Richard Rankin

Current human retina studies predominantly utilize post-mortem tissue, and the sample accessibility constraints make the characterization of the living human retina at single-cell resolution a challenge. Although single-nucleus RNA-seq expands the utility of frozen samples, it provides a nuclear-centric view, potentially missing key cytoplasmic information and transient biological processes. Thus, it is important to generate resources directly from living human retinal tissue to complement existing datasets. We profiled 106,829 single cells from nine unfrozen human retina samples. Living samples were collected within 10 min of therapeutic enucleation and four postmortem samples were collected within 6 h. After standardized dissociation, single-cell transcriptomes were generated using 10x Genomics 3’ RNA-seq and applied scVI to generate batch-corrected integrated atlas. Major cell types and subtypes were annotated through iterative Leiden clustering, canonical markers. Subsequent analyses included differential expression comparisons between cell states and regulon activity profiling to further characterize cellular identities and regulatory networks. Transcriptional dynamics were assessed using RNA velocity, and cell-cell signaling pathways were inferred with CellChat. Key findings were validated in independent samples from two additional donors (four samples) using the identical workflow. We contribute to establishing a reference for retinal cell type proportions and cellular states. Our analysis revealed ELF1-mlCone, a distinct cluster of mlCone photoreceptors identified by distinct transcriptional features. The presence and transcriptional features of this cluster were validated in independent samples. Additionally, by comparing living and post-mortem samples, our study highlights differences in transcriptional dynamics: living tissue preserved coherent RNA velocity streams, enabling clear dynamic state transitions, while post-mortem tissue exhibited disorganized patterns. These findings suggest that using living tissue can improve the capture of active cellular states and transitions. Our atlas provides a single-cell reference contrasting living versus early postmortem human retina, integrating cell type composition, transcriptional diversity, and functional insights. It may serve as a useful resource for retinal research and for understanding aspects of human retinal biology, particularly given its inclusion of living tissue and diverse pathological states.

Diabetic retinopathy (DR) remains a principal cause of vision impairment worldwide, involved complex retinal cellular pathophysiology that remains incompletely understood. To elucidate cell-type-specific molecular signatures underlying DR, we generated a high-resolution single-cell transcriptomic atlas of 297,121 retinal cells from 20 Chinese donors, including non-diabetic controls (26.4%), diabetic without retinopathy (23.4%) and DR (50.2%). Following rigorous quality control, batch-effect correction, and clustering and annotation, 10 major retinal cell populations were delineated. Differential expression analyses across disease states within each cell type yielded candidate gene sets, which were further refined via a multi-stage machine-learning pipeline combining L1-regularized logistic regression and recursive feature elimination with cross-validation, alongside bootstrap stability selection. Resulting cell-type-specific classifiers achieved high accuracy (79–95%) and AUCs (0.85–0.99) in distinguishing DR disease states. Enrichment analyses implicated immune activation, oxidative stress, neurodegeneration and synaptic dysfunction pathways across multiple cell types in retina. Integrating 567 unique marker genes from all cell types, a general multilayer perceptron classifier achieved 95.31% overall accuracy on held-out test data, demonstrating the translational potential of these signatures for non-diabetic controls, diabetic without retinopathy and DR classification. This high-resolution atlas and the accompanying analytic framework provide a robust computational framework for biomarker discovery, mechanistic insight and targeted intervention strategies in diabetic retinal diseases.

The human retina exhibits complex cellular heterogeneity which is critical for visual function, yet comprehensive ethnic-specific references are scarce in ophthalmic transcriptomics. The lack of single-cell RNA sequencing (scRNA-seq) data from Asian populations particularly Chinese donors imposes significant limitations in understanding population-specific retinal biology. We constructed the first comprehensive single-cell transcriptomic atlas of the human retina from Chinese donors, generated through high-throughput scRNA-seq of ∼290,000 viable cells obtained from 18 fresh retinal specimens (living donor and post-mortem specimens). Our multi-level analysis identified 10 distinct retinal cell types, encompassing all major neuronal lineages, Müller glia, astrocytes, microglia. Detailed subcluster analyses further revealed extensive heterogeneity, identifying distinct subtypes within several cell populations such as 7 amacrine cell subtypes and 14 bipolar cell subtypes. Concurrently, through systematic analysis of delineated subtype-specific molecular programs, we mapped their associated biological signaling pathways, functions, and mechanistic processes. This analysis explained the critical involvement of these subpopulations in core biological processes including synaptic organization, neurotransmission, and phototransduction cascades, potentially governing retinal homeostatic regulation and disease mechanisms. Single-cell transcriptomic atlas of the human retina from Chinese donors describes a comprehensive cellular landscape, encompassing major cell types and subtypes including neuronal, glial, and immune populations. This ethnic-specific atlas provided an important reference for understanding retinal development, cellular interaction and disease pathogenesis in Chinese populations, addressing a longstanding gap in ophthalmic transcriptomic resources.

