Novel variants in CDH2 are associated with a new syndrome including Peters anomaly

Abstract

Peters anomaly (PA) is a congenital corneal opacity associated with corneo-lenticular attachments. PA can be isolated or part of a syndrome with most cases remaining genetically unsolved. Exome sequencing of a trio with syndromic PA and 145 additional unexplained probands with developmental ocular conditions identified a de novo splicing and three novel missense heterozygous CDH2 variants affecting the extracellular cadherin domains in four individuals with PA. Syndromic anomalies were seen in three individuals and included left-sided cardiac lesions, dysmorphic facial features, and decreasing height percentiles; brain magnetic resonance imaging identified agenesis of the corpus callosum and hypoplasia of the inferior cerebellar vermis. CDH2 encodes for N-cadherin, a transmembrane protein that mediates cell-cell adhesion in multiple tissues. Immunostaining in mouse embryonic eyes confirmed N-cadherin is present in the lens stalk at the time of separation from the future cornea and in the developing lens and corneal endothelium at later stages, supporting a possible role in PA. Previous studies in animal models have noted the importance of Cdh2/cdh2 in development of the eye, heart, brain, and skeletal structures, also consistent with the patient features presented here. Examination of CDH2 in additional patients with PA is indicated to confirm this association.

Introduction

Peters anomaly (PA) is a type of anterior segment dysgenesis characterized by central corneal opacity, absence of Descemet’s membrane, and iridocorneal (type I) or corneo-lenticular (type II) adhesions¹. PA is thought to be caused by a failure of lens vesicle separation from the corneal ectoderm and/or defects in later development of the cornea and nearby tissues leading to the formation of adhesions². While PA can occur as an isolated ocular condition, up to 72% of bilateral cases were found to have systemic anomalies¹. The best known cause of syndromic PA is classic Peters plus syndrome, characterized by PA, short stature, brachydactyly, dysmorphic facial features, and variable cardiac, genitourinary, and craniofacial defects, and caused by pathogenic variants in B3GLCT³. Other occasional genetic causes have been identified, including 8q21 deletions and pathogenic variants in TFAP2A, FLNA, PAX6, PITX2, CYP1B1, FOXC1, and WDR37, however the majority of cases of syndromic PA remains unexplained^4-7. Here we report identification of novel variants in CDH2 in four patients with PA.

Materials and Methods

Human sample analyses

This human study was approved by the Institutional Review Board of Children’s Hospital of Wisconsin with written informed consent obtained for every participant including publication of photographs.

Exome sequencing/analysis and Sanger sequencing were performed as previously described; SNP & Variation Suite (SVS) software (Golden Helix) was used⁷. Splicing predictions were made using Human Splicing Finder (http://www.umd.be/HSF/) and the resulting DNA sequences were translated through EMBOSS Transeq (https://www.ebi.ac.uk/Tools/st/emboss_transeq/).

Mouse N-cadherin immunostaining

Mouse embryos were obtained from timed-matings (E10.75/E11.0 and E17.5), fixed in 4% paraformaldehyde, and embedded in OCT compound following sucrose infiltration. Cryosections (10mm) through the eye region were prepared on glass slides and subjected to immunofluorescent labeling using the following antibodies [(a-N-cadherin, BD Biosciences, 610921, 1:400) (a-b-catenin, Santa-Cruz Biotechnology, sc-7199, 1:200) (a-mouse-594, Life Technologies, A21203, 1:1000) (a-rabbit-488, Life-technologies, A11008, 1:1000) (a-Collagen I, Abcam, AB34710, 1:200)]. Slides were first immersed in 100mM Tris and heated in a pressure cooker for 15 minutes, then rinsed in PBS. Primary antibodies were incubated in 4% powdered milk in PBS overnight at 4°C, rinsed in Phosphate-buffered saline (PBS) 3x for 30 min, and incubated in secondary antibody for 2 hours at room temperature. Imaging was performed on a widefield inverted microscope (Zeiss) with a 40x oil objective.

Results

Patient 1 (Figure 1, Table 1) is a Caucasian male with syndromic PA. Ocular features included bilateral PA with iridocorneal adhesions, right mild anterior cortical cataract, eccentric pupils and closed angle glaucoma; he underwent successful bilateral corneal transplants (donor corneas) and pathology of the corneal buttons showed bilateral bullous keratopathy, central stromal lamellar disorganization, and central absence of Descemet’s membrane. He also had mild dysmorphic facial features, mild cognitive delays and language disorder, brain anomalies (see below), mildly tortuous aortic arch with coarctation, and mild pectus excavatum. Growth parameters showed decreasing height percentiles (length of 68.7 cm (50–75^th centile) at 6 months and 89 cm (3–10^th centile) by 34 months) with normal weight and head circumference. Trio analysis of exome data identified rare variants of interest in two genes: compound heterozygous variants in HSPG2, which were also present in three unaffected siblings and thus considered benign, and a de novo heterozygous splicing variant in CDH2 (encoding N-cadherin, a cell-surface glycoprotein), c.702+1G>A, present in 10/23 reads, confirmed by Sanger sequencing and not present in the parents or four unaffected siblings (Figure 1P).

