Expanding the Spectrum of FOXC1 and PITX2 Mutations and Copy Number Changes in Patients with Anterior Segment Malformations

Abstract

Purpose.: Anterior segment dysgenesis (ASD) comprises a heterogeneous group of developmental abnormalities that affect several structures of the anterior segment of the eye. The main purpose of this study was to assess the proportion of FOXC1 and PITX2 mutations and copy number changes in 80 probands with ASD.

Methods.: The patients were examined for FOXC1 and PITX2 copy number changes and mutations using MLPA (multiplex ligation-dependent probe amplification) and direct sequencing. Subsequently, the identified copy number changes were fine-mapped using high-resolution microarrays. In the remaining mutation-negative patients, sequencing of the FOXC1 andPITX2 3′ untranslated regions (UTRs) and three other candidate genes (P32, PDP2, and FOXC2) was performed.

Results.: Thirteen FOXC1 and eight PITX2 mutations were identified, accounting for 26% (21/80) of the cases. In addition, six FOXC1 and five PITX2 deletions were found, explaining 14% (11/80) of the cases. The smallest FOXC1 and PITX2 deletions were 5.4 and 1.6 kb in size, respectively. Six patients carrying FOXC1 deletions presented with variable extraocular phenotypic features such as hearing defects (in 4/6) and mental retardation (in 2/6). No further genetic defects were found in the remaining mutation-negative patients.

Conclusions.: FOXC1 and PITX2 genetic defects explain 40% of our large ASD cohort. The current spectrum of intragenic FOXC1 and PITX2 mutations was extended considerably, the identified copy number changes were fine mapped, the smallest FOXC1 and PITX2 deletions reported so far were identified, and the need for dedicated copy number screening of the FOXC1 and PITX2 genomic landscape was emphasized. This study is unique in that sequence and copy number changes were screened simultaneously in both genes.

Anterior segment dysgenesis (ASD) comprises a heterogeneous group of developmental abnormalities affecting several structures of the anterior segment of the eye. There are numerous classifications, largely based on clinical features, that generally fail to take into account either the embryologic derivatives of the affected structures or the underlying genetic defect. ¹ Such clinical classification systems are hampered by the variability in clinical manifestations and by overlapping features between the subgroups. ^{1 –3} As the underlying molecular genetics of the different subgroups are being unraveled, it becomes even more difficult. 

Axenfeld-Rieger malformations (ARMs) represent a subgroup of ASD and refer to a group of autosomal dominant inherited disorders, affecting structures in the anterior eye segment derived from the neural crest. Apart from the classic ocular features, such as abnormal angle tissue, a hypoplastic or malformed iris, multiple pupils (polycoria), an elongated pupil (corectopia), and/or a posterior embryotoxon, ARM is often associated with systemic defects including dental, facial, and/or periumbilical abnormalities. ^1,4 The major disease burden, however, brings with it the high risk of glaucoma, which develops in approximately 50% of ARM patients, potentially leading to visual loss or even blindness. ^1,4 ARM is genetically heterogeneous and has been associated with mutations in at least two transcription factor–encoding genes (FOXC1, PITX2) and with two known additional genetic loci (13q14, 16q24). ^{5 –8} The association between ARM and PAX6 in a patient with Rieger syndrome has been revised recently and appears to be incorrect. ^9,10 It has been postulated that disease-causing mutations and copy number changes in FOXC1 and PITX2 account for less than 40% of the patients with ARM. However, this estimation should be interpreted with care, as it is based on studies involving small cohorts. Moreover, most studies did not include copy number analysis, which may have led to an underestimation of the contribution of FOXC1 and PITX2 in the molecular pathogenesis of ARM (reviewed in Ref. 4 and references therein). To date, 46 intragenic FOXC1 and 41 intragenic PITX2 mutations have been described, reflecting substantial allelic heterogeneity. ⁴ In addition, several chromosomal rearrangements and copy number changes have been described in the FOXC1 and PITX2 genomic landscape (Ref. 11 and references therein and Refs. 12 –15). Both FOXC1 gene deletions and duplications have been associated with ARM, underlining the need for a correct gene dosage for normal development of the anterior eye segment. ^16,17  

