Complex balanced intrachromosomal rearrangement involving PITX2 identified as a cause of Axenfeld-Rieger Syndrome

Abstract

Axenfeld-Rieger Syndrome (ARS) type 1 is a rare autosomal dominant condition characterized by anterior chamber anomalies, umbilical defects, dental hypoplasia, and craniofacial anomalies, with Meckel’s diverticulum in some individuals. Here, we describe a clinically ascertained female of childbearing age with ARS for whom clinical targeted sequencing and deletion/duplication analysis followed by clinical exome and genome sequencing resulted in no pathogenic variants or variants of unknown significance (VUS’s) in PITX2 or FOXC1. Advanced bioinformatic analysis of the genome data identified a complex, balanced rearrangement disrupting PITX2. This case is the first reported intrachromosomal rearrangement leading to ARS, illustrating that for patients with compelling clinical phenotypes but negative genomic testing, additional bioinformatic analysis are essential to identify subtle genomic abnormalities in target genes.

INTRODUCTION

Axenfeld-Rieger Syndrome (ARS) is an autosomal dominant condition that affects between 1 in 50,000 to 1 in 100,000 live births (1). Axenfeld Rieger anomaly (ARA) is the typical ocular phenotype, recognized clinically by characteristic iris anomalies that include polycoria and corectopia due to iris hypoplasia, posterior embryotoxon, and iridocorneal adhesions along with an increased risk of glaucoma (2, 3). Approximately 70% of cases are caused by pathogenic variants in PITX2 or FOXC1, both of which encode transcription factors important during human development; the genetic etiology of the other 30% of cases remains unknown (4). ARS is divided into subtypes based on systemic effects outside the characteristic ocular phenotype. ARS Type 1, caused by variants in PITX2, is characterized by highly penetrant dental deformities (hypodontia and oligodontia) and umbilical defects (redundant periumbilical skin, umbilical hernia); up to 90% of individuals with a clinical diagnosis of ARS Type 1 can be explained by variants in PITX2 (4). The other major form of ARS, type 3, is caused by variants in FOXC1 and associated with a broad range of highly variable additional anomalies, most often heart and hearing defects, along with enamel hypoplasia, skeletal/joint anomalies, early hypotonia, and white matter changes in the brain (4). Our proband presented clinically with an ARS type 1 phenotype, suggesting a variant causing haploinsufficiency of PITX2. Panel, exome, and genome sequencing, however, were negative, prompting further bioinformatic analysis of the raw genome data.

PITX2 is expressed in various developing tissues including the eye, the brain, the pituitary gland, the craniofacial region, and the heart (5, 6). It has three main isoforms: PITX2A (NM_153427.2), PITX2B (NM_153426.3), and PITX2C (NM_000325.6). The isoforms have different N-terminal regions but share the final two exons encoding the homeodomain. PITX2A is the original transcript used in most literature reports, but PITX2C was recently identified as the MANE transcript (7). Commonly identified pathogenic variants in PITX2 include intragenic missense, splicing, frameshift, and nonsense variants affecting the shared C-terminal region and deletion of one or more coding exons (4); Deletion of a regulatory region in the upstream gene desert (chr4:110926950-111115957) or translocations misplacing the regulatory region to other parts of the genome have also been identified in individual with ARS type 1 (4, 8-12) and the importance of this upstream region in driving pitx2 expression was also confirmed in zebrafish studies (12). Intragenic variants in PITX2 are the most commonly identified type of pathogenic variant with over 106 different intragenic variants identified, but genomic deletions, including non-coding regulatory region deletions, account for roughly 17% of cases (4).

METHODS

Clinical Sequencing

Deletion / duplication analysis was performed by GeneDx (Gaithersburg, MD). Genomic DNA (gDNA) was collected from whole blood. An exon-level oligo array CGH was performed for the coding exons of PITX2 and GJA1. For FOXC1, Multiplex Ligation-dependent Probe Amplification (MLPA) was performed to detect copy number variants (CNVs). Singlet exome sequencing was performed by GeneDx. gDNA was collected from blood and coding regions were purified using GeneDx’s proprietary capture system. Paired-end reads were sequenced on an Illumina NGS platform. Genome sequencing was performed by Perkin-Elmer Genomics (Pittsburg, PA). gDNA was collected from whole blood of the trio of proband and parents. 2x150 bp reads were sequenced using Illumina NGS platform, resulting in an average depth of 29X.

Human Subjects

Trio raw genome sequencing data generated by Perkin-Elmer Genomics was reanalyzed with a focus on genomic rearrangements. Reanalysis was performed with informed consent through the Center for Individualized Medicine at Mayo Clinic. This study was approved by the Mayo Clinic Institutional Review Board (IRB).

