Size Characterization and Quantification of Exosomes by

Article pubs.acs.org/ac

Size Characterization and Quantification of Exosomes by Asymmetrical-Flow Field-Flow Fractionation Simona Sitar,† Anja Kejžar,‡ David Pahovnik,† Ksenija Kogej,§ Magda Tušek-Ž nidarič,⊥ Metka Lenassi,‡ and Ema Ž agar*,† †

National Institute of Chemistry, Laboratory for Polymer Chemistry and Technology, Hajdrihova 19, 1000 Ljubljana, Slovenia University of Ljubljana, Faculty of Medicine, Institute of Biochemistry, Vrazov Trg 2, 1000 Ljubljana, Slovenia § University of Ljubljana, Faculty of Chemistry and Chemical Technology, Department of Chemistry and Biochemistry, Večna pot 113, 1000 Ljubljana, Slovenia ⊥ National Institute of Biology, Večna pot 111, 1000 Ljubljana, Slovenia

Downloaded via TUFTS UNIV on June 29, 2018 at 14:03:49 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

‡

ABSTRACT: In the past few years extracellular vesicles called exosomes have gained huge interest of scientific community since they show a great potential for human diagnostic and therapeutic applications. However, an ongoing challenge is accurate size characterization and quantification of exosomes because of the lack of reliable characterization techniques. In this work, the emphasis was focused on a method development to size-separate, characterize, and quantify small amounts of exosomes by asymmetrical-flow field-flow fractionation (AF4) technique coupled to a multidetection system (UV and MALS). Batch DLS (dynamic light-scattering) and NTA (nanoparticle tracking analysis) analyses of unfractionated exosomes were also conducted to evaluate their shape and internal structure, as well as their number density. The results show significant influence of cross-flow conditions and channel thickness on fractionation quality of exosomes, whereas the focusing time has less impact. The AF4/UV-MALS and DLS results display the presence of two particles subpopulations, that is, the larger exosomes and the smaller vesicle-like particles, which coeluted in AF4 together with impurities in early eluting peak. Compared to DLS and AF4-MALS results, NTA somewhat overestimates the size and the number density for larger exosome population, but it discriminates the smaller particle population.

E

index, and broad size distribution.12 The most commonly used techniques for exosome characterization are fluorescence based detection, nanoparticle tracking analysis (NTA), resistive pulse sensing, dynamic light-scattering (DLS), atomic force microscopy (AFM), as well as transmission and scanning electron microscopy (TEM, SEM).12−17 Currently, the most commonly used method for characterization of exosome size and concentration is NTA that covers the size range between 30 and 1000 nm.15,18 NTA is based on tracking the vesicles, which are continuously moving because of Brownian motion, by lightscattering using a light microscope. Although NTA is a relatively simple technique and convenient to use, it has many drawbacks, for example, two populations can only be resolved if their particle diameters differ by at least 1.5-fold. In addition, the quality of NTA data strongly depends on the applied software settings and, consequently, this technique demands a skilled operator. These drawbacks make qualification of NTA as a quality control method very difficult. A technique with a strong potential to overcome the above-

xosomes are natural nanometer-sized membrane vesicles secreted by fusion of the multivesicular body with the plasma membrane of the cell.1,2 They are released by normal and neoplastic cells into their environment, and are present in most, if not all, biological fluids as well as in cell culture media.3,4 Exosomes consist of a lipid bilayer membrane surrounding a small cytosol, and contain various molecular constituents of their origin cell, including proteins and nucleic acid material like micro RNA.5 These vesicles have an important role in many physiological functions, ranging from intercellular communications, coagulation, modulation of immune response, angiogenesis to cell survival, and others.6−10 In the last years exosomes have gained strong clinical and scientific interest due to their promising potential in human diagnostic and therapeutic applications. They have been studied extensively as biomarkers for cancer and neurodegenerative diseases11 since exosomes present in body fluids (like blood) are more accessible than parent cells. Therefore, they are superior diagnostic alternative to invasive and painful needle biopsy.10 One of the most critical limitations to be faced with is a lack of suitable, that is, reliable and accurate, techniques for exosome characterization and quantification since straightforward detection is hampered by their small size, low refractive © 2015 American Chemical Society

Received: April 29, 2015 Accepted: August 20, 2015 Published: August 20, 2015 9225

DOI: 10.1021/acs.analchem.5b01636 Anal. Chem. 2015, 87, 9225−9233

Article

Analytical Chemistry

Table 1. Cross-Flow Conditions Applied for the Separation of Exosome Sample by AF4 with Corresponding Figures That Graphically Represent the Experimentsa

a

exp. no.

