Heterozygous Mutations of OTX2 Cause Severe Ocular Malformations

Patients and Controls

Patients with ocular malformation spectrum defects (n=333) were recruited, with informed consent and with appropriate ethics review committee approvals. Detailed case histories of individuals with OTX2 coding-region sequence changes are available in appendix A (online only). Control DNAs (n=164) from apparently healthy adults were anonymized; they matched the patients in ethnicity (apart from patient 7, who was of African origin [see the “Results” section]). We sequenced 94 DNAs in full, and a further 70 partially, over the region of the proline missense mutations.

DNA Preparation

Genomic DNA was prepared from peripheral blood by use of a Nucleon DNA extraction kit (Tepnel Life Sciences). Buccal cell lysates were prepared from mouthwash samples by resuspending the cell pellet in 100 μl of 0.05 M NaOH, heating in a boiling water bath for 20 min, and then neutralizing with 12.5 μl of 1 M Tris (pH 7.6).

PCR and Sequencing

The human OTX2 cDNA (GenBank accession number NM_172337) is encoded by three exons that are located in BAC AL161757 (GenBank) at positions 59950–60217 (exon 1), 61214–6138 (exon 2), and 6322–64868 (exon 3). The cDNA numbering is derived from the human OTX2 cDNA (GenBank accession number NM_172337). We designed four genomic amplicons to cover the OTX2 coding region and some flanking intronic sequences in each case: amplicon 1 contained exon 1, amplicon 2 contained exon 2, and overlapping amplicons 3a and 3b covered the coding part of exon 3. PCR primers are shown in table 1. PCRs were performed in 96-well plates. For amplicons 1 and 2, 50 ng genomic DNA was amplified in a volume of 50 μl containing 1× Qiagen HotStarTaq PCR buffer, 1.5 mM MgSO₄, 5 μl PCR enhancer buffer (Invitrogen), 0.2 mM dNTPs, 250 nM forward primer, 250 nM reverse primer, and 2 U of HotStarTaq polymerase. For amplicons 3a and 3b, the PCR enhancer buffer was omitted and was replaced with 10 μl of Q solution (Qiagen). Thermal cycling was performed using an MJ PTC225 machine. PCR conditions were 1 cycle of 95°C for 15 min; 35 cycles of 94°C for 30 s, 55°C for 30 s, and 72°C for 1 min; and 1 cycle of 72°C for 5 min.

PCR products were run on a 2% agarose gel to ensure adequate yield and to check for the absence of nonspecific products. Unincorporated primers and dNTPs were removed by incubating 5 μl of PCR product with 1 μl of ExoSapIT (exonuclease I and shrimp alkaline phosphatase enzyme mix [USB]) for 45 min at 37°C, followed by 20 min at 80°C to inactivate the enzyme.

Direct Sequencing

Sequencing of PCR products was performed in a 96-well plate format, as described elsewhere (Aijaz et al. 2004), with the use of forward and reverse primers as shown in table 1. Individual sequence traces were inspected using Chromas software (Technelysium). Multiple sequence traces were aligned and compared using the phredPhrap program from the Consed package (see the Laboratory of Phil Green Web site). All changes were verified by performing a second PCR and sequencing reaction on the patient’s DNA. Family members were analyzed if samples were available.

RNA In Situ Hybridization

The expression pattern of OTX2 was investigated by nonradioactive RNA in situ hybridization in human embryos from Carnegie stages (CS) 14–22, which correspond approximately to mouse Theiler stages 18–23 (E11.25–E15.25). Human embryos were obtained from the Medical Research Council/Wellcome Trust Human Developmental Biology Resource, with full ethical approval. Preparation of embryo sections and nonradioactive RNA in situ hybridization was performed as described elsewhere (Lai et al. 2003).

The template for the OTX2 probe was a 1.1-kb fragment that was PCR amplified from exon 3 with the use of primers hOTX2A (5′-GCG GAT CCA GTG CTC CTG TGT CTA TCT GGA G-3′) and hOTX2D (5′-GCG AAT TCC AGA GGT GGA GTT CAA GGT TGC-3′) and that was cloned into pGEM-3Z. The construct was linearized with BamHI and was transcribed with T7 RNA polymerase to synthesize the digoxygenin-labeled antisense riboprobe.

