BDNF in the dentate gyrus is required for consolidation of "pattern-separated" memories

Abstract

Successful memory involves not only remembering information over time, but also keeping memories distinct and less confusable. The computational process for making representations for similar input patterns more distinct from each other has been referred to as "pattern separation." In this work, we developed a set of behavioral conditions that allowed us to manipulate the load for pattern separation at different stages of memory. Thus, we provide experimental evidence that a brain-derived neurotrophic factor (BDNF)-dependent pattern separation process occurs during the encoding/storage/consolidation, but not the retrieval stage of memory processing. We also found that a spontaneous increase in BDNF in the dentate gyrus of the hippocampus is associated with exposure to landmarks delineating similar, but not dissimilar, spatial locations, suggesting that BDNF is expressed on an "as-needed" basis for pattern separation.

Graphical abstract

Figure 1

The Spontaneous Location Recognition Task (A) Cartoon depicting the apparatus used for the SLR task. Two or three objects were used according to the different conditions in which the task was run. Walls are drawn shorter than actual size for illustrative purposes. (B) Schematic of the SLR. (C) Percentage of time exploring each of the locations during the sample phase of the SLR task. Rats spent equal amount of time exploring each of the three locations during the sample phase. This indicates that the differences in the discrimination ratio cannot be explained by preferential exploration of the more separated location (A1) during the sample phase. (D) Discrimination ratios during the choice phase for the novel and familiar conditions. ^∗∗∗p < 0.001; n = 8. (E) Discrimination ratios during the choice phase 24 hr after the sample phase for trials in which object A5 was kept in a familiar location whereas A4 was moved either a small (50°) or a large (120°) distance. Discrimination ratios were significantly different from zero. ^∗∗p < 0.01, ^∗∗∗p < 0.001, one-sample t test; n = 7. Data are expressed as the mean ± SEM.

Figure 2

BDNF Activity in the DG Is Required for Memory Consolidation of Similar, but Not Dissimilar, Spatial Representations (A) Schematic of the SLR task for the similar (s-SLR) or dissimilar (d-SLR) conditions depicting the time points at which IgG or anti-BDNF was infused. (B) Coronal section indicating representative infusion sites in the DG. (C) Percentage of time exploring each of the locations during the sample phase of the s-SLR task for control IgG- or anti-BDNF injected rats. (D–F) BDNF antibodies or control IgGs (1 μg μl⁻¹ /0.5 μl side) were injected into the DG either 15 min before (D) or 5 min after (E) the sample phase. Injection of anti-BDNF into the DG 6 hr after the sample phase had no effect on s-SLR performance (F). ^∗∗p < 0.01, ^∗∗∗p < 0.001; n = 7. Data are expressed as the mean ± SEM.

Figure 3

BDNF Expression in the DG Is Required for Memory Consolidation of Similar, but Not Dissimilar, Spatial Representations (A) Schematic of the SLR task. (B) Effect of the infusion of BDNF antisense oligonucleotides (4 nmol μl⁻¹/0.5 μl side; BDNF ASO) or BDNF scrambled missense oligonucleotides (4 nmol μl⁻¹/0.5 μl side; BDNF MSO) in the DG on BDNF steady-state levels 7 hr or 24 hr after injection. Top left: representative blots of BDNF and actin protein levels in the DG, CA3, or CA1 regions 7 hr after oligonucleotide injections. Bottom left: representative blots for BDNF and actin protein levels in the DG 24 hr after oligonucleotide injections. Right: quantification of BDNF expression after ASO or MSO injection. ^∗∗∗p < 0.001; n = 4. (C) Exploration time during the sample phase or time spent exploring each of the locations 2 hr after BDNF ASO or BDNF MSO injection into the DG. (D) Effect of the injection of BDNF ASO or BDNF MSO into the DG 2 hr before the sample phase during a choice phase 24 hr later in the s-SLR or the d-SLR version of the task. ^∗∗∗p < 0.001; n = 7. Data are expressed as the mean ± SEM.

Figure 4

BDNF in the DG Is Not Required During Retrieval (A) Schematic of the SLR task. (B) Effect of BDNF antibodies (1 μg μl⁻¹ /0.5 μl side) injected into the DG 15 min before the choice phase on the s-SLR task compared to control IgGs (1 μg μl⁻¹ /0.5 μl side). p > 0.1; n = 7. Data are expressed as the mean ± SEM.

Figure 5

Exploration of Similar, but Not Dissimilar, Spatial Locations Is Associated with Increased BDNF Levels in the DG (A) Schematic illustration of the task configurations. (B) Total exploration time for each object in the small (top) and large (bottom) separation conditions. (C) Coronal brain section at coordinate −3.96 from bregma depicting the areas isolated for BDNF protein measurements. Tissue was punched and homogenized for SDS-PAGE. (D) Top: BDNF and actin protein levels in the DG and CA1 regions of rats subjected to the different conditions and corresponding representative blots. Bottom: Zif268 and actin protein levels in the DG and CA1 regions of rats exposed to the different conditions and corresponding representative blots. ^∗p < 0.05, ^∗∗p < 0.01; n = 6. (E) BDNF expression in the DG and CA1 after exposure to three objects. We used the same conditions as in the sample phase during the SLR task. BDNF and actin protein levels in the DG and CA1 regions of rats subjected to the different conditions and corresponding representative blots. ^∗p < 0.05; n = 8. Data are expressed as the mean ± SEM.

Figure 6

BDNF Enhances Consolidation of Similar Spatial Representations (A) Schematic of the extrasimilar SLR task (xs-SLR). The task was similar to the s-SLR except that in the xs-SLR task, two of the objects were brought even closer together during the sample phase, resulting in poor performance of control animals during the choice phase 24 hr later. (B) Percentage of time exploring each of the locations during the sample phase of the xs-SLR task. (C) Effect of recombinant human BDNF (0.5 μg μl⁻¹ /0.5 μl side; rhBDNF) or saline injected into the DG 5 min after the sample phase on performance during the choice phase 24 hr later. ^∗∗p < 0.01; n = 7. Data are expressed as the mean ± SEM.