Effect of Sample Concentration on Nanoparticle Tracking Analysis of Small Extracellular Vesicles and Liposomes Mimicking the Physicochemical Properties of Exosomes

Abstract

For quantitative analysis, data should be obtained at a sample concentration that is within the range of linearity. We examined the effect of sample concentration on nanoparticle tracking analysis (NTA) of small extracellular vesicles (sEVs), including exosomes, by comparing NTA results of sEVs with those obtained for polystyrene nanoparticles (PSN) and liposomes, which mimic lipid composition and physicochemical properties of exosomes. Initially, NTA of PSN at different concentrations was performed and the particle sizes determined were validated by dynamic light scattering. The major peak maxima for PSN mixtures of different sizes at the higher particle numbers were similar, with some fluctuation of the minor peak maxima observed at the lower particle number, which was also observed for sEVs. Sample concentration is critical for obtaining reproducible data for liposomes and exosomes and increasing the sample concentration caused an increase in data variability because of particle interactions. The inter-day repeatability of particles sizes and concentration for sEVs were 7.47 and 4.51%, respectively. Analysis of the linearity range revealed that this was narrower for sEVs when compared with that of liposomes. Owing to the use of liposomes that mimic the lipid composition and physicochemical properties of exosomes and proteinase-treated sEVs, it was demonstrated that these different analytical results could be possibly caused by the protein corona of sEVs. Consideration of the sample concentration and linearity range is important for obtaining reproducible and reliable data of sEVs.

Introduction

The development of nanotechnology-related medicines has increased over recent years.^1,2) Encapsulation of larger molecules, such as nucleic acids, as well as small chemical entities, into nanoparticles is an attractive approach for delivering materials to cells and has attracted growing interest, e.g., as demonstrated recently for the delivery of severe acute respiratory syndrome coronavirus2 (SARS-CoV2) mRNA using lipid nanoparticles.^3–6)

In addition to the use of artificial nanomedicines such as lipid nanoparticles, e.g., liposomes, naturally occurring nanoparticles in humans, small extracellular vesicles (sEVs) including exosomes,^7–9) have attained significant attention as potential new therapeutics.^10,11) sEVs encapsulate various transmitters such as microRNA (miRNA), mRNA and proteins, and transport these biomolecules to recipient cells.^8,12,13) For example, tumor-derived exosomes play multiple roles in tumor growth and metastasis,¹⁴⁾ and reports have shown that exosome production in cancerous plasma increases significantly when compared with that of healthy samples.^15–18) The number of CD9/HER2 double-positive exosomes have been reported to increase significantly in sera of patients with breast cancer and ovarian cancer when compared with patient sera from healthy donors or patients with glaucoma or interstitial lung disease/pulmonary fibrosis, as determined by ExoCounter, which combines the properties of nanobeads with an optical disc technology.¹⁵⁾ The number of HER2-positive exosomes have been shown to increase in samples from breast cancer patients by use of the nano-plasmonic exosome (nPLEX) assay, which is based on (i) transmission surface plasmon resonance through periodic nanohole arrays,¹⁶⁾ (ii) a surface plasmon resonance platform ¹⁷⁾ or (iii) a current electrohydrodynamic method.¹⁸⁾ These results indicate that sEVs are suitable makers to monitor particular diseases.¹⁹⁾ Therefore, determining sEV levels should provide valuable information about the disease status of patients. Improving the method to purify sEVs and determining the exact properties of sEVs is required when establishing an analytical method to measure sEV levels.²⁰⁾

Nanoparticle tracking analysis (NTA) is an attractive tool to characterize sEVs because particle number and size distributions can be measured.²¹⁾ NTA is a method that captures the movement of individual nanoparticles using optical video images, calculates the diffusion coefficient and then uses the diffusion coefficient in combination with the Stokes–Einstein equation to calculate the hydrodynamic diameters of the sEVs.²¹⁾ Dynamic light scattering (DLS) is an analytical method that measures the same physical property, i.e., hydrodynamic diameter, as NTA.²²⁾ Because DLS uses fluctuations of scattered light to calculate the size of molecules, the results of DLS analysis are very sensitive to the presence of large particles because the intensity of the scattered light is proportional to the square of the volume of the particles.²³⁾ NTA provides better resolution of the size distribution when compared with that of DLS because of the different detection approaches used.²⁴⁾ Moreover, NTA quantifies the nanoparticles that are analyzed.