In 2006, FOXC1 and PITX2 were shown to be co-expressed in the periocular mesenchyme. ¹⁸ Moreover, biochemical and subcellular localization assays demonstrated that these proteins physically interact with each other in a common complex on the chromatin. Apparently, PITX2 negatively regulates the functional activity of FOXC1, and, furthermore, functional assays have shown that PITX2 loss-of-function mutants lose their ability to inhibit FOXC1. ¹⁸ Additional studies designed to identify and define other members of the PITX2 and FOXC1 molecular pathways are critical in shaping our understanding of the molecular pathogenesis of ARM and in identifying novel candidate genes. Several co-expressed genes, direct protein interaction partners, upstream transcriptional regulators, and downstream transcriptional targets have been identified, such as P32, PDP2, and FOXC2. ^{19 –24} The splicing factor SF2-associated protein (P32), for example, was isolated as a putative FOXC1-interacting protein by using a human trabecular meshwork (HTM) yeast two-hybrid (Y2H) cDNA library assay. ²³ Huang et al. ²³ demonstrated that this multicompartmental and multifunctional protein inhibits FOXC1 mediated transcriptional activation in a dose-dependent manner. Moreover, an impaired interaction was noted between the mutant FOXC1 protein p.Phe112Ser and P32. The pyruvate dehydrogenase phosphatase 2 gene (PDP2), in turn, was identified as a direct downstream target of PITX2 in a hormone-inducible PITX2 expression system in nonpigmented ciliary body cells (NPCE) (Strungaru MH et al. IOVS 2007;48:ARVO E-Abstract 3212). PDP2 encodes an enzyme from the pyruvate dehydrogenase complex (PDC), which is implicated in pyruvate decarboxylation within the mitochondria. FOXC2 is also thought to be involved in the FOXC1 pathway, as FOXC1 and FOXC2 show coordinated function and are co-expressed in the mesenchyme from which the ocular drainage structures are derived. ²⁴  

The primary purpose of this study was to assess the prevalence of disease-causing FOXC1 and PITX2 mutations and copy number changes in a large cohort of 80 probands with ASD, mainly of Belgian-Dutch origin. Moreover, we sought to further characterize the identified copy number changes using high-resolution targeted microarrays, to assess the potential correlation between the extent of the deletions and disease severity. 

In patients without coding FOXC1/PITX2 intragenic mutations or copy number changes we sequenced the 3′UTRs, as these may harbor important regulatory pathogenic or modifying variants. In addition, genes encoding proteins involved in the same pathways as FOXC1 and PITX2, were considered to be plausible candidate genes in the remaining patients. To this end, we explored the putative involvement of three such candidate genes, P32, PDP2, and FOXC2, in the molecular pathogenesis of ASD. 

Methods

Patients

The cohort consists of 80 diagnostic referrals for ARM derived from different ophthalmology clinics and genetics centers. More than 60% of the cohort originates from Belgium or the Netherlands. The study complied with the guidelines set forth in the Declaration of Helsinki. 

The clinical records were reinspected for those patients in whom we identified a molecular defect involving FOXC1 or PITX2. Of the 32 mutation-positive patients 12 (37.5%) were examined locally. 

Direct Sequencing of PITX2a, FOXC1, P32, FOXC2, and PDP2

Primers were designed or selected from the literature to amplify the coding exons and intronic splice site junctions of PITX2a (NM_153427.1), FOXC1 (NM_001453.2), P32 (NM_001212.3), PDP2 (NM_020786.1), and FOXC2 (NM_005251.2) (primers available on request). PITX2 consists of seven exons and can give rise to four alternative transcripts (PITX2A, B, C, and D). ²⁵ We sequenced the coding region of PITX2A. In addition, primers were designed to generate overlapping amplicons covering the 3′UTR of FOXC1 and PITX2 (primers available on request). PCR products were sequenced in both directions with dye-termination chemistry (BigDye ver. 3.1 terminator cycle sequencing kit; Applied Biosystems, Inc. [ABI], Foster City, CA). Sequencing reactions were loaded on a genetic analyzer (model 3100 or 3730 with Sequencing Analysis ver. 3.7 and SeqScape ver. 1.1; ABI). PITX2a and FOXC1 were sequenced in all patients. The amplicons in which we identified amino acid–altering variants were sequenced in 100 control samples to evaluate their significance. The 3′UTRs were sequenced in 40 patients negative for coding FOXC1/PITX2 mutations or copy number changes. The remaining patients were not screened for the 3′UTRs because of a lack of sufficient DNA. Downstream screening of P32, PDP2, and FOXC2 was performed for a pilot group of 31 probands without identifiable FOXC1/PITX2 changes. 

