Direct stochastic optical reconstruction microscopy (dSTORM) is used to bypass the typical diffraction limit of light microscopy and to view exosomes at the nanometer scale. It can be employed in both two and three dimensions to characterize exosomes.
Extracellular vesicles (EVs) are released by all cell types and play an important role in cell signaling and homeostasis. The visualization of EVs often require indirect methods due to their small diameter (40-250 nm), which is beneath the diffraction limit of typical light microscopy. We have developed a super-resolution microscopy-based visualization of EVs to bypass the diffraction limit in both two and three dimensions. Using this approach, we can resolve the three-dimensional shape of EVs to within +/- 20 nm resolution on the XY-axis and +/- 50 nm resolution along the Z-axis. In conclusion, we propose that super-resolution microscopy be considered as a characterization method of EVs, including exosomes, as well as enveloped viruses.
Extracellular vesicles (EVs) are membrane-bound vesicles released by all cell types. They contain lipids, proteins, metabolites, and nucleic acids and transfer these materials locally between cells and distally between tissues and organs. There are three primary subtypes of EVs: apoptotic bodies, microvesicles, and exosomes1,2. Here, we focus our discussion on exosomes and their associated proteins.
Exosomes are secreted vesicles originating from the inward budding of early endosomes into the multivesicular body (MVB). The MVB then fuses with the plasma membrane, releasing the exosomes into the extracellular space to travel to other cells3,4. Exosomes exist on a spectrum of sizes ranging from 40 to 150 nm and are enriched with endosomal transmembrane proteins known as tetraspanins (CD9, CD63, CD81), membrane-bound endosomal sorting complex required for the transport (ESCRT), and lipid raft-associated proteins1,2, 5,6,7.
Characterizing the biochemical makeup of exosomes has become a popular field for researchers to better understand their functional nature. Many methods exist for visualizing and characterizing exosomes, including nanoscale flow cytometry, nanoparticle tracking analysis (NTA), scanning and transmission electron microscopy (TEM), surface plasmon resonance, resistive pulse sensing, and traditional light microscopy, each of which contains intrinsic pros and cons8,9. TEM and cryo-EM can achieve nanometer-based resolution, but often require dehydrating and freeze-fracture steps, thereby shrinking or lysing EVs10,11. NTA relies on light scattering, allowing for the characterization of hundreds of EVs at a time, but is an indirect measurement of particle size and cannot easily distinguish between EVs, viruses, and protein aggregates12,13,14,15,16. Nanoscale flow cytometry employs light scattering from an excitation path, which can then be translated into size measurements, but is an emerging technology, and there is little consensus on what size of particles are within the linear range of detection for various instruments12,17,18.
Traditional light microscopy using fluorescent proteins or dyes has been one of the most heavily employed techniques for visualizing subcellular compartments, protein complexes, and signaling machinery within a cell. While this technique proves useful in visualizing the localization of complexes, the diffraction limit of traditional light microscopy (around 250-400 nm) prevents the clear resolution of proteins or structures in the typical size range of an exosome (40-150 nm)12,19,20.
Super-resolution microscopy, namely, direct stochastic optical reconstruction microscopy (dSTORM), distinguishes itself from conventional light microscopy by employing the photoswitchable properties of specific fluorophores and detecting these blinking events to reconstruct images down to nanometer precision21. Photoswitching events are collected using a high-framerate detection camera over the course of tens of thousands of individual exposures, and a point spread function is used to map with high confidence the exact location of the photoswitching fluorophore19,20,22. This allows dSTORM to bypass the diffraction limit of light microscopy. Several groups have reported the use of super-resolution techniques for visualizing and tracking exosomes and their associated proteins22,23,24,25. The final resolution depends on the biophysical properties of the fluorophore, but often ranges from +/-10-100 nm along the XY-axis, allowing single-molecule resolution.
The ability to resolve individual fluorophores at this scale on the XY-axis has revolutionized microscopy. However, there is little data on the three-dimensional (3-D) dSTORM of an exosome. Therefore, we sought to establish a standard operating procedure (SOP) for dSTORM-based visualization and characterization of purified EVs, including exosomes to nanometer precision in 3-D.