cross-flow gradient

1 2 3 4 5 6 7 8

0.061 mL/min2 (1.0−0.09 mL/min in 15 min) 0.027 mL/min2 (0.5−0.09 mL/min in 15 min) 0.011 mL/min2 (0.25−0.09 mL/min in 15 min) 0.005 mL/min2 (0.25−0.09 mL/min in 30 min) 0.0036 mL/min2 (0.25−0.09 mL/min in 45 min) 0.15 mL/min2 (1−0.25 mL/min in 5 min) and 0.0036 mL/min2 (0.25−0.09 mL/min in 45 min) 0.35 mL/min2 (2−0.25 mL/min in 5 min) and 0.0036 mL/min2 (0.25−0.09 mL/min in 45 min) 0.55 mL/min2 (3−0.25 mL/min in 5 min) and 0.0036 mL/min2 (0.25−0.09 mL/min in 45 min)

graphically shown exp. Figure Figure Figure Figure Figure Figure Figure Figure

1a 1a 1a and Figure 1b 1b 1b 2 2 2

The detector and the focus flow rates were 1.0 and 1.5 mL/min, respectively.

determined. The results were compared to those obtained by TEM, DLS, and NTA techniques.

mentioned problems and to become a quality control method is asymmetrical-flow field-flow fractionation (AF4) coupled to a multi-angle light-scattering detector (MALS).19 AF4 is a subclass of field-flow fractionation techniques20 developed by Giddings in 1966. It separates macromolecules and particles according to their diffusion coefficient. Separation is achieved in a channel consisting of two plates separated by a spacer. The upper plate is impermeable, whereas the bottom plate is permeable, made of porous frit and covered by a semipermeable membrane with a defined pore-size. The highresolution separation is achieved within the parabolic flow profile passing through the channel, against which a perpendicular cross-flow is applied. The analyzed molecules/ particles are driven by the cross-flow toward the channel bottom plate or accumulation wall. However, because of the counteracting Brownian motion of the particles, they reach equilibrium position away from the accumulation wall. Small particles with high diffusion coefficient float closer to the channel center and are displaced by the faster flow stream of the parabolic flow profile and, consequently, elute earlier than larger particles with smaller diffusion coefficient, which drives them closer to the accumulation wall where the flow is slower. With AF4-MALS the average size and size distribution of macromolecules and supramolecular assemblies can be determined in an accurate and reliable manner. AF4 system covers broad separation range from few nanometers up to micrometers and, thus, it is suitable for exosome separation. Because a wide variety of running eluents, including organic and salty aqueous buffer, can be applied, AF4 technique in combination with a light-scattering detection is suitable for separation and characterization of a wide range of biological and nonbiological analytes, that is, proteins,21−23 plasmids,22 polysaccharides,22,24 and virus-like particles.25,26 Recently, AF4 technique coupled to different detectors was used for fractionation of exosomes harvested from the immortalized human mesenchymal27 or neural28 stem cell culture for further protein analysis by mass spectrometry, and for size separation and characterization of exosomes from an aggressive mouse melanoma cell culture;29 however, the studies are still limited. The purpose of this work was to apply AF4 in combination with a multidetection system (UV and MALS) for characterization and quantification of extracellular vesicles called exosomes. The main focus was directed toward the optimization of experimental parameters, that is, focusing time, cross-flow rate, and channel thickness to achieve efficient fractionation of exosomes. From the fractograms, the average size, the size distribution, the shape, the internal structure, and the number density of exosomes (number per mL) were

■

EXPERIMENTAL SECTION Exosome Sample. Lyophilized exosome standard HBMBLCL21-30 purified from culture supernatant of EBV transformed lymphoblastoid B cell line was purchased from HansaBioMed, Tallinn, Estonia. Prior to the measurements, exosome pellets were suspended in water, following the manufacturer’s instructions. Asymmetrical-Flow Field-Flow Fractionation Coupled to a Multidetection System (AF4/UV-MALS). AF4 was performed at room temperature on an Eclipse 3+ system (Wyatt Technology Europe, Dernbach, Germany) connected to an isocratic pump, an online vacuum degasser and an autosampler (all Agilent Technologies 1260 series, USA). Exosome sample was separated in a channel with a trapezoidalshaped spacer with a thickness of 480, 350, 250, and 190 μm, a tip-to-tip length of 152 mm and an initial channel breadth of 21 mm that decreased to a final 3 mm. A 10 kDa regenerated cellulose membrane was utilized on the accumulation wall. The fractionated particles were detected with a UV detector at 280 nm (Agilent Technologies, USA), and a multi-angle lightscattering (MALS) detector (DAWN HELEOS, Wyatt Technology, USA) at 658 nm, calibrated using toluene, and normalized with bovine serum albumin protein as an isotropic scatterer standard. The PBS (phosphate buffered saline, pH 7.4) as a running eluent, composed of 137 mM NaCl, 2.68 mM KCl, 10.14 mM Na2HPO4, and 1.84 mM KH2PO4, was supplemented with 0.02% w/v sodium azide (NaN3) as a bactericide, and filtered through a Nylon-66 membrane with a pore-size of 0.45 μm (Supelco Analytical, USA). Between the HPLC pump and the AF4 channel an additional filter with a pore size of 0.1 μm was placed (PEEK Inline Filter Holder). Thirty micrograms of exosome standard was resuspended in 250 μL of water to prepare a suspension with a concentration of 0.12 μg/μL. Then, 20 μL of exosome suspension (2.4 μg) was injected in the focus mode using the focus flow of 1.5 mL/ min and the injection flow of 0.2 mL/min over 5 min. After injection, the sample was typically focused for additional 7 min. Besides, the shorter and longer focusing times, that is, 2 and 12 min, were also tested. After focusing, the sample was eluted at the detector flow rate of 1.0 mL/min using various cross-flow regimes, listed in Table 1. In all experiments, the last two steps, that is, elution and elution plus injection, included washing the channel and injection loop without any cross-flow. For the data acquisition and evaluation Astra 5.3.4.20 software was utilized. The size of the exosomes was expressed by two different radii, that is, the root-mean-square radius 9226


