Label-free imaging and biomarker analysis of exosomes with plasmonic scattering microscopy

Exosome analysis is a promising tool for clinical and biological research applications. However, detection and biomarker quantification of exosomes is technically challenging because they are small and highly heterogeneous. Here, we report an optical approach for imaging exosomes and quantifying their protein markers without labels using plasmonic scattering microscopy (PSM). PSM can provide improved spatial resolution and distortion-free image compared to conventional surface plasmon resonance (SPR) microscopy, with the signal-to-noise ratio similar to objective coupled surface plasmon resonance (SPR) microscopy, and millimeter-scale field of view as a prism-coupled SPR system, thus allowing exosome size distribution analysis with high throughput. In addition, PSM retains the high specificity and surface sensitivity of the SPR sensors and thus allows selection of exosomes from extracellular vesicles with antibody-modified sensor surfaces and in situ analyzing binding kinetics between antibody and the surface protein biomarkers on the captured exosomes. Finally, the PSM can be easily constructed on a popular prism-coupled SPR system with commercially available components. Thus, it may provide an economical and powerful tool for clinical exosome analysis and exploration of fundamental issues such as exosome biomarker binding properties.

a, Bright field, PSM and SPR images of the A431 cells in the prism coupled PSM system. Because the laser is used in this system to achieve high incident intensity for PSM, the interference effect covers the SPR images, making it only suitable for ensemble SPR measurement. b, SPR images of the A431 cells in another optimized prism coupled SPR imaging system. It can be seen that the PSM provides a high spatial resolution to image the cell adhesion sites, which has been revealed by total internal reflection fluorescence microscopy (Journal of cell science 2010, 123(21), 3621-3628). These results show that the PSM can provide higher spatial resolution than traditional SPR imaging systems. Fig. S3 a, Experimental SPR curve and PSM scattering curve. The intensity is achieved by averaging the intensities of all pixels in the raw image. b, PSM intensity response during changing the PBS buffer to 80% PBS buffer in water, where the PSM intensity variation is ~5.18 grayscales. The standard deviation σ of PSM intensity measuring PBS buffer is ~0.063 grayscale. The refractive index variations between PBS buffer and 80% PBS buffer can be estimated to be (46 mDeg)/(130 Deg/RIU) ~ 3.54 × 10 -4 RIU, where 46 mDeg is the ensemble SPR intensity difference between 100% and 80% PBS buffer (Nat Methods 2020, 17, 1010-1017, 130 Deg/RIU is the ensemble SPR sensitivity factor, and RIU represents refractive index unit. Then, the sensitivity factor (SF) of PSM channel can be estimated by (5.18 grayscales)/( 3.54 × 10 -4 RIU) ~ 1.46 × 10 4 grayscales/RIU. Finally, the refractive index resolution of PSM for ensemble measurements can be determined to be σ /SF = (0.063 grayscale)/( 1.46 × 10 4 grayscales/RIU) ~ 4.3 × 10 -6 RIU, which is comparable to most ensemble SPR sensors (Chemical Reviews 2008, 108 (2), 462-493). c, SPR intensity response during changing the PBS buffer to 80% PBS buffer in water, where the PSM intensity variation is ~6.02 grayscales. The standard deviation σ of SPR intensity measuring PBS buffer is ~0.11 grayscale. Using the same protocol as Fig. S3b, the sensitivity factor SF of SPR channel can be estimated by ~ 1.70 × 10 4 grayscales/RIU, and the refractive index resolution of SPR channel can be estimated to be ~ 6.4 × 10 -6 RIU, which is comparable to most ensemble SPR sensors (Chemical Reviews 2008, 108 (2), 462-493).

Fig. S4
Size distributions of extracellular vesicles from different cells measured by nanoparticle tracking analysis instrument (NanoSight NS300, Malvern Panalytical, Malvern, UK). The solid lines are Gaussian fittings. The EV sample was diluted before measurement. The dilution factor for achieving ~50 vesicles in one frame and the mean diameter of the vesicles are marked in the figures.

Fig. S5
Ensemble PSM and SPR measurement of 5 × 10 10 mL -1 HeLa EVs binding to the goat anti mouse IgG antibody. Compared with Fig. 2, we can see that the nonspecific binding is very weak.

