Magnetic beads (CSMN Beads-100, MBs, 100 nm) were purchased from So-Fe Biomedicine (Shanghai, China). The rest of chemical reagents used in the synthesis of nanodevices were purchased from Sinopharm Chemical Reagent Co. (Shanghai, China) and Sigma (Shanghai, China). SUM-1315 cells line was purchased from the Chinese Academy of Sciences (Shanghai, China). DMEM medium (11965092), Penicillin/streptomycin (15140122), and FBS (10099141C) were purchased from GIBCO (Shanghai, China). Anti-CD9 (13,403) primary antibodies were purchased from Cell Signaling Technology (Shanghai, China). Anti-CD63 (ab22595) and anti-calnexin (ab134045) primary antibodies were purchased from Abcam (Shanghai, China). Antirabbit HRP secondary antibodies (a0208), Enhanced-chemiluminescence (ECL) kits (P0018FS), and BCA kit (P0012S) were purchased from Beyotime. Biotechnology (Shanghai, China). CD63 aptamer with azide group was synthesized by Sangon (Shanghai, China), with the following sequence: 5′-CAC CCC ACC TCG CTC CCG TGA CAC TAA TGC TA-3′-N₃ (Zhang Z et al., 2019).

Magnetic nanoparticle solution (600 μl, 5 mg/ml) was added to a 1.5 ml centrifuge tube. The nanoparticles were precipitated to bottom of the centrifuge tube with magnets, and supernatant was removed. After being washed with water three times, the nanoparticles were redispersed in 3 ml of 2-morpholinoethanesulfonic acid (MES) buffer with ultrasound. 1-(3-Dimethylaminopropyl)-3-ethylcarbodiimide hydrochloride (EDC) (dissolved with MES, 2.5 mg/ml, 2 ml) and N-hydroxy succinimide (NHS) (dissolved with MES, 5 mg/ml, 1 ml) were then added, followed by sonication for 15 s. The centrifuge tube was sealed by sealing film and shaken at 37°C for 45 min. After that, the nanoparticles were deposited to the bottom of the centrifuge tube and washed twice with phosphate buffered saline (PBS) buffer. The nanoparticles were redispersed in 6 ml PBS. Compound 5 (the compound codes mentioned in this paper were shown in Figure 2) in dimethyl sulfoxide (DMSO) (60 μl, 100 mg/ml) was added to the centrifuge tube and mixed by ultrasound. The mixture was oscillated at 37°C for 12 h. After removal of the supernatant, the CD63 aptamer with azide group was added into the nanoparticles’ aqueous solution at a 1:1 M ratio. 1 mol% copper sulfate pentahydrate and 5 mol% sodium ascorbates were added and solution was shaken at room temperature for 8 h to get the final magnetic bead system (MBs-Apt₆₃).

SUM-1315 cells tested negative for mycoplasma were cultured in DMEM supplemented with 10% FBS and 1% penicillin−streptomycin. All cells were cultured in an incubator under 5% CO₂ at 37°C.

The aptamer-MBs mixture system was loaded into each well of the agarose gel and separated via electrophoresis. Target oligonucleotides were detected with Gel-Red kits.

Exosomes were extracted from the supernatant of SUM-1315 cells or serum in the following steps: centrifugation was performed at 4°C at 2,000 × g for 10 min, and the supernatant was kept and centrifuged at 10,000 × g for another 30 min to further remove the debris. These samples were transferred to UC tubes and centrifuged at 110,000 × g for 75 min, followed by removal of the supernatant. The precipitates were resuspended and diluted with 1×PBS, and filtered through 0.22 µm membrane. The product was centrifugated again at 110,000 × g for 75 min, and the supernatant was discarded. The pellets were resuspended with 1×PBS and stored at −80°C for further use.

Magnetic beads (0.1 μg) were added to every 500 μl of exosome suspension or serum to capture exosomes. The binding process was carried out on a shaker. After 20 min, the EP tube was put in a tube rack with a permanent magnet placed against the tube bottom. The supernatant was discarded. After 5 min of precipitation, and the pellets were resuspended in PBS and placed under UV light for 20 min. After elution of the exosomes, the bare magnetic beads were removed by permanent magnets after of 5 min’s adsorption time, and the exosome-containing supernatant was retained in 500 μl PBS.

