneal epithelia of Le‐Cre ^Tg/− Vangl2 ^fl/+ littermates (n = 5) exhibited irregularities of stratification, with the epithelium being 1–2 cells thick and with areas of cytoplasmic covering only. Scale bar: 20 μm.

Vangl2‐null corneas exhibit reduced or partially absent corneal epithelium basement membrane

Histological analysis (Fig. 1) suggested the normal basal/wing/squamous cell apical‐basal arrangement of the epithelium was disrupted in Le‐Cre ^Tg/− Vangl2 ^fl/fl corneas, compared with controls. The basement membrane of mutant corneal epithelia was disrupted and sometimes undetectable in mutant corneal epithelia – this was apparent by H&E staining (n > 25), and immunohistochemistry using antibodies against extracellular matrix (ECM) proteins laminin and collagen IV. Partial or total absence of collagen IV and laminin was reproducibly observed in Le‐Cre ^Tg/− Vangl2 ^fl/fl corneal epithelial (n = 6) but not in the controls (n = 9; Figs 3 and S1). A preliminary TEM analysis of Le‐Cre ^Tg/− Vangl2 ^fl/fl corneas (n = 2) was performed alongside control littermates, Le‐Cre ^Tg/− Vangl2 ^fl/+ (n = 2) and Le‐Cre ^−/− Vangl2 ^fl/fl (n = 2), which confirmed disruption of the basement membrane. Control Le‐Cre ^Tg/− Vangl2 ^fl/+ and Le‐Cre ^−/− Vangl2 ^fl/fl eyes both exhibited a distinct basement membrane underlying the corneal epithelium and also a clear morphological difference in cell shape and cytoplasm density between the polarised basal epithelial cells and the overlying apical cells (Fig. S2). Strikingly the basement membrane was also much thinner in the Le‐Cre ^Tg/− Vangl2 ^fl/fl mutant corneas and partially absent. Cells in the mutant corneas were arranged in a disorganised manner, there was no clear morphological differentiation between basal and apical cells, and the basal cells showed the same cytoplasmic density as more superficial cells (Fig. S2).

Figure 3.

Corneal epithelial basement membrane deficiency in adult Vangl2‐mutant mice. Haematoxylin and eosin (H&E) staining (top), and immunohistochemistry for extracellular matrix (ECM) proteins laminin (middle) and collagen IV (bottom) in adult Le‐Cre ^Tg/− Vangl2 ^fl/+ control mice (left) and Le‐Cre ^Tg/− Vangl2 ^fl/+ littermates (right). Arrows point to the basement membrane. In controls, the basement membrane is visible in H&E‐stained controls as a strongly stained lamina immediately underneath the basal surface of the epithelial cells, sitting on a more weakly stained ECM. This is reproducibly not visible or patchy in mutants (right). Laminin and collagen IV staining in mutants is absent or patchy and thin, confirming the light microscopy and supplementary transmission electron microscopy (TEM). DAPI staining (blue) visualises cell nuclei. Scale bar: 40 μm.

Morphometric analysis of mutant corneas

A morphometric analysis was performed on Le‐Cre ^Tg/− Vangl2 ^fl/fl corneas and controls. The mean diameter of Le‐Cre ^Tg/− Vangl2 ^fl/fl, Le‐Cre ^Tg/− Vangl2 ^fl/+ and Cre ^−/− corneas was measured. Consistent with previous studies showing that Cre expression is not always neutral to phenotype, it was found that Le‐Cre ^Tg/− eyes are smaller than Cre‐negative eyes, irrespective of Vangl2‐genotype (Adams & van der Weyden, 2001; Dorà et al. 2014; Fig. 4A). This is presumed to be either a toxic effect of Cre or, more likely, a negative effect of reduced Pax6 availability due to Pax6 binding sites in the Le‐Cre promoter. For this reason, only Le‐Cre ^Tg/− Vangl2 ^fl/+ mice were used as controls for morphometric analysis.

Figure 4.

