Patrick Yu-Wai-Man

The molecular programs that establish specialized retinal regions during development are essential for high-acuity vision, yet how their disruption contributes to human visual disorders remains poorly understood. Bosch-Boonstra-Schaaf optic atrophy syndrome (BBSOAS), caused by pathogenic variants in NR2F1 and characterized by visual impairment, provides an opportunity to investigate these mechanisms. Using single-cell RNA sequencing of three complementary Nr2f1 mouse models, including two carrying patient-specific mutations, we identified a shared Nr2f1 -dependent transcriptional program enriched in retinoic acid (RA) pathway genes. Loss or mutation of Nr2f1 disrupted the spatial organization of RA signaling, most prominently by expanding the dorso-equatorial Cyp26a1 expression domain into ventral retina and reducing ventral determinants such as Vax2 . These molecular changes were associated with altered dorso-ventral distribution of S-and M-opsin-expressing cone photoreceptors. We further demonstrate that human NR2F1 binds a conserved regulatory region upstream of CYP26A1 , supporting its direct role in regulating local RA availability. Finally, high-resolution optical coherence tomography in individuals with BBSOAS revealed reproducible foveal abnormalities, including a smaller and shallower foveal pit and increased central retinal thickness, consistent with foveal hypoplasia. These findings uncover a previously unrecognized retina-intrinsic component of BBSOAS visual pathology and establish an NR2F1-RA/CYP26A1 regulatory axis linking developmental retinal regionalization to human foveal specialization.

Diabetic retinopathy (DR) arises from intertwined inflammatory, metabolic, and hypoxia-driven angiogenic programs, yet upstream regulators coordinating these processes remain incompletely defined. Here, we used an integrative multi-omics and experimental framework to identify cathepsin H (CTSH) as a candidate causal driver of proliferative DR (PDR). By combining GWAS, eQTL, pQTL, and mQTL datasets with Mendelian randomization, summary-data-based Mendelian randomization, and Bayesian colocalization, CTSH emerged as the strongest genetically supported candidate across discovery and validation analyses. In the UK Biobank (UKB), circulating CTSH was elevated in diabetic retinopathy and independently predicted incident disease. Single-cell transcriptomic analyses localized CTSH predominantly to myeloid compartments within fibrovascular membranes and linked CTSH-high states to inflammatory, hypoxic, and angiogenic programs. In high-glucose-stimulated THP-1 monocytes, CTSH promoted reactive oxygen species accumulation, NF-κB activation, and increased IL-6, TNF-α, HIF-1α, and VEGF expression, whereas CTSH silencing reversed these effects. Structure-guided virtual screening identified Eriocitrin as a lead CTSH-binding candidate. In db/db mice, intravitreal Eriocitrin improved inner-retinal function, restored OCTA-derived vascular metrics, and partially rescued retinal structure, with efficacy comparable to anti-VEGF treatment across several endpoints. Molecular analyses further showed coordinated suppression of inflammatory, hypoxic, angiogenic, and NF-κB signaling. Together, these findings identify CTSH as an upstream immunometabolic regulator of DR-related inflammatory and angiogenic biology, with the strongest genetic support observed for PDR, and support CTSH targeting as a potential multi-pathway therapeutic strategy beyond VEGF inhibition.

