PAX6 molecular analysis and genotype–phenotype correlations in families with aniridia from Australasia and Southeast Asia

Participant recruitment

Twenty-nine individuals from 18 families were enrolled. The study complied with the tenets of the Declaration of Helsinki and the Association for Research in Vision and Ophthalmology (ARVO) statement on human subjects. Informed consent was obtained from participants or their legal guardian in the local language. Institutional board approval was obtained from the Southern Adelaide Clinical Human Research Ethics Committee, and further approval was obtained from local institutions in Cambodia, Bhutan, and Sri Lanka.

Most Australian and New Caledonian cases were part of the Australian and New Zealand Registry of Advanced Glaucoma as described previously [18]. They received a complete ocular examination, including corrected visual acuity, slit-lamp biomicroscopy of the anterior segment, and review of the fundus. In children, when performed, refraction was measured with a retinoscope following cycloplegic dilation with 1% cyclopentolate. Central corneal thickness (CCT) was measured in a minority of patients using an ultrasound pachymeter (Pachymate, DGH Technology, Exton, PA).

Fifteen patients from seven families were enrolled as a result of surveys of schools for the blind in Cambodia [19], Sri Lanka [20], and Bhutan [21] that evaluated the causes of childhood blindness within those communities. Data were collected from school staff, children, relatives, and medical records. A detailed eye examination was conducted by a team of Australian and local optometrists and ophthalmologists as previously described [19–21]. Briefly, the examination included assessment of near and distance visual acuity, the anterior segment was assessed with slit-lamp biomicroscopy, and the posterior segment was evaluated with indirect or direct fundus ophthalmoscopy. Children who had distance vision recorded as better than no perception of light underwent refractive testing and low-vision assessment. Four affected relatives of a family from Bhutan provided samples for DNA analysis but could not be examined.

Genetic analysis

DNA was extracted from whole blood anticoagulated with EDTA using the QIAcube (Qiagen, Doncaster, Australia) automated system and reagents according to the manufacturer’s protocols. Alternatively, for patients either needle phobic or remote from a pathology collection center, including Cambodian, Bhutanese, and Sri Lankan participants, individuals provided a saliva specimen collected in an Oragene® DNA Self-Collection Kit (DNA Genotek Inc., Kanata, Canada), and the DNA was extracted according to the manufacturer’s instructions. The entire coding sequence and intron–exon boundaries of PAX6, encoding the longer transcript NM_001604.5, was amplified with PCR using primer pairs detailed in Appendix 1. Each PCR was performed using 100 ng of purified genomic DNA as a template in a reaction mix containing 1X final concentration of AmpliTaq Gold 360 Master Mix. The following PCR conditions on a Veriti (Life Technologies, Mulgrave, Australia) thermal cycler were used as standard for all templates with both primer pairs: Step 1, 95 °C for 10 min; Step 2, 95 °C for 30 s, 55 °C for 30 s, 72 °C for 1 min, repeated for 40 cycles; and Step 3, 72 °C for 7 min. PCR amplified products were prepared for DNA sequencing with the ExoSAP method using a 10 μl sample of each PCR reaction treated with 5 U of Exonuclease I (Biolabs, Ipswich, MA) and 1 U of Shrimp Alkaline Phosphatase (USB Corporation, Cleveland, OH) to remove residual primers and dNTPs. Bidirectional BigDye® Terminator Cycle Sequencing (Life Technologies) reactions of the appropriate template and PAX6 PCR primer were resolved and base called on an Applied Biosystems 3130XL Genetic Analyzer (Life Technologies).

Detection of the sequence variants was performed using Mutation Surveyor™ v4.0.11 (SoftGenetics LLC, State College, PA). All forward and reverse sequence trace files were assembled by the program against the PAX6 GenBank reference (assembly GRCh37.p13). Software programs Sorting Intolerant From Tolerant (SIFT) and PolyPhen-2 HumVar were used to predict the potential impact of amino acid substitutions on the protein. Conservation among mammalian species was assessed using Homologene. Genome Aggregation Database r2.0.2 (gnomAD) was used to compare allelic frequencies to the population frequencies.