Sequence Variation Evaluation and Nomenclature of Mutations

The presence of all mutations was confirmed on a second PCR product. Segregation analysis of disease alleles was performed if possible. Thorough bioinformatic evaluation of novel variants was performed (Alamut software, ver.1.5; Interactive Tiosoftware, Rouen, France). Alamut provides for each variant the HGVS (Human Genome Variation Society, Genomic Disorders Research Centre, Carlton South, VIC Australia) nomenclature. For missense variants, it calculates the Grantham distance and automatically fills in the queries for PolyPhen and SIFT prediction servers, based on the UniProt ²⁶ protein identifiers and FASTA sequences of several orthologs, respectively (Polyphen, http://genetics.bwh.harvard.edu/pph/ provided in the public domain by the Division of Genetics, Department of Medicine, Brigham and Women's Hospital, Harvard Medical School, Boston, MA, and the Bork Group, EMBL, Heidelberg, Germany; SIFT [sorting tolerant from intolerant], http://www.fhcrc.org/ provided in the public domain by the Fred Hutchinson Cancer Research Center, Seattle, WA; FASTA, http://www.ebi.ac.uk/tools/fasta/ provided in the public domain by the European Bioinformatics Institute, European Molecular Biology Laboratory, Heidelberg, Germany). Variants with a major impact on protein level—for example, through the introduction of stop codons, frame shifts or large deletions, duplications or insertions—were classified as pathogenic. Variants within the 3′UTR that were not registered as a known SNP in the Ensembl Genome Browser ²⁷ were classified as putative pathogenic. The numbering of DNA sequence variants was performed according to the HGVS Mutation Nomenclature Recommendations (http://www.hgvs.org/mutnomen/recs.html/ provided in the public domain by the Human Genome Variation Society). The FOXC1 mutations are described according to RefSeq: NM_001453.2. The PITX2 mutations are described according to the PITX2 transcript variant 1 (RefSeq: NM_153427.1) (www.ncbi.nlm.nih.gov/locuslink/refseq/ provided in the public domain by the National Center for Biotechnology Information, Bethesda, MD). 

Multiplex Ligation-Dependent Probe Amplification

MLPA permits relative quantification of changes in copy number of specific genomic regions. ²⁸ The P054 probe mix includes probes for the TWIST1, FOXL2, FOXC1, FOXC2, ATR, PITX2, and OA1 genes. More specifically, it contains three probes for PITX2, two probes for FOXC1, and two probes for FOXC2. MLPA was performed in accordance with the manufacturer's instructions (MRC Holland; Amsterdam, The Netherlands). 

Copy Number Analysis Using 250K SNP Arrays

Three PITX2 deletions identified using MLPA were subsequently delineated using 250K NspI SNP arrays (Affymetrix, Santa Clara, CA). Processing of the samples, quality control, and SNP copy number assessment was conducted by an Affymetrix Service Provider (CNAT 4.0 software; DNA Vision, Charleroi, Belgium). The results were subsequently visualized in arrayCGHbase. ²⁹  

Copy Number Analysis with Targeted High-Resolution Arrays

During the study, it became possible to design a custom-targeted 60K oligonucleotide (Agilent Technologies, Palo Alto, CA) targeting a region of 5 Mb around FOXC1 (chr6:1–5000000; UCSC, Human Genome Browser, March 2006/ provided by the University of California Santa Cruz at http://genome.ucsc.edu) and a region of 5 Mb around PITX2 (chr4:109258234-114257994; UCSC, Human Genome Browser, March 2006). Using this array, we characterized four FOXC1 deletions, two partial FOXC1 deletions, one PITX2 deletion, and one partial PITX2 deletion, originally identified through MLPA. 

Hybridizations were performed according to the manufacturer's instructions, with minor modifications. ³⁰ The results were subsequently visualized in arrayCGHbase. ²⁹  

Results

FOXC1 and PITX2 Screening

Molecular screening of FOXC1 and PITX2 in 80 probands with ASD resulted in the identification of mutations in 32 (40%) of 80 patients. In particular, an FOXC1 mutation/copy number change was detected in 19 (24%) patients, and a PITX2 mutation/copy number change was detected in 13 (16%) patients. The mutations are listed in Table 1 with novel mutations represented in bold. Copy number changes are listed in Table 2. 

Table 1.

 

