By secreting extracellular vesicles (EVs), including exosomes and microvesicles, into the extracellular milieu, cells can convey complex biological messages between each other. These vesicles are generally thought to be static packages lacking the flexibility of their parental cells in terms of motility and the ability to change shape. However, cryo-electron micrographs reveal the presence of actin-like filaments in a subpopulation of EVs, raising the question if these vesicles could possess motile capabilities similar to that produced by actin in cells.

We here show that fluorescently labeled EVs change their shape in a matter of minutes, regardless of whether they are isolated from human body fluids, mouse tissue or cell culture of human cells or yeast. Our findings therefore cast doubt on movement being confined to cells, suggesting that some EVs indeed have an intrinsic capacity to move. This novel observation showing morphological plasticity among EVs adds another level of complexity to the already multifaceted vesicular secretome, and may lead to new ways in which we perceive these nano-carriers of intercellular signals.

(A & B) Cryo-electron micrographs of human ejaculate.

(C) Cryo electron micrograph of exosomes isolated from HMC-1 cells.

(D) Quantitative analysis of tubular vesicle length in human ejaculate samples.

(E-K) Light microscopy images were acquired with a 63X optical magnification and are represented here with a 200X digital magnification. Red scale bar = 0.5µm. White arrow indicates non-filamentous tubular vesicle and green arrow indicates filamentous tubular vesicles.


Extracellular Vesicles (EVs) are lipid membrane vesicles secreted by most studied cells and are involved in cell-to-cell communication, being able to convey messages to recipient cells by transfer of proteins, lipids and nucleic acids[1][2] [3][4][5] [6][7]. Furthermore, EVs are found to play key roles when studied in relation to physiology, oncology as well as immunology. When EV morphology is studied with cryo-electron microscopy (cryo-EM), most vesicles are round, but they can also have multiple other shapes.

A small subpopulation of extracellular vesicles also contain filamentous structures that are very reminiscent of polymerized actin[13][17]. Actin has also been found in multiple mass-spectrometric analysis of the EV proteome, as is evident in the database EVpedia[18]. As actin is a part of the cellular cytoskeleton crucial for cellular motility[19], we hypothesized that if EVs contain actin, they might also have the capacity to move. In this report, we present several observations showing that vesicles can take on shapes which are not energetically favorable (elongated), and that the induced shape is not permanent, but rather flexible and/or reversible.


Our objective was to evaluate if EV have the ability to change shape and/or move.

Results & Discussion

Cryo-EM showed tubular vesicles in EVs isolated from human ejaculate (Fig. 1A-B) and in exosomes isolated from a human mast cell (HMC-1) culture (Fig. 1C). Some are seen to contain internal filamentous structures that follow their same elongation direction (Fig. 1A-C; green arrows). However, some large tubules with elongated shapes are void of filamentous structures (Fig. 1B; white arrow).

The length of the filamentous tubular vesicles was quantified in human ejaculates, where the most instances of tubular filamentous structures were observed. They were larger than the average EV with a size of 1370±420 nm (n=18; Fig. 1D)[16]. The observation that vesicles can obtain an elongated morphology as well as contain luminal filaments prompted further investigation into the potential motility of this subcategory of EV’s. EVs were isolated either by centrifugation at 16500 × g (usually called microvesicles) or 118000 × g (usually called exosomes) from several sources.

The vesicular nature of these preparations was both assessed with thin-section electron microscopy and flow cytometry to confirm the presence of the common EV markers CD9, CD63 and CD81 on vesicles derived from HMC1 cells (Suppl. Fig. 1 and 2). EVs from all sources were allowed to settle onto glass bottom dishes, and were fluorescently stained before being visualized with epifluorescence microscopy. When examined in this manner, most EVs were static, non-moving fluorescent structures. However, time-lapse series revealed that a small fraction of these vesicles, larger than the average EV, and in a similar size range of the tubules measured in figure 1D, could undergo morphological changes (Fig. 1E-K and Suppl. Videos 1-8).