Artificial beads have been used to examine factors that affect NTA,^25,26) and there are several reports that have addressed device-dependent and -independent sources of NTA measurement error and uncertainty.^{22,25,27–30)} Most of these studies use polystyrene nanoparticles (PSN) as artificial beads. However, sEVs are complex soft lipid nanoparticles that are multimodal in size and the use of polystyrene nanoparticles as a model of sEV is not necessarily ideal. Thus, using a model material with similar characteristics to sEVs is required for determining factors that affect NTA of sEVs. We have recently developed liposomes that mimic the lipid composition and physical properties, e.g., size, zeta potential and membrane rigidity, of naturally occurring exosomes to clarify physicochemical factors that affect the cellular internalization of exosomes.³¹⁾ Therefore, in this study, we used liposomes as a model of exosomes to standardize NTA.

In this study, we examined the effects of sample concentration on particle size and particle number. The resulting data should be procured at a sample concentration within the linearity range for quantitative analysis.³²⁾ The effect of sample concentrations on the analysis of PSN was examined initially. We then analyzed PSN mixtures of different particle sizes. Subsequently, liposomes, which mimic naturally occurring exosomes, were examined at different concentrations. Finally, the effect of sample concentration on the reliability of sEV analysis was studied. Based on the results for PSN, liposomes and sEVs, aspects that need to be considered when analyzing sEVs by NTA were identified.

Experimental

Materials

A stock suspension of PSN (certified reference material (CRM) 5701-a, with a certified mean diameter and uncertainty of 118.5 ± 1.8 nm as measured by DLS, The National Metrology Institute of Japan) was purchased from FUJIFILM Wako Pure Chemical Corporation (Osaka, Japan). PSN (200 nm) were also purchased from Thermo Fisher Scientific K.K. (Waltham, MA, U.S.A.). 1,2-Distearoyl-sn-glycero-3-phosphocholine (DSPC) and 1,2-dioleoyl-sn-glycero-3-phospho-L-serine (sodium salt) (DOPS) were purchased from Avanti Polar Lipids (Birmingham, AL, U.S.A.). Cholesterol (Chol) was obtained from Sigma-Aldrich (St. Louis, MO, U.S.A.).

Liposome Preparations

Liposomes that mimicked the physical characteristics of exosomes were fabricated based on our previous report.³⁰⁾ The lipid composition (mol%) of liposomes was DSPC/Chol/DOPS (40/40/20 mol%) and were prepared using a modified Bangham method.³³⁾ Briefly, the desired amounts of lipids were mixed in chloroform and dried by evaporation at 70 °C to create a thin homogeneous lipid film. The dried film was hydrated with 0.5 mL of 5% (w/w) aqueous glucose solution under mechanical agitation at 70 °C for 5–10 min. The hydrated lipid solutions were freeze-thawed five times using a dry ice-methanol slush (–78 °C) and a water bath (70 °C). The solution was then passed 21 times through a Mini Extruder (Avanti Polar Lipids) equipped with a 100 nm polycarbonate filter. The total lipid content was calculated by using an enzymatic assay kit that uses choline oxidase [Phospholipid C-test from Wako Pure Chemical Corporation].

Extraction of sEVs from HepG2 Cells, HeLa Cells, and K562 Cells

HepG2 cells (American Type Culture Collection, Manassas, VA, U.S.A.) and HeLa cells (Health Science Research Resources Bank, Osaka, Japan) were cultured in advanced Dulbecco’s Modified Eagle’s Medium (DMEM; Thermo Scientific). K562 cells (American Type Culture Collection) was cultured in advanced RPMI 1640 Medium (Thermo Scientific). Cells were grown in a humidified incubator at 37 °C and 5% CO₂.

Secreted sEVs from the HepG2 cell, HeLa cells, and K562 cells culture medium were isolated by a MagCapture™ Exosome Isolation Kit PS (FUJIFILM Wako Pure Chemical Corporation) in accordance with the manufacturer’s instruction.³¹⁾ Briefly, HepG2 cells were cultured in serum-free DMEM (advanced DMEM, Thermo Fisher Scientific) supplemented with Gluta Max (Thermo Fisher Scientific). The conditioned medium was collected after 48 h, centrifuged to eliminate cells and cell debris and filtered through a 0.22 µm membrane. After ultrafiltration with a Vivaspin 20 concentrator (GE Healthcare Life Sciences, Marlborough, MA, U.S.A.), the concentrated sEVs were captured on affinity magnetic beads, washed and eluted. The eluent was filtered through a 0.45 µm filter to remove the magnetic beads. The positive marker proteins CD63, Alix, and Hsp70, and the negative marker protein apolipoprotein B were used to confirm the presence of sEVs^20,34,35) (Supplementary Fig. 1). The sEV concentration was determined by measuring the total protein amount using a protein quantification kit (wide range) from Dojindo Laboratories (Kumamoto, Japan).

