A Novel CRYBB2 Splicing Mutation Is Associated With Lens Extracellular Matrix Remodeling and Vascular Alterations in Congenital Cataract

Abstract

Purpose

We identify the genetic cause of autosomal dominant congenital cataract in a large family and define how a CRYBB2 splice-site variant perturbs lens development and associated ocular phenotypes.

Methods

Whole-exome sequencing and Sanger sequencing were used for variant identification and segregation. Wild-type (WT) or mutant CRYBB2 was expressed in lens epithelial cells to assess splicing and downstream transcriptional changes by RNA sequencing and quantitative reverse transcription PCR (qRT-PCR). Extracellular matrix (ECM) organization and cell–matrix interactions were examined by immunostaining and adhesion assays. In zebrafish, crybb2 disruption was evaluated using lens structural assays, hyaloid vasculature imaging in Tg(flk1:EGFP) larvae, tracer distribution assays, and mRNA rescue. Single-cell RNA sequencing at 48 hours post-fertilization (hpf) was performed to profile lens and vascular populations and infer pathway-level changes.

Results

We identified a novel CRYBB2 splice-site variant in affected members of a family with congenital cataract, consistent with an association with the disease in this pedigree. The variant disrupted normal splicing and was associated with altered βB2-crystallin–linked transcriptional programs and ECM homeostasis, including changes in collagen IV/laminin deposition and impaired cell–matrix adhesion. In zebrafish, crybb2 disruption caused lens fiber cell differentiation defects and was accompanied by abnormal hyaloid vascular patterning and altered tracer distribution around the lens; these phenotypes were partially improved by crybb2 mRNA rescue. Single-cell transcriptomic analyses suggested coordinated changes in ECM and junction/adhesion-related pathways across lens and vascular compartments, consistent with a lens-associated, non-cell-autonomous influence on adjacent vasculature.

Conclusions

A novel CRYBB2 splice-site mutation is associated with autosomal dominant congenital cataract and disrupts lens homeostasis with accompanying changes in hyaloid vascular remodeling. These findings expand the pathogenic spectrum of CRYBB2 and support a model in which lens abnormalities are associated with altered lens–vascular interactions during development.

Congenital cataracts (CC) represent a significant cause of visual impairment and blindness in children, with a global prevalence of approximately 4.24 per 10,000 live births.¹ In specific populations in Australia, the incidence rate has been reported to be 2.2 per 10,000.² Beyond affecting visual development in children, CC can also lead to other ocular conditions such as amblyopia and strabismus, which in turn impact children's daily activities and academic performance. Moreover, the lengthy and intricate treatment and rehabilitation process for CC imposes substantial economic and psychological burdens on families.³ From a societal perspective, the high incidence and significant disability associated with CC increase the consumption of public health resources and have adverse effects on socioeconomic development.²

In recent years, the genetic research on CC has made significant progress with the development of genetic testing technologies. Studies have shown that CC exhibit high genetic heterogeneity, involving mutations in multiple genes. To date, at least 52 genes and loci have been identified that can cause isolated CC.⁴ The proteins encoded by these genes include crystallins (such as crybb2 and CRYGC), membrane transport proteins, cytoskeletal proteins, and transcription factors.⁵ Among them, mutations in the crybb2 gene are one of the important pathogenic factors for CC. The βB2-crystallin protein encoded by the crybb2 gene is a key protein for maintaining the transparency and normal function of the lens, and its mutations can lead to abnormal protein folding, aggregation, or degradation, thereby affecting the normal development of the lens.³ In addition, mutations in the CRYGC gene have also been confirmed to be associated with CC. For example, a study found that the p.Gly129Cys mutation in the CRYGC gene leads to cataract formation by disrupting the tertiary structure and stability of the protein.⁶ Moreover, some studies have found in a zebrafish model that knockout of the foxe3 gene can cause abnormal development of the lens and anterior segment of the eye, resulting in increased differentiation of lens fiber cells and leading to cataracts, microphthalmia, and significant changes in the expression of genes associated with human cataracts.⁷

Although crystallins are traditionally considered lens structural proteins, emerging evidence indicates that certain β-crystallins perform important functions in non-lens ocular tissues. Notably, studies of the Nuc1 rat, which carries a mutation in the βA3/A1-crystallin gene, demonstrated that βA3/A1-crystallin is expressed in retinal astrocytes and is required for proper retinal vascular remodeling during development.⁸ In this model, mutation of βA3/A1-crystallin disrupted astrocyte morphology and was associated with defective vascular patterning and increased vascular leakage. These findings highlight that closely related crystallins can exert extra-lenticular roles in ocular vascular development. Whether CRYBB2 shares similar non-lens functions or instead primarily acts within the lens with secondary consequences for adjacent tissues remains unclear.

Zebrafish (Danio rerio) has become an indispensable vertebrate model in ophthalmology because of its transparency during embryogenesis, rapid eye development, and genetic tractability. Notably, the zebrafish eye shares highly conserved structural and developmental pathways with humans, including lens morphogenesis, retinal layering, and hyaloid vasculature formation, making it ideally suited for modeling congenital lens and vascular disorders.⁹ Recent studies have leveraged zebrafish to unravel the molecular underpinnings of crystallin-related cataracts. For example, transgenic or mutant zebrafish expressing pathogenic human crystallin variants have recapitulated lens opacity and fiber-cell disorganization, providing insight into protein-folding failure and aggregation in cataractogenesis.¹⁰

Single-cell RNA sequencing (scRNA-seq) has profoundly transformed genetic disease research by enabling unprecedented resolution of transcriptional dynamics across individual cell types. Unlike bulk RNA-seq, which aggregates gene expression signals, scRNA-seq uncovers heterogeneous cellular responses to a mutation, pinpointing both primary disease-driving cell populations and secondary compensatory or stress-responsive cells. This is especially critical in developmental disorders, where pathology often emerges from disrupted differentiation or fate decisions—insights that are accessible via trajectory and pseudotime analyses inherent to single-cell approaches. Zebrafish models have greatly benefited from scRNA-seq because of several key advantages: optical transparency, rapid development, ease of genetic manipulation, and conservation of over 70% of human protein-coding genes, providing a unique window into developmental and disease mechanisms across tissues.¹¹ Recent zebrafish atlas studies spanning dozens of developmental stages have produced comprehensive cellular maps—tools that are essential for interpreting gene mutation effects at the single-cell level.¹² Moreover, scRNA-seq has been used effectively in zebrafish to resolve cellular heterogeneity in tissues like kidney hematopoietic compartments and brain, revealing rare cell types and dynamic immune responses in genetic or environmental perturbations.¹³ In the context of monogenic diseases, scRNA-seq has been increasingly integrated with CRISPR-based perturbations (e.g., Perturb-seq) to map mutation-specific gene expression changes at scale, offering a powerful framework for genotype-phenotype mapping.¹⁴

In this study, we took advantage of the zebrafish model to explore how mutations in the crybb2 gene affect lens development. βB2-crystallin, encoded by crybb2, is essential for lens transparency and normal function. We found that the c.450-2A>G mutation disrupted protein folding and interfered with the proper differentiation of lens fiber cells. To further probe the molecular basis, we combined molecular dynamics simulations with single-cell RNA sequencing, which allowed us to trace the effects of the mutation at both the structural and transcriptional levels. By capturing cell-specific changes across developmental stages, single-cell analysis helped pinpoint how crybb2 mutations reshape lens biology. Together, these approaches not only deepen our understanding of the pathogenic mechanisms of crybb2 mutations but also provide a foundation for developing gene-based therapeutic strategies in congenital cataract research.

Materials and Methods

Patients and Clinical Data

This study investigated a five-generation Chinese family with 11 affected individuals, including one deceased member. The proband was a 31-year-old woman born to consanguineous parents who, along with her husband, sought genetic counseling before pregnancy. Her primary complaint was blurred vision since childhood, which was later clinically diagnosed as CC. Pedigree analysis indicated that the disease followed a pattern consistent with autosomal dominant inheritance. The severity and onset of visual impairment varied among affected family members, ranging from early childhood to late adulthood. Detailed ophthalmological examination data were available for the proband and her affected parents. For other family members, only limited clinical history was available through family records and interviews. Comprehensive phenotypic evaluation was not available across the entire pedigree.

Family members were included in the genetic analysis if DNA samples were available, and informed consent was obtained. Individuals without available DNA samples were not included in the sequencing analysis. This study focused on cross-sectional clinical observations within the pedigree, and no formal longitudinal follow-up dataset was available for quantitative analysis. Because clinical information was not uniformly available across all family members, analyses involving pedigree phenotypes were descriptive, and no statistical imputation for missing clinical data was performed.

Genomic DNA Extraction and Variant Analysis

Genomic DNA was extracted from peripheral blood of family members using the QIAamp DNA Blood Mini Kit (QIAGEN) according to the manufacturer's protocol. Whole-exome sequencing (WES) was performed on the proband and her parents (IV-6, III-6, and III-7) using the Nano WES Human Exome V1 kit (Berry Genomics) on an Illumina NovaSeq 6000 platform, generating 150 bp paired-end reads. Sequencing reads were aligned to the human reference genome (GRCh38) using BWA, with indel realignment, base quality recalibration, and variant calling performed using GATK. Variants were annotated with ANNOVAR and the Enliven Variants Annotation Interpretation System (Berry Genomics).¹⁵ Candidate variants were filtered based on allele frequency (<0.05) in public databases including 1000 Genomes, ExAC, gnomAD, ESP6500, and HGMD. Functional predictions were performed using SIFT, PolyPhen-2, MutationTaster, GERP++, and SPIDEX. Pathogenicity was further assessed with ClinVar, OMIM, HGMD, HGVS, and PubMed. Variant classification followed the ACMG guidelines, categorizing variants into five classes: pathogenic, likely pathogenic, variant of uncertain significance (VUS), likely benign, and benign.^15,16

Sanger Sequencing

The variants identified by WES, Sanger sequencing was performed to confirm the candidate variants in each independent genomic DNA sample (IV-6, III-6, and III-7). The segregation analysis of the candidate variants was traced using other affected individuals (III-3, III-6, III-7, III-8, III-9, IV-4, IV-6, IV-8, IV-10, IV-11, V-4). Primers were designed using Primer 6 software (version 6.25) for standard PCR assays. For the crybb2 mutation (GenBank NM_000496.3, c.450-2A>G, splicing), the primers were designed as follows: forward 5′-AGGAAGAAAGCAGAGGCTCAG-3′ and reverse 5′-TTGGTGCCACTGCATGTCGC-3′. PCR amplification and sequencing were carried out using standard protocols on an ABI 3500 Genetic Analyzer (Thermo Fisher Scientific, Waltham, MA, USA).¹⁷

Mutation Site Validation and Protein Analysis

In this study, we used multiple approaches to validate and analyze the mutation sites. Initially, Sanger sequencing was conducted to verify the results obtained from WES, thereby ensuring the accuracy of the sequencing outcomes. Subsequently, various prediction tools, including CADD, DANN, and Polyphen2, were used to evaluate the pathogenic potential of the identified mutations. Furthermore, the BLAST online analysis tool was used to compare the homology of amino acids among different species, which helped to elucidate the evolutionary conservation of the mutation sites. The AlphaFold2 online prediction platform and PyMol software were used to analyze and compare protein conformations. Specifically, in silico predictions indicated that the c.450-2A>G mutation causes intron 5 retention, leading to a frameshift and a premature stop codon. The resulting predicted truncated amino acid sequence, which lacks the C-terminal Greek key motifs, was used to construct the mutant protein model. Finally, molecular dynamics (MD) simulations were performed using these WT and truncated mutant models to investigate the structural alterations in the protein.¹⁸ The stability of the protein structure was assessed using root mean square deviation (RMSD), and binding mode analysis was conducted to further explore the potential effects of the mutation on protein structure and function.

