Au23(CR)14 nanocluster restores fibril Aβ’s unfolded state with abolished cytotoxicity and dissolves endogenous Aβ plaques

Abstract The misfolding of amyloid-β (Aβ) peptides from the natural unfolded state to β-sheet structure is a critical step, leading to abnormal fibrillation and formation of endogenous Aβ plaques in Alzheimer's disease (AD). Previous studies have reported inhibition of Aβ fibrillation or disassembly of exogenous Aβ fibrils in vitro. However, soluble Aβ oligomers have been reported with increased cytotoxicity; this might partly explain why current clinical trials targeting disassembly of Aβ fibrils by anti-Aβ antibodies have failed so far. Here we show that Au23(CR)14 (a new Au nanocluster modified by Cys-Arg (CR) dipeptide) is able to completely dissolve exogenous mature Aβ fibrils into monomers and restore the natural unfolded state of Aβ peptides from misfolded β-sheets. Furthermore, the cytotoxicity of Aβ40 fibrils when dissolved by Au23(CR)14 is fully abolished. More importantly, Au23(CR)14 is able to completely dissolve endogenous Aβ plaques in brain slices from transgenic AD model mice. In addition, Au23(CR)14 has good biocompatibility and infiltration ability across the blood–brain barrier. Taken together, this work presents a promising therapeutics candidate for AD treatment, and manifests the potential of nanotechnological approaches in the development of nanomedicines.


INTRODUCTION
A hallmark sequence of events in Alzheimer's disease (AD) is the misfolding, fibrillation and accumulation of amyloid-β (Aβ) peptides, resulting in cellular dysfunction, loss of synaptic connections and brain damage [1][2][3][4]. Over the past three decades, the inhibition of Aβ fibrillation and the disassembly of deposited Aβ fibrils have been the magnets for searching promising therapeutics for AD treatment [5][6][7][8][9]. A number of inhibitors (including βand γ -secretase inhibitors) for inhibiting Aβ production were discontinued in phase ii or iii clinical trials due to their low efficacy and serious side effects [10]. Anti-Aβ antibody-based immunotherapy for disassembling the mature Aβ fibrils was once expected to be the first radical treatment of AD [11]. However, prior studies have indicated that the soluble Aβ oligomers, as the most toxic species, might reappear during the disassembly process to induce more neurotoxicity (i.e. the 'dust-raising' effect) [12][13][14][15][16]. One approach to ameliorate the toxicity of soluble Aβ oligomers is to promote their aggregation by, for example, chiral silica nanoribbons and star-shaped poly(2-hydroxyethyl acrylate) nanostructures [17,18]. Also, graphene quantum dots are reported to drive the peptide fibrillization off-pathway to eliminate the toxic intermediates, which points to the potential of using zero-dimensional nanomaterials for in vivo mitigation of a range of amyloidosis types [19]. Recently, polymer-peptide conjugates and curcumin-gold nanoparticles (AuNPs) with hydrodynamic diameters of 10-25 nm have been shown to disassemble exogenous Aβ fibrils in vitro, but they failed to restore the natural unfolded state of Aβ from the misfolded β-sheets [20][21][22]. However, the β-lactoglobulin 'coronae' of the

RESEARCH ARTICLE
AuNPs are reported to enable X-ray destruction of islet amyloid polypeptide (IAPP) amyloids, providing a viable new nanotechnology against amyloidogenesis [23]. The small molecule epigallocatechin gallate (EGCG) presents the capability to prevent aggregation and remodel amyloid fibrils, which could also convert mature amyloid fibrils to amorphous protein aggregates that are less toxic to cells, implying the possibility of reducing the toxicity of amyloid fibrils by remodeling their molecular structures [24][25][26]. Therefore, the treatment of AD needs to explore new materials that are able to dissolve endogenous Aβ plaques and abolish the proteotoxicity of Aβ fibrils by restoring their natural unfolded state from the misfolded β-sheets.

