Highlights
- •Intersecting eCLIP and exRNA dataset predicts extracellular RBPs in human biofluids
- •Deconvolution of RNA bound by RBPs associates RBPs with exRNA carrier classes
- •Generated RNA-RBP coverage map of more than 6,900 biofluid samples across whole human genome
- •Differential release of extracellular RBPs and their isoforms across different cell types
Summary
Although the role of RNA binding proteins (RBPs) in extracellular RNA (exRNA) biology is well established, their exRNA cargo and distribution across biofluids are largely unknown. To address this gap, we extend the exRNA Atlas resource by mapping exRNAs carried by extracellular RBPs (exRBPs). This map was developed through an integrative analysis of ENCODE enhanced crosslinking and immunoprecipitation (eCLIP) data (150 RBPs) and human exRNA profiles (6,930 samples). Computational analysis and experimental validation identified exRBPs in plasma, serum, saliva, urine, cerebrospinal fluid, and cell-culture-conditioned medium. exRBPs carry exRNA transcripts from small non-coding RNA biotypes, including microRNA (miRNA), piRNA, tRNA, small nuclear RNA (snRNA), small nucleolar RNA (snoRNA), Y RNA, and lncRNA, as well as protein-coding mRNA fragments. Computational deconvolution of exRBP RNA cargo reveals associations of exRBPs with extracellular vesicles, lipoproteins, and ribonucleoproteins across human biofluids. Overall, we mapped the distribution of exRBPs across human biofluids, presenting a resource for the community.
Introduction
Extracellular RNAs (exRNAs) are RNAs that exist in biofluids outside of cells and can be associated with a variety of carriers that includes vesicles, lipoproteins, and free ribonucleoproteins (RNPs; RNA-binding proteins plus the RNA bound to them).1 Because exRNAs offer insights into the cell from which they originate, they provide a bounty of information that can be leveraged for liquid biopsy detection and tracking of normal and disease processes. Previously, in an effort to compare data across studies, which is often confounded by lab-specific variables, deconvolution revealed 6 major cargo types (CTs) that could be related to high- and low-density vesicles (HDVs and LDVs, respectively), lipoprotein particles (LPPs), and RNPs.2 However, specific RNA-binding proteins (RBPs) carrying exRNA (exRBPs) were not identified, with the exception of AGO2. AGO2 was associated with CT3A and CT3B CTs by sequencing AGO2 pull-downs followed by correlating the RNAs in the AGO2 profile with RNAs in the CT3A and CT3B profiles. In contrast, despite CT3C also likely being associated with an RNP carrier type, it showed little correlation with any reference CT tested, opening the question about the specific exRNA carrier involved. Given that there are over 2,000 RBPs in human cells,3 and three of six CTs (CT3A, CT3B, and CT3C) are related to RNPs, we concluded that the extent to which RBPs serve as exRNA carriers was worth exploring. This major gap in knowledge limited our understanding of the role of RBPs in exRNA biology and their potential as liquid biopsy biomarkers that bind and export exRNA fragments of human genes into accessible human biofluids.
A major role of RBPs as exRNA carriers would be expected based on their role in the biogenesis, localization, loading, and transport of exRNA.4,5,6,7,8,9,10 Generally, exRNA research has focused largely on vesicular exRNAs because of the difficulties associated with identifying the diversity of RBPs (∼2,000) that can potentially serve as carriers of exRNA. While mass spectrometry experiments led to the discovery of RNP structures like exomeres and supermeres,11,12 the specific RBPs involved in these structures and their exRNA cargo remain unknown. New opportunities to address these questions became possible following the release of enhanced crosslinking and immunoprecipitation (eCLIP) profiles as part of the ENCODE project. These eCLIP profiles allowed precise detection of binding sites of 150 RBPs across the human genome.13 While these results reflect intracellular RBP-RNA interactions, we hypothesized that similar interactions may be observable extracellularly as well. By combining the cellular eCLIP profiles from ENCODE with human small RNA sequencing (RNA-seq) from different biofluids in the exRNA Atlas, it became possible to determine which RBPs play a role as exRNA carriers.
The most direct combined analysis approach to identify an exRBP is to ask whether exRNA profiles in the Atlas contain a significant amount of exRNA fragments in the regions predicted by eCLIP to bind with the RBP. While informative, this direct method alone has limited power to establish an association of a specific exRNA fragment with a specific RBP in extracellular space. Moreover, we assumed that the power to detect a footprint at the level of an RBP could be enhanced because the relative amounts of any RBPs and their cargo vary from sample to sample because of biological and experimental variables. Specifically, we reasoned that, when the expression levels of exRNA fragments carrying the binding signature of the same RBP covary across diverse conditions, that “correlation footprint” indicates a specific interaction between that RBP and those exRNA fragments. The interaction might either be direct, as in an extracellular RNP complex, or indirect, resulting from a role of the RBP in the export of the exRNAs into extracellular space. By combining evidence from individual fragments, we predicted that we could obtain evidence about any possible role of the RBP as an exRNA carrier. We applied this “correlation footprint” strategy to develop a map of candidate exRBPs and their exRNA cargo.
Here, we present analyses of this exRBP map resource as well as its experimental validation. Extending correlation analysis to specific RNA biotypes, we identified biotypes carried by the exRBPs, placed the exRBPs within specific human biofluids, and associated them with different classes of extracellular vesicles (EVs), lipoproteins, and RNP particles. We also identified exRBPs within different classes of EVs produced by red blood cells (RBCs). Cognizant of the vast opportunities for further exploration of exRBP biology and their potential as disease biomarkers, we offer an exRNA Atlas resource with exRBP coverage information for all 150 eCLIP-profiled RBPs available for all 6,930 human small RNA-seq datasets.
Results
Extending the exRNA Atlas resource with RBP exRNA profiling information
To identify candidate exRNA fragments carried by RBPs, we analyzed ENCODE eCLIP data for 150 RBPs (as described in STAR Methods). Guided by the hypothesis that RNA bound inside cells would also be found in the biofluids and are identifiable in exRNA profiles, we intersected eCLIP RBP binding sites from ENCODE with human small RNA-seq data derived from cerebrospinal fluid (CSF), plasma, saliva, serum, and urine from the exRNA Atlas (Figures 1A–1C).