Morphometric analysis of Vangl2‐mutant corneas. Corneal epithelial diameter (top) and thickness (bottom) were measured. Le‐Cre‐positive corneas were slightly but very significantly smaller than Le‐Cre‐negative corneas, irrespective of Vangl2 genotype. Conditional knockout of Vangl2 lead to increased mean thickness of the corneal epithelium, both centrally and peripherally, accompanying the cellular, morphological and molecular defects described in this paper. * represents P < 0.05. **** represents P < 0.0001.

Transverse medial sections from serially sectioned eyes were stained with H&E, and the thickness of the corneal epithelium was measured in central and peripheral regions as described in the Materials and methods On average, the epithelium of Le‐Cre ^Tg/−; Vangl2 ^fl/fl corneas was significantly thicker in both central and peripheral regions when compared with Le‐Cre ^Tg/−; Vangl2 ^fl/+ controls (central corneal epithelium thickness ± SEM: Le‐Cre ^Tg/−; Vangl2 ^fl/+ 25.85 ± 2.15 μm, Le‐Cre ^Tg/−; Vangl2 ^fl/fl 41.61 ± 6.76 μm; t‐test: n = 10; P = 0.0396; peripheral corneal epithelium thickness ± SEM: Le‐Cre ^Tg/−; Vangl2 ^fl/+ 25.55 ± 1.43 μm, Le‐Cre ^Tg/−; Vangl2 ^fl/fl 37.62 ± 4.72 μm; t‐test: n = 10; P = 0.0157; Fig. 4B).

On the basis of these data, it was suggested that one of the primary defects in corneal epithelia that are null for Vangl2, and which should therefore lack all PCP signalling, may in fact be a failure of normal apical‐basal epithelial polarity. Further molecular analysis was performed on mutant corneas and controls to investigate this.

Cell proliferation and cell death in the corneal epithelia of Vangl2‐deficient mice

In normal corneal epithelia, cell proliferation is restricted to the basal cell layer, and apoptotic events are normally only detected in superficial apical cells prior to desquamation. It was considered possible that the defects observed above may result from loss of apical and basal identity of cells in which case patterns of proliferation and apoptosis may be disrupted. TUNEL labelling was therefore performed to label apoptotic cells in the corneal epithelium of Le‐Cre ^Tg/−; Vangl2 ^fl/fl mice and Le‐Cre ^Tg/−; Vangl2 ^fl/+ control animals. Apoptotic nuclei were restricted to cells in the superficial layer of both control and Le‐Cre ^Tg/−; Vangl2 ^fl/fl corneas. No TUNEL‐positive cells were detected in the basal or wing layers of the corneal epithelium in either Vangl2‐deficient or control mice (Fig. 5), which reflects normal shedding from the epithelial surface (Ren & Wilson, 1996).

Figure 5.

Molecular analysis of cell proliferation and apoptosis in Vangl2‐mutant corneal epithelia. (A–C) Terminal deoxynucleotidyl transferase dUTP nick‐end labeling (TUNEL) analysis of apoptotic cell death. In both Le‐Cre ^Tg/− Vangl2 ^fl/+ control corneas (A) and Le‐Cre ^Tg/− Vangl2 ^fl/fl conditional knockouts (B) apoptosis (green fluorescent labelling) was restricted to the most superficial cell layer, with no apoptotic events noted in any other layer of the corneal epithelium (n = 3 of both genotypes). (C) A positive control – tissue section digested with DNAse1 to create double‐stranded breaks in DNA of all cells. (D, E) Immunohistochemical analysis of cell proliferation in corneal epithelium of mice after a single injection of BrdU. In both Le‐Cre ^Tg/− Vangl2 ^fl/+ control corneas (D) and Le‐Cre ^Tg/− Vangl2 ^fl/fl conditional knockouts (E), DNA replication was restricted to the basal layer of the corneal epithelium, with no proliferation events more apically.

