Catalytically active gold clusters with atomic precision for noninvasive early intervention of neurotrauma

Background Neurotrauma is a worldwide public health problem which can be divided into primary and secondary damge. The primary damge is caused by external forces and triggers the overproduction of peroxides and superoxides, leading to long-lasting secondary damage including oxidative stress, wound infection and immunological reactions. The emerging catalysts have shown great potential in the treatment of brain injury and neurogenic inflammation, but are limited to biosafety issues and delivery efficiency. Results Herein, we proposed the noninvasive delivery route to brain trauma by employing highly active gold clusters with enzyme-like activity to achieve the early intervention. The decomposition rate to H2O2 of the ultrasmall gold clusters is 10 times that of glassy carbon (GC) electrodes, indicating excellent catalytic activity. The gold clusters can relieve the oxidative stress and decrease the excessive O2·− and H2O2 both in vitro and in vivo. Besides, gold clusters can accelerate the wound healing of brain trauma and alleviate inflammation via inhibiting the activation of astrocytes and microglia through noninvasive adminstration. decrease the peroxide and superoxide of brain tissue. Conclusions Present work shows noninvasive treatment is a promising route for early intervention of brain trauma. Supplementary Information The online version contains supplementary material available at 10.1186/s12951-021-01071-4.

Herein, we investigated a noninvasive administration to treat brain trauma at the early stage with gold clusters of well-defined structure and high catalytic selectively. Electrochemical assay unraveled their high catalytic activities toward hydrogen evolution reaction (HER) and oxygen evolution reaction (OER), and further revealed their excellent activity toward the reduction of O 2 and H 2 O 2 . Furthermore, noninvasive administration of the gold clusters distinctly improved the wound healing on TBI mice and Morris water maze tests further confirmed significant recovery on learning ability and spatial memory with Au 24 Cu 1 and Au 24 Cd 1 treatment. Ex vivo assay further proved the ability of gold clusters on mitigating oxidative stress and inhibiting neuroinflammation. Together with the results of acceptable biocompatibility, gold clusters showed promising potential as noninvasive therapeutics against TBI.

Results and discussion
The Au 25 cluster is composed of 13 gold atoms as the core and 6 Au 2 (SG) 3 as the outer shell. The catalytically active sites of the cluster are mainly located on the surface. Single atom Cu or Cd replaces the Au in one of the S-Au-S, indicative of synthesizing Au 24 Cu 1 and Au 24 Cd 1 , respectively. After single atom Cu and Cd substitution, the valence electron structure of the cluster is altered, thereby changing the catalytic performance of the cluster [23]. Figure 1a illustrated the effects of MPA-protected Au 25 , Au 24 Cu 1 and Au 24 Cd 1 clusters in brain trauma via catalytic systems. TEM images illustrated the homogenous distribution of the clusters at about 2 nm (Fig. 1b,  c), similar with L-NIBC-coated gold clusters [46]. The hydrodynamic sizes determined by dynamic light scattering (DLS) were 1.98, 1.92 and 2.58 nm for Au 25 , Au 24 Cd 1 and Au 24 Cu 1 , respectively, a little larger than the statistical diameter of TEM. After incubation in water for 24 and 48 h, the size of clusters changed negligibly, revealing the favorable stability (Fig. 1d, e). In addition, the zeta potentials of all clusters were around − 35 mV (Fig. 1f ), consistent with our previous work [26,47]. The concentration Cd and Cu elements in Au 24 Cd 1 and Au 24 Cu 1 clusters account for 3% and 5% of the total metal as determined by an accurate inductively coupled plasma mass spectrometry (ICP-MS), respectively, further confirming the single-atom substitution in Au 25 clusters (Additional file 1: Fig. S1). X-ray photoelectron spectroscopy (XPS) confirmed that the dominant state of Au in the clusters is the Au (0) state, and the peaks of Cu 2p 3/2 (931.8 eV) and Cd 3d 5/2 (404.8 eV) are on the reducing side of Cu (0) (id. =932.2 eV) and Cd (0) (id. =405.3 eV), respectively, indicating the successful single-atom substitution in Au 25 clusters (Additional file 1: Fig. S2). These results display the ultrasmall size and good colloid stability of clusters, manifesting the potential in biological applications.