Bioinformatic analysis

Raw genome sequencing data was reanalyzed for the proband and parents using BIMA (13) to map the sequencing data and SVAtools (14) to call structural variants. These two algorithms have proven performance for detecting junctions of chromosomal rearrangements. The SVAtools annotation includes strand assignment for each breakpoint of a junction.

Breakpoint sequencing confirmation

Reference DNA sequences spanning the minimal 5′ and maximal 3′ positions of the approximate breakpoints involving the PITX2 gene were identified as determined by SVAtools. Primers were designed around the predicted PITX2 breakpoint junctions using IDT’s PrimerQuest Tool (Primer sequences are available upon request). PCR was performed on proband’s DNA, as well as a negative control, with a C-KAPA2G Robust HS PCR Master mix (Roche, Basel, CH) using a touchdown PCR program. Results were visualized on a 2% agarose gel in a UV light box, and amplicon size for the proband was compared to the expected size given primer placement (~237 bp and ~221bp). PCR product was purified with the AMPure XP kit (Beckman Coulter, Brea, CA). Sanger sequencing was performed on a 3730xl DNA Analyzer using an ABI Cycle Sequencing Kit v1.1 (Thermo Fisher Scientific, Waltham, MA). The resulting sequences were analyzed using Mutation Surveyor v5.0.0 (SoftGenetics, State College, PA) and mapped to the GRCh38 genome using the BLAT function in the UCSC genome browser to determine precise breakpoints.

RESULTS

Clinical History

An adult woman presented to the clinical genomics service with the goal of identifying a genetic etiology. She had elevated intraocular pressure, anterior chamber abnormalities including guttata in both eyes, diffuse, patchy iris atrophy in both eyes, posterior embryotoxon in both eyes and corectopia in the left eye (Figure 1A-B). Optic nerve cupping consistent with glaucoma was noted in the right eye. She noted declining vision right eye greater than left. She had significant visual field loss related to glaucoma in her right eye (Figure S1) as well as thinning of the retinal nerve fiber layer in both eyes, right greater than left (Figure S2). She had significant dental anomalies with oligodontia and microdontia. As an infant, she had redundant periumbilical skin that was surgically corrected. She was otherwise non-dysmorphic, had normal developmental trajectory, and held an advanced degree. She had normal hearing and no evidence of congenital heart disease. Her clinical diagnosis was Axenfeld-Rieger Syndrome, and her phenotype was consistent with ARS type 1, suggestive of a pathogenic variant in PITX2 because of the presence of typical dental and umbilical malformations. Prior clinical testing including a gene panel with deletion/duplication analysis, whole exome sequencing, and whole genome sequencing did not identify any rare variants in PITX2.

Figure 1. Anterior segment anomalies yielding a clinical diagnosis of Axenfeld-Reiger Syndrome.

Posterior embryotoxoin in both left (A) and right (B) eyes, corectopia left eye.

Complex Intrachromosomal Rearrangement

Based on the compelling clinical phenotype suggestive of ARS type 1, further analysis was performed on raw genomic data with a focus on the two known ARS genes. Review of the SVAtools results identified a complex, balanced intrachromosomal rearrangement of chromosome 4 (Figure 2). The rearrangement involved seven junctions between 4q24-25 and 4q31.3-32.1 (Figure 2A-B). The breakpoints of these junctions disrupt four non-overlapping genes, ARHGEF38, PAPSS1, FAM160A1, and PITX2 (Figure 2B). Of these genes, only PITX2 is currently associated with human disease in the Online Mendelian Inheritance in Man (OMIM) database. The rearrangement was not identified in either of the proband’s parents, and therefore is de novo. No structural variants affecting FOXC1 were detected in the proband.

Figure 2. Complex intrachromosomal rearrangement of chromosome 4.

(A) Bioinformatic analysis identified a rearrangement between two regions of chromosome 4. Region 1 (R1) includes portions of 4q24 and 25 while region 2 (R2) includes portions of 4q31.3 and 32.1. (B) Seven junctions were identified which disrupt four non-overlapping genes, ARHGEF38, PAPSS1, PITX2, and FAM160A1. Only PITX2 is associated with human disease. The lettered segments are not to scale. Inverted segments are shown in blue. Breakpoint positions shown are in GRCh38. (C) Disruption of PITX2 by a 601 kb insertion. The six exons comprising PITX2 are shown as boxes. The homeodomain is shown in gray in exons 5 and 6. The gene structures of isoforms PITX2A and PITX2B are directly disrupted by the 601kb insertion. While the PITX2C coding region is left intact, it is separated from upstream regulatory elements that are necessary for expression.

