A method of separating extracellular vesicles from blood shows potential clinical translation, and reveals extracellular vesicle cargo gremlin-1 as a diagnostic biomarker

Abstract

Extracellular vesicles (EVs) have potential as minimally invasive biomarkers. However, the methods most commonly used for EV retrieval rely on ultracentrifugation, are time-consuming, and unrealistic to translate to standard-of-care. We sought a method suitable for EV separation from blood that could be used in patient care. Sera from breast cancer patients and age-matched controls (n = 27 patients; n = 36 controls) were analysed to compare 6 proposed EV separation methods. The EVs were then characterised on 8 parameters. The selected method was subsequently applied to independent cohorts of sera (n = 20 patients; n = 20 controls), as proof-of-principle, investigating EVs' gremlin-1 cargo. Three independent runs with each method were very reproducible, within each given method. All isolates contained EVs, although they varied in quantity and purity. Methods that require ultracentrifugation were not superior for low volumes of sera typically available in routine standard-of-care. A CD63/CD81/CD9-coated immunobead-based method was most suitable based on EV markers' detection and minimal albumin and lipoprotein contamination. Applying this method to independent sera cohorts, EVs and their gremlin-1 cargo were at significantly higher amounts for breast cancer patients compared to controls. In conclusion, CD63/CD81/CD9-coated immunobeads may enable clinical utility of blood-based EVs as biomarkers.

Keywords: Clinical utility; Extracellular vesicles; Gremlin-1; breast cancer; standard-of-care.

Fig. 1

Study design overview BC patients’ and healthy controls’ sera pools were prepared to establish if an optimal method for clinical utility could be identified, when comparing six proposed EV separation methods. Following extensive characterisation of EVs, the selected method was applied to an independent cohort of sera samples (n = 20 from patients; n = 20 from controls). Estimates of EVs amounts, and a cargo protein of interest, were evaluated.

Fig. 2

Immunoblotting for EV biomarkers and potential albumin contamination and ELISAs for potential ApoB contamination EV biomarkers and albumin contamination in EVs from the traditional DIFF-UC method, n = 3 separations (A). Representative immunoblots following the other 5 methods i.e. controls (B) and BC (C), summarised in (D). ELISA of ApoB contamination of controls (E) and BC (F) EVs. Hs578Ts(i)₈ cell lysate was the control. Graphs are mean of n = 3±SEM experiments. One-way ANOVA; *P<0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Fig. 3

Particle concentration, size and protein concentration of EVs Concentration (particles/mL) (A, B), size (C, D), and protein quantification (E, F) of EVs of controls and patients, respectively. Graphs represent mean±SEM of n = 3 independent experiments. One-way ANOVA was used as statistical test. *P < 0.05, ***P < 0.001, ****P < 0.0001.

Fig. 4

TEM analysis of separated EVs TEM analysis of EVs from control (top panel) and patient (bottom panel) sera using DIFF-UC (A), PEG (B), NBI (C), SEC (D), Stemcell (E), and Miltenyi (F) methods. Scale bars are 500 nm (top images of each panel) and 100 nm (bottom images of each panel).

Fig. 5

Characteristics of the methods and resulting EVs A summary of the results of each of the 8 characteristics used to compare the 6 separation methods. The results are qualitatively graded by low (+), moderate (++), high (+++). Particle concentration and size are divided into two columns, representing control and patients separately.

Fig. 6

Gremlin-1 in EVs separated using Stemcell method EVs from sera of BC patients (n = 20) and age-matched controls (n = 20) were separated using the Stemcell kit. Protein concentration of lysed EVs (A). Gremlin-1 content of lysed EVs (B). Graphs represent mean±SEM of n = 3 independent experiments. µg/mL reports the amount of EV protein per millilitre (mL) of lysed EV suspension. Paired t-test was used as statistical test. *P < 0.05.