Exosome engineered upconversion nanoparticles with targeting moieties enhance Aβ aggregate clearance and mitigate neurotoxicity in Alzheimer’s disease

Abstract

Alzheimer’s disease is an increasingly prevalent neurodegenerative disorder characterized by the accumulation of misfolded Aβ aggregates. The efficacy of upconversion nanomaterials in photodynamic therapy has been proven, yet their applicability is constrained by their inherent toxicity. In this study, exosomes derived from stem cells, combined with targeting molecules, were utilized to encapsulate upconversion nanoparticles for targeted removal of pathogenic amyloid aggregates in Alzheimer’s disease. Evidence from AFM and TEM imaging, fluorescence spectroscopy, and cell-based experiments demonstrates that the clearance efficiency of amyloid aggregates is enhanced, and the cytotoxicity induced by misfolded Aβ aggregates is significantly reduced in nerve cells. Notably, the viability of neural PC12 cells is partially restored even in the presence of toxic amyloid aggregates.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-23597-5.

Introduction

Alzheimer’s disease (AD) is a brain disorder that progressively destroys brain functions. The incidence of AD has been steadily increasing and imposes a significant burden on patients and society. To date, no effective therapeutic treatment is yet available. One of the pathological hallmarks of AD is the accumulation and deposition of misfolded Aβ aggregates, specifically Aβ(1–40) and Aβ(1–42), leading to the formation of so-called senile plaques in the brain^1,2. The Aβ(1–40) and Aβ(1–42) peptides are derived from the proteolytic cleavage of the amyloid precursor protein (APP) and possess a rich secondary structure of β-sheets, which are prone to self-assembly, forming neurofibrillary tangles (NFTs). This ultimately triggers neurotoxicity and leads to neuronal death³. Consequently, inhibiting protein amyloidosis and, furthermore, eliminating existing amyloid aggregates represent potential therapeutic targets for AD. Studies have demonstrated that rare earth-doped upconversion nanoparticles (UCNPs) can bind to various photosensitizers, generating reactive oxygen species (ROS) under near-infrared (NIR) light irradiation, which can non-invasively degrade protein aggregates^4–6. However, despite the numerous advantages of UCNPs^7–10, their biological applications remain limited in clinical practice due to concerns regarding biosafety^11–13. To enhance the biocompatibility of UCNPs, numerous researchers have undertaken a wide range of surface modification strategies^14–18. Currently, these modified UCNPs still necessitate at least one week for complete metabolism within the human body¹⁹.

Biomimetic surface engineering of nanomaterials demonstrates superior biological performance compared to conventional synthetic systems, particularly in terms of reduced immunogenicity, prolonged systemic circulation, and enhanced biocompatibility. Capitalizing on these inherent advantages, we developed human mesenchymal stem cell-derived exosome (hucMSC-exo)-coated upconversion nanoparticles (UCNPs) to leverage the natural biomimicry of exosomal membranes. Notably, the exosomal coating confers dual therapeutic functionalities for Alzheimer’s disease (AD) intervention: Aβ clearance and neurorestoration.

Exosome-borne surface proteins facilitate active binding and subsequent degradation of amyloid-β (Aβ) aggregates through lysosomal pathways^20–23. Meanwhile, exosomes-coating could play a beneficial role in AD treatment by directly releasing bioactive contents to further restore neuronal functions^20–26. Targeted delivery of exosomal cargo (e.g., neurotrophic miRNAs, growth factors) to compromised neurons promotes synaptic plasticity and mitigates tau hyperphosphorylation^24–26. Furthermore, exosomes have been shown to help drugs cross the blood‒brain barrier (BBB) and increase intracranial drug concentrations in AD treatment²⁷. All these advantages of stem cell exosomes would benefit the PDT treatment by UCNP nanomaterials. To enhance the amyloid targeting and minimize the dosage of the nanomaterials, a targeting peptide, (DSPE-PEG2000-Mal)-CGGGKLVFF (hereafter abbreviated as DP), is designed with the 3’ end of the peptide KLVFF^28,29_, which can recognize and target Aβ(1–40). The DSPE moiety spontaneously integrates into the exosomal lipid bilayer via hydrophobic interactions. PEG2000 spacer optimizes ligand presentation by minimizing steric hindrance between the targeting peptide and exosomal surface proteins. The KLVFF motif specifically binds to Aβ(1–40) aggregates through β-sheet structural complementarity, while endogenous exosomal adhesion molecules (e.g., tetraspanins) mediate neuronal recognition^30–32 (Scheme 1).

