SeCQDs were fabricated by a hydrothermal treatment according to the previous report (Li et al., 2017). Briefly, SeCys (200 mg) was dissolved in deionized water (12 ml). Then the pH of the solution was adjusted to pH 8.5 by NaOH to make sure the sample was completely dissolved. The solution was transferred to a stainless autoclave and heated at 80°C for 24 h in an oven. After that, the suspension was centrifuged at 12,000 rpm and the supernatant was collected and dialyzed. Then brown powder was obtained after freeze-dry of the SeCQDs solution.

To detect •OH, 5,5-dimethyl-1-pyrroline-N-oxide (DMPO) was used as the spin trapping agent to trap •OH. A solution containing DMPO (100 mM), H₂O₂ (100 mM) and SeCQDs aqueous solution (5, 10, 50 μg ml⁻¹) in PBS (10 mM, pH 7.0) in a quartz cuvette was exposed to UV light (365 nm, 60 mW cm⁻²) for 7 min. Then, the solution was transferred into a capillary tube, which was mounted onto the EPR spectrometer for scanning.

The formation of •OH was also monitored by the colorimetric assay and fluorescent assay, in which methylene blue (MB) and terephthalate (TA) were utilized as the dyes, respectively. TA can react with •OH radicals to generate a highly fluorescent product, 2-hydroxyl terephthalic acid (TAOH). To detect •OH, 400 μl of PBS (pH 7.0) containing H₂O₂ (100 mM) and MB (20 μM) or TA (40 μM) was added to the SeCQDs solution (0, 5, 10, 50 μg ml⁻¹) and exposed to UV light (365 nm, 60 mW cm⁻²) for 3 min. After that, the samples were centrifuged at 12,000 rpm for 5 min. The supernatants were collected for measurements.

Aβ40 peptides (100 μM) with or without various concentrations of SeCQDs or SeCys (5, 50 μg ml⁻¹) were incubated at 37°C for 7 days in aggregation buffer. At different times, aliquots of each sample were taken for fluorescence measurements. The final concentration of Aβ40 used for measurements was kept at 1 μM, and the thioflavin T (ThT) concentration was 10 μM. The excitation wavelength was 444 nm, and the emission intensity at 482 nm was used for analysis.

The peptide samples (10 μl) were spotted onto carbon-coated copper grids and stained with 1.5% (w/v) phosphotungstic acid (pH 7.4). Grids were air-dried before analysis on the transmission electron microscopy (TEM).

Samples for NMR were run in aqueous HEPES buffer with 10% ²H₂O added. Samples containing Aβ40 were run at 0.2 mM. SeCQDs were incubated with Aβ40 for 2 h at 37°C. NMR measurements were carried out on a Bruker 600-MHz AVANCE NMR spectrometer equipped with a triple channel cryoprobe at 5°C. The concentration of SeCQDs was 50 μg ml⁻¹.

To prepare different peptide samples, Aβ40 peptides (100 μM) with or without various concentrations of SeCQDs (5, 50 μg ml⁻¹) were incubated at 37°C for 7 days in aggregation buffer. The generation of ROS in PC12 cells (rat pheochromocytoma, American Type Culture Collection) was monitored using 2′,7′-dichlorofluorescein (DCF) diacetate (Beyotime, China). This nonfluorescent and cell-permeable dye can be converted to the anionic but nonfluorescent form DCFH by intracellular esterases. On the action of intracellular ROS, DCFH would be oxidized into its highly fluorescent form DCF, whose fluorescence intensity correlates with the amount of intracellular reactive oxygen radicals. To perform the test, PC12 cells which pretreated with different peptide samples for 12 h were incubated with 20 μM DCF diacetate for 60 min at 37°C. The cells were then rinsed with PBS solution. The fluorescence intensity was monitored on a fluorescence spectrofluorometer with excitation and emission wavelengths of 488 and 525 nm, respectively.

