Sporadic and Familial Congenital Cataracts: Mutational Spectrum and New Diagnoses Using Next-Generation Sequencing.

Congenital cataracts are a significant cause of lifelong visual loss. They may be isolated or associated with microcornea, microphthalmia, anterior segment dysgenesis (ASD) and glaucoma, and there can be syndromic associations. Genetic diagnosis is challenging due to marked genetic heterogeneity. In this study, next-generation sequencing (NGS) of 32 cataract-associated genes was undertaken in 46 apparently nonsyndromic congenital cataract probands, around half sporadic and half familial cases. We identified pathogenic variants in 70% of cases, and over 68% of these were novel. In almost two-thirds (20/33) of these cases, this resulted in new information about the diagnosis and/or inheritance pattern. This included identification of: new syndromic diagnoses due to NHS or BCOR mutations; complex ocular phenotypes due to PAX6 mutations; de novo autosomal-dominant or X-linked mutations in sporadic cases; and mutations in two separate cataract genes in one family. Variants were found in the crystallin and gap junction genes, including the first report of severe microphthalmia and sclerocornea associated with a novel GJA8 mutation. Mutations were also found in rarely reported genes including MAF, VIM, MIP, and BFSP1. Targeted NGS in presumed nonsyndromic congenital cataract patients provided significant diagnostic information in both familial and sporadic cases.

Introduction

Congenital cataracts are responsible for approximately 10% of childhood blindness worldwide [Gilbert et al., 1993; Foster et al., 1997; Gilbert and Foster, 2001; Muhit and Gilbert, 2003]. Management is often difficult due to the risk of amblyopia in the developing visual system and complicated by the presence of glaucoma with attendant further loss of vision [Swamy et al., 2007]. Congenital cataracts occur due to a disruption to the lens microarchitecture or the accumulation of protein aggregates in the lens [Shiels and Hejtmancik, 2007]. There are autosomal‐dominant, autosomal‐recessive, and X‐linked genetic forms of congenital cataracts, which may be isolated (nonsyndromic) or associated with systemic disease or syndromes (syndromic). In only approximately 18% of cases is there a known family history [Wirth et al., 2002], so a significant proportion are sporadic cases where it is not known if there may be an underlying genetic cause for the lens abnormality.

Mutations in over 30 genes are known to cause nonsyndromic forms of congenital cataracts [Hejtmancik, 2008; Huang and He, 2010]. These code for a variety of lens structural proteins including crystallin, gap junction, intermediate filament, and lens membrane proteins, as well as transcription factors, and there are over 150 other cataract loci [Shiels et al., 2010]. Around 18% of patients with congenital cataracts also have microcornea, defined by horizontal corneal diameter (HCD) <10 mm at birth in an eye of normal axial length (AXL) [Hansen et al., 2007]. There can be other associated ophthalmic abnormalities including microphthalmia, anterior chamber abnormalities, and retinal degeneration. Some syndromic forms of congenital cataracts can be subtle with associated systemic features such as learning difficulties and facial features presenting or becoming apparent only in later childhood [Slavotinek, 2011]. Owing to the marked genetic and phenotypic heterogeneity, genetic testing has been a difficult, expensive, and time‐consuming task using conventional Sanger sequencing. When this was undertaken on a research basis, mutations were identified in known cataract genes in 33%–50% of familial nonsyndromic cases [Hansen et al., 2007; Kumar et al., 2011; Sun et al., 2011], but sporadic cases were not examined.

For such a heterogeneous condition, high‐throughput sequencing techniques offer a much more time and cost‐effective method of mutation detection. Using whole‐exome sequencing (WES) with target gene analysis, mutations were identified in approximately 50% of congenital nonsyndromic cataract families with a clear autosomal‐dominant inheritance pattern [Reis et al., 2013; Prokudin et al., 2014]. Application of targeted next‐generation sequencing (NGS) in 36 families with a high proportion of consanguineous and syndromic cases led to a mutation detection rate of 75% [Gillespie et al., 2014].

In this study, to investigate NGS in nonsyndromic cases with a variety of inheritance patterns including a high proportion of sporadic cases in nonconsanguineous families, we applied targeted NGS of cataract genes in a cohort of 46 apparently nonsyndromic congenital cataract families, the largest series of congenital cataract cases studied to date. Half the cases were sporadic, and the remainder had apparent autosomal‐dominant or autosomal‐recessive inheritance patterns. Mutations were identified in 17 genes and we found a high mutation detection rate of approximately 70% in these sporadic and familial apparently nonsyndromic cases. New diagnoses and/or new information about inheritance patterns were identified in almost two‐thirds of the cases where pathogenic variants were identified.

Materials and Methods

Patients

Forty‐six probands with apparently isolated, nonsyndromic bilateral congenital cataracts were investigated. These patients were seen in the genetic eye clinic of a major pediatric referral hospital over a period of 12 years. The majority of the families were from a Caucasian background (37/46) and the remainder had Middle Eastern (7/46) or Asian (2/46) ethnicity, representing the diverse cultural background of the Western Sydney region. Patients were selected on the basis of DNA sample availability, and preferably with parental and other family member samples also available. Approximately half the cases (22/46) were familial, 19 with a likely autosomal‐dominant mode of inheritance (Families 3–6, 8, 10, 11, 14, 16, 19, 22, 23, 26, 31–33, 38–40), although one of these (Family 14) was in a consanguineous family, and in three families the pedigree was suggestive of autosomal‐recessive inheritance (Families 1, 24, and 46), with consanguinity in two of these (Families 24 and 46). The other 24 of the 46 probands were sporadic cases (Supp. Fig. S1). Most patients had bilateral congenital cataracts diagnosed in the 1st year of life, and some also had microcornea and additional ophthalmological features such as microphthalmia, nystagmus, and glaucoma (Supp. Table S1). Genomic DNA was isolated from leukocytes of peripheral venous blood in all cases, except for the proband from Family 40 where only a saliva sample could be obtained. Two probands were previously investigated using a WES approach and were independently investigated in this study (Families 8 and 11 in this study correspond to Families 4 and 2, respectively, in Prokudin et al. [2014]). Ophthalmological details and samples for genomic DNA extraction were collected from family members when available (Supp. Fig. S1). All experiments were approved by the Human Research Ethics Committee of Sydney Children's Hospital Network, Sydney, Australia.