To evaluate the electrocatalytic activities of the Au 25 , Au 24 Cu 1 and Au 24 Cd 1 clusters, a standard three-electrode system were employed to conduct the electrocatalytic activities toward HER and OER [48][49][50]. Figure 2a shows cyclic voltammetry (CV) curves of glassy carbon (GC) electrodes modified with as prepared Au 25 , Au 24 Cu 1 and Au 24 Cd 1 clusters. Compared with a blank GC electrode, clusters achieve larger negative current density toward HER (Fig. 2a), which were further verified by linear sweep voltammetry (LSV) measurement (Fig. 2b). As shown in Fig. 2c, Au 24 Cu 1 clusters demonstrated a higher efficiency at the potential of −0.4585 V in catalysis, significantly better than Au 25 (SC 12 H 25 ) 18 [8] and PtCo cluster [10]. OER exhibits similar results with HER that all clusters improved the performance of the electrochemical reaction (Fig. 2d, f ). Au 24 Cd 1 shows the lowest current onset on CV curves at 1.3 V, followed by Au 24 Cu 1 and Au 25 , which is a little different from the results of HER. Besides, Au 24 Cd 1 can reach 0.023 mA/cm 2 at a potential of 1.3254 V, while that of NiFe clusters is less than 0.01 mA/cm 2 , indicating better superiority to NiFe cluster [11]. The results of HER and OER further demonstrated that the catalytic performance were altered by the valence electron structure change via different singleatom substitution [23].
We also evaluated the in vitro catalytic activities of Au 25 , Au 24 Cu 1 and Au 24 Cd 1 clusters in H 2 O 2 reduction and oxygen reduction reaction (ORR) [51,52]. As shown in Fig. 3a, the current density of the GC electrode modified by Au 24 Cu 1 clusters can reach -1.48 mA/cm −2 at a potential of -0.8 V, a little higher than Pt 12 clusters [18]. The results were further verified by LSV measurement (Fig. 3b), demonstrating that all clusters can enhance the electrocatalytic activity and Au 24 Cu 1 clusters exhibit the best activity for ORR among all clusters. For the reduction of H 2 O 2 , only imperceptible reduction current was observed in the presence of H 2 O 2 on the GC electrode, consistent with previous works [51]. Compared with the GC electrode at -0.8 V, Au 24 Cu 1 clusters show the current destiny of -1.74 mA/cm −2 , and Au 24 Cd 1 clusters exhibit that of -1.56 mA/cm −2 (Fig. 3d, e), slightly better than noble metal (Ag, Pd, Au, Pt) on Graphene/ZnO multihybrid nanoarchitectures [19]. Figure 3c, f quantitatively demonstrate the improvement of all clusters on O 2 and H 2 O 2 reduction, indicating excellent catalytic activities. Au 24 Cu 1 presented the predominance toward both reactions, and Au 24 Cd 1 and Au 25 followed toward H 2 O 2 and O 2 reduction, respectively.
Since the remarkable catalytic activities inspired us to investigate their biological responses in cells and braininjured mice, we conducted biological inspections to evaluate their activities in vitro and in vivo. Combined with our previous work [26], Au 24 Cd 1 and Au 24 Cu 1 clusters conferred catalytic selectivity and enzyme-like activities, beneficial to reduce the oxidative stress induced by brain injury (Fig. 4a). Au 25 , Au 24 Cu 1 and Au 24 Cd 1 exihibit good biocompatibility at the concentration of 50 µg/   [26].
To investigate the biological activity of clusters, imaging of H 2 O 2 -treated neuron cells with or without Au 25 , Au 24 Cd 1 and Au 24 Cu 1 clusters was obtained (Fig. 4b).
