Introduction

Peters anomaly (OMIM 604229) is a rare congenital disorder characterized by corneal opacities, adhesions (iris lens and/or lens cornea) due to malformations of the posterior corneal stroma and absence of Descemet's membrane and the corneal endothelium.1 It is sometimes regarded as a sporadic condition with a low risk of recurrence and is often attributed to aneuploidy or fetal alcohol syndrome,2, 3 though reports of inherited cases associated with other ocular anomalies, such as congenital cataracts,4 dysgenesis of the irido-corneal angle,5 and cataract-microcornea syndrome (OMIM 116150) are common. Peters anomaly can also be associated with other malformations, including Krause–Kivlin syndrome (Peters plus syndrome; OMIM 261540) and Axenfeld–Rieger syndrome (OMIM 602482). Peters anomaly and other ocular malformations that affect the anterior chamber of the eye (including the cornea, iris, lens, and iridocorneal angle), belong to a spectrum of disorders collectively known as anterior segment dysgenesis (ASD).

Primarily, mutations in genes that encode transcription factors that show expression in the eye or other structural components have been implicated in ASD. For example, genes encoding transcription factors expressed during the migration of neural crest cells in eye development, such as PAX6,6 PITX2,7, 8 FOXC1,9 FOXE3,10, 11 and PITX3,12, 13 have been associated with disorders such as aniridia, Axenfeld–Rieger syndrome, Peters anomaly, and congenital cataracts. Genes that encode structural components of the eye, such as the crystallins (ie, CRYAA)14 or the gap junction gene GJA8,15 have also been associated with various forms of ASD, such as congenital cataracts and microcornea-cataract syndrome, respectively.

We previously reported a proband diagnosed with congenital Peters anomaly and a family history of a variable form of autosomal dominant ASD, including microcornea, congenital cataracts, and scleralization of the iris.16 In this paper, we present the progression of the ocular phenotype in this family observed over a 30-year period and describe the functional candidate approach used to determine the causative gene.

Materials and methods

Clinical investigation of proband and relatives

We studied a four-generation family from the island of Newfoundland with mild-to-severe forms of ASD segregating as an autosomal dominant trait (Figure 1). The proband was diagnosed at birth in 1979 with Peters anomaly.16 Long-term clinical and ophthalmological examination on family members has since been carried out (Table 1), and has included visual acuity measurements, refraction, direct and indirect ophthalmoscopy, slit lamp examination, tonometry, gonioscopy, corneal diameter measurements, axial length measurements (ultrasound), and external eye photographs as previously described.16 In all, 31 of 66 known family members have participated in this study, and of these, we have identified 11 cases of ASD across six sibships (Figure 1). Because of the variability of ocular phenotypes seen in this family, we conducted repeated, complete slit lamp examination of the cornea, iris, angle and lens, and measurement of the corneal diameter on a number of individuals. These examinations provided evidence for minimal diagnostic criteria, which included either corneal scleralization, lens opacities, strands in the angle, or small corneal diameter (<11 mm). This study was approved by the Human Investigations Committee, Memorial University, St John's, Newfoundland, Canada (HIC no. 02.116).

Figure 1
figure 1

A four generation Newfoundland family segregating an autosomal dominant form of anterior segment dysgenesis. The proband (PID III-6: arrow) has congenital Peters anomaly.

Table 1 Review of ocular findings in family members

Selection and screening of functional candidate genes

Genes involved in the structural development of the eye or with previous association with ASD, such as Peters anomaly and related disorders (ie, Axenfeld–Rieger syndrome, Krause–Kivlin syndrome), were sequentially screened for mutations. PAX6 encodes a transcription factor, which has been implicated in aniridia and Peters anomaly.6, 17 PITX2 (RIEG1) is primarily associated with Axenfeld–Rieger syndrome, a syndromic form of ASD with similar ocular phenotypes,8 though some mutations have been identified as causing non-syndromic Peters anomaly.7 PITX2 interacts with the transcription factor FOXC1,9, 18 and mutations in FOXC1 produce a similar phenotype as seen with PITX2 mutations. PITX3 has been associated with autosomal-dominant congenital cataracts and Peters anomaly.12, 13, 19 Mutations in B3GALTL, a galactosyltransferase gene, cause Peters plus syndrome, which consists of Peters anomaly, short stature, cleft lip, cleft palate, and mental retardation.20 Though there have been no reports of incomplete or atypical forms of Krause–Kivlin syndrome showing only the ocular phenotype, B3GALTL was screened in this family to check for possible hypomorphic alleles, which may cause the non-syndromic phenotype seen in this family. CRYAA, a member of the crystallin family, which constitutes the major structural protein in lens fiber cells has been associated with autosomal-dominant cataracts, microcornea, and corneal opacities.14, 21 GJA8 encodes a gap junction protein, which regulates ion and crystallin concentrations within the lens fiber cells.15 This gene has been associated with cataract-microcornea syndrome, and various other forms of cataracts. CYP1B1 is a member of the cytochrome P450 superfamily, and mutations in this gene have been attributed to primary congenital glaucoma, and Peters anomaly.22, 23, 24 Finally, FOXE3 is a lens-specific transcription factor which has been implicated in cases of ASD, including autosomal-dominant Peters anomaly,10 congenital cataracts,11 and autosomal-recessive congenital primary aphakia.25

