snRNA data of Brain section from Parkinson Mouse Model based on inducible expression of human a-syn constructs

Using 6-months, 16-months, and 24-months old mice of a inducible expression of human a-syn constructs based Parkinson mouse model, we produced a single nucleus RNA dataset by cutting 0mm Bregma to -5mm Bregma. The Chromium 3’ Single Cell Library Kit (10x Genomics) was used and Sequencing was performed on a NovaSeq 6000. 

Paired 150bp snRNA-seq was performed using the 10X Genomics Gene Expression (GEX) 3´protocol with an Illumina 9000 sequencer. For the alignment of reads, a custom reference was created by adding the sequences of the S1/S2 transgene and the CamkIIa promoter to the mm10 mouse reference genome. Count matrices were obtained using the cellranger count 7.1 pipeline, including introns. Six samples were mapped using the bwUni2.0 High-Performance Computing infrastructure.

The unfiltered count matrices were loaded into R and corrected for ambient mRNA using SoupX 1.6.0 with default settings, adjusting “tfidfMin” settings between 0.9 and 1.3 depending on the sample. Seurat objects were created for each sample and subsequently merged. Cells were filtered out based on the following criteria: number of unique molecular identifiers (nUMI) < 2500, number of genes (nGene) < 1500, mitochondrial gene percentage > 3%, ribosomal gene percentage > 1.5%, or log10(Genes/nUMI) < 0.85. Subsequently, doublets and sex-doublets were removed using scDblFinder 1.16.0 and cellXY 0.99.0.

Normalization was performed using the SCTransform function on 4000 variable features with glmGamPoi method implemented in Seurat 5.0.1, and top 50 embeddings were obtained via scVI (scvi-tools 1.1.1) integration for sex, age, batch, and number of pooled animals. Clustering was done using the Leiden algorithm and visualized with Uniform Manifold Approximation and Projection algorithm (UMAP). Clusters represented by few samples, less than 100 cells, or a single batch, and not of conditions of interest, were removed. Clusters driven by ribosomal or mitochondrial genes, as well as markers of hindbrain and olfactory cell types, were also discarded. The steps from normalization onward were repeated until no further clusters needed removal. Final integration was performed using harmony 1.2.0 with an integration diversity penalty (theta) of 2, followed by final clustering based on the top 30 harmony components and UMAP visualization. Each subsequent clustering for annotation of sub-cell types was computed following the same procedure.

Clusters were annotated in a hierarchical manner using literature, the Mouse Brain Atlas (mousebrain.org), the Human Protein Atlas, and markers identified via the FindConservedMarkers function in Seurat. First, neurons and non-neuronal cells were distinguished using mainly canonical markers, such as but not limited to Rbfox3 (neurons), Mbp (oligodendrocytes), Acsbg1 (astrocytes), Pdgfra (oligodendrocyte precursor cells), Cx3cr1 (microglia), Colec12 (vascular cells), and Ttr (choroid plexus cells). Neurons were further classified into Vglut1, Vglut2, GABA, cholinergic, and dopaminergic neurons. Vglut1 and GABA neurons were further sub-annotated.