Small and long RNA transcriptome of whole human cerebrospinal fluid and serum as compared to their extracellular vesicle fractions reveal profound differences in expression patterns and impacts on biological processes.

BACKGROUND: Next generation sequencing (NGS) of human specimen is expected to improve prognosis and diagnosis of human diseases, but its sensitivity urges for well-defined sampling and standardized protocols in order to avoid error-prone conclusions. METHODS: In this study, large volumes of pooled human cerebrospinal fluid (CSF) were used to prepare RNA from human CSF-derived extracellular vesicles (EV) and from whole CSF, as well as from whole human serum and serum-derived EV. In all four fractions small and long coding and non-coding RNA expression was analyzed with NGS and transcriptome analyses. RESULTS: We show, that the source of sampling has a large impact on the acquired NGS pattern, and differences between small RNA fractions are more distinct than differences between long RNA fractions. The highest percentual discrepancy between small RNA fractions and the second highest difference between long RNA fractions is seen in the comparison of CSF-derived EV and whole CSF. Differences between miR (microRNA) and mRNA fractions of EV and the respective whole body fluid have the potential to affect different cellular and biological processes. I.e. a comparison of miR in both CSF fractions reveals that miR from EV target four transcripts sets involved in neurobiological processes, whereas eight others, also involved in neurobiological processes are targeted by miR found in whole CSF only. Likewise, three mRNAs sets derived from CSF-derived EV are associated with neurobiological and six sets with mitochondrial metabolism, whereas no such mRNA transcript sets are found in the whole CSF fraction. We show that trace amounts of blood-derived contaminations of CSF can bias RNA-based CSF diagnostics. CONCLUSIONS: This study shows that the composition of small and long RNA differ significantly between whole body fluid and its respective EV fraction and thus can affect different cellular and molecular functions. Trace amounts of blood-derived contaminations of CSF can bias CSF analysis. This has to be considered for a meaningful RNA-based diagnostics. Our data imply a transport of EV from serum to CSF across the blood-brain barrier.