Zebrafish Feeding

Zebrafish (AB strain) were maintained and bred under standard laboratory conditions according to the guidelines of the China Zebrafish Resource Center. Detailed husbandry parameters are provided in the Supplementary Methods.

Microscopic Manipulation of Zebrafish

Microinjection and microscopic imaging procedures for zebrafish embryos were performed using standard micromanipulation systems. Detailed protocols are described in the Supplementary Methods.

Generation of Crybb2 Knockout (Crybb2^+/−) Zebrafish Model

The crybb2 knockout zebrafish model was generated using the CRISPR/Cas9 system. Briefly, specific single-guide RNAs targeting the crybb2 gene were designed and co-injected with Cas9 protein into wild-type (AB strain) zebrafish embryos at the one-cell stage. Stable mutant lines were subsequently established and verified through standard outcrossing and genotypic screening. Detailed procedures regarding single-guide RNA design, microinjection parameters, and mutant line establishment are provided in the Supplementary Methods.

Genotyping of Crybb2 Knockout Zebrafish

Routine genotyping of the crybb2 mutant zebrafish was performed via PCR amplification of genomic DNA extracted from caudal fin clips. Specific primer sequences and detailed procedures are available in the Supplementary Methods.

Quantitative Real-Time PCR Analysis

Total RNA extraction, reverse transcription, and quantitative real-time PCR were performed to quantify gene expression across different developmental stages, with β-actin serving as the internal reference. Primer sequences and specific thermal cycling conditions are detailed in the Supplementary Methods.

Lentiviral Construction, Transduction, and Crybb2 Minigene Splicing Analysis

A short hairpin RNA targeting the crybb2 gene was cloned into a lentiviral interference vector and validated by sequencing. Lentiviral particles were packaged in HEK293T cells using standard helper plasmids and collected 48 hours after transfection. Viral titers were quantified by qPCR, and aliquots were stored at –80°C. After transduction, total RNA was extracted with TRIzol reagent (Invitrogen, Carlsbad, CA, USA), and cDNA was synthesized using the PrimeScript RT kit (Takara Biotechnology Co., Kyoto, Japan). To assess alternative splicing, PCR was performed with primers flanking the targeted minigene region, and products were analyzed on 2% agarose gels to identify abnormal splice variants.

Western Blot Analysis

The translational consequences of the CRYBB2 c.450-2A>G mutation, an in vitro expression system was used. Briefly, FLAG-tagged wild-type and mutant CRYBB2 expression plasmids were transiently transfected into HEK293T cells, followed by protein extraction and Western blot analysis using specific antibodies against FLAG and the internal control, β-actin. Detailed experimental procedures are provided in the Supplementary Methods.

Ultrastructural Observation of Zebrafish Lens Tissue by TEM

Zebrafish lens tissues were processed for transmission electron microscopy (TEM) to evaluate the ultrastructure of lens fiber cells. Detailed fixation, embedding, and ultrathin sectioning protocols are provided in the Supplementary Methods.

Histological Analysis of Zebrafish Lens Tissue by H&E Staining

Histological evaluation of lens architecture was conducted using standard hematoxylin and eosin (H&E) staining on paraffin-embedded sections, as detailed in the Supplementary Methods.

Immunofluorescence Staining

Immunofluorescence staining on paraffin-embedded sections was performed to analyze specific protein expression. The primary antibodies used included Anti-Lens Fiber Cell Marker (Zl-1), MIP, VEGFR2/KDR (Flk-1), ELAVL3 (HuC/D), Rhodopsin, and ZNF259. Detailed staining procedures and antibody concentrations are listed in the Supplementary Methods.

In Situ Hybridization of Zebrafish Embryos

Whole-mount in situ hybridization using a DIG-labeled ath5 probe was performed to assess early retinal development. Detailed procedures are available in the Supplementary Methods.

M. mRNA Rescue in Crybb2^+/− Zebrafish

The full-length crybb2 coding sequence was cloned into the pCS2(+) vector (Beijing Tsingke Biotech Co., Ltd., Beijing, China), and capped mRNA was synthesized from the linearized plasmid using the mMESSAGE mMACHINE SP6 Transcription Kit (Thermo Fisher Scientific) according to the manufacturer's instructions. Approximately 150 pg of synthesized crybb2 mRNA was microinjected into the cytoplasm of one-cell stage zebrafish embryos derived from crybb2 heterozygous crosses. Injected embryos were raised under standard conditions, and phenotypic rescue efficiency was subsequently evaluated by lens morphology and vascular assays.

Fluorescent Dye Injection

At five days post-fertilization (dpf), fluorescent dye injection was performed on zebrafish larvae.¹⁹ A volume of 10 nL of tetramethylrhodamine, methyl ester, perchlorate (TMR-MEP) (10 mg/mL) was injected into the atrium of zebrafish. This small-molecule dye readily penetrates vascular walls and enters adjacent tissues, including the lens-associated hyaloid vasculature.²⁰ One hour after injection, the distribution and accumulation of the dye within the lens-associated hyaloid vasculature of zebrafish larvae were examined using a laser scanning confocal microscope (Olympus FV3000; Olympus, Tokyo, Japan).

In Vivo Imaging of the Hyaloid Vessels in Zebrafish

Before imaging, with Tg (flk1: EGFP) zebrafish larvae at 5 dpf were embedded in confocal dishes containing 1% low-melting-point agarose (Sigma-Aldrich, St. Louis, MO, USA). Approximately 20–30 optical sections were captured per eye using a laser scanning confocal microscope (Olympus FV3000), and the images were processed using Olympus FV3000 software.

Imaging Analysis of the Hyaloidvascular Density and Perfusion Area

The density of hyaloid vessels in zebrafish at 5 dpf was measured using Adobe Photoshop 2020 software (Adobe Systems Inc., San Jose, CA, USA). The hyaloid vessels were outlined by two independent observers who were masked to the experimental groups.²¹ The ratio of the pixel area of the lassoed vessels to the total pixel area within the selected region (indicated by a white dashed circle in Supplementary Fig. S22A) was calculated as the vitreous vascular density. The perfused and total areas of the vitreous vasculature were measured by analyzing binarized particles using ImageJ software (version 1.50i; National Institutes of Health, Bethesda, MD, USA).²² The investigator performing image quantification was masked to genotype and experimental groups.

Single-Cell RNA Sequencing and Transcriptomic Analysis of Crybb2-Related Zebrafish Embryos

In this study, scRNA-seq was applied to analyze zebrafish embryonic cells, with the aim of exploring the cellular heterogeneity and transcriptomic alterations associated with crybb2 gene disruption during early development, particularly in ocular and neural tissues. For each experimental group, approximately 200 zebrafish embryos were collected and pooled into a single biological replicate. The pooled samples for both groups were processed concurrently in the same batch through a standardized protocol involving washing, mechanical dissociation, and enzymatic digestion to obtain single-cell suspensions. Single-cell capture and cDNA synthesis were performed using the 10 × Genomics Chromium platform, targeting an estimated 15,000–20,000 cells per sample. Library preparation was conducted with the Chromium Single Cell 3′ Reagent Kit (Library and Gel Bead Kit v3.1; 10x Genomics, Pleasanton, CA, USA), and high-throughput sequencing was carried out on the Illumina NovaSeq 6000 platform. Raw sequencing data were processed using the Cell Ranger software (v6.0.1), with reads mapped to the zebrafish reference transcriptome (GRCz11; Ensembl, Cambridge, UK). Subsequent dimensionality reduction and clustering were performed using the Seurat R package (v4.1.0), including principal component analysis and t-distributed stochastic neighbor embedding. Cell populations were annotated based on the expression of known marker genes. To assess the effects of crybb2 mutation, differentially expressed genes were identified using the Wilcoxon rank-sum test and further analyzed with the MetaDE R package. Functional enrichment, biological process categorization, and upstream regulatory network analysis were performed using Ingenuity Pathway Analysis software to elucidate crybb2-related signaling alterations and developmental consequences.

Statistical Analysis

Statistical analyses and graphing were performed using GraphPad Prism 8 software. All statistical data are presented as mean ± SEM. Before parametric statistical testing, data distribution was assessed using the Shapiro–Wilk test. Comparisons between two groups were conducted using the Student's t-test. For single-factor comparisons involving three or more groups, a one-way ANOVA was used, followed by Tukey's post-hoc test for multiple comparisons. For experiments involving two independent variables or repeated measures, a two-way ANOVA or two-way repeated-measures (RM) ANOVA was used, followed by Sidak's post-hoc test. All reported P-values from ANOVA are adjusted P-values to control the family-wise error rate. Where appropriate, effect sizes and 95% confidence intervals were calculated to support statistical interpretation. The levels of statistical significance are as follows: *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001; NS indicates no significant difference.

Results

Identification of a Novel Splice-Site Mutation in Crybb2 in a CC-Affected Family

A pedigree of the family was constructed, revealing a typical autosomal dominant inheritance pattern of CC (Fig. 1A). Notably, multiple instances of consanguineous marriage were documented within the pedigree: the proband's parents (III-6 and III-7) were first cousins, her maternal grandparents (II-3 and II-4) were also consanguineous, and her maternal uncle (III-8) and aunt (III-9) were likewise first cousins. Clinical histories and genotyping results of the available family members are summarized in Supplementary Table S1. Within the available family members, the CRYBB2 c.450-2A>G variant was identified in affected individuals and was not detected in tested unaffected relatives, consistent with an association between this variant and congenital cataract in this family. Among the family members for whom clinical information was available, carriers generally presented with congenital or early-onset bilateral lens opacities, whereas unaffected tested relatives did not have documented congenital cataract. Detailed ophthalmic examinations were available for the proband (IV-6) and her affected parents (III-6 and III-7). Representative clinical images of these genotyped family members are shown in Figure 1B.

