Adult-born hippocampal neurons promote cognitive flexibility in mice

Abstract

The hippocampus is involved in segregating memories, an ability that utilizes the neural process of pattern separation and allows for cognitive flexibility. We evaluated a proposed role for adult hippocampal neurogenesis in cognitive flexibility using variants of the active place avoidance task and two independent methods of ablating adult-born neurons, focal X-irradiation of the hippocampus, and genetic ablation of glial fibrillary acidic protein positive neural progenitor cells, in mice. We found that ablation of adult neurogenesis did not impair the ability to learn the initial location of a shock zone. However, when conflict was introduced by switching the location of the shock zone to the opposite side of the room, irradiated and transgenic mice entered the new shock zone location significantly more than their respective controls. This impairment was associated with increased upregulation of the immediate early gene Arc in the dorsal dentate gyrus, suggesting a role for adult neurogenesis in modulating network excitability and/or synaptic plasticity. Additional experiments revealed that irradiated mice were also impaired in learning to avoid a rotating shock zone when it was added to an initially learned stationary shock zone, but were unimpaired in learning the identical simultaneous task variant if it was their initial experience with place avoidance. Impaired avoidance could not be attributed to a deficit in extinction or an inability to learn a new shock zone location in a different environment. Together these results demonstrate that adult neurogenesis contributes to cognitive flexibility when it requires changing a learned response to a stimulus-evoked memory.

FIGURE 1

The active place avoidance task (A) A mouse walking on a rotating circular platform in a room with multiple visual cues learns to avoid a 60° region of the room that has been designated as the shock zone. The shock zone and the corresponding section of the platform where the shocks are delivered are indicated by shading. (B) Overhead view of the mouse sitting directly opposite the stationary shock zone as the platform rotates clockwise. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

FIGURE 2

Ablation of adult hippocampal neurogenesis. Doublecortin immunostaining in the granule cell layer of the dentate gyrus was reduced in (A) X-ray treated mice (F_1,8 = 49.21, *P < 0.01) and (B) GFAP-TK TG mice treated with ganciclovir (F_1,8 = 90.30, *P < 0.01). Representative photomicrographs of doublecortin immunoreactivity in the subgranular zone of the dentate gyrus of a WT mouse (C) and a GFAP-TK TG mouse (D) are shown. Scale bars, 200 mm. Error bars represent S.E.M. X-ray-treated (N = 5); sham-irradiated (N = 5); GFAP-TK TG (N = 5); WT (N = 5). [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

FIGURE 3

Irradiation impaired cognitive flexibility required during conflict trials. (A) Schematic of behavioral procedures. (B) Example behavior. The path (gray) of a representative X- and sham-irradiated mouse. Open red circles indicate where the animal was shocked during each trial. The shock was off during pretraining. (C) Initial training: X- (N = 13) and sham-irradiated (N = 13) mice did not differ in the average number of times they entered the inactive shock zone during pretraining (F_1,24 = 1.47, P = 0.24) or the active shock zone during six training trials (F_{1, 24} = 1.72, P = 0.20). The significant effect of trial confirmed that performance in both groups improved with training (F_{5, 20} = 10.69, P < 0.0001). (D) Conflict condition: After the location of the shock zone was switched, irradiated mice entered the new location of the shock zone significantly more than controls (F_{1, 101} = 14.98, P < 0.01). The significant effect of trial indicates that both groups improved with additional training (F_5,101 = 30.53, P < 0.01), although the lack of an interaction suggests that performance of irradiated mice was worse than controls across all six trials (F_5,101 = 2.23, P = 0.14). Error bars represent S.E.M. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

FIGURE 4

Color-coded time-in-location maps for each treatment group during active place avoidance. Group-averaged time-in-location maps for each 10-min trial were computed to visually assess the patterns of avoidance behavior, which evolved with training experience. A blue-to-red scale was used, where blue represented the locations with the lowest dwell time and red represented the locations with the maximum dwell times. (A) Pretraining and each trial of initial training are shown. The same color category assignments were used for all maps, where the minimum time for each color assignment in seconds was: 0.09 (light blue), 2.45 (green), 6.09 (orange), 10.36 (red). (B) Each conflict trial is shown. The same color category assignments were used for all maps, where the minimum time for each color assignment in seconds was: 0.25 (light blue), 1.75 (green), 6.50 (orange), 15.25 (red). X-ray-treated (N = 13); sham-treated (N = 13). [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

FIGURE 5

Irradiated mice were not impaired in extinguishing avoidance responses. (A) Schematic of behavioral procedures. (B, left) Initial training: Irradiated (N = 8) and sham-irradiated (N = 8) mice did not differ in the total amount of time they spent in the active shock zone during six trials of initial training (F_1,14 = 0.40, P = 0.54) (B, right) Extinction training: Groups did not differ in the total amount of time they spent in the inactive shock zone during 15 trials of extinction training (F_1,14 = 1.25, P = 0.28). [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