Scheme 1.

Exosome-coated upconversion materials are used to efficiently degrade amyloid aggregates.

Experiment and materials

Materials and instruments

The following reagents were purchased from Aladdin Co., Ltd., China: yttrium oxide (Y₂O₃), yttrium oxide (Yb₂O₃), erbium oxide (Er₂O₃), rose Bengal (RB), sodium hydroxide (NaOH), ammonium fluoride (NH4F), 1-octadecene (1-ODE), 3-aminopropyltriethoxysilane (APTES), N-hydroxysuccinimide (NHS), 1-ethyl-(3-dimethylaminopropyl) carbodiimide hydrochloride (EDC), polyvinylpyrrolidone (PVP, Mw = 40000), 1,1,1,3,3,3-hexafluoro-2-propanol (HFIP), oleic acid (OA) and tetraethyl silicate (TEOS). CD spectra were measured on a spectropolarimeter (JASCO, Hachioji City, Japan) with model No. PTC-348W1 (JASCO). X-ray diffraction (XRD) was performed on a Rigaku D/MAX-2500PC diffractometer using Cu Ka radiation (l = 1.5406 Å). All AFM measurements were performed with a Multimode 8 AFM apparatus (Bruker, USA).

Synthesis of UCNPs

In this study, ytterbium (Yb) and erbium (Er) codoped UCNPs were synthesized by a hydrothermal method. The synthesis steps were as follows: 0.04 mmol of Y₂O₃, 0.09 mmol of Yb₂O₃ and 0.01 mmol of Er₂O₃ were added to a closed three-necked flask with dilute hydrochloric acid and stirred magnetically until the solution became transparent. Subsequently, the solution was evaporated to produce 1 mmol of LnCl₃ (18% Yb, 2% Er). OA (12 mL) and 1-ODE (15 mL) were mixed in a three-necked flask and heated to 150 °C under nitrogen protection to dissolve LnCl₃ until the solution became bright yellow. After cooling to room temperature, 10 mL of methanol solution containing 5 mmol of NaOH and 8 mmol of NH₄F was added to the mixed mixture and stirred for 30 min. Then, the mixture was heated to 75 °C and 105 °C to remove methanol and water, respectively. Finally, the mixture was heated to 300 °C under nitrogen protection and maintained for 1 h. To precipitate the product, the solution was mixed with 30 mL of ethanol after cooling. The UCNPs were obtained through centrifugation, separation, washing, and drying.

Isolation and characterization of exosomes derived from human umbilical cord MSCs

With the informed consent of the puerpera, fresh human umbilical cord tissue was obtained from the Fourth People’s Hospital of Zhenjiang. This research project conforms to the principles outlined in the Declaration of Helsinki and complies with relevant policies and regulations of China. We strictly adhere to all stipulations of the aforementioned documents in conducting experimental research. Medical Ethics Review Document of Zhenjiang Maternity and Child Health Hospital was included in S2 of Supporting information. All of the experiments in this study were approved by the Medical Ethics Committee of Jiangsu University (IRB approval protocol number: 2020161). Primary isolation of human umbilical cord-derived mesenchymal stem cells (hUC-MSCs) was performed according to methods previously established in our laboratory³³. Briefly, fresh umbilical cord tissue was processed into 1 mm³-sized tissue blocks within 6 h after collection, affixed to sterile dishes supplemented with medium (α-MEM) containing 10% fetal bovine serum (Gibco, Grand Island, USA) and cultured at 37 °C in 5% CO₂. After approximately 10 days, the primary cells were digested with 0.25% trypsin and then subcultured. All the experimental protocols were approved by the Medical Ethics Committee of Jiangsu University (2012258).