PC12 cells were cultured in DMEM (Gibco BRL) medium supplemented with 5% fetal bovine serum (FBS), 10% horse serum (HS) in a 5% CO₂ humidified environment at 37°C. Cells were plated at 7 000 cells per well on poly-l-Lysine coated 96-well plates in fresh medium. After 24 h, Aβ40 (10 μM) that had been aged with or without various concentrations of SeCQDs were dispensed into the PC12 cells, and the cells were further incubated for 24 h at 37°C. Cytotoxicity was measured by using MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide, Sigma-Aldrich) assay. Absorbance values of formazan were determined at 570 nm with a Bio-Rad model-680 microplate reader.

Seven-week-old male C57BL6/J mice with body weights between 18 and 20 g were obtained from the Experimental Animal Center of the Chinese Academy of Medical Sciences. All the protocols and procedures for animal handing were carried out following the guidelines of the Hebei committee for care and use of laboratory animals, and were approved by the Animal Experimentation Ethics Committee of the Hebei Medical University. After intravenous injection of 100 μl SeCQDs (1 mg ml⁻¹), the main organs were collected after 6 h. The Se content of the samples was measured by ICP-MS (Agilent 7800 ICP-MS). The data points shown are the mean values ± SEM from three independent experiments (Three mice were used for each group.).

For the biocompatibility study, the main organs were collected and dissected to make paraffin section 14 days after i. v. injection of SeCQDs (75 μg per mouse). Then, Hematoxylin-eosin (H&E) staining assay was conducted.

Adult male SD rats (320 ± 30 g) were obtained from Laboratory Animal Resources, Chinese Academy of Sciences. All the protocols and procedures for animal handing were carried out following the guidelines of the Hebei committee for care and use of laboratory animals, and were approved by the Animal Experimentation Ethics Committee of the Hebei Medical University. The rats were housed under standard condition with a light exposure of 12 h light/12 h dark cycle and free access to food and water. The rats were denied to access to food for 12 h before the test. After that, they were anesthetized by intraperitoneal injection of sodium pentobarbital (45 mg kg⁻¹). Aβ40 oligomers (1.0 mg ml⁻¹, total volume of 10 μl, 2 μl min⁻¹) were injected into each side of the hippocampus. The position of cornu ammonis area 1 (CA1) was obtained by subtracting 3.5 mm from the anteroposterior position, 2.0 mm from the mediolateral position, and 3.0 mm from the dorsoventral position. The experimental rats were housed for 3 weeks to establish the AD rat model. Aβ40 oligomers were freshly prepared by dissolving Aβ40 in 0.1% trifluoroacetic acid to give a concentration of 10 μg μl⁻¹, followed by incubation for 7 days at 37°C. The rats in control groups were injected with saline instead of Aβ40 oligomers. In the treating group, SeCQDs (100 μg ml⁻¹, total volume of 100 μl) or SeCys (100 μg ml⁻¹, total volume of 100 μl) were administrated to AD model rats via tail vein injection every 2 days for a consecutive 21 days after the models had been established for 3 days.