1 Propagation and maintenance of cell lines
2 Exosome isolation and purification
3 Fixation and preparation
4 Direct stochastic optical reconstruction microscopy calibration
5 Visualization of EV in three dimensions
6 Post-acquisition modification and EV tracing
The goal of this study was to evaluate the effectiveness of super-resolution microscopy in visualizing individual EVs with nanometer resolution in three dimensions (3-D). To analyze the shape and size of individual EVs, we employed photoswitchable dye and incubated the EVs with a far-red, membrane intercalating dye, and removed excess dye through chromatography29. The affinity-captured anti-CD81 and red-stained EVs were then viewed in the super-resolution microscope under the 640 nm excitation laser. Following the calibration of the microscope that produced an average error of 16 nm on the XY-axis and 38 nm on the Z-axis (Figure 1A,B), the purified U2OS EVs were successfully visualized with a resolution of up to 20 nm on the XY-axis and 50 nm along the Z-axis.
Individual EVs visualized through dSTORM in 3-D photoswitched throughout the 10,000 frame exposure as the laser power was increased and were readily apparent in the acquired image (Figures 2A,B). Post-acquisition image correction of the Z-plane, photon counts, sigmas, and localization precision of the reconstructed image allowed for the clear resolution of the EV in 3-D (Figure 2C,D). The EV in Figure 2C photoswitched during only the first 7,000 frames, as seen by the legend in the upper-right corner. This is a result of photobleaching that may have been caused by raising the laser power too quickly. The histogram confirms that the majority of photoswitching events occurred within a 100 nm radius (Figure 2E), validating that the visualized EV is an exosome and that the isolation of EVs of a small diameter was successful.
Size distribution analysis was performed on other individually traced EVs using a line histogram tool and XYZ plane view tool to confirm that the majority of photoswitching events occurred within a 100 nm radius of the center (Figure 3A,C), further validating dSTORM's ability to visualize EVs of a small diameter. As seen by a 3-D visualization tool, error along the Z-axis is increased, producing an elongated final image of the EV along the axial axis (Figure 3D, Video 1). Photoswitching events were not correlated with EV size (Figure 3E), demonstrating that dSTORM-based characterization can be used for small EVs such as exosomes and small enveloped viruses less than 100 nm in diameter.
The correct parameters for Z-plane and sigma are integral to the proper resolution of the membrane in 3-D. Additionally, the correct exposure time, number of frames captured, and initial laser level are crucial to producing an image of an individual EV with a resolved membrane. Further investigation should be done on how to optimize the microscope and set both pre and post-acquisitional parameters to capture the highest resolution images of EVs.
Figure 1: Calibration of the super-resolution microscope in three dimensions using 100 nm microspheres. (A) Field of view in grayscale from calibration of the microscope with microspheres after post-acquisition image corrections. Scale bar in the lower right. (B) Absolute calibration errors on the XY-axis and the Z-axis were obtained through channel mapping calibration and 3-D mapping calibration, respectively. N = 10 biological replicates. Please click here to view a larger version of this figure.
Figure 2: dSTORM of a single EV in three dimensions. (A) Grayscale field of view of CD81+ affinity-purified EVs stained with photoswitchable red membrane intercalating dye and excited with the 640 nm excitation laser. Scale bar in the lower right. (B) Field of view from A labeled according to channel color. (C) Single EV from the zoomed-in view of the white box in A, labeled according to the frame index. The heat map in the upper right indicates the frame in which the photoswitching events were recorded. Scale bar in the lower right. (D) EV from C labeled according to channel color. (E) Size distribution analysis was created by bisecting the EV on the dashed line shown in D and separating the photoswitching event into 15.4 nm bins using a line histogram tool. Please click here to view a larger version of this figure.
Figure 3: Distribution of photoswitching events during the acquisition of a single EV in three dimensions. (A) Reconstructed image of a CD81+ affinity-purified EV labeled with photoswitchable red membrane intercalating dye and excited with the 640 nm excitation laser. Scale bar in the lower right. (B) Box and whisker plot of average diameters of CD81+ affinity-purified EVs obtained using the microscope's Line Histogram Tool along the XY-axis. (mean = 104 nm, standard deviation = 28 nm). (C) Location of individual photoswitching events along the XY dimension of the EV in A, recorded throughout the 10,000 frame exposure. (D) Location of individual photoswitching events along the XZ-axis of the EV in A, recorded throughout the 10,000 frame exposure. (E) Scatterplot of the number of photoswitching events recorded on individual EVs of varying diameters. R-squared value of 0.1065 demonstrates no correlation between the number of detected photoswitching events and EV diameter. Please click here to view a larger version of this figure.