The H 2 O 2 stimulation can significantly elevates signals, indicating the presence of excessive amount of ROS and O 2 ·− [53]. All clusters can decrease the ROS and O 2 ·− signals, while Au 24 Cd 1 shows the best clearance efficiency. Meanwhile, Au 24 Cu 1 displays better clearance capability against ROS than Au 25 and Au 24 Cd 1 , suggesting higher selectivity for ROS. The related quantification further confirmed the remarkable biological catalytic activity and laying the groundwork for in vivo utilization (Fig. 4d, e). In addition, H 2 O 2 induces decreases in cell viability (∼78%) due to oxidative stress and inflammation [54], while Au 25 , Au 24 Cd 1 and Au 24 Cu 1 clusters can rescue cell viability back to 90%, indicative of great potential to protect nerves and lower the lethality from H 2 O 2 -induced cytotoxicity (Fig. 4c). All the in vitro results manifest good catalytic activities of Au 24 Cd 1 and Au 24 Cu 1 clusters on ROS scavenging, revealing their potential as biocatalysts and suggesting further investigation in vivo. Catalytic mechanism and in vitro evaluation of the clusters. a Schematic illustration of the catalytic mechanism and selectivity of the clusters. Cu and Cd as single active sites exhibit superiorities against reactive oxygen species (ROS) and reactive nitrogen species (RNS) respectively, resulting in free radical scavenging and inspire their use in brain injury. b Fluorescence images of ROS (green) and O 2 ·− (red) levels induced by 100 µM H 2 O 2 with or without clusters treatment, illustrating the high activity of scavenging ROS of the clusters. c HT22 cell viability under treatment of H 2 O 2 and treated with or without the clusters (n = 5 per group, data are presented as mean ± SD), which showed the ability of the clusters to rescue the lethality induce by H 2 O 2 . d, e Quantitative analysis of Au 25 , Au 24 Cu 1 and Au 24 Cd 1 against O 2 ·− and ROS, respectively (n = 3 independent experiments, data are presented as mean ± SD). Data are analyzed by one-way ANOVA with Tukey test (adjusted p values are shown, *p< 0.05, ***p< 0.001, compared with the Con group) The primary injuries triggers a casade of biochemical reactions, leading to the long-lasting sencondary injuries. The secondary brain injuries generate harmful molecules and cytokines, leading to acute or chronic neuronal damage, memory impairment and inflammation. Traditional intravenous administration have shown great poteintial in brain diseases, but toxicity remains a major concern for clinical translation [15,28]. Therefore, the above in vitro preliminary results inspired us to treat TBI noninvasively and it is reasonable to develop a noninvasive routine to treat TBI at the early stage. The clusters were dropped in the injured area of TBI mice. The wound size was significantly reduced to healthy levels after clusters treatment, whereas the untreated mice only showed a partial recovery (Fig. 5a, b). We further evaluated the oxidative stress-related indicators, including SOD, GSH/ GSSG, MDA, and H 2 O 2 in TBI mice [28,55]. Brain injuries generate excessive ROS in tissues, consume lots of SOD and GSH, and produce harmful lipid peroxidation ( Fig. 5c-f ). SOD and GSH/GSSG are sharply decreased on day 7 post-injury and slightly increased on day 14 post-injury, and MDA and H 2 O 2 are relatively serious on day 7 post-injury and a little alleviated on day 14 postinjury for TBI groups, indicating severe oxidative stress after brain injuries. However, clusters can effectively decrease the MDA and H 2 O 2 levels, suggesting the ROS elimination of clusters, and can siginificantly recovered the SOD and GSH/GSSG levels, maintaining the balance of ROS levels.
Moreover, the behavior tests were conducted to evaluate the spatial learning and memory abilities by Morris water maze (Fig. 6). All mice were trained to learn to search for the platform during the acquisition phase on days 13-17 and 26-30 including the total distance travelled and the latency to the hidden platform. Compared with TBI groups, both travel distance and latency to platform gradually decrease with training after clusters treatment (Fig. 6c, d), indicating that clusters can effectively improve the motor function following brain injury. Au 24 Cu 1 or Au 24 Cd 1 clusters show better results than Au 25 , indicating the efficiency enhancement by Cu and Cd single-atom substitution. Figure 6e, f show that platform crossings and average distance obviously decrease in TBI group and can almost return back to normal levels with cluters treatment after injury. These facts imply that clusters can effectively improve the learning ability and spatial memory of mice with TBI-induced brain injury.