Article


recalculated for exosome concentration of 0.12 μg/μL, so the results can be compared to these determined by the AF4MALS. Transmission Electron Microscopy (TEM). The exosome sample was visualized by TEM using negative staining method. Four μL of exosome suspension (0.12 μg/μL) was applied on Formvar-coated and carbon stabilized copper grid and stained with 1% (w/v) water solution of uranyl acetate. Sample was examined with Philips CM 100 electron microscope (FEI, The Netherlands), operating at 80 kV, and images were recorded with Orius 200 camera (Gatan) using DigitalMicrograph Software (Gatan Inc.,Washington, DC, USA).

(Rrms) and the geometric radius (Rgeom), which were obtained from the MALS (multi-angle light-scattering) detector without the need of the solute concentration and the sample refractive index increment. The Rrms radii of the fractionated exosomes were calculated using the data from 15 angles from the MALS detector, applying, depending on exosome size, either the Zimm first or the Debye third order models. The Rgeom was determined by the Astra Particle template assuming spherical exosome shape.30 The amount of eluting exosomes was evaluated solely from MALS data by the Astra Number density template31 using exosome refractive index of 1.39.32 Dynamic Light-Scattering (DLS). DLS measurements were conducted using a 3D cross-correlation spectrometer from the LS Instruments GmbH (Fribourg, Switzerland), where the two coherent incident light beams are generated with a 20 mV He−Ne laser operating at 632.8 nm. The DLS measurements of exosome sample were performed at 90° angle. To determine the hydrodynamic radii (Rh) of particles the autocorrelation function of intensity (G2(t)) was converted into the correlation function of the scattered electric field (g1(t)) by using the Siegert’s relation.33 For the monodisperse particles having small diameter as compared to the wavelength of incident light and for the hard spheres of any size, the translational diffusion coefficient (D) is calculated from the 2 following equation: g1(t) = e−Dq t. Then, the Rh of the particles is obtained from the D via the Stokes−Einstein equation: Rh = kT/6πη0D, where k is the Boltzmann constant, T is the temperature and η0 is the solvent viscosity. For the particles studied herein, a multiexponential fit to g1(t) was used since the particles showed bimodal size distribution. The multiexponential fit was based on the original inverse Laplace transform program CONTIN developed by Provencher.34 The sample solution was prepared in the same way as for the AF4 measurements, that is, 30 μg of exosome standard was resuspended in 250 μL of water to a final concentration of 0.12 μg/μL, and was allowed to equilibrate for 20 min before characterization. Nanoparticle Tracking Analysis (NTA). NTA measurements were performed with a NanoSight LM10 apparatus (Amesbury, U.K.) with a 488 nm Blue Laser Module. When the beam of the laser light-scattering microscope hits the vesicles, the scattered light is detected by a sensitive charged-couple device camera, operating at 30 frames per second, which visualizes and records the moving vesicles. Individual vesicles, which are moving under Brownian motion, are identified and tracked by NTA software and transformed to a particle size based on the following formula derived from the Stokes− Einstein equation: (x , y)2 = 2kBT/3rhπη; where kB is the

■

RESULTS AND DISCUSSION AF4/UV-MALS: Determination of Optimum CrossFlow Rate. Cross-flow is an important parameter defining the separation quality within AF4 channel. According to AF4 theory, the equilibrium height of the analyzed particles over the accumulation wall is determined by equilibrium between the strength of the applied cross-flow and the counteracting particle motion because of diffusion. Therefore, a selection of an appropriate cross-flow rate is crucial for method development as it directly influences the separation quality. To study the influence of cross-flow rate on the exosome fractionation various cross-flow conditions were tested. The literature reports unsuccessful separation of liposome particles, which are similar to exosomes, by AF4 using the isocratic cross-flow conditions.36 Therefore, the AF4 method was optimized by applying various linear cross-flow gradients: initially, one and, later, two different linear cross-flow gradients in one run. All tested cross-flow conditions are listed in Table 1. The fractograms of exosome sample obtained at three different linear cross-flow gradients, that is, 0.061, 0.027, and 0.011 mL/min2 (the cross-flow decreased within 15 min from 1.0 to 0.09, 0.5−0.09, and 0.25−0.09 mL/min, respectively), are shown in Figure 1a (Exp. No. 1−3 in Table 1). When fast cross-flow gradients (0.061 and 0.027 mL/min2) were applied, most of the sample eluted from the channel with field release, that is, when the cross-flow had been turned to zero flow rate at 32 min (black-1 and red-2 curves in Figure 1a). These results indicate that almost entire sample retained in the channel at too fast gradients. At the slowest cross-flow gradient, that is, 0.011 mL/min2, more reproducible elution behavior of exosome sample was observed, however, a small amount of sample still eluted with the zero cross-flow rate (blue curve-3 in Figure 1a). To achieve elution of the entire sample within the operating cross-flow, the AF4 fractionation was performed at the crossflow gradient from 0.25 to 0.09 mL/min within two ramp times; one within 30 min (cross-flow gradient of 0.005 mL/ min2; exp. no. 4 in Table 1) and the other one within 45 min (cross-flow gradient of 0.0036 mL/min2; exp. no. 5 in Table 1). The resulting elution profiles at 90° angle show that the entire sample eluted from the channel under operating cross-flow only at the slowest cross-flow gradient of 0.0036 mL/min2 (green curve-1), whereas at the cross-flow gradient of 0.005 mL/min2 (magenta curve-2) a part of the sample still left in the channel when the cross-flow had been turned off at 47 min (Figure 1b). On the basis of these results the slowest cross-flow gradient of 0.0036 mL/min2 was used throughout the further study. However, at all applied cross-flow conditions the sample eluted immediately after the void peak that appears due to pressure fluctuations when the focus mode is switched to the elution mode. Since in such cases it is not possible to set

