Suitability of RNA for NGS analysis after co-extraction with metabolites from limiting biological samples.

Availability of biological material is often a limiting factor in generating robust, multi-omics datasets. Additionally, regional tissue heterogeneity obscures the ability to directly compare adjacent regions. This is particularly true of mass-spectrometry based metabolomics where analytes can't be amplified. Milligram quantities of tissue are needed to extract sufficient quantities of metabolites to detect important, lower abundant metabolites. Researchers often pool several individual samples or compare datasets from different animals for RNAseq and metabolomics, lowering experimental power and creating batch effects. We compared RNA quality and quantity, RNAseq expression profiling, and dispersion metrics among mouse liver samples 2 days post injection with Lymphocytic choriomeningitis virus (LCMV) or vehicle (VEH). The LCMV model was selected to provide high biological contrast with which to assess RNAseq results as a function of extraction method. Frozen liver samples were pulverized in liquid nitrogen and split into 2 aliquots, one for RNA extraction (RNA) and one for sequential metabolite extraction (80% MeOH) followed by RNA extraction (metRNA). This allowed for intra-individual correlations between RNA and metRNA. RNA quantity and integrity measurements were not statistically significant between metRNA and RNA, and the top differentially expressed genes between VEH and LCMV were identical. This data suggests that methanol extraction does not influence RNA quality, integrity, or gene expression analysis. We observed that metRNA was associated with increased intra-replicate dispersion. This suggests that prior metabolite extraction may increase RNAseq variability and have deleterious effects on statistical power. Co-extraction of metabolites and RNA is viable for multi-omics of a single sample, but consideration of loss of statistical power due to increased variability should be considered.