Seven kinds of AuNCs on inhibiting Aβ fibrillation
First, the effects of these seven kinds of AuNCs on inhibiting Aβ fibrillation were investigated by coincubating 20 μmol·L −1 Aβ 40 with each kind of AuNCs at the same concentration (25 mg·L −1 ). The concentrations were selected based on their solubility and biological relevance from our preliminary experiments. The standard thioflavine-T (ThT) binding fluorescence assay was employed to record the fibrillation kinetics. As shown in Fig. 1A, the fibrillation kinetics of 20 μmol·L −1 Aβ 40 without AuNCs showed a standard S-curve (black curve); the formation of preformed/mature Aβ 40 fibrils was confirmed by atomic force microscopy (AFM) images ( Fig. 1B; Fig. S1 in the online supplementary material). Cys-AuNCs had no inhibitory effect (red curve); CSH-AuNCs (orange curve), p-MBA-AuNCs (yellow curve) and MPA-AuNCs (green curve) showed partial inhibition. Consistent with our previous studies, GSH-AuNCs showed complete inhibition of Aβ 40 fibrillation (blue curve). Encouragingly, NIBC-AuNCs (cyan curve) and CR-AuNCs (purple curve) were also able to completely inhibit Aβ 40 fibrillation, which was further verified by AFM images (no fibrils could be found in Fig. 1C-E). Moreover, in situ real-time circular dichroism (CD) spectra were used to record the conformational transition of Aβ 40 in the fibrillation process. As shown in Fig. 1F, in the absence of AuNCs, Aβ 40 had undergone a misfolding process from an unfolded state (negative peak at 198 nm) into a β-sheet structure (negative peak at 220 nm). Interestingly, GSH-AuNCs (Fig. 1G), NIBC-AuNCs (Fig. 1H) and CR-AuNCs (Fig. 1I) could maintain the unfolded state of Aβ 40 peptides throughout the incubation. It should be noted that the seven AuNCs used have similar particle sizes (1.6 ± 0.5 nm), and their transmission electron microscope (TEM) images and the UV-visible absorption spectra are shown in Fig. S2.

Seven kinds of AuNCs on the dissolving of mature Aβ fibrils
Since inhibition of fibrillation and dissolution of fibrils should be considered as two discrete events, then whether these AuNCs could dissolve preformed/mature Aβ 40 fibrils was investigated by using CD and AFM. Freshly prepared Aβ 40 (20 μmol·L −1 ) were pre-incubated at 37 • C for 72 h. The preformed Aβ 40 fibrils were then coincubated with 50 mg·L −1 individual AuNCs for 48 h. The formation and dissolution of Aβ 40 fibrils were recorded by CD. The data showed that the peak at 220 nm did not change between 72 h and 120 h when treated with Cys-AuNCs ( Fig. 2B 1 ), CSH-AuNCs ( Fig. 2C 1 ), p-MBA-AuNCs (Fig. 2D 1 ), MPA-AuNCs (Fig. 2E 1 ), GSH-AuNCs (Fig. 2F 1 ) and NIBC-AuNCs (Fig. 2G 1 ), and that the fibrils were intact (Fig. 2B 2 -G 2 ), indicating no dissolution of the mature Aβ 40 fibrils. The failure of GSH-AuNCs to dissolve Aβ 40 fibrils confirmed that inhibition of fibrillation and dissolution of fibrils are two discrete events. Most excitingly, when treated by CR-AuNCs, the peak at 220 nm (i.e. β-sheet) disappeared and the peak at 198 nm (i.e. unfolded state) resurfaced (Fig. 2H 1 ). The dissolution of   the mature Aβ 40 fibrils by CR-AuNCs was further confirmed by AFM observation (Fig. 2H 2 ). The perfect overlay of CD curves of 0 h and 120 h demonstrated that CR-AuNCs could completely dissolve the mature Aβ 40 fibrils, and fully restore the unfolded state of Aβ 40 peptides from β-sheet structure.