FIGURE 6

Irradiated mice were specifically impaired in trials involving a conflict. (A) Schematic of behavioral procedures. (B) Initial training: X- (N = 25) and sham-irradiated mice (N = 25) did not differ in the average number of times they entered the shock zone during the first two trials of initial training (F_1,48 = 0.0006, P = 0.98). (C) Conflict condition: Irradiated mice (N = 13) entered the new location of the shock zone significantly more than sham-irradiated mice (N = 13) during the first two conflict trials (F_1,24 = 6.85, P < 0.05). (D) No conflict-initial conditions: X- (N = 6) and sham-irradiated mice (N = 6) did not differ in the average number of times they entered the shock zone during two trials that were identical to the initial training trials (F_1,10 = 1.88, P = 0.20). (E) No conflict-new conditions: X (N = 6) and sham-irradiated mice (N = 6) did not differ in the average number of times they entered a new location of the shock zone during two trials that occurred within a novel context (F_1,10 = 1.74, P = 0.22). Error bars represent S.E.M. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

FIGURE 7

Irradiation impaired cognitive flexibility required with the addition of a shock zone. (A) Schematic of behavioral procedures. (B) Initial training: X- (N = 8) and sham-irradiated (N = 7) mice did not differ in the average number of times they entered the stationary shock zone during six initial training trials (F_1,13 = 0.67, P = 0.43). (C, D). After the conditions of the task were changed, there was a significant effect of treatment (sham vs. irradiation) (F_1,312 = 5.88, P < 0.05) and a significant treatment ×shock zone (initial vs. new) interaction (F_1,312 = 4.70, P < 0.05). Post-hoc tests: Average entrances of irradiated mice into initial stationary shock zone = average entrances of sham mice into initial stationary shock zone. Average entrances of irradiated mice into new rotating shock zone > average entrances of sham mice into new rotating shock zone. Error bars represent S.E.M. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

FIGURE 8

Irradiation did not impair initial learning of the simultaneous two-frame avoidance task variant. (A) Schematic of behavioral procedures. (B, C) Compared to sham controls (N = 7), irradiated mice (N = 8) were not impaired in learning to avoid the stationary shock zone or the rotating shock zone across all 12 trials. The effect of treatment (sham vs. irradiation), (F_1,312 = 0.05, P = 0.82) and the treatment × shock zone (stationary vs. rotating) interaction (F_{1, 312} = 0.38, P = 0.54) were not significant. Error bars represent S.E.M. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

FIGURE 9

Genetic ablation of adult neurogenesis impaired cognitive flexibility required during conflict trials. (A) Example behavior. The path (gray) of a representative WT and GFAP-TK TG mouse. (B) Initial training: GFAP-TK TG (N = 12) and WT (N = 8) mice did not differ in the average number of times they entered the inactive shock zone during pretraining (F_1,18 = 2.93, P = 0.10) or the active shock zone during six initial training trials (F_1,18 = 0.80, P = 0.38). Avoidance behavior improved across trials in both groups (F_{5, 14} = 6.42, P < 0.01). (C) Conflict condition: After the location of the shock zone was switched, GFAP-TK TG mice entered the new shock zone location significantly more than WT mice (F_{1, 18} = 4.66, P < 0.05). There was a significant effect of trial (F_{5, 14} = 7.60, P < 0.01) and no significant interaction (F_{5, 14} = 1.19, P = 0.36). (D) Average number of times each group entered the shock zone during the first two initial training and conflict trials. There were significant effects of shock zone location (F_{1, 36} = 11.37, P <0.01) and genotype × shock zone location interaction (F_{1, 36} = 4.62, P < 0.05). Post-hoc tests confirmed that GFPAP-TK TG mice entered the switched shock zone more than WT controls (*P < 0.05). Error bars represent S.E.M. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

FIGURE 10

Impaired cognitive flexibility was associated with upregulation of Arc protein in the dentate gyrus. (A, left) Representative ×10 (top) and × 20 (bottom) photomicrographs of Arc+ cells in the dorsal GCL of a sham mouse tested in the No conflict condition. Scale bar, 200 mm (top) and 50 mm (bottom). Multiple photographs were taken at × 20 (top) and × 40 (bottom) and reconstructed. (A, right) Quantification of Arc+ cells in the dorsal half of the GCL. There were significant effects of behavioral condition (F_2,25 = 33.20, P < 0.01) and treatment × behavioral condition interaction (F_2,25 = 3.51, P < 0.05). Post-hoc tests: irradiated mice in the conflict condition (N = 5) > sham mice in the conflict condition (N = 6) = sham mice in the no conflict condition (N = 5) = irradiated mice in the no conflict condition (N = 5) >Naïve irradiated mice (N = 5) = Naive sham mice (N = 5), (B, left) Representative ×10 photomicrograph of Arc+ cells in the ventral GCL of a sham-irradiated mouse in the no conflict condition. Scale bar, 200 mm. Multiple photographs were taken at ×20 and reconstructed. (B, right) Quantification of Arc+ cells in the ventral half of the GCL. There was a significant effect of behavioral condition (F_{2, 25} = 14.07, P < 0.01) but not treatment (F_1,25 = 0.85, P = 0.37) or treatment ×behavioral condition interaction (F_{2, 25} = 0.28, P = 0.76). Post-hoc tests: Mice in the no conflict condition = mice in the conflict condition > naïve mice. Error bars represent S.E.M. *P < 0.05. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]