SeCQDs were synthesized by a hydrothermal treatment of SeCys according to the previous report (Li et al., 2017). The TEM images showed that SeCQDs were well dispersed with a diameter of approximately 25 nm (Figure 2A). A typical lattice spacing of 0.32 nm was clearly observed in the high-resolution TEM (HRTEM) images (Figure 2B), which was similar to the bulk graphite (002 facet) (Li et al., 2017; Rosenkrans et al., 2020). The peak in X-ray diffraction (XRD) pattern of SeCQDs at 22° was also assigned to the (002) plane (Supplementary Figure S1) (Rosenkrans et al., 2020). The relatively small size revealed the possibility of SeCQDs to penetrate the BBB and be used as agents for AD treatment. The UV-Vis absorption spectrum of SeCQDs showed two typical absorption peaks at approximately 280 and 340 nm due to the existence of multiple electron transitions (Supplementary Figure S2) (Li et al., 2017). The high degree of crystallinity was further confirmed by Raman spectrum analysis, in which the crystalline G band at 1,605 cm⁻¹ was stronger than the disordered D band at 1,380 cm⁻¹, with a G-to-D intensity ratio (IG/ID) of 1.2 (Figure 2C). Stretching vibrations for O-H or N-H (Bonds between 3,100 cm⁻¹ and 3,600 cm⁻¹), C=O (1,615 cm⁻¹), C-N (1,381 cm⁻¹) and C-O (1,154 cm⁻¹) bonds were observed in Fourier translation infrared (FT-IR) spectrum (Figure 2D), demonstrating the formation of polyaromatic structures and the existence of free carboxyl and amino groups at the edge of SeCQDs. The composition of SeCQDs was also analyzed using X-ray photoelectron spectroscopy (XPS), which revealed that SeCQDs were primarily composed of carbon, oxygen, nitrogen, and selenium (Figure 2E, Supplementary Figure S3). In addition, the stability of SeCQDs in different solutions, including water, PBS (pH 7.4) and cell culture medium (DMEM supplemented with 10% fetal bovine serum) was also investigated since it was a key factor for their practical applications in biological systems. As shown in the dynamic light scattering (DLS) studies and TEM results, the particle size (Supplementary Figure S4A) and morphology (Supplementary Figure S4B-D) of SeCQDs did not change in all these solutions after standing for 7 days, revealing the good stability of SeCQDs.

Following the synthesis and analysis of SeCQDs, we next evaluated their ROS-scavenging activity (Figure 3A). Among all the ROS, ⋅OH as one of the most damaging species, can directly react with almost all biomolecules including DNA, proteins and membrane lipids (Li et al., 2017). Thus, we employed ⋅OH as a model to investigate the radical scavenging efficiency of SeCQDs using ESR spectroscopy and colorimetric as well as fluorescent assay. As shown in Figure 3B, in the presence of the spin trap, DMPO, a strong typical signature of DMPO-HO⋅ appeared in the system of H₂O₂ with UV irradiation. However, the amount of produced ⋅OH radical decreased significantly upon addition of SeCQDs. And the inhibition effect on the formation of ⋅OH radical was highly dependent on the concentration of SeCQDs. The same results were obtained from the colorimetric and fluorescent assay, in which MB (Zhu et al., 2020) and TA (Karim et al., 2018) were utilized as the dyes, respectively (Figures 3C,D). The discoloration of MB and the decrease in fluorescence intensity of TAOH produced via the oxidation of TA by ⋅OH both indicated the ROS-scavenging activity of SeCQDs.

ThT fluorescence assay was employed to examine the influence of SeCQDs on the aggregation of Aβ. Aβ40, the most abundantly produced Aβ isoform, was chosen as the protein model, which has been widely used for in vitro amyloidogenesis study (Gao et al., 2019; Li et al., 2020). As a benzothiazole dye, ThT can selectively bind to the β-sheet region present in Aβ fibrils, resulting in a strong increase in its fluorescence. The ThT fluorescence curves in Figure 4A clearly indicated the inhibition effect of SeCQDs on Aβ aggregation. When fresh Aβ40 alone incubated at 37°C, the ThT fluorescence displayed a standard sigmoidal curve, consistent with the nucleation-dependent polymerization model (Gao et al., 2019). However, the increasing trend of ThT fluorescence was greatly suppressed upon introduction of SeCQDs, which indicated that the aggregation process of Aβ was inhibited by SeCQDs. A control experiment was also carried out to clarify that the fluorescence of ThT could not be affected by the addition of SeCQDs with the concentration used in the inhibition study (Supplementary Figure S5). In addition, SeCQDs suppressed Aβ aggregation in a dose-dependent manner (Figure 4A). Critically, compared with SeCQDs, SeCys displayed little inhibition effect on Aβ aggregation (Supplementary Figure S6). The inhibition on Aβ40 aggregation was further evaluated by circular dichroism (CD) spectra. As demonstrated in Supplementary Figure S7, SeCQDs could inhibit structural transition from the native Aβ40 random coil to the β-sheet conformation in solution. While, Aβ40 retained monomeric forms at the start stage of the experiment and SeCQDs themselves did not show any obvious signal in this range, which eliminated the influence of Aβ40 and SeCQDs themselves on the inhibition results (Supplementary Figure S8). TEM analysis also showed that in the absence of SeCQDs, Aβ monomers assembled into mature fibrils, while, SeCQDs predominantly inhibited Aβ fibrillization (Figures 4B–D).

