This study was approved by the Research Ethics Committee of Wenzhou Medical University Laboratory (approval number: 2021–239-k-209) and adhered to the tenets of the Declaration of Helsinki. A total of 40 individuals with clinically confirmed bilateral congenital cataracts were recruited from the Pediatric Cataract Center of Wenzhou Medical University Eye Hospital (Wenzhou, China). Written informed consent was obtained from all adult participants and from the parents or legal guardians of minors prior to enrollment. All data were handled in a de-identified manner. Participants were assigned study-specific codes, and no personally identifiable information was included in the manuscript or Supplementary Material. Pedigrees presented in the Supplementary Material are anonymized and do not contain information sufficient to identify individual participants. Detailed family histories and medical records were carefully collected. The presence and type of cataract phenotype in both affected and unaffected individuals were confirmed by slit-lamp biomicroscopy. Patients with a history of intrauterine infection, drug exposure, metabolic disorders, or malnutrition were excluded. For genomic DNA analysis, a 2 mL sample of peripheral venous blood or oral mucosal tissue was collected. Genomic DNA was extracted using either the QIAGEN Blood DNA Kit (QIAGEN, Germany) or the Invitrogen™ MagMAX™ DNA Multi-Sample Ultra 2.0 Kit (Thermo Fisher Scientific, Norway), following the manufacturer’s instructions. All probands presented with bilateral congenital cataracts, identified at birth or diagnosed within the first year of life, with timing supported by medical records, a consistent parental report of onset, or both. Some patients also showed microcornea and other ocular features, including microphthalmia, nystagmus, and glaucoma. In addition, a few patients exhibited extraocular manifestations, such as dental dysmorphologies, proteinuria, micrognathia, and polycystic kidney disease (Supplementary Table S1).

For genetic analysis, genomic DNA from affected individuals underwent WES. Library preparation was performed using the Twist Human Core Exome Kit (Twist Bioscience, United States), and sequencing was carried out on the NovaSeq 6000 platform (Illumina, San Diego, United States). Sequence reads were aligned to the human reference genome (hg19/GRCh37). The protocols for next-generation sequencing and downstream data analysis, including copy-number variation analysis, were described previously (Huang et al., 2017). In summary, variants were filtered to retain only novel variants that were absent from the public control databases Kaviar (https://db.systemsbiology.net/kaviar/) and the Genome Aggregation Database (gnomAD v4.1.0, http://gnomad.broadinstitute.org). In addition, rare variants with a gnomAD allele frequency <0.0001 were kept.

Variants were scrutinized for potential pathogenic clinical significance based on the Association for Clinical Genomic Science (ACGS) Best Practice Guidelines for Variant Classification in Rare Disease 2024 (v1.2) (Durkie et al., 2024) and the ClinGen Sequence Variant Interpretation (SVI) Group’s recommendations (Abou Tayoun et al., 2018; Ghosh et al., 2018). This analysis was based on a comprehensive review of previous literature reports, along with computational, functional, and population data. Confirmed variants underwent annotation using ANNOVAR (http://wannovar.wglab.org/), and respective minor allele frequencies were assessed in dbSNP (http://www.ncbi.nlm.nih.gov/projects/SNP), 1000 Genomes (http://www.1000genomes.org/), Exome Aggregation Consortium (ExAC) databases (http://exac.broadinstitute.org/), gnomAD (gnomAD v4.1.0, http://gnomad.broadinstitut.e.org/), and the PSI Gene Chinese-specific database. Additionally, prediction algorithms such as PolyPhen-2 (Adzhubei et al., 2010) (version 2.2.2, 2012, http://genetics.bwh.harvard.edu/pph2/), MutationTaster (Schwarz et al., 2010) (version 2, 2012, http://www.mutationtaster.org/), MutationAssessor (Reva et al., 2011) (http://mutationassessor.org), SpliceAI (de Sainte Agathe et al., 2023), REVEL (Ioannidis et al., 2016), and CADD (Rentzsch et al., 2019) (http://cadd.gs.washington.edu) and disease and phenotype databases including Online Mendelian Inheritance in Man (OMIM; http://www.omim.org), ClinVar (http://www.ncbi.nlm.nih.gov/clinvar), the Human Gene Mutation Database (HGMD; http://www.hgmd.org), and Human Phenotype Ontology (HPO; https://hpo.jax.org/app/) were used for variant annotation and interpretation. The above database was accessed by February 2024.