Boltzmann constant and (x , y)2 is the mean-squared speed of a particle with a hydrodynamic radius of Rh in a medium of defined viscosity and temperature.18 NTA is appropriate for sizing and enumerating the vesicles with diameter from 30 to 1000 nm at a concentration ranging from 108 to 109 particles/ mL.15,18,35 For the NTA measurements, 30 μg of exosome standard was resuspended in 30 μL of water to a final concentration of 1 μg/ μL. One microliter of this suspension was diluted to 1 mL with the particle-free PBS buffer and transferred into the NTA sample chamber via the syringe pump. The settings for the shutter and the camera gain were 800 and 350, respectively. During the measurements 7 videos within 60 s were captured by the software. The number of particles per mL was 9227


Article


Figure 2. AF4-MALS fractograms at 90° angle for the exosome sample separated at two linear, successively applied cross-flow gradients in a single run. Initial cross-flow gradients applied over 5 min were 1.0− 0.25 (black line-1), 2.0−0.25 (red line-2), and 3.0−0.25 mL/min (blue line-3). The second cross-flow gradient was in all experiments the same, that is, 0.25−0.09 mL/min in 45 min.

AF4/UV-MALS: Influence of Focusing Time. To optimize the separation quality in AF4, the sample constituents are, before the elution step, positioned in their steady-state equilibrium level by optimal focusing/relaxation step. By this means, the resolution is improved, the band broadening reduced, and the premature elution of analytes prevented. When focusing time is too short, the sample band under the injection port in the channel is wider than necessary, resulting in nonoptimal resolution. On the other hand, excessive focusing can cause interaction of the analyte’s molecules with the accumulation wall or even with each other.36,37 To investigate the impact of focusing step prolongation, the exosome sample was fractionated using identical elution conditions (the initial cross-flow gradient of 3−0.25 mL/min in 5 min followed by the lower cross-flow gradient of 0.25−0.09 mL/min in 45 min; the detector flow of 1 mL/min; the focus flow of 1.5 mL/min), but varying focusing time, that is, 2, 7, and 12 min, after the focusinjection step (Figure 3). With longer focusing only marginal changes in the peak shape and the sample mass recovery were observed, indicating absence of sample adsorption to the accumulation wall. Besides, the values of MALS derived Rrms

Figure 1. AF4-MALS fractograms of exosome sample obtained at different cross-flow conditions: (a) linear gradients over 15 min (crossflow turned-off at 32 min) 1.0−0.09 mL/min (black-1), 0.5−0.09 mL/ min (red-2), 0.25−0.09 mL/min (blue-3) and (b) linear gradient 0.25−0.09 mL/min in 15 min (blue-3) with cross-flow turned-off at 32 min; linear gradient 0.25−0.09 mL/min in 30 min (magenta-2) with cross-flow turned-off at 47 min; and linear gradient 0.25−0.09 mL/ min in 45 min (green-1) with cross-flow turned-off at 62 min. The asterisk (*) designates switching from elution to elution + injection step to flush the channel and injection loop.

suitable baseline the AF4 method was further modified in a way to better separate the peak of exosome sample from the void peak. For this purpose two linear cross-flow gradients were successively applied in a single run. We tested three different initial cross-flow gradients, that is, 0.15, 0.35, and 0.55 mL/min2 (cross-flow decreased over 5 min from 1.0 to 0.25, 2.0−0.25, and 3.0−0.25 mL/min, respectively), followed by the second cross-flow gradient which was the same in all experiments, i.e. 0.25−0.09 mL/min in 45 min (exp. no. 6−8 in Table 1). At all three initial gradients, the elution profiles at 90° angle indicate well separated exosome peak from the void peak (Figure 2). The shape of the elution curves does not change noticeably, meaning that the strength of the initial cross-flow gradient did not cause more pronounced peak broadening. Moreover, the area under the peak recorded by either MALS or UV detectors did not change a lot, indicating constant sample mass recovery without any losses due to sample passing through- or sticking onto the membrane. Based on these results the highest initial cross-flow gradient of 0.55 mL/min2 was used throughout the further study.

