Optimal isolation of extracellular vesicles from pleural fluid and profiling of their microRNA cargo

Abstract

Pleural effusion occurs in both benign and malignant pleural disease. In malignant pleural effusions, the diagnostic accuracy and sensitivity of pleural fluid cytology is less than perfect, particularly for the diagnosis of malignant pleural mesothelioma, but also in some cases for the diagnosis of metastatic pleural malignancy with primary cancer in the lung, breast or other sites. Extracellular vesicles (EVs) carry an enriched cargo of microRNAs (miRNAs) which are selectively packaged and differentially expressed in pleural disease states. To investigate the diagnostic potential of miRNA cargo in pleural fluid extracellular vesicles (PFEVs), we evaluated methods for isolating the extracellular vesicle (EV) fraction including combinations of ultracentrifugation, size-exclusion chromatography (SEC) and ultrafiltration (10 kDa filter unit). PFEVs were characterized by total and EV-associated protein, nanoparticle tracking analysis and visualisation by transmission electron microscopy. miRNA expression was analyzed by Nanostring nCounter® in separate EV fractions isolated from pleural fluid with or without additional RNA purification by ultrafiltration (3 kDa filter unit). Optimal PFEV yield, purity and miRNA expression were observed when PFEV were isolated from a larger volume of pleural fluid processed through combined ultracentrifugation and SEC techniques. Purification of total RNA by ultrafiltration further enhanced the detectability of PFEV miRNAs. This study demonstrates the feasibility of isolating PFEVs, and the potential to examine PFEV miRNA cargo using Nanostring technology to discover disease biomarkers.

Keywords: exosomes; extracellular vesicles (EVs); microRNA (miRNAs); microvesicles; nanostring; pleural fluid.

FIGURE 1

Method matrices for isolating pleural fluid extracellular vesicles from cell‐free pleural fluids. (a) Pleural fluid was spun at 600–800 x g, resulting in cell‐free pleural fluid (CFPF). CFPF was processed to generate MV–EXO, pMV and pEXO fractions (Preparations A—H). The table lists input volumes of pleural fluid used for eight different EV preparations employing ultracentrifugation, size‐exclusion chromatography and/or ultrafiltration steps. (b) Total protein yield (μg) in EV fractions prepared from four pleural fluid samples (cases PF6238, PF6290, PF6466 and PF6468) using eight different methods for EV isolation.

FIGURE 2

Western blotting for comparison of PFEV isolation preparations. (a) The expression of Albumin, FLOT1 and CD9 in PFEV recovered from Preparations A—H in samples PF6238 and PF6290. 2 μg protein was loaded per lane, except for lane 5 (0.7 μg –due to technical error) and lane 7 (1.4 μg—insufficient sample). (b) Expression of Albumin, FLOT1 and CD9 in PFEV of PF7432 (Lanes 1, 2), PF6129 (Lanes 3, 4) and PF 21153 (lanes 5, 6). All lanes were loaded with 2ug total protein. Lanes 1, 3, and 5 represent PFEV prepared by SEC–UF from 0.5 mL CFPF, while lanes 2, 4, and 6 were prepared by UC–SEC–UF from 4 ml CFPF. Table on the right: The input volume of cell‐free pleural fluid for PFEV isolation and the resulting protein yield were presented.

FIGURE 3

PFEV characterisation by nanoparticle tracking analysis and transmission electron microscopy. (a) NTA: PFEV of PF6129 isolated by UC alone showed 5.94 × 10⁸ ± 1.14 × 10⁷ particles per mL final EV volume, whereas UC‐SEC yielded 1.21 × 10⁸ ± 2.93 × 10⁶ particles per mL. The 10^th, 50^th and 90^th percentiles of the averaged EV particle sizes were 94.9 nm, 130.3 nm and 214.5 nm for UC, whereas 100.9 nm, 145.0 nm and 229.9 nm for UC‐SEC. (b) TEM: PFEV of PF6238 and PF6246 isolated by UC–UC or UC–SEC–UC preparation were evident as indicated within the red boxes. Images of UC–SEC–UC derived PFEVs show well‐defined cup‐shaped particles of varying sizes, with lower background in comparison with corresponding samples isolated by UC–UC.

FIGURE 4

Heatmap of miRNA expression in PF7278; and PF6066, PF6238, PF6246 and PF7045. (a) The fold change in miRNA raw count values between sample d and e was presented. (b) 275 miRNAs were presented at Log₁₀ transformed raw count expression value. Higher expression of most miRNAs was observed with the application of SEC, RNA purification, more significantly by increasing the volume of CFPF used in PFEV isolation. (c) Heatmap of miRNA expression at log₁₀ transformed raw count expression value. PFEV of four pleural fluid cases (PF6066, PF6238, PF6246, PF7045) were isolated by Method 1 (UC–qEV–UC, RNAPur from 5 mL CFPF input volume) or Method 2 (UC–UC from 16 mL CFPF input volume). The number of miRNAs with raw count values < 50 or ≥ 50 was presented.

FIGURE 4

Heatmap of miRNA expression in PF7278; and PF6066, PF6238, PF6246 and PF7045. (a) The fold change in miRNA raw count values between sample d and e was presented. (b) 275 miRNAs were presented at Log₁₀ transformed raw count expression value. Higher expression of most miRNAs was observed with the application of SEC, RNA purification, more significantly by increasing the volume of CFPF used in PFEV isolation. (c) Heatmap of miRNA expression at log₁₀ transformed raw count expression value. PFEV of four pleural fluid cases (PF6066, PF6238, PF6246, PF7045) were isolated by Method 1 (UC–qEV–UC, RNAPur from 5 mL CFPF input volume) or Method 2 (UC–UC from 16 mL CFPF input volume). The number of miRNAs with raw count values < 50 or ≥ 50 was presented.