Cell-specific RNA quantification in human SN DA neurons from heterogeneous post-mortem midbrain samples by UV-laser microdissection and RT-qPCR

Abstract

Cell specificity of gene expression analysis is from particular relevance when the abundance of target cells is not homogeneous in the compared tissue samples, like it is the case, e.g., when comparing brain tissues from controls and in neurodegenerative disease states. While single-cell gene expression profiling is already a methodological challenge per se, it becomes even more prone to artifacts when analyzing individual cells from human post-mortem samples. Not only because human samples can never be matched as precisely as those from animal models, but also, because the RNA-quality that can be obtained from human samples usually displays a high range of variability. Here, we detail our most actual method for combining contact-free UV-laser microdissection (UV-LMD) with reverse transcription and quantitative PCR (RT-qPCR) that addresses all these issues. We specifically optimized our protocols to quantify and compare mRNA as well as miRNA levels in human neurons from post-mortem brain tissue. As human post-mortem tissue samples are never perfectly matched (e.g., in respect to distinct donor ages and RNA integrity numbers RIN), we refined data analysis by applying a linear mixed effects model to RT-qPCR data, which allows dissecting and subtracting linear contributions of distinct confounders on detected gene expression levels (i.e., RIN, age). All these issues were considered for comparative gene expression analysis in dopamine (DA) midbrain neurons of the Substantia nigra (SN) from controls and Parkinson’s disease (PD) specimens, as the preferential degeneration of SN DA neurons in the pathological hallmark of PD. By utilizing the here-described protocol we identified that a variety of genes—encoding for ion channels, dopamine metabolism proteins, and PARK gene products—display a transcriptional dysregulation in remaining human SN DA neurons from PD brains compared to those of controls. We show that the linear mixed effects model allows further stratification of RT-qPCR data, as it indicated that differential gene expression of some genes was rather correlated with different ages of the analyzed human brain samples than with the disease state.