Preparation of small extracellular vesicles (sEVs): To eliminate endogenous sEVs from fetal bovine serum (FBS), ultracentrifugation was performed at 100,000 ×g (4 °C, 16 h) using a Beckman Coulter ultracentrifuge. The resulting sEVs-depleted FBS was subsequently used to formulate culture medium. Upon reaching 80% cell confluency, the conventional medium was replaced with the sEVs-free medium. Post-culture, the conditioned medium was subjected to sequential centrifugation to remove contaminants: Cell debris removal: 300 ×g, 10 min; Dead cell elimination: 2,000 ×g, 10 min; Organelle clearance: 10,000 ×g, 30 min. The clarified supernatant was further ultracentrifuged at 100,000 ×g for 90 min to pellet sEVs, which were then resuspended in phosphate-buffered saline (PBS) for downstream applications.

Characterization of sEVs: Size distribution: Nanoparticle Tracking Analysis (NTA) was employed to measure sEVs diameter. Morphological assessment: Transmission Electron Microscopy (TEM, FEI Tecnai 12, Philips) was utilized to visualize sEVs ultrastructure.

Modification of UCNPs

After UCNPs preparation, the UCNPs were coated with a layer of mesoporous silica shell and rose bengal (RB) as a photosensitizer. After that, the silica-coated UCNPs-RB nanoparticles were wrapped in exosomes and modified with targeting peptides, and finally, the products were used to degrade the prepared Aβ(1–40) aggregates.

For the organosilica-shell coating, 10 mL of cyclohexane containing 0.1 mL of CO-520 and 10 mg of UCNPs were magnetically stirred for 10 min until the solution became transparent. Later, 80 µL of aqueous ammonia solution, 0.4 mL of CO-520 and 40 µL of TEOS were added to the above solution and stirred overnight. Then, 100 µL APTES was added to obtain UCNPs@SiO₂-NH₂. The final products were collected after washing and drying. First, 10 mg of UCNPs@SiO₂-NH₂ and 0.2 g of PVP were mixed in 10 mL of deionized water. After full dispersion, the mixture was heated to 96 °C and stirred for 6 h. After the same collection methods, the UCNPs@ROS nanoparticles were prepared.

For photosensitizer loading, 20 mL of EDC solution (50 mM) was mixed with 15 mg of NHS, 16 mg of UCNPs@ROS and 10 mg of RB. Hydrochloric acid (0.1 M) was used to adjust the pH to 5.0, and then the solution was magnetically stirred in the dark overnight. After washing and dialysis for 24 h, the RB/UCNPs@ROS were collected.

Exosomes and targeting peptide encapsulation: According to previous reports, exosomes can be used to encapsulate nanomaterials by the mechanical extrusion method^34,35. Briefly, exosome-UCNP mixtures (1:1 ratio) were sequentially extruded through 400 nm, 200 nm, and 100 nm polycarbonate membranes with 33 passes per membrane. The exosome-encapsulated nanoparticles (RB/UCNPs@ROS@Exo) were then stored at -80 °C until use. After a co-incubation with RB/UCNPs@ROS@Exo at room temperature for 0.5 h, the targeting peptide (DSPE-PEG2000-Mal)-CGGGKLVFF, DP) could self-assemble on the surface of the exosomes. The final RB/UCNPs@ROS@exosome-DP nanoparticles were obtained.

Preparation of Aβ(1–40) aggregates

HFIP was used to dissolve the dry powder of Aβ(1–40) (1 mg/mL). The suspension was shaken in several centrifuge tubes for 12 h at room temperature. After being sealed with parafilm, the tubes were kept for 2 h to remove HFIP. After incubation at 37 °C for 24 h, Aβ(1–40) aggregates were obtained.

Detection of ¹O₂

A solution of SOSG (1 × 10^− 4 M, Maokang Biotechnology, Shanghai) was added to the prepared RB/UCNPs@ROS@Exo solution at a final concentration of 1%. After being irradiated under NIR (FUV 6BK, Guangzhou Bangwo Electronics Technology Co., Ltd.) for 20 min, the fluorescence at 485 nm was measured.

Cytotoxicity assessment of the nanomaterials on PC12 cells

The cytotoxicity of the prepared RB/UCNPs@ROS@Exosome-DP was analyzed by a CCK-8 assay. Pheochromocytoma (PC12) cells were cultured at a density of 8 × 10³ (96-well plate) for 24 h. Then, the fresh cell culture medium was mixed with the prepared nanomaterials at different concentrations (0, 10, 25, 50, 100 µg/mL) and subjected to a CCK-8 assay. In the CCK-8 assay, PC12 cells were incubated with the abovementioned culture medium mixture and CCK-8 reagent (10 µL). The absorbance at 450 nm was measured after incubation for 2 h.