Summary of Identified FOXC1 and PITX2 Mutations with Associated Phenotypes

Mutation (cDNA)	Mutation (Protein)	Protein Domain	YOB; Sex	Type; Segregation	Ocular Features	Extraocular Features
FOXC1
c.286dupG	p.Asp96GlyfsX210	FHD	1973; F	S; Mother − Father/	Bilateral glaucoma, bilateral posterior embryotoxon and angle abnormalities, cataract and corneal opacities in right eye	None reported
c.316C>T	p.Gln106X	FHD	1985; M	S; Mother/Father/	Enucleation of the right eye due to severe glaucoma. Left eye: posterior embryotoxon, hazy peripheral cornea, limited ectropion uveae at pupillary border, limited iris hypop lasia, anterior synechiae	None reported
c.325A>G	p.Met109Val	FHD	1978; M	?; Mother − Father +	Unilateral corectopia, blue sclerae	Maxillary hypoplasia, dental anomalies, hearing loss, learning difficulties
c.335del	p.Phe112SerfsX69	FHD	1983; F	F; Mother + Father/	Bilateral glaucoma and buphthalmos, Rieger anomaly	Mild dental anomalies, unilateral hearing impairment
c.392C>A	p.Ser131X	FHD	1978; M	F; Mother + Father − Brother −	Bilateral glaucoma, left eye enucleated	Dental anomalies, epilepsy
c.392C>G	p.Ser131Trp	FHD	1984; F	?; Mother/Father/	Iris hypoplasia, Rieger anomaly	Congenital scoliosis, dental anomalies
c.412A>G	p.Lys138Glu	FHD	1975; M	?; Mother/Father/	Normal tension glaucoma, clear cornea, clear lens, posterior embryotoxon	None reported
c.605delC	p.Pro202ArgfsX113		2001; F	F; Mother + Father − Brother +	Unilateral glaucoma, unilateral Haab's striae, (mild iris hypoplasia), unilateral posterior embryotoxon	Maxillary hypoplasia anomaly, hypertelorism
c.692_696del	p.Gly231ValfsX73	ID	1995; F	S; Mother − Father +	Glaucoma, Axenfeld-Rieger anomaly, no other data obtained	No data obtained
c.780dup	p.Asp261ArgfsX45	ID	?; M	F; Mother + Father/	No data obtained	No data obtained
c.816_817delinsG	p.Ser272ArgfsX43	ID	1979; M	S; Mother − Father −	No data obtained	No data obtained
c.980_981del	p.Glu327AlafsX200	ID	1977; F	?; Mother/Father/	No data obtained	No data obtained
c.1491C>G	p.Tyr497X	AD2	2006; M	F; Mother/Father +	Glaucoma, no other data obtained	Hypertelorism
PITX2
c.137_138del	p.Phe46TyrfsX152	HD	2002; M	S; Mother − Father/	Unilateral glaucoma, bilateral posterior embryotoxon, bilateral polycoria, corectopia, anterior synechiae, bilateral angle abnormalities	Maxillary hypoplasi a, umbilical anomaly, dental anomalies
c.172T>C	p.Phe58Leu	HD	1994; M	S; Mother − Father −	Bilateral glaucoma, irregular pupil, anterior synechiae	Maxillary hypoplasia, umbilical anomaly, dental anomalies, tachycardia, hyperextension of elbow and finger joints, history of multiple sprained ankles, fifth finger clinodactyly and brachymesophalangia
c.175C>T	p.Gln59X	HD	?; M	S; Mother − Father −	Rieger syndrome, no other data obtained	No data obtained
c.224G>A	p.Trp75X	HD	1999; M	S; Mother − Father −	Unilateral glaucoma, bilateral posterior embryotoxon, bilateral corectopia	Maxillary hypoplasia, umbilical anomaly, dental anomalies, bilateral retention of th e testis
c.253–11A>G	/	/	1998; M	F; Mother + Father/Sister: +	Unilateral posterior embryotoxon, bilateral corectopia, pigmentation on anterior lens capsule	Maxillary hypoplasia, umbilical anomaly, dental anomalies, short stature
c.282G>A	p.Trp94X	HD	1994; M	F; Mother + Father − Sister: +	Severe iridocorneal dysgenesis with corneal endothelium decompensation and edema, unilateral corectopia	Maxillary hypoplasia, dental anomalies, umbilicus not examined, hear ing loss, hypertelorism, speech and language delay, motor retardation, short palatum
c.301C>T	p.Gln101X	ID1	2003; F	S; Mother − Father −	No glaucoma, Right eye: leucoma and eccentric irregular pupil. Left eye: transparent cornea, polycoria	Maxillary hypoplasia, umbilical anomaly, dental anomal ies
c.304C>T	p.Gln102X	ID1	1981; F	S; Mother/Father/	No data obtained	Redundant umbilical skin, dental anomalies

 Novel mutations are shown in bold. YOB, year of birth; F, familial; S, sporadic; +, positive for the mutation; −, negative for the mutation;/, not available for testing.

Table 2.