Figure 1.

Clinical, genetic, and structural characterization of a five-generation family with CC. (A) Pedigree of a five-generation family with CC. Generations are labeled I–V. Symbols indicate sex and phenotype; the proband is marked with a black arrow. (B) Representative ophthalmic examinations of the proband (IV-6) and her affected parents (III-6 and III-7). The clinical imaging includes ophthalmic ultrasound examinations (for IV-6, III-6, and III-7), scanning laser ophthalmoscopy (for IV-6), and slit-lamp anterior segment photography (for III-6 and III-7). (C) Schematic diagram of CRYBB2 protein domains, with the splice-site mutation c.450-2A>G mapped to the intron-exon boundary. (D) Conservation analysis of the CRYBB2 locus across multiple species, illustrating cross-species sequence alignment. (E) AlphaFold-predicted three-dimensional structural models of WT and mutant CRYBB2 proteins, with Greek key motifs and terminal regions color-coded. Merged structural overlay is also shown. (F) Molecular dynamics simulations comparing WT and mutant CRYBB2 proteins, including RMSD trajectories.

WES of the proband (IV-6) and her parents (III-6, III-7) identified a heterozygous candidate variant in CRYBB2 (NM_000496.3): c.450-2A>G in intron 5. This splice-site mutation was detected in both the mother and father, who were heterozygous carriers. The c.450-2A>G alteration is predicted to disrupt the canonical splice acceptor site, potentially leading to abnormal splicing of exon 6 and consequent alterations in protein structure. The CRYBB2 protein (NP_000487.1), normally composed of 205 amino acids, belongs to the β/γ-crystallin “Greek key” family, which is highly conserved. Figure 1C shows the position of the c.450-2A>G variant within CRYBB2, lying upstream of the fourth “Greek key” motif. Conservation analysis confirmed that this nucleotide is highly conserved across species (Fig. 1D). The variant was absent from public databases, including 1000Genomes, gnomAD, and ExAC. Sanger sequencing confirmed the presence of the c.450-2A>G variant in affected family members who were tested and its absence in tested unaffected relatives (Supplementary Fig. S1). No other pathogenic variants in CC-related genes were identified in the proband. Based on the ACMG/AMP standards and guidelines, this novel CRYBB2 variant was classified as pathogenic.²³

MD Simulation of the CRYBB2 Variant

MD simulations revealed marked structural alterations in the N-terminal arm of the CRYBB2 mutant protein compared with the wild-type (WT) structure, changes that are predicted to compromise overall protein stability. The three-dimensional structures of WT and mutant proteins are shown in Figure 1E, with the superimposed models highlighting conformational differences between the two proteins. Notably, the mutation appears to disturb the arrangement of the β/γ-crystallin “Greek key” domains 1–4, which are critical for the structural and functional integrity of the protein. The RMSD analysis further supports these observations (Fig. 1F). For the WT protein, structural stabilization occurred around 200 ns, with RMSD values fluctuating near 4 Å, indicating a relatively stable conformation. In contrast, the mutant protein consistently displayed higher RMSD values, particularly during the mid and late phases of the 500-ns simulation, suggesting greater conformational flexibility and reduced stability. Moreover, the mutant trajectory exhibited more pronounced fluctuations throughout the simulation.

Generation of Crybb2 Mutant Zebrafish via CRISPR/Cas9 and Genotypic Characterization

Using CRISPR/Cas9-mediated gene editing, we established crybb2 mutant zebrafish carrying targeted mutations at the crybb2 locus (Supplementary Fig. S2). The upper panel displays the WT genomic sequence, whereas the lower panel shows the mutant sequence with the specific base alteration indicated on the right. The verified mutant genotype provided the basis for subsequent phenotypic characterization of crybb2 deficiency in zebrafish. At 6 dpf, crybb2^+/− mutant larvae exhibited abnormal lens morphologies compared with WT (Supplementary Fig. S2A). These morphological differences, including irregular lens shape and visible structural alterations, were documented under a stereomicroscope.

Expression Analysis of Crybb2 MRNA in WT and Mutant Zebrafish

We analyzed crybb2 mRNA expression in wild-typeand crybb2^+/− mutant zebrafish at different developmental stages (24, 48, 72, and 120 hours post-fertilization [hpf]) (Supplementary Fig. S3A). Total RNA was extracted from pooled embryos, and three primer sets targeting the intron 5 region of crybb2 were designed for RT-PCR analysis. Lane 1 and Lane 2 used primer pairs spanning the intronic sequence and adjacent exons. As expected, no amplification products were detected in either WT or crybb2^+/− samples, because these primers do not cover a complete exon-exon junction. In Lane 3, primers specific to exons 5 and 6 amplified a 268-bp band (126 bp from exon 5 and 142 bp from exon 6) in all WT samples, indicating normal splicing. In contrast, Lane 4 from crybb2^+/− mutants showed no detectable band. The theoretical product size in the presence of intron retention was 2462 bp (126 bp exon 5 + 2194 bp intron 5 + 142 bp exon 6), which exceeds the detectable range of conventional agarose gel electrophoresis. This absence of visible amplification therefore reflects splicing disruption in the mutant group. Lanes 5 and 6 amplified β-actin (internal control), producing consistent bands across all samples and time points, confirming RNA integrity and proper RT-PCR performance.

Quantitative real-time PCR further confirmed these findings (Supplementary Fig. S3B). At 24 hpf, crybb2 expression in mutant larvae was significantly upregulated compared with WT controls (P = 0.0456). At 48 hpf, expression became low in mutants, with a significant difference from controls (P = 0.0122). At 72 hpf, mutant expression keptthis level. By 120 hpf, expression was again markedly lower in mutants, with an extremely significant reduction (P = 0.0052).

Validation of the crybb2 Splicing Mutation in a Cell-Based Model

The effect of the crybb2 c.450-2A>G mutation on RNA processing, expression plasmids carrying either WT or mutant sequences were constructed and transfected into HEK293T cells. Total RNA was extracted, reverse-transcribed, and amplified by PCR to evaluate splicing efficiency (Supplementary Fig. S4). Lane 1 (positive control, plasmid DNA) displayed a band of ∼169 bp, consistent with the expected size of the crybb2 cDNA fragment used in plasmid construction. Lanes 2 and 4, corresponding to cells transfected with WT crybb2 plasmids, showed clear 169 bp bands, indicating successful splicing and generation of the mature transcript. In contrast, Lanes 3 and 5, corresponding to cells transfected with the mutant plasmid (c.450-2A>G), produced a band of ∼2109 bp, representing the unspliced transcript containing exon 5, intron 5, and exon 6. No 169 bp spliced product was detected in these lanes, indicating that the mutation completely prevented intron removal and the formation of mature crybb2 mRNA.

In Vitro Expression Profiling Reveals Aberrant Mutant CRYBB2 Protein Forms

The translational consequences of the c.450-2A>G mutation and address the structural anomalies predicted by our models, we performed Western blot analysis using an in vitro HEK293T cell expression system. Cells were transfected with FLAG-tagged WT or mutant CRYBB2 constructs. As illustrated in Supplementary Figure S5, the WT construct produced a single distinct band at the expected molecular weight. In striking contrast, the mutant construct yielded an aberrant protein profile characterized by a prominent doublet. These biochemical findings support the presence of aberrant and likely truncated CRYBB2 protein variants, which may contribute to lens pathology.

Ocular and Lens Enlargement in crybb2^+/− Mutant Zebrafish

In a detailed assessment of zebrafish larvae at 6 dpf, comparing WT with heterozygous crybb2^+/− mutant types, it was observed that mutations in the crybb2 gene significantly impacted the development of the eyes and lenses. Specifically, the eyes of crybb2^± mutants were significantly larger than those of the wild type (P = 0.0002), suggesting that the crybb2 gene may play a crucial role in regulating ocular growth (Fig. 2B upper panel). The significant increase in lens diameter (p < 0.0001) further confirmed the pivotal role of the crybb2^+/− mutation in lens development, with this enlargement potentially affecting vision and the focusing ability of the eye (Fig. 2B lower panel). The significant increase in the ratios of eye diameter to body length (P = 0.0001) and lens diameter to body length (P = 0.0173) indicated that, relative to body size, the eyes and lenses of crybb2^+/− mutant zebrafish developed excessively, which may reflect the role of the crybb2 gene in coordinating the balance between ocular and overall body development.

Figure 2.

Characterization of WT and crybb2^+/− zebrafish larvae. (A) Stereomicroscopic imaging of WT and crybb2^+/− zebrafish eyes at 6 dpf. Images were captured using a Leica M205 FA stereomicroscope. (B) Measurements of eye and lens diameters and their ratios to body length in zebrafish larvae at 6 dpf. Quantification was performed on 10 larvae per group. Images were processed and analyzed using ImageJ software. (C) Survival analysis of WT and crybb2^+/− larvae from 1 to 20 dpf. Survival curves were plotted and analyzed using GraphPad Prism 8 with the Kaplan-Meier method. (D) Representative locomotor trajectories of zebrafish larvae at 7 dpf. (E) Quantitative analysis of locomotor activity at 7 dpf, including traveled distance, swimming speed, maximum acceleration, activity states, and resting time. Behavioral tracking was conducted using EthoVision XT software (Noldus Information Technology, Leesburg, VA, USA).

Survival Analysis of Crybb2^+/− Zebrafish Embryos

We observed that mutant zebrafish exhibited abnormal mortality during the early stages of embryonic development. Consequently, we recorded and analyzed the survival of embryos over a 20-day post-fertilization period (Fig. 2C). The reduced survival rate of crybb2^+/− mutant larvae highlights the detrimental effects of the crybb2 mutation on the early development and survival of zebrafish. The survival curves indicate that the crybb2 gene plays a significant role in maintaining normal developmental processes, and its dysfunction can lead to additional developmental abnormalities. This data provides statistical evidence of the adverse effects of the crybb2 mutation, emphasizing the importance of further research to elucidate the molecular mechanisms underlying these survival differences.

Locomotor Behavior of Crybb2^+/− Zebrafish

Locomotor trajectories of WT and crybb2^+/− zebrafish larvae were recorded at 7 dpf (Fig. 2D). WT larvae displayed trajectories concentrated in the central arena, with irregular but organized paths characteristic of exploratory swimming behavior. In contrast, crybb2^+/− larvae exhibited stereotyped circular swimming patterns restricted to the periphery of the arena. Quantitative analysis confirmed that the proportion of time spent near the boundary was significantly greater in mutants compared with WT (P = 0.0019). Behavioral parameters further supported these differences (Fig. 2E). The average traveled distance in WT larvae was ∼3,500 mm, whereas mutants traveled less than 2,800 mm (P = 0.0031). Average swimming speed was 0.75 mm/s in WT compared with <0.60 mm/s in mutants (P = 0.0044). Maximum acceleration reached ∼1700 mm/s² in WT but dropped below 1,500 mm/s² in mutants (P = 0.0064). The cumulative duration of highly mobile states was ∼35 s in WT, significantly longer than ∼20 s in mutants (P = 0.0044). Similarly, cumulative mobile time was reduced in mutants (<100 s) compared with ∼120 s in WT (P = 0.0070), whereas resting time was extended in mutants (∼3500 s) relative to WT (∼3400 s) (P = 0.0104). Analysis of forward swimming angle relative to the arena center showed no significant difference between groups (P = 0.6278).

