Pipeline schematic.

(A) The ArcCreERT2 x eYFP tagging strategy allows for labelling of Arc-expressing cells with eYFP following injection of 4-OHT. (B) Indelible labelling of Arc+ ensembles with eYFP followed by immunolabelling for c-Fos+ allows for identification of a co-labeled ensembles active during two distinct timepoints. (C) A graphical summary of the components of the workflow. (D) eYFP+ and c-Fos+ cells are immunolabelled and imaged across brain-wide coronal sections. (E) eYFP+ and c-Fos+ populations are automatically segmented, and co-labeled cells are identified in ImageJ/Fiji. Images are automatically pre-processed and flattened for registration alignment downstream. (F) The object-oriented infrastructure of SMARTTR package in R allows for importation of segmentation data, registration, and mapping, and statistical analysis and visualization using a user-friendly API. Representative schematics were created using BioRender.com.

Labelling and automatic identification of ensembles active during the acquisition and expression of learned helplessness.

(A) Experimental design. (B) Average habituation activity (IR beam breaks) decreases in the shock group. (C) Prior administration of inescapable shocks increases subsequent escape latency across 30 trials of shocks. (D) Average escape latency across trials 11-30 is higher in the inescapable shock (IS) group (n=6) compared to the context training (CT) group (n=5). (E) Representative hippocampal image showing eYFP+ cells (F) and c-Fos+ cells (G), and their overlap (H) in the dDG. (I) The consecutive image processing steps optimized for auto-segmentation of eYFP+ cells. (J) The consecutive image processing steps optimized of auto-segmentation of c-Fos+ cell. (K) Using the 3D MultiColoc plugin, all possible overlaps are calculated between the segmented eYFP+ and c-Fos+ objects in ImageJ (middle left). Results are exported and later thresholded by percent volume overlap relative to segmented c-Fos+ object (middle right) to identify co-labeled cells (left). ***p < .001. Error bars represent ± SEM. Figures of behavioral timelines were created using BioRender.com.

Network level analysis reveals enhanced sensory and affective processing during learned helplessness acquisition.

(A-B) Regional cross correlation heatmaps of eYFP+ volume normalized cell counts in context trained (CT) and inescapable shock (IS) mice. Significant values are p < 0.01. (C) Functional networks constructed after thresholding for the strongest and most significantly correlated or anti-correlated connections (r > 0.9, p < 0.01). (D) Average degree centrality does not differ between IS and CT groups. (E) Degree frequency distributions are right-tailed. (F-H) Mean clustering coefficient, global efficiency, and mean betweenness centrality do not differ between the CT and IS networks. (I) The top node degree values in descending order for the CT (white, top) and IS (green, bottom) networks indicate which regions are most highly connected. (J) The top node betweenness values in descending order for the CT (white, top) and IS (green, bottom) networks indicate which regions are most influential in directing “information flow.” (K) Volcano plot of the Pearson correlation differences (rIS – rCT) for all individual regional connections against their p-values calculated from a permutation analysis. Points intersecting or within the upper left or right quadrant represent the regional relationships with the greatest change (|correlation difference| > 1), that were most significant (p < 0.01). (L) A parallel coordinate plot highlighting individual significantly changed regional correlations between groups, as well as the direction of their change. Error bars represent ± SEM.

Network-level analysis reveals influential altered functional connectivity of the substantia nigra during learned helplessness expression.

(A-B) Regional cross correlation heatmaps of c-Fos+ expression in context trained (CT) and inescapable shock (IS) groups. Significant values are p < 0.005. (C) Functional networks constructed after thresholding for the strongest and most significantly correlated or anti-correlated connections (r > 0.9, p < 0.005). (D) Highest individual node degree values in descending order for the CT (white, top) and IS (red, bottom) networks indicate which regions are most highly connected. (E) Node betweenness values in descending for the CT (white, top) and IS (red, bottom) networks. (F-I) Pearson correlation distributions of the dorsal CA1 (dCA1), ventral medial nucleus of the thalamus (VM), subthalamic nucleus (STN), and substantia nigra, reticular part (SNr). Distributions between CT and IS groups significantly differ in the SNr. (J-K) Volcano plot and parallel coordinate plots highlighting the permuted correlation differences (rIS – rCT) that show the greatest change (|correlation difference| > 1), and are most significant (p < 0.01) between the CT and IS groups. ****p < .0001. Error bars represent ± SEM.

Network analysis of reactivated inescapable shock ensembles during learned helplessness reveals altered functional connectivity.

(A-E) Representative images of regions identified for targeted analysis, including isocortical regions (anterior cingulate area, ACA; agranular insula, AI; retrosplenial area, RSP), dorsal hippocampal regions (dorsal dentate gyrus, dDG; dorsal CA3, dCA3; and dorsal CA1, dCA1), ventral hippocampal regions (ventral dentate gyrus, vDG; ventral CA3, vCA3; and ventral CA1, vCA1), and amygdalar areas (lateral amygdalar nucleus, LA; basolateral amygdalar nucleus, BLA; basomedial amygdalar nucleus, BMA; central amygdalar nucleus, CEA; medial amygdalar nucleus, MEA). Representative region overlays were manually drawn. (F-I) The ACA, AI, dDG, dCA3, and MEA show significantly decreased reactivation activity (co-labeled cells / eYFP+ cells) in the mice exposed to inescapable shock compared to context training. (J) LA reactivation activity in both context trained (CT) and inescapable shock (IS) mice shows positive correlation to escape latency. (K-L) Correlation heatmaps of reactivation activity in CT and IS mice. (M) Functional networks constructed after thresholding for the strongest and most significant correlated or anti-correlated connections (r > 0.9, p < 0.01). (N) Top individual node degree values in descending for the CT (white, top) and IS (yellow, bottom) networks. (O) Highest node betweenness values in descending for the CT (white, top) and IS (yellow, bottom) networks. (P-S) Pearson correlation distributions of the dCA1, AI, ACA, secondary motor area (MOs), and substantia nigra, reticular part (SNr). Distributions between CT and IS groups significantly differ in the AI. (T-U) Volcano plot and parallel coordinate plots highlighting the permuted correlation differences (rIS – rCT) of functional connections of reactivated activity showing the greatest change (|correlation difference| > 1), and are most significant (p < 0.01) between the CT and IS groups. **p < .01, ****p < .0001. Error bars represent ± SEM.