To examine whether disrupted stratification of the corneal epithelium in Vangl2‐deficient corneal epithelia was linked to defects in epithelial cell proliferation, BrdU staining was performed in order to observe proliferating cells within the corneal epithelium. Mice were treated with a single intraperitoneal injection of BrdU, to label cells in S‐phase, for 2 h before the mice were killed and eyes taken for BrdU immunohistochemistry. In both Le‐Cre ^Tg/− Vangl2 ^fl/fl mice eyes, Le‐Cre ^Tg/− Vangl2 ^fl/+ and all Le‐Cre ^−/− controls (n = 6–12 of each genotype), BrdU incorporation was restricted to cells in the basal layer of the epithelium only (Fig. 5D,E). No suprabasal DNA synthesis was observed within mutant mice and therefore no evidence of loss in apical‐basal controlled mitotic division.

These data suggest that although the morphological differentiation between apical and basal epithelial cells was often lost in Le‐Cre ^Tg/− Vangl2 ^fl/fl corneas, cell identity (basal or apical) was maintained, and hence proliferation continued to be restricted to the basal layer and apoptosis to the superficial cell layers.

Localisation of apical‐basal and cell fate markers in the control and mutant corneal epithelium

The basal lamina is secreted by corneal epithelial cells, so the disruption of the basement membrane in Vangl2‐null corneas could result from defective subcellular apical‐basal organisation. The membrane‐associated scaffold protein Scribble has been previously described as an important apical‐basal organiser within Drosophila cells through its role, along with the other proteins in its complex, Discs large and Lethal Giant Larvae, in the subcellular localisation of proteins (Bilder & Perrimon, 2000; Harris & Lim, 2001). Scribble has previously been shown to be expressed within the corneal epithelium of mice (Nguyen et al. 2005). In order to further study apical‐basal polarity in the conditional Vangl2‐knockout corneas, Scribble immunostaining was performed on Le‐Cre ^Tg/− Vangl2 ^fl/fl eyes and compared with Le‐Cre ^Tg/− Vangl2 ^fl/+ littermates. Scribble was found to be localised primarily to the lateral and apical boundaries of the basal epithelial cells of control Le‐Cre ^Tg/− Vangl2 ^fl/+ mice, and was generally absent from the basal plasma membrane of these cells (Fig. 6; n = 5 corneas). In Le‐Cre ^Tg/− Vangl2 ^fl/fl eyes Scribble localisation was moderately but consistently disrupted, with localisation bleeding into the basal boundary of many cells (n = 5 corneas; Fig. 6B). This suggests disruption of apical‐basal polarity in those cells, associated with partial or total loss of the basement membrane described above. PI3K is an important enzyme for corneal epithelial cellular motility and intracellular trafficking (Zhao et al. 2006). It is activated by growth factors acting upon cells and initiating signalling through downstream effectors such as protein kinase B (Katso et al. 2001). Xu et al. (2010) grew mammary epithelial cells upon a laminin‐rich ECM; cells grown on the ECM became polarised and displayed a localisation of PI3K to the basal side. Cells grown in the absence of laminin remained unpolarised with no localisation of PI3K in the cell. Immunohistochemistry was performed to investigate PI3K localisation in control and mutant corneal epithelia and to determine whether loss of the basement membrane affected distribution of the protein. When Le‐Cre ^Tg/− Vangl2 ^fl/fl mice (n = 3) were compared with Le‐Cre ^Tg/− Vangl2 ^fl/+ littermates (n = 5), fluorescence was highest in the cytoplasm or cell membranes of basal cells of Le‐Cre ^Tg/− Vangl2 ^fl/+ mice eyes as expected; however, Le‐Cre ^Tg/− Vangl2 ^fl/fl eyes exhibited only low protein levels in basal cells, similar to that observed in more apical cells (Fig. 6).

Figure 6.

Cell polarity and cell fate markers in corneal epithelia of Vangl2‐mutants. Immunohistochemical labelling of Scribble, phosphatidylinositol‐4,5‐bisphosphate 3‐kinase (PI3K), P63, cytokeratin‐12 and Pax6 in Le‐Cre ^Tg/− Vangl2 ^fl/+ control corneas and Le‐Cre ^Tg/− Vangl2 ^fl/fl conditional knockouts. Asterisks denote selected Vangl2‐mutant cells with atypical localisation of Scribble to the basal boundary, suggesting disrupted apical‐basal polarity of the basal epithelial cells. PI3K is upregulated in the basal cells of control epithelia but not of Vangl2 knockout cells. Other markers of cell fate and apical‐basal identity are expressed normally in mutant epithelia. n = 3–9 for all markers. Scale bar: 40 μm.

