Vanderbilt Institute of Chemical Biology



Discovery at the VICB






A Revealing New Look at Extracellular Vesicles

By: Carol A. Rouzer, VICB Communications
Published:  May 2, 2019


Better methods of purification lead to dramatic new insights regarding the composition and function of extracellular vesicles and nanoparticles.


Cells release an array of extracellular vesicles of various shapes and sizes. Thought at first to be primarily a mechanism for removal of cellular debris, we now know that these vesicles  play a role in the transfer of molecules between cells and in intercellular communication. There are two main classifications of extracellular vesicles based on their mechanism of formation and release (Figure 1). Exosomes (~40 nm -150 nm in diameter) result from the invagination of endosomal membranes leading to the formation of intraluminal vesicles (ILVs) within multivesicular endosomes (MVEs). Release of ILVs to the extracellular environment as exosomes occurs upon fusion of the MVE with the plasma membrane. Microvesicles (~150 nm - 1000 nm in diameter) result through the outward budding of the plasma membrane. In addition to these vesicular structures, cells also release non-vesicular nanoparticles of an equally varied composition and function. Through extensive studies, researchers have identified a large number of proteins, lipids, and nucleic acid species in preparations of exosomes (Figure 2), leading to the hypothesis that these vesicles play a role in the transport of secreted extracellular RNA and RNA-binding proteins, dsDNA and histones, as well as a range of metabolic enzymes. However, most exosome preparations actually contain a mixture of exosomes and other small extracellular vesicles (sEVs), in addition to some non-vesicular material. This led Vanderbilt Institute of Chemical Biology member Robert Coffey and his laboratory to devise better methods of isolating and characterizing exosomes and non-vesicular particles known as exomeres. As a result of their work, we find that we must rethink what we believed we knew about these particles and their function [D.K. Jeppesen, et al. (2019) Cell, 177, 428; and Q. Zhang, et al. (2019) Cell Reports, 27, 940].



FIGURE 1. Mechanisms of formation of exosomes and microvesicles. Exosomes are formed through invagination of endosomal membranes to form intraluminal vesicles (ILVs) within multivesicular endosomes (MVEs). When MVEs fuse with the plasma membrane, their contents, including the ILVs are released into the surrounding environment, and the ILVs are designated exosomes. Microvesicles, in contrast, are formed through direct outward budding of the plasma membrane. Reprinted by permission from Springer Nature from G. van Niel, et al. (2018) Nat. Rev. Mol. Cell Biol., 19, 213. Copyright 2018, Springer Nature.





FIGURE 2. A large number of protein, nucleic acid, and lipid species have been reported to be components of exosomes. Reprinted by permission from Springer Nature from G. van Niel, et al. (2018) Nat. Rev. Mol. Cell Biol., 19, 213. Copyright 2018, Springer Nature.



The investigators started with a focus on exosomes. To obtain a more highly purified preparation of these particles from the culture medium of DKO-1 human colon carcinoma and Gli36 human glioblastoma cells, they first cleared the medium of cellular debris and large particulate matter. They then subjected it to centrifugation, first at 15,000 x g for 40 min and then at 120,000 x g for 4 h. These two procedures yielded pellets of large extracellular vesicles (p15, lEVs) and small extracellular vesicles (p120, sEVs), respectively (Figure 3).  Next, they further purified the p120 fraction by centrifugation on high-resolution iodixanol gradients. From this procedure, they obtained a lower density fraction that contained sEVs bearing classical exosome markers (CD63, CD81, and CD9) and exhibiting a cup-shaped morphology typical of exosomes purified by centrifugation (Figure 4, left). A higher density fraction contained non-vesicular particles (NVs).



