Motor cortex microcircuit simulation based on brain activity mapping

Abstract

The deceptively simple laminar structure of neocortex belies the complexity of intra- and interlaminar connectivity. We developed a computational model based primarily on a unified set of brain activity mapping studies of mouse M1. The simulation consisted of 775 spiking neurons of 10 cell types with detailed population-to-population connectivity. Static analysis of connectivity with graph-theoretic tools revealed that the corticostriatal population showed strong centrality, suggesting that would provide a network hub. Subsequent dynamical analysis confirmed this observation, in addition to revealing network dynamics that cannot be readily predicted through analysis of the wiring diagram alone. Activation thresholds depended on the stimulated layer. Low stimulation produced transient activation, while stronger activation produced sustained oscillations where the threshold for sustained responses varied by layer: 13% in layer 2/3, 54% in layer 5A, 25% in layer 5B, and 17% in layer 6. The frequency and phase of the resulting oscillation also depended on stimulation layer. By demonstrating the effectiveness of combined static and dynamic analysis, our results show how static brain maps can be related to the results of brain activity mapping.

Figure 1

(A) Connection probability (p_ij) between cell types ranged from 0 to 80% (gray level). (B) Average AMPA (for E cells), GABA (for I cells) weights between cell types. For E2 → STR, E2 → SPI, E2 → I5L connections, weights were set in a depth-location-dependent manner.

Figure 2

Betweenness centrality view of network. Node size is proportional to the population betweenness centrality; edge thickness is proportional to the edge betweenness centrality. Self-connections are not shown.

Figure 3

Connectivity strength view of network: edge thickness is proportional to the normalized strength (product of weight and convergence normalized by population size). Node size is proportional to population count. Self-connections are shown as small looping arrows.

Figure 4

A single stimulation produces either transient activation (left column) or sustained activation (right column), with different patterns depending on the layer of stimulation. Stimulation was given at time 0. Transient activity is produced by subthreshold activation: 12% for L2/3 (A), 53% for L5A (C), 24% for L5B (E), and 16% for L6 (G). Sustained responses are generated at threshold: 13% for L2/3 (B), 53% for L5A (D), 25% for L5B (F), and 17% for L6 (H). Results are representative of 4000 simulations.

Figure 5

Minimum stimulation threshold for sustained oscillations by stimulation layer. Average across 25 trials. Boxes encompass lower and upper quartiles, with lines indicating the median. Whiskers show maximum 1.5× interquartile range; outliers are indicated by ticks.

Figure 6

Oscillations of sustained responses continue indefinitely. (A–D) Same activations as shown in Figure 4 shown extended in time. (E) Oscillations differ in phase depending on site of stimulation. Gaussian smoothed local population response in output cells (SPI and STR populations), computed from 40 ms to avoid transients.

Figure 7

Spiking activity of STR and I5 only network demonstrated a pyramidal-interneuron network gamma (PING) mechanism. Sixty percent of neurons were brought to threshold with STR↔I5 connections either left intact (A) and removed (B). Activity from 250 ms to 500 ms after stimulus is shown in order to remove any stimulus artifact and allow the network to reach steady-state dynamics. The inset shows the firing activity of A between 250 ms and 310 ms.

Figure 8

Rasters showing network activity with 54% L5A activation where I5→STR connection is left intact (A; control), strength is set to half the typical value (B) or the connection is cut (C), where E2→I5L connection alone is cut (D) or it is cut in addition to halving the strength of I5→STR (E) and where a SPI→STR connection is introduced (F).