Identifying avalanches in simulated mouse primary motor cortex (M1)

Connectivity among the neurons participating in avalanches recorded in vitro and in vivo has been unknown. We set out to observe avalanches in a realistic simulation of a cortical column from the mouse primary motor cortex (M1) and to investigate mechanisms underlying the generation of avalanches. We applied sustained 0.23nA current to a 40um x 1350um x 40um volume of a 300um x 1350um x 40um simulated cortical column. About 2.2s after stimulus onset, self-sustained asynchronous irregular (AI) activity appeared. We binned spikes into 1ms bins during AI activity and found no gaps in spiking when all of neurons were included. We decided to look for avalanches in populations by cell type. Using the same stimulus protocol, Pyramidal Tract neurons in layer 5B (PT5B) were responsive only after the onset of AI activity across the simulated cortical column. PT5B activity in raster plots resembled avalanches so we binned PT5B responses into 1ms bins and computed avalanches as continuous activity every millisecond from one neuron to next until activity stoped for 1ms or more. The PT5B population power law value was -1.97 from 7.2s of AI activity. The total number of avalanches observed was 1,122 and the longest duration was 50ms.We applied 0.22nA for 5s to the simulated cortical column in the same manner as above to see if PT5B avalanches remained if the stimulus was turned off. AI activity appeared at 8s or 3s after the stimulus was turned off. PT5B responses were placed into 1ms bins from a 2s period resulting in 249 avalanches, 32ms longest duration avalanche, and power law value of -2.22. The same experiment with 180s runtime resulted in 172s of AI activity, 27,938 avalanches, 50ms longest duration avalanche, and power law value of -2.74. Avalanche number 4 shown in the accompanying diagram has a duration of 6ms (from 8,514ms to 8,519ms) and size of 15 unique neurons. The red lines in the diagram show direct connections between the PT5B neurons. Our ability to record from every neuron in our simulated M1 enables us to investigate connectivity underlying avalanche activity in normal and diseased brain.