This morphological change could be demonstrated in cell culture derived microvesicles (Fig. 1E-F and Suppl. Video 1-2) as well as exosomes (Fig. 1G-H and Suppl. Video 3-4). The reshaping ability was also demonstrated in vesicles isolated from biological fluids and tissue, as movement was observed in serum microvesicles (Fig. 1I and Suppl. Video 5 and 8), and in exosomes isolated from mouse lungs (Fig. 1J and Suppl. Video 6). Furthermore, morphological change was also observed in a microvesicle preparation from cell culture of the budding yeast Sacchaomyces cerevisiae (Fig. 1K and Suppl. Video 7) which further supported that this phenomenon is evolutionary conserved. Several different modes of EV movements were observed. Most of the morphological changes seemed to be of the nature where an elongated structure rounds up and shifts into a round structure (e.g. Fig. 1E).

Another movement consisted in what appeared as a smaller vesicle moving inside a larger, stationary, vesicle (Fig. 1F). Some vesicles showed a motility where it appears that a smaller vesicle, or a bulge on the vesicle, glides along another vesicle (Fig. 1J). Finally, in some instances vesicles stretch out flexible protrusions (Fig. 1G, H, and K), a sort of movement which likely requires energy.

As morphological reshaping can be observed in EVs isolated from multiple sources, it can be argued that this is a conserved function inherited to a portion of the EVs from the parent cell. Whether this is a function deliberately afforded to the vesicle or merely a stochastically acquired ability, is at this point unknown.

Multiple factors have been proposed to regulate membrane curvature and thus also membrane shape, either individually or in concert with one another. Examples of such factors are membrane protein crowding, asymmetrical (conical-) lipid distribution, protein motif insertion (amphipathic helix), and BAR (Bin/Amphiphysin/Rvs) domains, to name a few[20][21][22][23][24][25]. An alternative to the aforementioned factors is the presence of polymerized actin within the vesicles. The fashion by which the vesicles appear to lose an energetically demanding elongated structure and change to a round form could arguably be ascribed to the de-polymerization of cytoskeletal filaments, the de-attachment of stabilizing proteins or the dispersion of aggregated membrane proteins or lipids.

The notion that actin is the driving force of the morphological change is strengthened by the presence of filaments which bare resemblance to polymerized actin within a subpopulation of vesicles, as visualized by cryo-EM. In addition, the length of the filamentous tubules seen in cryo-EM micrographs seem to match the perceived size of the shape-changing EVs observed in the wide field microscope. On the other hand, since EM reveals elongated vesicles that lack any filamentous structures, one can speculate that some of the non-actin factors can dictate EV membrane curvatures.

The suggestion that EVs can attain forms other than round spheres has been shown by electron microscopy, and is generally accepted. Our observation that their shape is flexible is however novel, and the implications of this observation are intriguing. An EV with these properties could putatively engage multiple receptors on a recipient cell by their elongated shape, facilitating receptor crosslinking or delivering multiple signals at joint locations on a cell. As curvature has been proposed to play a role in the process of membrane fusion, one can also speculate that an increased curvature in the vesicles would promote this process and thus promote uptake of EV cargo by a target cell[26].

Therefore, an ability to change shape may alter the effectiveness of both signaling and uptake, and shape could be a regulator of these processes. Lastly, if extracellular vesicles contain actin filaments capable of polymerizing and depolymerizing, then these vesicles could have motile capabilities. This would grant them the ability to travel between cells without the need for any moving fluid nor the need of recipient cells to be motile themselves. The vesicles that elongated protrusions suggests that there is an energetically demanding process at play. It has previously been shown that vesicles attained from ejaculate have the capability to produce ATP, which is required for actin polymerization.

Speculatively, actin polymerization could also facilitate cell entry by forcing the vesicles through the cell membrane of a recipient cell, reminiscent of the way in which vaccinia virus can spread between cells[28]. Targeted delivery of an EV to a recipient cell would be a much more specific cellular communication tool than randomly releasing cellular messages inside vesicles to the extracellular space.

Even though the observed events of EV shape-changes are relatively few in vitro, one might speculate that this is a more commonly occurring event in vivo, where conditions could be more favorable for such a phenomenon to occur. An environment rich in extracellular matrix, cells both near and far, and different chemical gradients and compositions could have profound effect on EVs carrying the machinery required for shape-change or intrinsic motility.


We report a novel feature of EVs, namely an ability to shape change. This is demonstrated across multiple samples and is thus indicative of a conserved function. Although having proposed a number of possible explanations, the mechanism underlying this shape change remains to be elucidated.