Instrument Configuration

A nanosight NS300 (Malvern Instruments Ltd., U.K.) equipped with a 532-nm laser, CMOS camera and syringe pump were used. NTA 2.3 (software build 033) was used for data collection and analysis. All video capture and analysis conditions used (camera shutter, camera gain, detection threshold, flow rate) were identical for all samples in an individual experiment, except when the sample concentration was varied. The sample concentration was calculated using nominal weight per mL, the weight (lipid) per mL, and protein weight per mL for PSN, liposomes, and sEVs, respectively. Here, the camera gain and threshold were adjusted to ensure that particles were clearly recognized. The data acquisition time was 60 s. The sample volume analyzed for NTA using the NS300 was approximately 56 pL.²⁵⁾ Therefore, the number of captured particles was sufficient to ensure the reproducibility error of the analyzed results was minimized. The flow rate of the syringe pump was sufficiently slow to ensure Brownian motions were captured. Although the syringe speed of 30 and 50 did not lead to significant differences in the particle number of liposomes (e.g., 3.30 × 10⁸ and 3.36 × 10⁸ particles/mL for the 0.1 µg/mL liposome sample, respectively), the speed of the syringe did cause differences in the particle number for sEV samples, and the syringe speed of 50 gave smaller particle numbers (e.g., 9.02 × 10⁷ and 6.86 × 10⁷, for 0.1 µg/mL sEVs when using syringe speeds of 30 and 50, respectively). Therefore, we used a syringe speed of 30 for the analysis of sEVs. The PSN were diluted with water to the indicated concentrations. The valid tracks were more than 3000 for PSN and liposome analyses, except when examining sample concentrations. The liposomes were diluted with PBS to the indicated concentrations. The viscosity of the solutions at each temperature was entered into the software. The valid tracks were more than 1100 for sEV analysis.

Measurement of Solvent Viscosity

The viscosity of phosphate buffered saline (PBS) at analytical temperatures was measured according to the Japanese Pharmacopoeia³⁶⁾ using glass viscometers (Shibata Science Technologies, Tokyo, Japan).

DLS Measurements

The hydrodynamic diameters and polydispersity index (PDI) values of the PSN and liposomes were measured at 25 °C by DLS combined with cumulant analysis with a Horiba nanoPartica SZ-100V2 (Horiba, Kyoto, Japan) equipped with software (SZ-100 for Windows, ver. 2.40). Mean values and standard deviations (S.D.) were obtained from three measurements. Sample data used are presented in Supplementary Table 1.

Transmission Electron Microscopy (TEM) Measurement

Approximately, 5 µL of the sample was placed on Parafilm. Then, a carbon-coated 400-mesh copper grid was positioned on top of the drop for 10 s and washed with a droplet of distilled water. The grid was contrasted by adding a drop of 2% uranyl acetate on Parafilm and incubating the grid on top of the drop for 10 s, and the excess liquid was removed gently using an absorbing paper. After drying, the samples were subjected to TEM observation (HITACHI H-7600) at 100 kV.

Results and Discussion

Polystyrene Analysis

Sample Concentration of PSNs

We studied the effect of PSN sample concentration from 0.1 to 10 µg/mL on the mean particle size and particle number (Table 1). Although the relative standard deviation (RSD) of particle numbers at PSN sample concentrations of 0.1 and 0.2 µg/mL were relatively large, i.e., 5.57 and 4.31%, respectively, at other PSN sample concentrations small RSDs were obtained for both the mean particle size and particle number. Interestingly, the mean particle sizes were small at PSN sample concentrations of 5 and 10 µg /mL. This is shown clearly from the representative size distribution at each sample concentration, as shown in Fig. 1a. Weak shoulder peaks were observed at smaller particle sizes for PSN sample concentrations of 5 and 10 µg/mL (Fig. 1a), indicating that small intermolecular distances cause two particles to be recognized as one particle by the camera and the speed of Brownian motion increased by two-fold (Fig. 1a). The linearity of the particle number (particles/mL) against sample concentration (0.1 to 10 µg /mL) is presented in Fig. 1b. Linearity was not observed above a sample concentration of 5 µg/mL (Fig. 1b), which coincided with the appearance of the weak shoulders peaks in the distributions at sample concentrations of 5 and 10 µg/mL. In contrast, the linearity from 0.1–1 µg/mL was good with a correlation coefficient (r) of 0.9934, which is acceptable (Fig. 1c). Thus, based on the RSDs and linearity range we used 1 µg/mL PSN for testing repeatability.

Table 1. The Effect of PSN Sample Concentrations

Sample concentration (µg/mL)	Particle size (nm)		Particle number (particles/mL)
	Ave.	RSD (%)	Ave.	RSD (%)
0.1	109.6	0.64	1.5E + 08	5.57
0.2	107.0	0.97	2.7E + 08	4.31
0.5	107.2	0.44	5.4E + 08	1.99
1	111.2	0.68	8.6E + 08	1.24
5	101.4	0.81	1.6E + 09	0.61
10	101.8	0.90	2.4E + 09	1.04

Sample, 120 nm PSN; n = 3. Camera level, 10; threshold, 7, 8, 9, 14 and 40 for 0.1, 0.2, 0.5, 1 and 5 µg/mL samples, respectively. Camera level, 9; threshold, 45 for the 10 µg/mL sample. Syringe speed, 50.