Diabetic retinopathy (DR) is a major cause of vision loss worldwide. Here, we conduct single-cell RNA sequencing of twenty human retina samples (from living and post-mortem donors) across non-diabetic, diabetic, and DR states to create a comprehensive transcriptomic atlas. We identify two stable microglial populations—homeostatic and inflammatory—that exist along a functional continuum, plus a neutrophil cluster within C1QA+ myeloid cells with dynamic transitions occurring throughout disease progression. Module-level analysis reveals divergent transcriptional trajectories: homeostatic microglia maintain energetic programs while selectively upregulating stress elements, whereas inflammatory microglia layer additional pro-inflammatory programs onto preserved biosynthetic foundations. Eleven co-expression modules organize into two major axes: an inflammatory-stress axis, and a regulatory/metabolic-motility axis, with a stable translation module persisting across disease stages. Cell communication analysis further highlights sophisticated neural-immune interactions, particularly between photoreceptors and microglia. Our findings provide insights into the complex cellular dynamics of DR progression and suggest potential therapeutic targets for early intervention.

Background: Current human retina studies predominantly utilize post-mortem tissue, and the sample accessibility constraints make the characterization of the living human retina at single-cell resolution a challenge. Although single-nucleus RNA-seq expands the utility of frozen samples, it provides a nuclear-centric view, potentially missing key cytoplasmic information and transient biological processes. Thus, it is important to generate resources directly from living human retinal tissue to complement existing datasets. Methods: We profiled 106,829 single cells from nine unfrozen human retina samples. Living samples were collected withinLJ10LJmin of therapeutic enucleation and four postmortem samples were collected within 6LJh. After standardized dissociation, single-cell transcriptomes were generated using 10x Genomics 3’ RNA-seq and applied scVI to generate batch-corrected integrated atlas. Major cell types and subtypes were annotated through iterative Leiden clustering, canonical markers. Subsequent analyses included differential expression comparisons between cell states and regulon activity profiling to further characterize cellular identities and regulatory networks. Transcriptional dynamics were assessed using RNA velocity, and cell-cell signaling pathways were inferred with CellChat. Key findings were validated in independent samples from two additional donors (four samples) using the identical workflow. Results: We contribute to establishing a reference for retinal cell type proportions and cellular states. Our analysis revealed ELF1-mlCone, a distinct cluster of mlCone photoreceptors identified by distinct transcriptional features. The presence and transcriptional features of this cluster were validated in independent samples. Additionally, by comparing living and post-mortem samples, our study highlights differences in transcriptional dynamics: living tissue preserved coherent RNA velocity streams, enabling clear dynamic state transitions, while post-mortem tissue exhibited disorganized patterns. These findings suggest that using living tissue can improve the capture of active cellular states and transitions. Conclusions: Our atlas provides a single-cell reference contrasting living versus early post-mortem human retina, integrating cell type composition, transcriptional diversity, and functional insights. It may serve as a useful resource for retinal research and for understanding aspects of human retinal biology, particularly given its inclusion of living tissue and diverse pathological states.

We obtain a computable a posteriori error bound on the broken energy norm of the error in the Fortin–Soulie finite element approximation of a linear second order elliptic problem with variable permeability. This bound is shown to be efficient in the sense that it also provides a lower bound for the broken energy norm of the error up to a constant and higher order data oscillation terms. The estimator is completely free of unknown constants and provides a guaranteed numerical bound on the error.