PAX6 was analyzed for copy number variation (CNV) with multiplex ligation-dependent probe amplification (MLPA) using the SALSA MLPA P219-B2 PAX6 probemix (MRC Holland, Amsterdam, the Netherlands) according to the manufacturer’s instructions. The copy number present in the original DNA specimen was determined from the relative amplitude of each amplicon product detected using the ABI 3130xl Genetic Analyzer and the data analyzed using Peak Scanner v2.0 (ThermoFisher Scientific, Scoresby, Australia). Breakpoints of the identified PAX6 deletions were further characterized using single nucleotide polymorphism (SNP) array (Illumina Infinium CytoSNP-850K BeadChip, Illumina PTY, Scoresby, Australia) in some patients. The SNP array was also performed for individual 13 following inconclusive MLPA results. Mosaic changes were detected by assessing for changes in probe intensities along with a shift in the allele frequencies of the SNP probes as previously described [22]. The percentage of mosaicism was estimated by comparing the expected allele frequency under a deletion model to various mosaic levels [22].

Data were analyzed with PASW Statistics, Rel. 18.0.1.2009 (SPSS Inc., Chicago, IL). Data are presented as mean ± standard deviation. The Mann–Whitney U test was used for the assessment of differences in nonparametric data. A p-value ≤ 0.05 was considered significant.

Phenotype

Twenty-nine patients from 18 families were evaluated in this study, comprising 14 male and 15 female patients. The mean age of the whole cohort was 26.4±14.9 years (range 9–58 years), while the mean age of the Australasian cohort was 31.2±15.9 years (range 11–58 years) and the mean age of the Southeast Asian cohort was 19.9±11.3 years (range 9–51 years). The age difference is explained by the fact that patients from Southeast Asia were children recruited from schools for the blind. The phenotype of the 25 examined individuals is described in Table 1. A family history of aniridia was reported in 10/18 probands (56%). Additional family members were available for genetic testing for five of the probands, and the pedigrees are illustrated in Figure 1A.

All individuals had total iris aplasia based on available documentation. Some of the most commonly associated complications are shown in Figure 2. After iris anomalies, the most commonly associated features were foveal hypoplasia (14/18, 78%) and nystagmus (17/19, 90%). Cataract or a history of cataract extraction was identified in 72% (18/25). Cataract extraction was performed in 50% (9/18) with a mean age at surgery of 20.4±13.9 years. Posterior subcapsular or cortical and posterior subcapsular cataracts were the predominant forms of lenticular opacification. Subluxation of the lens was identified in 36% (9/25). Glaucoma was identified in 44% (11/25), and glaucoma filtration surgery or glaucoma drainage device implantation was required by 36% (4/11) with a mean age at glaucoma surgery of 13.0±5.60 years. CCT was obtained in 12 individuals; the mean CCT was high at 645±74.0 μm (range 526–779 μm). Keratopathy was present in 48% (12/25; Figure 2), including one individual with limbal stem cell failure and complete corneal opacification who required corneal transplantation (1/12, 8%). The mean age at keratopathy diagnosis was 19.3±10.5 years (range 9–34 years). A range of refractive errors were identified within the cohort with refraction ranging from −13.00 to +7.30 (excluding individuals 15, 16, and 17 who had refractions of +8.00 to +11.00 secondary to subluxed lenses). All individuals from family 12 were myopic, including two individuals (12a and 12c) with high myopia associated with chorioretinal atrophy affecting the macula. Visual acuity ≤ 1.00 logMAR was recorded in 65% (31/48) of eyes, including 40% (8/20) of eyes from individuals aged ≤20 years compared with 82% (23/28) of eyes from individuals aged >20 years.

The prevalence of keratopathy was statistically significantly higher in the Australasian cohort (11/14, 78.6%) than in the Southeast Asian cohort (1/11, 9.1%, p=0.002). The prevalence of other aniridia-associated features, including foveal hypoplasia, nystagmus, cataracts, lens dislocation, and glaucoma, were not statistically significantly different between the two cohorts.

PAX6 genetic results

Sequence variants in the PAX6 gene were found in 11 of the 18 probands (61.1%) and included seven previously reported and four novel variants. Eight variants were predicted to introduce a premature termination codon, one variant resulted in a C-terminal extension variant, one variant affected the initiation codon, and one was a missense variant (Table 2). Chromosomal deletions were found in an additional six probands (33.3%). No sequence variants or deletions in PAX6 were identified in individual 14. In total, PAX6 accounted for 94.4% (17/18) of the probands with aniridia in this study.