Figure 3. UV fractograms of exosome sample obtained at the focusing time of 2 min (black-1 curve), 7 min (red curve-2), and 12 min (blue curve-3) after the focus-injection step. The asterisk (*) designates switching from elution to elution + injection step to flush the channel and injection loop. 9228


Article

Analytical Chemistry and Rgeom radii of exosome sample did not change significantly with focusing duration (Figure 4), which again indicate

experiments, we tested, at otherwise the same experimental conditions, the spacers with the thickness of 190, 250, and 480 μm (Figure 5). The 190 (black curve-1) and 250 μm (red

Figure 4. AF4-MALS fractograms (normalized) of exosome sample together with (a) Rrms and (b) Rgeom radii obtained at the focusing time of 2 min (black curve-1), 7 min (red curve-2), and 12 min (blue curve3) after the focus-injection step. The asterisk (*) designates switching from elution to elution + injection step to flush the channel and injection loop.

Figure 5. (a) AF4-MALS fractograms of exosome sample obtained with different spacer types: 190 μm (black curve-1), 250 μm (red curve-2), 350 μm (blue curve-3), and 480 μm (magenta curve-4). (b) Normalized LS signals at 90° for exosome sample together with Rgeom radii determined by using 350 μm (blue curve-3) and 480 μm (magenta curve-4) spacers. The asterisk (*) designates switching from elution to elution + injection step to flush the channel and injection loop.

negligible effect of focusing time on elution behavior of exosome sample. On the basis of these results the focusing time of 7 min was used throughout the further study. AF4/UV-MALS: The Effect of Channel Thickness. Channel geometry, that is, the channel width and thickness, is one of the fundamental parameters affecting sample fractionation. Channel width influences the shape of the eluting peak, but it is not a crucial parameter that determines the fractionation quality. On the other hand, channel thickness (a distance from the accumulation to the upper channel wall: normally between 50 and 500 μm) defined with the spacer type, controls the parabolic velocity profile inside the channel, and thus it has a substantial impact on the fractionation quality. In addition, the spacer thickness determines the loading capacity; the higher the spacer thickness is the greater is the sample loading capacity, however at the detriment of resolution. In our case, the sample overloading was not the issue since our aim was, due to small sample availability, to inject as small amount of the sample as possible and still get satisfactory LS and UV responses. The influence of channel thickness on fractionation behavior of exosome sample was studied using four different spacer types. In addition to the 350 μm spacer used in the first set of

curve-2) spacers gave in the fractograms only a single narrow peak with poor separation by size. With the 350 (blue curve-3) and 480 μm (magenta curve-4) spacers, an improved separation was observed, however, in the case of 480 μm spacer a part of the sample eluted from the channel after the cross-flow had been turned-off (Figure 5a). In addition, the average values of Rrms and Rgeom of exosome sample determined with 480 μm spacer were smaller from the corresponding values determined with 350 μm spacer, which provided fractionation of the entire sample under the operating cross-flow (Figure 5b). In summary, the optimal method for exosome fractionation by AF4 includes the spacer with 350 μm thickness, the channel flow rate of 1 mL/min, the focus flow rate of 1.5 mL/min, the focusing time prior elution step of 7 min, and the use of two successive, linear cross-flow gradients, that is, high initial crossflow gradient of 0.55 mL/min2 (cross-flow decreasing from 3.0 to 0.25 mL/min in 5 min) that allows good sample separation from the void peak, followed by the lower cross-flow gradient of 0.0036 mL/min2 (cross-flow decreasing from 0.25 to 0.09 mL/ min in 45 min) that enables fractionation of the entire sample 9229


Article


of significantly diverse particle size, corresponding to smaller and larger exosome populations. Exosomes of small size, hardly visible at low magnification (Figures 7a and 7b, marked with red narrow arrowheads), appeared as vesicles with physical diameter of about 15 nm (Figure 7c, marked with red arrowheads). Larger variety of exosomes exhibited more irregularly shaped vesicles (Figures 7a and 7b, marked with green arrows). Importantly, aggregates of differentially stained particles of various size (from 20 to 100 nm in diameter at the largest particle dimension) and shapes were also detected by TEM (Figure 7a, marked with yellow arrowheads). Their identity is unknown, but could represent the aggregated bovine serum albumins, which are routinely added to the growth media for maintaining cells and are a known contaminant of vesicle preparations,38 or the antibodies, which are normally released from human lymphoblastoid B cell line used for the production of exosomes in the present study. Thus, it seems that high intensity UV signal of the early eluting peak in AF4/UV-MALS fractogram of exosome sample is a cumulative effect of coeluted small size exosomes and protein impurities that absorb UV light with absorbance maxima at 280 nm, which was also the wavelength of the UV detector connected to AF4. Comparison of AF4-MALS, DLS, and NTA Results. The radii of two different particle populations detected by AF4MALS were determined from the angular variation in scatteredlight intensity and are expressed as the root-mean-square radius (Rrms) or the geometric radius (Rgeom). The Rgeom was determined using a sphere model since the shape of the exosomes in suspension is spherical.30 The average size of smaller particles, consisting of small vesicles and impurities which coeluted in AF4 fractogram under the peak with apex at 23.7 min, was determined to be 23 (24) nm for Rrms (Rgeom), whereas that of the larger exosome subtype eluted under the peak with apex at 33.8 min, was 113 (127) nm for Rrms (Rgeom) (Figure 6). Previously, similar size distribution with average radii of 15 and 81 nm was observed for the exosomes isolated from the aggressive mouse melanoma cell culture line.29 The average hydrodynamic radius, Rh, of unfractionated exosomes was determined also by DLS in batch-mode at 90° angle using CONTIN algorithm, and NTA. A comparison between the radii obtained by AF4-MALS, batch-DLS and NTA is reasonable since the vesicle’s scattering mass is distributed on a thin shell surrounding the spherical core, and consequently, the Rgeom determined by MALS and the Rh determined by DLS or NTA should be comparable.39