Detoxification experiment

The Aβ(1–40) aggregate solution was added to microplates containing the RB/UCNPs@ROS@Exosome-DP nanoparticles and to PC12 cells, which were then incubated for 24 h under NIR light. After that, cell viability was assayed using a CCK-8 cell counting kit (Beyotime) under different conditions.

ThT assay

A total of 25 µL of amyloid protein Aβ (1–40) used in the experiment was mixed with 25 µL of ThT (1 mM) and 150 µL of deionized water. After thorough mixing, the solution was added to a 96-well plate (Nunc 96 flat bottom generic), and absorbance or fluorescence spectra were measured using a microplate reader with excitation at 450 nm and emission at 485 nm.

CD spectroscopy

At designated time points during the degradation process, 150 µL of Aβ (1–40) aggregate solution was sampled and transferred to a quartz cuvette with a 1 mm path length. Circular dichroism (CD) spectra were acquired at room temperature using a spectropolarimeter (J-815) with a scanning range of 190–250 nm, a slit width of 2 nm, and a scanning speed of 50 nm/min.

Atomic force microscope (AFM) analysis

100 µL of Aβ (1–40) aggregate solution was deposited onto a glass slide and air-dried at room temperature. All AFM images were acquired using a scanning probe microscope (Multi-Mode VIII, Bruker) with a resolution of 512 × 512 pixels.

Fourier transform infrared (FTIR) spectroscopy

An adequate amount of nanoparticle powder was finely ground and placed directly under the IR light source. FTIR spectra were recorded using a Nicolet iS50 spectrometer in the wavenumber range of 4000–400 cm⁻¹.

X-ray diffraction (XRD) analysis

The synthesized UCNPs powder was thoroughly ground and compacted into pellets. XRD patterns were collected using a D8 Advance diffractometer (Bruker) with a 2θ scanning range of 5–80°, a step size of 0.02°, and an operating voltage of 40 kV.

Western blot

Protein markers of sEVs were analyzed by Western blotting under denaturing conditions. Total protein lysates (20 µg per lane) were subjected to 12% SDS-PAGE gels (Bio-Rad Mini-PROTEAN system) and transferred to a PVDF membrane (Millipore) by semidry Western blotting. Following blocking with 5% nonfat milk in Tris-buffered saline containing 0.1% Tween-20 (TBST) for 1 h at 25 °C, the membranes were incubated overnight at 4 °C. The primary antibodies used in this study were: CD9 (Proteintech, USA, 60232-1-Ig), CD81 (Proteintech, USA, 18250-1-AP), HSP70 (CST, USA, 2679 S), Calnexin (CST, USA, 4872 S). Subsequently, membranes were washed three times with 1 x TBST (10 min/wash) and then incubated with horseradish peroxidase (HRP)-conjugated goat anti-rabbit/mouse IgG secondary antibody (USA, Invitrogen, 31460/31430) at 37 °C for 1 h. The protein bands on the membrane were detected using an enhanced chemiluminescence detection system (General Electric Company).

Results and discussion

Characterization of modified UCNPs and ROS production

For the preparation of the UCNPs, we optimized the synthesis conditions based on previous reports³⁶. First, we synthesized UCNPs and UCNPs@SiO₂ with a diameter of 13.1 ± 0.1 nm by a thermal decomposition method. The upconversion luminescence was characterized by a fluorescence spectrophotometer (F-4500), as shown in Fig. 1a. The emission fluorescence spectrum presented two obvious emission peaks at 522 nm (²H_11/2→⁴I_15/2) and 542 nm (⁴S_3/2→⁴I_15/2)^37,38. The SEM and EDS shown in Fig. 1b were used to confirm the morphology and elemental composition of the UCNPs. The UCNPs were coated with a silica layer on the surface to improve their biocompatibility and inhibit fluorescence quenching. Figure 1c-d showed the TEM images of UCNPs and UCNPs@SiO₂. The diameter of UCNPs@SiO₂ nanoparticles was measured to be approximately 25.8 ± 0.1 nm and the thickness of the silica layer was approximately 6.8 ± 0.2 nm. The XRD spectra shown in Fig. 1e confirmed the elemental composition of the UCNPs and the hexagonal (β) phase structure compared with the standard card JCPDS:28-1192^39,40. The FTIR spectrum, as shown in Fig. 1f, revealed successful silica coating and subsequent amination, which was intended to improve the dispersity and solubility of the material^41,42. Then, to enhance the high loading capacity of the photosensitizers, UCNPs@SiO₂ were etched into rattle-structured UCNPs@ROS via a previously described method⁴³.