Fig. 1. The Effect of Sample Concentration on the Size Distribution of PSNs (a) and the Linearity of Particle Numbers against Sample Concentrations (b and c)

Analytical conditions: data acquisition time, 60 s; n = 3; sample, 120 nm PSN; camera level, 10; threshold, 7, 8, 9, 14 and 40 for 0.1, 0.2, 0.5, 1 and 5 µg/mL samples, respectively; camera level 9 and threshold 45 for the 10 µg/mL sample; syringe speed, 50.

We then examined the intra-day and inter-day repeatability under the above optimized conditions. Intra-repeatability of particle size and particle number were 0.24 and 2.05%, respectively, and these values are acceptable. The mean particle diameter calculated by NTA was 109.3 nm, which is smaller when compared with the value of 120 nm obtained by DLS analysis (Supplementary Table 1). This result is reasonable, because, in general, the mean size of an intensity-weighted distribution is larger than the corresponding number-weighted distribution,^23,37,38) and the particle size obtained by NTA yielded a number-weighted distribution whereas DLS data gave an intensity-weighed distribution.

The intra-day repeatability 0.24% for mean particle sizes obtained by NTA was comparable to the repeatability obtained by DLS (RSD = 0.87%). The inter-day repeatability of the mean particle size, 1.22%, was also comparable to results obtained by DLS (0.93%). The intra-day and inter-day repeatability of particle numbers were also acceptable, and the values were 2.05 and 2.88%, respectively. These results indicated that PSN analysis gave similar repeatability when compared with the results obtained by other analytical methods using Brownian motion.

Analysis of Nanoparticle Mixtures with Different Particle Sizes

We then studied the repeatability of the resolution and particle numbers of PSN with different particle sizes because sEVs exist as a mixture of different sizes in vivo. The nominal 120-nm PSN and 200-nm PSN were mixed at a ratio of 4 : 1 (w/w) to give a total sample concentration of 1 µg/mL, which is the optimized concentration, and this sample was analyzed by NTA. Table 2 shows the repeatability results when three consecutive analyses using the same samples were performed. Figure 2 shows the distribution of the mixtures. As shown in Table 2, the ratio of the two different sized particle numbers was 4.25 and the RSD was 3.89%. This repeatability was similar to the result obtained for single-sized particle analysis. The RSDs of the peak maxima for particle sizes were 1.07 and 3.64% (RSD), with some fluctuation of the peak maxima observed at the lower particle numbers. In contrast, the RSD of the mean particle size was satisfactory with an RSD of 0.80%. This fluctuation of the peak maximum at lower particle numbers should be noted when multimodal samples such as sEVs are analyzed.

Table 2. Analysis of PSN Mixture

Ratio of particle numbers		Size at peak maximum (nm) Major peak		Size at peak maximum (nm) Minor peak		Mean particle size (nm)
Ave.	RSD (%)	Ave.	RSD (%)	Ave.	RSD (%)	Ave.	RSD (%)
4.25	3.89	107.8	1.07	183.2	3.64	121.2	0.80

Samples are a mixture of 120 nm PSN (0.8 µg/mL) and 20 nm PSN (0.2 µg/mL). Data acquisition time: 60 s; camera level, 10; threshold, 11; syringe speed, 50. n = 3.

Fig. 2. The Size Distribution of PSN Mixtures

Overlay of three analyses. Samples were mixtures of 120 nm PSNs (0.8 µg/mL) and 20 nm PSNs (0.2 µg/mL). Analytical conditions: data acquisition time, 60 s; camera level, 10; threshold, 11; syringe speed, 50 (n = 3).

Analysis of Liposomes

Effect of Liposome Sample Concentration

Using the knowledge obtained from analysis of standard PSN, we then analyzed liposomes, which were prepared such that the lipid composition and physicochemical properties, e.g., size, zeta potential and rigidity, were similar to those of exosomes.³¹⁾ Although liposomes composed of a single lipid (i.e., POPC³⁸⁾), two lipids (90–120 nm DOPC/CHOL (55 : 45 mol/mol)³⁹⁾) or three components (EPC, DOPE and DOTAP²²⁾) have been examined, there are no reports that have examined liposomes composed of lipids that mimic the lipid composition of exosomes. Thus, in an effort to understand aspects that are important when analyzing sEVs, we prepared liposomes composed of lipids found in exosomes.