 

Summary of Identified FOXC1 and PITX2 Deletions with Associated Phenotype types

Deletion	Max Size; Position (chr6)*	YOB; Sex	Type; Segregation	Ocular Features	Extraocular Features
FOXC1
FOXC1_1	∼5.4 kb; 1553924–1559364	2003; M	S; Mother − Father −	No glaucoma, megalocornea, bilateral posterior embryotoxon	Hypertelorism, maxillary hypoplasia, no cardiac anomaly, no growth retardation, no umbilic al anomaly, no dental anomalies
FOXC1_2	∼34 kb; 1551415–1585522	1973; F	F; Mother + Father/Children: −	Right eye: glaucoma, posterior embryotoxon, megalocornea, Haab's striae, corectopia, iris strands, ectropion uveae	Hypertelorism, maxillary hypop lasia, progressive hearing loss, no umbilical anomaly, no dental anomalies, normal intelligence
				Left eye: glaucoma, megalocornea, Haab's striae, corectopia, iris strands, polycoria, nuclear cataract
FOXC1_3	∼84 kb; 1552945–1636775	1982; F	F; Mother − Father −	Bilateral glaucoma, posterior embryotoxon, atrophic iris	No data obtained
FOXC1_4	∼2.6 Mb; 0–2646377	1966; M	F; Mother/Father/Brother +	Glaucoma, Descemet membrane ruptures, limited peripheral cornea nebulae	Maxillary hypoplasia, hearing loss due to middle ear malformations, tooth extraction required because of maxillary dental crowding with normal number of teeth, no cardiac anomaly, no umbilical anomaly, normal intelligence
				Right eye: corticonuclear cataract, iris hypoplasia, upward pupillary displacement, anterior synechiae
				Left eye: discrete anterior subcapsular cataract, iris hypoplasia, polycoria
FOXC1_5	∼3.4 Mb; 566884–3960186	1978; F	F; Mother/Father/	Right eye: glaucoma, partial posterior embryotoxon	Mild mental retardation, middle ear hearing loss, eczema, no growth retardation, no umbilical anomaly
				Left eye: glaucoma, partial posterior embryotoxon, anterior synechiae in area of cystic schisis of inferior iris
FOXC1_6	∼4.7 Mb; 0–4749872	1994; F	S; Mother + Father/Brother −	Bilateral irregular pupil, iridocorneal adhesions, prominent Schwalbe's line	Minor maxillary hypoplasia, no cardiac anomaly, no growth retardation, no umbilical anomaly, dental anomalies, small mouth, small nose, chronic glue ear and hearing loss, hyperlax long fingers, IQ 56 with developmental and speech delay, premature pubarche and axillarche
PITX2
PITX2_1	1.6 kb; 111760308–111761945	2006; M	S; Mother − Father −	Unilateral microcornea	Maxillary hypoplasia, umbilical anomaly, dental anomalies, no cardiac anomalies, no growth retardation
PITX2_2	286 kb; 111648252–111934227	2008; F	S; Mother − Father −	Right eye: mild atrophic iris	Maxillary hypoplasia, umbilical anomaly, dental anomalies, normal intelligence, no cardiac/renal anomalies, no growth ret ardation
				Left eye: corectopia, atrophic iris, iridocorneal adhesions
PITX2_3	∼1.1 Mb; 111161726–112223083	?; M	F; Mother/Father/	Bilateral glaucoma, bilateral posterior embryotoxon, unilateral polycoria, bilateral corectopia	Maxillary hypoplasia, umbilical anomaly, dental anomalies, normal intelligence, no cardiac anomalies, no growth retardation
PITX2_4	∼2.5 Mb; 110200973–112725989	1960; M	S; Mother/Father/	Severe bilateral glaucoma, anterior segment anomalies.	Umbilical anomaly, dental anomalies, normal intelligence
				No other data
PITX2_5	∼2.8 Mb; 110322934–113076304	1974; M	S; Mother/Father/	No glaucoma, asymmetric ocular anomalies	Umbilical anomaly, dental anomalies, no cardiac anomaly, intelligence not specified
				Right eye: posterior embryotoxon, dyscoria, Fuchs endothelial dystrophy
				Left eye: polycoria, corectopia, anterior synechiae

 YOB, year of birth; F, familial; S, sporadic; +, positive for the mutation; −, negative for the mutation;/, not available for testing.

*  Chromosomal position according to the UCSC, Human Genome Browser, March 2006.