under the operating cross-flow. This method results in successful fractionation of exosomes as demonstrated by a typical fractogram shown in Figure 6.

Figure 6. Normalized AF4-MALS (solid line) and AF4-UV (dashed line) fractograms of exosome sample in PBS buffer, pH = 7.4, together with geometric radius, Rgeom (red, filled crcles), and root-mean-square radius, Rrms (blue, empty circles). The fractionation conditions were: 350 μm spacer, channel flow rate of 1 mL/min, focus flow rate of 1.5 mL/min, focusing time prior elution step of 7 min, a high initial crossflow gradient of 0.55 mL/min2 (cross-flow decreased from 3.0 to 0.25 mL/min in 5 min) followed by a lower cross-flow gradient of 0.0036 mL/min2 (cross-flow decreased from 0.25 to 0.09 mL/min in 45 min). The asterisk (*) designates switching from elution to elution + injection step to flush the channel and injection loop.

The elution profile of exosome sample recorded by the LS detector displayed a broad size distribution, with two not well resolved peaks with maxima at elution time of 23.7 and 33.8 min (Figure 6). The intensity of LS signal of the early eluting peak was much lower as compared to that of the later eluting peak, indicating the presence of particles of significantly diverse size, as shown also by their measured radii. On the contrary, the intensity of the UV signal was very high in the region corresponding to early eluting peak, but it significantly reduced for the later eluting peak. Similar results were previously reported for exosomes isolated from the aggressive mouse melanoma cell culture line.29 TEM Microscopy. To clarify the AF4/UV-MALS results the particles in exosome sample were visualized using TEM and negative staining method (Figure 7). As already observed by AF4-MALS analysis, TEM examination confirmed the presence

Figure 7. TEM micrographs of exosome sample were taken at various magnifications (see scale bars). (A) Large vesicle (green arrows) and many small vesicles (red narrow arrowheads) which are hardly detect at low magnification; some other structures might represent protein/protein aggregates and are seen in background (yellow arrowheads). (B) One large (green arrow) and two small (red arrowheads) vesicles. (C) Structure and size of small vesicles (red narrow arrowheads) were determined at higher magnification. 9230


Article


for the smaller than for the larger particles, which indicated higher concentration of smaller particles (Figure 8b). By calculating the parameter ρ (ρ = Rrms/Rh) from the values of Rrms (MALS) and Rh (DLS) we inferred on the exosome shape and internal structure.40,41 The ρ value for the larger exosome subtype was determined to be 0.84 (Table 2). This value is slightly higher from ρ = 0.75, which is characteristic of a solid sphere structure of uniform density. Based on these results the larger particles correspond to the exosomes containing denser internal cargo, for example, nucleic acids and proteins, which resulted in a solid sphere structure of nonuniform density. The shape factor could not be determined for the smaller particle population since its composition was found to be nonuniform. The results of size characterization of exosome sample by AF4-MALS and DLS were compared to the results obtained by an alternative NTA technique, which enables determination of the average Rh of particles, as well as the number of particles per mL. In contrast to AF4-MALS and DLS (Figures 6 and 8), NTA analysis of the unfractionated exosomes revealed only one broad (from 50 to 450 nm), asymmetrical peak with the shoulder prone toward larger particle size (Figure 9). The

The correlation function obtained from DLS was bimodal with two considerably different average values of relaxation times, which pointed to the presence of two particle subpopulations in exosome suspension (Figure 8), as already

Figure 8. (a) Normalized intensity correlation function, |G2(t)|, of the scattered light intensity and (b) the corresponding size distribution by intensity (black, solid curve) and number (red, dotted curve) for exosome sample measured by batch-DLS at scattering angle of 90°.