Figure 3 shows the size distribution of liposomes examined at different sample concentrations (i.e., 0.1 to 10 µg/mL). At a sample concentration of 10 µg/mL, the peak shapes were broad with a multimodal distribution (Fig. 3). As with PSN, the appearance of a peak at smaller sizes around 50 nm was observed, which may be because of false recognition by the instrument due to small intermolecular distances. Additionally, different from the PSN results, a peak representing a larger species appeared at a sample concentration of 10 µg/mL, which indicates interactions among the liposomes. The intra-repeatability of the analyzed liposomes was acceptable, except at a concentration of 10 µg/mL, for both mean particle size and particle number (Table 3).

Fig. 3. The Effect of Sample Concentration on the Size Distribution of Liposomes (a) and the Linearity of Particle Numbers against the Sample Concentrations (b and c)

Samples, liposomes. Analytical conditions: data acquisition time, 60 s; n = 3. Other analytical conditions were: 0.1 µg/mL liposomes: camera level, 16; threshold, 12; syringe speed, 50. 0.2 µg/mL liposomes: camera level, 16; threshold, 12; syringe speed, 50. 0.5 µg/mL liposomes: camera level, 16; threshold, 19; syringe speed, 50. 1 µg/mL liposomes: camera level, 10; threshold, 8; syringe speed, 50. 4 µg/mL liposomes: camera level, 10; threshold, 10; syringe speed, 50. 10 µg/mL liposomes: camera level, 10; threshold, 10; syringe speed, 50.

Table 3. The Effect of Liposome Concentrations on the Analyzed Results

Sample concentrations (µg/mL)	Particle size (nm)		Particle number (particles/mL)
	Ave.	RSD (%)	Ave.	RSD (%)
0.1	93.6	0.77	2.6E + 08	3.62
0.2	96.5	1.01	3.4E + 08	3.77
0.5	101.1	0.66	8.9E + 08	1.35
1	104.9	0.78	1.7E + 09	1.26
4	117.2	0.40	2.5E + 09	1.42
10	92.7	4.47	3.2E + 09	10.0

Analytical conditions: data acquisition time, 60 s; n = 3. The conditions were: 0.1 µg/mL liposome: camera level, 16; threshold, 12; syringe speed, 50. 0.2 µg/mL liposome: camera level, 16; threshold, 12; syringe speed, 50. 0.5 µg/mL liposome: camera level, 16; threshold, 19; syringe speed, 50. 1 µg/mL liposome: camera level, 10; threshold, 8; syringe speed, 50. 4 µg/mL liposome: camera level, 10; threshold, 10; syringe speed, 50. 10 µg/mL liposome: camera level, 10; threshold, 10; syringe speed, 50.

As with PSN, the mean particle sizes obtained by NTA were slightly smaller than the values obtained by DLS (Supplementary Table 1) at all liposome sample concentrations examined. As stated above in “Sample Concentration of PSNs,” this result is rational, because the mean size of an intensity-weighted distribution is larger than the corresponding number-weighted distribution.^23,37,38) However, when the liposome particles distribution is large (e.g., larger than PDI = 0.2), it is possible that DLS analysis gives values that are smaller than those obtained by NTA because of the lower detection limit of NTA (>30 nm) when compared with that of DLS.²²⁾ The linearity of the observed particle number against sample concentration is shown in Fig. 3b. Linearity between 0.1 to 1 µg/mL (r = 0.9984, Fig. 3b) was observed; however, linearity was not observed above a sample concentration of 1 µg/mL (Fig. 3c). Furthermore, the particle size increased as the sample concentration increased (Table 3), suggesting particle interactions occurred more often at the higher sample concentrations. The particle sizes of PSNs and liposomes at 1 µg/mL PSN were 111.2 nm and 104.9 nm, respectively, and they were similar. However, the linearity from 0.1 to 1.0 mg/mL was better in PSNs than in liposomes. Therefore, it is suggested that factors other than size cause differences in linearity between the results of PSNs and liposomes. However, at higher concentrations, the particle sizes increased, which may have been caused by particle interactions. Therefore, the particle size may indirectly affect the linearity of particle numbers. Interestingly, the mean particle size at a sample concentration of 10 µg/mL was smaller than the mean particle size at other sample concentrations (Table 3), which corroborates the results shown in Table 1 for PSN data.

Analysis of the Inter-Day Repeatability of Liposomes

The inter-day repeatability of particle size and particle number was good at 0.5 µg/mL. Here, linearity was established, and the RSDs were 1.03 and 3.94% for mean particle size and particle number, respectively. These results showed that optimizing the sample concentration should be considered when analyzing particle numbers. Moreover, selecting the appropriate sample concentration ensures that repeatable results are obtained, which enables the quantitative evaluation of liposomes and comparison between particular group samples.