FIGURE 3. Methods for purifying p120 fractions and exosomes. Culture medium recovered from a cell line of interest is subjected to centrifugation (pre-clearing) to remove cellular debris and large extracellular vesicles (lEVs). Then, centrifugation at 120,000 x g for 4 h yields a pellet (p120) that is washed and lysed for further study. This pellet contains exosomes as well as other small extracellular vesicles (sEVs) and some contaminating non-vesicular (NV) material. Alternatively, the cleared medium is treated with magnetic beads bearing antibodies targeting an exosomal membrane protein (CD63, CD81, or CD9). Exosomes bind to the antibodies and are recovered with a magnet. After washing, the exosomes are lysed, and the lysate is separated from the beads.


Proteomic profiling of sEVs and NVs from the iodixanol gradients revealed that the most abundant proteins in the sEV fraction were syntenin-1 and ALIX, both of which were well-characterized exosome proteins from prior studies. The NV fraction contained metabolic enzymes (GAPDH, PKM, ENO1) and cytosolic proteins such as HSP90 and tubulins. The NV fraction also contained histones H2A and H3. Notably, these proteins that were present in the NVs had previously been reported to be exosome components, but they were not found in the sEV fraction in this study.


In general, large RNA species were present at low concentrations, whereas small RNAs were enriched in extracellular samples as compared to cells. The various extracellular particle fractions contained different distributions of various small RNA species, with miRNAs most abundant in the NVs and tRNA fragments enriched in lEVs and sEVs. A number of RNA-binding proteins previously reported to be components of exosomes were not present, or present only at very low levels. YRNAs and vault RNAs were present with their concentration highest in the NV fractions.

Argonaute proteins, which are involved in processing miRNAs were present in the p120 fraction, although most proved to be associated with NVs rather than sEVs. Heat shock proteins HSP90 and HSC70, which are also involved in miRNA processing were present in both sEV and NV fractions. To better define the localization of these proteins, the researchers devised a method to isolate exosomes that was not based on centrifugation. Instead, the investigators trapped sEVs that expressed the exosome markers DC63, CD81, or CD9 on magnetic beads coated with antibodies directed against those proteins. They then used magnets to recover the beads, washed them to remove unbound contaminants, and finally eluted and lysed the purified exosomes (Figure 3). These vesicles did not contain Argonaute proteins or HSP90, indicating that, contrary to prior reports, exosomes are not a source of miRNA-processing proteins.



FIGURE 4. Negative stain transmission electron micrograph of sEVs (left) and DNPs (right) from Gii36 cells. Reproduced under the Creative Commons CC BY-NC-ND license from Q. Zhang, et al. (2019) Cell Reports, 27, rep.2019.01.009.


The researchers' ability to purify exosomes from other sEVs led them to discover that β1-integrin, a protein associated with tumor metastasis, was present predominantly in non-exosomal sEVs. Exosomes also contained relatively little of RAB proteins thought to be involved in the sorting and trafficking of MVEs. Similarly, although annexins were present at high levels in sEVs, only annexins AIX and AVII were present in true exosomes. Annexins A1 and A2 clearly co-purified with other fractions, and the observation that A1 was present in structures budding from the plasma membrane indicated that they were likely components of microvesicles. The researchers were able to confirm that exosomes contain the epidermal growth factor receptor (EGFR), as previously reported, but this protein was also present in non-exosomal sEVs. In contrast, ALIX and syntenin-1 were confirmed to be strongly associated with exosomes.

A function previously attributed to exosomes is the export of cytoplasmic dsDNA and associated histones from the cell. However, the researchers demonstrated that both histones and dsDNA localized primarily to the NV fraction in their experiments. Further work using structured illumination microscopy (SIM) revealed that, although cytoplasmic dsDNA co-localized with MVEs containing the exosome marker CD63, it was not present within ILVs that would eventually become exosomes. The finding that these MVEs also contained LC3B, a protein involved in the degradation of cytoplasmic chromatin via the autophagy pathway suggested that they were the result of the fusion of an endosomal MVE with an autophagosome containing DNA and histones to yield an amphisome. Subsequent fusion of the amphisome with the plasma membrane would lead to release of exosomes as well as free DNA and histones to the extracellular environment.