Fig. 1.

The characterization of NaYF₄:Yb, Er. (a) Fluorescence emission spectrum. (b) SEM image and EDS spectra. (c,d) TEM characterization of the UCNPs and UCNPs@SiO₂, respectively. (e) The XRD pattern. (f) The FTIR spectra of UCNPs, UCNPs@SiO₂-NH₂, RB and RB/UCNPs@ROS. (g) The UV‒Vis spectra of RB/UCNPs@ROS (black line) and RB (red line). (h) The fluorescence spectra of UCNPs, UCNPs@SiO₂ and RB/UCNPs@ROS. (i,j) The fluorescence spectra of singlet oxygen sensor green (SOSG) incubated with different concentrations (10, 25, 50, 100 µg/mL) of RB/UCNPs@ROS.

Next, RB was successfully loaded onto UCNPs@ROS as photosensitizers via EDC/NHS chemistry, as verified by the UV-Vis spectra of RB/UCNPs@ROS (black line) and RB (red line). As shown in Fig. 1g, the UV-Vis spectrum of RB/UCNPs@ROS showed a redshift in adsorption from the 550 nm peak for RB to the 568 nm peak, indicating the covalent bonding of RB. The significant quenching of the upconversion fluorescence also supported the binding of RB to the UCNPs@ROS, as shown in Fig. 1h. This result suggested that the upconversion fluorescence emission was absorbed by RB and converted to reactive oxygen species (ROS). The fluorescence of SOSG revealed that the dominant kind of generated ROS was singlet oxygen (¹O₂), in accordance with previous reports⁴⁴. Figure 1i–j shows that the quantity of generated ¹O₂ measured by SOSG fluorescence was closely related to the concentration of the RB/UCNPs@ROS nanomaterials. As the concentration of nanomaterials increased, the intensity of the fluorescence emission at 525 nm increased.

Characterization of exosomes and their encapsulation on UCNPs@SiO₂

Mesenchymal stem cells were isolated from the human umbilical cord according to a previously reported protocol³³. The stemness of human umbilical cord mesenchymal stem cells (MSCs) was evaluated by adipogenic induction and osteogenic induction. As shown in Figure S3, the adipogenic and osteogenic differentiations of MSCs were successfully confirmed by Oil Red staining for the formation of lipid droplets in MSCs (red arrows indicate lipid droplets) and Alizarin Red staining for the formation of mineralized nodules in MSCs (red arrows indicate calcium nodules). Exosomes were subsequently isolated and purified from the mesenchymal stem cell culture supernatant according to a previously reported protocol⁴⁵. Optical microscope (Figure S3), transmission electron microscopy (TEM) (Fig. 2a), nanoparticle tracking analysis (NTA) (Fig. 2b), atomic force microscopy (AFM) (Fig. 2c) and western blotting (Fig. 2d) were used to identify the morphology, particle size distributions and surface marker expression levels of the isolated exosomes. TEM and NTA revealed that the exosomes displayed a typical cup-shaped structure approximately 30–200 nm in diameter. The western blotting results indicated that the exosomes expressed exosome-specific biomarkers, such as CD9, CD81 and HSP70.

Fig. 2.

The characterization of human umbilical cord mesenchymal stem cells (MSCs) and their derived exosomes (HucMSC-Ex). (a) Representative TEM image of human stem cell-derived exosomes (scale bar = 100 nm). (b) The size analysis by NTA. (c) Atomic force microscopy was used to detect the morphology and height of HucMSC-Ex. (d) Western blot (WB) was performed to detect the expression of proteins markers in HucMSC and HucMSC-Ex. The raw data were included in S1 of the Supporting Information.