Figure 9. Size distribution and number of particles per mL for the exosome sample obtained by NTA.

observed by AF4-MALS. The identified fast and slow relaxation modes were associated with the two types of scattering particles, i.e. the smaller particles with the average size of ∼23 nm and the larger ones with the average size of ∼135 nm (Table 2). An intensity correlation function, normalized between one and zero, (|G2(t)|), and the corresponding distribution of Rh by intensity and number are shown in Figure 8. The peak for the Rh distribution by number is much higher

average hydrodynamic size determined by NTA was 152 ± 38 nm, which is, within the experimental error and considering the NTA detection limit, a good approximation to the average Rh value determined for larger exosome subpopulation by DLS (135 nm). The number of particles per mL in the exosome sample was determined by AF4-MALS and NTA (Figures 9 and 10). The number density of larger exosome subpopulation determined by AF4-MALS was ∼1.1 × 1010 particles/mL (Table 2 and Figure 10). Determination of particle number density for the smaller exosome subpopulation was not possible since it coeluted with protein impurities, having different refractive index value. With NTA it was determined to be ∼2.87 × 1010 particles/mL in the whole exosome standard (Figure 9). This value is roughly comparable to the AF4-MALS value for the larger exosome subpopulation, indicating that particles of smaller size (below 50 nm) were largely not taken into account in calculation by NTA. The obtained results are not surprising considering the detection limit of NTA and the fact that technique is highly dependent on the human factor since optimization of several parameters (e.g., shutter/gain, minimum

Table 2. Size, Shape, and Number Density of Particle Subpopulations Determined by AF4-MALS (Rgeom, Rrms) and Batch-DLS (Rh) Rgeoma (nm)

Rrmsb (nm)

Rh (nm)

ρ (Rrms/Rh)

method

MALS

MALS

DLS

MALS and DLS

MALS

small particles large particles

24

23

23

127

113

135

0.84

1.1 × 1010

number of particles/mL

a

Rgeom was determined using a sphere model. bFor small and large particles, the Zimm 1st and the Debye 3rd models, respectively, were used. 9231


Article


(2) Hood, J. L.; Wickline, S. A. Wiley Interdiscip. Rev. Nanomed. Nanobiotechnol. 2012, 4, 458−467. (3) Hurley, J. H.; Boura, E.; Carlson, L. A.; Rozycki, B. Cell 2010, 143, 875−887. (4) Zhang, H. G.; Grizzle, W. E. Clin. Cancer Res. 2011, 17, 959−964. (5) Van den Boorn, J. G.; Dassler, J.; Coch, C.; Schlee, M.; Hartmann, G. Adv. Drug Delivery Rev. 2013, 65, 331−335. (6) Simons, M.; Raposo, G. Curr. Opin. Cell Biol. 2009, 21, 575−581. (7) Lee, T. H.; D’Asti, E.; Magnus, N.; Al-Nedawi, K.; Meehan, B.; Rak, J. Semin. Immunopathol. 2011, 33, 455−467. (8) Van der Pol, E.; Böing, A. N.; Harrison, P.; Sturk, A.; Nieuwland, R. Pharmacol. Rev. 2012, 64, 676−705. (9) Gyorgy, B.; Szabo, T. G.; Pasztoi, M.; Pal, Z.; Misjak, P.; Aradi, B.; et al. Cell. Mol. Life Sci. 2011, 68, 2667−2688. (10) Vlassov, A. V.; Magdaleno, S.; Setterquist, R.; Conrad, R. Biochim. Biophys. Acta, Gen. Subj. 2012, 1820, 940−948. (11) El Andaloussi, S.; Mäger, I.; Breakefield, X. O.; Wood, M. J. Nat. Rev. Drug Discovery 2013, 12, 347−357. (12) Van der Pol, E.; Coumans, F.; Varga, Z.; Krumrey, M.; Nieuwland, R. J. Thromb. Haemostasis 2013, 11, 36−45. (13) Hood, J. P.; Pan, H.; Lanza, G. M.; Wickline, S. A. Lab. Invest. 2009, 89, 1317−1328. (14) Hood, J. L.; San, R. S.; Wickline, S. A. Cancer Res. 2011, 71, 3792−3801. (15) Dragovic, R. A.; Gardiner, C.; Brooks, A. S.; Tannetta, D. S.; Ferguson, D. J.; Hole, P.; et al. Nanomedicine 2011, 7, 780−788. (16) Sokolova, V.; Ludwig, A. K.; Hornung, S.; Rotan, O.; Horn, P. A.; Epple, M.; Giebel, N. Colloids Surf., B 2011, 87, 146−150. (17) Müller, G. J. Bioanal. Biomed. 2012, 4, 46−60. (18) Filipe, V.; Hawe, A.; Jiskoot, W. Pharm. Res. 2010, 27, 796−810. (19) Wahlund, K. G.; Giddings, J. C. Anal. Chem. 1987, 59, 1332− 1339. (20) Giddings, J. C. Sep. Sci. 1966, 1, 123−125. (21) Litzen, A.; Wahlund, K. G. J. Chromatogr. 1989, 476, 413−421. (22) Wahlund, K. G.; Litzen, A. J. Chromatogr. 1989, 461, 73−87. (23) Yohannes, G.; Sneck, M.; Varjo, S. J. O.; Jussila, M.; Wiedmer, S. K.; Kovanen, P. T.; Oeoerni, K.; Riekkola, M. L. Anal. Biochem. 2006, 354, 255−265. (24) Wittgren, B.; Wahlund, K. G. J. Chromatogr. 1997, A760, 205− 218. (25) Wei, Z.; Mcevoy, M.; Razinkov, V.; Polozova, A.; Li, E.; CasasFinet, J.; Tous, G. I.; Balu, P.; Pan, A. A.; Mehta, H.; Schenerman, M. A. J. Virol. Methods 2007, 144, 122−132. (26) Chuan, Y. P.; Fan, Y. Y.; Lua, L.; Middelberg, A. P. J. Biotechnol. Bioeng. 2008, 99, 1425−1433. (27) Oh, S.; Kang, D.; Ahn, S. M.; Simpson, R. J.; Lee, B. H.; Moon, M. H. J. Sep. Sci. 2007, 30, 1082−1087. (28) Kang, D.; Oh, S.; Ahn, S. M.; Lee, B. H.; Moon, M. H. J. Proteome Res. 2008, 7, 3475−3480. (29) Petersen, K. E.; Manangon, E.; Hood, J. L.; Wickline, S. A.; Fernandez, D. P.; Johnson, W. P.; Gale, B. K. Anal. Bioanal. Chem. 2014, 406, 7855−7866. (30) Arraud, N.; Linares, R.; Tan, S.; Gounou, C.; Pasquet, J. M.; Mornet, S.; Brisson, A. R. J. Thromb. Haemostasis 2014, 12, 614−627. (31) Wyatt, P. J.; Weida M. J. U.S. Patent 6,774,994 B1. (32) Gardiner, C.; Ferreira, Y. J.; Dragovic, R. A.; Redman, C. W. G.; Sargent, I. L. J. Extracell. Vesicles 2013, 2, 19671. (33) Berne, B.J.; Pecora, R. Dynamic Light Scattering; Wiley: New York, 1976. (34) http://s-provencher.com/index.shtml (accessed Jan 26, 2011). (35) Hole, P.; Sillence, K.; Hannell, C.; Maguire, C. M.; Roesslein, M.; Suarez, G.; Capracotta, S.; Magdolenova, Z.; Horev-Azaria, L.; Dybowska, A.; Cooke, L.; Haase, A.; Contal, S.; Manø, S.; Vennemann, A.; Sauvain, J.-J.; Staunton, K. C.; Anguissola, S.; Luch, A.; Dusinska, M.; Korenstein, R.; Gutleb, A. C.; Wiemann, M.; Prina-Mello, A.; Riediker, M.; Wick, P. J. Nanopart. Res. 2013, 15, 2101. (36) Hupfeld, S.; Ausbacher, D.; Brandl, M. J. Sep. Sci. 2009, 32, 1465−1470. (37) Litzen, A. Anal. Chem. 1993, 65, 461−470.