DNA isolation and bi-directional sequencing

Genomic DNA was isolated from peripheral leukocytes26 and screened for mutations in the proband and four relatives with ocular phenotypes (Figure 1: pedigree identifier (PID) II-5; II-14; III-2; III-6; IV-1), representing the full range of phenotypes found in this family, and two unaffected individuals (PIDs: II-10; II-15). Oligonucleotide primers were designed using Primer 3 software (http://frodo.wi.mit.edu/) to amplify all coding regions, and UTRs of the following genes: PAX6 (NM_001604), PITX2 (NM_153426), PITX3 (NM_005029), CYP1B1 (NM_000104), FOXC1 (NM_001453), B3GALTL (NM_194318), CRYAA (NM_000394), GJA8 (NM_005267), and FOXE3 (NM_012186). Amplicons were run on a 1% agarose gel stained with SYBR Safe (Invitrogen by Life Technologies, Carlsbad, CA, USA) for size verification. Primer sequences and PCR conditions are available upon request.

Amplicons were purified using 50% S-300HR sephacryl (Amersham Biosciences, Uppsala, Sweden) and Multiscreen HTS filter plates (Millipore Corporation, Billerica, MA, USA). Bi-directional sequencing was carried out with BigDye Terminator Kit v 3.1 (Applied Biosystems, Foster City, CA, USA), and samples were run on an ABI 3130 × l DNA Analyzer. Electropherograms were inspected manually for quality and imported into Mutation Surveyor software v 3.2 (Transition Technologies, Toronto, ON, Canada) to detect sequence variants. All sequencing variants were checked for co-segregation with the ASD phenotype in the family. Only one allele (FOXE3 c. 959G>T) co-segregated with the disease phenotype. Population frequency of this variant was determined using ethnically matched controls from a previous study.27

Total RNA was extracted from Epstein–Barr virus-transformed B lymphocytes from one affected individual (PID IV-1) with Trizol (Invitrogen by Life Technologies, Carlsbad, CA, USA); this was followed by DNaseI treatment (Ambion, Austin, TX, USA). Complementary DNA (cDNA) synthesis was carried out; PCR was carried out using primers surrounding the stop codon of FOXE3, and the PCR products were analyzed by size fractionation using agarose gel electrophoresis and direct sequencing. A set of primers designed to amplify the intron between exons 5 and 6 of GAPDH was used as an internal control to check for the presence of genomic DNA contamination. These primers amplify a 170 bp fragment with a cDNA template, and a 278 bp fragment using a genomic template (Figure 3a). Primers and experimental conditions are available upon request.

Results

Long-term clinical follow-up

The proband (PID III-6) is a 30-year-old male born at 35 weeks gestation (Figure 1). Bilateral dense corneal opacities were noted at birth and were initially attributed to maternal prenatal infection. Examination under anesthetic showed corneal opacification of the proband's right eye from the temporal side past the midline and hazy cornea nasally (Table 1; Figure 2a). Fine adhesions from the collarette of the iris to the cornea obscured the angle. The left cornea of the proband was more densely opaque with a large central adhesion from the cornea to the lens and peripheral adhesions from the iris to the cornea (Figure 2b). Both globes were of normal size, but corneal measurements were 10 mm horizontally and 9 mm vertically (N≥11 mm). At 6 months of age, the proband had a corneal transplant of the more severely affected eye (left) and optical iridectomy of the right eye. Pathology review of the left corneal button indicated absent Descemet's membrane, partial absence of Bowman's membrane, and thinning of the central cornea consistent with Peters anomaly. The corneal graft was rejected 6 weeks post-operatively and the eye became phthisical.

Figure 2
figure 2

Eye phenotype of the proband (PID III-6) over time. (a) Picture of the proband's right eye at 7 weeks showing corneal opacities consistent with Peters anomaly. (b) Surgical drawing of the proband's left eye showing keratolenticular adhesion, and corneal leukoma (pictures a and b are taken from Green and Johnson16 with permission. (c) Proband's right eye at 30 years of age. (d) Proband's phthisical left eye at 30 years of age.