Analysis of sEVs

We then analyzed sEVs extracted from HepG2 cells.³¹⁾ Methods to quantify lipids are typically insensitive for determining small amounts of sEVs, unless specialized techniques such as total reflection Fourier-transform IR spectroscopy are used.²⁰⁾ In general, quantification of sEVs using total protein amounts is a typical alternative approach.²⁰⁾ We analyzed sEVs with the protein concentrations so that the particle numbers become similar to those of liposomes.

The repeatability results showed that the distribution of the main peaks was similar, whereas variation in the distribution of minor peaks was observed, which was also observed for PSN mixtures with different particle sizes (Figs. 2 and 4a). The variability of the particle size of sEVs analyzed by NTA was rational, judging from the TEM-derived image of the sample with a mode value of approximately 60–70 nm (Fig. 4b). The effect of sample concentration was examined. Increasing the sample concentration caused an increase in the variability of particle numbers (Table 4). The mean particle size increased as the sample concentration increased (Table 4). As shown in Fig. 4, the particle distribution for large particle sizes increased as the sample concentration increased, which was presumably caused by particle interactions. In contrast to the PSN and liposome analyses, a new weaker peak with a smaller particle size did not appear at higher sample concentrations, probably because of the detection limitations of NTA.²²⁾ Therefore, in this case, 0.1 or 0.2 µg/mL was a more appropriate concentration to quantify sEVs. The inter-day repeatability of 0.1 µg/mL sEVs was 7.47 and 4.51% for mean particle size and particle numbers (n = 3), respectively, which were larger than those observed for PSN and liposome analyses. Based on the size distribution and reliability of the analyzed data, the linearity range was anticipated to be smaller when compared with the data for liposomes. The linearity between sample concentrations (0.1–1 µg/mL) and observed particle numbers was r = 0.9523 (Fig. 4c), which was inferior when compared with the liposome results (r = 0.9984) (Fig. 3b). We also analyzed sEVs collected from K562 and HeLa cells. The size distribution at 0.1 µg/mL was similar to that of HepG2 cells (Supplementary Fig. 2). Although the correlation coefficient (r) was different depending on the cell type, the coefficients of K562 cells and HeLa cells (0.9860 and 0.9454, respectively) were smaller than those of PSN and liposomes (Supplementary Fig. 2). In this study, we used liposomes that mimic the lipid composition, rigidity, size and zeta potential of exosomes. Therefore, the differences in the analytical linearity observed in this NTA between liposomes and sEVs may be caused by structural differences, e.g., the existence of proteins or sugar chains on the surface of sEVs. In addition to sample interactions, as observed for the liposomes, protein modifications may also promote interactions among sEVs, which possibly reduce the linearity range. This possibility is supported by the data corresponding to the linearity of particle numbers against the sample concentrations of sEVs upon proteinase treatment,⁴⁰⁾ which improved from r = 0.9523 to r = 0.9959, and the linearity is similar to that of liposomes (r = 0.9984) (Supplementary Fig. 3).

Fig. 4. The Effect of Sample Concentrations on the Size Distribution of sEVs Collected from HepG2 Cells (a), TEM-Derived Image of sEVs (b) and the Linearity of Particle Numbers against the Sample Concentrations (c)

Samples, sEVs from HepG2 cells. NTA analytical conditions: data acquisition time, 60 s; n = 3. Other analytical conditions were: 0.1 µg/mL: camera level, 16; threshold, 5; syringe speed, 30. 0.2 µg/mL: camera level, 16; threshold, 12; syringe speed, 30. 0.4 µg/mL: camera level, 16; threshold, 12; syringe speed, 30. 1 µg/mL: camera level, 16; threshold, 12; syringe speed, 30.

Table 4. The Particle Distribution of sEVs Extracted from HepG2 Cells

Sample concentration (µg/mL)	Particle size (nm)		Particle number (particles/mL)
	Ave.	RSD (%)	Ave.	RSD (%)
0.1	84.0	4.14	9.07E + 07	1.27
0.2	99.4	1.41	2.10E + 08	3.17
0.4	106.8	6.41	3.23E + 08	8.96
1	108.2	0.38	4.78E + 08	14.3

Analytical conditions: 0.1 µg/mL: camera level, 16; threshold, 5; syringe speed, 30. 0.2 µg/mL: camera level, 16; threshold, 12; syringe speed, 30. 0.4 µg/mL: camera level, 16; threshold, 12; syringe speed, 30. 1 µg/mL: camera level, 16; threshold, 12; syringe speed, 30.