In FOXC1 we identified 13 mutations in 13 unrelated probands: 3 nonsense, 7 frameshift, and 3 missense mutations (Table 1, Fig. 1). To the best of our knowledge, 12 of 13 are novel. In general, nonsense and frameshift mutations were considered to be pathogenic. The effect of three missense mutations (p.Met109Val, p.Ser131Trp, and p.Lys138Glu) on protein function was more difficult to assess, and computational programs were used to predict their pathogenic character (Supplementary Material S1). All three mutations are located within the highly conserved DNA-binding forkhead domain (Supplementary Material S2) and probably affect the DNA-binding capacity. Although the computational predictions sometimes contradict each other, all three substitutions were regarded as pathogenic because of their location in the highly conserved forkhead domain (Supplementary Material S2), the overall outcome of the different programs and the Grantham score (Supplementary Material S1), and their absence in the 100 control samples. Sequence analysis of the FOXC1 3′UTR of 40 patients in whom no pathogenic mutation was identified in the FOXC1 coding region, led to the identification of several registered SNPs, but no putative pathogenic variants were found. Furthermore, we identified two partial and four total FOXC1 gene deletions with MLPA, which were further characterized by targeted microarray (Figs. 2, 3; Table 2). There was a scattered location of the breakpoint regions, and the deletions appeared to be at maximum 4.7 Mb, 3.4 Mb, 2.6 Mb, 84 kb, 34 kb, and 5.4 kb in size. 

Figure 1.

 

Overview of all known and novel FOXC1 mutations. Shown (drawn to scale) is the FOXC1 coding region, with all known (black lettering) and novel (orange lettering) FOXC1 mutations. Red: The DNA-binding FHD. Gray: The 5′ and 3′ UTRs (not drawn to scale).

Figure 2.

 

Overview of delineated FOXC1 deletions. The FOXC1 region with custom tracks showing the extent of the molecularly defined deletions reported by Chanda et al. ¹² and identified in the current study. Horizontal red bars: locations and sizes of deletions; horizontal green bars: locations and sizes of duplications. (Drawn according to the UCSC, Human Genome Browser, March 2006; University of California Santa Cruz.)

Figure 3.

 

Copy number analysis by targeted arrays (Agilent; Palo Alto, CA). Blue box: array CGH profiles for the FOXC1 region (chromosome [chr]6:0-6,000,000) showing five samples with FOXC1 deletions. Blue vertical line: location of FOXC1. Green box: profiles for the PITX2 region showing two samples with PITX2 deletions. The location of PITX2 are marked in the first (green vertical line) and second (green rectangle) profiles. Colored dots: log₂ ratios of individual oligonucleotides (red: deleted; black: normal; green: duplicated).

Within PITX2, we identified eight different mutations: five nonsense, one frameshift, one missense, and one splice site (Table 1; Fig. 4). Overall, six of eight mutations are novel. As for FOXC1, the nonsense and frameshift mutations identified in PITX2 are believed to have a major effect on the protein function and, hence, were considered pathogenic. The missense mutation p.Phe58Leu is located within the evolutionarily conserved DNA-binding homeodomain (Supplementary Material S2) and Polyphen and SIFT analysis assign this substitution as “possibly pathogenic” and “not tolerated,” respectively (Supplementary Material S1). Although the Grantham score suggests a minor impact, the overall results are in favor of a pathogenic character. Moreover, this variant is de novo, as it was not found in the DNA from both unaffected parents, and it was absent in 100 control samples. Sequence analysis of the PITX2 3′UTR revealed several known SNPs and one putative pathogenic variant, which was not registered as a known SNP, in a patient of Caucasian origin. However, the de novo character of this latter variant could not be assessed because of the lack of parental DNA. Furthermore, one partial and four total PITX2 deletions were identified by MLPA. The partial deletion spanned one of four MLPA probes and appeared to be de novo. The PITX2 deletions were further delineated with 250K SNP arrays (Affymetrix) or targeted 60K arrays (Agilent). This method enabled characterization of the extent of all five deletions (Fig. 5). The breakpoints were scattered and the maximum size of the deletions were 2.5 Mb, 2.8 Mb, 1.1 Mb, 286 kb, and 1.6 kb (Figs. 3, 5; Table 2). 

Figure 4.

 

Overview of all known and novel PITX2 mutations. A scale drawing of the PITX2 coding region, with all known (black lettering) and all novel (orange lettering) PITX2 mutations. Red: the DNA-binding homeodomain (HD). Gray: 5′ and 3′ UTRs (not drawn to scale). The exonic regions are represented by boxes and the separating intronic regions (not drawn to scale) are depicted by horizontal lines between two subsequent boxes.

Figure 5.

 

Overview of delineated PITX2 deletions. The PITX2 region (4q25–q26) with custom tracks showing the extent of all molecularly defined deletions reported in the literature ^{13 –15} and in the present study. Horizontal red bars: locations and sizes of deletions. Red: minimally deleted regions; pink: regions harboring the breakpoints. (Drawn according to the UCSC, Human Genome Browser, March 2006; University of California Santa Cruz.)

The corresponding ocular and extraocular phenotypes of patients with established genotypes are provided in detail in Tables 1 and 2. 