Further markers of cell fate and apical‐basal identity were assayed in mutant corneas. Immunohistochemical analysis of localisation of the ΔN isoform of the tumour suppressor protein P63, previously shown to be a marker of basal corneal epithelial cells in the mouse (Collinson et al. 2002; Ramaesh et al. 2005), revealed no difference between control Le‐Cre ^Tg/− Vangl2 ^fl/+ (n = 4) and mutant Le‐Cre ^Tg/− Vangl2 ^fl/fl eyes (n = 3), with basal cells of both genotype exhibiting strong staining and apical cells very weak or no staining. The definitive corneal epithelial marker, cytokeratin‐12, was localised to cytoplasm of all cells in all corneal epithelia, irrespective of genotype (n = 4 of each), and the ocular surface marker, Pax6, was localised to nuclei of all cells (n = 4 of each genotype; Fig. 6). Loss of either of these markers would suggest failure of corneal epithelial identity.

These data, together with the defects to the basement membrane described above, suggested that there was no loss of corneal epithelial identity in the Le‐Cre ^Tg/− Vangl2 ^fl/fl mutants, but that the basal epithelial cells had partially or totally lost their apical‐basal polarity.

Loss of Vangl2 causes cells to exhibit migration defects

Wound healing efficiency in embryonic skin is regulated by a PCP pathway (Caddy et al. 2010). We previously showed that corneal epithelial cells exhibited slower rates of scratch‐wound healing when Vangl2 was inactivated by a tamoxifen‐inducible Cre (Findlay et al. 2016). To determine whether the Vangl2‐deficient model employed in this study showed a similar defect, monolayers of Le‐Cre ^Tg/− Vangl2 ^fl/fl, Le‐Cre ^Tg/− Vangl2 ^fl/+ and Le‐Cre ^−/− Vangl2 ^fl/fl corneal epithelial cells were cultured as previously described (Leiper et al. 2006) and scratch‐wound assays were performed. Vangl2–null Le‐Cre ^Tg/− Vangl2 ^fl/fl cells healed significantly more slowly (26.2 ± 3.5 μm h⁻¹ – mean ± SEM; n = 7) than Le‐Cre ^Tg/− Vangl2 ^fl/+ controls (39.4 ± 3.7 μm h⁻¹; n = 9; Fig. 7A). These data confirmed a role for Vangl2 in the migration during wound healing of corneal epithelial cells.

Figure 7.

Planar migration defects in Vangl2‐deficient corneal epithelial cells. (A) Wound healing rate of in vitro cultured adult mouse corneal epithelial cells. The rate of migration of Le‐Cre ^Tg/− Vangl2 ^fl/fl cells following scratch‐wounding was significantly slower than that of Le‐Cre ^Tg/− Vangl2 ^fl/+ and Le‐Cre ^−/− Vangl2 ^fl/fl cells (one‐way anova: F = 3.720, P = 0.0309). (B) Cathodal migration [expressed as forward migration index (FMIX), as described in Materials and methods], for human corneal epithelial cells after VANGL2 knockdown to ~30% of normal protein levels. The response of cells cultured at low density (individual cells; light grey) was not significantly different from that of confluent cells (dark grey). * represents P < 0.05

