Social and novel contexts modify hippocampal CA2 representations of space

Abstract

The hippocampus supports a cognitive map of space and is critical for encoding declarative memory (who, what, when and where). Recent studies have implicated hippocampal subfield CA2 in social and contextual memory but how it does so remains unknown. Here we find that in adult male rats, presentation of a social stimulus (novel or familiar rat) or a novel object induces global remapping of place fields in CA2 with no effect on neuronal firing rate or immediate early gene expression. This remapping did not occur in CA1, suggesting this effect is specific for CA2. Thus, modification of existing spatial representations might be a potential mechanism by which CA2 encodes social and novel contextual information.

Figure 1. CA2 encodes spatial information.

(a) Place field maps for three neurons and electrode tracks for CA1 and CA2. (b) Spatial information calculated for CA1 and CA2 neurons. CA2 neurons carried less spatial information than CA1 neurons (median(CA1)=2.191, median(CA2)=1.536, U=741, P<0.001, Mann–Whitney test, n=61 CA2 neurons and 31 CA1 neurons). (c) CA2 place cells had a higher average firing rate than CA1 place cells (median (CA1)=0.5146, median(CA2)=1.354, U=468, P<0.0001, Mann–Whitney test, n=61 CA2 neurons and 31 CA1 neurons). (d) CA1 and CA2 place cells had similar peak firing rates (median(CA1)=9, median(CA2)=9, U=860, P>0.05, Mann–Whitney test, n=61 CA2 neurons and 31 CA1 neurons). (e) Representative smFISH probing for Arc and Pcp4 in dorsal hippocampus from a rat that was killed 30 min after a 10-min exploration. (f) Time course of the percentage of cells with Arc foci in CA1 and CA2 following spatial exploration (main effect of time F(3,25)=11.09, P<0.0001; main effect of subregion F(1,25)=19.47, P<0.001, two-way paired ANOVA with Bonferroni post hoc tests denoted on graph; n=5–9 rats per condition). (g) Representative images of data in f. (h) High-magnification images of neurons outlined by white boxes in g. Arrows denote Arc foci throughout z-plane. HCC, home cage controls. Scale bar, 500 μm (a), 200 μm (e, left image), 50 μm (e, right image and g), 10 μm (h). *P<0.05; **P<0.01; ***P<0.001, ****P<0.0001. Error bars are s.e.m.

Figure 2. CA2 activity levels are not modified by social stimuli.

(a,b) Average firing rate before and during exposure to social stimuli for CA1 and CA2. No differences in average firing rates were detected for CA2 neurons during familiar exposure (median(pre-social)=0.5850, median(familiar)=0.5271, W=43.0, P>0.05, n=49 CA2 neurons) or novel exposure (median(pre-social)=0.5278, median(familiar)=0.5280, W=−162.0, P>0.05, n=55 CA2 neurons, Wilcoxon signed-rank test) or for CA1 neurons during familiar exposure (median(pre-social)=0.3210, median(familiar)=0.3800, W=336.0, P>0.05, n=55 CA1 neurons) or novel exposure (median(pre-social)=0.2875, median(familiar)=0.3682, W=431.0, P>0.05, n=58 CA1 neurons, Wilcoxon signed-rank test). (c,d) Time course of percentage of cells with Arc foci in CA1 and CA2 following spatial and social exploration (CA1: main effect of time F(2,48)=25.13, P<0.0001; main effect of condition F(2,48)=3.198, P=0.0497, two-way ANOVA. Bonferroni post hoc tests revealed no significant difference between conditions at any time point; however, context versus novel at 15 min trended towards significance (P=0.055). CA2: main effect of time F(2,48)=18.72, P<0.0001; main effect of condition F(2,48)=3.400, P=0.0416, two-way ANOVA. Bonferroni post hoc results are shown for the main effect of condition. See Supplementary Table 1 for post hoc tests for the main effect of time. HCC conditions are shown for comparison but were not included in the statistics. n=5–9 rats per condition). (e) Immunostaining of calbindin-1 (CALB1) in dorsal hippocampus and CA1 to delineate the target layer of CA2 axons and representative smFISH images probing for Calb1 and Arc across conditions. (f) Percentage of Calb1+ or − cells expressing Arc foci across conditions (main effect of condition F(2,13)=19.59, P=0.0001, two-way paired ANOVA with Bonferroni post hoc tests; n=5–6 rats per condition). (g) Per cent total Arc foci+cells that are Calb1+ or Calb1− (main effect of Calb1 expression F(1,13)=107.8, P<0.0001, two-way paired ANOVA with Bonferroni post hoc tests). (h) Arc mean nuclear fluorescence per cell measured in Arc foci+cells as a function of Calb1 expression (main effect of Calb1 expression F(1,13)=56.94, P<0.0001; main effect of condition F(2,13)=8.155, P=0.0051, two-way paired ANOVA with Bonferroni post hoc tests). Scale bar, 200 and 50 μm (left and right/lower). #,*P<0.05; ##,**P<0.001; ###,***P<0.001; ####,****P<0.0001. Error bars are s.e.m.