While they were using their highly purified preparations to demonstrate that many previously reported properties and functions of exosomes were, in fact, incorrect, the Coffey lab also embarked on a more detailed study of the NV component of particulate matter released by cells. In this case, they first cleared culture medium obtained from DiFi human colorectal carcinoma, Gli36  human glioblastoma, and MDCK canine kidney cells to remove cellular debris and lEVs. They then subjected it to centrifugation at 167,000 x g for 4 h and then for 16 h. The pellet obtained after 4 h contained sEVs with a morphology similar to that of exosomes, whereas the pellet obtained after 16 h contained distinct non-vesicular nanoparticles (DNPs) of ~39 nm - 71 nm in diameter (Figure 4). The relative quantities of sEVs and DNPs in the medium from the different cell types varied.

Proteomics analysis of the sEV and DNP fractions from DiFi cells revealed 1,741 proteins in common, 322 proteins found only in the sEV fraction, and 40 found only in DNPs.  The DNPs were enriched for proteins involved in metabolism, glycan processing, and amyloid formation. In contrast, sEVs were enriched for integrins, annexins, and EGFR. Common proteins included those involved in vesicle biogenesis and trafficking, suggesting that they shared a common mechanism of origin. Gene Set Enrichment Analysis suggested that the sEV fraction proteins were associated with IL2-STAT5 signaling and apical junction formation, whereas the DNP fraction proteins were associated with glycolysis and mTORC1 signaling. These findings were consistent with the characteristics of exosomes and exomeres (a previously reported non-vesicular extracellular nanoparticle). Analysis of nucleic acid and lipid content of the two fractions also suggested that the sEV and DNP fractions comprised, predominantly, exosomes and exomeres, respectively. As prior purifications of exomeres required asymmetric flow field-flow fractionation, a difficult procedure available to few laboratories, this method for obtaining the particles was a major step forward in the field.

α2,6-Sialylated N-glycans are a frequent glycan modification in cancer cells. Hence, the researchers were interested to find that the enzyme ST6Gal-1, which catalyzes α2,6-sialylation of glycans, was present in both exomeres and exosomes. Fluorescence-activated cell sorting analysis demonstrated that the enzyme was only present in exosomes that contained high levels of EGFR and CD81. An in vitro assay revealed that ST6Gal-1 in both exomeres and exosomes was enzymatically active, and both particulate fractions were able to transfer the enzyme to SW948 and SW48 cells, which do not naturally express it. Following a 48 or 72 h exposure to exosomes or exomeres, the cells also contained higher levels of α2,6-linked sialic acid than unexposed control cells, with the largest change resulting from exomere exposure. Focusing on β1-integrin, a well-characterized substrate of ST6Gal-1, the researchers demonstrated a rapid transfer of the preformed α2,6-sialylated protein from exosomes to target cells, followed by relatively low levels of de novo sialylation of cellular β1-integrin. In contrast, exomeres contained little to no preformed α2,6-sialylated β1-integrin, so no direct transfer occurred when they were added to the cells. However, over time, de novo synthesis catalyzed by transfered ST6Gal-1 led to increased levels of the sialylated protein in recipient cells that were sustained over longer time periods than those observed in exosome-treated cells. Thus, the researchers concluded that both exosomes and exomeres are capable of transferring active enzymes to target cells. In the case of ST6Gal-1, exomeres are the more potent of the two particles in this function.

Amphiregulin is a ligand for the EGFR. To determine if exosomes and/or exomeres are capable of transferring signaling molecules to recipient cells, the investigators isolated both particulate fractions from Amphiregulin-overexpressing MDCK cells and non-expressing parental cells. After confirming the presence of both mature and immature forms of AREG in exosomes and exomeres from the overexpressing, but not the parental cells, they demonstrated that all forms could be transferred to DiFi cells upon particle exposure. Furthermore, exposure of DiFi cells to AREG-containing exosomes or exomeres led to a robust increase in AREG expression. A smaller and more transient increase in AREG gene expression occurred upon exposure to purified recombinant AREG protein (rAREG). Similarly, AREG-containing, but not parental exosomes and exomeres resulted in a marked increase in EGFR-dependent target protein phosphorylation in DiFi cells. Again, rAREG elicited a substantially smaller response.