MRI Analysis

MRI scanning was approved by the local institutional ethics committees. All patients scanned for the present study provided written informed consent. Patient 5 and his unaffected mother underwent MRI scanning. MRI acquisition parameters were TE = 4.2, TI = 450, TR = 15, NEX = 1, flip angle = 20°, acquisition matrix 256 × 128, and field of view 24 cm; 124 contiguous slices were produced, with voxel dimension 0.9375 mm × 0.9375 mm × 1.5 mm. An equivalent sequence was used for patient 4B. These data were reformatted in multiple planes to allow careful examination of regions of interest. Additionally, signal changes were inspected on T2 and FastFLAIR sequences (T2 and PD sequence: TE1 = 30, TE2 = 120, TR = 2000, NEX = 1, acquisition matrix 256 × 128, field of view 24 cm × 18 cm, and slice thickness 5 mm contiguous; FastFLAIR sequence: TE1 = 152, TE2 = 2200, TR = 10002, NEX = 1, acquisition matrix 256 × 128, field of view 24 cm, and slice thickness 5 mm contiguous). Patient 4B underwent similar high-resolution T1-weighted MRI scanning. For patients 1, 2, and 6, existing MRI data were reviewed. For patient 3A, CT scan data were reviewed.

Psychometric Testing

Patient 5 and his mother underwent standard neuropsychometric testing, omitting any tests requiring visual function. This component of the study was also approved by the local institutional ethics committee and was undertaken after written informed consent was obtained.

OTX2 Mutation Analysis

Of the 333 patients analyzed, 8 showed sequence variation in the OTX2 coding region. The changes are distributed throughout the ORF (fig. 1). Additional family members were analyzed, if samples were available, for a total of 11 affected individuals in eight families (fig. 1). Detailed case histories of affected individuals are presented in appendix A (online only), and the phenotypes are summarized in table 2. In the course of the study, we uncovered evidence of mosaicism; therefore, when possible, we also analyzed oral epithelial DNA (obtained by mouthwash or buccal scrape).

Patient 1, with bilateral anophthalmia, has a de novo frameshifting insertion (c.635insGC) at codon 155 in exon 3, the A in the SIWSPA motif (figs. 1 and 2 and table 2). Subsequent out-of-frame translation of 22 amino acids is predicted to terminate at a TGA codon, beginning at c.701. Neither the parents nor the unaffected brother carry the mutation (fig. 1). Identical results were obtained with buccal cell DNA from all four individuals.

Patient 2, with bilateral optic-nerve aplasia (MIM 165550) and microphthalmia (fig. 3 and table 2), has a de novo transversion, c.436C→G, in exon 3 (fig. 1). This is predicted to produce an arginine-to-glycine (R89G) change at position 52 of the homeodomain. Neither parent carries the change in lymphocyte DNA (fig. 1).

Patient 3A, with severe bilateral microphthalmia, has an N-terminal domain frameshifting mutation, c.252delC, in codon 27 of exon 1 (figs. 1 and 2 and table 2). Out-of-frame translation is predicted until the first base of a termination codon is encountered at c.320 in exon 2. A fetus (patient 3B [fig. 1]), terminated after ultrasound diagnosis of affected status, also carried the same mutation. The phenotypically normal mother of patients 3A and 3B was found to have the mutation at low levels in blood and buccal DNA before the existence of the affected fetus was revealed. The unaffected father and maternal grandparents do not have the mutation in blood DNA (fig. 1). These data indicate that the mother is a gonosomal (gonadal and somatic) mosaic carrier of the mutation (Edwards 1989).

Siblings 4A and 4B—diagnosed with microphthalmia and severe developmental delay and with LCA, respectively (figs. 2 and 4 and table 2)—carry a transversion, c.708T→A, in exon 3 (fig. 1). The tyrosine-to-termination (TAT→TAA) change at codon 179 is in the C-terminal domain. Their mother (patient 4C) is a gonosomal mosaic for the Y179X mutation (fig. 1). She has a later-onset phenotype resembling pigmentary retinopathy (fig. 4).

Patient 5, with bilateral anophthalmia, has a frameshifting deletion, c.288delCC, in codons 39–40 in exon 2, at the start of the homeodomain (figs. 1 and 2 and table 2). This mutation is predicted to generate a 46-aa nonsense peptide before reaching a TAA termination codon, beginning at c.429 in exon 3. The phenotypically normal mother of this patient has full representation of the same mutation in both lymphocyte and buccal DNA, indicating that she is probably a nonmosaic mutation carrier. The unaffected father, brother, and sister do not carry the mutation (fig. 1).

