Engram-specific transcriptome profiling of contextual memory consolidation

Abstract

Sparse populations of neurons in the dentate gyrus (DG) of the hippocampus are causally implicated in the encoding of contextual fear memories. However, engram-specific molecular mechanisms underlying memory consolidation remain largely unknown. Here we perform unbiased RNA sequencing of DG engram neurons 24 h after contextual fear conditioning to identify transcriptome changes specific to memory consolidation. DG engram neurons exhibit a highly distinct pattern of gene expression, in which CREB-dependent transcription features prominently (P = 6.2 × 10^-13), including Atf3 (P = 2.4 × 10^-41), Penk (P = 1.3 × 10^-15), and Kcnq3 (P = 3.1 × 10^-12). Moreover, we validate the functional relevance of the RNAseq findings by establishing the causal requirement of intact CREB function specifically within the DG engram during memory consolidation, and identify a novel group of CREB target genes involved in the encoding of long-term memory.

Fig. 1

Activity-dependent, sustained expression of Arc::dVenus in DG granule cells. a Experimental setup. Arc::dVenus mice were fear conditioned and the number of dVenus⁺ cells was measured in the DG at successive time-points; 1 h (n = 5), 5 h (n = 7), 8 h (n = 5), 14 h (n = 5) and 24 h (n = 7), after training. Home-cage (HC) controls (n = 5) serve as a baseline. b Number of dVenus⁺ cells per 1.3 mm² section in the DG, at specific time-points after fear conditioning. Analysis of variance: effect of training history over baseline (HC): F_(1,33) = 13.102, P = P = 1.0 × 10⁻⁵; post hoc LSD: HC vs. 1 h: P = 2.2 × 10⁻⁵, HC vs. 5 h: P = 1.9 × 10⁻⁵, HC vs. 8 h: P = 4.0 × 10⁻⁵, HC vs. 14 h: P = 6.0 × 10⁻⁶, HC vs. 24 h: P = 4.5 × 10⁻⁸. c Representative images of the DG from fear conditioned mice at each successive time-point after fear conditioning. Scale bar: 200 μm. d Animals were implanted with microendoscopes to longitudinally monitor in vivo dVenus fluorescence in the DG (n = 3). e Percentage of dVenus⁺ cells at 5 h that also express dVenus 24 h after fear conditioning. f Percentage of dVenus⁺ cells at 24 h that also expressed dVenus 5 h after fear conditioning. g Representative microendoscopy images of dVenus⁺ cells at 5 and 24 h. Colored arrows indicate cells expressing dVenus at both time-points. Scale bar: 100 μm. h Percentage of dVenus⁺ cells in the DG that also express endogenous Arc in home-cage controls (HC, n = 4), no shock controls (NS, n = 4) or fear conditioned animals (FC, n = 4). Multivariate analysis of variance: F_(2,12) = 40.2, P = 0.0003, post hoc LSD: HC vs. NS: P = 0.006, HC vs. FC: P = 0.0001, NS vs. FC: P = 0.003. i Representative images demonstrating co-expression of endogenous Arc and dVenus.   *P < 0.05, **P < 0.01, ***P < 0.001. Data are presented as mean ± SEM. Scale bar: 200 μm. Source data are provided as a Source Data file

