Conditioned rats acquired passive avoidance behavior
shows the step-down latency of rats during passive avoidance training (Day 1 and Day 2) and retrieval (Day 3). Over the two-day training, conditioned rats learned to stay on the platform to avoid the 60-mmHg CRD, showing greater step-down latencies than controls (Day 1, F(1,20) = 1.6, P = 0.22; Day 2, F(1,20) = 6.8, P = 0.017, mixed model ANOVA). The learned PA behavior was retrieved on Day 3 in the absence of the colorectal balloon. Conditioned rats showed significantly greater step-down latencies on the first trial (37 ± 6 s in conditioned rats vs. 15 ± 4 s in controls, P = 0.01, Student's t-test), as well as on the second trial (36 ± 4 s in conditioned rats vs. 21 ± 5 s in controls, P = 0.03, Wilcoxon rank test). The presence of a balloon during training combined with repeated exposure to the platform may have caused habituation. This may account for the increase in step-down latency in the control rats during training, as well as the drop in step-down latency in both groups on Day 3 (without the balloon) as compared to trial 1, Day 2 (with the balloon). Step-down latency on Day 3 was likely a more accurate measurement of contextual recall of PA, without the nonspecific effect of an inserted balloon.
Acquisition and retrieval of step-down passive avoidance
Functional brain activation during the retrieval of passive avoidance
Brain areas showing significant differences in rCBF between the conditioned and the control group are depicted in and summarized in . Compared to controls, conditioned animals showed increased rCBF bilaterally in a wide range of areas including medial PFC subregions (ventral cingulate, Cg2; right dorsal cingulate, Cg1; retrosplenial, RS, equivalent to posterior cingulate in primates; PrL), aINS, nucleus accumbens (NAcc), amygdala (including basolateral amygdala, BLA; amygdalopiriform transition area, APir; cortical amygdaloid nuclei, PLCo, PMCo), and dorsomedial periaqueductal gray(DMPAG). In addition, increased rCBF was noted bilaterally in primary motor cortex (M1), primary and secondary somatosensory cortices (S1, S2), as well as the anterior dorsal caudate putamen (adCPu), lateral caudate putamen (lCPu), dorsal hippocampus (dHPC), lateral septum (LS), and the cerebellum (vermis, CbVermis; hemisphere, CbHemis). White matter tracts, including the forceps minor of the corpus callosum (fmi) and external capsule (ec) bilaterally, and left anterior commissure (AC) also showed increased rCBF in conditioned rats. Significant decreases in rCBF were noted in the conditioned rats in the inferior colliculus (IC) and pontine nuclei (Pn) bilaterally, and in the entorhinal cortex (Ent) in the right hemisphere.
Key brain regions showing significant differences in regional cerebral blood flow between the conditioned and the control rats
Brain regions showing significant changes in regional cerebral blood flow in conditioned rats compared to controls
Shaded cells in depict regions that also showed changes in rCBF in response to acute CRD as we previously reported [68
]. Regions showing activation in both studies included the Cg1, PrL, aINS, S1, S2, M1, and adCPu, with differences noted in the extent of activation in these regions. Important differences in regional brain activation were also noted between these studies. In response to acute CRD, but not retrieval of CRD-conditioned PA, broader cortical areas, including the auditory and visual areas, as well as the central and lateral nuclei of the amygdala showed activation. In contrast, the BLA, dHPC, NAcc, DMPAG, and cerebellum showed activation only in the current study.
Functional connectivity of brain networks in the control rats
An inter-regional correlation matrix of rCBF was constructed for the control group and visualized as a heatmap in . Significant correlations (P < 0.05) were interpreted as functional connections and marked with white dots. The matrix is symmetric across the diagonal line from upper left to lower right, which itself reflects the trivial correlation of ROIs to themselves. There were 47 significant positive correlations and 15 significant negative correlations. A cluster of midline cortical ROIs, including bilateral Cg2, RS, PrL/PFC, and right Cg1, were strongly and positively correlated with each other. Within this cortical cluster, all except PrL/PFC were significantly positively correlated with striatal ROIs (including bilateral NAcc and adCPu), but negatively correlated with the amygdala. Intra-structural positive correlations were noted within the striatum, and the cerebellum. In addition, PrL/PFC, Cg2, RS, S1, NAcc, adCPu, dHPC, amygdala and the cerebellar hemispheres showed strong, positive cross-hemisphere correlations.
Functional connectivity analysis of network activation during retrieval of passive avoidance
Graph theoretical analysis revealed organization of the functional network and hubs (, ). The cortical cluster (, red vertices) is clearly shown in the center of the network, with its negative connections (dashed lines) to the amygdala and the cerebellum (, blue vertices). ROIs in the above-mentioned cortical cluster (Cg2, right Cg1, RS, PrL) were identified as hubs together with bilateral adCPu. In addition, CbHemis, dHPC, and LS were identified as hubs by their high betweenness centrality.
Hubs of the functional brain network
Functional connectivity of brain networks in the conditioned rats
shows the inter-regional correlation matrix for the conditioned group. There were fewer connections in the conditioned group than in the controls, 25 positive and 16 negative significant correlations. Certain similarities in the functional connectivity pattern between the two groups were noted, including the intra-structural positive correlations in the cortex and the striatum, as well as positive correlations between the cortex and striatum, and negative correlations between the cortex and amygdala. Meanwhile, there were important group differences (). In the cortex, M1 and PrL/PFC showed more connectivity with other cortical areas, whereas RS showed less connectivity in the conditioned group. Strong positive connections were seen between the amygdala and cerebellar hemispheres. The amygdala was negatively connected to PrL/PFC in the conditioned group, whereas in the controls the amygdala was negatively connected to RS. In addition, M1, S1, PrL/PFC, Cg2, RS, aINS, adCPu, dHPC, amygdala and the cerebellar hemisphere showed positive cross-hemisphere correlation.
Changes in interregional correlation in the conditioned group compared to the controls
Graph analysis revealed a cortical cluster, with RS seemingly removed from the core (, red vertices). The amygdala and cerebellum in conditioned animals formed a separate cluster, and were negatively connected to the cortical cluster (PrL/PFC, Cg1, Cg2). The NAcc, PrL/PFC, and aINS were shown to be crucial to the network structure, with the highest betweenness centrality. In addition, cingulate cortex (Cg1, Cg2), adCPu, and amygdala were also identified as network hubs by graph analysis ().