The proband was re-examined at 30 years of age by a clinical geneticist (BF; Figure 2c and d) and found to be of normal intelligence. His height was 172 cm (25th percentile); ratio of upper to lower segment 0.98; arm span 170 cm; weight 107 kg (>95th percentile); and head circumference was 62 cm (this information was not available for other relatives). There was no facial dysmorphism apart from features related to a phthisical left eye due to post-operative complications: the left eyebrow was lower and the left palpebral fissure was shorter than that of the right eye. His palate was intact with a single uvula, and the philtrum was well developed (philtrum length 2 cm, 50–75th percentile). Ears were large (length of 7.3 cm), but were normal in position and contour. The only other minor physical anomaly was clinodactyly of the right second and third fingers and of the left second finger. The lack of extra-ocular phenotypes seen in this patient allowed us to rule out Krause–Kivlin syndrome or other related syndromes as a possible diagnosis for this individual.

The father of the proband (PID II-14) was examined at 27 years of age (the year the proband was born), and shown to have small posterior subcapsular and central nuclear cataracts, first noted at age seven (Table 1). His visual acuity was 6/7.5-3 in the right eye and 6/9 in left eye. He had bilateral microcornea (corneal diameter of 10 mm) and scleralization and vascularization of the cornea, particularly superiorly. On gonioscopy, fine iris processes were noted extending over the trabecular meshwork. Subsequently, he had cataract extractions at ages 39 and 40 because of decreasing visual acuity.

Three paternal aunts (PIDs II-5; II-7; II-16), who had lens opacities documented in childhood, had cataract extractions before 1979 (in their 20s or 30s) when mature cataracts developed. All three individuals presented with microcornea (corneal diameters of 8.5–10.5 mm), mild-to-moderate scleralization of the cornea with varying degrees of vascularization, particularly superiorly and inferiorly (Table 1). Three paternal cousins (PIDs III-2; III-3; III-7) had cataract extractions at ages 16–40 years. The cataracts were originally described as anterior polar, anterior cortical, nuclear, and posterior subcapsular cataracts. Six of these seven affected family members had favorable results post-operatively, although one had vitreous hemorrhage, choroidal detachment, and temporary hypotony of one eye. Individual PID II-7 had serious post-operative complications, including retinal detachments with failed repair, failed corneal grafting, and enucleation of a painful blind eye with hand movement vision only in the remaining eye (Table 1).

The youngest affected family members are PID IV-1 (6 years old) and PID IV-2 (10 years old; Figure 1). Individual PID IV-2 has anterior polar cataracts detected at birth, but has not yet required cataract extraction. Individual PID IV-1 had a complicated pre- and postnatal course with extreme premature birth at 24 weeks gestation. He was hospitalized for 4½ months during which time he had necrotizing enterocolitis with perforation of the bowel, and successful treatment of retinopathy of prematurity. Because of the size and central location of his anterior polar cataracts, he had cataract extractions at 8 months of age (Table 1). He also had bilateral iridectomies because of the central corneal opacities. His central acuity is recorded as 6/18, but his course continues to be complicated with a recent diagnosis of autism.

A paternal uncle who died of pneumonia at 6 months of age in the 1940s was likely affected, as he had ‘white eyes’ and was registered blind with the Canadian National Institute for the Blind (PID II-9). Eight other paternal aunts and uncles and 15 paternal cousins had normal corneal diameters and clear corneas and lenses upon examination (although some had a prominent arcus suggesting elevated cholesterol levels).

Interestingly, there is no evidence that either of the grandparents had any form of ASD (DNA not available). For instance, the paternal grandfather of the proband (PID I-1) was examined and had corneal diameters of 11 mm, mild corneal scleralization, inferior arcus, and faint lens opacities, including anterior polar specks, and anterior and posterior cortical spokes but retained 6/7.5 and 6/6 visual acuity at 77 years of age. Two of his brothers had clear corneas and lenses in their 60s or 70s. The paternal grandmother of the proband (PID I-2) had corneal diameter of 11.5 mm, mild central endothelial corneal changes (Fuch's dystrophy), arcus senilis, and faint dot and spoke lens opacities when examined at 71 years. She subsequently had mature cataracts extracted at ages 78 and 81. On examination, her brother, sister, niece, and nephew all had clear corneas and lenses, and were considered unaffected. The phenotypes seen in the grandparents can all be attributed to the aging process. It is possible that one had a subclinical phenotype or that gonadal mosaicism was present.