Multiple protein sequence alignments were conducted using T-COFFEE and Jalview (Waterhouse et al., 2009) to assess cross-species conservation. Online resources such as UniProt (https://www.uniprot.org) and SMART (smart.embl-heidelberg.de) were utilized to analyze alterations in protein properties for assessing secondary structure. Protein and nucleotide sequences were visualized using IBS (Illustrator for Biological Sequences). Three-dimensional (3D) models of both wild-type and mutant proteins were generated using the SWISS-MODEL server program. PyMOL was used to prepare illustrations (Supplementary Figure S1).

Validation of candidate variants by Sanger sequencing was performed in all probands and available family members. Primers were designed to amplify the specific DNA fragments of interest, and polymerase chain reaction (PCR) was carried out under standard conditions. The PCR products were then sequenced on an ABI 3730xl DNA Analyzer (Applied Biosystems, United States). The resulting sequences were compared with the corresponding reference sequences using MutationMapper software.

WES was performed on genomic DNA from 40 probands with congenital cataracts to detect disease-associated variants. In total, 12.36 billion bases were generated, with an average of 73.25 million reads per chip. This provided approximately 98.2% coverage of the targeted regions and an average sequencing depth of 93.26× for each sample (Supplementary Table S3).

Overall, 19 of the 40 cases (47.5%) were familial. Pedigree analysis of these 19 families indicated autosomal dominant inheritance in 18 families, while the remaining family exhibited an X-linked dominant pattern. Pathogenic or likely pathogenic variants were identified in 15 of the 40 probands (Table 2). The variant detection yields were 47.4% (9/19) in familial cases and 28.6% (6/21) in sporadic cases (Figure 1). These pathogenic/likely pathogenic variants were distributed across 12 genes previously implicated in congenital cataract. Variants in crystallin genes (CRYAA, CRYBB2, CRYBB3, and CRYGS) were observed in 10.0% (4/40) of the cohort. Notably, variants were detected in PAX6 in three families and GJA8 in two families. In addition, single families harbored variants in BCOR, FTL, FYCO1, MIP, NHS, and COL11A1 (Figure 2). Overall, 15 pathogenic/likely pathogenic variants were identified, including 7 novel variants and 8 previously reported variants. Variant classification was performed in accordance with the ACGS guidelines for sequence variant interpretation; all variants were classified as pathogenic or likely pathogenic. Most familial cases were associated with autosomal dominant mutations in crystallin genes, except for one family carrying an X-linked NHS mutation. Sporadic cases were mainly explained by autosomal dominant mutations in a broader set of genes, including crystallin, gap junction, and transcription factor genes. In addition, de novo mutations in the X-linked gene BCOR were detected in unrelated families, and one autosomal recessive case was associated with a FYCO1 mutation. Seven variants were classified as “variants of uncertain significance” and were identified in four familial and three sporadic cases (Supplementary Table S3; Supplementary Figure S2). The remaining six familial cases and fourteen sporadic patients with congenital cataracts had no variants of interest detected in this analysis (Supplementary Figure S3).

Variants in the crystallin genes were the most frequent mutations identified in this study (Zhuang et al., 2019). Pathogenic/likely pathogenic variants were detected in four probands: three familial and one sporadic case (Table 2). All familial cases were consistent with autosomal dominant inheritance, whereas the sporadic case was most consistent with a de novo autosomal dominant variant. All variants identified in these cases were missense mutations (Figure 3). The majority of variants localized to the Greek key motifs of crystallin proteins, which are essential for correct protein folding and the maintenance of lens transparency (Vendra et al., 2013).

Two probands in our cohort carried pathogenic variants in the gap junction gene GJA8 (Table 2). One of these cases was sporadic and harbored a de novo heterozygous missense variant in exon 2 of GJA8 (OMIM 600897), NM_005267.5:c.133T>A, p. (Trp45Arg). The proband presented with esotropia, nystagmus, and posterior staphyloma (Supplementary Table S1). Although a different nucleotide change at the same position in GJA8 (OMIM 600897), NM_005267.5:c.133T>C, p. (Trp45Arg) has been reported previously, the affected individuals in that family showed a distinct phenotype. In addition to congenital cataracts, they presented with microcornea, microphthalmia, and a posterior capsule defect, indicating marked clinical heterogeneity compared with our proband (Zhang H. et al., 2019).