Cytotoxicity assay of different UCNPs with surface modifications

As shown in Fig. 3, the cytotoxicity of the modified UCNPs was assessed in L929 (Fig. 3a) and PC12 cells (Fig. 3b) at concentrations of 10 µg/mL, 25 µg/mL, 50 µg/mL, and 100 µg/mL.The cells without nanomaterials served as a control for 100% cell viability. For PC12 cells, unmodified UCNPs exhibited significant cytotoxicity, with cell viability decreasing from 62% to 33% as the concentration increased. In comparison, UCNPs coated with silica layers did not significantly mitigate this toxicity. However, encapsulation within exosomes significantly improved the viability of PC12 cells (RB/UCNPs@ROS@Exo). The cell viability remained above 80% as the nanoparticle concentration increased from 10 µg/mL to 50 µg/mL. When the concentration of RB/UCNPs@ROS@Exo was raised to 100 µg/mL, the PC12 cell viability was 71%, which was considerably higher than that of UCNPs (35%) or silica-modified UCNPs (29%). Cytotoxicity experiments were also performed on mouse fibroblast cells (L929). The cytotoxicity for L929 cells was similar to that of PC12 cells.

Fig. 3.

Cytotoxicity assay of the UCNPs, UCNPs@SiO₂ and RB/UCNPs@ROS@Exosome at different concentrations (10, 25, 50, 100 µg/mL) using a CCK-8 assay with (a) L929 cells and (b) PC12 cells.

To assess cytotoxicity differences among the three nanomaterials (UCNPs, UCNPs@SiO₂, RB/UCNPs@ROS@Exo), a one-way analysis of variance (ANOVA) was performed, with statistical significance set at *p* < 0.05. Exosomal encapsulation significantly reduced the cytotoxicity of UCNPs toward both L929 and PC12 cells across the tested concentrations. ANOVA revealed a significant overall effect of nanomaterial type on cell viability (*p* < 0.001). Post hoc pairwise comparisons using Tukey’s HSD test indicated that: For L929 cells, cell viability of UCNPs (Mean ± SD = 55.8% ± 11%) was significantly lower than that of RB/UCNPs@ROS@Exo (Mean ± SD = 110.0% ± 16%) (Mean Difference = 54.2%, adjusted *p* < 0.001). For PC12 cells, cytotoxicity of UCNPs (Mean ± SD = 48.8% ± 13%) was also significantly lower than that of RB/UCNPs@ROS@Exo (Mean ± SD = 87.0% ± 15%) (Mean Difference = 38.2%, adjusted *p* < 0.001). Furthermore, RB/UCNPs@ROS@Exo exhibited significantly lower cytotoxicity than UCNPs@SiO₂ for both L929 and PC12 cells (adjusted *p* < 0.01 for each). Tukey’s test also showed no significant differences in cytotoxicity between different concentrations of UCNP-based materials. Collectively, these results demonstrate that exosomal encapsulation effectively enhances the biocompatibility of UCNP-based materials.

Application of modified UCNPs for the photodegradation of Aβ(1–40) aggregates

Deposited amyloid aggregates are considered pathogenic proteins associated with Alzheimer’s disease (AD). Here, we utilized prepared Aβ(1–40) aggregates as the target of AD treatment, UCNPs coated with exosomes and targeting peptide (RB/UCNPs@ROS@Exosome-DP) as photodynamic therapy (PDT) agents to degrade preformed Aβ(1–40) aggregates to examine the potential applicability of exosomes and targeting peptide encapsulation in AD treatment. Firstly, we prepared Aβ(1–40) aggregates by incubating a 1 mg/mL Aβ(1–40) solution in a 37 °C thermomixer for 24 h. The growth of the amyloid fibrillar aggregates was monitored using a Thioflavin-T (ThT) assay⁴⁶,^47. As shown in Fig. 4a, the growth curve reached a steady state after approximately 1000 min, showing that the Aβ(1–40) monomers self-assembled into amyloid fibrils, which are rich in β-sheet secondary structures. The TEM and AFM images in Fig. 4b, c show the morphology of typical fibrillar aggregates. As shown in Fig. 4d, the height of the fibrils was measured to be 7.1 ± 0.3 nm.

Fig. 4.

Characterization of Aβ(1–40) aggregates. (a) ThT assay of the amyloid aggregate growth curve; the Aβ(1–40) aggregates were characterized by (b) TEM and (c) AFM; (d) The corresponding height distribution in AFM images.