Candidate gene screen reveals causative variant in FOXE3

Screening of the nine functional candidate genes in seven individuals (five affected; two unaffected PIDS: III-6, II-14; II-5; III-2; IV-1; II-10; II-15) revealed 45 sequence variants (Table 2). Segregation analysis showed co-segregation between the ASD trait and a non-stop mutation within the transcription factor gene FOXE3 (c.959 G>T: p.X320L; Figure 3b) causing the elimination of the functional opal stop codon (UGA) at the 3′ end of FOXE3. The next available stop codon is 213 bps downstream (in the 3′ UTR) predicting the addition of 72 amino acid residues to the C-terminus of FOXE3.

Table 2 Summary of 45 sequencing variants identified in nine functional candidate genes sequenced in five affected members with a range of anterior segment dysgenesis from mild phenotype to Peters anomaly, and two unaffected members
Figure 3
figure 3

(a) Agarose gel (1%) stained with SYBR Safe showing PCR amplification of FOXE3 from cDNA from a lymphoblastoid cell line of PID IV-1 (in quadruplicate). The internal control of GAPDH amplifying a 170 bp fragment to check for presence of cDNA (and absence of gDNA) is shown below. A genomic control is shown to the right, and the 278 bp fragment amplified from GAPDH exclusively in gDNA. (b, top) Electropherogram of the novel, non-stop mutation in FOXE3 (c.959 G>T: p.X320L) identified in all affected family members. (Bottom) Electropherogram of the region surrounding the stop codon of FOXE3 from the cDNA of an affected patient (PID IV-1) isolated from a lymphoblast cell line.

We then genotyped all available members of the extended family and found that the variant co-segregated with the disease phenotype (Table 2). This variant was not detected in 141 ethnically matched controls. To test whether this mutation would be present in the mRNA of FOXE3, RNA was isolated from a lymphoblast cell line of individual PID IV-1 and converted to cDNA. FOXE3 is a lens-specific gene expressed only in the lens epithelium,10 as a result this approach was not guaranteed to succeed. However, we were able to amplify a 458 bp cDNA product from FOXE3 surrounding the c.959 G>T mutation (Figure 3a). Direct sequencing revealed that the c.959 G>T mutation was absent in the cDNA (Figure 3b), suggesting that the mRNA transcribed from the ‘non-stop’ allele may be degraded before being translated, or it is possible that the RNA is not transcribed at all.

Discussion

Peters anomaly is a rare congenital disorder often regarded as a sporadic condition with a low risk of recurrence and attributed to aneuploidy or fetal alcohol syndrome. Inherited cases associated with malformations of the anterior chamber of the eye have also been described. In 1986, we first reported a family with an autosomal dominant form of ASD with the proband exhibiting Peters anomaly.16 This four generation family from Newfoundland has had extensive clinical assessment over the course of 30 years, revealing a progression over time of the ASD phenotype in affected family members. For example, the father of the proband (PID-II-14) had mild cataracts at the age of 7 and required cataract extractions at the ages of 39 and 40 because of decreasing visual acuity. Other affected family members (PIDs II-5; II-7; II-16) have also shown a non-static phenotype with mild cataracts noted at young ages progressing to mature cataracts causing decreased visual acuity, making extractions necessary at a later age.

The anterior chamber is formed by three successive waves of embryonic cells derived from the neural crest. Aberration in the migration of these cells can cause anterior chamber defects, such as microcornea, corneal opacities, or congenital cataracts.5 Using a functional candidate gene approach, we looked for mutations in this family, in genes previously associated with ASD or Peters anomaly and identified a novel non-stop mutation in FOXE3 (c.959 G>T: p.X320L; Figure 3b), which segregated with the variable ocular phenotype, and was absent from 282 ethnically matched chromosomes. Though rare, non-stop mutations are reported in a number of cases involving various diseases, such as in non-classic 3-β-HSD congenital adrenal hyperplasia,28 and excessive hemorrhaging because of mutation of coagulation FX.29