Sequencing Analysis of Selected Candidate Genes

Downstream screening of a pilot group of 31 probands without identifiable FOXC1/PITX2 changes consisted of sequencing three candidate genes P32, PDP2, and FOXC2. Apart from known SNPs, we identified one sequence variant within PDP2 (c.1083G>A, p.Arg361His). However, this variant was not classified as pathogenic, based on several in silico predictions. Polyphen predicted this variant to be benign, and according to SIFT predictions, the protein substitution is tolerated. In addition, the Grantham distance of this substitution was only 29. Because of the lack of pathogenic sequence variants in this pilot group of patients, we refrained from screening additional patients. 

Discussion

Molecular screening of FOXC1 and PITX2 in a large cohort (80 probands with ASD) led to the identification of the underlying molecular defect in 40% of the patients. This study is unique in that we simultaneously screened for sequence changes (coding region and 3′UTRs) and copy number changes in both genes. Twelve novel FOXC1 and six novel PITX2 mutations extend the spectrum of 46 and 41 reported intragenic FOXC1 and PITX2 mutations, respectively. Of note, most of the reported mutations in these two genes are unique, emphasizing their allelic heterogeneity. 

The single-exon gene FOXC1 is located on chromosome band 6p25 and encodes a forkhead transcription factor. It is a member of a gene family that is characterized by an evolutionarily conserved 110 amino acid (AA), DNA-binding forkhead domain (FHD). Apart from this FHD, FOXC1 contains a terminal protein sequence of unknown function that is conserved up to Xenopus. ² We identified six novel mutations within and six novel mutations downstream of the FHD. When all known and novel FOXC1 mutations were taken into account, missense mutations were the most common (45%), whereas frameshift mutations were overrepresented outside the FHD. 

The PITX2 gene, located on 4p25, encodes a member of the bicoid class of homeodomain (HD) proteins and was the first gene to be associated with ARM. ⁷ This gene consists of seven exons and can give rise to four alternative transcripts (PITX2A, B, C, and D). ²⁵ The PITX2A protein, which is expressed in the developing eye, can be subdivided into five regions; two activation domains (AD), one homeodomain (HD), and two inhibitory domains (ID). ³¹ In general, frameshift and nonsense mutations in PITX2 are scattered along the coding region, whereas most missense mutations are restricted to the DNA-binding HD. We identified five mutations that are presumed to affect the DNA-binding capacity of the HD severely. Furthermore, we identified two nonsense mutations within the ID1. The latter mutations left the HD intact, but completely ablated ID1, ID2, and AD2. Taking into account all known and novel PITX2 mutations, missense mutations appeared to be the most prominent (45%). It is noteworthy that more than half of the mutations were located within the HD (62%). 

Figure 6.

 

Anterior segment photographs from patients with FOXC1 genetic defects. (A–C) Ocular images from a patient carrying a partial FOXC1 deletion (FOXC1_2, Table 2). (A) Ectopic pupil and posterior embryotoxon in a right eye. (B) Iris strands in a right eye. (C) Peripheral polycoria and ectropion uveae in a left eye. (D) Anterior synechiae and iris cyst in the left eye from a patient carrying a relatively large FOXC1 deletion (FOXC1_5, Table 2). (E, F) Ocular images of an eye of a patient with a FOXC1 mutation c.316C>T (Table 1). (E) Image of the left eye showing posterior embryotoxon and limited iris hypoplasia. (F) Anterior synechiae in the left eye.

Both FOXC1 and PITX2 mutations can be associated with a wide spectrum of anterior segment phenotypes that do not differ significantly. Moreover, there is inter- and intrafamilial variable expressivity for certain point mutations, ^32,33 and mutations in either FOXC1 or PITX2 are prevalent in different clinical subgroups of the ASD spectrum (Supplementary Material S3; Fig. 6). ^6,34  

In this study, MLPA revealed six FOXC1 and five PITX2 deletions. The FOXC1 deletions can be added to the relatively large number of chromosomal rearrangements involving 6p25, but only a handful of these have been accurately characterized at the molecular level ^12,35 (as reviewed in Ref. 11). Recently, 10 duplications and deletions spanning FOXC1 were delineated in detail using targeted oligonucleotide arrays and detailed junction analysis. These rearrangements were nonrecurrent and a spectrum of recombination, DNA repair and replication were presumed to underlie these 6p25 rearrangements. ¹² In this study, we further assessed the extent of the deletions with a targeted microarray and compared their location and extent with those described by Chanda et al. ¹² (Figs. 2, 6). 