Among the eight variants predicted to result in premature termination codons, five were nonsense variants, two were frameshift variants, and one was a splicing variant (Table 2). Two were novel, p.(Pro81HisfsTer12) and p.(Gln74Ter), whereas the remainder have been previously reported. Variant p.(Pro81HisfsTer12) segregated in two affected family members of individual 1a and was absent from his maternal grandparents. The variant was confirmed to be de novo with molecular analysis (Figure 1B). Splicing variant c.1183+1G>A was predicted to result in the loss of the last exon. One variant was predicted to result in a string of lysine extension at the end of the PAX6 protein (p.(Ter423LeuextTer15)), leading to translation into the 3′ untranslated region (UTR). This variant was present in individual 12a from Cambodia and segregated in four affected children. Variant p.(Met1?) affected the initiation codon, was novel, and was considered deleterious. The removal of the initiation codon is expected to result in an absence of translation or translation starting at a cryptic site downstream and resulting in a truncated protein missing the N-terminal region that includes the paired DNA-binding domain. Finally, the missense variant, p.(Ile29Thr), has not been previously reported in the literature and was not present in the gnomAD populations database. The missense variant was predicted to be deleterious by SIFT and PolyPhen-2, and was highly conserved among species. As a result, this novel variant was regarded as affecting protein function.

The six chromosomal deletions are described in Table 3. MLPA of the PAX6 gene revealed a partial terminal deletion involving exons 14 and 15 of the gene in individual 3a that segregated in the proband’s affected daughter (3b). The WT1 gene is proximal of PAX6 and therefore, was not deleted. The karyotype of individual 7 revealed a deletion of the short arm of chromosome 11 (del11(p13-p15)). Heterozygous deletion of the entire PAX6 gene was confirmed with MLPA. This patient was obese (weight >90^th percentile) and had mild intellectual impairment and immunoglobulin A (IgA) nephropathy in addition to the typical features of aniridia, compatible with a diagnosis of WAGRO syndrome. The SNP array of individuals 16 and 18 revealed a deletion of 795 kb and 503 kb of chromosome 11, respectively, both encompassing PAX6 but not the WT1 gene. The SNP array of individual 13 identified a 480 kb deletion of chromosome 11 that was 100 kb distal of the PAX6 gene. Finally, the SNP array of individual 5 identified a 96 kb interstitial deletion at chromosome 11p13 involving PAX6 but not WT1. The expected probe intensities showed a decrease that was less than the expected decrease for a complete loss of one copy of the region, and the allele frequency differed from what was expected under a deletion model, both suggesting mosaicism. The estimated level of mosaicism for the deleted cell line was 30%. There was no association with the aniridia-associated features or the visual acuity between individuals with deletions in PAX6 compared with those with sequence variants after accounting for age.

Phenotype

Variants in the PAX6 gene have been identified in individuals with aniridia from different populations [14,23-25]. This study is the largest study that includes patients with aniridia from Australia, and the first report of patients with aniridia from New Caledonia, Cambodia, Sri Lanka, and Bhutan. Aniridia is an autosomal dominant disorder with high penetrance but variable expressivity [26]. Abnormalities of the iris are the most common feature of aniridia, and are characterized by either the absence of a visible iris or rudimentary remnant. Foveal hypoplasia is present in the majority of patients [2,23,27], as shown in this study (78%), and results in decreased visual acuity and nystagmus [1]. Progressive loss of visual function often results from complications, such as cataracts, glaucoma, and corneal opacification, that lead to worsened vision with increasing age. In this cohort, vision ≤ 1.00 logMAR defined as severe visual impairment according to the World Health Organization [28] was present in 65% of eyes, and visual acuity progressively deteriorated with age.

Cataract is the most common complication requiring surgery in aniridia, usually observed in 50–90% of patients [1-3,23,27,29,30]. Cataracts were present in 72% of the cohort. Small lens opacities may be present from birth, but they usually do not impair the vision until the teens or early adulthood [1,3]. This is reflected in the mean age of cataract extraction at 20 years in this cohort. Lens dislocation is usually less common [3,23,30] but was present in 36% of the cohort.

The prevalence of glaucoma in aniridia is reported as 6% to 75% [1], although more recent studies suggest a prevalence around 45–65% [23,27,29,30] with an onset commonly in childhood or early adulthood [27,30]. Glaucoma was present in 44% of the cohort, and 36% needed glaucoma surgery at a mean age of 13 years. Glaucoma in aniridia is often difficult to treat and may require multiple interventions to control the intraocular pressure (IOP) to slow progression of the disease [31,32]. The mechanisms causing glaucoma in patients with aniridia are thought to result from developmental defects in the anterior chamber and impair the aqueous humor outflow through the trabecular meshwork and Schlemm’s canal. However, different explanations have been provided regarding the nature of these developmental defects, including progressive angle closure and obstruction of the trabecular meshwork by the residual iris stroma extending anteriorly [33], absence of Schlemm’s canal [1], and abnormalities of the canal’s inner layer or the trabecular meshwork [32]. Heterozygous Pax6-deficient mice show impaired development of the trabecular meshwork and Schlemm’s canal [34,35], and growing adhesions between the cornea and the iris periphery, resulting in complete closure of the iridocorneal angle, increased IOP, and development of glaucoma [35].