To achieve a balance between a sufficient quantity of nanomaterials and acceptable side effects, various concentrations of nanomaterials were tested in photodegradation experiments. According to SOSG fluorescence measurements, as depicted in Fig. 1i-j, RB/UCNPs@ROS nanoparticles at a concentration of 25 µg/mL could produce a large amount of ROS while still maintaining excellent biosafety for PC12 cells. We used 25 µg/mL RB/UCNPs@ROS@Exosome-DP for photodegrading preformed Aβ(1–40) aggregates. As shown in Fig. 5a-d, AFM and TEM images revealed morphological changes in the Aβ(1–40) aggregates during the photodegradation process. The long fibrils of the Aβ(1–40) aggregates gradually shattered and eventually degraded into small unstructured particles. As shown in Fig. 5e-f, quantitative analysis of AFM morphology—specifically measuring the coverage area ratio of amyloid fibrils to the total image area and the height distribution of Aβ(1–40) aggregates—demonstrated a time-dependent inhibitory effect of nanomaterials on amyloid aggregation. The hydrophobic residues within the hydrophobic core are critical for fibrillization. The degradation of fibrils may expose hydrophobic residues within the core, which aligns with previous reports demonstrating that organic-inorganic hybrid nanoparticles can selectively recognize the hydrophobic core in β-sheet conformations, thereby blocking β-sheet propagation^48,49.

Fig. 5.

AFM and TEM characterization of the photodegradation of amyloid aggregates under NIR light. (a–d) AFM and TEM characterization of Aβ(1–40) aggregates photodegraded with RB/UCNPs@ROS@Exo (25 µg/mL) for different durations (0 h, 1 h, 2 h, and 3 h). (e) Height distribution analysis of the fibrils. (f) Change in the coverage ratio of the fibrils in the AFM images. Secondary structure of amyloid aggregates during the photodegradation process was assayed by (g) CD spectra and (h) ThT fluorescence spectra of Aβ(1–40) aggregate undergoing degradations for different durations. (i) The ThT fluorescence intensity of the Aβ(1–40) aggregates.

Analysis of the secondary structure of amyloid aggregates

During the degradation process, the secondary structure change during photodegradation for 3 h was investigated by analyzing the circular dichroism (CD) spectrum. As shown in Fig. 5g, the sigmoid CD curve indicated the existence of a β-sheet structure in the prepared amyloid aggregates⁵⁰. After photodegradation, the positive peak of the CD curve gradually decreased from 2.2 mdeg at 195 nm to 0.46 mdeg at 198 nm, indicating the decomposition of the secondary structure of the Aβ(1–40) aggregates. As shown in Fig. 5h-i, the ThT fluorescence at 485 nm immediately decreased after photodegradation for 0.5 h, confirming that the secondary structure of the prepared Aβ(1–40) aggregates was significantly disrupted.

Detoxification of amyloid aggregates

We further investigated whether the targeting peptide and exosome-encapsulated UCNPs (RB/UCNPs@ROS@Exosome-DP) could ameliorate the cytotoxicity of amyloid aggregates and improve the viability of PC12 cells. Firstly, the cytotoxicity of preformed Aβ(1–40) aggregates to cells was tested. After being incubated with PC12 cells for 24 h, the Aβ(1–40) aggregates showed high toxicity to PC12 cells, as indicated by a significant decrease in cell viability to 44%, as shown in Fig. 6. To test the detoxification effect of the UCNPs, Aβ(1–40) aggregates and RB/UCNPs@ROS@Exosome-DP nanoparticles were mixed and coincubated with PC12 cells for 24 h in microplate wells under NIR light. In the experiment, the effects of different factors, including the Aβ(1–40) aggregates, the exosome coatings, the presence of targeting peptides and the NIR light irradiation, on the detoxification of amyloid aggregates on PC12 cells were systematically investigated.

Fig. 6.

Cell viability of PC12 cells under different conditions measured by the CCK-8 assay(Laser group: 1.0 W/cm² for 15 min). Different letters above the bars denote statistically significant differences between the UCNP derivatives.