To date, only 14 microscopic or submicroscopic deletions spanning PITX2 have been reported, and the extent of these was further defined at the molecular level in only 6 of 14 cases. ^{4,13 –15} We delineated five PITX2 deletions by using arrays and subsequently compared them with deletions described previously (Fig. 5). The location of the breakpoints is scattered, which is indicative of the absence of a recombination hotspot. Such nonrecurrent rearrangement events may be explained by nonhomologous end joining (NHEJ), alternative NHEJ (alt-NHEJ), fork stalling and template switching (FoSTeS), or variations on these models as possible underlying mechanisms. ^12,36  

Of interest, there appeared to be no phenotypic differences between patients with intragenic PITX2 mutations and PITX2 deletions (Tables 1, 2). Furthermore, it should be noted that the patients carrying relatively large deletions had normal intelligence and did not display other obvious extraocular congenital malformations (Table 2). Based on the latter observation, we presume that no other dosage-sensitive genes are located in the PITX2 neighboring region. 

Our study not only provides substantial molecular and clinical data, but also demonstrates that deletions in the FOXC1 and PITX2 region can most easily be detected using MLPA with a commercially available probe mix. However, other techniques can be used to identify such copy number changes, including microsatellite analysis, quantitative PCR (qPCR), and copy number screening with microarrays. Microsatellite analysis has been successfully applied to identify PITX2 copy number changes by Lines et al., ¹⁵ but the major drawback for this is the need for parental DNA. ¹⁵ A better option is copy number screening using qPCR, which is flexible and comes at a relatively low average cost. ³⁷ High-density (targeted or genomewide) oligonucleotide arrays may be regarded as the ultimate alternative to MLPA and qPCR, but these arrays are still relatively expensive and require more hands-on time. On the other hand, they allow the assessment of the extent of the copy number change. Finally, the use of genomewide microarrays may lead to the identification of novel candidate regions. 

Recently, Tümer and Bach-Holm ⁴ generated a decision-tree aimed at selecting the most cost-effective diagnostic strategy for patients with a tentative diagnosis of ARM. However, in our diagnostic setting we perform direct sequencing of the open reading frames (ORFs) and MLPA in parallel. In our view, this is the most cost- and time-efficient strategy, leading to the identification of the underlying molecular defect in at least 40% of the diagnostic referrals for ARM. 

The large proportion of molecularly unsolved ASD patients indicates the involvement of other genes in the molecular pathogenesis of anterior segment malformations. Hence, mutation screening of three plausible candidate genes (P32, PDP2, and FOXC2) was undertaken in a pilot group of molecularly unsolved patients, but did not reveal any pathogenic mutation. Future strategies toward disease gene identification may include massive parallel sequencing of a spectrum of genes involved in ocular development, or high-resolution genomewide microarray-based copy number profiling for the identification of disease causing copy number changes. The latter strategy was applied with success in Peters Plus syndrome, an autosomal recessive disorder characterized by anterior eye chamber abnormalities, disproportionate short stature, and developmental delay. ³⁸ To our knowledge, a systematic genomewide search for new copy number variations (CNVs) in ASD patients has not yet been performed, and might lead toward the identification of novel ASD disease genes. 

In conclusion, there were underlying FOXC1/PITX2 genetic defects in 40% of our large ASD cohort. We considerably extended the current spectrum of intragenic FOXC1/PITX2 mutations and copy number changes, identified the smallest FOXC1/PITX2 deletions reported so far, and emphasized the need for dedicated copy number screening of the FOXC1 and PITX2 genomic landscape in ASD. Finally, mutations in the 3′UTR of FOXC1and PITX2, and in the ORF of P32, PDP2, and FOXC2 did not play a major role in the ASD cases examined in this study. 

Supplementary Materials

Text s01, PDF - Text s01, PDF 

Footnotes

 Supported by Specialisatiebeurs from Institute for the Promotion of Innovation through Science and Technology in Flanders (IWT-Vlaanderen) and Funds for Research in Ophthalmology (FRO) (BD). EDB is a senior clinical investigator of the Fund for Scientific Research (FWO).

Footnotes

 Disclosure: B. D'haene, None; F. Meire, None; I. Claerhout, None; H.Y. Kroes, None; A. Plomp, None; Y.H. Arens, None; T. de Ravel, None; I. Casteels, None; S. De Jaegere, None; S. Hooghe, None; W. Wuyts, None; J. van den Ende, None; F. Roulez, None; H.E. Veenstra-Knol, None; R.A. Oldenburg, None; J. Giltay, None; J.B.G.M. Verheij, None; J.-T. de Faber, None; B. Menten, None; A. De Paepe, None; P. Kestelyn, None; B.P. Leroy, None; E. De Baere, None

The authors thank the clinicians for providing patient samples and the families who participated in this study.