Extracellular Vesicle Purity: Particle-to-Protein Ratios

By Benn Bluestein-Veyra, M.Sc. Organic Chemistry · Updated 20 October 2026

Extracellular vesicle purity describes how much of a preparation consists of vesicles and how much consists of co-isolated free protein, protein aggregates and lipoproteins. The field most often scores it with ratios between bulk measurements, led by the particle-to-protein ratio Webber and Clayton proposed in 2013.^[1][2]

This page covers how those ratios are built, what raises and lowers them, which co-isolates dominate blood and culture media, how isolation methods change the result, and what MISEV2023 asks laboratories to report.^[3]

Extracellular vesicle purity is expressed as a ratio of bulk measurements

A vesicle preparation is a mixture of lipid-bilayer vesicles and non-vesicular extracellular particles that share their size, density or composition. MISEV2023 notes that these particles can greatly outnumber vesicles in biological matrices, so most separation methods recover both.^[3]

MISEV2018 proposed that ratios between quantification methods give useful measures of purity. It named the protein to particle, protein to lipid and RNA to particle ratios as candidate metrics, and noted that their transfer across different assays was still being established.^[2]

The same guidelines grouped separation methods by recovery and specificity. Polymer precipitation and lengthy ultracentrifugation recover most extracellular material, while density gradients and affinity capture recover fewer components with higher specificity.^[2]

The particle-to-protein ratio scores purity with two routine assays

Webber and Clayton divided the particle count from nanoparticle tracking analysis by the protein concentration from a micro BCA assay, giving particles per microgram of protein (Fig. 1). They proposed it as a routine quality check that any laboratory could apply to every preparation.^[1]

They tested the ratio by adding particle-free bovine serum albumin to a purified preparation. Protein rose while particle counts held steady, and a halving of the ratio corresponded to roughly 40 to 50 percent more non-vesicular protein in their assay.^[1]

From their culture and biofluid samples the authors proposed three ranges. Ratios above 3 × 10¹⁰ particles per microgram indicated high purity, 2 × 10⁹ to 2 × 10¹⁰ low purity, and ratios below 1.5 × 10⁹ an impure preparation.^[1]

• Unpurified medium from standard culture flasks gave about 1 × 10⁷ particles per microgram; pelleting raised it about 60-fold and a wash a further 4.6-fold.^[1]
• Flotation on a sucrose cushion gave the highest values in the study, approaching 3.4 × 10¹⁰ particles per microgram.^[1]
• Pellet-and-wash preparations from fresh serum and urine reached about 9.7 × 10⁸, around 34-fold below the sucrose cushion value.^[1]
• Medium with vesicle-depleted fetal bovine serum alone gave about 1 × 10⁶, which the authors set as the background of non-cell-derived particles.^[1]

The ratio also reflects how many particles the source contains. Serum gave higher ratios than urine despite its far higher protein content, because unpurified urine held about 800-fold fewer particles than serum.^[1]

The authors stated two assumptions behind the method. It treats every counted particle as a vesicle, and it assumes each vesicle carries a comparable amount of protein.^[1]

Lipid measurements add a membrane-specific denominator

Every vesicle has a lipid bilayer, so total lipid offers a measure tied to membranes. Osteikoetxea and colleagues adapted the sulfo-phospho-vanillin assay to a 96-well format and reported protein to lipid ratios for apoptotic bodies, microvesicles and exosomes from several cell lines and plasma.^[4]

Each subpopulation had a characteristic ratio, with protein to lipid rising as vesicle diameter increased. A high-speed pellet with an abnormally high ratio contained no vesicular structures by electron microscopy, and washing red cell microvesicles brought an elevated ratio back to the expected value by the third wash.^[4]

Visnovitz and colleagues later replaced organic-solvent lipid standards with an aqueous liposome standard, which raised sensitivity about tenfold. They reported that lipid-based measurements of vesicles can be more reliable than protein-based ones.^[5]

The same group related lipid to particle measurements. They calculated total membrane surface per milliliter from resistive pulse sensing diameters and counts, set it against measured lipid, and proposed pairing the lipid assay with protein or particle measurements.^[5]

Mihály and colleagues derived a spectroscopic protein to lipid ratio from the amide and CH stretching bands of infrared spectra. The ratio distinguished vesicle subpopulations and also carried information on aggregated proteins, lipoproteins and immune complexes in the sample.^[6]

Ghosal and colleagues applied the same reasoning to non-vesicular extracellular particles. These particles had significantly higher protein to lipid ratios than vesicle subtypes and were enriched in cholesteryl esters and triacylglycerols.^[7]

Lipoproteins are the principal particulate co-isolates of blood

Lipoproteins overlap with blood vesicles in size, density or both (Fig. 2). MISEV2023 notes that every class except HDL overlaps vesicles in size, HDL overlaps them in density, and some lipoprotein subtypes outnumber blood vesicles by orders of magnitude.^[3][8]

Sódar and colleagues found by flow cytometry that most particles in the vesicle size range in platelet-free plasma lacked common vesicle markers. They identified most of them as lipoproteins, predominantly LDL, which co-purified with vesicles and associated with them on mixing in vitro.^[9]

Yuana and colleagues showed the density side of the overlap. HDL isolated between 1.063 and 1.21 g/mL shares the 1.13 to 1.19 g/mL density of plasma vesicles, and electron microscopy of that fraction found about one vesicle for every hundred HDL particles.^[10]

Karimi and colleagues summarized the size side. Chylomicrons range from 75 to 1200 nm and VLDL from 30 to 80 nm, both overlapping vesicles, while IDL, LDL and HDL fall within 5 to 35 nm.^[11]

The structural proteins of these particles serve as MISEV category 3 purity markers.^[3] Apolipoprotein A-I is the major protein of HDL (Fig. 3a),^[12] and apolipoprotein B100 wraps the LDL particle as a continuous beta-sheet belt beside a globular N-terminal domain (Fig. 3b).^[13][14]

Lipoproteins enter the ratios in two ways. Particles in the counting range add to the numerator, as Sódar and colleagues showed for LDL, and MISEV2023 notes that total lipid measurements may overestimate vesicles when lipoproteins are present.^[3][9]

Fig. 3 | Apolipoprotein A-I (residues 44 to 243) and Apolipoprotein B100 from human LDL.

