NG00162 – Oligodendrocytes single-cell whole genome and RNA sequencing

Characterizing the mechanisms of somatic mutations in the brain is important for understanding aging and disease, but little is known about the mutational patterns of different cell types. We performed whole-genome sequencing of 86 single oligodendrocytes, 20 mixed glia, and 56 single neurons from neurotypical individuals (0.4 to 104 years old) and compared the rates and signatures of somatic single nucleotide variants (sSNVs) and small insertions and deletions (indels) from each cell type. We further correlated this data with single-cell RNA (scRNA-seq) and chromatin accessibility (scATAC-seq) data generated from the same brains to compare the mutagenic processes in glia and neurons.

single-cell whole genome sequencing (scWGS):

Fluorescence-activated nuclear sorting (FANS) was used to isolate SOX10 cells from fresh frozen human brain tissue from the prefrontal cortex. Whole-genome amplification was performed using MDA or PTA following manufacturer guidelines. Libraries for sequencing were generated using the KAPA HyperPlus kit (Roche) using dual indexes and were sequenced across 5 lanes of Ilumina NovaSeq6000 (2x150bp), targeting 20x coverage (75Gbp)/sample. SCAN2 was used to identify single-cell somatic mutations.

single-cell RNA sequencing (scRNA-seq):

Sequencing libraries were prepared using the 10X Genomics Chromium Next GEM Single Cell Reagent Kit v3.1 with nuclear pellets from fresh frozen human brain tissue from the prefrontal cortex of 2 individuals. Each library preparation was submitted for paired-end single indexing sequencing on Illumina HiSeqX or NovaSeq6000 targeting ~50,000 read pairs per nucleus. The data was demultiplexed using bcl2fastq. scRNA-seq FASTQ files were processed using the 10X Genomics cellranger count pipeline for gene expression to perform alignment to hg19, barcode counting, UMI counting, and generation of feature-barcode matrices. Cell Ranger filtered count matrices were used for downstream analysis using Seurat 3.0. Each library was further filtered for cells with > 200 and < 3000 genes and <5% mitochondrial genes, and genes with <10,000 UMI counts and >3 cells. RNA counts were normalized using the LogNormalize method and the 2,000 most highly variable features were identified using the vst method. Data were scaled by regressing out the percentage of mitochondrial genes. Non-linear dimensional reduction and clustering was then performed. DoubletFinder was used to remove doublets using optimal parameters as per the paramSweep function. Finally, cell-type identities were assigned to each cluster in the Uniform Manifold Approximation and Projection (UMAP) based on expression of known brain cell-type markers.

single-cell ATAC sequencing (scATAC-seq):

Nuclei were obtained from the same brain region as used for single-cell whole-genome amplification. Nuclei derived from different individuals were processed for transposition separately, before loading to the 10x Chromium Controller for GEM generation, barcoding, and library construction, as per manufacturer instructions. Libraries were submitted for paired-end dual index sequencing on one flow cell of Illumina S2 NovaSeq6000 (100 cycles) to obtain ~50,000 reads per nucleus. Sequencing data were demultiplexed using bcl2fastq and mkfastq. cellranger-atac count v1.1.0 was run separately on the resulting FASTQ files for each scATAC-seq library (one per individual) with default parameters and the vendor-provided hg19 reference. Results from the individual library analyses were then merged by cellranger-atac aggr –normalize-depth. scATAC-seq data were analyzed by Signac v1.1.0 and Seurat v3 following the authors’ instructions.