The scarcity of the events described is a concern which either mirrors a true rate by which morphological changes occur in vesicles, or it is a consequence of the conditions in which the experiments were conducted, which may not be optimal for extracellular vesicle intrinsic motility. The former might be true since actin was observed in 1.3% of the total vesicles in human ejaculate (Cryo-EM, data not shown). By a rough estimate however there appears to be fewer EVs with shape changing capabilities than there are actin containing vesicles.

Furthermore, a clear morphological distinction of the vesicles is made difficult by the inadequate resolution of the wide field microscope as well as the minute size of the vesicles themselves. Lastly, the lipophilic membrane dye used has at least two shortcomings: it bleaches quickly, making time series acquisition difficult, and it is theoretically also prone to forming micelles or aggregates which could be mistaken for vesicles. However, controls with only the dye, without vesicles, were performed and only limited fluorescence was detected.


The machinery that allows for morphological changes of extracellular vesicles needs to be further analyzed, as well as the possible ability of these vesicles to move from one place to another. Elucidating which factors are necessary for morphological plasticity in the vesicles, or describing the subpopulation which has this plasticity, should take precedence, as this would aid further investigations into the biological role of this sub-population of vesicles.

If vesicles do have the capabilities to migrate then the factors dictating such behavior and the subsequent control of that can have vast implications for extracellular vesicle biology, and putatively for the therapeutic potential of EVs.


HMC-1 culture

The human mast cell line (HMC-1) was cultured in IMDM complemented with 10% EV-depleted fetal bovine serum (FBS), 100 units/ml streptomycin, 100 units/ml penicillin, 2 mM L-glutamine and 1.2 U/ml alpha-thioglycerol (all regents were from Sigma, St Louis, MO, USA). FBS was depleted of EVs by ultracentrifugation at 118,000 × gavg for 18 h at 4°C (Type 45 Ti rotor, 38,800 rpm, k-factor 178.6, Beckman Coulter, Brea, CA, USA).

Mouse lung tissue

Lungs from one mouse (C57BL/6) were dissected and immediately immersed into a petri dish filled with RPMI (Sigma Aldrich) supplemented with 100 units/ml Penicillin, 100 µg/ml Streptomycin and 2 mM L-glutamine (all regents were from HyClone GE healthcare Life Sciences, Logan, Utah, USA). The lungs were sliced into small cubes (1–2 mm) and incubated with Collagenase D (Roche Stockholm, Sweden) (2 mg/ml) and DNase I (Roche, city, country) (40 U/ml) dissolved in RPMI plain medium (Sigma Aldrich) for 30 min at 37°C. Chopped lung pieces were filtered (70 µm pore size) and the flow through put in culture for 72 h at 37°C in RPMI (Sigma Aldrich) supplemented with 100 units/ml Penicillin, 100 µg/ml Streptomycin, 2 mM L-glutamine (all regents were from HyClone), 20 µg/ml Carbenicillin and 10% FBS (Gibco Invitrogen Corporation, Carlsbad, CA, USA). FBS was depleted of EVs prior to use as described earlier.

Blood serum

20 ml of whole blood was collected from a healthy volunteer into BD vacutainer SST tubes (BD biosciences, New Jersey, USA) and left to clot for 1 h at room temperature to be able to collect serum. The blood was then centrifuged at 1880 × g for 10 min. The serum was transferred to ultracentrifuge tubes and further handled as described below.

Yeast culture

Cells of strain BY4741 were grown to OD600 0.5 in synthetic complete medium. Vesicles were isolated from 900 ml of culture using a centrifugation-based protocol. Cells were removed by pelleting them first at 400 × g for 10 min and at 600 × g for 15 min. The supernatant was used for further vesicle isolation.

EV Isolation

All EVs were isolated by differential ultracentrifugation. Cells were pelleted at 300 × g for 10 min. Supernatant was transferred to ultracentrifuge tubes and centrifuged at 16,500 × gavg for 20 min to pellet what is usually referred to as microvesicles. The supernatant was again carried over to new ultracentrifuge tubes and centrifuged at 118,000 × gavg for 3.5 h at 4°C to pellet what is usually referred to as exosomes. Both microvesicle and exosome pellets were resuspended in phosphate buffered saline (PBS) and used immediately for downstream experiments. For EM experiments. Exosomes were further purified by floatation on a density gradient.