Molecular composition and structure of CR-AuNCs
To ascertain their molecular composition and structure, CR-AuNCs were characterized using various technical platforms ( Fig. 3 and Fig. S3). The electrospray ionization mass spectrometry (ESI-MS) analysis showed a single distinct peak at 8397.9925, indicating that CR-AuNCs had a formula of Au 23 (CR) 14 ( Fig. 3A). The formula was further confirmed by thermogravimetric analysis (Fig. 3B). The weight loss of 46.0% meant that the CR weight ratio agrees well with the formula of Au 23 (CR) 14 (calculated loss: 46.0%). In addition, the high resolution TEM analysis showed that the Au 23 (CR) 14 had a spherical morphology (Fig. 3C), where the shape was regular with a clear lattice fringe (inset of Fig. 3C).

The process detail and possible mechanisms of Au 23 (CR) 14 dissolving the preformed Aβ 40 fibrils
To gain more insights into the dissolution process, preformed Aβ 40 fibrils were co-incubated with 50 mg·L −1 Au 23 (CR) 14 for 48 h. The dissolution dynamics were monitored by ThT assay. The fluorescence intensity declined continuously during 48 h incubation ( Fig. 4A), indicating a gradual process of dissolution. The gradual dissolution of Aβ 40 fibrils had also been evidenced by AFM studies (Fig. 4B 1 . The apparent sizes of the samples were assayed by dynamic light scattering (DLS). The DLS results showed that the apparent sizes of the samples decreased from over 1000 nm to less than 10 nm . The in situ real-time CD spectra revealed that the peak at 220 nm was continuously shifted to 198 nm ( Fig. 4D), indicating that the dissolution of Aβ 40 fibrils by Au 23 (CR) 14 is a dynamic process accompanied by a conformational transition from a β-sheet structure to an unfolded state. The native PAGE results showed one band with a molecular weight less than 6.5 kDa (Fig. 4E, 48 h), directly demonstrating that Au 23 (CR) 14 completely dissolves Aβ 40 fibrils into monomers (∼4.2 kDa).
To explore possible mechanisms of how Au 23 (CR) 14 , but not the other six kinds of AuNCs, could dissolve Aβ 40 fibrils, the zeta potentials of Aβ 40 fibrils, individual AuNCs and Aβ 40 fibrils, together with individual AuNCs, were measured. The median of the zeta potential of mature Aβ 40 fibrils was −41 ± 2 mV (black curves in Fig. 4F and Fig. S4). Cys-AuNCs, CSH-AuNCs, p-MBA-AuNCs, MPA-AuNCs, GSH-AuNCs, NIBC-AuNCs and Au 23 (CR) 14 have a zeta potential of −32, +36, −49, −57, +2, −34 and +68 mV, respectively (blue curves in Fig. 4F and Fig. S4). After addition of AuNCs, while the mixtures with the other six kinds of AuNCs showed a negative zeta potential with a median value from −44 to −18 mV, the mixture with Au 23 (CR) 14 showed a positive zeta potential with a median value of +34 mV (red curves in Fig. 4F and Fig. S4). These data suggest that Au 23 (CR) 14  Aβ 40 fibrils more strongly than other AuNCs. In consideration of Aβ 40 monomers with a net charge of negative 2.7 at physiological pH (7.4) [33], and the existence of a guanidine group in the residue of CR that could be protonated in a wide range of pH [34], the strong electrostatic interaction between Aβ 40 and Au 23 (CR) 14 might drive the gradual dissolution of mature Aβ 40 fibrils. The above results strongly suggest that Au 23 (CR) 14 dissolve the preformed/mature Aβ 40 fibrils from misfolded β-sheets into the unfolded monomer state through strong electrostatic interactions.

