Atlasing single nuclei RNA sequencing from postmortem brain in ADRD

National Institute On Aging

Investigators

Linked publications, trials & patents

Abstract

We performed 5’ GEX 10x Genomics single nuclei sequencing for each brain region, PFC, PCC, OC. After processing using the Cell Ranger counts pipeline, we obtained 985,072 cells for the PFC sequenced to 65,166 reads/cell, for PCC we obtained 905,208 cells and 37,806 reads/cell, and for OC 938,447 cells and 43,074 reads/cell. For comparisons with our data, we processed PFC samples from ROSMAP [1] through our pipeline and obtained 2,437,524 cells with 39,522 reads/cell. To enable a comprehensive evaluation of lncRNA expression in the context of AD, we constructed a custom reference transcriptome by integrating lncRNAs from LncBook [2] with GENCODE v47 annotations. We prioritized RNAs annotated in GENCODE v47 and used this as the foundation for our reference. We then added lncRNAs from LncBook v2.1 (https://ngdc.cncb.ac.cn/lncbook/downloads) which contains the largest number of annotated human lncRNA. To ensure compatibility with Cell Ranger, which uses both intronic and exonic regions in its single nuclei pipeline, we excluded any LncBook RNAs that overlapped existing annotated genes. This filtering step was necessary because overlapping gene features can lead to ambiguity in read assignment, and such features are typically excluded from analysis by Cell Ranger to maintain consistency. Cell Ranger builds its single nuclei reference by collapsing overlapping annotations to ensure that each read maps unambiguously to a single gene, which simplifies quantification and reduces potential errors. Minor differences between our reference, GENCODE and the default Cell Ranger annotations are primarily due to category shifts in GENCODE annotations, removal of overlapping genes, and the exclusion of certain RNA biotypes, such as pseudogenes and other non-coding RNAs, from the standard Cell Ranger gene set. To characterize baseline gene expression profiles across cell types, we focused on the Not AD group (n=21) representing individuals without significant AD pathology. As has been previously described, excitatory neurons have the greatest diversity of genes [3,4], and in our dataset, that is followed by interneurons and then glial cell types. The same trend is present for each of the major RNA biotypes across each dataset. Layer 5 Extratelencephalic neurons (L5ET), are a rare pyramidal neuronal cell type that projects to subcortical regions, L5ETs account for ~0.30% of cells detected in PFC and PCC and 0.17% of cells detected in the ROSMAP PFC. L5ETs have the largest transcriptomic diversity, including the highest number of lncRNAs, protein-coding genes and pseudogenes across the PFC, ROSMAP PFC and PCC datasets. Notably, we did not detect L5ETs in the OC. We have also found a number of lncRNAs that may be cell types and disease state dependent. We have also worked with our collaborator, Yanhong Shi, to identify cell changes in microglia and astrocytes in subjects with AD and different ApoE genotypes as well as different CLU genotypes. Papers are currently being drafted.