NGS, Variant Prioritization, and Validation

Thirty‐eight probands (from Families 1–38) were analyzed using the Illumina TruSeq Custom Amplicon (Version 1.5; Illumina Inc., 2011–2013, CA, USA) approach. As part of our work in developmental eye disease, a panel of amplicons targeted the coding exons and 10 bp flanking intronic sequences were designed to include 32 known congenital cataract disease genes (Supp. Table S2 – contains RefSeq [NCBI] accession numbers). Data were based on the GRCh37/hg19 version of the reference genome. Nonsyndromic and some syndromic congenital cataract disease genes were included owing to our clinical observation that some forms of syndromic cataracts may be subtle and missed by referring pediatricians and ophthalmologists, especially in infancy. A capture‐based approach was also used in this study when it became available, using the Illumina TruSight and TruSight One Clinical Exomes (Illumina Inc., 2013–2014). These both used the Nextera capture method (Illumina Inc.) designed to focus on disease‐causing genes including those in HGMD (http://www.hgmd.org/) and OMIM (http://www.omim.org/). Both clinical exome panels included the 32 congenital cataract genes we targeted in our custom amplicon (TruSeq) approach. The Illumina TruSight Clinical Exome was performed on two probands (Families 39 and 40) and one additional family member (father from Family 6). This captured 2,761 OMIM‐identified disease genes. The expanded TruSight One Clinical Exome, targeting 4,813 disease genes, was also used on 23 probands. These included 17 mutation‐negative patients from the TruSeq custom amplicon cohort (Families 1, 3, 4, 5, 7, 9, 11, 15, 17, 19, 21, 26, 29, 30, 34, 36, 38) and six probands (from Families 41 to 46). The custom amplicon samples were sequenced on the Illumina MiSeq (Illumina Inc.) with 2×250 bp paired‐end reads (Ramaciotti Centre, Sydney, Australia). The clinical exome samples were sequenced with the Illumina HiSeq 2500 (Illumina Inc.) using a 2×150 bp paired‐end read protocol (Ramaciotti Centre).

Alignment was undertaken using NextGene software (v2.4.1, 2015, SoftGenetics, PA, http://www.softgenetics.com/NextGENe.html), and this was also used for variant identification. Variants in the 32 cataract genes were annotated using Annovar [Wang et al., 2010] (March 22, 2015 Build, http://annovar.openbioinformatics.org) and Alamut‐Batch (Version 1.4, 2015; Interactive Biosoftware, Rouen, France, http://www.interactive‐biosoftware.com/alamut‐batch/) (Table 1).

Table 1

Sporadic and Familial Cases with Likely Causative Variants

Family (proband)	Inheritance, before/after testing	Gene Refseq ID	Nucleotide change	Predicted amino acid change	Protein domain	In silico: SIFT, MutTaster, PolyPhen, PhyloP, ExAC	Segregation	Novel	ACMG classification
Family 2 (II:1)	#Sporadic/new AD	GJA8 NM_005267.4	c.119C>T	p.Ala40Val	TM1	D, D, P, high, nil	Yes: de novo in proband	Yes	LP
Family 6 (II:1)	#AD/X‐linked	NHS NM_198270.3	c.2707delG, hemizygous	p.Glu903Asnfs*4	Nil	Nil	Yes: hemizygous in affected brothers, heterozygous in mother	Yes	P
Family 6 (I:1)	#AD/likely new AD	MAF NM_001031804.2	c.819G>C	p.Glu273Asp	b‐Zipper	D, D, P, mod, nil	N/K: parental samples unavailable, not present in sons	Yes	LP
Family 7 (II:3)	#Sporadic/X‐linked	BCOR NM_001123385.1	c.4390_4393del	p.Glu1464Profs*19	Nil	Nil	Yes: de novo in proband. Not found in son.	Yes	P
Family 8 (I:2)	AD/AD	CRYGCNM_020989.3	c.497C>T	p.Ser166Phe	4th Greek Key	D, D, P, high, nil	Yes: present in affected sons	Prokudin et al. (2014)	P
Family 10 (III:3)	AD/AD	PAX6 NM_0001604.4	c.1119del	p.Thr374Profs*5	PST rich	Nil	Yes: present in affected sibling and mother	Peter et al. (2013)	P
Family 12 (II:1)	#Sporadic/likely new AD	CRYBB1 NM_001887.3	c.368G>A	p.Arg123His	3rd Greek key	D, D, P, mod, nil	N/K: Not present in mother, father's sample unavailable	Yes	LP
Family 13 (II:2)	#Sporadic/new AD	CRYBB2 NM_000496.2	c.556T>C	p.Ser186Pro	4th Greek key	D, D, P, mod, nil	De novo; not found in parents or two unaffected siblings	Yes	LP
Family 14 (II:2)	#AD?/AR	GJA8 NM_005267.4	c.89dupT, homozygous	p.Ile31Hisfs*18	TM1	Nil	Yes: consanguineous family, mother heterozygous, father unavailable	Yes	P
Family 16 (I:2)	AD/AD	CRYAA NM_000394.3	c.142T>G	p.Tyr48Asp	N‐terminal	T, D, P, mod, nil	Yes: present in affected son	Yes	LP
Family 18 (II:1)	#Sporadic/new AD	PAX6 NM_0001604.4	c.239T>A	p.Ile80Asn	Paired	D, D, P, high, nil	Yes: de novo in proband	Yes	LP
Family 19 (II:2)	AD/AD	GJA3 NM_021954.3	c.260C>T	p.Thr87Met	TM2	D, D, P, high, nil	Found in all affected family members	Guleria et al. (2007)	P
Family 20 (II:2)	#Sporadic/new AD	GJA8 NM_005267.4	c.134G>C	p.Trp45Ser	TM1	D, D, P, high, nil	Yes: de novo in proband	Ma et al. (2005)	P
Family 22 (II:3)	AD/AD	CRYGC NM_020989.3	c.328_329delinsT	p.Pro110Serfs*37	3rd Greek key	Nil	Yes: present in affected children	Yes	P
Family 23 (I:1)	AD/AD	CRYAA NM_000394.3	c.61C>T	p.Arg21Trp	N‐terminal	D, D, B, high, nil	Yes: present in affected sons, absent in unaffected son	Devi et al. (2008)	P
Family 24 (II:8)	AR/AR	CRYBB3 NM_004076.4	c.493G>C, homozygous	p.Gly165Arg	4th Greek key	D, D, P, high, nil	Yes: homozygous in three affected siblings, heterozygous in both parents	Riazuddin et al. (2005)	P
Family 25 (II:1)	#Sporadic/likely new AD	VIM NM_003380.3	c.15del	p.Val6Cysfs*26	Head	Nil	N/K: parental samples unavailable	Yes	LP
Family 26 (II:4)	AD/AD	GJA3 NM_021954.3	c.176C>T	p.Pro59Leu	EC1	D, D, P, high, nil	N/K: parental samples unavailable	Bennett et al. (2004)	P
Family 27 (II:1)	#Sporadic/likely new AD	MAF NM_001031804.2	c.915C>T	p.Cys305Trp	b‐Zipper	D, D, P, high, nil	N/K: parental samples unavailable	Yes	LP
Family 28 (II:2)	#Sporadic/new AD	MIP NM_012064.3	c.97C>T	p.Arg33Cys	EC1	D, D, P, high, nil	Yes: de novo in proband	Gu et al. (2007)	P
Family 29 (II:1)	#Sporadic/?new AD	GJA3 NM_021954.3	c.7G>C	p. Asp3His	SP	T, D, P, high, nil	N/K: parental samples unavailable	Yes	LP
Family 30 (II:1)	#Sporadic/?new AD	MAF NM_001031804.2	c.880C>T	p.Arg294Trp	b‐Zipper	D, D, P, high, nil	N/K: parental samples unavailable	Yes	LP
Family 31 (III:1)	AD/AD	CRYGD NM_006891.3	c.70C>A	p.Pro24Thr	1st Greek key	T, B, B, weak, nil	Yes: present in affected grandmother	Burdon et al. (2004)	P
Family 32 (I:2)	AD/AD	MIP NM_012064.3	c.597_598ins GGGAACATTCCACT	p.Asn200Glyfs*12	EC3	Nil	Yes: present in affected daughter	Yes	P
Family 33 (II:2)	AD/AD	CRYAB NM_001885.2	c.320G>T	p.Arg107Leu	Small heat shock protein	T, D, P, high, nil	Yes: present in affected son and father	Yes	LP
Family 35 (II:1)	#Sporadic/new AD	CRYBB2 NM_000496.2	c.343C>A	p.Pro115Thr	3rd Greek key	D, D, B, high, nil	Yes: de novo in proband	Yes	LP
		CRYBB2 NM_000496.2	c.355G>A	p.Gly119Arg	3rd Greek key	D, D, P, mod, nil	Yes: de novo in proband	Yes	LP
Family 37 (II:1)	#Sporadic/likely new AD	CRYBB2 NM_000496.2	c.583T>G	p.Trp195Gly	C‐terminal	D, D, P, mod, nil	Likely de novo; mutation not found in mother, father unavailable	Yes	LP
Family 39 (II:1)	AD/AD	CRYBB3 NM_004076.4	c.634T>C	p.*212Argext*40	C‐terminal	Nil	Yes: present in affected father	Yes	LP
Family 40 (III:3)	#AD/X‐linked	NHS NM_198270.3	c.3624C>A	p.Cys1208*	Nil	Nil	Yes: hemizygous in affected male, heterozygous in mother and sister	Liao et al. (2011)	P
Family 41 (II:1)	#Sporadic/new AD	CRYGD NM_006891.3	c.448dup	p.Asp150Glyfs*3	4th Greek key	Nil	Yes: de novo in proband	Yes	P
Family 44 (II:1)	#Sporadic/new X‐linked	BCOR NM_001123385.1	c.1136_1139del	p.Val379Alafs*62	Nil	Nil	Yes: de novo in proband	Yes	P
Family 45 (II:1)	#Sporadic/new AD	GJA8 NM_005267.4	c.151G>A	p.Asp51Asn	EC1	D, D, P, high, nil	Yes: de novo in proband	Yes	LP
Family 46 (II:2)	AR/AR	BFSP1 NM_001195.4	c.1492del	p.Ser498Leufs*24	Tail	Nil	Mother heterozygous	Yes	P
		BFSP1 NM_001195.4	c.812T>C	p.Ile271Thr	2nd coil	D, D, P, high, rs147718368 0.0003625 (44/121392)	Not found in mother, father unavailable	Yes	LP