Video 1: 3-D image of a single CD81+ affinity-purified EV labeled with photoswitchable red membrane intercalating dye and excited with the 640 nm excitation laser. The color scheme is labeled according to depth along the Z-axis. Error along the Z-axis is elongated. Please click here to download this Video.
Variable | Min | Max |
Photon Count | 200 | 10,000,000 |
Z-position (nm) | -300 | 300 |
Localization Precision X (nm) | 0 | 40 |
Localization Precision Y (nm) | 0 | 40 |
Precision Sigma X (nm) | 0 | 40 |
Precision Sigma Y (nm) | 0 | 40 |
Sigma X (nm) (Channel 0, 640 nm laser) | 100 | 350 |
Sigma Y (nm) (Channel 0, 640 nm laser) | 100 | 350 |
Sigma X (nm) (Channel 1, 473, 561 nm lasers) | 100 | 350 |
Sigma Y (nm) (Channel 1, 473, 561 nm lasers) | 100 | 350 |
Frame Index | 0 | 10,000 |
Table 1: Parameters for the super-resolution microscope during 3-D dSTORM acquisition.
EVs have become a popular area of study due to their important role in many intracellular processes and cell-to-cell signaling1,30. However, their visualization proves to be difficult as their small size falls below the diffraction limit of light microscopy. Direct stochastic optical reconstruction microscopy (dSTORM) is a direct method of visualization that bypasses the diffraction limit by capturing photoswitching events of individual fluorophores over time and reconstructing an image based on these blinking events21,31. Super-resolution microscopy has been successfully performed in 3-D on several cellular structures such as actin filaments, microtubules, receptors embedded in the plasma membrane, and viral proteins in infected cells32,33,34,35,36. The purpose of this study was to evaluate the efficacy of super-resolution microscopy, specifically dSTORM, to visualize EVs in 3-D with nanometer resolution. We employed photoswitchable membrane intercalating dye to successfully visualize individual EVs from U2OS cells through dSTORM in up to +/- 20 nm resolution on the XY-axis and +/- 50 nm resolution on the Z-axis. Our previous work has shown that EVs from multiple cell lines and primary fluids have similar size distribution profiles as analyzed by NTA and TEM26,37. The use of a dSTORM-based characterization of EVs can add further confidence to this phenomenon, as well as potentially identify subpopulations in a single field of view. Further refinement of this method is warranted. One notable advantage to dSTORM is that the sample preparation does not require harsh or damaging steps that may alter the structure of the EVs. Our results further demonstrate that the biochemical nature of EVs is maintained during EV purification with anti-CD81 beads and acidic glycine26,37. We fixed the EVs with paraformaldehyde to avoid membrane permeabilizations caused by other fixatives such as methanol and ethanol. This allowed us to conclude that dSTORM accurately captures the morphology of EVs as they exist in solution. The use of Capto Core 700 is necessary, however, to purify the EVs effectively away from contaminants such as albumin and polyethylene glycol to below regulatory requirements38. One notable limitation of the protocol is that the binding efficiency between the EV and slides is not 100%, so some EVs are lost during sample preparation. Further investigation should be done on the efficacy of adhesive-coated slides to better bind EVs.
While the preparation of EVs for visualization is straightforward, the parameters during acquisition and especially during post-acquisitional modifications vary substantially from sample to sample, depending on the intensity and stability of the EVs and the fluorophore. One area of variability in the experiment is the intensity of the excitation lasers that must be delicately raised throughout exposure to maximize the signal but prevent photobleaching. Photobleaching, or when a fluorophore loses its ability to fluoresce, is a significant limitation throughout super-resolution microscopy12,39. To prevent photobleaching, the concentration of the enzyme in the buffer can be increased to 10 mM to better scavenge oxidizing molecules and prevent photobleaching. Additionally, setting the excitation laser power to a low initial level and slowly raising it throughout exposure to maintain a high signal is critical to prevent photobleaching.
