Digital museum of retinal ganglion cells with dense anatomy and physiology

Abstract

Most digital brain atlases have macroscopic resolution and are confined to a single imaging modality. Here we present a new kind of resource that combines dense maps of anatomy and physiology at cellular resolution. The resource encompasses 5 almost 400 ganglion cells from a single patch of mouse retina, and a digital "museum" provides a 3D interactive view of each cell's anatomy as well as graphs of its visual responses. To demonstrate the utility of the resource, we use it to divide the inner plexiform layer of the retina into four sublaminae defined 10 by a purely anatomical principle of arbor segregation. We also test the hypothesis that the aggregate neurite density of a ganglion cell type should be approximately uniform ("density conservation"). Finally, we find that ganglion cells arborizing in the inner marginal sublamina of the inner plexiform layer ex- 15hibit significantly more sustained visual responses on average.Calcium imaging followed by 3D electron microscopy (EM) has become established as a powerful approach for obtaining anatomical and physiological information about the same neurons. The approach has been used to study visual neurons in mouse cortex 20 [Bock et al., 2011, Lee et al., 2016 and retina [Briggman et al., 2011], and oculomotor neurons in larval zebrafish hindbrain [Vishwanathan et al., 2017]. These prior studies have been limited to tens of neurons sparsely sampled from a single individual. Here we present as a resource the anatomy and physiology of almost 25 400 ganglion cells (GCs) densely sampled from a single patch of mouse retina.To facilitate exploratory data analysis using the resource, we have constructed the Eyewire Museum (http://museum. eyewire.org), where every reconstructed ganglion cell can be in-30 teractively viewed along with its visual response properties. Due to its cellular and subcellular resolution, the Museum is novel relative to traditional brain atlases, which typically divide the brain into macroscopic regions [Lein et al., 2007, Amunts et al., 2013, Zingg et al., 2014. Because it fuses anatomy with physiology, 35 the Museum is also novel relative to traditional atlases of neuronal morphologies such as neuromorpho.org [Ascoli et al., 2007] and wormatlas.org [Hall et al., 2007]. Previous dense EM reconstructions in the mouse retina [Helmstaedter et al., 2013] and larval zebrafish olfactory bulb [Wanner et al., 2016] were limited to 40 anatomy only. A recent large-scale calcium imaging study contains the visual responses of more than 11,000 neurons from 50 retinas, but less than 1% of the cells had their dendritic arbors reconstructed [Baden et al., 2016].To illustrate the utility of our resource, we use it to reveal new 45 principles of retinal organization. First, we show how to optimally subdivide the inner plexiform layer (IPL) of the retina using the purely anatomical principle that arbors should segregate into distinct sublamina. For GC dendritic arbors, segregation is maximized by subdividing the IPL into two marginal sublamina flanking a...

Molecular identification of sixty-three amacrine cell types completes a mouse retinal cell atlas

“…For BCs and RGCs, we documented the existence of 15 and 46 types, respectively (Shekhar et al, 2016;Tran et al, 2019). These numbers correspond well to those obtained from recent high-throughput electrophysiological, ultrastructural and molecular studies (Baden et al, 2016;Bae et al, 2018;Franke et al, 2017;Greene et al, 2016;Rheaume et al, 2018).…”

Single-cell profiles of retinal neurons differing in resilience to injury reveal neuroprotective genes

“…Clusters were highly reproducible across biological replicates and three different computational approaches (Figure S1C-E). The number of estimated molecular types agree well with recent studies based on physiological, morphological and molecular methods (Baden et al, 2016;Bae et al, 2018;Rheaume et al, 2018) (Table S1 and Figure S1F). Finally, we compared the relative frequencies of RGC groups labeled immunohistochemically (IHC) in retinal whole mounts to their frequencies in the scRNA-seq data and found a striking correspondence (Figure 1I).…”

“…Analysis of 35,699 adult RGC transcriptomes revealed 45 cell clusters. Several lines of evidence indicate that the transcriptomically defined clusters correspond to cell types: (a) The number of types is similar to those from recent large-scale surveys based on serial section electron microscopy (≥35; (Bae et al, 2018), optical imaging of electrical activity (≥32; (Baden et al, 2016) and scRNAseq of neonatal retina (40; (Rheaume et al, 2018)). (b) Several clusters could be assigned to types based on previously known markers.…”

Cell types of the human retina and its organoids at single-cell resolution: developmental convergence, transcriptomic identity, and disease map

“…Primates are the only mammals with a fovea. Retinal cells in both periphery and fovea can be divided into morphologically (Bae et al, 2018), functionally (Baden et al, 2016;Dacey et al, 2003;Roska and Werblin, 2001) and transcriptomically (Macosko et al, 2015;Peng et al, 2019;Shekhar et al, 2016;Siegert et al, 2012) different cell classes that are further divisible into cell types. The neural retina contains five layers (Dowling, 2012).…”