Variants are heterozygous unless otherwise specified. # In the “inheritance, before/after testing” column indicate families where new information was obtained regarding the inheritance pattern in the family from the next‐generation sequencing work. Protein domains: SP, signal peptide; TM, transmembrane; EC, extracellular; IC, intracellular domains. In Silico: D, damaging; P, pathogenic; B, benign; T, tolerated. The computational calculations of SIFT, MutationTaster, PolyPhen, and PhyloP were performed in Alamut‐Visual (Version 2.6, Jan 2015; Interactive Biosoftware, http://www.interactive‐biosoftware.com/alamut‐visual/). ExAC rs number and minor allele frequency provided where available. Segregation: N/K, not known. ACMG classification as per Richards et al. (2015); P, pathogenic; LP, likely pathogenic.

Variants with coverage <15X were removed. Variants were annotated for minor allele frequencies in the dbSNP [Smigielski et al., 2000] (Build 144, May 2015, http://www.ncbi.nlm.nih.gov/projects/SNP), 1000genomes [Abecasis et al., 2012] (Phase 3, 2013, http://www.1000genomes.org/), Exome Variant Server (NHLBI GO Exome Sequencing Project [ESP], Seattle, WA; Version 2, July 2013, http://evs.gs.washington.edu/EVS/) and Exome Aggregation Consortium (ExAC) databases (Version 0.3, January 2015, Cambridge, MA, http://exac.broadinstitute.org/), and heterozygous variants with minor allele frequencies >0.01 were filtered out. Variants were analyzed for possible pathogenic clinical significance according to the 2015 American College of Medical Genetics and Genomics (ACMG) guidelines [Richards et al., 2015], based on a combination of previous reports in the literature and computational, functional, and population data. Nonsense, frameshift, and canonical splice‐site variants were considered strongly indicative of pathogenicity. For missense mutations, Alamut Visual (Version 2.6, Jan 2015; Interactive Biosoftware, http://www.interactive‐biosoftware.com/alamut‐visual/) was used for the individual variant analyses, providing computational algorithms for SIFT [Ng and Henikoff, 2001] (Version 1, http://sift.jcvi.org/), PolyPhen‐2 [Adzhubei et al., 2010] (Version 2.2.2, 2012, http://genetics.bwh.harvard.edu/pph2/) and MutationTaster [Schwarz et al., 2010] (Version 2, 2012, http://www.mutationtaster.org/), as well as conservation with PhyloP scores [Pollard et al., 2010]. Variants that were classified as pathogenic or likely pathogenic according to the ACMG guidelines were validated using Sanger sequencing in the proband and segregation was performed in family members when available. Paternity testing in Family 6 used the PowerPlex 16 HS System (Promega, Madison, WI) run on an ABI3730 capillary electrophoresis system (Australian Genome Research Facility (AGRF), Westmead, Australia). Primers for Sanger sequencing were designed on Primer3 (http://biotools.umassmed.edu/bioapps/primer3_www.cgi), and are available upon request. Nucleotide numbering uses +1 as the A of the ATG translation initiation codon in the reference sequence, with the initiation codon as codon 1. All variants reported in this manuscript have been submitted to ClinVar (http://www.ncbi.nlm.nih.gov/clinvar/).

Results

Targeted Region Analysis

The samples using the TruSeq custom amplicon approach (probands 1–38) had a mean depth of coverage of the targeted regions of 940X across all samples, with on average 91% of amplicon regions being covered by ≥ 15X (Supp. Table S3A). The 26 patients sequenced using the TruSight clinical exome approaches had an average coverage of 160X across the targeted regions, and 95% of captured regions covered ≥ 15X (Supp. Table S3B).

While the TruSeq amplicon‐based method achieved a very high average coverage of 940X, there were several gaps with poor coverage (Supp. Table S3C and D). These gaps were mainly in genes that were very GC rich, or contained repetitive regions, which were not adequately covered by the amplicon PCR‐based TruSeq method. Compared with this, the capture‐based TruSight library approach had a lower average coverage of 160X, but this was more even, with fewer and smaller gaps (Supp. Table S3D). Gaps in GJA3 and MAF underwent Sanger sequencing, due to the importance of gap junction genes in cataracts, and our discovery of three new mutations in MAF in this cohort. This led to detection of only one additional mutation, a GJA3 mutation in Family 29.