Patient 6, with bilateral anophthalmia, has the transition c.466C→T in exon 3 (figs. 1 and 2 and table 2). Glutamine (CAA) is substituted by a termination codon (TAA) at the second amino acid of the C-terminal domain (Q99X). The phenotypically normal father shows full expression of the same mutation in his blood DNA, suggesting that he is a constitutive mutation carrier. The unaffected mother does not have the mutation (fig. 1).

Patient 7, with bilateral microphthalmia and left sclerocornea, has the transversion c.568C→A in exon 3 (figs. 1 and 2 and table 2). This results in replacement of proline (CCC) with threonine (ACC) at position 133 (P133T). The patient’s phenotypically normal mother and brother have the same change, but the patient’s unaffected father and sister do not (fig. 1). Identical results were obtained with DNA from blood and oral epithelium from all five family members. P133 occurs at the end of a highly conserved hexapeptide motif, SGQFTP, in the C-terminal domain of vertebrate Otx2, Otx1, and Crx proteins (Germot et al. 2001).

Patient 8, with unilateral anophthalmia, has the transversion c.571C→G in exon 3 (figs. 1 and 2 and table 2). This replaces proline (CCC) with alanine (GCC) at amino acid position 134 (P134A). The patient’s similarly affected mother does not have this variant in blood or oral epithelium. The father was not available for analysis. The patient’s unaffected maternal grandmother and maternal great-grandmother do not have the change, nor does a maternal half brother (fig. 1).

Ninety-four control individuals (188 chromosomes) were examined by direct sequencing for the presence of all the coding-region variants. An additional 70 individuals were sequenced for the region containing prolines 133 and 134. None of the changes was observed in the controls. Ethnic matching was not feasible for family 7, who are of African origin, and therefore no conclusion can be drawn about the frequency of this change in the population from which patient 7 comes.

We detected three SNPs during the course of the present study. All were located in noncoding regions: two newly reported sites (submitted to NCBI dbSNP) in the first intron, 268+12 C/T (dbSNP accession number ss35522250) and 269–70 C/A (dbSNP accession number ss35522251), and one already-known variant near the start of the 3′ UTR (c.1050 G/A) (dbSNP accession number rs171978).

MRI Data and Neuropsychometric Testing of OTX2 Mutation Carriers

Three of the patients with OTX2 mutations have severe learning disabilities; these disabilities are associated, in two patients, with frequent seizures (table 2). These neurological phenotypes, together with the extensive expression of OTX2 in the brain, indicated that MRI scans and neuropsychometric testing might be informative, as shown for PAX6 (Sisodiya et al. 2001).

Patient 4B, patient 5, and the unaffected mother of patient 5 had MRI scans. Patient 5 and his mother also underwent neuropsychometric testing. Original diagnostic MRI data were reviewed for patients 1, 2, and 6, and CT data were reviewed for patient 3A. For patients 2 and 3A, only ocular data were available.

In four of the six patients for whom data are available, the optic nerves and optic chiasm are abnormal (fig. 3 and table 2). The nerves and chiasm are absent in patients 2 and 6 and are hypoplastic in patient 5. In patient 1, the optic nerve is absent on the anophthalmic side and is hypoplastic on the microphthalmic side; the chiasm is thin.

Patients 3A and 5 both had slightly enlarged lateral ventricles. A striking cerebral anomaly is evident in patient 6, who has bilateral abnormal hippocampal morphology (fig. 5). A similar hippocampal malformation was seen, on review, in patient 1.

Patient 5 functions intellectually just within the average ability range (Wechsler Adult Intelligence Scale–Revised [WAIS-R]: verbal IQ [VIQ] 91). His verbal memory was good and was in keeping with his intellectual level (verbal learning and recall 25th–50th percentile). His performance on tests of executive skills was average. His 55-year-old mutation-carrying mother has no eye phenotype and functions intellectually at a level comparable to that found for her son (VIQ 96). Her memory was not as good as that of her son: after correcting for age, her scores were in the 10th–25th percentile. Executive skills were average.

Patient 2 was described as having cognitive and language skills at least age-appropriate for a blind child when a neurodisability assessment was performed at age 4.5 years.

RNA In Situ Hybridization

The developing human eye at CS16–CS19, like that of the mouse at a corresponding age, shows strong OTX2 expression in the RPE layer and weaker OTX2 expression in the neural retina (fig. 6F, 6G, and 6H and data not shown). Mice have low-level expression in the E11 neural retina, which becomes restricted to the neuroblastic layer by E17.5 and which remains present in the inner nuclear layer into the postnatal period (Martinez-Morales et al. 2001; Nishida et al. 2003).