At four months post-fertilization, adult zebrafish were also evaluated for locomotor activity (Supplementary Figs. S6A. S6B). WT fish exhibited broad and evenly distributed trajectories throughout the arena, whereas crybb2^+/− mutants showed restricted swimming patterns concentrated at the periphery with reduced central exploration. Quantitative analysis showed that the swimming area covered by mutants was significantly smaller than that of WT (P = 0.0034). Consistent with these trajectory differences, behavioral parameters indicated reduced activity in mutants. The average traveled distance in WT adults was ∼5000 cm, compared with <4200 cm in mutants (P = 0.0349). Mean swimming speed was ∼3.5 cm/s in WT and <2.5 cm/s in mutants (P = 0.0381). Maximum acceleration was ∼130 cm/s² in WT versus ∼105 cm/s² in mutants, a nonsignificant trend (P  = 0.0740). Highly mobile state duration averaged ∼520 s in WT compared with ∼420 s in mutants, without statistical significance (P = 0.0706). Similarly, mobile time was slightly higher in WT (46 s) than in mutants (38 s), with no significant difference (P = 0.0706). Resting time was comparable between groups, both averaging ∼1400 s (P = 0.1335). Forward swimming angle distributions were symmetrical and consistent across groups, with no significant differences (P = 0.5891).

Ultrastructural Abnormalities of Lens Fiber Cells in Crybb2^+/− Zebrafish

TEM was used to examine the ultrastructure of lens fiber cells in WT and crybb2^+/− zebrafish at 3 dpf, 5 dpf, and 4 mpf. At 3 dpf (Fig. 3A), WT zebrafish lens fibers displayed the typical transparent morphology, with highly ordered cellular structures and almost complete clearance of organelles (Fig. 3AA'). Higher magnification of the boxed region (Fig. 3AB') confirmed that the cytoplasm was homogeneous and free of dark-stained structures, indicating proper terminal differentiation. In contrast, crybb2^+/− mutants (Fig. 3AC') exhibited numerous retained organelles. Enlarged images of the boxed regions (Figs. 3AE', 3AG') revealed abundant mitochondria (M, red arrows), lysosomes (L, green arrows), and mitochondria-lysosome fusion structures (M/L), indicating persistence of organelles that were absent in WT fibers. At 5 dpf (Supplementary Fig. S7), WT lens fibers (Supplementary Figs. S7A, S7B) again appeared transparent and well aligned, with no visible organelles. In crybb2^+/− mutants (Supplementary Fig. S7D), residual organelles remained evident. Enlarged panels (Supplementary Fig. S7–E, S7-J) highlighted increased clusters of mitochondria (M, red arrows) and a greater number of autophagic lysosomes (L, green arrows). In addition, the nuclei of mutant fiber cells (N, blue arrows) showed increased electron density, further distinguishing them from the organelle-free nuclei-depleted WT fibers. By 4 mpf (Supplementary Fig. S8), the ultrastructural differences were more pronounced. WT adult zebrafish (Supplementary Figs. S8A, S8B) retained transparent, well-ordered lens fibers, with no detectable nuclei or organelles. In contrast, crybb2^+/− mutants exhibited disrupted fiber cell alignment (yellow arrows, Supplementary Fig. S8D), with frequent residual nuclei (N, blue arrows), swollen and irregular mitochondria (M, red arrows), numerous autophagic lysosomes (L, green arrows), and mitochondria-lysosome fusion complexes (M/L) (Supplementary Figs. S8E–G). These features were consistently absent in WT controls.

Figure 3.

Ultrastructural, histological, and immunofluorescence analysis of lens development in WT and crybb2-knockout zebrafish. (A) TEM images of lens fiber cells in WT and crybb2^+/− zebrafish at 3 dpf. N, nucleus (blue arrows); M, mitochondria (red arrows); L, lysosome (green arrows); M/L, mitochondria-lysosome fusion structures (yellow arrows), lens fiber cells. (B) H&E staining of lens sections from WT, crybb2 morphants (MO), and rescue (MO + mRNA) zebrafish at 48, 72, 96, and 120 hpf. Retinal layers were labeled as follows: PL, photosensitive layer; INL, inner nuclear layer; GCL, ganglion cell layer; OPL, outer plexiform layer; IPL, inner plexiform layer; LEC, lens epithelial cells; eLFC, elongated lens fiber cells. (C) Relative positive area ratio of lens opacity was measured in zebrafish from 24 to 120 hpf. The y-axis represents the relative opacity area ratio, and the x-axis indicates developmental time points. Data are presented as mean ± SEM. Statistical analysis was performed using one-way ANOVA followed by Tukey's post-hocmultiple comparisons test (*P < 0.05, **P < 0.01). (D) Immunofluorescence staining of zebrafish lenses at 72 hpf with Zl-1 (green) and Mip (red). Overlays show colocalization of Zl-1 and Mip signals (merged images). Scale bars are indicated in each panel.

Histological Analysis of Lens Transparency in WT, Crybb2^+/−, and Rescue Zebrafish

H&E staining was performed on lenses from WT, crybb2^+/−, and crybb2^+/− + crybb2 mRNA zebrafish at 48, 72, 96, and 120 hpf (Figs. 3B, 3C). At 48 hpf, crybb2^± larvae exhibited pronounced opacity in the central region of the lens, whereas WT lenses were transparent. At 72 hpf, lens opacity persisted in the knockout group, while WT lenses remained clear, and the crybb2^+/− + mRNA group showed only mild opacity. Retinal layers, including the ganglion cell layer, inner nuclear layer, inner plexiform layer, and outer plexiform layer (OPL), were well developed and distinguishable in all groups. At 96 hpf and 120 hpf, lens transparency in the knockout group gradually improved but remained inferior to that of WT. Quantitative analysis of the lens opacity area ratio from 24 to 120 hpf using ImageJ software showed significantly higher values in the crybb2^+/− group compared with WT at all time points (*P < 0.05). In contrast, the crybb2^+/− + mRNA group exhibited significantly lower opacity area ratios than the knockout group (**P < 0.01).

Immunofluorescence Analysis of Lens Fiber Cells

Immunofluorescence staining with Zl-1 (green) and Mip (red) was performed to examine lens fiber cell structure and protein distribution in WT and crybb2^± zebrafish at 72 hpf and 5 dpf (Fig. 3D; Supplementary Fig. S9). At 72 hpf, WT zebrafish lenses exhibited continuous and well-organized Zl-1 staining in lens fiber cells, outlining a regular cellular framework (Fig. 3DA). Mip labeling in WT lenses was evenly distributed across the fiber cell membranes (Fig. 3DA′). In contrast, crybb2^+/− mutants displayed disrupted Zl-1 staining with fragmented and irregular fiber cell labeling (Fig. 3DB). Correspondingly, Mip staining in mutants (Fig. 3DB′) appeared diffuse and uneven, with focal areas of increased signal intensity consistent with lens opacity. At 5 dpf, WT zebrafish maintained clear Zl-1 staining in the cortical lens region, with distinct epithelial-fiber cell boundaries (arrow, Supplementary Fig. S9). Mip labeling also remained uniform and continuous, consistent with normal fiber cell maturation. In crybb2^+/− mutants, Zl-1 staining was less defined, with scattered and irregularly arranged fiber cells visible in the central lens region (dashed arrow). Mip staining in mutants was diffuse and irregular, with areas of signal aggregation, further indicating disturbed lens structure compared with WT.

Crybb2 Deficiency Is Associated With Alterations in Hyaloid Vessel Morphology, Tracer Distribution, and Flk-1 Expression

Confocal imaging of Tg(flk1:EGFP) zebrafish larvae at 5 dpf revealed significant abnormalities in hyaloid vasculature after crybb2deficiency (Fig. 4A). WT larvae exhibited a characteristic basket-like network of vitreous vessels, with dense branching and regular distribution around the lens. Incrybb2^+/− larvae, the vasculature was markedly sparser, with fewer branches, irregular orientation, and reduced density. In the rescue group (crybb2^+/− + crybb2 mRNA), vascular morphology was partially restored toward the WT pattern. White dashed lines outline the vascular regions, and enlarged boxed areas highlight the morphological differences. Quantitative analysis of vascular density confirmed these morphological findings (Fig. 4B). Crybb2^+/− larvae showed significantly reduced vascular density compared with WT (0.951 ± 0.025 vs. 1.027 ± 0.025, P = 0.0147) and the rescue group (0.951 ± 0.025 vs. 1.009 ± 0.019, P = 0.0178). To assess vascular transport and tracer distribution, lumen formation was evaluated by TMR-MEP injection at 5 dpf (Fig. 4C). In WT larvae, TMR-MEP fluorescence was evenly distributed within the hyaloid vessels and confined to the lumen, indicating normal vascular confinement. A similar pattern was observed in the rescue group. By contrast, crybb2^+/− larvae displayed stronger TMR-MEP retention within the vessels, with slower diffusion into surrounding regions. Overlay images showed higher colocalization of Flk1-GFP (green) and TMR-MEP (red) signals, consistent with altered tracer clearance and impaired lumen diffusion. Flk-1 (VEGFR2/KDR) immunofluorescence further revealed stage-specific differences in vascular endothelial development (Fig. 4D; Supplementary Fig. S10). At 48 hpf, WT larvae (Fig. 4DA) showed strong Flk-1 expression (green) surrounding the lens, whereas crybb2^+/− mutants (Fig. 4DB) exhibited weaker and discontinuous signals with blurred vessel boundaries (dashed arrow). The DAPI panels (A′, B′) confirmed nuclear distribution, while merged images (A′′, B′′) showed a clear and organized vascular network in WT compared with the sparse, diffuse pattern observed in mutants. At 72 hpf (Supplementary Fig. S10), WT larvae displayed even stronger Flk-1 expression and more clearly defined vessel structures, whereas crybb2^+/− mutants continued to show weak, diffuse signals and smaller vascular cavities. The corresponding DAPI and merged panels confirmed robust vascular development in WT versus underdeveloped vasculature in mutants.

Figure 4.