Likely Causative Variants in 73% of Familial and 68% of Sporadic Congenital Cataract Cases

Overall, likely disease‐causative variants were found in 33 cases in the 46 families we studied, indicating a mutation detection rate of over 70% using NGS (Table 1). This included one family (Family 6), where a mutation in a different cataract gene was found in the affected father, with confirmed paternity, compared with the causative disease gene identified in his two sons. There was a mutation detection rate in the familial cases of ∼73% (16/22), and a detection rate of 68% in the sporadic cases (17/25, including the affected father with a separate mutation in Family 6). Out of the 34 mutations found, 23 (∼68%) were novel. The mutations were spread over 17 out of the 32 genes analyzed (Fig. 1A). Variants in seven crystallin (CRYAA, CRYAB, CRYBB1, CRYBB2, CRYBB3, CRYGC, and CRYGD) and both gap junction genes (GJA3 and GJA8) were the most frequently identified in our cohort, accounting for 20/33 (∼61%) cases (Table 1; Fig. 1A). In addition, there were likely causative variants found in: MAF in three cases; PAX6, MIP, NHS, and BCOR in two cases each; and VIM and BFSP1 in one case each. The majority of familial cases were due to autosomal‐dominant mutations in crystallin or gap junction genes, with two autosomal‐recessive cases due to mutations in these genes, and two families were found to have X‐linked NHS mutations (Fig. 1B). Sporadic cases were mostly due to de novo autosomal‐dominant mutations in a wide variety of genes including crystallins, gap junctions, MAF, MIP, and VIM, and there were two de novo X‐linked BCOR cases (Fig. 1B). In almost two‐thirds (20/33) of the cases where likely causative variants were identified, this resulted in new information about the inheritance pattern for the family (Table 1). Eight variants of unknown significance were found, five of which were the only variants in that family (Supp. Table S4). According to the ACMG mutation guidelines, five of these were reclassified as of uncertain significance, one was likely benign, and two were benign. Nine families had no variants of interest.

Figure 1

Genes and inheritance patterns identified with NGS in apparently nonsyndromic sporadic and familial congenital cataract cases. A: Mutations were found in 17 different genes in apparently nonsyndromic sporadic and familial congenital cataract cases. These were in genes that encoded crystallins, gap junction, transcription factor, other structural proteins, and X‐linked syndromal proteins. The relative proportions are illustrated in this diagram. B: Mutation‐positive cases in familial and sporadic cases and the inheritance patterns. The 16 mutation‐positive familial cases comprised two X‐linked (XL, dotted), three autosomal‐recessive (AR, horizontal stripes), and 11 autosomal‐dominant (AD, diagonal stripes) diagnoses. In the sporadic cases with mutations, the inheritance pattern was revised to two cases of de novo X‐linked mutations in BCOR, and 15 de novo AD cases, including the father from Family 6 who had a separate genetic answer (mutation in MAF) to his two affected sons with NHS.

Novel Missense, “Run‐On,” and Frameshift Mutations in Key Domains of the Crystallin Genes

Thirteen of the patients with likely causative mutations were found to have variants in the crystallin genes (13/33, 39%), eight in familial and five in sporadic cases, including 10 novel variants (Fig. 1A and B; Table 1). Seven of the eight familial cases were AD, with one AR. The crystallin gene mutation contribution in our AD families (50%, 7/14), is consistent with previous reports [Hansen et al, 2007; Reis et al 2013]. All sporadic cases were shown to be likely new AD cases. Most of these mutations disrupted the Greek Key domains of the crystallin proteins, vital in the protein's folding and maintenance of lens transparency [Vendra et al., 2013].

Two novel missense mutations on the same allele were found in CRYBB2 in Patient II.1 from Family 35 (Fig. 2). Both missense mutations (c.343C>A p.(Pro115Thr) and c.355G>A p.(Gly119Arg)), occurred on the same allele, as revealed by the NGS short read data and were de novo in this sporadic case (Fig. 2D and E). These mutations occurred in exon 5 of CRYBB2 encoding the third Greek Key domain. A number of single missense mutations have been identified in this region of CRYBB2, as well as heterozygous nonsense mutations, but our patient is more severely affected than other reported cases [Litt et al., 1997; Gill et al., 2000; Vanita et al., 2001; Hilal et al., 2002; Yao et al., 2005; Bateman et al., 2007; Devi et al., 2008; Lou et al., 2009; Santhiya et al., 2010] with marked reduction in eye size (HCD 7–8mm and AXL 17mm at age 2 years) and moderately severe visual impairment with VA 6/80 and bilateral nystagmus. In our patient, it is likely that the mutation of both amino acids p.([(Pro115Thr; Gly119Arg)]), in particular the very small nonpolar glycine to a hydrophilic large arginine, within a short distance in this Greek Key motif may have a more severe impact on protein folding than only one mutation alone.

Figure 2

Two novel missense CRYBB2 mutations on the one allele in proband in Family 35, and other CRYBB2 mutations found in this study. A: This schematic shows the encoded domain structure of CRYBB2 (NM_000496.2). This protein contains four “Greek Key” domains between the N and C terminals. Mutations found in this study are illustrated above the schematic, with the two novel missense p.(Pro115Thr) and p.(Gly119Arg) found in Family 35, p.(Ser186Pro) found in Family 13, and p.(Trp195Gly) in Family 37. B: Pedigree from Family 35, with proband highlighted in II:1. C: Protein sequence alignments show that both affected residues, highlighted in gray, are in a highly conserved part of the third Greek Key domain. D: The proband had two de novo missense CRYBB2 mutations, confirmed on Sanger sequencing (red arrows). E: Next‐generation reads reveal that the variants (highlighted in blue) were both consistently in the same reads, showing that they are in cis.

In our series, we report the first identified “run‐on” mutation affecting CRYBB3 [Graw, 2009]. In an affected father and son in Family 39 (Supp. Fig. S1; Table 1), the novel heterozygous missense mutation changed the stop codon of CRYBB3 to arginine and was followed by “run on” into the 3′UTR and an additional 40 residues (c.634T>C p.(*212Argext*40)). This is predicted to cause a lengthened, abnormal CRYBB3 protein likely to disrupt protein folding and function due to the abnormally large product. Two novel heterozygous frameshift mutations in crystallins were identified, one in CRYGC and the other in CRYGD. The novel heterozygous deletion and insertion mutation in CRYGC (c.328_329delinsT p.(Pro110Serfs*37)) is predicted to lead to a frameshift mutation causing familial autosomal‐dominant congenital lamellar cataracts in Family 22 (Supp. Fig. S1; Table 1). The mutation leads to an abnormally short protein, as does a premature stop codon mutation in the same region in another autosomal‐dominant cataract family [Yao et al., 2008], with both mutations predicting loss of the third and fourth Greek Key domains of CRYGC. In the sporadic case in Family 41, a novel de novo heterozygous frameshift mutation was found in CRYGD, (c.448dup p.(Asp150Glyfs*3)) leading to a premature stop codon and deletion of most of the fourth Greek Key and all of the C‐terminal domains of the protein.