The central corneal thickness influences IOP measurements with thicker CCT leading to overestimation of IOP [36]. The mean CCT from individuals with aniridia in this study was high (645±74.0 μm) compared to the mean in healthy Australians (540±33.0 μm) [37] and in healthy Asian populations (504–556 μm) [37]. Other studies previously reported thick mean CCT measurements in children and adults with aniridia [15,27,38,39]. These findings indicate that IOP measurement can be less reliable in patients with aniridia and that examination of the optic nerve is important for detecting and managing patients’ glaucoma.

Aniridia-associated keratopathy is a serious complication developing in 50–90% of patients and has been reported as the most frequent complication by previous studies [23,27,29,40]. This complication often presents in late childhood or early adulthood and less commonly among younger patients [3,29]. In the present cohort, keratopathy was present in 48% of individuals with a mean age at diagnosis of 19 years (range 9–34 years). Early features involve the peripheral epithelium, and aniridia-associated keratopathy leads to progressive opacification and vascularization of the cornea to complete corneal sclerosis and vascularization as the complication progresses centrally [29]. The keratopathy is believed to result from insufficiency of the limbal stem cell population [41]. Additionally, corneal surface problems may be complicated by abnormal tear film and Meibomian gland dysfunction [12]. In this study, keratopathy was more frequent in the Australasian cohort than in the Southeast Asian cohort. Although this may reflect true ethnic phenotypic variability, an alternative explanation could be reduced access to surgical procedures to address aniridia complications such as cataract, lens dislocation, and glaucoma, as surgical procedures are known to accelerate the development of keratopathy [42]. The latter explanation is supported by the finding that only one of the Southeast Asian individuals had a surgical procedure (cataract surgery in one eye) and could account for the lower prevalence of keratopathy in this study compared with previous reports.

Genotype–phenotype correlations

We identified 11 different variants in PAX6, of which four were novel and eight introduced a premature termination codon. A family history of aniridia was present in ten probands. A de novo inheritance was confirmed molecularly in one family. De novo variants in PAX6 have previously been reported although the prevalence is not known [2,43]. Similar to other studies [13-15], we did not find a statistically significant association between the phenotype of individuals with variants resulting in premature termination codons compared with individuals with deletions in PAX6. The absence of phenotypic difference is consistent with the fact that truncated proteins are subject to nonsense-mediated decay [13], resulting in PAX6 haploinsufficiency in a similar manner to deletions in PAX6. Interestingly, nonsense suppression drugs have been shown to alter PAX6 dosage, inhibit disease progression, and reverse some ocular phenotype in PAX6-transgenic mice [44]. This therapeutic approach is promising for PAX6-associated aniridia because the majority of cases are explained by variants resulting in premature termination codons.

Although variants and gene deletions resulting in PAX6 haploinsufficiency do not present genotype–phenotype correlations, other types of PAX6 variants display associations. An extension variant, p.(Ter423LeuextTer15), resulting in a longer open reading frame and translation into the 3′ UTR, was identified in five affected individuals from the same family (individuals 12a–e). This variant has been reported among different populations with aniridia [14,15,25]. Partial aniridia and milder phenotypes have also been associated with PAX6 variants in the proline-threonine-serine-rich (PST) region [26]. However, none of the patients in this study with variants in the PST region had partial aniridia. Hingorani et al. reported that glaucoma was unusual and the iris anomalies were milder in individuals with extension variants in PAX6 compared to premature termination codon variants [2]. Interestingly, none of the five individuals carrying an extension variant in this study had glaucoma despite their age range (12–51 years) falling within the range of typical glaucoma onset. All individuals from this family had myopia, including three with high myopia (<–6.0 diopters) and two with chorioretinal atrophy consistent with degenerative myopia. Other studies have also noted an association between PAX6-associated aniridia and high myopia [2,45]. Interestingly, myopia has been reported in other individuals with a similar C-terminal extension variant as family 12 [2]. A genome-wide association study previously identified significant linkage to myopia on chromosome 11p13 in Caucasians, with strong evidence of linkage, but not association, for genetic markers covering the PAX6 gene [46]. Ng et al. identified two dinucleotide repeats in the PAX6 P1 promoter that increased transcriptional activity of PAX6 and were associated with high myopia in a Chinese population [47]. Additional studies showed conflicting results for the association of PAX6 polymorphisms with myopia, and a recent meta-analysis concluded that PAX6 may confer only a small effect, if any, to myopia development [48]. Further studies are needed to clarify the potential involvement of PAX6 in myopia.

