Cataract is a leading cause of reversible childhood blindness worldwide and is largely attributable to mutations in crystallin-encoding genes of the α-, β-, and γ-crystallin families. Crystallins account for over 90% of total lens proteins and are indispensable for maintaining lens transparency and long-term proteostasis. These three crystallin families display distinct structural organizations and functional specializations: α-crystallins function as small heat shock proteins (sHSPs) with molecular chaperone activity, whereas β- and γ-crystallins serve primarily structural roles, characterized by conserved Greek-key motifs that support dense protein packing and appropriate refractive properties of the lens. Despite these differences, mutations across α-, β-, and γ-crystallins converge on common pathogenic mechanisms by perturbing protein stability, intermolecular interactions, and solubility, thereby promoting protein misfolding, aggregation, or aberrant phase behavior. Such alterations disrupt lens protein homeostasis, increase light scattering, and ultimately lead to lens opacification. Cataract-related imaging and genetic data are increasingly being analyzed using emerging computational approaches, among which artificial intelligence (AI) represents a rapidly developing direction that may further support the interpretation of disease phenotypes and molecular variants in future studies. In this review, we synthesize current knowledge on human crystallin variation, pathogenic mechanisms of crystallin gene mutations, and cataract phenotypes, and discuss how AI-assisted approaches may further refine mechanistic interpretation and molecular characterization of cataract.
Ma et al. (Mon,) studied this question.