Comparative analysis of extracellular vesicle isolation methods from human AML bone marrow cells and AML cell lines

Abstract

Cellular crosstalk between hematopoietic stem/progenitor cells and the bone marrow (BM) niche is vital for the development and maintenance of myeloid malignancies. These compartments can communicate via bidirectional transfer of extracellular vesicles (EVs). EV trafficking in acute myeloid leukemia (AML) plays a crucial role in shaping the BM microenvironment into a leukemia-permissive niche. Although several EV isolation methods have been developed, it remains a major challenge to define the most accurate and reliable procedure. Here, we tested the efficacy and functional assay compatibility of four different EV isolation methods in leukemia-derived EVs: (1) membrane affinity-based: exoEasy Kit alone and (2) in combination with Amicon filtration; (3) precipitation: ExoQuick-TC; and (4) ultracentrifugation (UC). Western blot analysis of EV fractions showed the highest enrichment of EV marker expression (e.g., CD63, HSP70, and TSG101) by precipitation with removal of overabundant soluble proteins [e.g., bovine serum albumin (BSA)], which were not discarded using UC. Besides the presence of damaged EVs after UC, intact EVs were successfully isolated with all methods as evidenced by highly maintained spherical- and cup-shaped vesicles in transmission electron microscopy. Nanoparticle tracking analysis of EV particle size and concentration revealed significant differences in EV isolation efficacy, with exoEasy Kit providing the highest EV yield recovery. Of note, functional assays with exoEasy Kit-isolated EVs showed significant toxicity towards treated target cells [e.g., mesenchymal stromal cells (MSCs)], which was abrogated when combining exoEasy Kit with Amicon filtration. Additionally, MSC treated with green fluorescent protein (GFP)-tagged exoEasy Kit-isolated EVs did not show any EV uptake, while EV isolation by precipitation demonstrated efficient EV internalization. Taken together, the choice of EV isolation procedure significantly impacts the yield and potential functionality of leukemia-derived EVs. The cheapest method (UC) resulted in contaminated and destructed EV fractions, while the isolation method with the highest EV yield (exoEasy Kit) appeared to be incompatible with functional assays. We identified two methods (precipitation-based ExoQuick-TC and membrane affinity-based exoEasy Kit combined with Amicon filtration) yielding pure and intact EVs, also suitable for application in functional assays. This study highlights the importance of selecting the right EV isolation method depending on the desired experimental design.

Keywords: AML; EV isolation methods; bone marrow niche; extracellular vesicles; intercellular communication.

Figure 1

Schematic overview of EV isolation procedures. Extracellular vesicle (EV) isolation from MOLM-13 cells and human AML cells using different commonly used EV isolation methods. Cells were cultured for 4 days in vesicle-free FBS prior to harvesting conditioned medium. Samples were purified by two centrifugation steps and filtering through a 0.45-µm membrane. The purified supernatant was processed by four different isolation methods as depicted in the flowchart. d, days; h, hours; h-o, hands-on; min, minutes.

Figure 2

Western blot and transmission electron microscopy of MOLM-13-derived EVs. (A) Western blot analysis of MOLM-13 cells and MOLM-13-derived EVs (40 µg/lane) obtained after different isolation methods. As positive vesicle markers CD63, HSP70, and TSG101 were analyzed (⁺ indicates long, ⁺⁺ short exposure time). Cytochrome c and β-actin serve as non-EV markers. BSA (2 µg/lane) was used as purity control for non-EV co-isolating structures. The shown Western blot is a representative figure of n = 3 biological replicates from three independent experiments. (B) Transmission electron microscopy of MOLM-13-derived vesicles showed EV-sized cup-shaped structures for every isolation method. Scale bar in the right lower corner was set to 150 nm. The arrowheads display vesicle-like structures, whereas the arrow points out garland-like clotted formations.

Figure 3

Characterization of MOLM-13-derived EVs according to different isolation methods. Comparison of MOLM-13-derived EVs isolated with different isolation methods using nanoparticle tracking analysis (NTA) (A–C) and protein measurement (D, E). Histograms representing the mean particle size (A), total particle amount (B), and particle concentration (C) after exoEasy Kit (n = 4), exoEasy Kit + Amicon filtration (n = 4), ExoQuick-TC (n = 4), and UC (n = 3) isolation procedures. Histograms representing total protein amount (D) and protein concentration (E) after exoEasy Kit (n = 4), exoEasy Kit + Amicon filtration (n = 4), ExoQuick-TC (n = 4), and UC (n = 3) isolation procedures. Histograms representing the protein/particle ratio calculated by dividing the total protein amount by the total particle number from each paired sample (F) (n = 3 for UC, n = 4 for other isolation methods). *p<0.05, **p<0.01.

Figure 4

Characterization of primary AML-derived EVs according to different isolation methods. Nanoparticle tracking analysis of size (A), total particle number (B), and EV concentration (C) from primary AML-cell-derived EVs isolated by different isolation methods (n = 5 for exoEasy Kit, n = 6 for exoEasy Kit with filter units and ExoQuick-TC, and n = 3 for UC). Measurement of the total protein amount (D) and protein concentration (E) from primary AML EVs with different isolation methods (n = 3). *p<0.05, **p<0.01.

Figure 5

Assessment of EV isolation procedure compatibility with functional assays. (A) Histograms representing FACS viability assays of MOLM-13 (I) and EL08-1D2 cells (II) after treatment with different EV isolation buffers for 48 h (n = 5), and representative images of EL08-1D2 cells after 48 h treatment with the different isolation buffers (III). (B) FACS analysis of latex beads loaded with CD63-eGFP expressing MOLM-13-derived EVs following different isolation methods (n = 4). (C, D) Internalization assay of MOLM-13-derived EV (GFP^- EV) or CD63-eGFP expressing MOLM-13-derived EV (GFP⁺ EV) into target cells (EL08-1D2, n = 10 independent experiments; MSC, n = 4 different donors; HSPCs, n = 6 different donors) with or without additional heparin (hep) treatment. Quantification of GFP-positive cells (C) and corresponding confocal microscopy images (scale bar set to 20 µm) (D). *p<0.05, **p<0.01, ****p<0.0001.