AAV DJ-Syn-GFP

Episodic memories are stored in distributed neurons, but how individual neurons are determined to be fit for a coding ensemble remains unclear. Here, researchers identified thousands of regions that are predominantly bistable (CpG methylated or unmethylated) within neuronal genomes established during development in the mouse hippocampal dentate gyrus. Reducing DNA methylation and the proportion of methylated epitopes in bistable regions impairs novel context-induced neuronal activation and spatial memory. Conversely, increasing the frequency of methylated and unmethylated epitopes in bistable regions enhances intrinsic excitability and spatial memory but impairs spatial working memory, suggesting that the methylated-unmethylated epitope balance established during development in bistable regions is critical for normal neuronal excitability and hippocampal cognitive function. Researchers propose a model in which epigenetic bistable regions generate neuronal heterogeneity and that specific combinations of exonic epialleles determine the fitness of the coding ensemble by modulating neuronal excitability.

Adult mice were injected bilaterally in the dDG with either DNMT3A expressing AAV-DJ-SYN-DNMT3a-GFP virus or control AAV-DJ-SYN-GFP virus, titrated to achieve a relatively sparse expression pattern (Figure 1a). Twenty days later, animals injected with both Dnmt3a and control viruses exhibited GFP expression in the DGC. Overexpression of DNMT3A (OE) resulted in population-level hypermethylation of a large number of CpG sites, albeit restricted to a small subset of cells. These CpGs, like IMR CpGs, showed predominantly IM in control cells (Figure 1b, pink), but were relatively hypermethylated in OE neurons, showing a rightward shift in methylation (Figure 1b, blue). Unaffected CpGs in OE cells exhibited a bimodal distribution, with no clear rightward shift in their methylation profile (Figure 1c), indicating that hypermethylation in OE cells is primarily restricted to IM CpGs. Notably, the hypermethylated CpGs clustered into 13,745 DMRs (>2 CpGs), which largely overlapped with previously identified IMRs (67.38%, Fig. d). Furthermore, most genes associated with OE-DMRs were also IMR genes (Figure 1e). Similar to the hypomethylation in cKO, the hypermethylation in OE was partial, as not all IMR genes were affected and only some of the IMRs in genes with multiple IMRs were hypermethylated. Overall, these data indicate that IMRs can be hypermethylated, at least partially, by the OE of Dnmt3a.

Figure 1. Overexpression of DNMT3A in DGCs increases methylation at a subset of IMRs. (Odell S C, et al., 2020)