Single cell methylation, RNA, and ATAC sequencing of ventral hippocampal brain tissue from inbred mouse strains

Despite advances in genetics over the last two decades, relating genetic variants to behavior remains a fundamental challenge. Particularly in mice, progress is frustrated by the low resolution of genetic mapping, and consequently by the thousands of candidate causal variants that must be screened. To assess the utility of DNA methylation marks in discovering causative variants we examined their relationship to genetic variation by generating genome-wide, base-pair resolution maps of multiple cell types from the hippocampus of eight fully sequenced inbred strains of Mus musculus. We find that the impact of sequence variation, as mediated by methylation changes, depends on local CpG sequence density. At CpG sequence densities of less than 40 CpG/Kb, cells compensate for loss of methylated sites by methylating additional sites to maintain methylation levels. At higher CpG sequence densities, the exact location of a methylated site becomes more important suggesting that variants affecting methylation sites in regions of high CpG density will have a greater effect on transcription. We test this hypothesis by assessing the effect of mutations on single cell transcript abundance for two inbred strains (C57BL/6J and DBA/2J). The effect of mutations that alter methylated CpG sites is restricted to regions of high sequence density, and from single cell ATAC-seq data, localizes predominantly to enhancers. With approximately tenfold fewer mutations in high than low sequence density CpG regions, candidate variants affecting transcript abundance, and hence downstream phenotypes, including behavior, can be prioritized based on CpG sequence density. Furthermore, our findings imply that DNA methylation influences the likelihood that mutations occur at specific sites in the genome, supporting the view that the distribution of mutations is not random. Adult male animals (Jackson Laboratories) were euthanized at 10-16 weeks old in an isoflurane chamber and decapitated. The brain was removed and the ventral region of the hippocampus was microdissected, snap frozen in dry ice, and stored at -80oC until processing. Tissue from ~2 animals were combined into a single tube and considered a replicate, with 2 replicates per strain for snmC-seq2, snRNA-seq, and snATAC-seq experiments. snmC-seq2 on microdissected tissue as previously described (Luo et al 2018). We selected for a 75-25 enrichment of neuronal vs non-neuronal nuclei during FACS sorting using NeuN-488/DAPI counterstains. For snRNA-seq, single nuclei suspension and library generation were completed at the Cedars Sinai Applied Genomics, Computation and Translational Core and followed the 10X protocol for the Chromium Next GEM Automated Single Cell 3’ Library and Gel Bead Kit v3.1 (cat# PN-100014) as described except for the following modifications: Suspensions from cell nuclei were generated using the recommended method from the 10X scMultiome protocol (CG000375 Rev C) to lyse cells and obtain nuclei. Following single nuclei suspension generation, nuclei were counterstained for 7-AAD and NeuN-405 antibody (Novus Biologicals, 1:200) and sorted on a MACSQuant Tyto prior to GEM generation. We selected for a 75-25 split of NeuN+/7-AAD+ nuclei for neurons and NeuN-/7-AAD+ for non-neuronal nuclei respectively. We captured ~10,000 nuclei per genotype per region per replicate on a single 10X GEM chip. All downstream library preparation was done according to the 10X protocol (CG000286) and sequenced on a Novaseq 6000 with a target of ~40-50k reads per nucleus. For snATAC-seq, single nuclei suspension and library generation were completed at the Cedars Sinai AGCT core and followed the 10X protocol for Next GEM scATAC-Seq v1.1 (PN-1000175) as described except for the following modifications: Nuclei suspensions were generated using the recommended method from the 10X scMultiome protocol (CG000375 Rev C) to lyse cells and obtain nuclei Following single nuclei suspension generation, nuclei were counterstained for 7-AAD and sorted on a MACSQuant Tyto prior to GEM generation. NeuN was not used for neuronal enrichment due to dye incompatibility between our NeuN antibody and a nuclear counterstain. After the sort, we carried out permeabilization of nuclei as per the protocol. We aimed to capture 10,000 nuclei per well x 8 wells, for a total of 80,000 nuclei over 8 total samples (~10,000 nuclei per genotype per region per replicate). All downstream library preparation was done according to the 10X protocol (CG000209) and sequenced on a Novaseq 6000 with a target of >35k reads per nucleus.