The inhibition behavior and binding models between SeCQDs and Aβ were further confirmed by NMR spectroscopy (Zagorski and Barrow, 1992; Amir et al., 2012; Li et al., 2014; Valiente-Gabioud et al., 2018). Compared with that of Aβ alone, the ¹H NMR signals of all His proton lines and Tyr10 lines in Aβ were broadened and the resonances of Lys and Glu were shifted to high-field upon treated with SeCQDs (Figure 5). The changes in chemical shift indicated an altered chemical environment due to the direct interaction with SeCQDs or due to a structural change of Aβ that occurred upon binding. The negative charged SeCQDs (Supplementary Figure S9) can easily bind to the cationic cluster HHQK of Aβ via electrostatic interactions. In addition, the paired α-carboxyl and amino groups existed on the edge of SeCQDs camouflaged them as large amino acids, which can also trigger multivalent interactions with these amino acids. Due to the charged nature and bulky size, SeCQDs can cause the structure of Aβ to be changed, endowing them with inhibition effects on Aβ aggregation.

Having demonstrated the ability of SeCQDs to inhibit the formation of ROS and Aβ aggregation, we next investigated whether they could reduce the production of ROS in cells caused by Aβ aggregates as well as the cytotoxicity of Aβ aggregates. As shown in Figure 6A, the content of ROS in cells treated with Aβ aggregates increased significantly up to 236%, relative to that in the untreated control cells. However, upon pretreatment with SeCQDs at the concentration of 0.5 μg ml⁻¹ or 5 μg ml⁻¹, the intracellular level of ROS largely decreased to 173 and 139%, respectively, indicating SeCQDs can greatly reduce the intracellular production of ROS in a dose-dependent manner. Critically, there was no significant difference between the generation of ROS in SeCQDs solely treated cells and the untreated cells (Supplementary Figure S10). The effect of SeCQDs on the cytotoxicity of Aβ aggregates was also investigated. Figure 6B indicated that compared to the untreated cells, the cell viability was reduced to 54% upon treatment with Aβ aggregates for 24 h. SeCQDs protected PC12 cells from Aβ aggregates-induced cell death in a concentration-dependent manner. Importantly, the cytotoxicity of SeCQDs was also examined in PC12 cells. As shown in Supplementary Figure S11, SeCQDs showed no toxicity in our experimental conditions, suggesting that SeCQDs can be utilized as potential agents for AD treatment. It is well known that Aβ aggregates can cause damage to mitochondrial structure and function, which is one of the neurotoxic mechanisms of Aβ (Godoy et al., 2017). Mitochondrial membrane potential (MMP) was used to evaluate the function of mitochondria via the fluorescent probe, JC-1 (Liu et al., 2017; Shen et al., 2020). Supplementary Figure S12 showed that SeCQDs can significantly alleviate the Aβ induced depolarization of MMP. All these results proved the neuroprotective effect of SeCQDs on PC12 cells.

To ensure the possibility and safety for AD treatment, it is of importance to study the in vivo biodistribution and biocompatibility of SeCQDs. To investigate the accurate distribution of SeCQDs in main organs, SeCQDs were intravenously injected into C57BL/6 mice, and the main organs were collected after 6 h injection to determine the remained Se by ICP-MS (Figure 7A). The accumulation of SeCQDs in the brain revealed the potential capability of SeCQDs to cross the BBB.