Crybb2 deficiency impairs hyaloid vessel development and permeability in zebrafish. (A) Confocal imaging of hyaloid vasculature in Tg(flk1:EGFP) zebrafish larvae at 5 dpf. White dashed lines outline vitreous vascular structures; boxed regions indicate magnified views. (B) Quantification of vascular density across experimental groups, analyzed using ImageJ. Each group included at least eight larvae (n ≥ 8). Statistical significance was determined using one-way ANOVA followed by Tukey's post-hoc multiple comparison test. Data are presented as mean ± SEM. (C) Assessment of tracer distribution by intravenous injection of TMR-MEP (red) at 5 dpf, imaged together with flk1:EGFP-positive vasculature (green). Scale bar: 50 µm. (D) Immunofluorescence staining for Flk-1 (VEGFR2/KDR, green) at 48 hpf. Nuclei were counterstained with DAPI (blue). Z-stack confocal images acquired at magnification ×60. Scale bar: 60 µm.

Effects of Crybb2 Deficiency on Retinal Neuron Development

Whether crybb2 affects retinal neuronal development, we first examined ath5 expression at 36 hpf and HuC/D²⁴ labeling of retinal ganglion cells (RGCs) at 48 hpf and 4 dpf (Supplementary Fig. S11). In WT embryos, ath5 showed strong and localized expression in the developing retina (A, A′), whereas crybb2^+/− mutants exhibited slightly reduced signals with a similar distribution (B, B′). At 48 hpf, HuC/D immunostaining revealed abundant RGCs in WT retinas (C), whereas mutants displayed a modest reduction (D). By 4 dpf, WT larvae showed HuC/D-positive neurons extending into the optic tectum (C′, C′′), whereas mutants exhibited weaker labeling in this region (D′, D′′). The arrows in the figures indicate the stained neurons. Ventral views confirmed that neuronal development in non-retinal regions remained normal, suggesting that crybb2 disruption produced only localized and relatively mild changes in early retinal neurogenesis.

We then systematically evaluated the organization of three classes of vision-related neurons²⁵—amacrine and ganglion cells, rod photoreceptors, and double cone photoreceptors—by immunofluorescence staining of retinal cryosections at 5 dpf (Supplementary Fig. S12). HuC/D staining showed that amacrine and ganglion cells in the inner nuclear layer and ganglion cell layer remained complete and orderly in crybb2^+/− larvae (B), with no significant differences compared with WT (A) or rescue larvae (C). Rod photoreceptors, detected in the outer plexiform layer, exhibited nearly identical spatial distribution across WT, knockout, and rescue groups (D–F). Double cone photoreceptors in the knockout group (H) also displayed normal distribution patterns, consistent with WT (G) and the rescue group (I). Collectively, these results indicate that crybb2 deficiency caused only mild reductions in early HuC/D-positive neurons, whereas the overall spatial distribution and structural organization of amacrine cells, ganglion cells, rod photoreceptors, and double cone photoreceptors remained largely unaffected by 5 dpf.

Single Cell Atlas of WT and Crybb2^+/− Zebrafish Embryos and Subclustering of Lens and Vascular Populations

Transcriptional alterations associated with crybb2 mutation, we performed scRNA-seq on pooled WT and crybb2^+/− zebrafish embryos at 48 hpf. After data preprocessing, quality control metrics were examined, including detected gene numbers, transcript counts, and mitochondrial content. The distributions were consistent between WT and mutant samples, confirming comparable dataset quality (Supplementary Fig. S13A). In addition, highly variable genes were identified for both groups, and the overall profiles showed strong overlap, suggesting similar transcriptional complexity across WT and mutants (Supplementary Fig. S13B).

Unsupervised clustering of single cell transcriptomes from WT and crybb2^+/− zebrafish embryos at 48 hpf identified 25 major cell populations (Fig. 5A). These clusters covered the principal embryonic lineages, including neuronal, mesenchymal, chondrocytic, epithelial, and ocular cell types. Differential expression analysis revealed distinct marker profiles for each cluster: neuronal clusters were enriched for elavl family members and snap25a; mesenchymal and chondrocytic clusters expressed sfrp2, twist2, and col2a1a; early neuroepithelial cells showed high expression of sox3 and cldn5a; whereas epithelial groups were marked by tmsb1 and cldni (Supplementary Table S2). These expression patterns confirmed that the single-cell atlas comprehensively captured the major embryonic tissues and provided a robust basis for downstream focused analyses.

Figure 5.

Single-cell atlas and intercellular communication changes upon crybb2 knockout. (A) UMAP visualization of broad cell type annotations from single-cell transcriptomic data. (B) Cell-cell communication networks inferred by CellChat, comparing control and crybb2 knockout groups, shown by interaction number (top) and interaction strength (bottom). (C) UMAP visualization of sub-cluster annotations within selected vascular- and lens-related populations. (D) Dot plot summarizing significantly enriched ligand-receptor interactions between sub-clusters in control and crybb2 knockout conditions.

Given that experimental results indicated significant effects of crybb2 deficiency on lens development and vitreous vasculature, we next performed higher-resolution subclustering of cluster 12 (Lens_Fiber) and cluster 22 (Vascular/Endothelial). This analysis resolved nine transcriptionally distinct subgroups (Fig. 5B). The lens compartment was divided into Lens_Fiber_1, enriched for structural crystallins and aquaporins and representing early fiber differentiation, and Lens_Fiber_2, characterized by multiple γ-crystallin genes indicative of more mature fiber cells. The vascular lineage was further resolved into Endothelium_Active, enriched for angiogenic regulators; E ndothelium_Arterial, defined by arterial markers; Vascular_Progenitor, expressing early endothelial lineage genes; and Vascular_Pericytes, marked by perivascular support cell signatures. In addition, three accessory populations were identified, including Proliferating cells with cell cycle gene enrichment, Neuronal-like cells expressing neuronal markers, and Macrophage-like cells with lysosomal and myeloid features. Together, these results consider that at 48 hpf both the lens and vascular compartments are transcriptionally heterogeneous, encompassing progenitor, differentiated, and auxiliary cell states (Supplementary Table S3).

Crybb2 Expression Across Ocular Lineages

We examined the distribution of crybb2 transcripts across ocular cell populations in the single-cell RNA sequencing dataset. Crybb2 expression was readily detected in lens fiber cell clusters, particularly within the Lens_Fiber_2 population. In contrast, crybb2 transcripts were rarely observed in vascular-related populations, including Endothelium_Active, Endothelium_Arterial, Vascular_Progenitor, and Vascular_Pericytes.

Quantitative assessment confirmed that no endothelial or vascular progenitor cells exhibited detectable crybb2 expression (0% of cells with UMI >0 in these clusters), whereas expression was restricted to a small subset of lens fiber cells (Supplementary Fig. S14; Supplementary Table S4). Similar patterns were observed across control and mutant samples. These findings indicate that crybb2 expression at 48 hpf is largely confined to lens lineages, arguing against a strong endothelial cell-autonomous role at this developmental stage.

Differential Remodeling of Lens and Vascular Signaling Networks in Crybb2^+/− Zebrafish

Cell-cell communication analysis using CellChat representin marked differences between WT and crybb2^+/− embryos rather than direct evidence of molecular interaction (Fig. 5C). Compared with WT, both Lens_Fiber_1 and Lens_Fiber_2 displayed markedly reduced outgoing and incoming interaction strengths, with several edges either weakened or absent. This finding is consistent with the biology of lens fiber cells, which undergo terminal differentiation and downregulation of most signaling pathways. Although TEM and immunofluorescence analyses revealed structural disorganization, organelle retention, and autophagosome accumulation in mutant fiber cells, these abnormalities are more consistent with intrinsic defects in cellular homeostasis rather than direct evidence of active intercellular signaling. As a result, lens fiber cells did not show enhanced communication in CellChat despite their clear morphological disruption. In contrast, vascular populations exhibited pronounced changes in communication patterns. The Vascular_Progenitor and Endothelium_Active clusters showed strengthened reciprocal interactions, with pathways such as ESAM, ANGPTL, and collagen signaling prominently upregulated, whereas VEGF signaling activity was reduced. These alterations align with the altered tracer distribution observed in crybb2 mutants by TMR-MEP injection assays and endothelial immunostaining, supporting an association between that vascular compartments actively rewire their signaling networks in response to crybb2 deficiency.

Remodeling of Lens and Vascular Pathways in Crybb2^+/− Embryos

The molecular consequences of crybb2 knockout, we performed CellChat pathway-level information flow analysis, which aggregates ligand-receptor interactions into curated signaling pathways and quantifies their overall activity (Fig. 5D). This comparison revealed a pronounced shift in the balance of developmental, vascular, and ECM-associated signals between WT and crybb2^+/− embryos. Pathways that normally support ocular growth and vascularization, including VEGF, FGF, and EDN, were significantly reduced in crybb2^+/− embryos. The attenuation of VEGF signaling is consistent with diminished Flk-1 expression and impaired vascularization, while loss of FGF activity aligns with developmental defects in the lens and retina. Adhesion and ECM remodeling pathways such as FN1 and NCAM were also downregulated, mirroring the TEM findings of disrupted lens fiber organization and insufficient structural remodeling. By contrast, several pathways were abnormally upregulated in crybb2^+/− mutants. NOTCH signaling showed strong activation, reflecting altered control of cell fate decisions in retinal progenitors and lens epithelial cells. ANGPTL and ESAM signaling were markedly increased, both closely linked to endothelial junctions and vascular permeability, thereby corroborating the experimental observation of enhanced vitreous vessel leakage on TMR-MEP injection. COLLAGEN signaling was also elevated, consistent with ECM accumulation and organelle retention in lens fibers, whereas BMP signaling was aberrantly high, indicating sustained morphogenetic activity beyond normal developmental windows.

Aberrant Endothelial Adhesion and ECM Remodeling Drive Vascular Signaling Rewiring in Crybb2^+/− Zebrafish

CellChat network signal pathway analysis highlighted several signaling pathways that were differentially regulated in crybb2^+/− embryos compared to controls (Fig. 6A). The most striking change was in ESAM signaling, which was strongly enriched in mutants, specifically in Endothelial_Active→ Endothelial_Active interactions. Given that ESAM is an endothelial adhesion molecule frequently associated with barrier dysfunction, its selective upregulation provides compelling molecular evidence for the altered tracer distribution observed in TMR-MEP injection assays. A second major difference was the activation of the VEGF pathway, which was absent in controls but robustly present in mutants. VEGF interactions were detected both as an autocrine loop within Endothelial_Active cells and from Endothelial_Active to Vascular_Progenitors, indicating enhanced self-stimulation of endothelial cells. Such autocrine VEGF loops are characteristic of vascular remodeling processes and are strongly linked to permeability increases.

Figure 6.