OTX2 is widely expressed throughout the telencephalon and mesencephalon at CS14, with a sharp cutoff at the midbrain-hindbrain boundary (fig. 6A). OTX2 expression was observed in the lamina terminalis and the floor of the telencephalon at CS16 (fig. 6B and 6C), sites that have also been reported in mouse (Simeone et al. 1993; Mallamaci et al. 1996; Simeone 2000). At CS22, OTX2 was expressed in the choroid plexus, the dorsal thalamus, and the roof of the mesencephalon (fig. 6E). In the developing nasal structures, OTX2 is seen in the olfactory epithelium of the nasal pits at CS18 (fig. 6D), again mirroring the mouse patterns.

Spectrum of OTX2 Mutations and Genetic Mechanisms

We found eight different heterozygous OTX2 exonic changes in the patient cohort. Five of the eight changes are predicted to cause premature protein truncation either through direct creation of a stop codon or through a frameshift (fig. 1 and table 2). The only mutation predicted to lead to haploinsufficiency through nonsense-mediated decay (reviewed by Holbrook et al. [2004]) is in family 3, in which a stop codon is predicted in exon 2. In the other four families (1, 4, 5, and 6), termination is predicted in the last exon—making it likely that the truncated protein is expressed and that it may function in a dominant negative manner (Nussbaum et al. 2001).

The R89G de novo missense change found in patient 2 causes substitution of an invariant homeodomain arginine that is normally in contact with DNA phosphate residues. Missense mutations of the homologous residue have been identified in seven different homeodomain proteins, three of them (PITX2 [MIM 601542], SIX3 [MIM 603714], and CRX) associated with eye disease (Swaroop et al. 1999; D’Elia et al. 2001; Laflamme et al. 2004).

The mutations found in patient 1 (635insGC) and patient 2 (R89G) arose de novo. In siblings 3A and 3B (252delC) and siblings 4A and 4B (Y179X), the sequence changes were inherited from mothers who were found to carry, at a low level in blood DNA, the same mutation as their offspring. In both cases, the mother transmitted the mutation to two affected children, providing strong evidence that these mothers are gonosomal mosaics with a high frequency of mutant cells in their germ lines (Edwards 1989). The parents of the mosaic individual should not have the mutation, as observed in family 3 (fig. 1). Gonosomal mosaic individuals may or may not express clinical features themselves, depending on the proportion and distribution of mutant cells in their body. The mosaic mother in family 3 has undergone a full eye examination, including high-resolution multifocal electroretinogram (ERG), and is ophthalmologically normal. The mosaic mother in family 4 (patient 4C) has progressive retinal dystrophy (fig. 4E and 4F), with a self-reported history of undefined visual problems from an early age. Typically, an affected mosaic individual has a milder phenotype than his or her constitutive offspring (Zlotogora 1998). This seems to be true for patient 4C, since she has retinal pathology without the structural eye defects observed in her children.

Families 1–4 fit a model in which OTX2 haploinsufficiency (loss of function in one allele) alone causes structural eye malformations. However, families 5 and 6 provide evidence of incomplete penetrance, since predicted protein-truncating mutations identified in bilateral anophthalmic children are also found in an unaffected mother and father, respectively (fig. 1 and table 2).

Results from families 7 and 8 are difficult to interpret. Both have missense substitutions at adjacent highly conserved proline residues, but the functional consequences of these sequence changes are unknown. Both families show discordance of phenotype and genotype. For family 7, no ethnically matched controls were available, so the frequency of the P133T change in the population of origin of family 7 is unknown. In family 8, the affected son has the P134A change, but his very similarly affected mother does not, making it highly unlikely that this sequence variant is a major contributor to the phenotype. We shall now focus on the loss-of-function mutations.

Two different models, based on recent insights from other human genetic diseases, may explain incomplete penetrance. First, if OTX2 is the only locus involved, the level of gene expression may be affected by stochastic and/or genetic factors. This model assumes that an abnormal phenotype will be manifest if OTX2 expression falls below a certain critical threshold level and if loss of function of one copy reduces gene expression levels close to this threshold. Stochastic fluctuations in transcription levels, possibly influenced by environmental factors but independent of genetic background, may reduce expression levels below the threshold. Evidence of this comes from the observation that genetically identical Otx2-knockout mice have highly variable phenotypes (Matsuo et al. 1995). Genetic factors at the mutant locus—specifically, regulatory polymorphisms of the normal allele—can also determine penetrance by controlling transcription levels (Gouya et al. 2002; Vithana et al. 2003). If the mutant allele is in trans to a strongly expressing normal allele, transcription is maintained above the threshold, which explains the clinically unaffected status of constitutive mutation carriers, but a weakly expressing normal allele leads to expression below the threshold required for normal development, and the abnormal phenotype is revealed (Gouya et al. 2002). The human OTX2 locus spans >500 kb and has numerous distant enhancers that could be targets for regulatory polymorphisms (Kurokawa et al. 2004b).