GJA3 and GJA8 Mutations Including a Novel Missense Variant with a Severe Phenotype, and a Novel Recessive Frameshift Mutation

Seven of the patients with likely causative mutations in our study were found to have variants in the gap junction genes, GJA3 and GJA8 (7/33, 21%), three in familial and four in sporadic cases, including four novel variants (Table 1; Figs. 1 and 3C). The novel de novo heterozygous GJA8 missense mutation (c.151G>A p.(Asp51Asn)) in the proband in Family 45 (Supp. Fig. S1; Table 1; Fig. 3A–D), was associated with the most severe phenotype yet described in a patient with a gap junction mutation. This patient was markedly affected with bilateral microphthalmia, left more than right, congenital cataracts and sclerocornea of the left eye. The GJA8 missense mutation (c.151G>A p.(Asp51Asn)) occurred in a highly conserved amino acid of the first extracellular domain (EC1) of GJA8 (Fig. 3D). All the other reported mutations in GJA8, including those in the same domain, are associated with congenital cataracts ± microcornea, with a variety of phenotypes, intrafamilial variability, and none with associated sclerocornea or this degree of microphthalmia [Minogue et al., 2009; Beyer et al., 2013; Ge et al., 2014]. We postulate that the p.(Asp51Asn) mutation in this patient may highlight the importance of the Asp 51 amino acid in the EC1 domain in the function of GJA8, and note that the Gja8 p.Ser50Pro mutation in this region in mice is associated with altered interaction with the normal allele, interrupting the process of primary lens fiber cell formation [Xia et al., 2006]. Also, it broadens the phenotypic correlation of GJA8 mutations, highlighting the importance of screening this gene, not only in cataract patients, but also where the main phenotype may be listed as anterior segment abnormality or microphthalmia.

Figure 3

GJA8 mutation in Family 45 caused a severe cataract, sclerocornea, microphthalmia phenotype, and other GJA8 mutations found in this study. A: Pedigree of family 45, with the proband (II:1) highlighted. He was found to have a heterozygous missense mutation in the first extracellular domain (EC1) of GJA8, p.(Asp51Asn), and has the most severe GJA8 ocular phenotype reported to date. B: Left eye image of the proband in Family 45, showing his severe microphthalmia and sclerocornea. C: The schematic shows the encoded domain structure of GJA8 (NM_005267.4). This protein spans the cellular membrane and has a signal peptide (SP), four transmembrane (TM), two extracellular (EC), and two intracellular (IC) domains. Mutations found in GJA8 in this study are shown above the schematic, including the p.(Asp51Asn) mutation in Family 45 in the EC1 domain. The p.(Ile31Hisfs*18) recessive mutation found in Family 14 is the earliest frameshift mutation yet reported in this gene. Other recessive mutations reported in the literature are shown below the gene, and the star highlights recessive mutations. The p.(Ala40Val) and p.(Trp45Ser) heterozygous mutations were found in Families 2 and 20, respectively. D: Protein sequence alignments show that the affected residue in the proband of Family 45, highlighted in gray, lies in a highly conserved region of the EC1 domain.

A novel GJA8 frameshift mutation was identified in Family 14 (Supp. Fig. S1; Table 1). The affected mother and daughter were referred with suspected autosomal‐dominant cataracts, although it was noted that the daughter's parents were consanguineous. The affected daughter was found to have a novel homozygous (c.89dupT p.(Ile31Hisfs*18)) mutation in the first transmembrane domain of GJA8 (Table 1; Fig. 3C). She had congenital cataracts and microcornea, whereas her heterozygous mother had a milder phenotype with nuclear dot opacities. The father was unavailable for analysis. While both dominant and recessive mutations in GJA8 can cause cataracts, recessive mutations are less frequently reported. Interestingly, two out of the three reported recessive mutations are frameshift mutations [Ponnam et al., 2007; Schmidt et al., 2008], and one is a missense mutation [Ponnam et al., 2013] (Fig. 3C). The frameshift mutation we have identified is the earliest recessive mutation in the protein (Fig. 3C). As in our case, affected homozygous individuals in the literature all had severe dense congenital cataracts. The extent to which heterozygous carriers in the literature were affected is variable, with some reported as unaffected [Ponnam et al., 2007], or variably affected with mild nuclear dot opacities [Schmidt et al., 2008; Ponnam et al., 2013] as in the heterozygous mother in our family.

Mutations in X‐Linked Syndromal Cataract Genes NHS and BCOR in Four Families

The diagnosis of mutations in the X‐linked syndromal cataract genes NHS and BCOR highlights the genetic heterogeneity of congenital cataracts. Probands from Families 6 and 40 were diagnosed with Nance Horan syndrome (NHS) due to NHS mutations found in this study (Supp. Fig. S1; Table 1; Figs. 4A–E and 5A) after initial referrals with a presumed diagnosis of autosomal‐dominant cataracts. The proband from Family 6 was the elder of two brothers, aged 9 and 4 years, with congenital cataracts and microcornea (Fig. 4A). The father had cerulean cataracts (Fig. 4S and T), the mother had clear lenses and the family was offered clinical genetics review but declined. The proband in Family 6 had a novel frameshift hemizygous mutation in NHS (c.2707delG, p.(Glu903Asnfs*4)) (Fig. 5A). The mutation was also present in his affected brother and was heterozygous in their mother. When the brothers were reviewed, the typical facial and dental features were apparent, and they were beginning to exhibit intellectual delay and attention‐deficit consistent with a diagnosis of NHS. The male proband in Family 40 (Supp. Fig. S1; Fig. 4B) had congenital cataracts, left foot insertional polydactyly, and mild learning difficulties. His older sister and mother had been found to have juvenile‐onset cataracts (Fig. 4C–E). On clinical genetics review, he was found to have facial and dental features suggestive of NHS. A previously reported nonsense mutation (c.3624C>A, p.(Cys1208*)) in NHS [Huang et al., 2007] (Fig. 5A) was found in the proband and segregated appropriately in an X‐linked pattern in his sister and mother, again providing the correct diagnosis and inheritance pattern information for this family. While the digital anomaly of brachymetacarpalia has been seen in NHS [Ding et al., 2009], this is the first report of insertional or any form of polydactyly in this condition.

Figure 4

Phenotypic variability of the cataracts and associated ocular phenotypes in this study. This figure shows images of congenital cataracts and other ocular features found in this study and highlights the diverse heterogeneity of congenital cataracts, as well as the importance of accurate phenotyping, fundoscopy, and electrophysiology to clarify the diagnosis. Images are from families diagnosed with mutations in NHS, Families 6(A) and 40(B‐E); PAX6, Families 10(F) and 18(G‐R); MAF, father from Family 6(S&T); MIP, Family 32(U); and VIM, family 25(V). Image A displays the right sided cortical cataract from the proband (II:1) from Family 6, diagnosed with NHS. The other family with NHS (Family 40) is represented in images B–E. B: Right eye from proband (III:3, Family 40), with postsurgical changes following cataract extraction. The variable features demonstrated in obligate carrier females from this family are demonstrated in C, with subcapsular cataract from the mother (II:3) and D (right) and E (left) coralliform cataracts in the sister (III:1). Image F: Cortical congenital cataracts of proband (II:1) from Family 10, diagnosed with a missense PAX6 mutation in the paired domain. Images from Family 18, also diagnosed with a PAX6 mutation, are represented in images G–R, revealing the panocular features of PAX6‐related disease. These are from the proband (III:3) showing: G: anterior polar cataract; J: mild limbal stem cell failure with peripheral corneal pannus; M: altered macular reflex (*); P: OCT of foveal hypoplasia (red bracket) with loss of the foveal pit. The mother in Family 18 (II:2) also demonstrates multiple PAX6‐associated abnormalities with: H: limbal stem cell failure with increased corneal pannus; K: mild iris hypoplasia and mild ectropion uvea (arrow); N: reduced macular reflex (*); Q: OCT showing foveal hypoplasia with loss of the foveal pit (red bracket). The sister in Family 18 (III:2) also has: I: anterior polar cataract; L: ectropion uvea (arrow); O: reduced macular reflex (*); and R: OCT showing loss of foveal pit (red bracket). S (right) and T (left): Cerulean blue dot cataracts from the father (I:1) from Family 6, found to have a novel missense MAF mutation. U: Lamellar cataract of proband (II:1) in Family 32, found to have a novel frameshift mutation in MIP. V: The image demonstrates bilateral total cataracts as they appeared on presentation at 11 months of age in the proband of Family 25 (II:1), found to have a novel frameshift mutation in VIM.