It was considered possible in light of data presented above that the migration defect shown by Vangl2‐mutant corneal epithelial cells arises not through a classic PCP pathway but as a secondary consequence of apical‐basal defects through a non‐PCP mechanism. This was tested by knocking down VANGL2 by siRNA in human corneal epithelial cells in vitro and comparing planar migration of individual isolated cells with that of confluent cells after exposure to a physiological electric field of 200 mV mm⁻¹ that produces robust planar migration towards the cathode (Soong et al. 1990; Zhao et al. 1996; Farbourd et al. 2000; Findlay et al. 2016). Classic PCP activity requires cell–cell contact, so any effect of VANGL2 knockdown on isolated cells in vitro must represent a non‐PCP mechanism of action. Human corneal epithelial cells were transfected with a validated siRNA that causes knockdown of VANGL2 to about 30% of normal levels and which we showed previously causes a significant reduction in the ability of cells to migrate cathodally in an applied electric field (Findlay et al. 2016). Control cells were transfected with a negative control nonsense siRNA. Planar migration in an applied electric field was quantified as the FMIX representing the mean cosine of the angle θ of each cell's direction of travel towards the cathode, with FMIX = −1 representing perfect cathodal migration and FMIX = 0 representing mean random migration (see Materials and methods). It was found that there was no significant difference between the behaviour of isolated individual cells compared with that of cells in confluent sheets (Fig. 7B). The fact that planar migration was retarded in VANGL2‐knockdown cells even in the absence of cell–cell contact shows that a classic PCP mechanism was not involved. It suggests that VANGL2 has a non‐PCP function, associated with apical‐basal polarity, which leads to migration defects in the absence of normal VANGL2 dosage.

Discussion

Disrupted stratification of the Vangl2‐deficient corneal epithelium

In this study, the core PCP gene, Vangl2, was genetically disrupted in the corneal epithelium and a previously unsuspected role in apical‐basal patterning was identified. The stratification of the epithelium was delayed in neonates and badly disrupted in adult mice, with a disorganised arrangement of cells and partial loss of the epithelial basement membrane. Basement membrane defects were shown by light microscopy, immunohistochemistry and TEM. Immunohistochemical and TUNEL analysis showed that while the apical‐basal polarity of the individual basal epithelial cells was disrupted (demonstrated by mislocalisation of Scribble protein to the basal edge of basal cells) and cellular morphology was abnormal, the apical or basal identity of cells was maintained in mutant epithelia. Hence, proliferation was restricted to the basal layer and apoptosis to the superficial layer of the mutant epithelium. Mutant corneal epithelial cells showed a defect in corneal epithelium wound healing, but this may be secondary to the disruption of the basement membrane.

The role of PCP pathway proteins in adult vertebrate tissues is poorly known. Whereas classically, the PCP pathway controls the behaviour of epithelial cells in the plane of their basement membrane during embryological development, examples of PCP genes controlling apical‐basal behaviour of epithelial cells also exist. For example, the uterine epithelium of Vangl2 ^Lp/Lp mutant mouse embryos is disrupted with loss of apical‐basal polarity of columnar epithelial cells (Vandenberg & Sassoon, 2009). The columnar epithelium of Vangl2 ^Lp/Lp mutant uteri at E18.5 is composed of multiple cell layers rather than a single monolayer of columnar epithelial cells, reminiscent of the phenotype in Vangl2‐null corneal epithelia. Similar findings of apical‐basal disruption of epithelia in Vangl2 mutants were reported by Yates et al. (2010): airway lumina of the lungs of Vangl2 ^Lp/Lp homozygotes at E14.5 were found to be surrounded by disrupted epithelium. In wild‐type lung sections, airway lumina are surrounded by a single layer of epithelium made up of aligned and organised cells. In contrast, in the lungs of Vangl2 ^Lp/Lp homozygotes the lumina are demarcated by multi‐stratified, disorganised epithelial cells that are not aligned uniformly (Yates et al. 2010). Dysregulation or loss of Vangl2 and PCP activity have been shown to mediate increase of matrix metalloprotease activity, loss of epithelial morphology and metastasis in some cancers, a phenotype consistent with the epithelial disruption in Le‐Cre ^Tg/+ Vangl2 ^fl/fl corneas (Cantrell & Jessen, 2010; Puvirajesinghe et al. 2016).

