Evaluation of freezing methods for cell preservation in scRNA-seq
Identifying potential neurodegenerative disease (NGD) biomarkers in cerebrospinal fluid (CSF) may allow NGD diagnosis of living patients, as opposed to confirming NGDs during patient autopsy. However, the sample size and statistical power required to search for biomarkers in CSF is difficult to achieve, because of the current requirement that fresh CSF be processed immediately; this, coupled with the rarity of donors, makes using 10X Genomics kits needlessly expensive and inefficient. This study aims to determine the viability of cells in frozen CSF to enable studies with larger sample sizes. A total of 54 CSF samples were collected from 26 people and batched into 6 experimental groups, each treated with a different cryopreservative: 11 fresh (control), 3 methanol (10X recommended), 8 CryoStor, 11 Recovery (commercial agents), 15 FBS+DMSO, and 6 FBS+DMSO+DNAse (homebrews). Samples were processed using the 10X Chromium 5’ kit, and the resulting scRNA-seq data was analyzed in Python using the scanpy toolkit. During quality control, red blood cells (>0.01 HB gene expression) and dead cells (>5% mitochondrial or <10% ribosomal gene expression) were removed. This revealed that a majority of cells preserved with methanol had died. Additionally, 7 out of 8 CryoStor samples were poor quality, so no conclusion may be reliably drawn about the effectiveness of this freezing method. However, the remaining preservation approaches (Recovery, FBS+DMSO, and FBS+DMSO+DNAse) maintained all cell types identified in fresh CSF. Further, t-tests suggest no significant differences exist between cell type distributions in fresh CSF and CSF treated with these freezing methods. Because person-to-person variation in CSF cell type distribution is a prominent finding, it is valuable to further investigate the medical history of persons sampled.