Figure 1. Clinical images (A-O) and novel CDH2 alleles (P,Q) of individuals with syndromic and isolated Peters Anomaly.

Facial photographs of Patient 1 at 3 months of age before corneal transplant (A) and at 3 years of age (B-C) and of Patient 3 in infancy (D) and at 7 years of age (E). Optical coherence tomography of the eye showing iridocorneal adhesions (F) and retinal images showing mild optic nerve cupping but otherwise normal retinal development (G-H) from Patient 1. Brain MRI images from Patient 1 (I-K) and 3 (L-O). Sagittal noncontrast T1-weighted image from Patient 1 (I) shows complete agenesis of the corpus callosum with absence of the cingulate gyrus and radial orientation of cerebral sulci. There is minimal hypoplasia of the inferior cerebellar vermis (white arrow). Axial T2-weighted image (J) shows a parallel orientation of the lateral ventricles. Coronal T2-weighted image (K) shows lateral convexity of the frontal horns due to callosal agenesis and absence of the septum pellucidum. There is also incomplete rotation of the hippocampal formations (black arrows). Sagittal noncontrast T1-weighted image from Patient 3 (L) shows complete agenesis of the corpus callosum and hypoplasia of the inferior cerebellar vermis (white arrow). There is extradural tissue displacing the torcula suggestive of a thrombosed dural sinus malformation (asterisk). Coronal (M) and axial (N and O) T2-weighted images demonstrate nonsegmented appearance of the striatum bilaterally (black arrows) and numerous foci of subependymal gray matter heterotopia (white arrowheads). There is also incomplete rotation of the hippocampal formations (M). Pedigree of Patient 1 (specified with an arrow) with CDH2 de novo allele or wild type (WT) indicated for each family member tested (left) and sequencing traces for Patient 1 and his parents (right) (P). Schematic drawing of CDH2 protein (N-cadherin) (O) with known domains of CDH2 shown and positions of the identified human variants in Patient 1-4 indicated (red arrows). (1-25, signal peptide; 26-159, propeptide; 160-267, 268-382, 383-497, 498-603 and 604-714, EC domains I-V, correspondingly; 725-745, transmembrane domain, 746-906, cytoplasmic domain).

Table 1.

Summary of genetic and clinical information of affected patients.

CDH2 allele info	Patient 1	Patient 2	Patient 3	Patient 4
Nucleotide change	c.702+1G>A	c.485T>A	c.1574A>G	c.1807C>T
Predicted effect	Mis-splicing/ truncation	p.(Val162Asp)	p.(Asp525Gly)	p.(Pro603Ser)
Prediction software statistics	−2.59 (MaxEnt)	Damaging (5/5)†	Damaging (5/5)†	Damaging (5/5)†
gnomAD frequency	NP (0/250,000)	NP (0/251,000)	NP (0/251,000)	NP (0/251,000)
Inheritance	de novo	U	U	Inherited
EYE
Peters anomaly	+	+	+	+
Glaucoma or elevated IOP	+	+	+	+
FACE
Hypertelorism	−	+	+	−
Flat nasal bridge	+	+	+
Thin upper lip	+	+	+	−
Low set or posteriorly rotated ears	+	+	+	−
Upturned ear lobes	+	−	+	−
Small mouth with downturned corners	−	+	+	−
Low anterior hairline	−	+	+	−
BRAIN/NEUROLOGICAL
Cognitive delays	+	U	+	−
Mild hypotonia	+	+	+	−
Clenched hands/ hypertonic extremities	−	+	+	−
Agenesis corpus callosum	+	+	+	−
Hypoplasia of cerebellar vermis	+	U	+	−
Incomplete hippocampal rotation	+	U	+	−
Absent septum pellucidum	+	U	+	−
Other		Dilatation of ventricles	Heterotopia, Dural sinus malformation (likely)
CARDIAC
Left-sided cardiac lesion	+	−	+	−
TORSO
Widely spaced nipples	−	+	+	−
Other	Pectus excavatum	−	Sacral cleft	Umbilical hernia
GROWTH
Decreasing height percentiles with age	+	U	+	−