Figure 3. CA2 globally remaps on exposure to social and contextual stimuli.

(a) Example place maps for CA2 neurons (top) and CA1 neurons (bottom) on exposure to each of the four conditions depicted in the diagrams on top. In the first epoch of each of the four conditions, animals explored the open-field arena with no stimulus present. In the second epoch, animals were exposed to no stimulus (control), a familiar rat, a novel rat or a novel object. Stimulus rats were placed in a clear plexiglass chamber and novel objects were placed inside the open-field arena. For each place field map pair, the peak firing rate for the pair (in Hz, top) and the spatial correlation value (bottom) is listed to the left and heat maps for both place field maps are scaled to the peak rate. (b,c) Analyses of remapping of firing fields by CA2 neurons on the four exposures listed, including spatial correlations to measure global remapping (b) and changes in peak firing rates (c). CA2 place cells globally remapped their representation of space in response to familiar and novel rats and novel objects compared with control conditions with no social exposure (H(3,168)=13.32, P<0.01, Kruskal–Wallis test with Holm–Bonferroni correction; Fig. 3a,b, n=31, 37, 54 and 46 CA2 neurons for control, familiar social, novel social and novel object exposures). Global remapping occurred in CA2 without any significant changes in peak firing rate (H(3,168)=1.649, P>0.05, Kruskal–Wallis test, same n values as b). Spatial correlation values (d) and changes in peak firing rates (e) for CA1 neurons, which showed no global remapping (H(2,144)=5.434, P=0.0661, Kruskal–Wallis test, n=28, 59 and 57 neurons for control, familiar social and novel social, respectively) or changes in peak firing rate (H(2,144)=2.895, P>0.05, Kruskal–Wallis test, same n values as d) during control or social stimulation. *P<0.05; **P<0.01. Error bars are s.e.m.

Figure 4. CA2 place cell remapping in response to social stimulation was replicated in a different strain of rats.

Example colour-coded CA2 (upper) and CA1 (lower) place cell maps (dark blue indicates 0 Hz and dark red indicates peak rate for each cell) for the three behavioural paradigms illustrated above. Example maps are shown for all recorded CA2 and CA1 cells on a given day in two example rats for each condition (dotted horizontal lines indicate separation between cells from different rats). The peak rate for each cell is indicated to the left of its map. In the control condition (left column), rats explored an arena containing an empty cage and CA2 place cell maps changed gradually over time. In the social stimulation condition (middle column), a familiar rat was placed in the cage during the second and third behavioural sessions. Most CA2 place cells abruptly remapped in response to this social stimulation. The original maps did not return when the familiar rat was removed from the cage in the fourth recording session. In contrast, CA1 place cells exhibited stable maps across all sessions. In the control familiar object condition, a toy rat was placed in the cage during the second and third behavioural sessions. Neither CA1 nor CA2 cells exhibited consistent firing changes in response to the toy.

Figure 5. CA2 spatial correlations decrease with time and decrease further with social stimulation.

Spatial correlations for CA2 (a) and CA1 (b) place cells for control, familiar animal and familiar object (stuffed animal toy) conditions. Spatial correlations were calculated between the first and second sessions (A1–A2), the second and third sessions (A2–A3), the third and fourth sessions (A3–A4) and the first and fourth sessions (A1–A4). For both CA2 and CA1, temporally adjacent sessions (A1–A2, A2–A3 and A3–A4) had significantly greater spatial correlations than temporally distant sessions (A1–A4). These significant effects of time (two-way ANOVAs: effect of time, CA2 (n=96 cells): F(1,378)=15.2, P<0.001; CA1 (n=253): F(1,1066)=11.9, P=0.001) are indicated with ***P<0.001 and **P=0.001. (a) CA2 place cells remapped in response to social stimulation, as evidenced by significantly lower spatial correlations in the social stimulation condition than in the other conditions (repeated-measures ANOVA: effect of condition, F(2,93)=8.5, ***P<0.001; control: n=11 neurons, familiar animal: n=47, familiar toy: n=38). In addition, for A1–A4 comparisons, spatial correlations were significantly lower for the social condition than the other conditions (one-way ANOVA: F(2,93)=7.4, ##P=0.001; n-values same as above). (b) CA1 place cells did not show significant remapping in response to social stimulation (repeated-measures ANOVA: effect of condition, F(2,250)=0.6, P=0.6, control: n=26 neurons, familiar animal: n=134, familiar toy: n=93). Error bars are s.e.m.