Background: Lenadogene nolparvovec is an intravitreal gene therapy for patients with Leber hereditary optic neuropathy (LHON) carrying the m.11778G>A MT-ND4 variant. Idebenone, a synthetic coenzyme Q ₁₀ analogue, is the only approved treatment for LHON. To estimate the relative effects of both treatments, we performed two matching adjusted indirect comparisons (MAICs) between idebenone aggregated data from the LEROS study and expanded access programme (EAP), and lenadogene nolparvovec individual data from the REFLECT study (bilateral treatment). Methods: Matching covariates included age at disease onset, sex, baseline best-corrected visual acuity (BCVA), and time from vision loss to treatment. The outcomes of interest were clinically relevant recovery (CRR) from nadir at 24 months, time to initial CRR and change from baseline BCVA at 24 months. Results: For the MAIC LEROS versus REFLECT (effective sample size (ESS): 77), a statistically higher CRR at 24 months was observed with lenadogene nolparvovec compared with idebenone (60.4% vs 35.4%; OR=2.78, 95% CI 1.53 to 5.06; p=0.001). No statistically significant difference was observed for time to initial CRR and change from baseline BCVA at 24 months. For the MAIC EAP versus REFLECT, there was a low overlap between the two populations related to a difference in the time from vision loss to treatment. In a post hoc sensitivity analysis comparing EAP to RESCUE data (ESS: 33), CRR was 39.0% for idebenone versus 69.5% for lenadogene nolparvovec (OR=3.59, 95% CI 1.42 to 9.06; p=0.011). Conclusion: The two MAICs demonstrated a clinically meaningful higher visual recovery at 24 months with lenadogene nolparvovec than with idebenone in patients with LHON due to the m.11778G>A MT-ND4 variant.

Glaucoma is a leading cause of irreversible blindness, yet the circulating proteins and metabolic pathways that causally contribute to different glaucoma subtypes remain poorly defined. We analyzed baseline plasma proteomics in 1485 glaucoma cases (447 primary open‑angle glaucoma [POAG], 177 primary angle-closure glaucoma [PACG], 120 normal-tension glaucoma [NTG]) in the UK Biobank using Cox models with graded adjustment. We then integrated five independent protein quantitative trait loci resources with FINLAND R12 genome-wide association study data to perform two‑sample Mendelian randomization (MR) and cross‑cohort meta‑analysis for overall glaucoma and each subtype. To prioritize effector genes and pathways, we conducted summary-data-based Mendelian randomization (SMR) using eQTLGen and two‑step mediation MR using metabolite quantitative trait loci data for ∼1400 plasma metabolites from the Canadian Longitudinal Study on Aging cohort. In fully adjusted Cox models, 484 proteins were associated with incident glaucoma, 135 with NTG, 59 with POAG, and 1 with PACG (false discovery rate <0.05). Multicohort MR and meta‑analysis identified eight proteins with robust causal effects: NRP2, TSPAN1, and HAVCR2 for overall glaucoma; NRXN3 for PACG; MANSC4 for NTG; and LTBP2, CD69, and SMAD1 for POAG. SMR supported NRP2 (overall glaucoma) and SMAD1 (POAG) as causal genes. Mediation MR revealed that sphingomyelins, acylcarnitines, and bile acid-related metabolites partially mediated the effects of several proteins, defining shared (e.g., sphingolipid) and subtype‑specific metabolic pathways. By integrating epidemiologic, proteomic, genetic, and metabolomic data, we identify convergent systemic protein and metabolic signatures associated with glaucoma susceptibility and its clinical subtypes. These findings nominate NRP2, SMAD1, and related pathways as promising biomarkers and therapeutic targets and support a systems‑level view of glaucoma pathogenesis beyond intraocular pressure alone.

Inherited optic neuropathies (IONs), such as Leber hereditary optic neuropathy (LHON) and autosomal dominant optic atrophy (ADOA), typically lead to irreversible severe vision loss due to mitochondrial dysfunction causing retinal ganglion cell degeneration. Although current treatment options are limited, substantial progress has been made recently in our understanding of the molecular genetic pathways that lead to retinal ganglion cell loss. Clinical trials for LHON have demonstrated the efficacy of idebenone, an oral neuroprotective agent, and gene replacement therapy using allotopic gene expression. Early phase clinical trials are underway for ADOA caused by variants in the nuclear gene OPA1 using innovative techniques to modulate gene expression in a variant-agnostic manner. In this review, we have critically appraised a range of therapeutic strategies, including gene editing and stem cell-based optic nerve regeneration, with a discussion of the barriers to translation. Future studies focussing on understanding genetic heterogeneity, disease variability and optimising patient selection for clinical trials are essential to improve patient management and fast track transformative therapies for IONs.