Figure 5

Mutations in NHS and BCOR revealed syndromic conditions, and PAX6 mutations indicated more complex ocular phenotypes. A: Exonic structure of NHS (NM_198270.3) with previously reported mutations marked by symbols below the gene. KEY: ^, nonsense; *, frameshift; #, splice site; ∼, missense mutations. The solid black lines delineate reported exonic deletions. Mutations found in this study are demonstrated above the gene, from Families 6 and 40. B: Exonic structure of BCOR (NM_001123385.1) with previously reported mutations marked by symbols below the gene. KEY: ^, nonsense; *, frameshift; #, splice site; ∼. missense mutations. The solid black lines delineate reported exonic deletions. Mutations found in this study are demonstrated above the gene, from Families 44 and 7. C: Exonic and protein domain structure of PAX6 (NM_0001604.4). The exons are numbered, with the alternatively spliced exon 5a highlighted above the gene. Below the gene, the protein domains are highlighted (UTR, untranslated region; PST, Proline, Serine, Threonine rich domain). The two mutations found in this study, from Families 18 and 10, are marked above the gene. Due to the large number of mutations found in this gene, they are not illustrated in the figure. A curated database of PAX6 mutations can be found at LOVD (http://www.lovd.nl/PAX6).

In two families, de novo novel frameshift mutations were found in BCOR, the gene responsible for oculo‐facio‐cardio‐dental (OFCD) syndrome (Fig. 5B). In Family 44 (Supp. Fig. S1), the female proband had sporadic cataracts and cleft palate. She was found to have a novel de novo heterozygous frameshift deletion in BCOR (c.1136_1139del p.(Val379Alafs*62)). She was subsequently diagnosed with atrial septal defect and was later found to have dental and facial features consistent with OFCD. In Family 7, the female proband with microphthalmia, cataracts and glaucoma, was found to have a de novo novel frameshift BCOR mutation (c.4390_4393del p.(Glu1464Profs*19)). This mutation was not found in her son who had aniridia and had been found in a separate study to have a paternally inherited heterozygous mutation in PAX6 [Willcock et al., 2006] (Supp. Fig. S2).

Mutations in PAX6 in Two Families with Complex Cataract Phenotypes

Two heterozygous variants were found in PAX6, a familial frameshift mutation in Family 10 and a de novo novel missense mutation in Family 18 (Supp. Fig. S1; Table 1; Figs. 4F–R and 5C). They add to the increasing evidence that patients with a variety of eye phenotypes such as cataract, glaucoma, anterior segment dysgenesis (ASD), and nystagmus should have PAX6 testing performed, and this gene should be included in NGS panels for congenital cataracts. This is particularly important if clinical findings are atypical or visual impairment more severe than expected. The proband from Family 10 was referred with an atypical presentation of cataracts, with additional nystagmus and congenital ptosis and astigmatism. A frameshift mutation in PAX6 (c.1119del, p.(Thr374Profs*5)) was identified and his affected sister and mother had the same PAX6 mutation and displayed similar clinical features, although they lacked nystagmus (Fig. 4G–R). This mutation is recently reported in a family with a similarly complex and variable cataract phenotype including cataracts, ptosis, iris hypoplasia, corneal opacification, and diabetes [Peter et al., 2013]. A novel de novo PAX6 missense mutation (c.239T>A p.(Ile80Asn)) was identified in Patient II.1 in Family 18 (Supp. Fig. S1; Table 1; Figs. 4F and 5C). He also had nystagmus and reduced visual acuity and required strabismus surgery. The PAX6 mutation explains his complex phenotype and may provide an explanation for the poorer than expected outcome following his cataract surgery.

Three New Mutations Found in the Basic Zipper Region of MAF

Three novel heterozygous mutations in MAF were identified in Families 6, 27, and 30 (Fig. 1; Table 1), one in the EHR (extended homology region, EHR), and two in the bZIP (b‐zipper) domain (Fig. 6A and B). The father in Family 6 had cerulean cataracts (Fig. 4S and T), and we undertook NGS in this man when the X‐linked cause of his boys’ cataracts was found. This led to identification of a novel heterozygous missense mutation in MAF (c.819G>C p.(Glu273Asp)), the first identified in the EHR. The proband from Family 30 had sporadic congenital cataracts and a heterozygous missense c.880C>T p.(Arg294Trp) mutation in the basic region of MAF, very close to where previous mutations were reported (Fig. 6B). Another novel heterozygous mutation (c.915C>T p.(Cys305Trp)) was found in Patient II.1, from Family 27 (Fig. 6B). This patient had sporadic congenital cataracts and severe glaucoma requiring surgical intervention.

Figure 6

Novel mutations identified in MAF, VIM, BFSP1, and MIP. A: Schematic showing the encoded structure of MAF (NM_001031804.2). MAF has an N‐terminal and b‐zipper domain, and the detailed section of b‐ZIP reveals three evolutionarily conserved subdomains with the EHR, basic and leucine zipper regions. B: The b‐zipper domain is expanded to reveal these three subregions. Previously reported mutations in the basic and leucine zipper regions are demonstrated above the figure. The extensive amino acid conservation of the b‐zipper region is revealed, and the highly conserved three new mutations found in this study in MAF, highlighted in gray, from Families 6, 30, and 27. C: Schematic of encoded structure of VIM (NM_003380.3), which comprises a head, rod, and tail domain. The rod domain has three coil subdomains; 1A, 1B, and 2. A novel frameshift mutation was found in the proband of Family 25 (II:1), in the head domain (labeled above head domain). The only other mutation, a heterozygous missense mutation in the rod domain, is also marked, below the rod domain. D: Schematic of encoded structure of BFSP1 (NM_001195.4), which comprises a head, rod, and tail domain. Similar to VIM, the rod domain of BFSP1 also has coil subdomains 1A, 1B, and 2. Compound heterozygous mutations were found in the proband from Family 46, highlighted above the second coil and tail domains. The other mutations reported in the literature are labeled below the gene; a recessive frameshift mutation in the rod domain, and a heterozygous missense mutation in the tail domain in an autosomal‐dominant family. E: Schematic of encoded structure of MIP (NM_012064.3), which is a protein with four cytoplasmic (C) domains, six transmembrane (TM) domains, and three extracellular (EC) domains. The two mutations found in this study from Families 28 and 32 are illustrated above the gene, along with other mutations reported in the literature to date below the gene.