Another familial case carried a missense variant in GJA8 (OMIM 600897), NM_005267.5:c.131T>C, p. (Val44Ala). In addition to congenital cataracts, the proband also presented with intermittent exotropia (Supplementary Table S1). This variant has been reported previously and functionally validated in in vitro cell-based assays (Zhu et al., 2014). The missense variants in the gap junction gene GJA8 (OMIM 600897; NM_005267.5:c.131T>C, p. (Val44Ala); NM_005267.5:c.133T>A, p. (Trp45Arg)) are located in extracellular loop 1, close to the TM1/EC1 boundary (Figure 4a). These substitutions are predicted to selectively disrupt hemichannel gating while having less effect on fully formed gap junction channels. Dysfunctional hemichannels have been shown to contribute to the development of human congenital cataracts (Zhu et al., 2014; Beyer et al., 2013).

A novel nonsense variant in MIP (OMIM 154050), NM_012064.4:c.657C>A, p. (Tyr219Ter), was identified in familial case #5. This pedigree spanned four generations and included nine individuals, comprising six affected and three unaffected members; one affected individual was deceased. Clinical examination of all available family members revealed isolated lamellar cataracts in all affected individuals (Supplementary Table S1). Aquaporin 0 (AQP0), also known as the major intrinsic protein of the lens, is encoded by the MIP gene. The NM_012064.4:c.657C>A variant changes a highly conserved tyrosine codon (TAC) to a stop codon (TAA) at amino acid position 219 of AQP0, p. (Tyr219Ter) (Figure 4b). This nonsense change removes the entire intracellular C-terminal domain and produces a prematurely truncated protein. As a consequence, AQP0 is predicted to lose its function as a water channel in the cell membrane, which may lead to a congenital cataract phenotype (Song et al., 2015).

X-linked syndromic cataracts were identified in two of the fifteen families. A hemizygous frameshift variant in the NHS (OMIM 300457), NM_001291867.2:c.766dup, p. (Leu256Profs*21), was identified in an individual from another family case #6 (Figure 5a). This variant has been reported previously (Chen et al., 2021). This frameshift mutation is predicted to result in a truncated protein. NHS is associated with X-linked Nance–Horan (Burdon et al., 2003) syndrome (Kammoun et al., 2018). Nevertheless, the proband presented only with bilateral cataracts and nystagmus, without clinical evidence of microphthalmia, dental anomalies, or other characteristic craniofacial features. This apparently incomplete phenotypic spectrum may reflect age-dependent penetrance, delayed manifestation of associated features, or both, given the proband’s young age (Supplementary Table S1).

In sporadic case #3, a de novo frameshift mutation, BCOR (OMIM 300485), NM_001123385:c.4862del, p. (Pro1621Argfs*53), was identified (Table 2). This mutation is responsible for X-linked oculo-facio-cardio-dental (OFCD) syndrome (Ng et al., 2004; Fan et al., 2009). The female proband (III:3) presented with bilateral posterior polar cataract, lower eyelid inversion, a broad nasal tip, and a patent foramen ovale (PFO). Dental anomalies included dysmorphologies of the teeth, delayed eruption, and features consistent with OFCD syndrome. Neither her parents nor her fraternal twin sister, who was also conceived via in vitro fertilization, tested positive for this mutation. The BCOR NM_001123385:c.4862del, p. (Pro1621Argfs*53) mutation is predicted to delete the entire PCGF1-binding domain. This domain is essential for interaction with PCGF1, a component of the polycomb group (PcG) multiprotein BCOR complex. This interaction is required to maintain the transcriptionally repressive state of BCL6 and CDKN1A (Figure 5b) (Junco et al., 2013). Additionally, the proband inherited maturity-onset diabetes of the young type 2 (MODY2) from her mother, associated with the GCK (OMIM 138079), NM_000162.5:c.766G>A, p. (Glu256Lys) mutation (Emelyanov et al., 2017).

In PAX6, two heterozygous variants were detected. A familial splice mutation (OMIM 607108), NM_001368894.2:c.400-1G>A was found in Family #7, and a novel frameshift mutation (OMIM 607108), NM_001368894.2:c.542dup, p. (Val182Glyfs*32) was identified in Family #8 (Figure 5d). Furthermore, a sporadic case (#5) revealed a deletion of approximately 144.7 kb at chromosome 11p13. This region contains two RefSeq protein-coding genes, ELP4 and PAX6, and was classified as “pathogenic” by ACGS. The variant caused a partial deletion of intron 4 and the 3′ UTR of the PAX6 gene. qPCR validation confirmed that the proband was heterozygous, while his parents remained unaffected (Figure 5e). As a critical transcription factor, mutations in PAX6 have the potential to affect various structures during development. PAX6 mutations are characterized by the partial or complete absence of the iris, often accompanied by other ocular abnormalities such as cataracts and glaucoma (Zhang et al., 2011), corneal degeneration and microphthalmia (Lin et al., 2011), optic-nerve malformations (Azuma et al., 2003), and foveal hypoplasia and nystagmus (Khan and Aldahmesh, 2008). PAX6 truncations are widely believed to be associated with aniridia, primarily due to haploinsufficiency (Dubey et al., 2015).

