Single-nuclei RNA-seq provides comprehensive transcriptomic classification of human retinal cell types.

Abstract

Purpose : The human retina is composed of many different neuronal and non-neuronal cell types, with their fraction in the tissue varying dramatically, ranging from 75% to less than 0.5%. Significant cell heterogeneity is observed within many retinal cell types. However, the number of cell subtypes and their molecular signature remain unknown. Our study aims at generating the first version of human retinal cell atlas reference by characterizing the transcriptome and open chromatin profile for all cell types in the human retina.

Methods : Single-nuclei RNA-seq and single-nuclei ATAC-seq are carried out to profile healthy human retina from six individual donors using the 10x Genomics technologies. Each donor retina is dissected into three areas: the fovea, macula, and peripheral retina. A fractionation protocol was developed to enrich nuclei from rare neuron types, such as bipolar cells, amacrine cells, and retinal ganglion cells. Integrative data analysis is performed to identify cell subtypes, marker genes, chromatin signature, and transcription factors and modules. RNA in situ hybridization is performed to validate novel cell populations.

Results : A transcriptome profile is generated for over 300K nuclei, leading to the identification of over 70 cell types in the retina. Notably, the numbers of bipolar cell clusters (13) and amacrine cell clusters (39) both exceeded the previously reported primate and human studies. Through integration of the single-nuclei RNA-seq and single-nuclei ATAC-seq data, key transcription factors and transcription modules were identified at both major- and sub- cell-type level. Moreover, three-way comparison among mouse, monkey and human retina single-cell transcriptomic data revealed conserved and lineage specific cell types during evolution. Finally, it has been observed that genes associated with different human retina diseases show distinct cell type specific gene expression profiles, providing insight to potential disease mechanisms at cell subtype resolution.

Conclusions : The study represents the most comprehensive single-cell transcriptome and single-cell chromatin accessibility profiling for the human retina to date. Over 300K single nuclei are profiled and over 70 types of cells are identified, making it a high-quality dataset that can serve as the first version of a human retina cell atlas reference.

This is a 2020 ARVO Annual Meeting abstract.