Transcriptome analysis of single-cell suspension from adult mouse brain tissue
RNA sequencing has been used to investigate multiple complex diseases, including PTSD, schizophrenia, and Alzheimer’s disease. However, the limits of homogenate RNA sequencing only allow for detecting drastic expression changes; a small subpopulation of cells exhibiting a change likely will not draw attention when placed against the noise of the rest of the sample. To combat this limitation, single-cell RNA sequencing is becoming widely-used in several applications, including cancer and tissue-specific analysis of RNA. Currently, very few established protocols exist on the dissociation of adult nervous tissue into viable single-cell suspensions, and all work only to varying degrees of success. This is primarily due to the extreme interconnectivity of brain tissue, widespread myelination, and dense extracellular matrix.
Our group examined two dissociation methods, mechanical dissociation via Dounce homogenization and the Miltenyi GentleMacs system, and determined their efficacy in single-cell RNA sequencing. Mouse cortex was rapidly dissected, and homogenized by both methods. Then, the samples were cleaned using a debris-removal kit and the mRNA of individual cells were barcoded using the 10x Chromium system. cDNA libraries were prepared and sequenced with Illumina’s MiSeq platform, then analyzed using the 10X Genomics Cell Ranger and a marker-consensus based cell type classifier.
We report that each method proved to be both informative in some aspects and limiting in others. Neither method provided a representative sample of the expected cell types in the brain. Dounce homogenization yielded an estimated successful cell capture count of 2627 with 25.7% mapping rate. With GentleMacs homogenization, 64.7% reads were mapped correctly to the transcriptome. However, the estimated successful cell capture count amount to only 227.
Both methods presented their own strengths and weaknesses. The Dounce homogenizer was more reliable for cell counting, however, a low signal to noise ratio indicated the presence of high background contamination. The gentleMacs cell suspension had a low number of cells detected, but this could be attributed to the difficulty in counting the suspension. Despite this weakness, the background noise was substantially lower than the dounce homogenizer sample. With improvements in the counting methods we predict that the gentleMacs will produce high quality single cell sequencing data.