Extracellular vesicle isolation for small RNA analysis: a methodological review.

Extracellular vesicles (EVs) are lipid-membrane bound, cell-derived nanoparticles secreted by all cell types under physiological and pathological conditions. They are found in most biological fluids including plasma, serum, urine, saliva, and cerebrospinal fluid (CSF). EV transport packages of proteins, lipids, and nucleic acids including microRNAs that reflect the physiological state of their cell of origin. As a result, EVs have emerged as powerful vehicles for biomarker discovery and liquid biopsy.

However, isolating EVs in a form compatible with downstream next-generation sequencing (NGS), particularly for small RNA analysis, remains a significant technical challenge. In this review, we list and evaluate major EV isolation techniques, their application to various biofluids, and their compatibility with small RNA NGS.

Ultracentrifugation (UC)

UC is a traditional method involving sequential centrifugation steps to remove cells and debris before pelleting EVs at ~100,000 x g. UC remains widely used for cell culture media and CSF. Michel et al.¹ demonstrated successful recovery of EVs for small RNA-seq from CSF using this approach.

Input volume: 5-50 mL of biofluid or culture supernatant
Typical yield: ~10⁸–10⁹ EV particles; ~50–200 ng RNA.

Density gradient centrifugation

Layering a sucrose or iodixanol density gradient post-UC enhances purity. Welsh et al.² and Buschmann et al.³ applied this method successfully to plasma and urine samples, resulting in high-quality small RNA profiles.

Input volume: ~1–5 mL of plasma or urine.
Typical yield: ~10⁸ EV particles; ~50–150 ng RNA.

Size exclusion chromatography (SEC)

SEC separates EVs by size using porous resin-packed columns. It preserves EV structure and efficiently separates them from soluble proteins. Gaspar et al.⁴ showed excellent miRNA profiling from plasma using qEV columns. Exo-spin™ kits from Cell Guidance Systems⁵, which combine precipitation and SEC, have been used in hundreds of EV studies and have shown consistent results in plasma and CSF for small RNA profiling.

Input volume: 0.5–1 mL of plasma/serum; up to 5 mL of CSF.
Typical yield: ~10⁹ EV particles; 20–100 ng RNA.

Precipitation-based methods

Polymer-based kits like ExoQuick® offer simplicity and speed but co-purify large amounts of non-exosomal proteins and other material, as well as carried-over precipitant. Wang et al.⁶ found significant biases in small RNA profiles from plasma and urine using precipitation. Boulestrau et al.⁷ successfully used this approach on saliva.

Input volume: 0.5–1 mL of biofluid
Typical yield: ~108 EVs (variable count); 50–200 ng RNA.

Ultrafiltration and Tangential Flow Filtration (TFF)

These membrane-based approaches concentrate and purify EVs by size. Jia et al.⁸ showed that TFF can isolate intact EVs from large volumes of urine, yielding clean small RNA suitable for sequencing.

Input volume: 10–100 mL of urine or conditioned media
Typical yield: ~109–1010 EV particles; 200–500 ng RNA.

Immunoaffinity captur

Using antibodies against EV surface proteins (CD63, CD9, CD81), this method enables highly specific capture.

Input volume: 0.1–0.5 mL of biofluid
Typical yield: ~107–108 EV particles; 10–50 ng RNA.

Microfluidic platforms

Emerging microfluidic devices allow EV capture from small sample volumes with minimal processing. Using this approach Boulestrau et al.⁷ demonstrated improved consistency and small RNA yield from saliva.

Input volume: 10–200 μL of biofluid
Typical yield: ~106–107 EV particles; 5–20 ng RNA.

Acoustic and dielectrophoretic separation

These label-free techniques separate EVs using acoustic waves or electric fields. Lin et al.⁹ used this approach to isolate EVs from whole blood, preserving their native RNA cargo.

Input volume: 0.5–2 mL of whole blood
Typical yield: ~108 EV particles; 50–150 ng RNA.
 

Table 1. Summary of EV isolation methods

Future developments

The ability to isolate and characterize EV subpopulations based on their protein composition is driving the next wave of innovations. Platforms like NanoView utilize interferometric imaging and fluorescent tagging to detect and phenotype individual EVs. ExoView further extends this concept by immobilizing EVs on chip surfaces with antibody capture, enabling multiplexed analysis of surface markers and associated RNA content. Immunomagnetic Capture with barcoded beads and mass cytometry is enabling scalable, high-specificity isolation workflows. These advances will make it possible to integrate EV surface profiling with transcriptomic data at the single-vesicle level, which can be a game-changer for mechanistic studies.

Conclusion

No one-size-fits-all approach exists for EV isolation. Method selection must consider sample type, throughput needs, and the analytical endpoint. For small RNA sequencing , SEC-based kits like Exo-spin™, density gradients, and immunoaffinity methods offer the highest purity. Continued advances in isolation specificity, especially through protein marker-based subpopulation sorting, will open new frontiers in functional EV genomics.