The normal interaction of mouse Foxe3 with genes responsible for cellular differentiation and proliferation provides an explanation for the presence of small lenses and other eye malformations seen in humans. In mice, Foxe3 encodes a DNA-binding transcription factor expressed during the formation of the lens placode and assists in the formation of the lens itself.30 Mutations in Foxe3 reduce the ability of the transcription factor to bind DNA, causing formation of a small lens.10 Sometimes the anterior lens epithelium does not separate from the cornea, resulting in keratolenticular adhesions,10 which we observed in the proband with Peters anomaly. One explanation is that this adhesion is caused by a dysregulation of genes responsible for apoptosis within the corneal stalk, which is thought to degrade during lens morphogenesis as a result of apoptosis.31 An alternative explanation is that mutation of Foxe3 may cause a dysregulation of cadherin proteins, such as E-cadherin, which are present in the corneal stalk.31 Foxe3 also controls the expression of Cryaα within the developing lens, and dysfunction of Cryaα expression reduces the solubility of the crystallin protein complex causing crystallization and potential cataract formation. Foxe3 is responsible for the downregulation of Prox1, which controls a gene responsible for blocking cell cycle progression (cyclin dependant kinase inhibitor; Cdkn1c). As Foxe3 expression becomes dysregulated, Prox1 expression increases, subsequently increasing Cdkn1c expression causing a reduction of cellular proliferation in the anterior lens.30, 32 Foxe3 also controls expression of platelet-derived growth factor receptor-α (Pdgfrα), which is responsible for lens fiber differentiation within the anterior lens epithelium.30 Post-natally, FOXE3 is expressed exclusively in the anterior lens epithelium which is the only site of cell proliferation within the lens. This restricted cellular expression is consistent with the non-syndromic phenotype observed in FOXE3 mutation carriers.30, 31

Mutations in FOXE3 and its involvement in human ASD and/or Peters anomaly were first identified by Semina et al,11 who described two patients with apparent autosomal dominant posterior embryotoxon and congenital cataracts caused by a heterozygous single nucleotide insertion which altered the 5′ terminal amino acids, and added an additional 111 amino acids to the FOXE3 protein.11 A second report of FOXE3 and its involvement in ASD and/or Peters anomaly was published by Ormestad et al10, who identified a heterozygous missense mutation (c.524 G>T: p.R90L) coding for the fork head DNA-binding domain of the FOXE3 protein of a single affected individual with familial Peters anomaly, though no other DNA samples from the family were available to confirm segregation. Recently, Iseri et al33 described four other families containing FOXE3 mutations, c.21_24del:p.M71IfsX216, c.146G>C:p.G49A, c.244A>G: p.M82V, c.958T>C p.X320ArgextX72, all of which are hypothesized to be pathogenic. Homozygous mutations within Foxe3/FOXE3 can also cause autosomal recessive congenital aphakia within mice (dyl mice)30 and humans respectively.25 Also, mice containing heterozygous mutations within Foxe3 showed histological ocular malformations upon investigation,10 further showing the ability of FOXE3 to cause dominant and recessive ocular conditions.

Two recent reports (Iseri et al 33 and Brémond-Gignac 34) identified similar non-stop mutations within FOXE3 (c.958T>C: p.X320ArgextX72 and c.959G>C: p.X320SerextX72 respectively) in patients with ocular diseases. Iseri et al33 reported a family with autosomal dominant ASD in a proband diagnosed with unilateral Peters anomaly as well as in four family members segregating congenital cataracts, microphthalmia, and iris coloboma. Brémond-Gignac et al34 reported a single patient with congenital cataract and no family history of ocular disease. Both of these studies predict a similar 72 amino acid extension of the FOXE3 protein, though no functional analyses were performed to confirm this. It is clear that functional studies must be conducted to confirm that an extended protein is produced, as our results suggest that these non-stop RNA species may be degraded. The mutation identified in the Newfoundland family (c.959 G>T) occurs in the stop codon of FOXE3 and according to our preliminary data from cDNA (lymphoblastoid cell line) of individual PID IV-1, the mRNA transcript containing the non-stop allele is degraded (or not transcribed) suggesting haploinsufficiency of FOXE3 (Figure 3). An alternative interpretation of the data is that the RNA containing the c.959 T allele may be more unstable than that of the wild type, making it under-represented in total RNA. This could also explain why the mutation was not detected by direct sequencing of cDNA, thus quantitative and translation based studies are required to confirm the absence of both the mutant RNA and the extended protein.

Semina et al11 also described an insertion which caused the addition of 111 amino acids to the FOXE3 protein. It is possible that this mutation would also undergo the same type of mRNA degradation that we observe. We are currently conducting quantitative studies of the FOXE3 c.959 G>T mutation within human cells to validate our hypothesis of RNA degradation.

In summary, our results implicate FOXE3 in the pathology of ASD and corroborate other recent reports of ASD cases with FOXE3 mutations. Therefore, mutations within FOXE3 may explain currently unsolved ASD cases and should be screened. Discovery of genes and mutations causing ASD in families is of particular clinical relevance. A molecular diagnosis of ASD cases through mutation screening can provide accurate risks of recurrence, especially in light of mild or subclinical phenotypes, which may show progression over time such as those seen in the family seen in this study.