PBS solution (20 µl) containing exosomes was added to a carbon-coated copper grid and adsorbed. After cleaning with PBS, the copper grid was fixed in 1% glutaraldehyde for 2 min. The excess solution was removed by using paper towel to blot the edges of each grid. Then the sample was negatively stained with saturated uranyl acetate solution and incubated at room temperature for 1min. The unevaporated solution is removed with paper towel. The samples were observed under a transmission electron microscope.

Nanoparticle tracking analysis was performed on ZetaVIEW S/N 17-310 and all NTA measurements were performed using the same setup to ensure consistent results. The exosomes were diluted in PBS to obtain 50 particles in each field of vision for optimal counting.

Dynamic light scattering (DLS) measurements were conducted on a Nano ZS instrument (Malvern, United Kingdom). All samples were resuspended in PBS, and measured with the same instrument setup.

Exosome suspension (10 μl) was loaded into SDS-PAGE and then separated by electrophoresis. Proteins were transferred onto polyvinylidene fluoride membranes and blocked with 5% skim milk. The primary anti-CD63 and anti-CD9 antibodies were used, followed by incubation with HRP-conjugated secondary antibodies. Target proteins were detected with enhanced-ECL kits.

BALB/C-nude mice (female, 4 weeks) were purchased from Model Animal Research Center, Nanjing University. The animal experiment was approved by the Ethics Committee of Nanjing Medical University. We confirmed that all animal experiments are complied with National Institutes of Health guide for the care and use of Laboratory animals (NIH Publications No. 8023, revised 1978). The animals were raised at 25°C with 60% humidity. 2 × 10⁶ of sum-1315 cells were inoculated into the back of each nude mouse. 10 days later, blood was collected from the posterior venous cluster of the eyeball. The blood was placed at room temperature for 2 h and centrifuged at 3000 rpm for 10 min at 4°C, and the upper serum was preserved. The serum of nude mice was stored at −80°C.

Exosomes were incubated with 20 µl fluorescently labeled antibody (CD63) at 37°C for 30 min in dark. Then 1 ml of pre-cooled PBS (4°C) was added, followed by centrifugation at 110,000 g for 70 min. The supernatant was carefully removed and 1 ml of pre-cooled PBS (4°C) was added to resuspend. The solution was then centrifuged at 110,000 g at 4°C for 70 min, and the supernatant was carefully removed, followed by resuspension of pellets in 50 µl of precooled PBS (4°C) for analysis.

Exosome suspension was treated with 7 M urea, 2%SDS, and 1× Protease Inhibitor Cocktail. Protein concentration was determined using the BCA protein detection kit, and then the sample was processed with nuclease digestion, reduction/alkylation, acetone precipitation, and lysine-C/trypsin digestion. Desalination was carried out in the monospin column using 0.1% trifluoroacetic acid (TFA) and acetonitrile. After vacuum drying, 0.1%FA was added to redissolve the sample, and a 1–2 μg of sample was analyzed. The separation was performed with Easy-NLC 1000 (Thermo Scientific, United States) using an analytical column (C18, 1.9 μm, 75 μm × 20 cm) at a flow rate of 300 NL/min. Orbitrap Fusion Lumos (Thermo Scientific, United States) was used as a mass spectrometer. Data Dependent Acquisition (DDA) model was used for tandem mass spectrometry. The full-scan resolution was 60,000 (FWHM), the mass-charge ratio range was set to M/Z 350-1800, and the impact energy was set to 30% in HCD fragmentation mode. The original mass spectrometry data were collected and analyzed using the mass informatics platform Proteome Discoverer 2.4 (Thermo Fisher).

Origin 2019 and GraphPad Prism 8.0 were used for statistical calculation of all histograms, and one-way anova was used for calculation of mean and variance.

The 2-nitrobenzyl groups were used as photo-cleavable linkers to anchor CD63 aptamers onto MBs. The detailed synthesis route is shown in Figure 2, and the chemical synthesis steps of photoresponsive ligands are described in Supplementary Material S1. The structure of MBs-Apt₆₃ is shown in Figure 1. Firstly, the photo-cleavable linkers were anchored onto the surface of MBs via a moderate amination reaction. To selectively capture exosomes, CD63 aptamers were finally conjugated with the MBs through click reaction. The final UV-responsive magnetic bead system (MBs-Apt₆₃) was characterized for Zeta potential, infrared spectroscopy, and dynamic light scattering (DLS) size (Supplementary Figure S2 and Supplementary Figure S10).