Au 23 (CR) 14 -mediated Aβ 40 fibril dissolution on cell viabilities
To investigate the effect of Au 23 (CR) 14 -mediated dissolution of Aβ 40 fibrils on cell viabilities, an AD cell model based on Aβ 40 fibril-induced cell deaths of PC-12 cells was adopted [35]. First, PC-12 cells were co-incubated with freshly preformed Aβ 40  20 μmol·L −1 monomers were used to cause a 50% decrease of cell viability based on our preliminary titration experiments. As shown in Fig. 5A, when treated with Aβ 40 fibrils alone, cell shrinkage started to appear in the 3rd hour, and then cells with reduced sizes and round shapes apparently increased from the 12th to the 48th hour. In contrast, when PC-12 cells were treated with Aβ 40 fibrils and 50 mg·L −1 Au 23 (CR) 14 , no obvious morphological changes were observed (Fig. 5B). The corresponding videos are shown in the online supplementary material. Second, a CCK-8 assay was used for quantifying cell viabilities. Freshly preformed Aβ 40 fibrils from 20 μmol·L −1 monomers were added into PC-12 cells with or without Au 23 (CR) 14 ; the cells were cultured and sampled at the 3rd, 6th, 12th, 24th and 48th hour for assaying their viabilities. No treatment was included as the blank control. As shown in Fig. 5C, the cell viability was not affected in the blank control group (gray bars), and the addition of preformed Aβ 40 fibrils alone caused a gradual decrease of cell viability to 50% (red bars). In contrast, when cells were cultured with 50 mg·L −1 Au 23 (CR) 14 together with preformed Aβ 40 fibrils, the cell viability decreased initially to 70% at the 12th hour and then started to increase, reaching almost 100% (same as the blank control) at the 48th hour (blue bars). These data collectively demonstrated that Au 23 (CR) 14 could fully abolish the cytotoxicity of Aβ 40 fibrils. As for the two phasic characteristics of cell viabilities in the Au 23 (CR) 14 treatment, we speculate that the toxic oligomers [36,37] were produced during the dissolution process and the cytotoxicity was fully abolished when the oligomers were completely dissolved into non-toxic monomers.

The capacity of Au 23 (CR) 14 for dissolving exogenous Aβ fibrils
The ultimate test is whether the capacity of Au 23 (CR) 14 for dissolving exogenous Aβ fibrils can be translated into dissolving the endogenous Aβ plaques. We obtained brain slices derived from the resected brain tissue of an adult transgenic mouse model of AD, where the brain slices contained endogenous Aβ plaques. The brain slices were co-incubated without (Fig. 6A 1 -A 3 ) or with (Fig. 6B 1 -B 3 ) Au 23 (CR) 14 for 24 h, and then the slices were stained with anti-Aβ antibodies for immunohistochemical analyses. Figure 6A 1 -A 3 shows that the hippocampus and the neocortex were present with a large amount of endogenous Aβ plaques (yellow-brown patches indicated by the arrows). Excitingly, the treatment with 50 mg·L −1 Au 23 (CR) 14 eliminated all yellow-brown patches (Fig. 6B 1 -B 3 ), demonstrating that Au 23 (CR) 14 could completely dissolve the endogenous Aβ plaques in the hippocampus and the neocortex. Furthermore, our data showed that Au 23 (CR) 14 did not affect cell viability at a concentration of as high as 100 mg·L −1 (Fig. S5), indicating good biocompatibility. In addition, the overcoming of the blood-brain barrier is one precondition of nanomaterials in treating neurological diseases [6]. Our data showed that Au 23 (CR) 14 particles were readily detected in the brain tissues when intraperitoneally injected into normal mice, demonstrating that Au 23 (CR) 14 is capable of overcoming the blood-brain barrier (Fig. 6D).

CONCLUSION
In conclusion, seven kinds of AuNCs (i.e. Cys-AuNCs, CSH-AuNCs, p-MBA-AuNCs, MPA-AuNCs, GSH-AuNCs, NIBC-AuNCs and Au 23 (CR) 14 ) were synthesized and adopted to investigate their effects on the dissolution of mature Aβ fibrils and the restoration of the unfolded state of Aβ peptides. Among the seven kinds of AuNCs tested, only Au 23 (CR) 14   into monomers, and fully abolish cytotoxicity by restoring the natural unfolded state of Aβ peptides from misfolded β-sheets. Furthermore, Au 23 (CR) 14 are able to completely dissolve endogenous Aβ plaques in the brain slices from transgenic AD model mice. In addition, Au 23 (CR) 14 have good biocompatibility and infiltration ability across the blood-brain barrier. The biodistribution of AuNCs in vivo has been reported in our recent paper published in Nanomedicine [38]. Compared with the chaperone-gold nanoparticle in vivo test on zebrafish, similar efficacies to dissolve Aβ plaques and cross the blood-brain barrier are achieved by Au 23 (CR) 14 based on a rodent model, further indicating the clinical potential of nanoparticles or nanoclusters against Alzheimer's symptoms [39]. The relevant behavioral pathology and neurodegeneration would be offered in subsequent research. This study provides a compelling nanotherapeutic candidate for AD treatment.