Novel Mutations in Lens Structural Genes VIM, BFSP1, and MIP

A novel heterozygous frameshift mutation was identified in the Intermediate Filament gene, VIM (c.15del p.(Val6Cysfs*26)) in Patient II.1, Family 25 (Supp. Fig. S1; Table 1; Figs. 4V and 6C). This is only the second reported mutation in VIM. Compound heterozygous mutations in BFSP1 were found in Family 46 (Supp. Fig. S2; Table 1), representing only the second recessive mutations reported in this gene (Fig. 6D). This family had two affected sons with cataract and microcornea from a consanguineous family. The proband was found to be compound heterozygous for mutations in BFSP1 (c.1492del p.(Ser498Leufs*24) and c.812T>C p.(Ile271Thr)). Segregation studies on the mother revealed she was heterozygous for the frameshift mutation, confirming that these are in trans. The other son and father were unavailable for study. A patient with sporadic cataracts (Patient II.2, Family 28) and a mother and daughter (Family 32) with autosomal‐dominant lamellar cataracts (Supp. Fig. S1; Table 1; Fig. 4U) were found to have heterozygous mutations in MIP (Fig. 6E). Our patient from Family 28 had a previously reported de novo missense mutation in MIP (c.97C>T p.(Arg33Cys)) [Gu et al., 2007], affecting the first extracellular loop of the protein (Fig. 6E). The affected individuals in Family 32 had a novel heterozygous variant in MIP, a 14‐bp insertion leading to a premature stop codon, c.597_598insGGGAACATTCCACT p.(Asn200Glyfs*12).

Discussion

This study highlights the benefits of an NGS approach with analysis of a targeted group of genes in the genetically heterogeneous condition of congenital cataracts. We achieved a mutation detection rate of over 70% in 46 sporadic and familial cases with apparently nonsyndromic, bilateral congenital cataracts. Similar detection rates were found in the sporadic and familial cases. Several novel mutations were found in a variety of genes including the crystallin and gap junction genes, more rarely reported disease genes including MAF, VIM, and MIP, and the syndrome‐associated genes, NHS and BCOR. This approach provides a significant advance for patients in provision of genetic diagnosis and new recurrence risk information, as well as identification of clinically unrecognized syndromic associations.

NGS Is a Powerful Tool in Sporadic and Familial Cataract Cases and in Identification of Subtle Syndromal Cases

Sporadic congenital cataract patients present diagnostic challenges when attempting to identify a genetic aetiology, and make up the majority of congenital cataract cases, since only around 18% of congenital cataract cases have a family history [Wirth et al., 2002]. While possible empiric‐based explanations for sporadic cases include de novo mutations in autosomal‐dominant genes, autosomal‐recessive inheritance, or X‐linked mutations, each of these possible modes of inheritance have vastly different recurrence risk likelihoods for the parents and affected individual, and are of limited use without a specific genetic diagnosis. Our study demonstrates that the majority of the mutations identified in sporadic bilateral congenital cataract cases were due to de novo heterozygous mutations in autosomal‐dominant genes (15/17, 88%) (Fig. 1B). In the remaining two sporadic cases, we found de novo X‐linked mutations in BCOR in females. This raises the possibility of asymptomatic or undiagnosed female carriers, which may not be easily diagnosed without the aid of an NGS approach. A high de novo mutation rate in this condition is important information for couples seeking reproductive information after an affected child. It indicates a low recurrence rate, with a small chance of gonadal mosaicism, and may decrease parental anxiety when planning for another child. Prenatal testing can be offered to exclude gonadal mosaicism, further enhancing reproductive confidence. In addition, probands with de novo heterozygous autosomal‐dominant mutations do themselves have a 50% chance of passing on their mutation to future offspring, whereas females with de novo X‐linked mutations are at risk of more severely affected male offspring or recurrent miscarriages in the case of BCOR mutations. In both cases, affected individuals may wish to utilize reproductive options such as prenatal testing or preimplantation genetic diagnosis. These results demonstrate that for sporadic cataracts, a targeted NGS strategy is useful for couples who have had an affected child, as well as the affected individuals themselves, to provide accurate recurrence and transmission risk counselling.

Similar difficulties exist in counseling families with familial cataracts, as pedigree information alone may not be accurate or extensive enough to provide accurate information for inheritance risk. In this study, three familial cases had revised genetic information due to the genetic results from this study (Table 1). This included a case of autosomal‐recessive GJA8 mutations (Family 14) and an X‐linked NHS mutation (Family 40) in families that appeared autosomal‐dominant on pedigree information. In Family 6, an unexpected answer with mutations in two different cataract genes, NHS and MAF, was found. The possibility of two entirely different genes responsible for the same condition running in a family is often overlooked, and we have now demonstrated this in Family 6, as well as in Family 7, where a BCOR mutation was found in the mother of this previously reported family, with a PAX6 mutation in other family members [Willcock et al., 2006]. These cases highlight the ability of an NGS‐based approach to provide accurate diagnosis that cannot be achieved through phenotyping alone.

This study has also shown the benefit of using an NGS approach in young patients and situations where syndromal diagnosis may not be clear‐cut or the clinical signs are still evolving. Male patients with NHS have bilateral severe congenital cataracts, and most have microcornea, as well as nystagmus, and glaucoma, which can worsen postoperatively [Ding et al., 2009]. Our findings in Families 6 and 40 highlight the need for increased clinical attention to this condition in young male patients where the full hand of clinical features may not be immediately obvious. Also, findings in the female carriers in these families demonstrated the variability and difficulty of clinical diagnosis in this condition where females may be entirely disease‐free or have mild lens opacities not obvious on initial review (Fig. 4C–E). Similarly, Families 7 and 44 highlight the importance of X‐linked syndromal diagnosis in females with BCOR mutations, associated with OFCD syndrome. Clinical features can be subtle especially in young patients, and cardiac status should be monitored.

NGS Identifies Novel Variants and Phenotypes Associated with Mutations in Crystallin and Gap Junction Genes, and in Rarely Reported Cataract Genes

Mutations in the crystallin and gap junction genes have a major role in maintaining lens transparency and were the most frequent mutations found in this study (Fig. 1). Two‐thirds (14/21) of these mutations were novel (Table 1), including two missense heterozygous mutations in CRYBB2 on the same allele in the proband from Family 35 with a severe phenotype (Fig. 2). Also, a striking novel phenotype of microphthalmia, sclerocornea, and cataracts was seen in the proband from Family 45, due to a novel missense mutation, p.(Asp51Asn) in GJA8 (Fig. 3).