We chose a red photoswitchable dye to stain the EVs due to its excitation wavelength, durability, and ability to withstand fixation29. However, the dye we chose may present issues during super-resolution microscopy when excited too quickly or at too high of an intensity. The fluorophores in the photoswitchable red membrane dye can photobleach after 10,000 frames or after the laser power exceeds 75.6 mW. Additionally, the membrane dye's intensity and ability to fluoresce decreases substantially after a week of storage at 4 °C. Finally, the membrane intercalating dyes have been shown to form micelles in an aqueous solution, so any excess dye still present in the EV sample after purification may be detected and mistaken for an EV if there is no other marker to identify the EV, such as a tetraspanin. Further investigation should be done using other membrane intercalating dyes to optimize for photo-stability31,40. Fluorescent antibodies conjugated to the EV may be a more suitable option as they tend to be more resistant to photobleaching and can amplify the signal41. However, the drawback of antibodies is that the signal from the fluorophore is slightly offset from the EV due to the distance from the epitope and fluorophore on the antibody.
Existing methods of EV visualization and characterization, such as NTA, flow cytometry, or EM, can require damaging preparation steps or are indirect methods of visualization. dSTORM requires little sample preparation, conserving the natural biochemical nature of EVs, and is a direct method of visualization that can bypass the diffraction limit and visualize EVs of a small diameter8,9,10,11,12,13,14,15,16,17,18,21,26. Other techniques of super-resolution microscopy can be optimized for EV characterization such as stimulated emission depletion (STED), spinning disc confocal microscopy (SDCM), and photo-activated localization microscopy (PALM)42,43. The current study exclusively focuses on dSTORM, but future work on optimizing super-resolution microscopy and comparing/contrasting these techniques is warranted. EVs have recently become a popular area of research as their role in virus progression has become more evident. Many evolutionarily distinct viruses, such as Epstein Barr Virus, HIV, and Hepatitis A Virus, have evolved to take advantage of the EV-signaling pathways to promote disease progression and evade the body's immune response37,44,45,46,47,48,49,50. These viruses have been shown to incorporate viral factors, such as mRNAs or viral proteins, into EVs that can then transfer these components to uninfected cells, while escaping immune detection6,51,52,53. These exosomal-associated viral factors can be detected and possibly employed as a biomarker for disease progression54,55. Therefore, dSTORM-based visualization of individual EVs and their associated proteins can be explored as a platform for disease biomarkers and, perhaps, disease progression of certain viruses54,55. Further research should be done to assess dSTORM's utility in visualizing both contents within an EV and proteins on its membrane.
In conclusion, we have demonstrated that super-resolution microscopy should be considered an effective technique for the visualization of EVs in 3-D with nanometer resolution. Results obtained by dSTORM are consistent with other EV characterizing techniques. A distinct advantage of dSTORM is the ability to directly visualize particles beneath the diffraction limit of light without dehydration or freeze fracture steps that can alter the biochemical nature of EVs.
The authors have nothing to disclose.
We would like to thank Oxford Nanoimaging for their constructive feedback and guidance. This work was funded by the 5UM1CA121947-10 to R.P.M. and the 1R01DA040394 to D.P.D.
15 µ-Slide 8 well plates | Ibidi | 80827 | |
1X PBS | Gibco | 14190-144 | |
1X Penicillin Streptomycin solution | Gibco | 15140-122 | |
50 mL conical tube | Thermo Fisher | 339652 | |
500 mL 0.22 µm vacuum filtration apparatus | Genesee | 25-227 | |
750 kDa hollow-fiber cartridge cutoff filter | Cytiva | 29-0142-95 | |
AKTA Flux S | Cytiva | 29-0384-37 | |
AKTA Start | Cytiva | 29022094-ECOMINSSW | |
Anti-CD81 magnetic beads | Thermo Fisher | 10616D | |
B-cubed buffer | ONI | BCA0017 | |
CellMask Red | Thermo Fisher | C10046 | |
Dubelco's Modified Eagle Medium | Thermo Fisher | 10566016 | |
Fetal Bovine Serum | VWR | 97068-085 | |
Frac 30 Fraction collector | Cytiva | 29022094-ECOMINSSW | |
Glycine pH=2.0 | Thermo Fisher | BP381-5 | |
HiTrap CaptoCore 700 Column | Cytiva | 17548151 | |
Molecular Biology Grade Water | Corning | 9820003 | |
Nanoimager | Oxford Nanoimaging | Custom | |
Paraformaldehhyde | Electron Microscopy Sciences | 15710 | |
Polyethylene glycol | Thermo Fisher | BP233-1 | |
RNase A | Promega | A797C | |
T175 Flasks | Genesee | 25-211 | |
Tetraspek microspheres | Invitrogen | T7279 | |
Tris- HCl pH=7.5 | Thermo Fisher | BP153-1 | |
Unicorn V | Cytiva | 29022094-ECOMINSSW |