Transcription factor profiling and signaling pathway analysis in retinal vascular and lens-associated subclusters following crybb2 knockout. (A) Pseudotime trajectory of single-cell transcriptomes, colored by subcluster identity, showing developmental bifurcation in the Endothelium_Active population. (B) Pseudotime trajectory colored by experimental group, displaying distribution of WT and crybb2^+/− knockout cells. (C) Heatmap of the top 50 transcription factors in endothelial cells. Rows represent TFs, and columns represent groups (WT vs. crybb2^+/−). Expression values are scaled by row (scale = “row”). (D) BEAM branch heatmap of endothelial pseudotime trajectory, showing fate-biased gene modules identified by branch-dependent expression analysis. Genes are ordered by pseudotime and grouped by module identity.

ECM remodeling was also markedly altered, with COLLAGEN-SDC4 signaling significantly enhanced in mutants. This signal originated predominantly from Vascular_Progenitors toward multiple recipient populations, whereas weaker but detectable Lens_Fiber→Vascular_Progenitor COLLAGEN-SDC4 interactions were also observed. This finding is consistent with the TEM evidence of ECM disorganization in the lens and suggests that abnormal lens fiber states may be associated with vascular changes via ECM-related signaling pathways. NOTCH signaling (mediated by jag2b/dll4-notch interactions) was present in both WT and mutants but was more widespread and involved a greater number of receptors in crybb2^+/− embryos. This broader NOTCH activation indicates enhanced endothelial plasticity and remodeling, which is consistent with the observed vascular phenotypes. Finally, the FGF pathway (fgf10a/fgf18a → fgfr) was readily detectable in controls but almost absent in mutants. Loss of FGF signaling likely reflects impaired vascular stabilization and reduced neurotrophic support, in line with the structural instability observed in both lens fibers and vitreous vasculature of crybb2^+/− embryos.

Ligand-Receptor Network Remodeling (ESAM, VEGF, Collagen-SDC4, DLL4-NOTCH) Under Crybb2 Deficiency

Analysis of scRNA-seq data revealed that several key signaling pathways related to vascular function and ECM organization were significantly altered in crybb2^+/− embryos compared with controls. In the ESAM pathway, esama expression was weak and sparse in controls (Supplementary Fig. S15A) but became markedly stronger and more widely distributed in mutants (Supplementary Fig. S15B), with signals concentrated in vascular-related clusters including Vascular_Progenitors, Endothelial_Active cells, and Pericytes, while Lens_Fiber clusters remained almost negative. This vascular-specific increase in esama provides molecular support for the increased endothelial adhesion and vascular permeability phenotypes observed in vivo. In the VEGF pathway, vegfaa expression was readily detected in vascular compartments of controls (Supplementary Fig. S16A) but was reduced in mutants (Supplementary Fig. S16B), consistent with the weakened VEGF signaling predicted by CellChat. In contrast, receptor expression was maintained or enhanced: kdr remained broadly expressed in endothelial populations, and flt1 was strongly upregulated in Vascular_Progenitors and Pericytes, indicating receptor-level compensation despite reduced ligand input. Collagen signaling exhibited the most pronounced remodeling, as shown by CellChat scatter plots (Supplementary Fig. S17A): in controls, Vascular_Progenitors acted as the main ECM signal recipients with little outgoing activity from lens clusters, whereas in mutants both Vascular_Progenitors and Endothelial_Active cells displayed markedly stronger incoming strengths and Lens_Fiber_1 showed a striking increase in outgoing signaling. FeaturePlot analysis confirmed that collagen genes (col4a1, col1a1b, col1a2, col2a1a) were broadly upregulated in both lens fiber and vascular compartments of mutants (Supplementary Fig. S17B). Cluster-specific violin plots further revealed that Lens_Fiber_1 was the primary ECM source with robust induction of all collagen genes, while Lens_Fiber_2 showed modest increases; among vascular populations, Vascular_Progenitors exhibited the strongest upregulation, followed by Endothelial_Active, Arterial, and Pericytes, whereas neuronal- and macrophage-like clusters remained unchanged (Supplementary Fig. S17C). Finally, the NOTCH pathway also showed selective activation in mutants (Supplementary Fig. S18AB): dll4, notch1a, and notch3 were all markedly upregulated in Endothelial_Active, Vascular_Progenitors, and Pericytes, whereas notch2 expression was unchanged, indicating a remodeling of the VEGF-NOTCH axis where reduced VEGF ligand availability is accompanied by increased NOTCH signaling. Together, these results suggest coordinated signaling changes associated with crybb2 knockout, characterized by ESAM upregulation in vascular compartments, reduced VEGF ligand input with compensatory receptor activity, active ECM remodeling driven by lens fibers and vascular cells through collagen gene upregulation, and enhanced NOTCH activation, which may be related to both lens fiber abnormalities and vascular remodeling.

Transcriptional Pathway Profiling by GSVA Reveals Lens-to-Vascular Remodeling Axis in Crybb2^+/− Zebrafish

The transcriptional programs underlying the phenotypes observed in crybb2^+/− zebrafish, GSVA enrichment analysis (Supplementary Table S5) was performed on major subclusters identified by scRNA-seq (Fig. 6B). Three representative clusters were selected for focused interpretation, reflecting the key functional stations of our study: Lens_Fiber_1, the primary site of crybb2 expression and the locus of ultrastructural abnormalities detected by TEM; Endothelium_Active, the vascular endothelial population directly linked to permeability changes and ESAM upregulation; and Vascular_Progenitors, an early endothelial precursor population in which CellChat predicted NOTCH activation. In Lens_Fiber_1, enrichment was observed for epithelial-mesenchymal transition, apical junction, WNT/β-catenin signaling, p53 pathway, and unfolded protein response, indicating enhanced ECM remodeling, polarity changes, and stress activation that are consistent with TEM findings of organelle retention and collagen accumulation and with the collagen-driven signaling inferred from CellChat. In Endothelium_Active, angiogenesis and NOTCH signaling were significantly upregulated, together with KRAS/PI3K/MAPK and WNT/β-catenin signaling, reflecting endothelial activation and remodeling that align with experimental evidence of vascular leakage and adhesion molecule expression. In Vascular_Progenitors, the enriched pathways included angiogenesis, NOTCH signaling, epithelial-mesenchymal transition, hypoxia, and MYC-related programs, suggesting that progenitor populations are transcriptionally primed for differentiation and remodeling under crybb2 deficiency, consistent with the altered vascular fate observed in Flk-1 immunostaining. Collectively, these results outline a stepwise pathological axis whereby Lens_Fiber_1 functions as the primary source of ECM and stress pathway activation, Endothelium_Active reflects direct endothelial dysfunction, and Vascular_Progenitors display altered developmental trajectories, thus providing a transcriptional framework consistent with an association between crybb2-dependent lens abnormalities and secondary vascular remodeling phenotypes.

Transcription Factor Programs in Endothelial Cells Are Rewired in the Crybb2 Knockout

Row-scaled transcription factor (TF) heatmaps restricted to endothelial cells showed a bidirectional reprogramming in the crybb2 knockout (Fig. 6C). Stress/hypoxia and remodeling regulators—ATF3, EPAS1b (HIF-2α), RXRAA, ZNF423, TBX3a, and FOXP1b—were relatively elevated in knockout fish, whereas endothelial/ocular differentiation-linked TFs—KLF17, PAX6b, RORAB, PROX1a, and ZNF292b—were reduced, consistent with a shift from a maturation-maintaining program toward stress/remodeling control in the endothelium (Fig. 6C). At single-cell resolution, TF expression resolved into group-specific, state-defining patterns (Supplementary Fig. S19A). Knockout cells showed coherent activation of regulators across subsets (notably TWIST1a and ACTB1, with KLF17 displaying cluster-restricted enrichment), while controls preferentially expressed FOXO family factors and ZEB2b, indicative of baseline homeostatic/differentiation states (Supplementary Fig. S19A). The single-cell view also helps reconcile the row-scaled average: TFs like KLF17 can appear overall reduced in the bulk endothelial comparison (Fig. 6C) yet remain focally retained or heightened within specific knockout subclusters (Supplementary Fig. S19A), underscoring a redistribution of TF activity rather than a uniform gain/loss. TF dynamics in endothelial clusters, we examined expression in vascular compartments, revealing consistent crybb2^+/−-specific upregulation of atf3, her6, cebpb, ssrp1a, and nfkb1 (Supplementary Fig. S19B). Among these, her6 is a canonical Notch downstream TF, strongly supporting the observed Notch signaling activation in CellChat and GSVA analyses. Atf3 and cebpb are stress/inflammation-related TFs, suggesting increased endothelial stress and activation. Ssrp1a, involved in chromatin remodeling, indicated broad transcriptional reprogramming. On the other hand, TFs downregulated in crybb2^+/− embryos, including foxc1a and smad3b, are essential for endothelial maturation and TGF-β pathway regulation, implying compromised vessel stabilization in the mutant background.

Lens Fiber Cell Terminal Differentiation and Alters Pseudotemporal Gene Expression Dynamics

We performed high-resolution differential expression and pseudotemporal trajectory analyses. In the volcano plot (Supplementary Fig. S20A), several key genes such as gnpt2b, dennd2da, LRRTM4.5, ahcy1, and si:dkey-117m1.4 were markedly downregulated, exhibiting high −log₁₀(P_adj) values, indicating strong statistical significance. Additional genes, including zic6, ing1, pcloa, and zfx, also fell within the log₂FC < −2.5 range, suggesting potential involvement in crybb2-related lens dysfunction, impaired autophagy, or developmental arrest. Single-cell level visualization confirmed the robustness of these findings. In the sample-specific expression heatmap (Supplementary Fig. S20B), multiple genes (si:dkey-117m1.4, gngt2b, zic6, zfx, dennd2da, ahcyl1) showed near-complete silencing in crybb2-deficient cells, while being strongly expressed in localized regions of the control lens fiber population. These genes are functionally linked to visual signal transduction (gngt2b), developmental transcriptional regulation (zic6, zfx), vesicle trafficking and metabolic control (dennd2da, ahcyl1), and potentially lens-specific regulation (si:dkey-117m1.4). Their absence in crybb2 mutants suggests a loss of terminal differentiation programs in lens fiber cells.

FeaturePlot analysis provided spatial context for these transcriptional changes (Supplementary Fig. S20C). In the control group, expression of gngt2b, zic6, and zfx localized to a restricted UMAP sub region, indicating a shared regulatory module likely associated with a late-stage lens fiber cell state. In crybb2 mutants, these expression hotspots were entirely absent, supporting the hypothesis that this terminal differentiation state is lost. The overlap of expression patterns among these genes further implies coordinated regulation within the same developmental program.

