Multiscale and multimodal reconstruction of cortical structure and function

Abstract

We present a semi-automated reconstruction of L2/3 mouse primary visual cortex from 3 million cubic microns of electron microscopic images, including pyramidal and inhibitory neurons, astrocytes, microglia, oligodendrocytes and precursors, pericytes, vasculature, mitochondria, and synapses. Visual responses of a subset of pyramidal cells are included. The data are being made publicly available, along with tools for programmatic and 3D interactive access. The density of synaptic inputs onto inhibitory neurons varies across cell classes and compartments. We uncover a compartment-specific correlation between mitochondrial coverage and synapse density. Frequencies of connectivity motifs in the graph of pyramidal cells are predicted quite accurately from node degrees using the configuration model of random graphs. Cells receiving more connections from nearby cells exhibit stronger and more reliable visual responses. These example findings illustrate the resource's utility for relating structure and function of cortical circuits as well as for neuronal cell biology.

DotMotif: An open-source tool for connectome subgraph isomorphism search and graph queries

“…We demonstrate the use of our tool on both near-complete (most or all of the organism's neurons are included in the graph) connectomes as well as on partial connectomes. Here, we compare connectomes of the invertebrate nematode C. elegans [23,20], the partial connectome of the invertebrate fruit fly from the Howard Hughes Medical Institute Janelia Hemibrain [2] project, and the partial vertebrate mouse visual cortex connectome from the Intelligence Advanced Research Project Activity (IARPA) Machine Intelligence from Cortical Networks (MICrONS) project [24,25,10]. Further dataset details are available in Table 1.…”

“…These simple circuits are hypothesized to be repeated and reused many times to perform a similar role. Though they might comprise only a small number of neurons, as shown in Figure 1, these simple circuits or graph motifs may be critical to understanding the functional role of larger structures in the brain, such as cortical columns or other modules of computational importance [8,9,10]. Due to the nature of available high-resolution imaging modalities such as electron microscopy, it is not always feasible to collect functional as well as anatomical data at synaptic resolution.…”

“…The C. elegans connectome was derived from the July 2020 C. elegans adjacency matrix at WormWiring.org, and synapses were filtered to only include chemical synapses for this analysis [20,21]. The MICrONS v185 graph was derived from the proofread soma subgraph at microns-explorer.org [10,22]. The Hemibrain v1.2 dataset was accessed through the neuPrint system [6].…”

“…Neuroscience questions may be reformulated as analyses on a graph, and the neuroscientist may add graph theory to the toolbox of strategies with which to understand the brain [8,11]. Such questions include searching for specific subgraph structures, investigating the connectivity of specific neuron cell types or categories, proofreading connectomes for accuracy, and generating summary statistics on the graph as a whole [12,13,3,14,15,16,10]. Many graphs generated by the connectomics community in recent years have spanned multiple gigabytes of hard drive space [2,10], rendering conventional graph toolkits, such as the common NetworkX library [17] (or its counterparts in other programming languages) under-powered to address the needs of the scientific community.…”

“…Such questions include searching for specific subgraph structures, investigating the connectivity of specific neuron cell types or categories, proofreading connectomes for accuracy, and generating summary statistics on the graph as a whole [12,13,3,14,15,16,10]. Many graphs generated by the connectomics community in recent years have spanned multiple gigabytes of hard drive space [2,10], rendering conventional graph toolkits, such as the common NetworkX library [17] (or its counterparts in other programming languages) under-powered to address the needs of the scientific community. These tools, which often require all graph data to be stored in RAM, would require impractically expensive compute hardware in order to run fully in-memory, and would require impractically long timelines in order to run while swapping data from memory to disk.…”