In Family #7, the proband (III:1) was referred with an embryonic nuclear cataract located nasally, accompanied by complete aniridia, nystagmus, and foveal hypoplasia. His affected father (II:1) and brother (III:2) harbored the same PAX6 variant and exhibited a comparable clinical phenotype. In Family #8, both the proband and her mother presented with complete aniridia, congenital cataract, and nystagmus. In contrast, the proband’s daughter currently manifests complete aniridia without evidence of congenital cataract, which may reflect age-dependent expressivity given her young age (9 months) (Figure 5d). Sporadic case #5 also presented with clinical symptoms of embryonic nuclear cataract, complete absence of the iris, nystagmus, and foveal hypoplasia, which are attributed to a partial deletion of PAX6. This PAX6 mutation accounts for his complex phenotype and may explain the suboptimal outcome following his cataract surgery (Supplementary Table S1) (Ma et al., 2016).

Family #9 consisted of 16 members across four generations. The proband (IV:4) was an 11-year-old female who presented with bilateral congenital coralliform cataracts. Her mother displayed a similar phenotype at an early age. Both the proband and her mother had a serum ferritin level of 2000.0 ng/ml. A previously reported mutation in FTL (OMIM 134790), NM_000146.4:c.-159G>C, known as the “Verona mutation” (Girelli et al., 1995), was found in both the proband and her mother. The affected individuals showed a heterozygous G>C change at position 41 from the transcription start site, within the third residue of the 5-base sequence (CAGUG) that characterizes the loop structure of the IRE. This FTL variant could be responsible for hereditary hyperferritinemia cataract syndrome (HHCS) (Meneses et al., 2011; Luscieti et al., 2013) (Figure 6a).

Novel compound heterozygous variants FYCO1 (OMIM 607182), NM_024513.3:c.3588–9T>A, and FYCO1 (OMIM 607182), NM_024513.4:c.2345_2346del, p. (Gln782Argfs?)* were identified in sporadic case #6. Parental segregation was confirmed (Figure 6b). The frameshift mutation c.2345_2346del, p. (Gln782Argfs?)* is predicted to truncate most of the coiled-coil region and result in the complete loss of the FYVE zinc-finger and GOLD domain. Additionally, the T-to-A transversion located at the conserved intron 11 donor splice site (c.3587 + 1G>T) may affect splicing (Li et al., 2018). All these variants are predicted to cause nonsense-mediated decay of the FYCO1 mRNA, leading to a loss of FYCO1 function (Chen et al., 2011). This loss occurs despite the need for turnover of large amounts of proteins and organelles during fiber cell differentiation (Chen et al., 2017).

In sporadic case #7, a splice variant in the COL11A1 gene (OMIM 120280), NM_001854.4:c.3114 + 1G>A was identified. This mutation has been classified as likely pathogenic (LP) by the ACGS. The proband presented with bilateral congenital cortical cataracts. The right eye exhibited lens opacity covering approximately one-third of the pupil area, while the left eye showed opacity affecting about two-fifths of the pupil area. Both parents were unaffected and did not carry the mutation (Figure 5c). The COL11A1 gene encodes type XI collagen, which is primarily expressed in cartilage, the lens of the eye, the cochlea, and other connective tissues (Nallanthighal et al., 2021). Mutations in this gene can lead to various clinical manifestations, particularly affecting the eyes, hearing, and skeletal system (Snead and Yates, 1999). Ocular manifestations of COL11A1 mutations primarily involve lens opacity and glaucoma (Wang et al., 2023). In the ocular system, type XI collagen plays a crucial role in maintaining the structural integrity and transparency of the lens. Mutations typically disrupt collagen synthesis or assembly, resulting in abnormal optical properties of the lens (Boothe et al., 2020). Consequently, this leads to lens opacity and, in some cases, can result in vision impairment or blindness.