Pseudotemporal heatmap analysis revealed distinct temporal expression modules in lens fiber cells (Supplementary Fig. S20D). Module 1 genes, including orym × family members, capn3a, and lim2.3, were highly expressed in early pseudotime but rapidly downregulated, consistent with early structural assembly and organelle clearance events. Module 2 genes (fabp7a, col1a2, hsp90ab1) displayed late-stage upregulation, suggesting activation of stress response, ECM remodeling, and membrane repair mechanisms-potentially as compensatory processes in crybb2 mutants. Module 3 genes (ppiaa, ofl1, marcks1a) showed moderate activation in mid-to-late pseudotime, reflecting residual metabolic and protein modification activity. Branch-specific BEAM analysis (Supplementary Fig. S20E) suggested that crybb2 deficiency is associated with altered developmental trajectories of lens fiber cells. Genes such as crygm family members, fabp4a, and tmsb4x exhibited branch-specific enrichment, indicating that crybb2 loss may disrupt the normal temporal progression of lens fiber maturation and may be associated with shifts toward alternative cellular states (Fate 2) relative to canonical terminal differentiation (Fate 1).

Crybb2^+/− Zebrafish Endothelial Transcriptional Reprogramming With Impaired Structural Maintenance

In vitreous vascular endothelial clusters, crybb2 loss produced two clear transcriptional shifts (Supplementary Table S6). Downregulated genes, enriched in controls, included matn1, col10a1a, SEZ6, KIRREL3.35, GRIN2A, mgat4b, klf17, and akap12a, indicating reduced ECM anchoring, polarity maintenance, and structural stability-consistent with altered VEGF/Notch input and increased permeability (Supplementary Fig. S21A). Upregulated genes, enriched in crybb2 mutants, were dominated by stress-response and cytoskeletal remodeling factors such as actb1/2, ybx1, hnrnpa0l/b, tpm4, myl12.1, tnmd, and tnfaip6, pointing to cytoskeletal reorganization and a stress-like endothelial state. GO analysis highlighted enrichment of vessel morphogenesis, lymphangiogenesis, and protein-folding stress responses, whereas KEGG pathways indicated activation of MAPK signaling, regulation of the actin cytoskeleton, and gap junction remodeling-linking transcriptional changes to impaired barrier integrity and endothelial destabilization (Supplementary Fig. S21B). These results suggest that crybb2 deficiency is associated with a shift in endothelial cell states, from a mature, polarity-stable phenotype toward a more stress-responsive and destabilized profile.

Fate-Biased Reprogramming of Endothelial Cells in Crybb2^+/− Zebrafish

Transcriptional consequences of crybb2 heterozygous deletion in vascular endothelial populations, we performed pseudotime trajectory analysis focusing exclusively on the Endothelium_Active subpopulation. The reconstructed trajectory revealed a clear bifurcation, indicating a potential fate decision event during endothelial maturation (Fig. 6D1). Both control and crybb2^+/− cells were distributed along the full trajectory, yet crybb2^+/− cells preferentially accumulated along the lower branch, suggesting a bias toward an alternative differentiation state (Supplementary Fig. S22A). When colored by pseudotime, the trajectory showed a smooth and continuous gradient from a central root toward both branches, confirming the robustness of the inferred developmental progression (Fig. 6D2).

Branch-specific expression dynamics assessed by BEAM analysis uncovered four gene modules with distinct temporal and fate-biased patterns (Supplementary Fig. S22B). Cluster 1 genes (e.g., cdh6, ebf3a, nav3) were enriched in fate 1 during early pseudotime, reflecting roles in adhesion and early vascular morphogenesis. Cluster 2 contained canonical endothelial identity genes such as flt1 and kdr, expressed across both fates but peaking in mid-pseudotime, consistent with their maintenance role. Notably, Cluster 3 (tgfbr3, itga1, spns2) and Cluster 4 (bmp6, cebpd, FLNA, prdx1) were fate 2 biased, associated with TGF-β signaling, ECM-cell interactions, oxidative stress responses, and cytoskeletal remodeling—features aligned with crybb2^+/−-specific endothelial remodeling.

Gene-level pseudotime trends further supported a progressive fate shift in crybb2^+/− endothelial cells (Supplementary Fig. S22C). Early endothelial identity markers such as kdr remained largely stable, indicating preserved lineage identity in the early phase. However, mid-to-late pseudotime showed marked upregulation of stress-associated atf3, structural regulator FLNA, and ECM remodeling factor bmp6, alongside tgfbr3 activation-hallmarks of fate deviation toward a stress- and remodeling-oriented terminal state. In contrast, cdh6, robustly induced in late pseudotime in controls, was almost absent in crybb2^+/−, pointing to impaired maturation and junctional stability. Static expression comparisons from violin plots confirmed these dynamic findings (Supplementary Fig. S22D). Cdh6, ebf3a, kdr, and tgfbr3 were significantly downregulated in crybb2^+/− cells, highlighting disruption of both fate 1 adhesion programs and fate 2 signaling capacity. Conversely, atf3 and cebpd were markedly upregulated, consistent with an activated stress response and organelle damage seen in ultrastructural analyses. Bmp6 and FLNA showed limited baseline differences but were late-phase upregulated in pseudotime, suggesting their effects manifest predominantly at terminal stages.

Crybb2 Knockout Remodels Cell Cycle Composition in Lens and Vascular Endothelial Subpopulations

After trajectory and fate analyses, we next assessed the impact of crybb2 knockout on cell cycle states across multiple subpopulations (Supplementary Fig. S23). Distinct alterations were observed in several clusters. In proliferating cells, the proportion of S-phase increased (50% vs. 42.9%) with a concomitant decrease in G2M, suggesting either enhanced DNA synthesis activity or S-phase arrest. In Endothelium_Active cells, the G1 fraction was reduced (55.7% vs. 65.3%) alongside a pronounced rise in S-phase (30.7% vs. 20.4%), indicating premature entry into DNA replication, consistent with the observed fate shift and tgfbr3 upregulation. Lens Fiber 1 and Lens Fiber 2 clusters displayed dramatic S-phase increases (from 0%→28% and 0%→40.6%, respectively), indicating terminal differentiation defects in agreement with ultrastructural findings of organelle retention and autophagosome accumulation. Endothelium_Arterial cells exhibited an emergent G2M-phase in the knockout group (6.4% vs. 0%), whereas vascular progenitor cells showed only a mild S-phase increase. Underscoring a critical role for crybb2 in coordinating cell cycle regulation and maintaining tissue homeostasis.

Discussion

CC is a complex genetic disorder, the pathogenesis of which involves gene mutations and abnormalities in multiple cellular signaling pathways. In this study, we explored the relationship between these phenotypes and behavioral changes using WES and a zebrafish model. Subsequently, single-cell sequencing analysis was conducted to comprehensively investigate the mechanisms underlying the CRYBB2 splicing mutation in CC, particularly its impact on the Wnt/β-catenin signaling pathway and oxidative stress response. Additionally, we analyzed in detail the effects of the CRYBB2 gene mutation on lens development, retinal neurodevelopment, and hyaloid vasculature development using the zebrafish model.

The βB2-crystallin protein encoded by the CRYBB2 gene is crucial for maintaining lens transparency.²⁶ The c.450-2A>G mutation identified in this study in affected abnormal splicing of exon 6, thereby affecting the structure and function of the CRYBB2 protein. This variant was identified in affected family members who were tested and was absent from public databases, supporting its classification as pathogenic according to ACMG/AMP criteria. Mutations in the CRYBB2 gene have been associated with various phenotypes of CC, and such mutations may lead to abnormal protein folding, aggregation, or degradation, thereby affecting normal lens development.²⁷ In our zebrafish model, crybb2 knockout not only impaired lens structure but also led to marked changes in vitreous vascular morphology and stability. Behavioral experiments further revealed that mutant zebrafish showed distinct movement patterns compared to wild-type controls: their swimming trajectories were restricted to peripheral regions with a reduced range of motion, suggesting altered visual input or refractive properties because of lens-vitreous abnormalities. Quantitative tracking using EthoVision XT indicated significantly lower travel distance, swimming speed, and maximum acceleration in mutants, pointing to reduced visual-guided locomotor performance. These findings highlight the pivotal role of crybb2 in maintaining the structural and optical integrity of the lens and vitreous, and suggest that disruption of this gene triggers secondary biomechanical and optical changes that impair visual behavior in zebrafish.^28,29

AlphaFold-based structural modeling predicted that the CRYBB2 mutant protein exhibits significant conformational alterations, particularly in the N-terminal arm and Greek key motifs, which are essential for βB2-crystallin stability and intermolecular packing in the lens fiber cell environment. These structural deviations could impair the correct folding and assembly of crystallin oligomers, thereby weakening the physical and optical stability of the lens. MD simulations further revealed reduced intramolecular hydrogen bonding and altered surface electrostatic distribution in the mutant protein, indicating a lower structural resilience under physiological fluctuations. These structural defects are consistent with findings that other congenital crystallin variants similarly compromise protein stability and solubility through MD‑detected conformational shifts and aggregation tendencies.²⁶ Regarding the molecular mechanism of the CRYBB2 c.450-2A>G mutation, our in vitro translation assay bridges the gap between the genetic splicing defect and the structural phenotype. This mutation generates robustly expressed, aberrant protein forms, presenting as a distinct doublet on Western blot. Quantitative analysis further showed that the total CRYBB2 protein level (including both aberrant isoforms) was significantly increased in the mutant group compared with the wild-type group (P = 0.0026, Supplementary Fig. S5B), indicating accumulation of aberrant CRYBB2 protein products. Consistent with the established pathomechanisms of autosomal dominant CC, these truncated mutant crystallins—which lack critical C-terminal Greek key motifs—are highly prone to misfolding. Rather than being eliminated, their robust expression and subsequent accumulation likely drive a dominant-negative effect by co-aggregating with normal, wild-type crystallins expressed from the unaffected allele in heterozygous patients. This pathogenic co-aggregation not only depletes the functional, highly soluble crystallin pool essential for lens transparency but also triggers cellular stress responses, ultimately driving the severe lenticular opacities observed in the patients and our zebrafish models.

In addition to their structural role in the lens, certain β-crystallins have been reported to exert functional roles outside of lens fiber cells. Notably, Sinha et al. demonstrated that βA3/A1-crystallin (Cryba3) is expressed in retinal astrocytes and is essential for normal retinal vascular remodeling during development. In Cryba3-deficient rodents, disruption of astrocyte morphology was accompanied by abnormal retinal vascular patterning and increased vascular leakage, indicating a cell-autonomous role of βA3/A1-crystallin in vascular development.⁸ These findings raise the important question of whether other closely related β-crystallins, including CRYBB2, may similarly participate directly in vascular or retinal cell function.