Hemolysis testing has been recognized as a classical assay for evaluating the cell damage effect of nanomaterials (Zhu et al., 2019; Zhou et al., 2020). As shown in Figure 7B, SeCQDs with the concentration varied from 0.1 μg ml⁻¹–50 μg ml⁻¹ only caused <1% haemolysis, which was considered to be biocompatible in accordance with ISO/TR 7406 (the permissible limit for hemolysis is 5%) (Chibhabha et al., 2020; Lima et al., 2021). The in vivo biocompatibility of SeCQDs was also analyzed by the change of body weight after injecting the drug (Figure 7C). No significant difference in body weight was observed between SeCQDs-treated mice and untreated mice within 14 days. On day 14, the main organs of the mice were collected and examined by Hematoxylin-eosin (H&E) staining (Figure 7D). H&E stained pathological sections of the main organs of SeCQDs treated mice including heart, liver, spleen, lung, kidney and brain exhibited no apparent lesions or abnormalities as compared with that from the untreated group, which demonstrated the excellent in vivo biocompatibility of SeCQDs.

Inspired by the excellent biocompatibility of SeCQDs and their in vitro inhibition effect on Aβ-induced cytotoxicity, we next investigated whether SeCQDs can be used for highly efficient in vivo AD treatment. Progressive cognitive declines, the main clinical symptoms of AD, have been reported to be directly correlated with Aβ-mediated synaptic deficits (Beckman et al., 2019; Hou et al., 2020; Samanta et al., 2021). In order to analyze the possibility of SeCQDs for in vivo AD therapy, the Morris water maze (MWM) test (Sanati et al., 2019; Sun et al., 2019; Huang et al., 2021) was first conducted to evaluate the efficiency of SeCQDs to ameliorate Aβ induced memory deficits. As shown in Figure 8A, B, all the rats exhibited progressive decline in escape latencies during the 5-days spatial learning training. Compared with the normal saline injected wild type rats (control group), Aβ-infused rats required much more time to find the hidden platform (Figure 8B) and spent decreased time in the target platform quadrant (Figure 8C) in the following the probe trial test, indicating that Aβ induced spatial learning and memory deficits in rats. In contrast, after administrated with SeCQDs, the Aβ40-infused rats showed shorter escape latencies and increased staying time in the targeted quadrant. Critically, compared with SeCys, SeCQDs showed obviously higher efficiency in improving the cognitive ability of Aβ40-infused rats.

Aβ deposition and neuronal loss in the hippocampus have been widely known as the key markers of AD. Aβ deposition in the brain was characterized by immunohistochemistry (IHC) of Aβ (Sun et al., 2019; Liu et al., 2019). As displayed in Figure 9, obvious Aβ deposition was observed around the neurons in Aβ-infused rats. However, Aβ plaques depositions were considerably decreased in SeCQDs-treated groups, indicating that administration of SeCQDs reduced Aβ accumulation in the brain. Nissl staining of hippocampal slices (Sun et al., 2019; Liu et al., 2019) showed that Aβ-infused rats had very few Nissl bodies. However, more neurons with restored integrity were observed in SeCQDs-treated AD model rats, which confirmed the protective effect of SeCQDs against Aβ induced neurodegenerative consequences (Figure 9). Critically, the protection effects were also observed in SeCys-treated groups but not as obvious as that of SeCQDs-treated group. Similar to other selenium compounds, short-term administration of SeCys also showed the potential in improving memory deficits and reducing Aβ plaques in AD model mice (Weekley and Harris, 2013; Du et al., 2016). However, the relative high toxicity limited its clinical application as therapeutic drug for AD treatment (Supplementary Figure S13). In contrast, with the intrinsic properties of both selenium and CQDs, SeCQDs possessed excellent biocompatibility and can significantly improve the reference memory deficit, inhibit Aβ accumulation and neuron degeneration in Aβ-treated rats. All these characteristics rendered SeCQDs as promising candidates for AD therapy.