Fig. S6
Ensemble PSM and SPR measurement of 5 × 10 7 mL -1 HeLa EVs binding to the anti-CD63 antibody. The curve is hard to fit because no obvious dissociation was observed within the measurement period.

Fig. S7
Ensemble measurement of 5 × 10 10 mL -1 HeLa EVs binding to the low-density anti-CD63 antibody. The low-density anti-CD63 modified sensor surface by incubating the gold surface with the solution mixing 20 nM BSA with 20 nM anti-CD63antibodies.

Note S1. Effective diameter correction
The surface plasmon field decreases exponentially from the surface (z-direction) into the solution. In other words, the scattering of the evanescent field by a finite-size object depends on the distance (z) from the surface. The effective scattering diameter D eff and volume V eff of the analyte can be given by where z is the distance from the gold surface, R is the radius of the analyte, D is diameter of analyte, and l = 100 nm is the decay length of the evanescent field ( Fig. S4;

Note S2. Signal-to-noise ratio analysis
To estimate the theoretical singal-to-noise ratio (SNR) limit for the PSM system, the total Rayleigh scattering intensity I total of one small object can be estimated by where n s and n m are the refractive indices of analyte and medium, λ is the incident wavelength, d is the analyte diameter, P is the incident light intensity, t is the average period. For the polystyrene nanoparticles measured by PSM here, P = 4 W cm -2 , n s = 1.58, n m = 1.33, λ = 660 nm, and t = 0.1 s. Considering the single photon energy of ~1.2398/(0.66 µm) eV and the 30 x intensity enhancement of surface plasmon field, the total scattering intensity of one object in the PSM system can be expressed as = 2 × 10 -5 × ( ( )) 6 ℎ .
( 3) The objective collects the scattering photons in perpendicular to the propagation direction of surface plasmon wave, and the collection efficiency can be calculated with the equation in spherical coordinate system of where θ and φ are the polar angle and azimuthal angle, respectively. The objective collection angle for the PSM can be calculated by = n ( = 1.33 ) , where NA is the objective numerical aperture, and n m = 1.33 is used to correct the effect of water refraction on the scattering light collection. For the objective with NA of 0.28, the collection efficiency is calculated to be ~1.1 %. For one polystyrene nanoparticle with effective diameter of 80.7 nm, the objective can collect ~60766 scattering photons, which can be converted to 32814 electrons after considering the camera quantum efficiency of 54% at incident wavelength of 660 nm. The camera sensitivity is ~2.4 electrons/grayscale. Thus, one polystyrene nanoparticle with real diameter of 93.7 nm and effective diameter of 80.7 nm can produce the total intensity of ~13672 grayscales in the image, agreeing with the experimental results of 13040 grayscales. The standard deviation of intensities of images recorded in the absence of nanoparticles is ~90 grayscales. Thus, the SNR of our system measuring 93.7 nm polystyrene nanoparticle is determined to be ~145.

Note S3. Refractive index correction
The extracellular vesicles, including the exosomes, own the refractive index of 1.39 (Journal of Extracellular Vesicles 2014, 3, 25361), and polystyrene nanoparticles own the refractive index of 1.58 (refractiveindex.info). Based on the equation S2, the extracellular vesicles will have ~16 times smaller intensity than the polystyrene nanoparticles with the same diameter.

Note S4. Exosome concentration estimation
On the basis of Fick's law of diffusion, one could expect the binding frequency of analytes decreased with time, as described in the classical Cottrell equation where f(t) is the binding frequency as a function of time, A is the area of the observation region, D and C are the diffusion coefficient and the concentration of nanoparticles, respectively (Anal. Chem. 2016Chem. , 88, 2380Chem. -2385. After integration, one can estimate the number of total binding events F(t) to be ( ) = 2 0.5 0.5 0.5 .#( 7) The analyte diffusion coefficient can be estimated with Stokes-Einstein equation where k B is the Bolzmann constant, T is temperature, η is the viscosity of the liquid, r is the hydrodynamic radius. For exosomes with diameter of ~100 nm, the diffusion coefficient can be estimated to be 4.9 × 10 -12 m 2 /s.