DBP5: Functional Connectomics: Terascale Reconstructions of Cortical Circuits
PI: Clay Reid, MD, PhD, Professor of Neurobiology, Allen Brain Institute
Funding Source(s): 5R01NS075436-05, Large-scale connectivity and function in a cortical circuit; 6/15/11-4/30/16 (Reid)
How the DBP acts as a driver of test bed:
This project has driven the development of our image alignment software to register and analyze image stacks of 10 to 100 terabytes comprising millions of raw images, and successfully resulted in the publication of the largest-to- date (to our knowledge) network of cortical neurons in Nature 2016.
This work builds on two existing projects with similar goals: to understand the relationship between the function of a cortical circuit and the fine-scale connections within the circuit. A fundamental but unsolved question in neuroscience is how specific connections between neurons underlie information processing in the brain. Even the smallest local circuit in the cerebral cortex consists of tens of thousands of neurons, each making thousands of connections. Perhaps the biggest reason we have yet an incomplete understanding of the cerebral cortex, or any other complex neural system, is that we don't have an actual wiring diagram of any cortical circuit. But even if we had a wiring diagram, we would need to know what each neuron in a circuit is doing: its physiology. For the past decade, we have been studying neurons in the visual cortex whose responses to sensory stimuli have been characterized with two-photon calcium imaging. This approach allows us to image the function of literally every neuron in a cube ½ millimeter on a side. In collaboration with the PSC, we then use high-resolution and high-throughput anatomical techniques to uncover the connections between these functionally characterized neurons.
We have been addressing these questions in the mouse visual cortex, an emerging model of visual processing that is amenable to genetic manipulation and in vivo imaging techniques. By combining functional and structural imaging (serial-section electron microscopy; ssEM) at increasingly large scales, we are collecting datasets that provide, for the first time, a complete physiological and structural overview of this important circuit in the cerebral cortex. We have recently demonstrated that this combined approach is both feasible and powerful enough to address outstanding problems in cortical neuroscience. It is now possible to study cortical circuits on their own terms: in all of their complexity and with datasets that are in many senses complete.
We have these specific aims:
To reconstruct a recurrent network of functionally characterized layer 2/3 neurons.
Recent studies have found highly interconnected subnetworks of cells embedded in local excitatory circuits in visual cortex. We hypothesize that these subnetworks are related to the functional organization of visual responses, such that layer 2/3 neurons with similar functional properties are selectively connected to each other. This project requires a dataset three times larger than that reconstructed in our previous work, and presents new computational and data-handling challenges.
To reconstruct a feed-forward and recurrent multi-layer network of cortical neurons.
To reconstruct a multi-layered network in the cortex, a dataset 5 to 20 times larger than the previous (hundreds of terabytes of raw data) will require an essentially new way to view and handle the data—VOV architecture described in Specific Aim 2 of TR&D3 — and thus will further engage the unique expertise of the MMBioS team.
Publications resulting from this work:
- Lee, W. A.; Bonin, V.; Reed, M.; Graham, B. J.; Hood, G.; Glattfelder, K.; Reid, R. C (2016) Anatomy and function of an excitatory network in the visual cortex. Nature 532: 370-374 PMID: 27018655, PMC4844839