A 22-year-old woman had an 8-year history of progressive bilateral vision loss and of diabetes mellitus. Her mother had diabetes and two first cousins had severe congenital deafness. On examination, her visual acuities were 6/36 bilaterally, with absent colour vision and gross optic disc pallor. Her HbA1c was 94 mmol/mol (20–42). Visual evoked potentials were bilaterally symmetrically prolonged with P100 140 ms. Optical coherence tomography showed significant thinning of the peripapillary retinal nerve fibre layer. MR scan of brain and orbits showed small optic nerves and chiasm. Genetic testing confirmed a diagnosis of autosomal recessive Wolfram syndrome-1. She is currently registered blind. We review the clinical and genetic features of Wolfram syndrome.

Dominant Optic Atrophy (DOA) is the most common inherited optic neuropathy and presents as gradual visual loss caused by the loss of retinal ganglion cells (RGCs). Over 60% of DOA cases are caused by pathogenic variants in the OPA1 gene, which encodes a mitochondrial GTPase essential in mitochondrial fusion. Currently, there are no treatments for DOA. Here, we tested the therapeutic potential of an approach to DOA using CRISPR activation (CRISPRa). Homology directed repair was used to introduce a common OPA1 pathogenic variant (c.2708_2711TTAGdel) into HEK293T cells as an in vitro model of DOA. Heterozygous c.2708_2711TTAGdel cells had reduced levels of OPA1 mRNA transcript, OPA1 protein, and mitochondrial network alterations. The effect of inactivated Cas9 fused to an activator (dCas9–VPR) was tested with a range of guide RNAs (gRNA) targeted to the promotor region of OPA1. gRNA3 and dCas9–VPR increased OPA1 expression at the RNA and protein level towards control levels. Importantly, the correct ratio of OPA1 isoform transcripts was maintained by CRISPRa. CRISPRa-treated cells showed an improvement in mitochondrial networks compared to untreated cells, indicating partial rescue of a disease-associated phenotype. Collectively, these data support the potential application of CRISPRa as a therapeutic intervention in DOA.

Primary mitochondrial disorders (PMDs) are a clinically heterogeneous group of genetic disorders that can affect many tissues, with a broad phenotypic spectrum ranging from isolated organ involvement to severe early-onset multisystem disease. Visual loss from optic atrophy is a frequent clinical manifestation of mitochondrial cytopathies. This study aimed to identify the missing heritability in previously unsolved cases of suspected isolated or syndromic optic neuropathy. Based on three recent reports on biallelic NSUN3 variants causing early-onset PMD, we explored in detail the genetic and clinical spectrum of NSUN3-associated disease. Affected individuals were analyzed by exome or genome sequencing. In silico variant analysis and functional assays were performed to investigate the consequences of the identified variants. Detailed phenotyping data were collected from medical records and direct questioning after the identification of candidate-likely pathogenic variants. Interrogation of exome and genome sequencing data led to the identification of six candidate NSUN3 variants in eight affected individuals from five unrelated families (including a previously reported case). A broad phenotypic spectrum was observed ranging from isolated optic atrophy to severe early-onset PMD. Identified NSUN3 variants impairing NSUN3 activity are located within the S-adenosylmethionine-dependent methyltransferases domain and loss of function variants were associated with a more severe phenotype. Remarkably, bilateral optic atrophy was a unifying clinical feature observed in almost all affected individuals. Pathogenic or likely pathogenic biallelic variants in NSUN3 disrupt mt-tRNA^Met methylation and mitochondrial translation leading to mitochondrial disease ranging from mild isolated optic atrophy to a severe multisystemic phenotype with possible limited life expectancy.