We measured the UV responsiveness of the magnetic beads with a concentration at the range of 0–10 μg/ml, while the time of exposure to UV light was between 0 and 30 min. In Figures 3A,B, it was found that within the range of 0–10 μg/ml, the light-responsive groups of magnetic beads were highly sensitive to and effectively cleaved by UV light. The CD63 aptamers were rapidly released from MBs-Apt₆₃ under irradiation, which also showed obvious time dependence. With increased exposure time, the amount of released free aptamers increased, and reached the maximum after 15 min of UV exposure.

To test the ability of MBs-Apt₆₃ to capture exosomes, we firstly extracted tumor exosomes from the cultural medium of breast cancer cell line SUM-1315 by UC. Western blotting analysis of exosomes showed the presence of CD9, CD63 (two common surface markers of exosomes) and calnexin (endoplasmic reticulum protein), indicating the successful preparation (Figure 3C). In DLS analysis, the exosomes showed an average size of 56 nm (Figure 3D), in line with characteristic particle size of exosomes (30–150 nm). Finally, we used TEM to observe the morphology of exosomes (Figures 3E,F), which displayed a typical lipid bilayer structure and a size in the range of 50–100 nm.

The process of isolating exosomes from cell supernatant or serum is shown in Figure 4A, which involves beads binding and exosome elution. DLS analysis was performed after MBs binding to exosomes and UV-triggered elution, which showed a size of 250 and 190 nm, respectively. (Figures 4B,C). In Figure 4C, the magnetic bead components were not removed in the system after ultraviolet dissociation. This reduced shift in maximum peak particle size from 256 to 190 nm indicated the elution of exosomes from the MBs after UV exposure. To further elaborate the binding and elution processes, TEM was used to observe the MBs-exosome complex before and after UV-triggered cargo release. Typical MBs bound to many exosomes were clearly seen before exposure to UV light (Figure 4D), whereas only purified exosomes could be observed after UV-induced release of exosomes and removal of magnetic beads by permanent magnets (Figure 4E).

To evaluating the performance of this novel MBs system in isolating serum exosomes, we established mice models of triple-negative breast cancer using human breast cancer SUM-1315 cells. We collected the serum when the tumor grew to about 800 mm³, and then isolated exosomes with MBs-Apt₆₃ and conventional UC respectively. We firstly analyzed the particle size distribution (Figure 5A) and particle concentration of exosomes (Figure 5B) with NTA. Results showed that the particle size distribution of exosomes isolated by MBs-Apt₆₃ and UC was similarly uniform, both of which were about 100 nm. While the serum concentration of exosomes acquired by UC is lower compared with that by MBs-Apt₆₃ (2.5 × 10⁹/ml versus 4.0 × 10⁹/ml) according to the particle count analysis, indicating higher efficiency of MBs-Apt₆₃ in extracting serum exosomes.

Next, we used the high sensitivity flow cytometry (HSFC) system to analyze the expression of CD63 on exosomes separated by UC and MBs-Apt₆₃ (Figure 5C). The results showed that the expression abundance of CD63 was 4.9% in exosomes extracted by MBs-Apt₆₃, which is much higher (0.4%) in exosomes separated by UC method. Therefore, the light strategy in MBs-Apt₆₃ system improved the purity of exosomes with high expression of CD63.

In exosome-based liquid biopsies, diagnostic value ultimately comes from biomarkers in exosomes. In this study, proteins in exosomes separated by UC and magnetic beads were both analyzed by liquid mass spectrometry. We conducted a database search and comparison according to Proteome Discoverer 2.4 system. Specific database search parameters are shown in Supplementary Table S2. The relevant quality control information is shown in Supplementary Figure S5. 1044 proteins were identified in the exosomes isolated by UC, while only 677 proteins were detected in the exosomes separated by magnetic beads (Supplementary Table S3). To verify that our optically controlled MBs-Apt₆₃ preserved more exosome-related proteins, we performed GO analysis on high-expression set of exosome proteins extracted by MBs-Apt₆₃ versus UC. The screening criteria for differential proteins between the two groups were:

(1), Inter-group ratio ≥ 4 (2), Unique Peptides ≥ 2.

Among the top 10 GO components that were significantly higher in exosomes extracted by MBs-Apt₆₃, three are related to exosome formation, including rough endoplasmic reticulum, early endosomes, and endocytic vesicles (Figure 5D), indicating a higher expression abundance of exosome-related proteins in CD63 positive exosomes separated by MBs-Apt₆₃.