Soluble proteins, aggregates and culture supplements shape the ratio

Blood carries high concentrations of albumin, immunoglobulins and fibrinogen, which are generally smaller and denser than vesicles.^[3] Karimi and colleagues described albumin as the main co-isolate of plasma vesicles prepared by ultracentrifugation, regardless of washing.^[11]

Protein aggregates can also enter particle counts, as Webber and Clayton and later Brennan and colleagues noted.^[1][15] Böing and colleagues reported aggregates co-isolated by differential centrifugation, and Linares and colleagues showed that high-speed centrifugation assembles vesicles of different phenotypes into mixed aggregates.^[16][17]

Some plasma proteins bind vesicles directly. Tóth and colleagues found that vesicles incubated in plasma acquired a protein corona that included apolipoproteins A-I, B, C-III and E, complement factors and immunoglobulin chains, which raised their density.^[18]

MISEV2023 concludes that some blood proteins and lipoproteins once defined as contaminants may be genuinely associated with vesicles and remain after gentle, rigorous separation. Their contribution to the protein term is then part of the vesicle itself.^[3][18]

In culture, fetal bovine serum supplies vesicles of its own. Shelke and colleagues found that an 18 h ultracentrifugation removed about 95 percent of RNA-containing serum vesicles, and that a 1.5 h protocol was insufficient.^[19]

MISEV2023 recommends processing and characterizing non-conditioned medium as a control, which shows how much the medium itself adds to vesicle measurements. Webber and Clayton used this control to set their background ratio.^[1][3]

The isolation method largely sets the ratio a preparation reaches

Lobb and colleagues compared four isolation methods on culture medium and plasma using resistive pulse sensing and protein assays. Polymer precipitation gave the highest particle yield and the lowest particle-to-protein ratio, while size-exclusion chromatography performed comparably to density gradient purification.^[20]

Böing and colleagues showed what size-exclusion chromatography removes. Fractions 9 to 12 held 61 percent of platelet-derived vesicles with under 5 percent of HDL and under 1 percent of protein, enriching vesicles 8-fold over HDL and 70-fold over protein.^[16]

Brennan and colleagues compared common methods on human serum. Size-exclusion columns and density gradient ultracentrifugation gave higher particle-to-protein ratios than the other methods, and plain ultracentrifugation gave a low ratio because of co-isolated protein.^[15]

In culture medium, Benedikter and colleagues found that ultrafiltration followed by size-exclusion chromatography tended to give a higher yield and vesicle-to-protein purity than ultracentrifugation.^[21] Kalra and colleagues reported that an iodixanol density gradient best removed abundant plasma proteins among three plasma methods.^[22]

Combining separation principles addresses lipoproteins (Fig. 4). Karimi and colleagues floated plasma vesicles on an iodixanol cushion before size-exclusion chromatography, which reduced lipoprotein particles about 100-fold and allowed 1187 proteins to be identified in plasma vesicles.^[11]

MISEV2023 asks for the measurements behind every ratio

MISEV2023 notes that total protein assays often overestimate vesicle concentration because of co-isolated protein, and that protein to particle ratios depend on the detection limits of each instrument. It therefore asks for absolute protein and particle concentrations alongside any reported ratio.^[3]

• Report the original measurements behind a ratio, together with the ratio.^[3]
• Report the lower concentration limit of detection of each assay.^[3]
• Keep protein concentrations within the linear range of the reported reference curve.^[3]
• State whether intact or disrupted vesicles were measured, including any detergent.^[3]
• Report output as particles unless upstream processing is highly specific for vesicles.^[3]

For lipid assays, the guidelines ask authors to state the detection limit and to consider how co-isolated particles such as lipoproteins influence the value.^[3] Category 3 proteins, including apolipoproteins and albumin, give the matching protein-level readout of purity.^[2][3]

The counting method defines the numerator of the ratio

A ratio is as comparable as the two assays that produce it. Bachurski and colleagues found that one tracking instrument reported bead concentrations 2.0 to 2.1-fold above their nominal values, while a second gave more accurate and repeatable vesicle concentrations.^[23]

Both instruments in that study reported peak vesicle diameters of 60 nm or larger, while electron microscopy and interferometric imaging resolved smaller vesicles.^[23] MISEV2023 notes that ratios depend on each instrument’s lower size limit, so they compare best within one counting method and setting.^[3]

The denominator varies with the assay as well. Vergauwen and colleagues found that a fluorometric assay measured dilute vesicle and albumin samples most accurately, and that recovery after ultrafiltration depended on the filter membrane.^[24]

Within one workflow, the ratio is a sensitive way to follow purity over time, and full reporting makes it interpretable between laboratories. Webber and Clayton anticipated that fluorescence-capable counting would refine the method by separating vesicles from other particles.^[1][3]

Where to go from here

The particle counting reference compares tracking analysis with resistive pulse sensing, the two instruments behind most numerators. The protein marker reference covers the MISEV categories, and the isolation methods reference describes the separations that set each ratio.

The MISEV2023 compliance and certificate of analysis references show how purity data are documented. Verified practices can request lot documentation, and our team is glad to discuss the methods described on this page.