Second, digenic or oligogenic inheritance may manifest as incomplete penetrance (Ming and Muenke 2002; van Heyningen and Yeyati 2004). An example of digenic inheritance is provided by a form of retinitis pigmentosa in which affected individuals have heterozygous mutations in both the ROM1 (MIM 180721) and RDS (MIM 179605) genes, which encode related components of an oligomeric photoreceptor protein. Mutations at only one locus do not lead to an abnormal phenotype; penetrance is dependent on cosegregation of both mutations (Kajiwara et al. 1994). Perhaps more relevant to the developmental complexity of the eye and brain is holoprosencephaly (HPE [MIM 142945 and MIM 157170]), a disease that frequently shows non-Mendelian segregation patterns (Ming and Muenke 2002). The complex inheritance patterns and range of phenotypes we observe for OTX2 are similar to those reported for SHH (MIM 600725) and SIX3, the major contributors toward HPE types 3 and 2, respectively (Nanni et al. 1999; Wallis et al. 1999). Whereas some HPE-associated SHH and SIX3 mutations have arisen de novo, others have been inherited from completely unaffected carrier parents (Nanni et al. 1999; Wallis et al. 1999). The possibility that interaction with modifier gene variants at other loci may be required for full penetrance is supported by the observation that, in three reported cases, affected individuals carry mutations in two HPE genes (SHH with TGIF [MIM 602630] or ZIC2 [MIM 603073]) (Nanni et al. 1999; Ming and Muenke 2002). Most HPE genes are involved in developmental signaling, and it is possible that reduction in the dosage of one protein can be buffered but two cannot, so that expression of an abnormal phenotype may depend on mutations in more than one gene (Ming and Muenke 2002).

Of six families with loss-of-function mutations, four have mutation-carrier parents (families 3, 4, 5, and 6 [fig. 1]), three of whom are unaffected. Two parents are probably constitutive mutation carriers, and two are gonosomal mosaics. This presents challenges for genetic counseling of families with OTX2 mutations. Estimating the risk of having an affected child is extremely difficult. The relationship between levels of somatic and germline mosaicism is unpredictable. Little is known about either the precise stage at which the germ line is determined in humans or the number of precursor cells recruited. However, the number of cells ancestral to the germ line in mammals is thought to be small, so an individual who is gonosomal mosaic is likely to have a significant proportion of germ cells carrying the mutation (Zlotogora 1998; Leuer et al. 2001). Consequently, a relatively high recurrence risk is predicted for the offspring of gonosomal mosaic individuals, even if the proportion of mutant alleles in blood DNA is low. This has been observed in practice by others (Zlotogora 1998; Leuer et al. 2001) and in the present study, with both gonosomal mosaic mothers passing the mutation to two affected offspring. Constitutive-mutation-carrier parents are expected to be at 50% risk of having an affected child; however, recurrence was observed only in families with a mosaic parent. The lack of penetrance of OTX2 mutations must also be taken into account; the recurrence risk will presumably be lower if, as we suggest, other factors are required for manifestation of an abnormal phenotype. The relevant factors are, however, already present in families with one or more affected offspring.

Unaffected parents may be constitutive or mosaic mutation carriers; therefore, it is essential to test them for any OTX2 mutation identified in affected probands. However, accurate estimates of the recurrence risk await a molecular definition of the factors that determine penetrance of these mutations.

Phenotypic Variability

OTX2 loss-of-function mutations are associated with a broad spectrum of ocular phenotypes, ranging from bilateral anophthalmia to mild microphthalmia with retinal abnormalities. Asymmetry between the two sides, also noted in other genetic ocular malformations (Fantes et al. 2003; Ragge et al. 2005), is frequently seen. The spectrum of neurological phenotypes is broad, ranging from severe developmental delay to normal cognitive development. Variable expressivity is illustrated by the different phenotypes of siblings 4A and 4B.