The basement membrane is an ECM composed of secretions from the overlying basal layer of epithelial cells, the basal lamina and reticular connective tissue (Paulsson, 1992; LeBleu et al. 2007). It was thin and, in patches, absent in Le‐Cre ^Tg/+ Vangl2 ^fl/fl mutant eyes. Basement membrane laminin has been shown to secondarily maintain polarisation in epithelial cells (Klein et al. 1988; Xu et al. 2010). PI3K has previously been shown to locate to the basal side of mouse mammary epithelial cells due to the influence of laminin in the ECM inducing apical‐basal polarity within the cells (Xu et al. 2010). The results obtained in this study suggest that when Vangl2 expression is lost within corneal epithelial cells they lose polarity and exhibit defects in secretion of basal lamina components from the basal side of cells. The basement membrane is known to be essential for providing chemical cues that aid in epithelial migration (Abrams et al. 2000; Teixeira et al. 2003, 2006). Its absence could be the source of the disorganisation and wound healing delay observed in the Le‐Cre ^Tg/+ Vangl2 ^fl/fl mutant eyes.

A reduction in cell adhesion may underlie reduced planar migration of Vangl2‐deficient cells in response to wounding or an applied electric field. For example, Oteiza et al. (2010) reported impaired cell adhesion linked to PCP dysfunction: inhibition of Wnt11 and Prickle‐1a in Zebrafish impaired cell–cell adhesion of the progenitor cells of Kupffer's vesicle. A direct assay of cell adhesion in conditional knockout cells using single‐cell force spectroscopy (Puech et al. 2006) would be informative in this respect.

Apical‐basal complexes have previously been shown to interact with PCP components (Djiane et al. 2005; Dollar et al. 2005; Mahaffey et al. 2013). Vangl2 and Scribble have been linked in the determination of planar polarity in mice: Scribble ^Crc/+ Vangl2 ^Lp/+ double‐mutants exhibit cochlear disorganisation comparable to that observed in Vangl2 ^Lp/Lp mutants (Montcouquiol et al. 2003). The localisation of Scribble along the apical‐lateral edge of the epithelial cell layer in control corneal epithelia is consistent with previous observations that Scribble localises with the tight junction protein ZO‐1 (Nakgawa & Huibregtse, 2000). The results obtained in this study showed partial mislocalisation of Scribble to the basal side of the innermost layer of the epithelium in Le‐Cre ^Tg/− Vangl2 ^fl/fl mice. This is in contrast to observations made in Drosophila by Courbard et al. (2009) who found no defect in the localisation of Scribble in either Vangl or Frizzled mutants; they also found that there were no differences in Vangl localisation in Scribble mutants. Yates et al. (2013) observed that Vangl2 ^Lp/Lp mutants exhibited no defects in Scribble localisation within the mammalian lung. The data from this study would therefore suggest that the molecular roles of Vangl2 may differ in different tissues.

This study has confirmed a novel, sustained role for the core PCP component, Vangl2, in generation and maintenance of stratified epithelial morphology in an adult vertebrate system. It also provides preliminary evidence that at least some of the planar defects observed arise through a non‐PCP mechanism, with loss or disruption of the normal epithelial basement membrane. We consider it possible that some or even all of the planar defects may be secondary to a primary failure of apical‐basal polarity in the epithelial cells. This may represent a more general model for how PCP signalling is effected in other systems, and warrants further investigation.

We previously showed (Dorà et al. 2014) that the Le‐Cre transgene itself imposes a phenotype on the mouse eye, independently of the presence or absence of floxed alleles of any gene. The phenotype varies in severity with genetic background, but Le‐Cre ^Tg/− mice have a tendency to microphthalmia, small disorganised lenses and irido‐corneal adhesion with small pupillary opening, though their corneal structure is normal (Dorà et al. 2014). During this study, Le‐Cre ^Tg/− mice had mild microphthalmia (Fig. 4A) and variable lens fibre disruption. For this reason, all controls in this study were Le‐Cre ^Tg/−, and significant effects of Vangl2 reported here only for comparison of Le‐Cre ^Tg/− Vangl2 ^fl/fl compared with Le‐Cre ^Tg/− Vangl2 ^fl/+ and/or Le‐Cre ^Tg/− Vangl2 ^+/+. We conclude, in agreement with previous work (Dorà et al. 2014), that Cre‐negative mice are not valid controls for studies that have used Le‐Cre to conditionally inactivate floxed genes in the corneal epithelium. Several previous studies have, however, used the Le‐Cre transgene to knock out floxed genes in adult eyes and claimed the subsequent defects including microphthalmia, disorganised small lenses or small pupillary openings were due to the conditionally knocked out gene and not the Le‐Cre transgene itself. Table 1 lists these studies. With one exception, all have used Le‐Cre‐negative mice as controls and have detailed morphological defects in Le‐Cre‐positive fl/fl mice that are consistent with the phenotype caused by the Le‐Cre transgene expression within the developing eye. A few studies also highlight the loss of endogenous Pax6 expression within the Le‐Cre ^Tg/− mutant cells but claim the cause is the knocked out gene, whereas Dorà et al. (2014) showed that the eye phenotype in Le‐Cre ^Tg/− mice was most likely caused by reduced levels of Pax6. In all of these studies it is possible that the phenotype attributed to the gene knockout was in fact, at least in part, caused by the Le‐Cre transgene, and we would suggest that the datasets as a whole should be interpreted in this light.