^†SIFT, Polyphen2 HVAR, MutationTaster, MutationAssessor, and FATHMM MKL;

IOP- intraocular pressure; NP- Not Present; U- unknown; Bold- present in 3 or more cases

The identified CDH2 allele is not present in gnomAD (https://gnomad.broadinstitute.org) and is predicted to result in loss of the donor splice site, reducing the MaxEnt score from 5.58 to −2.59. The nearest possible donor splice site with a MaxEnt score of 4.0 or higher is at position +290 and would lead to the introduction of 11 erroneous amino acids followed by a stop codon, as would all other downstream alternative donor splice sites. This predicted effect is likely to result in a loss-of-function, since it is expected to truncate the protein at its first extracellular cadherin (EC) domain. Analysis of the frequency of variants in CDH2 in gnomAD suggests that the gene is very intolerant to loss-of-function variants and somewhat intolerant of missense mutations⁸.

Review of exome data from an additional 145 probands with developmental ocular anomalies (including 36 with PA) identified three other novel (not present in ~250,000 alleles available in gnomAD) variants affecting the EC domains I and IV and predicted to be damaging by all five in silico analysis programs (Table 1) in 3/36 additional probands diagnosed with PA (Figure 1Q, Table 1). However, no novel variants were identified in the109 probands with other ocular conditions. Including Patient 1, CDH2 variants were identified in 11% (4/37) PA cases.

Patient 2 was previously described (Patient 6⁹). Briefly, he is a Jamaican male with bilateral PA and elevated intraocular pressure, along with dysmorphic facial features, hypotonia with clenched hands, brain anomalies (agenesis corpus callosum, hydrocephalus, and intraventricular hemorrhage), and widely spaced nipples. Growth parameters were normal at birth; unfortunately, the patient was lost to follow-up. He had a heterozygous damaging missense variant in CDH2, c.485T>A, p.(Val162Asp) that was present in 20/42 reads and seen in both directions.

Patient 3 is a female of mixed ancestry with syndromic PA (Figure 1). Ocular anomalies consisted of bilateral corneal opacity and extensive neovascularization with no view of the anterior or posterior segment; corneal transplant (keratoprosthesis) failed in both eyes. Non-ocular anomalies included dysmorphic facial features, hypotonia with hypertonic extremities at birth, cognitive delay noted at 7 years of age, structural brain anomalies (see below), left-sided complex cardiac lesion (coarctation of the aorta, atrial septal defect, persistent ductus arteriosus, and mildly hypoplastic left heart), widely spaced nipples, and sacral cleft. Growth parameters showed decreasing height percentiles with a birth length of 49 cm (50^thcentile) but <3^rd centile by 7 years of age with additional skeletal anomalies such as bowed legs and in toeing noted. She had a heterozygous damaging missense variant in CDH2, c.1574A>G; p.(Asp525Gly) seen in 52/80 reads and confirmed by Sanger sequencing; parental samples and family history are not available.

Patient 4 is an African American male with bilateral PA with iridocorneal adhesions, congenital glaucoma, failed corneal transplant (left) and umbilical hernia. No other syndromic anomalies were noted at 18 months. An older brother is reported to have non-syndromic bilateral PA but was unavailable. Patient 4 had a heterozygous damaging missense variant in CDH2, c.1807C>T; p.(Pro603Ser) seen in 20/39 reads and confirmed by Sanger sequencing; the variant was also present in the mother, who is apparently unaffected but without evidence of formal eye exam, and with family history of unilateral cataract and glaucoma. Unfortunately, the family was lost to follow-up so additional examination/testing could not be performed.

Brain MRI images were reviewed for Patients 1 and 3 and showed overlapping patterns of structural brain anomalies (Figure 1 I-O). Sagittal noncontrast T1-weighted image showed complete agenesis of the corpus callosum and hypoplasia of the inferior cerebellar vermis in both patients along with absence of the cingulate gyrus and radial orientation of the cerebral sulci in both patients and a suspected dural sinus malformation in Patient 3. T2-weighted images additionally showed incomplete rotation of the hippocampal formations along with absence of the septum pellucidum in both patients as well as a nonsegmented appearance of the striatum and subependymal gray matter heterotopia in Patient 3.

Review of the Decipher database¹⁰ generally supported this association. Within the publicly available data, we identified one additional case with a de novo missense variant and overlapping anomalies of the eye, nervous system, head or neck, and skeletal system, and six cases with deletion of CDH2, both inherited and de novo, and variable overlapping phenotypes including cognitive impairment in most and abnormalities of the eye and heart in one case each.