Confirmation of PITX2 Breakpoint

Primers were designed around the predicted breakpoint junctions in PITX2 which would result in PCR amplicons of 237 bp and 221 bp (Figure S3) if the positions of the predicted breakpoint junctions were accurate. Clinical Sanger sequencing of the PCR fragments confirmed that the breakpoints in PITX2 were at chr4:110631631 and chr4:110631654 (Figure S3) which were consistent with the bioinformatic breakpoint prediction. The 22 bases in between were likely deleted during the rearrangement.

Predicted effect

The rearrangement is predicted to result in the insertion of a 601,408 bp segment into the PITX2 region. The insertion falls between coding exons of PITX2 isoforms PITX2A and B but is upstream of PITX2C. For transcripts PITX2A and B, this rearrangement results in the insertion of coding regions for other genes within intron 2/3 and extends the size of the intron by more than 50x, from ~11,000 bp to more than 612,000 bp, which is well outside the described human intron size range (15). Both the excessive size of the derivative intron and the inclusion of coding regions from other genes makes it unlikely that proper splicing will occur for the PITX2A/B transcripts. If the transcripts end at the final coding exon prior to the breakpoint, the shortened transcript is likely to be subject to nonsense-mediated decay due to the lack of a natural stop codon; if the transcript escapes NMD, the protein will be non-functional since the DNA-binding homeodomain is missing (Figure 2C). While the coding region of PITX2C is unaffected, the rearrangement likely disrupts the interaction of the identified critical upstream regulatory elements (in segment e) with the PITX2C promoter (in segment d) (Figure 2B-C), either through the increased distance of the promoter from the regulatory elements (due to the insertion of the 601 kb segment b) or through changes in chromatin structure caused by the addition of coding regions and promoters from other genes within the insertion. Previously identified deletions and translocations with breakpoints that separate the upstream regulatory region of PITX2 from the promoter have been shown to be sufficient to cause ARS type 1, highlighting the essential role of this region in the normal expression of PITX2 (8, 11, 12, 16-18). While PITX2 expression was not able to be directly measured in this patient, a similar inversion separating a promoter of another transcription factor, TFAP2A, from its enhancer region was shown to reduce expression of the gene in a patient-specific hiPSC based model (19) and changes in gene expression due to position effects from balanced translocations outside the coding region have been confirmed for other genes (20, 21). Thus, this complex rearrangement is predicted to lead to a heterozygous disruption of all major PITX2 isoforms (A-C), which is consistent with the known haploinsufficiency mechanism of ARS type 1.

DISCUSSION

This case illustrates the limitations of clinical genomic sequencing and demonstrates the need for further adoption of highly sensitive SV-detecting pipelines in the clinical setting. Our proband presented clinically with an ARS phenotype that was highly suggestive of disruption of PITX2, but standard clinical sequencing did not identify a causative variant. Current clinical genome sequencing typically relies on short read sequences of 100-150bp in length that map back to a known consensus genome reference sequence. Systematic use of structural variant detection tools that are capable of detecting SVs in short read sequence data is limited for most clinical sequencing labs.

The algorithm SVAtools was originally developed by Mayo to analyze short read mate-pair sequencing data, and subsequently has been adapted to detect complex chromosomal rearrangements in short read standard WGS data (14). A major limitation of current structural variant calling tools is the large number of calls that are produced. Having a compelling phenotype and a limited number of genes that would be associated with this phenotype allowed the bioinformatics team to focus their analyses on the region of the genome known to cause ARS type 1. When the SVAtools pipeline revealed the possibility of a complex rearrangement affecting PITX2, we were able to confirm these results with Sanger sequencing in a CLIA-certified genomics laboratory to return these results back to the proband. These results provided the molecular answer for this patient’s clinical presentation and allowed for informed reproductive counseling.

Germline complex chromosomal rearrangements are an increasing area of interest as a cause of monogenic disease. Balanced events such as the one reported here are often not detectable by standard cytogenetic array technology and current clinical sequencing focuses on the identification and interpretation of SNVs, small deletions, insertions, or deletion-insertion events, as well as larger copy number variants. The complexity of identifying pathogenic complex structural rearrangements has limited the reporting of these events in many current clinical genome sequencing tests. The rate of SV formation is estimated to be higher than that SNV formation(22), thus there is an urgent need to improve the ability of laboratories to characterize disease causing SVs in the clinical setting. This case provides a potential model for SV identification from discovery of an SV affecting PITX2 using bioinformatic tools developed at Mayo leading to the development of a CLIA-certified diagnostic confirmation assay. This case also illustrates the need for additional bioinformatic scrutiny when a clinical phenotype with low genetic heterogeneity is strongly suspected, but next generation sequencing genomic testing does not identify likely causative variants.

DATA AVAILABILITY

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.