Abstract
Single-nucleus RNA sequencing (snRNA-seq) enables resolving cellular heterogeneity in complex tissues by using nuclei instead of cells, overcoming limitations of single-cell RNA sequencing and enabling analysis of frozen and hard-to-isolate tissues. Despite advances in isolation techniques, systematic evaluations of their effects on nuclear integrity and subsequent data quality remain lacking, a critical gap with profound implications for rigor and reproducibility. To address this, we compared three mechanistically distinct nuclei isolation strategies with brain tissue: a sucrose gradient centrifugation-based method, a spin column-based method, and a machine-assisted platform. All methods captured diverse cell types but revealed considerable protocol-dependent differences in cell type proportions, transcriptional homogeneity, and the preservation of cell-state-specific markers. Moreover, workflows differentially influenced contamination levels from ambient, mitochondrial, and ribosomal RNAs, with the machine-assisted method exhibiting the highest overall data quality. Our findings establish nuclei isolation methodology as a critical experimental variable shaping snRNA-seq data quality and biological interpretation.