Figure 10. AF4-MALS fractogram of exosome sample and calculated amount of particles (number density of particles per mL) for larger exosome population as a function of elution time.

particle size expected, detection threshold, capture duration) are required by the operator and can affect the final results to a large extent.18,35

■

CONCLUSIONS Crucial experimental parameters influencing exosome separation efficiency in AF4 were found to be the cross-flow velocity and the channel thickness, whereas the focusing time had the least influence. Optimal fractionation was obtained by the channel thickness of 350 μm, the channel flow rate of 1 mL/ min, the focus flow rate of 1.5 mL/min, the focusing time prior elution step of 7 min, and the use of two successive, linear cross-flow gradients, that is, high initial gradient of 0.55 mL/ min2 that allows good separation of exosome from the void peak, followed by lower cross-flow gradient of 0.0036 mL/min2 that enables fractionation of the entire sample under the operating cross-flow. AF4/UV-MALS and DLS results displayed the exosome sample to have broad size distribution and pointed to the presence of two particles subpopulations, that is, the larger exosomes and the smaller vesicle-like particles, which coeluted in AF4 together with impurities in early eluting peak. The larger exosomes have solid sphere structure of nonuniform density. The Rh and number density of exosomes determined by NTA revealed consistency in terms of size and number density determination of larger exosome population by DLS and AF4MALS, respectively.

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Phone: 00386 1 47 60 203. Fax: 00386 1 47 60 300. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS The authors gratefully acknowledge the financial support of the Ministry of Higher Education, Science and Technology of the Republic of Slovenia, and the Slovenian Research Agency (Project J3-5499 and Programs P2-0145, P1-0170 and P40165).

■

REFERENCES

(1) Schorey, J. S.; Bhatnagar, S. Traffic 2008, 9, 871−881. 9232


Article

Analytical Chemistry (38) Witwer, K. W.; Buzas, E. I.; Bemis, L. T.; Bora, A.; Lasser, C.; Lotvall, J.; Nolte-'tHoen, E. N.; Piper, M. G.; Sivaraman, S.; Skog, J.; Thery, C.; Wauben, M. H.; Hochberg, F. J. Extracell. Vesicles 2013, 2, 20360. (39) Jin, A. J.; Huster, D.; Gawrisch, K.; Nossal, R. Eur. Biophys. J. 1999, 28, 187−199. (40) Burchard, W.; Schmidt, M.; Stockmayer, W. H. Macromolecules 1980, 13, 1265−1272. (41) Tanford, C. Physical Chemistry of Macromolecules; Wiley & Sons: New York, 1967.

9233


Size Characterization and Quantification of Exosomes by

Recommend Documents