Deep Profiling of EV Long RNAs Reveals Biofluid-Specific Transcriptomes and Splicing Landscapes

Abstract

RNA profiling of extracellular vesicles (EVs) from human biofluids has historically been limited to small RNA species, with long RNAs--such as mRNA exons and long non-coding RNAs--remaining largely underexplored. Moreover, the dominance of hematopoietic-derived EVs in complex fluids like plasma has posed significant challenges for detecting low-abundance, tissue-specific transcripts. Here, we establish foundational transcriptomic maps of long RNAs in EVs from plasma, urine, and cerebrospinal fluid (CSF) using ultra-deep whole transcriptome sequencing (WTS), revealing both fluid-specific and shared expression and splicing signatures. We then introduce a targeted RNA capture method that enriches for all protein-coding and long non-coding transcripts, dramatically enhancing sensitivity for gene and splice variant detection. Applying this approach to brain-specific transcripts, we achieve >85-fold enrichment of target gene expression and, on average, 3.1-fold increase in detected splice junctions per gene compared to untargeted WTS. As a proof of concept, we apply this brain-targeted RNA panel to EVs from plasma in a Parkinsons disease cohort of 40 plasma samples and compare its performance to exome sequencing as well as untargeted WTS. This work advances EV transcriptomics into the long RNA domain and establishes a framework for high-sensitivity, noninvasive biomarker profiling across tissues and diseases.