To further explore the ability of AREG-containing exosomes and exomeres to trigger EGFR-dependent signaling, the researchers created intestinal organoids from mice expressing EGFR labeled with the Emerald green fluorescent tag (EGFR-Em). When they treated these organoids with AREG-containing exosomes or exomeres, they initially observed a diffuse increase in EGFR-Em fluorescence in the epithelial cells that was then followed by the redistribution of the fluorescence into intracellular punctate foci. These findings suggested an increase in EGFR-Em expression followed by receptor internalization. In contrast, rAREG elicited a rapid receptor internalization without increased expression, and parental cell exosomes and exomeres had no effect (Figure 5). 



FIGURE 5. Effects of AREG-containing exosomes and exomeres versus rAREG on EGFR-Em expression and distribution in intestinal organoids. Organoids were treated with AREG-containing exosomes or exomeres or rAREG as indicated for 5 min or 30 min. Fluorescence signals indicate the location of EGFR-Emerald (Em, green), β-catenin (βcat, red), and nuclei (nuc, blue). Beneath each fluorescence image is a high-magnification single channel image corresponding to the box in the dotted line. AREG-containing exosomes and exomeres initially cause an increase in EGFR-Em expression that is diffusely distributed, but later is seen in a punctate distribution within the cytoplasm, consistent with receptor internalization. rAREG causes a rapid redistribution of EGFR-Em into a punctate pattern. Reproduced under the Creative Commons CC BY-NC-ND license from Q. Zhang, et al. (2019) Cell Reports, 27, rep.2019.01.009.


The results suggested that exosomes and exomeres are a potent means by which to deliver AREG to cells. To further test this hypothesis, the researchers created colonic tumor organoids from mice bearing a stem cell-driven form of colonic cancer. They found that treating these organoids with rAREG led to an increase in the size, but not the number of organoids in culture. In contrast, AREG-containing exosomes and exomeres elicited marked increases in both the size and number of organoids. This was not observed upon exposure to parental exosomes or exomeres (Figure 6).




FIGURE 6. Effects of AREG-containing exosomes and exomeres versus rAREG on size and number of colonic tumor organoids. Colonic tumor organoids were treated with parental (PAR) exosomes and exomeres (lacking AREG), AREG-containing exosomes and exomeres, two different concentrations of rAREG, and rEGF (a positive control EGFR ligand).  rAREG caused an increase in the size, but not the number of organoids, whereas AREG-containing exosomes or exomeres caused an increase in both size and number of organoids. Parental exosomes and exomeres had no effect. Reproduced under the Creative Commons CC BY-NC-ND license from Q. Zhang, et al. (2019) Cell Reports, 27, rep.2019.01.009.


The findings of these two studies revolutionize our thinking about the composition and function of extracellular vesicles and non-vesicular particles. Foremost, they demonstrate that these particles are highly heterogeneous, and careful purification and characterization are a critical prior step to functional studies. Thus, we have learned that many components and roles previously attributed to exosomes are actually properties of NV contaminants or other sEVs that are typically present in preparations thought to be exosomes. We have also learned that the DNPs present in the NV fraction include exomeres and that these particles are capable of both transferring enzymatic activity and eliciting signaling responses in target cells. Clearly, much work remains to fully understand the role of extracellular vesicles and nanoparticles in intracellular communication in both health and disease.



View Cell and Cell Reports articles:

Reassessment of Exosome Composition


Transfer of Functional Cargo in Exomeres (Cell Reports)







The Vanderbilt Institute of Chemical Biology, 896 Preston Building, Nashville, TN 37232-6304, phone 866.303 VICB (8422), fax 615 936 3884
Vanderbilt University is committed to principles of equal opportunity and affirmative action. Copyright © 2014 by Vanderbilt University Medical Center

  • Areas of Interest