Fig. 2

Fear conditioning induces a unique transcriptional profile in DG engram cells. a Experimental setup. Nucleated patch aspiration was performed 24 h after fear conditioning (FC, n = 12 biological replicates), context-only exposure (NS, n = 4 biological replicates), or naïve home-cage controls (HC, n = 3 biological replicates). b Sample-to-sample principal component analysis. PC1 scores separated samples by state of activation (dVenus⁺ [green] vs. dVenus⁻ [magenta]) across all experimental groups, while PC2 separated samples based on their training history (fear conditioned group [FC] vs. naïve home-cage [HC] and no-shock [NS] controls). Orange rectangle delineates the corresponding PC1/PC2 isolated quadrant. c Differential expression between dVenus⁺ and dVenus⁻ cells for all genes with a raw P < 0.05. Dotted line indicates P_adj < 0.05 (FDR corrected). Genes that are upregulated in dVenus⁺ cells are in red, and genes that are downregulated in dVenus⁺ cells are in blue . The top 7 up and downregulated genes along with the total number of regulated genes with P_adj < 0.05 are labeled. d Log₂ fold change of a panel of known activity regulated genes between dVenus⁺ and dVenus⁻ cells 24 h after fear conditioning. PRGs primary response genes, SRGs secondary response genes. Data are presented as mean ± SEM. e Functional pathway enrichment with P < 0.01 of differentially expressed genes in the FC group. The enrichment of these pathways in the NS and HC groups is plotted alongside the FC group. Gray dotted line indicates significance threshold set at −log₁₀ P > 1.3 (P < 0.05, Fisher’s exact test), and blue dotted line indicates significance threshold set at −log₁₀ P > 2 (P < 0.01, Fisher’s exact test). f Gene ontology (GO) analysis of molecular function revealed Ion channel activity as overrepresented in the FC group (GO:0005216,       P = 2.7 × 10⁻⁵, FDR corrected Fisher’s exact test). Of the 40 genes in this GO class, 11 were K+ channels. The genes of these K+ channels are plotted in the right panel as a log₂ fold change between dVenus⁺ and dVenus⁻ cells. Data are presented as mean ± SEM

Fig. 3

Distinct activation of a CREB-dependent network exclusively in DG engram cells. a Fear conditioning-induced CREB-dependent gene network activation. Twenty-two of the top 50 significantly regulated genes after FC are part of the CREB network, of which 14 have direct transcriptional regulation. b Differential expression between dVenus⁺ and dVenus^–cells of the 22 genes identified in the CREB network. Dotted line indicates P_adj < 0.05 (FDR corrected). Red: Genes upregulated, Blue: Genes downregulated, in dVenus⁺ cells. c Multiplex RNA-scope validates the differential expression pattern of Arc, Atf3, and dVenus 24 h after fear conditioning. Left: Log₂ fold change of fluorescence intensity between Arc⁺ and neighboring Arc⁻ cells is reported for each gene (Arc + Atf3 + dVenus: n = 4). Analysis of variance: Arc: F_(1,7) = 10.19, P = 1.9 × 10⁻², Atf3: F_(1,7) = 39.58, P = 7.5 × 10⁻⁴, dVenus: F_(1,7) = 225.17, P = 6 × 10⁻⁶. Right: Representative images demonstrating positive and negative-control probes as well as co-expression patterns of Arc (green), Atf3 (red), and dVenus (cyan) in the DG of animals. DAPI (blue) labels all cells. 6×. d Multiplex RNA-scope experiments to validate the differential expression pattern of Arc, Penk, and Kcnq3 24 h after fear conditioning. Left: Log₂ fold change of fluorescence intensity between Arc⁺ and neighboring Arc⁻ a cell is reported for each gene (Arc + Penk: n = 4, Arc + Kcnq3: n = 4). Analysis of variance: Upper: Arc: F_(1,7) = 15.30, P = 7.8 × 10⁻³, Penk: F_(1,7) = 24.91, P = 2.5 × 10⁻³, Lower: Arc: F_(1,7) = 7.87, P = 3.1 x 10^-2, Kcnq3: F_(1,7) = 40.86, P = 6.9 x 10^-4. Right: Representative images demonstrating co-expression patterns of Arc (green) and Penk (red), or Arc (green), and Kcnq3 (red) in the DG of animals. DAPI (blue) labels all cells. c, d Double arrows: Arc⁺ cells, single arrows: neighboring Arc⁻ cells. *P < 0.05, **P < 0.01, ***P < 0.001. Data are presented as mean ± SEM. Scale bar: 20 μm. Source data are provided as a Source Data file. e Group-wise analysis of significantly regulated genes under direct transcriptional regulation of CREB. The overlap P value measures the enrichment of regulated genes from our data sets, compared to previously identified CREB targets. The activation z-score predicts the activation state of the upstream regulator (CREB in this case) based on the log₂-fold change values of CREB targets. z-scores greater than 2 or smaller than −2 are considered significant

