Between similarity and difference: network dynamics of the hippocampal- parahippocampal circuitry in pattern separation of male Wistar rats

Abstract

Introduction: Studies indicate that pattern separation for spatial and object information involves structures of the temporal cortex (lateral entorhinal and perirhinal cortices) and hippocampus (dentate gyrus and CA3), which are particularly sensitive to aging. However, little is known about how the hippocampal network, the anteroposterior axis of these regions, and the excitatory-inhibitory circuit contribute to the recognition and separation of object patterns.

Methods: This study investigated the expression of c-Fos and PV along the anteroposterior axis of the hippocampus in a multi-trial task to assess the recognition of novel objects and recognition of novel objects with different levels of similarity. Five groups of animals performed tasks with different similarity demands (NOR, DIST, 25, 50, 75%).

Results: The data showed that conditions of greater similarity led to increased c-Fos expression in CA3c and Hilus in the rostral hippocampus. Graph analysis revealed that hippocampal networks became more densely interconnected and efficient as object similarity increased. Furthermore, different patterns of cluster organization emerged depending on task demands. Besides, the granule cell layer along the dorsoventral axis exhibited greater activation of inhibitory neurons (PV+/c-Fos+) under conditions of higher similarity. Differential inhibitory/excitatory control of the DG-CA3 microcircuit network is seen across conditions. Modeling the DG layers revealed robust control of GCs through direct and indirect effects of interneurons present in the hilus and granule layer. Bidirectional direct and indirect effects of MCs on GCs were observed.

Discussion: These results contribute to our understanding of how brain networks and DG excitatory/inhibitory microcircuits are jointly engaged in object recognition memory and disambiguation of overlapping inputs.

Keywords: PV cells; dentate gyrus; graph analysis; hippocampal network; object recognition memory; pattern separation.

FIGURE 1

Experimental design for behavioral and immunofluorescence data. (A) Rats performed the novel object recognition (NOR) and similar novel object recognition (SNOR) tasks with multiple trials under different conditions: all similarity levels grouped (25, 50, and 75%) or separated by similarity level. (B) Representative schematic of object overlap calculation using Lego blocks, adapted from Johnson et al. (2017). (C) Assessment of neural areas recruited in each condition via c-Fos expression and PV+/c-Fos+ double labeling, followed by the construction of two interregional correlation matrices: all c-Fos+ cells (left) and activated PV+/c-Fos+ and c-Fos+ only cells (right).

FIGURE 2

Behavioral results for each experimental condition. Cumulative discrimination index per trial and total exploration time for each experimental condition. The discrimination index is shown in the main panels, and the total exploration time is presented in the upper right corner of each graph. One-sample t-tests were applied to evaluate the discrimination index performance against chance. Total exploration times were compared between conditions using ANOVA. Significant differences are indicated by p < 0.05; N = 6–7 animals per group. Error bars represent the standard error of the mean (SEM).

FIGURE 3

Representative delimitation of hippocampal and parahippocampal regions and c-Fos expression between groups in the rostral hippocampus and parahippocampal cortex. (A) Representative delimitation of hippocampal subregions along the rostrocaudal axis (rostral, medial, and caudal—subdivided into dorsocaudal and ventrocaudal) and parahippocampal areas selected for morphometric and neurochemical analyses. Blue lines indicate regional boundaries, and blue squares denote sampling sites in the superficial (sl) and deep (dl) cortical layers. Arrows in the lower panels (from left to right) indicate Pv+/c-Fos+ cells, and red arrowheads indicate Pv+/c-Fos- cells. The rightmost panel displays the merged image of the triple immunostaining. (B) Normalized c-Fos expression in subregions of the rostral hippocampus across different similarity conditions. (C) Normalized c-Fos expression in subregions and layers of the parahippocampal cortex for the various similarity conditions. Each color corresponds to a specific condition, as indicated in the legend. Each graph corresponds to the subarea of interest. Brackets above the graphs denote the pairs of conditions compared. Depending on data distribution, Kruskal-Wallis tests with Dunn’s post hoc comparisons, or planned two-way orthogonal contrasts, were applied. Here only areas with significant differences between groups were represented. *Significant differences are indicated by p < 0.05 ‡Denotes trends approaching significance. Sample sizes ranged from N = 6–7 animals per group. Error bars represent the standard error of the mean (SEM).