As shown in Fig. 6, an improvement in cell viability was observed after the addition of exosome-encapsulated nanoparticles even without the NIR laser or targeting peptide (the RB/UCNPs@ROS@Exo(L⁻) group), and this effect was enhanced by the addition of the targeting peptide to the nanoparticles (RB/UCNPs@ROS@Exo-DP). Under NIR light, exosomes and targeting peptide-modified nanoparticles (the group RB/UCNPs@ROS@Exo-DP(L⁺)) had the highest cell viability, with more than 82% of neural cells alive, which was significantly higher than detoxification by nanoparticles without targeting peptide (73% cell viability, RB/UCNPs@ROS@Exo(L⁺)) and degradation without NIR (64% cell viability, RB/UCNPs@ROS@Exo-DP(L⁻)). A one-way analysis of variance (ANOVA) was performed, with statistical significance set at *p* < 0.05. As shown in Fig. 6, post hoc pairwise comparisons using Tukey’s method indicated that among the different UCNP derivatives, RB/UCNPs@ROS@Exo-DP exhibited a significantly greater reduction in cytotoxicity compared to the others.

Mechanism of cellular uptake and detoxification

The RB/UCNPs@ROS@Exosome-DP nanoparticles likely exploit exosome-mediated endocytosis and targeting peptide for targeted delivery, followed by synergistic amyloid disaggregation (via exosomal chaperones/UCNP-PDT), antioxidant/anti-apoptotic effects (exosome content), and autophagy activation^20–26.

The targeted delivery of the hybrid exosome-UCNP nanomaterials

The active targeting via exosome surface markers and targeting molecules drive selective internalization into neuronal PC12 cells and amyloid-rich regions. Stem cell-derived exosomes naturally express adhesion molecules (e.g., integrins, L1CAM) that bind to receptors on neuronal cells or amyloid plaques, enabling receptor-mediated endocytosis. Exosome lipid bilayers may fuse with PC12 cell membranes, bypassing lysosomal degradation (critical for preserving UCNP functionality).

Amyloid clearance and cytoprotection

Exosomal content (miRNAs, chaperones, antioxidants) counteracts residual Aβ/ROS toxicity, restoring neuronal viability. Exosome-derived neuroprotection could be realized by anti-apoptotic factors (e.g., miR-21, survivin), Chaperones (HSP70, clusterin) to refold misfolded proteins or enhance amyloid phagocytosis, antioxidants (SOD, catalase) and trophic support to suppress the apoptosis of PC12 cells and reverse Aβ-induced cytotoxicity^51,52.

Targeted photodynamic therapy (PDT)

UCNPs convert low-energy photons to higher-energy emissions, activating photosensitizers to generate reactive oxygen species (ROS). Controlled ROS production likely disrupts β-sheet structures in Aβ aggregates, as evidenced by AFM/TEM showing reduced fibril density in Fig. 5. Exosomes and the targeting peptide localize UCNPs to amyloid sites and minimize off-target ROS damage, resolving the toxicity issue of free UCNPs.

Conclusion

In this study, upconversion nanoparticles with a small diameter of 13.1 ± 0.1 nm and narrow distribution were synthesized. After a series of surface modifications and encapsulation with stem cell-derived exosomes and targeting peptides, the UCNPs demonstrated excellent biocompatibility and enhanced ability to detoxify amyloid aggregates in PC12 cells. Macroscopically, the involvement of exosomes and targeting peptides enhanced the targeting clearance of pathogenic proteins in the cellular environment and restored cell viability, which suggests the potential application of these nanomaterials in treating relevant diseases.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Conceptualization, C. J. and J. W.; Data curation, C. J. and Z. M.; Formal analysis, C. J., Z. M. and Z. Z.; Funding acquisition, Z. Z.; Investigation, C. J., Z. M. P. W. and L. S.; Methodology, C. J., Z. Z. and J. W.; Resources, P. W. and H. Q; Supervision, Z. Z. and J. W. ; Validation, C. J. and Z. M.; Writing – original draft, C. J.; Writing – review & editing, Z. Z. and J. W.

Funding

This research was supported by the Natural Science Foundation of Universities in Jiangsu Province (Grant no. 21KJB220015), Taicang Technology Project (Grant no. TC2023JC32) and Innovation Team Funds of Suzhou Chien-shiung Institute of Technology (Grant no. 2023JXKYTD01).

Data availability

The original contributions presented in the study are included in the article, further inquiries can be directed to the corresponding author.

Declarations

Competing interests

The authors declare no competing interests.