To determine whether N-cadherin is present in the areas affected by PA, we performed N-cadherin localization studies in ocular tissue from mouse embryos at E10.75/E11.0 and E17.5. N-cadherin protein localization was observed to be restricted to the lens and not the surface ectoderm/corneal epithelium in E10.75-E11.0 embryos (Figure 2A-C). However, N-cadherin was detected at the locations of first cellular contacts between cells of the lens stalk region. At later stages (E17.5), N-cadherin was restricted to the cells of the developing lens and corneal endothelium and was localized to cell membranes that face each other (Figure 2D).

Figure 2. Mouse N-cadherin localization studies.

A-C. Immunostaining in mouse embryos at the time of lens vesicle separation from overlying corneal ectoderm (E10.75-E11.0). Higher magnification images (corresponding to the boxed area in A-C) of beta-catenin and N-cadherin are shown on the right. Please note N-cadherin at the locations of first cellular contact between the cells of the lens stalk region (arrows) and no positive staining of the surface ectoderm/corneal epithelium (arrowheads); however, beta-catenin is prominently localized at both locations. D. Immunostaining in the developing mouse cornea and lens at a later developmental time point (E17.5). Higher magnification images of the boxed area (d) are shown on the right. N-cadherin is restricted to the cells of the lens and corneal endothelium (arrowheads). The localization of collagen I between the lens epithelium and corneal endothelium shows the normal separation between the two tissues and the N-cadherin protein within each cell type during development.

Discussion

Review of exome data from individuals with mixed developmental ocular conditions identified four novel CDH2 variants affecting the extracellular cadherin (EC) domains I and IV of N-cadherin in unrelated patients with syndromic (3) or non-syndromic (1) PA. Altogether, novel CDH2 variants were identified in 4/37 (11%) of unexplained PA cases and 0/109 probands with other ocular conditions, suggesting that disruption of CDH2 is specifically associated with PA in our cohort.

Through calcium-dependent cell adhesion, N-cadherin plays an important role in the embryonic development. N-cadherin is a member of the cadherin family whose structure consists of 5 conserved extracellular cadherin domains (EC-I-V) followed by a single transmembrane domain and a cytoplasmic domain¹¹. The EC-I domain has been shown to be responsible for adhesion and specificity while the EC-IV domain promotes cell motility and transition¹². Cdh2 function is well conserved across species. Studies in mouse embryos have noted the importance of Cdh2 in the proper development of the heart, brain, and skeletal structures^13-16. The observed abnormalities of the cardiac outflow tract were consistent with the left-sided cardiac phenotypes noted in the patients reported here as well as a previously published inherited novel splicing variant in a patient with a congenital left-sided cardiac lesion¹⁷; additionally, two novel missense variants within EC-I and EC-III were identified in patients with arrhythmogenic cardiomyopathy¹⁸. In zebrafish, variants in the EC-I or EC-IV domains of cdh2 resulted in the glo, pac, and lyr phenotypes characterized by defects in the brain, somites and cardiac/circulatory system, along with retinal and lens defects^{19; 20}. Interestingly, the glo allele affects an amino acid immediately adjacent to the p.(Val162Asp) variant within the EC-I domain identified in Patient 2 while pac and lyr disrupt the EC-IV region affected in the other two patients presented here.

A role for N-cadherin in PA is supported by several lines of evidence in animal models. We detected N-cadherin at the relevant time points during eye development (see above), consistent with the previously published reports that identified Cdh2/N-cadherin expression in the lens placode, neural retina, and corneal endothelial cells²¹. Loss of Cdh2 specifically in the corneal endothelium in mouse resulted in corneal opacity²². Conditional deletion of Cdh2 from the presumptive lens ectoderm resulted in microphthalmia, iris hyperplasia, and small lens while loss of both Cdh2 and Cdh1 caused iridocorneal adhesions, absent corneal endothelium, and persistent lens stalk, similar to PA, as well as aphakia, and retinal anomalies²³. The expression of N-cadherin at the time of lens vesicle separation in the locations of first cellular contact between cells of the lens stalk region, as well as in the lens and corneal endothelial layer at later developmental stages is consistent with the hypothesis that abnormalities in lens vesicle separation and/or formation of the endothelial layer are likely causes of PA².

In conclusion, novel and predicted deleterious CDH2 alleles were identified in 11% of PA exomes in our study. Additional studies of CDH2 in patients with PA, especially in association with heart, brain, and growth defects, are highly recommended to confirm and further define this association.