Materials, cells and mice
Amyloid

Synthesis of AuNCs with different ligands
All AuNCs used in this work were synthesized on the basis of a method reported in our previous study with only a few minor changes in experimental parameters [19,30]. Take Au 23 (CR) 14 for example, 0.675 mmol (187 mg) of CR was dissolved in a 100 mL mixture of ultrapure water and ethyl alcohol (v/v = 1/2). Then a freshly prepared aqueous solution (6 mL) of HAuCl 4 (2.5 mmol·L −1 ) was slowly added into the pre-prepared mixture. The mixed solution was cooled to ∼0 • C in a cool bath for 18 h under a proper stirring frequency (340 rpm by mechanical agitation). Then, a fresh aqueous solution of NaOH (0.1 mol·L −1 , 18 mL) was added to the mixed solution. The reaction was maintained for 10 min and stirred vigorously (400 rpm  solution to react completely. The resulting mixed solution was collected and moved into an Amicon R Ultra-4 3K (MWCO: 3000) Centrifugal Filter device for centrifugal separation (RCF: 5000 × g, 30 min). Then the solution in the centrifuge tube was removed, and the solution in the filter device was washed by ultrapure water several times. Finally, the Au 23 (CR) 14 solution in the filter device was collected and lyophilized for further characterizations and experiments. The other six kinds of AuNCs were synthesized following similar conditions and operations.

Characterization
Nuclear magnetic resonance spectroscopy measurements Hydrogen ( 1 H) nuclear magnetic resonance (H-NMR) spectra of CR dipeptide were recorded on a Bruker AVANCE III 500 MHz spectrometer. A 100 mg·L −1 sample solution was added to the NMR tube, and the data were analysed by MestReNova.

Infrared spectroscopy measurements
Infrared (IR) spectra of AuNCs and the corresponding ligands were recorded on a Bruker Vertex 80v Fourier transform infrared (FT-IR) spectrometer. Lyophilized AuNCs and the corresponding ligands were directly used for IR measurement with ATR mode in a vacuum atmosphere at room temperature. Scanning range: 4000-400 cm −1 ; scan times: 64; vacuum degree: <5 hPa.

UV-visible spectroscopy measurements
UV-visible spectra of AuNCs were recorded on a Shimadzu UV-1800 UV-Vis spectrophotometer with a range of 300-1000 nm at a scan rate of 0.5 nm·s −1 . Lyophilized AuNCs were dissolved in water and then diluted to 200 μL with a concentration of 200 mg·L −1 . Then the sample was transferred into a high-quality quartz glass cuvette with a black wall for spectrophotometry (volume: 600 μL).

Mass spectrometry measurements
ESI-MS of CR-AuNCs was performed on a Nano electrospray ionization-quadrupole time-of-flight mass spectrometer (ESI-Q-TOF MS, Bruker) operating in the negative ion mode. The sample injection rate was 8 μL·min −1 . A capillary voltage of 4 kV was used for the ESI-MS (nebulizer: 1.5 bar, dry gas: 4 L·min −1 , 120 • C, m/z = 800-12 000). The ESI-MS spectra were obtained by accumulating for 5 min.