The OTX2 transcription pattern in human embryos is very similar to that described for mouse at the stages examined. This is not always the case (Fougerousse et al. 2000). The highly conserved expression pattern indicates conserved transcriptional control mechanisms. Analysis of the Otx2 loci in human, mouse, Xenopus, zebrafish, and Fugu demonstrated that many of the distant enhancers that control Otx2 expression are functionally conserved (Kurokawa et al. 2004a, 2004b). These elements, spread over hundreds of kilobases of genomic DNA, are possible targets for mutations/variations that might influence phenotypic severity and penetrance.

The ocular expression pattern of OTX2 correlates well with the range of eye phenotypes observed in patients with loss-of-function mutations. Otx2 interacts with Mitf to specify the RPE, which is essential for normal eye development (Martinez-Morales et al. 2003). Early RPE defects cause gross structural abnormalities of the eye (Raymond and Jackson 1995; Graw 2003). The anophthalmia and severe microphthalmia phenotypes in our cohort may result directly from loss of OTX2 function in the developing RPE.

In the developing neural retina, OTX2 regulates CRX, and CRX mutations cause LCA and retinal degeneration (Sohocki et al. 1998; Rivolta et al. 2001; Nishida et al. 2003). The retinal phenotypes in patients 4B (with an original diagnosis of LCA) and 4C (resembling pigmentary retinopathy) may be due to disturbed OTX2 regulation of CRX function.

The optic nerve and chiasm were reduced or absent in four of the six patients for whom MRI, CT, or ultrasound data were available (patients 1, 2, 5 and 6 [fig. 3 and table 2]). Optic-nerve aplasia, which is very rare, was observed bilaterally in patients 2 and 6, with absence of the chiasm, and unilaterally on the anophthalmic side in patient 1, who also shows a thin chiasm. Hypoplasia of optic nerves and chiasm was observed in patient 5. Although OTX2/Otx2 is expressed in the optic stalk and subsequently in the optic-nerve sheath, it is not expressed in the optic nerve itself (fig. 6) (Simeone et al. 1993; Bovolenta et al. 1997; Martinez-Morales et al. 2001). Hypoplasia of the optic nerve and chiasm may therefore be a secondary effect of abnormal retinal development. The optic nerve is formed when retinal ganglion-cell axons exit the globe. The majority of axons, encased in the optic sheath, project to the contralateral side of the brain, forming the optic chiasm in the ventral diencephalon (Rasband et al. 2003). Otx2 is expressed in retinal ganglion cells (Bovolenta et al. 1997), so OTX2 mutations may compromise the development of the optic nerve and the formation of the chiasm.

Pronounced hippocampal malformation was observed (fig. 5C) in patient 6, with bilateral anophthalmia, severe developmental delay, and seizures beginning in infancy. The hippocampus is derived from the alar plate, which develops from an OTX2-expressing domain of the neural plate (Fernandez-Garre et al. 2002). Patient 4A was the other individual in the cohort with severe developmental delay and seizures with teenage onset (table 2). Unfortunately, no MRI information is available for this patient. However, a hippocampal abnormality similar to that in patient 6 was observed in patient 1, a young child with severe developmental and speech delay, like patient 4B, but currently without seizures. The association of hippocampal malformation with seizures remains to be confirmed as additional patients are identified.

It is interesting to note that the ocular defects reported in heterozygous Otx2-knockout mice were usually accompanied by severe craniofacial malformations (Matsuo et al. 1995). Our data show that heterozygous OTX2 mutations in humans are associated with structural eye malformations in the absence of gross craniofacial abnormalities. However, the phenotypic composition of our study cohort leads to biased ascertainment of OTX2 mutations in ocular malformation cases. A broader human mutation study might reveal OTX2 mutations associated with craniofacial malformation syndromes or seizure-associated brain defects in the absence of eye anomalies.

We have presented evidence that heterozygous loss-of-function mutations in the OTX2 gene make a major contribution to the genetic basis of anophthalmia and microphthalmia in a small but significant proportion of patients. Two notable findings of our study are the lack of penetrance of OTX2 mutations and the high incidence of gonosomal mosaicism. The factors that influence de novo mutagenesis in the early embryo are unknown; it is unclear at present whether some genes or genomic regions are more susceptible than others following fertilization or whether gonosomal mosaicism is more common than previously recognized and has simply been underascertained. Of five nucleotide substitutions in our cohort, four are transversions and one is a transition: this unusual bias (Nachman and Crowell 2000) may hint at novel mutational mechanisms. Elucidation of the factors that influence OTX2 penetrance is a high priority and will assist with genetic counseling; in addition, further studies will help to unravel the complex genetic interactions that are responsible for the development of the human eye.