High concentrations of proteins or colloidal solutions are considered to be a non-Newtonian fluid.⁴¹⁾ In non-Newtonian fluids, diffusion and velocity do not display an inverse correlation. This phenomenon may partially cause an increase in the mean particle size. Therefore, it is important to measure size distributions at low sample concentrations where repeatable and reliable results can be obtained, and the linearity range should be carefully evaluated before quantification.

Conclusion

In this study, we clarified the effect of sample concentration on NTA analysis of sEVs by comparison with standard PSN and liposomes, which mimic the lipid composition and physicochemical properties of exosomes. We first studied the effect of sample concentration using PSN and validated the mean particle size under optimized conditions using DLS. Both for liposome and sEV analysis, an increase in sample concentration caused an increase in the variability of the measured data. Furthermore, the linearity range of particle numbers was narrower for the analysis of sEVs when compared with that of liposomes. Because we used liposomes that mimic the lipid composition and physicochemical properties of exosomes, these differences in analytical results may be caused by the protein corona of sEVs. To the best of our knowledge, the factor that affects the linearity of sEVs has been shown for the first time in this study. In summary, sEVs have gained attention as markers to monitor particular diseases, and NTA is an attractive tool to characterize the particle number and size distribution of sEVs; however, sample concentrations and evaluation of the linearity range must be considered depending on the cell type to obtain reproducible and reliable data.

Acknowledgments

This work was supported in part by the Research on Development of New Drugs from the Japan Agency for Medical Research and Development, AMED (No. 20ak0101074h2204 [K.S.-K.]), and JSPS KAKENHI (Grant No. 19K07039 [K.S.-K.]).

Conflict of Interest

The authors declare no conflict of interest.

Supplementary Materials

The online version of this article contains supplementary materials.