In contrast to the Cryba3 model, our single-cell RNA sequencing analysis revealed that crybb2 expression at 48 hpf is largely restricted to lens fiber cell populations and is undetectable in endothelial or vascular progenitor clusters. This spatial confinement argues against a primary endothelial cell-autonomous function of crybb2 during early zebrafish development. Instead, our data support a lens-initiated mechanism, whereby structural and extracellular matrix alterations in lens fiber cells secondarily influence adjacent vitreous vasculature. Thus, although βA3/A1- and βB2-crystallins belong to the same crystallin family, their extra-lenticular roles may differ substantially in tissue specificity and developmental context.

Our ultrastructural validation provides converging evidence that crybb2 knockout primarily affects the lens and vitreous compartments, rather than retinal neuron development. TEM of lens fiber cells in mutant zebrafish revealed persistent organelles, abnormal mitochondrial morphology, and frequent mitochondria-lysosome fusion bodies—signatures of impaired terminal differentiation and disrupted degradative clearance. These features are consistent with a critical requirement for autophagy in lens homeostasis and transparency,³⁰ as summarized in recent literature highlighting that deficient autophagic pathways compromise organelle elimination during fiber cell maturation, leading to cataractogenesis. Histological analysis using H&E staining revealed that crybb2 knockout zebrafish lenses displayed disorganized fiber cell arrangement, irregular cell boundaries, and focal vacuole-like changes, indicating disruption of normal lens architecture. These morphological changes align with congenital cataract phenotypes in other lens structural gene mutations, where disrupted fiber compaction compromises refractive index uniformity and transparency.^31,32 Immunofluorescence staining further confirmed that crybb2 loss alters ECM composition and fiber cell-fiber cell interactions. Specifically, there was marked upregulation and spatial redistribution of collagen IV and laminin within the lens capsule and adjacent vitreous interface, consistent with aberrant ECM remodeling. These changes likely enhance mechanical signaling to neighboring ocular tissues, as excessive ECM deposition and altered organization have been reported to perturb lens capsule elasticity and contribute to secondary vitreous remodeling.^33,34 Importantly, the observed ECM changes parallel the CellChat prediction in our single-cell analysis, where lens fiber cells in crybb2 mutants showed increased outgoing collagen and laminin signaling toward vascular endothelial clusters. These observations are consistent with a potential association between primary lens pathology and secondary vascular remodeling. The combination of H&E histology and ECM-targeted immunofluorescence thus offers strong morphological and molecular validation of the transcriptional programs identified in our scRNA-seq and pathway analyses.

Our vascular-focused molecular validation experiments revealed significant alterations in the vitreous hyaloid vasculature in crybb2 knockout zebrafish. Immunofluorescence staining for endothelial adhesion molecules showed increased ESAM and collagen IV deposition along the vascular basement membrane, suggesting endothelial activation and ECM remodeling. Such ECM enrichment, especially at the vitreoretinal interface, has been linked to altered vascular stiffness and altered paracellular transport in ocular vascular disease models.^35,36 Functional tracer injection assays using intravitreal injection of the fluorescent tracer TMR-MEP showed reduced tracer diffusion in crybb2 mutants compared to controls, suggesting altered paracellular transport. This phenotype is consistent with pathway-level associations identified by our scRNA-seq and CellChat analyses, which supported an association between upregulation of cell-cell junction and ECM-receptor interaction pathways in Endothelium_Active clusters. Altered tracer distribution may be due to tight junction reinforcement or abnormal basement membrane thickening—mechanisms also reported in models with VEGF pathway dysregulation.^37,38 Because crybb2 mRNA was ubiquitously expressed, this rescue experiment does not resolve tissue-specific requirements and should be interpreted as functional support rather than evidence of cell-autonomous action. As with all inference-based analyses, these results should be interpreted as associative rather than causal.

The discrepancy in phenotypic severity between human patients and zebrafish may be attributed to the accelerated developmental timeline of the zebrafish eye. Our scRNA-seq data confirms a broad, albeit low-level, expression of crybb2 in the early zebrafish ocular cup, which may render the teleost eye more susceptible to the dominant-negative effects of truncated CRYBB2 proteins. This suggests that the crybb2 mutant zebrafish serves as a sensitized model to uncover secondary developmental defects—such as vascular instability—that might be masked or prenatally resolved in human patients.

Despite the pronounced structural and functional alterations in the lens and vitreous compartments, our molecular and histological assessments revealed no substantial abnormalities in retinal neuronal or photoreceptor-related compartments of crybb2 knockout zebrafish. Single-cell RNA-seq analysis showed comparable distributions of major retinal cell types, including photoreceptors, bipolar cells, horizontal cells, and Müller glia, between mutant and control groups, with no significant changes in key marker gene expression. Immunofluorescence staining for retinal ganglion cells (HuC/D), photoreceptor markers (e.g., zpr1 for cones, zpr3 for rods), and Müller glial markers (GS) similarly revealed no notable differences in cell number, spatial arrangement, or morphology across genotypes. These observations suggest that the crybb2 mutation does not exert measurable effects on retinal neurogenesis or photoreceptor maintenance at the developmental stages examined. This aligns with prior reports that lens-specific crystallin mutations may produce cataract and secondary vitreous alterations without directly impairing retinal architecture.^39,40 It is plausible that any potential retinal effects are secondary, arising from optical quality loss and altered vitreous-retina signaling over longer periods, rather than direct disruption of retinal differentiation pathways.

Our single-cell RNA-seq analysis, combined with ultrastructural and molecular validation, supports a conceptual model in which crybb2-associated lens abnormalities may be associated with secondary changes in the vitreous vasculature. through altered ECM composition and paracrine signaling. GSVA pathway enrichment in Lens_Fiber_1 highlighted epithelial-mesenchymal transition, apical junction disruption, Wnt/β-catenin activation, p53 signaling, and unfolded protein response. TEM confirmed persistent organelles, mitochondrial-lysosome fusion bodies, and collagen deposition, consistent with terminal differentiation arrest and stress-induced ECM remodeling. Such ECM remodeling is known to alter biomechanical and biochemical cues in the ocular microenvironment, which may signal to adjacent vascular compartments.^41,42

Endothelium_Active and Vascular_Progenitor clusters showed upregulation of angiogenesis, VEGF, NOTCH, PI3K/MAPK, and Wnt/β-catenin pathways, alongside fate bias in pseudotime analysis toward stress-responsive trajectories.⁴³ BEAM branch-specific genes (e.g., tgfbr3, bmp6, cebpd, atf3) indicate activation of TGF-β and oxidative stress modules, aligning with altered tracer distribution observed in tracer injection assays. This transcriptional reprogramming suggests that altered ECM and growth factor milieu from the lens may destabilize endothelial junctions and promote maladaptive remodeling.⁴⁴ Cell cycle profiling showed increased S-phase fractions in the Lens_Fiber_1 and Endothelium_Active subpopulations, indicating altered cell cycle dynamics in these populations under crybb2-deficient conditions. This proliferative skew may be driven by Wnt/β-catenin and TGF-β crosstalk, both implicated in pathological angiogenesis and cataract-associated vascular changes.^45,46

A key finding of our study is the identification of secondary ocular defects, including hyaloid vascular remodeling and retinal alterations, in the crybb2^+/− zebrafish model. Our scRNA-seq analysis revealed that crybb2 is not endogenously expressed in the vascular endothelium or retina, suggesting that these phenotypes are not cell-autonomous. Instead, they likely arise from the disruption of the “Lens-Vascular Axis.” The lens is not merely an optical element but a crucial signaling center that secretes factors required for the maintenance of the surrounding ocular environment. Proteostatic stress and fiber degeneration associated with the CRYBB2 mutation may be accompanied by alterations in the lens secretome—particularly the reduced ESAM and ANGPTL signaling suggested by our CellChat analysis—which may contribute to hyaloid vascular instability. The prominence of these defects in zebrafish, as opposed to human patients, may be due to the intense developmental crosstalk between the lens and hyaloid vessels during the rapid embryogenesis of the teleost eye, providing a unique “window” to observe these lens-derived paracrine effects.

A limitation of our clinical study is the restricted availability of detailed objective ophthalmic imaging across the large pedigree, with comprehensive clinical imaging available primarily for the proband and her affected parents. Relying on self-reported medical histories for other family members may limit the complete characterization of the disease. Given that CRYBB2 mutations frequently exhibit variable expressivity and intrafamilial variability, the severe phenotype observed in the proband may not fully represent the clinical spectrum of the entire pedigree. Another limitation of our study is the use of a single pooled biological replicate per condition for the scRNA-seq analysis. Although pooling approximately 200 embryos per group helps mitigate individual variation and the samples were processed in a single batch to avoid technical batch effects, interpretations of subtle transcriptional differences should be made with caution. To address this, we ensured that major scRNA-seq findings—such as ECM remodeling and vascular signaling alterations—were orthogonally validated through independent functional and morphological assays.

Furthermore, it is important to acknowledge the limitations of the zebrafish model when interpreting the clinical translational implications of our study. The phenotypes of vascular instability and hyaloid vascular remodeling were mainly observed in zebrafish larvae. Given species-specific developmental differences—such as the prominent and sustained role of the hyaloid vasculature during early zebrafish development compared with its typical prenatal regression in humans—these findings should be interpreted with caution when considering potential relevance to human patients. Therefore, although our study provides a novel mechanistic framework for the lens-vascular axis in CC, its direct clinical translational value should be interpreted with caution. Future studies utilizing mammalian models and high-resolution ocular vascular imaging (such as OCTA) in clinical cohorts are required to validate these findings.

In this study, we present an integrated view of lens–vascular interactions in crybb2-deficient zebrafish by combining ultrastructural analysis, molecular validation, and single-cell transcriptomics. Our findings indicate that crybb2 deficiency is associated with impaired terminal differentiation of lens fiber cells, characterized by persistent nuclei and organelles, abnormal mitochondrial–lysosome interactions, and remodeling of the extracellular matrix. These primary alterations in lens structure and homeostasis coincide with transcriptional changes in adjacent vitreous vascular endothelial populations, including shifts in angiogenic and stress-associated signaling pathways and altered endothelial cellular states. In contrast, retinal neuronal compartments remained largely unaffected during early developmental stages, suggesting that crybb2-related pathology primarily originates within the lens and its immediate microenvironment.

Based on these observations, we propose a lens–vascular remodeling model in which structural and signaling changes arising from the diseased lens are associated with secondary alterations in hyaloid vascular organization and endothelial cell behavior (Supplementary Fig. S24). This framework expands current understanding of crystallin proteins beyond their classical refractive roles and highlights their potential contributions to maintaining ocular tissue homeostasis during development. More broadly, these findings suggest that CC accompanied by vascular abnormalities may involve coordinated changes across multiple ocular compartments. Further studies will be required to clarify the molecular mediators linking lens pathology to vascular remodeling and to determine whether similar mechanisms operate in human congenital cataract disorders.