One novel variant, p.(Met1?), was predicted to abolish the initiation codon and was considered deleterious. Although this variant has not been previously reported, other variants affecting the initiation codon of PAX6 have been described with aniridia [24,25,49]. In the present study, individual 15 carrying variant p.(Met1?) had total aniridia, cataracts, and glaucoma. The phenotype associated with the different variants affecting the initiation codon does not seem to differ from variants resulting in premature termination codons [24,25,49].

One proband carried a missense variant predicted to be deleterious in exon 5, p.(Ile29Thr)). Missense variants associated with aniridia occur more often in exons 5–6 that encode the paired domain important for DNA binding [13]. A previous study showed that missense variants can affect protein stability or impair DNA-binding ability [50]. More recently, Xie et al. demonstrated that missense variants can affect the ability of the protein to activate its target genes [51]. Two other variants affecting the same codon, p.(Ile29Val) and p.(Ile29Ser), have been reported in patients with aniridia in the Human PAX6 Allelic Variant Database. Missense variants have been described in individuals with a range of ocular phenotypes from aniridia to milder non-aniridia phenotypes, including optic nerve anomalies, Peters’ anomaly, corectopia, and foveal hypoplasia [13]. Missense variants have been associated with better visual acuity due to the lower rate of associated features [2]. In this study, individual 6 carrying the p.(Ile29Thr) novel variant had visual acuities better than 1.00 logMAR, but he presented total aniridia, glaucoma, cataracts, foveal hypoplasia, and nystagmus, which does not suggest a milder phenotype.

We identified five deletions of the PAX6 gene using MLPA, one partial and four total deletions, as well as one deletion distal of PAX6. The partial PAX6 deletion involved the last two exons of the gene, resulting in a truncated protein. Three total PAX6 deletions did not encompass WT1, therefore, confirming a diagnosis of isolated aniridia. The other total deletion was part of a larger chromosomal deletion on chromosome 11p (p13-p15) as shown by the karyotype. The phenotype of the patient (individual 7) included aniridia, obesity, mild intellectual impairment, and IgA nephropathy, which is consistent with a diagnosis of WAGRO syndrome. The karyotypes of both parents were normal suggesting a de novo occurrence of the chromosomal deletion.

We identified one deletion in mosaic in 30% cells from blood. Mosaic PAX6 deletions have been associated with milder phenotypes in some patients with low mosaicism: a 10 Mb deletion in 28% cells was reported in an individual with coloboma [52] and a 5.28 Mb deletion in 45% cells was identified in an individual with mild WAGR syndrome [53]. In comparison, bilateral aniridia and optic atrophy were present in two individuals with a 200 kb deletion in 50% and 60% cells [54]. In this study, individual 5 had bilateral total aniridia with a deletion identified in 30% cells. However, the phenotype associated with mosaic PAX6 deletion is determined by the ocular tissue whereas the deletion was detected in cells from blood. Therefore, it is possible that the deletion is present in a greater proportion in ocular tissues, accounting for the more severe phenotype.

A deletion distal of PAX6 was identified in one patient (individual 13). Deletions downstream of PAX6 that leave the PAX6 gene intact have been reported in some patients with aniridia [17,55]. These deletions abolish PAX6 expression and cause aniridia in patients due to the loss of an evolutionary conserved regulatory region containing tissue specific enhancers that are essential for PAX6 expression [17].

The detection rate is similar to other studies that reported intragenic variants and chromosomal rearrangements in approximately 90% of patients with aniridia [14,15]. No PAX6 variant or copy number variation could be demonstrated in one of the 18 probands (5.6%) with aniridia (individual 14 from Sri Lanka). His phenotype was not different from the other probands in the cohort. He had a visual acuity of hand motion in the right eye due to cataract and a subluxed lens, and had no perception of light in his left eye due to complication of cataract surgery resulting in a phthisical eye. His father was reported as having aniridia but was not available for examination. A single nucleotide substitution located in PAX6 regulatory elements has been previously reported in an individual with aniridia [56]. Therefore, it is possible that deletions or variants distal of PAX6, or variants in other genes, may account for the aniridia in the unsolved case.

In conclusion, we reported a detection rate of 94% for variants and deletions in PAX6 in individuals with aniridia from Australasia and Southeast Asia. To the best of our knowledge, this study is the first description of aniridia and PAX6 variations in individuals from Cambodia, Sri Lanka, Bhutan, and New Caledonia. These results support the observation of marked allelic heterogeneity and high prevalence of loss of function variants within PAX6-related aniridia. Molecular diagnosis can assist in the clinical management of affected patients and may be essential to future clinical trials.