References

• 1) Akinc A., Maier M. A., Manoharan M., Fitzgerald K., Jayaraman M., Barros S., Ansell S., Du X., Hope M. J., Madden T. D., Mui B. L., Semple S. C., Tam Y. K., Ciufolini M., Witzigmann D., Kulkarni J. A., van der Meel R., Cullis P. R., Nat. Nanotechnol., 14, 1084–1087 (2019).
• 2) Oku N., Biol. Pharm. Bull., 40, 119–127 (2017).
• 3) Mulligan M. J., Lyke K. E., Kitchin N., et al., Nature (London), 586, 589–593 (2020).
• 4) Pardi N., Hogan M. J., Pelc R. S., et al., Nature (London), 543, 248–251 (2017).
• 5) Richner J. M., Himansu S., Dowd K. A., Butler S., Salazar V., Fox J. M., Julander J. G., Tang W. W., Shresta S., Pierson T. C., Ciaramella G., Diamond M. S., Cell, 168, 1114–1125.e10 (2017).
• 6) Shin M. D., Shukla S., Chung Y. H., Beiss V., Chan S. K., Oscar A. O.-R., Wirth D. M., Chen A., Sack M., Pokorski J. K., Steinmetz N. F., Nat. Nanotechnol., 15, 646–655 (2020).
• 7) Johnstone R. M., Adam M., Hammond J. R., Orr L., Turbide C. V., J. Biol. Chem., 262, 9412–9420 (1987).
• 8) Mathieu M., Martin-Jaular L., Lavieu G., Théry C., Nat. Cell Biol., 21, 9–17 (2019).
• 9) Yamashita T., Takahashi Y., Takakura Y., Biol. Pharm. Bull., 41, 835–842 (2018).
• 10) Batrakova E. V., Kim M. S., J. Control. Release, 219, 396–405 (2015).
• 11) Tai Y. L., Chen K. C., Hsieh J. T., Shen T. L., Cancer Sci., 109, 2364–2374 (2018).
• 12) Raposo G., Nijman H. W., Stoorvogel W., Liejendekker R., Harding C. V., Melief C. J., Geuze H. J. B., J. Exp. Med., 183, 1161–1172 (1996).
• 13) Wolfers J., Lozier A., Raposo G., Regnault A., Théry C., Masurier C., Flament C., Pouzieux S., Faure F., Tursz T., Angevin E., Amigorena S., Zitvogel L., Nat. Med., 7, 297–303 (2001).
• 14) Khan S., Jutzy J. M. S., Valenzuela M. M. A., Turay D., Aspe J. R., Ashok A., Mirshahidi S., Mercola D., Lilly M. B., Wall N. R., PLOS ONE, 7, e46737 (2012).
• 15) Kabe Y., Suematsu Y., Sakamoto S., Hirai M., Koike I., Hishiki T., Matsuda A., Hasegawa Y., Tsujita K., Ono M., Minegishi N., Hozawa A., Murakami Y., Kubo M., Itonaga M., Handa H., Clin. Chem., 64, 1463–1473 (2018).
• 16) Im H., Shao H., Park Y. I., Peterson V. M., Castro C. M., Weissleder R., Lee H., Nat. Biotechnol., 32, 490–495 (2014).
• 17) Sina A. A., Vaidyanathan R., Dey S., Carrascosa L. G., Shiddiky M. J., Trau M., Sci. Rep., 6, 30460 (2016).
• 18) Vaidyanathan R., Naghibosadat M., Rauf S., Korbie D., Carrascosa L. G., Shiddiky M. J., Trau M., Anal. Chem., 86, 11125–11132 (2014).
• 19) Ramis J. M., Sci. Rep., 10, 8667 (2020).
• 20) Théry C., Witwer K. W., Aikawa E., et al., J. Extracell. Vesicles, 7, 1535750 (2018).
• 21) Malloy A., Carr B., Part. Part. Syst. Charact., 23, 197–204 (2006).
• 22) Filipe V., Hawe A., Jiskoot W., Pharm. Res., 27, 796–810 (2010).
• 23) Hassan P. A., Rana S., Verma G., Langmuir, 31, 1, 3–12 (2015).
• 24) Dragovic R. A., Gardiner C., Brooks A. S., Tannetta D. S., Ferguson D. J. P., Hole P., Carr B., Redman C. W. G., Harris A. L., Dobson P. J., Harrison P., Sargent I. L., Nanomedicines, 7, 780–788 (2011).
• 25) Tian X., Nejadnik M. R., Baunsgaard D., Henriksen A., Rischel C., Jiskoot W., J. Pharm. Sci., 105, 3366–3375 (2016).
• 26) Krueger A. B., Carnell P., Carpenter J. F., J. Pharm. Sci., 105, 1434–1443 (2016).
• 27) Gardiner C., Ferreira Y. J., Dragovic R. A., Redman C. W. G., Sargent I. L., J. Extracell. Vesicles, 2, 19671 (2013).
• 28) Maas S. L., de Vrij J., van der Vlist E. J., Geragousian B., van Bloois L., Mastrobattista E., Schiffelers R. M., Wauben M. H. M., Broekman M. L. D., Nolte-’t Hoen E. N. M., J. Control. Release, 200, 87–96 (2015).
• 29) Matsuura Y., Ouchi N., Nakamura A., Kato H., Phys. Chem. Chem. Phys., 20, 17839–17846 (2018).
• 30) van der Pol E., Coumans F. A. W., Grootemaat A. E., Gardiner C., Sargent I. L., Harrison P., Sturk A., van Leeuwen T. G., Nieuwland R., J. Thromb. Haemost., 12, 1182–1192 (2014).
• 31) Sakai-Kato K., Yoshida K., Takechi-Haraya Y., Izutsu K., Langmuir, 36, 12735–12744 (2020).
• 32) International Council for Harmonisation of Technical Requirements for Pharmaceuticals for Human Use, “VALIDATION OF ANALYTICAL PROCEDURES: TEXT AND METHODOLOGY Q2(R1).”: ‹https://database.ich.org/sites/default/files/Q2%28R1%29%20Guideline.pdf›, cited 25 July, 2021.
• 33) Bangham A. D., Standish M. M., Watkins J. C., J. Mol. Biol., 13, 238–252 (1965).
• 34) Park S. J., Kim J. M., Kim J., Hur J., Park S., Kim K., Shin H.-J., Chwae Y.-J., Proc. Natl. Acad. Sci. U.S.A., 115, E11721–E11730 (2018).
• 35) Zeringer E., Li M., Barta T., Schageman J., Pedersen K. W., Neurauter A., Magdaleno S., Setterquist R., Vlassov A. V., World J. Methodol, 3, 11–18 (2013).
• 36) Ministry of Health, Labour and Welfare, “Japanese Pharmacopoeia The 17th edition,” 2016.
• 37) Hallett F. R., Watton J., Krygsman P., Biophys. J., 59, 357–362 (1991).
• 38) Kim A., Ng W. B., Bernt W., Cho N.-J., Sci. Rep, 9, 2639 (2019).
• 39) Bachurski D., Schuldner M., Nguyen P.-H., Malz A., Reiners K. S., Grenzi P. C., Babatz F., Schauss A. C., Hansen H. P., Hallek M., von Strandmann E. P., J. Extracell. Vesicles, 8, 1596016 (2019).
• 40) Inder K. L., Ruelcke J. E., Petelin L., Moon H., Choi E., Rae J., Blumenthal A., Hutmacher D., Saunders N. A., Stow J. L., Parton R. G., Hill M. M., J. Extracell. Vesicles, 3, 23784 (2014).
• 41) Chhabra R. P., “Rheology of complex fluids,” ed. by Krishnan J. M., Deshpande A. P., Sunil Kumar P. B., Springer Publishing Company, New York, 2010, pp 3–34.