Fig. 4

Disruption of CREB function prevents regulation of CREB target genes. a Experimental design. Fos::tTA mice were injected with AAV5-TRE::EGFP-mCREB targeting the DG. b On-Dox FC group remained on Dox throughout the experiment, while the off-Dox (HC and FC) groups were placed back on Dox immediately after training. Animals were sacrificed 4 h post-training. c Representative images demonstrating expression of EGFP-mCREB in DG neurons after fear conditioning. FC training on Dox induced very low expression of Fos::tTa driven EGFP-mCREB. Among animals off Dox, fear-conditioned animals (FC) showed much higher EGFP-mCREB expression than HC controls. Scale bar: 100 μm. d Multiplex RNA-scope experiments validate the use of mCREB to disrupt the expression of CREB target genes. Log₂ fold change of fluorescence intensity between Arc⁺ and neighboring Arc⁻ cells is reported for the control vector and EGFP⁺ vs. EGFP⁻ cells for mCREB injected animals. Analysis of variance: Panel 1 and 2-Control vector (n = 4): Arc: F_(1,7) = 51.64, P = 3.7 × 10⁻⁴, Atf3: F_(1,7) = 22.16, P = 3.3 × 10⁻³. EGFP-mCREB vector (n = 4): EGFP: F_(1,7) = 79.13, P = 1.1 × 10⁻⁴. Panel 3 and 4-Control vector (n = 4): Arc: F_(1,7) = 14.45, P = 8.9 × 10⁻³, Penk: F_(1,7) = 42.60, P = 6.2 × 10⁻⁴. EGFP-mCREB vector (n = 4): EGFP: F_(1,7) = 96.21, P = 6.5 × 10⁻⁵. *P < 0.05, **P < 0.01, ***P < 0.001. Data are presented as mean ± SEM. Source data are provided as a Source Data file. e Representative images demonstrating co-expression patterns of Arc (green) and Atf3 (red) or Arc (green) and Penk (red) in animals injected with the control vector (panels 1 and 3) and EGFP-mCREB (cyan, panels 2 and 4) in the DG of animals injected with the EGFP-mCREB virus. DAPI (blue) labels all cells. Double arrows indicate Arc⁺/EGFP⁺ cells, while single arrows indicate neighboring Arc⁻/EGFP⁻ cells. Scale bar: 20 μm

Fig. 5

DG engram-specific disruption of CREB function impairs memory consolidation. a Experimental design. Fos::tTA mice were injected with AAV5-TRE::EGFP-mCREB (n = 8) or control AAV5-TRE::mCherry (n = 7) targeting the DG, and subsequently taken off Doxycycline prior to fear conditioning. Animals were placed back on Dox immediately after fear conditioning and tested for contextual memory 72 h later. b Freezing levels (%) during the training session, prior to footshock onset (pre) and following the termination of the last footshock (post). Analysis of variance: Control Pre vs. Post: F_(1,13) = 103.4, P = 3.0 × 10⁻⁷, mCREB Pre vs. Post: F_(1,15) = 163.8, P = 2.2 × 10⁻⁹, Control vs. mCREB (Post): F_(1,14) = 1.4, P = 0.26. c Mice injected with mCREB exhibited a significant contextual memory deficit when tested 72 h after training. Analysis of variance F_(1,14) = 11.41, P = 0.005. d Experimental design. Fos::tTA mice were injected with AAV5-TRE::EGFP-mCREB (n = 8) or control AAV5-TRE::mCherry (n = 8) targeting the DG, and subsequently taken off Doxycycline prior to exposure to a novel context. Animals were placed back on Dox immediately after and fear conditioned 24 h later followed by a contextual memory test 72 h after that. e Mice with mCREB expression in cells active during novel context exposure exhibited no memory deficit when tested 72 h after training. f, g The same number of DG cells express EGFP-mCREB after exposure to either the fear-conditioning context or a novel context or. f Representative images and g quantification of the number of EGFP-mCREB cells per 0.6 mm². Scale bar: 200 μm. n.s. not significant, **P < 0.01, ***P < 0.001. Data are presented as mean ± SEM. Source data are provided as a Source Data file