Generation of the ACT-DEPP dataset.

(A) Experimental schematic of generation of the ACT-DEPP dataset. (B) Distribution of nuclei number, experimental conditions, and sample metadata among major subclasses. Some neocortical and subiculum Subclasses appearing in Figure 1D have been omitted from this panel, as they were not analyzed in any further downstream analyses. (C) UMAP embedding of ACT-DEPP nuclei colored by subclass. Inset: Single-nucleus transcriptomes were mapped to the Allen Brain Cell (ABC) whole-brain atlas, which assigns cells in a hierarchical taxonomy to one class, subclass, supertype, and cluster. (D) As in (C), but colored by activity condition. (E-H) Sankey flow diagrams showing changes in ABC-taxon identity for nuclei before (+) and after (-) removing ARGs from single-nucleus transcriptomes. (G) and (H) show the effect at lower ABC taxa for the two largest neuronal subclasses, CA1 and DG. CA1-ProS, cornu ammonis 1-prosubiculum; L2/3 PPP, layer 2/3 pre-, pro-, parasubiculum; NP Sub, near-projecting subiculum; CT Sub, corticothalamic subiculum; L6 Ent, layer 6 entorhinal; DG, dentate gyrus granule cells; Sub-ProS, Subiculum-prosubiculum; Micro, microglia; Astro, astroglia; Oligo, oligodendrocyte; OPC, oligodendrocyte precursor cells; Epen, ependymal cell; NGF, neurogliaform cell; IMN, DG immature neuron; SST, GABAergic somatostatin-expressing interneuron; PV, GABAergic parvalbumin-expressing interneuron; GABAergic vasopressin internal peptide-expressing interneuron; Sncg, GABAergic Sncg-expressing interneuron; Lhx6, GABAergic Lhx6-expressing interneuron

The activity-dependent transcriptome is stimulus-specific and obscures cell identity in DG.

(A-B) Venn diagrams of shared DEGs between KA or EE vs SE at 30 min (A) or 6 hr (B) in CA1 (nEE30m = 3329 nuclei, nSE = 1907 nuclei, nKA30m = 354 nuclei) and DG (nEE30m = 7119 nuclei, nSE = 6754 nuclei, nKA30m = 961 nuclei). DEGs were defined as genes with absolute log2(FC) > 1.5 and FDR < 0.05 using pseudobulk expression (Methods). (C) Re-embedded UMAPs of excitatory neuronal nuclei colored by Subclass (left, top), activity condition (left, bottom), or Supertype (right). Cross-reference of the UMAPs reveals KA-specific populations marked by red arrows. (D) Distribution of Supertypes within each subclass across different conditions. Colors within each column represent proportions of different Supertypes within the labeled Subclass. (E) As in (D), but for DG Clusters only. (F-G) Single-nucleus, SC-transformed expression (Hafemeister & Satija 2019) of marker genes for DG Cluster identity, plotted by condition (F) or Cluster (G)

Discovery of cell-type-specific ERGs induced by novel experience.

(A) Change in expression in EE30m vs SE for all ERGs in excitatory neuronal subclasses; gene columns are ordered by hierarchical clustering, and genes encoding transcription factors are annotated with black ticks above heatmaps. For every Subclass, we performed pseudobulk DEG analysis at EE 30 min vs SE and EE 6 hr vs SE. The heatmap contains all DEGs—ascertained from any Subclass listed—that were classified as ERGs. ERGs are defined as DEGs whose fold-change in EE vs SE was higher after 30 min instead of 6 hr (Methods). (B) Volcano plot of DEGs in CA1-ProS nuclei at EE30m (n = 3329 nuclei) vs SE (n = 1907 nuclei). (C) Heatmap of upregulated ERGs in excitatory neuronal Subclasses, ordered by Subclass-specific induction (tau score; Methods). Canonical ERGs are annotated in black text, while novel example Subclass-specific genes are annotated in red text. (D) Violin plots of SC-transformed expression for example genes with high tau scores. (E-H) as in (A-D) but for inhibitory neuronal subclasses. (I-K) as in (A-D) but for immature neurons and glial subclasses.

Distribution of IEG expression across CA1-ProS Supertypes.

(A) Example plots demonstrating IEG induction thresholds using Fos, Arc, and Nr4a1. For a given gene, the induction threshold was defined as the 90th percentile of expression among SE nuclei; thresholds are represented as dotted lines. (B) Activation of single cells across activity conditions within CA1-ProS neurons. Active cells were defined as those that induced 3 or more IEGs, similar to the approach in Hrvatin et al. 2018 (Methods). (C) 3-D plots of ABC MERFISH brain slices (published data from Yao et al. 2023) with only CA1-ProS Supertypes plotted. (D-E) Activation percentages as in (B), de-aggregated by Supertype. (F) Gene-gene correlation matrices for IEG expression within active cells at EE 30 min. Dendrograms represent hierarchical clustering of genes.

Circadian-dependent gene expression among major hippocampal cell types.

(A) Pseudobulk expression of canonical clock genes across the circadian period in SE. Error bars represent s.e.m. (B-E) Above: UMAP embeddings of nuclei in circadian-DEG space, colored by ZT of collection (Methods). Below: Pseudobulk expression for circadian DEGs across the diurnal cycle. Clock genes (yellow), high-tau genes (red), and ARGs (black) are highlighted.

Interactions between activity- and circadian-dependent transcription in the hippocampus.

(A-D) Expression change between EE30m vs SE for IEGs across the circadian period. Baseline pseudobulk expression for each gene at each ZT is plotted on the right. (E-H) As in (A-D) but for canonical clock genes. (I) Expression change between EE30m vs SE for genes with significant ZT:activity_condition interaction (Methods). Baseline pseudobulk expression for each gene at each ZT is plotted on the right. (J) As in (I) but for DG. Note: astrocytes and oligodendrocytes had no significantly interacting genes between ZT and activity condition.