Table 1.

Studies that have used LE‐CRE mice to study eye development

Publication	Gene	Control	Phenotype observed
Zhao et al. (2015)	ALK3	Le‐Cre −/− mice	Observed a smaller lens in Le‐Cre Tg/− mice
Zhang et al. (2015b)	FGFr2	Le‐Cre −/− mice	Observed a thinner corneal epithelial layer when compared with wild‐type mice, and abnormal Pax6 immunostaining
Choi et al. (2014)	Smoothened	Le‐Cre −/− mice	Observed microphthalmia in Le‐Cre Tg/− mice
Li et al. (2014)	Frs2α and Shp2	Le‐Cre −/− mice	Found that deletion of either gene caused a decrease in lens size
Yamben et al. (2013)	Scribble	Le‐Cre −/− mice	Observed small, misshapen lenses with central opacity in Le‐Cre Tg/− mice
Gupta et al. (2013)	Klf4	Le‐Cre −/− mice	Observed small lens with central opacity in Le‐Cre Tg/− mice
Zhao et al. (2012)	ALK3	Le‐Cre −/− mice	Observed small lens with notable defects in fibre cell differentiation in Le‐Cre Tg/− mice
Saravanamuthu et al. (2012)	Notch2	Le‐Cre −/− mice	Observed microphthalmia, and reduced pupillary openings
Kenchegowda et al. (2011)	Klf5	Le‐Cre −/− mice	Observed microphthalmia and closed eyelids, with small lenses in Le‐Cre Tg/− mice
Joo et al. (2010)	Pnn	Le‐Cre −/− mice	Observed microphthalmia
Chen et al. (2009)	Cited2	Le‐Cre −/− mice	Observed microphthalmia and thin opaque cornea. Found Le‐Cre Tg/− Cited2 fl/fl mice have a decrease in Pax6 expression
Le et al. (2009)	Jagged1	Le‐Cre Tg/− Jagged1 fl/+ and Le‐Cre −/− mice	Although this paper compares homozygous conditional knockouts with their heterozygous counterparts, they still claim all observations – microphthalmia and decreased pupillary openings – are due to a reduced expression of Jagged1
Rowan et al. (2008)	Rbp‐J	Le‐Cre −/− mice	Microphthalmia, small pupillary opening
Jia et al. (2007)	Rbp‐J	Le‐Cre −/− mice	Observed microphthalmia and small lens
Liu et al. (2006)	Six3	Le‐Cre −/− mice	Observed lens defects, reduced size, cataracts and absence

Author contributions

DAP and ASF both performed experiments, analysed data and contributed equally to writing the manuscript. RvL performed experiments and contributed to writing the manuscript. JMC conceived the study, performed experiments and wrote the manuscript.

Supporting information

Fig. S1 Collagen IV immunostaining of Le‐Cre ^Tg/− Vangl2 ^fl/fl conditional knockout cornea showing that although Collagen IV was absent from the epithelial basement membrane (BM, arrowhead) it was still detectable in Descemet's membrane (DM).

Fig. S2 Transmission electron microscopy (TEM) of control and Vangl2‐mutant corneas.