FIGURE 4

Network analysis of all c-Fos+ activated cells in the hippocampal and parahippocampal regions under different similarity conditions. (A) Correlation matrix constructed from Spearman correlations between medial temporal lobe areas at different rostrocaudal-dorsoventral hippocampus levels for each experimental group. Warm colors indicate stronger positive correlations, cold colors indicate stronger negative correlations. *Significant correlations at p < 0.05. **Significant correlations at p < 0.01. ***Significant correlations at p < 0.001. (B) Graphs constructed for each group. Outer circle with different colors representing different portions of the rostrocaudal axis of the hippocampus, see matches in the upper right. Inside the graph, each circle corresponds to a node (region). Edges represent correlations between areas. Lines around a graph represent the number of clusters in the graph. In the upper right is explained, Edge thickness represents the strength of the correlation. Darker lines represent negative correlations. Colors represent different levels of the rostrocaudal axis along the hippocampus. Node colors represent centrality levels among them. Node size represents degree. Hubs in each network have their names highlighted in red. N = 6–7.

FIGURE 5

Interneuron recruitment (PV+/c-Fos+) between groups in the DG-CA3 microcircuit along the dorsoventral anteroposterior axis of the hippocampus. Normalized c-Fos expression in DG layers and different CA3 portions in the rostral, medial, caudal dorsal, and caudal medial hippocampus across experimental conditions. Each line of graphs in the figure represents a portion of the hippocampus. Each graph corresponds to a subarea of the hippocampus. Each color corresponds to a specific condition, as indicated in the legend. Brackets above the graphs denote the pairs of conditions compared. Depending on data distribution, Kruskal-Wallis tests with Dunn’s post hoc comparisons, or planned two-way orthogonal contrasts, were applied. *Significant comparisons considering p < 0.05. ‡ Trends approaching significance; N = 6–7. ± Standard error of the mean.

FIGURE 6

Network analysis between all c-Fos+ activated cells versus PV+/c-Fos+ inhibitory activated cells in the DG-CA3 microcircuit across different anteroposterior and dorsoventral portions of the hippocampus under varying similarity conditions. Graphs constructed for each group. The outer circle with different colors represents different anteroposterior portions of the hippocampus. The lower half of each graph, outlined with a black line, represents PV+ interneurons; the upper half represents excitatory neurons. Within the graph, each circle corresponds to a node (region). Edges represent correlations between areas. The thickness of the edges represents the strength of the correlation. Red lines represent negative correlations; blue lines represent positive correlations. Colors denote different anteroposterior hippocampal portions. Node colors represent their centrality level. Node size represents degree. Network hubs have their names highlighted in red. N = 6–7.

FIGURE 7

Representation of the Bayesian structural equation model depicting theoretical relationships regarding the modulation of all c-Fos+ activated cells and PV+/c-Fos+ activated interneurons located in the DG layers on the activity of cells in the granule layer. Each experimental condition is represented by a distinct structural equation model (conditions identified above each model). Activated cells - possible Mossy cells (MCs) - located in the Hilus are highlighted in yellow. Parvalbumin-positive (PV+) interneurons in both the Hilus and granule cell layer are highlighted in green, while activated cells in the granular layer (GCs) are shown in lilac. Solid lines represent modeled pathways, with circles at the end of each path indicating the target variable of the effect. Positive β coefficients are depicted in blue, and negative β coefficients in red. Circles at the end of each path represent the target location of the connection. Indirect pathways are illustrated by dashed arrows: indirect pathways from PV interneurons in the Hilus to GCs are shown in green, whereas indirect positive or negative pathways from MCs to GCs are shown in blue and yellow, respectively. Each modeled path is accompanied by its corresponding standardized β value. The β value of Direct effects betas are highlighted in bold, other β value are indirect effects. Total effect paths are shown at the bottom of each model. Total effect paths were computed as follows: (i) MC → GC (direct effect plus or minus indirect effects) and (ii) PV in Hilus → GC (direct effect plus or minus indirect effects), with the sign of the total effect determined by the direction of the indirect effect. Asterisks (*) indicate significant β values, defined as those whose credible intervals do not cross zero. Total effect values for MCs on GCs and PV interneurons in the Hilus on GCs are shown at the bottom of each model. Sample sizes ranged from N = 6–7.