Behavior depends on coordinated activity across multiple brain regions. Within such networks, highly connected hub regions are assumed to disproportionately influence behavioral output, although this hypothesis has not been systematically evaluated. Previously, by mapping brain-wide expression of the activity-regulated gene c-fos, we identified a network of brain regions co-activated by fear memory. To test the hypothesis that hub regions are more important for network function, here, we simulated node deletion in silico in this behaviorally defined functional network. Removal of high degree nodes produced the greatest network disruption (e.g., reduction in global efficiency). To test these predictions in vivo, we examined the impact of post-training chemogenetic silencing of different network nodes on fear memory consolidation. In a series of independent experiments encompassing 25% of network nodes (i.e., 21/84 brain regions), we found that node degree accurately predicted observed deficits in memory consolidation, with silencing of highly connected hubs producing the largest impairments.
Figure_1A
All pair-wise correlations of Fos counts for the fear memory.
Figure_1B
Weighted adjacency matrix for the fear memory network.
Figure_1C
All pair-wise correlations of Fos counts from control animals.
Figure_1D
Weighted adjacency matrix for the control network.
Figure_2A
Weighted adjacency matrix following application of the disruption propagation model starting from node Re.
Figure_2B
Global efficiency before and after the application of the disruption propagation model to the fear memory network targeting node Re.
Figure_2C
The change in global efficiency following application of the disruption propagation model to each individual node, and the degree of each node, in the fear memory network.
Figure_3B
The effect of CNO administration on Fos expression in across different zeitgeber times.
Figure_3C
The effect of CNO administration on Fos expression in different brain regions.
Figure_3D
The effect of prolonged CNO administration on Fos expression.
Figure_5B
Percent freezing levels for mice expressing hM4Di in 21 different brain regions and administered CNO or VEH.
Figure_6ABD
Difference in freezing between CNO and VEH treated mice, changes in global efficiency for the fear memory and control networks following application of the disruption propagation model, and relative Fos levels.
Figure_6C
Difference in freezing between CNO and VEH treated mice, and the change in global efficiency following application of the disruption propagation model to the anatomical network.
Figure_S1ABC
Node degrees and clustering coefficients for the fear memory network.
Figure_S1D
Node degrees for the control network.
Figure_S1E
Weighted adjacency matrix for the anatomical network built using the same p-value as the fear memory network (p < 0.01).
Figure_S1F
Weighted adjacency matrix for the anatomical network built to have the same density as the fear memory network.
Figure_S1G
Node degrees for the anatomical network thresholded using the same p-value as the fear memory network (p < 0.01).
Figure_S1H
Node degrees for the anatomical network thresholded to have the same density as the fear memory network.
Figure_S1IJ
Node degrees in the fear memory and control networks.
Figure_S1KL
Node degrees in the fear memory and anatomical networks.
Figure_S3AC
Shannon's entropy (H) for changes in the global efficiency and giant component size following application of the disruption propagation model to nodes of the fear memory network. The number of edge changes allowed during modeling was varied from 0 to 60.
Figure_S3BD
Shannon's entropy (H) for changes in the global efficiency and giant component size following application of the disruption propagation model to nodes of the fear memory network. The threshold applied following propagation was varied from 0 to 1.
Figure_S3E
Giant component size before and after the application of the disruption propagation model to the fear memory network targeting node Re.
Figure_S3F
The change in giant component size following application of the disruption propagation model to each individual node, and the degree of each node, in the fear memory network.
Figure_S4AB
The change in global efficiency and giant component size following either simple node deletion or the application of the disruption propagation model for each node in the fear memory network.
Figure_S4CD
Change in global efficiency and giant component size following simple node deletion, and node degrees, in the fear memory network.
Figure_S5A
Percent freezing in mice infused with a control virus and administered VEH or CNO.
Figure_S5BCDEF
Difference in freezing between CNO and VEH treated animals. Changes in the giant component size for the fear memory network following application of the disruption propagation model. Node degrees in the fear memory network. Changes in the global efficiency of the fear memory network following simple node deletion. Alpha critical values derived from the cascading failure model applied to the fear memory network. Volume of each region targeted in behavioral experiments.
Figure_S6A
Difference in freezing between CNO and VEH treated animals. Change in the global efficiency following application of the disruption propagation model to the control network with the same density as the fear memory network.
Figure_S6B
Difference in freezing between CNO and VEH treated animals. Change in the global efficiency following application of the disruption propagation model to the anatomical network with the same density as the fear memory network.
Cascading_failure_model
R implementation of the cascading failure model
Disruption_propagation_model
R implementation of the disruption propagation model
DPM.R