Principles of visual cortex excitatory microcircuit organization

Synapse-specific connectivity and dynamics determine microcircuit function but are challenging to explore with classic paired recordings due to their low throughput. We therefore implemented optomapping, a ∼100-fold faster two-photon optogenetic method. In mouse primary visual cortex (V1), we optomapped 30,454 candidate inputs to reveal 1,790 excitatory inputs to pyramidal, basket, and Martinotti cells. Across these cell types, log-normal distribution of synaptic efficacies emerged as a principle. For pyramidal cells, optomapping reproduced the canonical circuit but unexpectedly uncovered that the excitation of basket cells concentrated to layer 5 and that of Martinotti cells dominated in layer 2/3. The excitation of basket cells was stronger and reached farther than the excitation of pyramidal cells, which may promote stability. Short-term plasticity surprisingly depended on cortical layer in addition to target cell. Finally, optomapping revealed an overrepresentation of shared inputs for interconnected layer-6 pyramidal cells. Thus, by resolving the throughput problem, optomapping uncovered hitherto unappreciated principles of V1 structure.

Methods
We optogenetically stimulated indivudal excitatory neurons by spiral-scanning a 1040-nm femtosecond laser beam over cell expressing soma-targetted ChroME opsin while simultaneously recording a postsynaptic neuron in whole-cell configuration. This data was acquired using custom software called "jScan" (https://github.com/pj-sjostrom/jScan) in combination with MultiPatch (https://github.com/pj-sjostrom/MultiPatch ), running in Igor Pro (Wavemetrics Inc).

In offline analysis using custom software called "CMap" (see jScan github link above) running in Igor Pro, time-locked excitatory postsynaptic potentials (EPSPs) consistently evoked across 20 repetitions were taken to denote the existence of a synaptic connection. Initial EPSP magnitude due to a 30-Hz train of three laser pulses was used as a metric of synaptic strength. Short-term plasticity was measured from the EPSP train. The electrophysiology traces were processed with CMap so that EPSP amplitude etc was mapped onto the location of optogenetically stimulated presynaptic neurons. This process was repeated for several nearby fields-of-view (FOV), to create an overall 'optomap'. The data pertaining to each optomap (e.g., connectivity, EPSP amplitude, short-term plasticity...) was exported for each postsynaptic cell into individual data files and collected into folders for the same cell type (e.g., "L23 PCs", "L23 BCs", "L23 MCs", etc across L2/3, L5, and L6).

In subsequent offline analysis using custom software called "CMetaMap" (see jScan github link above) running in Igor Pro, these folders with extacted optomap data were compiled across the same cell types (again, "L23 PC", "L23 BC", "L23 MC", etc across L2/3, L5, and L6). For example, to view the data that went into Figure 3 of Chou et al., please use the most recent version of CMetaMap to sequentialy load the data from the folders "L23 PC", "L23 BC", and "L23 MC".

In addition, we provide Excel sheets for each figure in the paper. These Excel sheets contain the raw data points that went into the figure. Also see the Notes provided in each Excel sheet for additional information on the data structure.