This study also found six novel mutations in MAF, VIM, BFSP1, and MIP, highlighting the need to include these rarely reported genes in any multigene congenital cataract panel for examination of patients with this genetically heterogeneous condition. Three of these novel mutations were found in MAF (Fig. 6B). Up until this report, five mutations have been found in the basic region of MAF, in patients with isolated ocular abnormalities, all clustered around the b‐zipper domain [Jamieson et al., 2002, 2003; Vanita et al., 2006; Hansen et al., 2009; Narumi et al., 2014] (Fig. 6B). A variety of cataracts and ocular malformations have been reported in these patients including microcornea, iris coloboma, and ASD. The relative scarcity of reported mutations in this gene may reflect the difficulties in conventional Sanger sequencing, as it is very GC rich. In this report, we identified the first mutation affecting the EHR of MAF, a region that assists with specificity of DNA binding by MAF [Kerppola and Curran, 1994], as well as two novel mutations in the basic region of the DNA binding domain (Fig. 6B). We did not identify any mutations in the transactivating N‐terminal domain of MAF, where mutations have been recently reported in patients with syndromic cataracts, with additional features of deafness, intellectual disability, seizures, and dysmorphism [Niceta et al., 2015]. Our findings highlight an emerging genotype–phenotype correlation associated with MAF mutations, where heterozygous missense mutations in the b‐zipper domain are associated with isolated ocular findings, whereas heterozygous missense mutations in the N‐terminal transactivation domain are associated with a syndromic cataract phenotype.

We report the second cataract‐associated mutation in VIM in Family 25 (Figs. 4V and 6C). Vimentin is a highly conserved type III intermediate filament protein, expressed in mesenchymal cells and tissues including the lens [Herrmann and Aebi, 2004]. The previous Vimentin mutation is in a family with autosomal‐dominant pulverulent cataracts, c.596G>A p.(Glu151Lys) [Muller et al., 2009], and affects the alpha‐helical coil 1B rod segment of VIM. Cell‐based assays suggest this mutation leads to defects in Vimentin assembly [Muller et al., 2009]. In Family 25, a heterozygous frameshift mutation was found in the head domain. This mutation is predicted to lead to premature truncation of the protein proximal to the rod domain, therefore abolishing any functional protein.

In another lens structural protein with a similar domain structure, the beaded filament gene BFSP1, we report the second family with recessive mutations in this gene, in Family 46 (Fig. 6D). The beaded filament proteins BFSP1 and BFSP2 are exclusively expressed in the eye lens. They assemble together to form heteropolymeric filaments, to create the unique cytoskeletal structure of the beaded filament in all vertebrate lenses [Perng et al., 2007]. The first reported mutation was a recessive homozygous deletion of exon 6 [Ramachandran et al., 2007], predicted to lead to a frameshift and loss of part of the rod and full tail region of the protein p.(Thr246fs*6). The second was a heterozygous mutation in a five‐generation dominant Chinese family due to a missense mutation in exon 7 (c.1042G>A p.(Asp348Asn)) in the tail domain. Both mutations point toward the importance of the tail region in the protein's function. In our patient (Family 46), the missense mutation affects the rod domain and the frameshift is predicted to lead to loss of part of the tail domain, highlighting the critical nature of these regions in the functions of BFSP1 required for lens clarity (Fig. 6D).

While approximately 15 heterozygous missense, frameshift, and splice‐site mutations in the water channel gene MIP are reported in cataract patients [Shiels, 2012] (Fig. 6E), it is generally a less frequently investigated gene. In our study, we found two mutations in MIP, a previously reported mutation in a sporadic case (Family 28) and a novel frameshift mutation, p.(Asn200Glyfs*12), in a dominant family (Family 32, Fig. 4U). The frameshift mutation is the most proximal mutation yet identified in MIP, and would be expected to lead to complete loss of the sixth transmembrane domain and C‐terminal of this water transport protein (Fig. 6E).

NGS Affects Management in PAX6‐Associated Cataract Cases

In two congenital cataract cases in this study, mutations were identified in PAX6, leading to significant new information for the families and changes to management. Patient III.1 in Family 10 (Table 1; Figs. 4G–R and 5C; Supp. Fig. S1) was referred with his partner for prepregnancy genetic information, with possible diagnoses of autosomal‐dominant cataracts and congenital motor nystagmus. The diagnosis of a PAX6 frameshift mutation (c.1119del p.(Thr374Profs*5)) led to a unifying diagnosis explaining the patient's nystagmus and cataracts, and further clinical ophthalmic review revealed associated ophthalmic features of limbal stem cell failure and ectropion uvea, features previously identified in a family with “run‐on” PAX6 mutation affecting the stop codon [Willcock et al., 2006]. He also had subtle hypoplasia of the irides and loss of the foveal pit confirmed with optical coherence tomography (OCT) (Fig. 4G, J, M, and P). His affected mother was found to have the same mutation as well as the additional complication of glaucoma. Both the mother (Fig. 4H, K, N, and Q) and sister (Fig. 4I, L, O, and R) were also found subsequently to have evidence of iris and foveal hypoplasia on fundoscopy and OCT. Identification of a heterozygous PAX6 mutation in this family has made a significant difference in their management, with regular monitoring for PAX6‐associated complications such as glaucoma, limbal stem cell‐associated corneal abnormalities, and foveal hypoplasia, which all have a significant impact on long‐term vision. This result also clarified the recurrence risk advice for affected family members for this autosomal‐dominant condition. The proband in Family 18 had a de novo missense mutation in the paired domain of PAX6 (Figs. 4F and 5C), adding to the increasing evidence in the literature that missense mutations in the paired domain of PAX6 can lead to a milder nonaniridia phenotype [Park et al., 2012], possibly by affecting DNA binding to transcriptional targets. Similar to the PAX6 frameshift family, this result has important implications for this patient. He had surgery for a large exotropia before his molecular diagnosis, and following surgery he has persistent diplopia. This failure of fusion may be related to his underlying pan‐ocular disorder due to his PAX6 mutation.

Benefits of Higher Depth and Evenness of Coverage in Cataract Mutation Detection

Our targeted NGS approach highlights the benefits of improved evenness and depth of coverage to maximize the mutation detection rate. The capture‐based TruSight library approach had a lower average coverage, but this was more even with fewer and smaller gaps, compared with the TruSeq amplicon‐based approach (Supp. Table S3). Using the expanded TruSight One library, five additional mutations were found in the 17 original “TruSeq Negative” patients, including in key genes with gaps on the amplicon approach including BCOR, GJA3, and MAF. The higher average coverage and read depth, able to be achieved economically with a targeted approach, led to a greater detection rate than a previous study using WES [Reis et al., 2013] in familial congenital cataract patients (mutations found in 39% of families). A study using targeted NGS in a cohort with more consanguineous and overtly syndromic congenital cataract patients [Gillespie et al., 2014] compared with our study, found a similar mutation detection rate to our study. Our work shows that apparently nonsyndromic, sporadic, and familial cases of congenital cataracts, also benefit from a targeted NGS approach for improved diagnosis and management.

Conclusion

This study highlights the clinical utility of targeted NGS in the setting of genetically heterogeneous congenital cataract patients. The 70% mutation detection rate shows that NGS is a useful test for both familial and sporadic apparently nonsyndromic cases. This has led to revision of diagnoses, accurate recurrence risk counseling, with an impact on management for each family. Overall, it demonstrates that in patients with congenital cataracts, whether sporadic or familial, a NGS‐based test provides significant additional diagnostic information and is warranted to improve management and provide accurate genetic counseling.

Disclaimer: Supplementary materials have been peer‐reviewed but not copyedited.

Supporting Information

Click here for additional data file.