Thermal gravimetric analysis measurements
Thermal gravimetric analysis (TGA) was performed on a SETARAM TG-DSC 111 instrument. The sample was dried before TG measurement. The test was performed in flowing air with a temperature increasing rate of 1 • C·min −1 .
X-ray photoelectron spectroscopy analysis X-ray photoelectron spectroscopy (XPS) measurements were performed by an ESCALAB 250Xi with a focused monochromatic Al Kα X-ray (1350 eV) source for excitation. The binding energy (BE) scale is calibrated by using the O 1s peak at 530.14 ± 0.05 eV, the N 1s peak at 400.06 ± 0.05 eV, the C 1s peak at 283.42 ± 0.05 eV, the S 2p peak at 161.47 ± 0.05 eV and the Au 4f peak at 83 ± 0.05 eV for known reference foils.

TEM measurements
The TEM measurements of AuNCs were performed by using a Talos F200S TEM (Thermo Fisher, USA) with an accelerating voltage of 200 kV. The TEM images of the brain slices were performed by using a FEI Tecnai G20 TEM (FEI, USA) with an accelerating voltage of 200 kV. Blinded observation of samples with random selection of grid areas was implemented to reduce bias during imaging.

Zeta potential measurement
The zeta potentials of individual AuNCs (50 mg·L −1 ), Aβ fibrils (pre-incubated from 20 μmol·L −1 Aβ 40 ) and Aβ fibrils (20 μmol·L −1 Aβ 40 ) together with individual AuNCs (50 mg·L −1 ) were measured by using a Malvern Nano-ZS ZEN3600 zetasizer. In situ real-time circular dichroism spectroscopy CD spectra were recorded in the far-UV region from 190 to 250 nm by a JASC J-1500 Spectrometer, using a setup containing a step of 0.5 nm, a bandwidth of 1.0 nm, a speed of 50 nm·min −1 , a time per point of 1.0 s, an ultrasonic vibration of 600 rpm and an incubation temperature of 37 • C. The sample for experiment was collected in a quartz cuvette with a 1 mm optical path length; the cuvette was covered with a cap and sealed by sealing film. Each spectrum calibrated after subtraction of background signal was processed with a smoothing function of 30 points. The spectra data were recorded every 3 h.

In vitro inhibition or dissolution experiments
DLS tracks apparent size change during the dissolution of Aβ 40 fibrils DLS measurement was performed on a Malvern Nano-ZS ZEN3600 zetasizer. Mixtures of 20 μmol·L −1 Aβ 40 were pre-incubated in a high-quality quartz glass cuvette at 37 • C (in an incubator chamber) for 72 h. Then, 20 μL mixtures of Au 23 (CR) 14 were injected into the pre-incubated mixtures of Aβ 40 and incubated for another 48 h. The apparent size of the sample was recorded at the 0, 3rd, 6th, 12th, 24th and 48th hour.

Native PAGE
The Aβ states after co-incubation with or without Au 23 (CR) 14 were analysed via Native PAGE using 4-20% Tris-glycine gradient gels (BeyoGel). Native PAGE (no addition of β-mercaptoethanol and SDS) was adopted here for maintaining non-covalent bonds of samples. Samples of Aβ fibrils with or without Au 23 (CR) 14 (50 mg·L −1 ) were added to native loading buffer. Equal volumes of each sample (20 μL) were loaded onto gels along with Beyo-Color (Beyo) prestained molecular weight mark-ers and electrophoretically separated at 150 V. Gels were stained for total protein using a hypersensitivity Coomassie blue (BeyoBlue) according to the manufacturer's protocol. After incubation with decoloring solution three times (each for 3 h), the gel was detected by the gel imaging analysis system.

Cell experiments
PC-12 cells were incubated in DMEM medium supplemented with 10% fetal bovine serum (FBS) at 37 • C with 5% CO 2 . The cells were regularly subcultured to maintain them in logarithmic phase of growth. The cells were seeded in 96-well plates at a cell population of ∼10 000 cells per well and incubated for 24 h at 37 • C before further treatment. The viability of PC-12 cells was assessed by CCK-8 assay. Before being examined by using a Synergy TM MX Multi-Mode Microplate Reader at a wavelength of 450 nm, cells were treated with 100 mL DMEM contained 10% CCK-8 solution for ∼2 h.