Electron microscopy

Cryo-EM images were acquired for vesicles derived from ejaculate and HMC-1 cultures, with either of two electron microscopes. One was a Tecnai F30 electron microscope (FEI Company Ltd.) with a GATAN UltraCam detector, lens coupled, 4 K CCD camera attached to a Tridiem Gatan Image Filter (Gatan, Inc., Pleasanton, CA, USA). The images were acquired at 27,500 X magnification with -4 to -6 defocus. The other electron microscope used was a Philips CM120 BioTWIN.

Thin section electron microscopy: 16,500 × g pellet were resuspended in 20% bovine serum albumin (BSA) in PBS and high pressure frozen (Leica Empact I, Leica Microsystems) directly after isolation. Samples were freeze substituted at -90°C for 1 h (solution: 2% uranyl actetate (UA) in 10% methanol and 90% acetone[29]), then washed twice with acetone. Temperature was increased 3°C/h to -50°C where the infiltration of HM20 was performed before 48 h ultra violet light polymerization. 70 nm thick sections were cut and en-section stained with uranyl acetate and lead citrate before being imaged on a Leo 912AB Omega TEM operated at 120 kV.

Fluorescence microscopy

EV isolates from either HMC-1, blood, yeast or lung tissue were loaded onto glass bottom dishes and left to adhere to the surface for 15 min at room temperature. The dishes were gently washed 3 times with PBS. They were then stained with PKH67 by mixing of the lipid dye with the Diluent C in a 1:500 ratio, of which 150 µl was placed on each dish containing vesicles. The dye was left to incubate on the dishes for 2 min after which it was removed and then the dishes were washed 3 times with PBS and finally filled with 2 ml of fresh PBS. The samples were then immediately analyzed with an Axio Observer (Zeiss, Oberkochen, Germany) with which time lapse series were taken at a magnification of 63X.

Flow cytometry

Samples were incubated over night under gentle agitation with dynabeads coated with antibodies against CD63 (15 µg exosomes/70,000 beads/antibody; Human CD63 Isolation/Detection (from cell culture media), Thermo Fisher Scientific). The sample was washed with PBS containing 1% EV depleted FBS after which the samples were incubated with human IgG (Sigma-Aldrich) for 15 min at 4°C. The samples were again washed and then incubated with PE-labeled antibodies against CD9, CD63, CD81 as well as with a PE(phycoerythrin)-labeled isotype control (BD Bioscience, Erembodegem, Belgium) for 40 min with gentle agitation. Samples were again washed before analysis with a FACSVerse (BD Biosciences).

Funding Statement

This work was funded by grants from the Swedish Research Council, the Swedish Cancer Foundation, Ingabritt och Arne Lunbergs forskningsstifftelse, VBG Group Herman Krefting Foundation for Asthma and Allergy Research, and the Swedish Heart and Lung Foundation. The funders had no role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript. JLH was supported by a VR young investigator grant and the Göran Gustafsson Foundation for Research in Natural Sciences and Medicine.


The Centre for Cellular Imaging at the Sahlgrenska Academy, University of Gothenburg is acknowledged for the use of imaging equipment and the support from the staff. We thank Gunnar Nilsson at the Karolinska Institute, Stockholm, Sweden, for the kind gift of the human mast cell line HMC-1. We thank Gunnel Karlsson at the Department of Chemistry, University of Lund, Sweden for the assistance with Cryo-TEM.

Conflict of interest

The authors do declare conflicts of interest:

Jan Lötvall is an employee of Codiak BioSciences, a company developing exosomes as a platform for therapy in disease, and has several patents in the area of using exosomes as therapeutics and diagnostic tools. Aleksander Cvjetkovic, Rossella Crescitelli, Johanna Höög, Davide Zabeo and Cecilia Lässer do not have any conflict of interest.

Ethics Statement

Samples were collected with the approval of the Regional Ethical Approval Committee in Gothenburg, Sweden.

No fraudulence is committed in performing these experiments or during processing of the data. We understand that in the case of fraudulence, the study can be retracted by ScienceMatters.