Since brain injuries can lead to chronic inflammation, we were examined their therapeutic effects on the neuroinflammation. Figure 7a demonstrates the schematic illustration for the TBI-associated inflammatory responses and relevant modulation with the clusters [1,8]. IL-1β, IL-6 and TNF-α are all upregulated after injury, revealing strong local inflammation. Au 25 only shows minor downregulation toward the three inflammatory factors, while Au 24 Cd 1 and Au 24 Cu 1 exhibit exceptional efficiency to reduce their levels, manifesting that substitution of Cu and Cd atom has great influence on the catalytic activity (Fig. 7b, c). Au 24 Cu 1 performs better than Au 24 Cd 1 in downregulating overexpressed TNF-α, whereas Au 24 Cd 1 behaves better in mediating IL-1β, IL-6, implying catalytic selectivity to different inflammatory cytokines. ELISA kits further verified the immunoblot results that Au 24 Cd 1 and Au 24 Cu 1 are superior to Au 25 on inhibiting inflammatory factors in brain tissues including IL-1β, IL-6, and TNF-α (Fig. 7d, f ). Additionally, the immuno histochemical assay illustrated the conspicuous anti-inflammation effect of Au 24 Cd 1 and Au 24 Cu 1 against IL-6 ( Fig. 7 h), consistent with quantitative results in Fig. 7 g. Furthermore, significant differences on the pathological slices can be seen that the TBI group exhibited swelling of nerve cells and obvious infiltration of inflammatory cells; while the groups treated with clusters especially the Au 24 Cd 1 and Au 24 Cu 1 recovered to almost normal (Fig. 7i). These results demonstrated that Fig. 7 Inflammation levels in brain tissues. a Schematic illustration of catalytic activity on oxidative stress and inflammatory responses of the clusters. Expression levels of IL-1β, IL-6 (b), and TNF-α (c) in the brain tissues on day 30 post TBI with or without clusters (n =3 per group) were analyzed using western blotting. The Au 24 Cd 1 showed superiority on downregulating the IL-1β and IL-6, while the Au 24 Cu 1 had better effect toward TNF-α. d-f ELISA analysis of inflammatory factors on days 7 and 14 post TBI with or without treatment of clusters (n = 3 per group) further confirmed the abilities of the Au 24 Cu 1 and Au 24 Cd 1 on regulating inflammatory factors. All the samples were derived from the same experiment and blots were processed in parallel. Data are presented as mean ± SEM and compared with the control groups. g Quantitative analysis of the positive cell numbers in injured cortex with or without nanocluster treatment (n=3 per group). Analyzed by one-way ANOVA with Tukey test and compared with the Con groups. h The immunohistochemical assay illustrated the conspicuous effect of Au 24 Cd 1 and Au 24 Cu 1 against IL-6. i The pathological slices showed that the TBI group exhibiting apoptotic and swollen nerve cells and infiltration of inflammatory cells, while the groups treated with clusters especially the Au 24 Cd 1 and Au 24 Cu 1 recovered to almost normal. Data are presented as mean ± SEM and compared with the Con groups, analyzed by one-way ANOVA with Tukey test (adjusted p values are shown, *p< 0.05, **p< 0.005, ***p< 0.001) the clusters especially Au 24 Cd 1 and Au 24 Cu 1 can accelerate the wound healing of traumatically injured brain by inhibiting the inflammatory factors.
Brain injuries can also mediate the recruitment of microglia and astrocytes in the injuried areas, which are the key cellular mediators of TBI after brain injury [1]. As shown in Fig. 8a, lots of astrocytes are produced and activated along with inflammatory cytokines, indicating strong local inflammation. The Au 24 Cd 1 and Au 24 Cu 1 clusters can inhibit the activation of astrocytes and remarkably reduce the overexpression of IL-1β and IL-6. The relative quantification further exhibited the significant effect of Au 24 Cd 1 and Au 24 Cu 1 clusters (Fig. 8b-d).
Immune response associated IL-1β and IL-6 can be alleviated by Au 24 Cd 1 , while Au 24 Cu 1 shows better effect on reducing TNF-α, indicating their catalytic selectivity and potential selectivity towards the treatment of neuroinflammation. Also, it is noted that the performance of mice in Au 24 Cu 1 and Au 24 Cd 1 groups is superior to that in Au 25 groups, indicating that Au 24 Cu 1 and Au 24 Cd 1 clusters possess a better anti-inflammation property by efficiently eliminating excessive free radicals and inhibiting neuroglia activation. Additionally, hematology analysis and tissue toxicology analysis showed no toxicity in the long term, suggesting the good biological safety of clusters (Additional file 1: Figs. S4-S6).
The present work demonstrated the high catalytic activity of Au 24 Cd 1 and Au 24 Cu 1 clusters and their availability as TBI therapeutics. Unlike traditional nanozymes, gold clusters overcomed the water insolubility and exhibited the well-defined structure at the atomic level with ultrasmall size and good biocompatibility. Catalytic activity and selectivity of clusters can be further increased and availability as TBI therapeutics can also be improved via single ation substitution. Moreover, since noninvasive administration can directly and rapidly release the biocatalysts into the damaged area, it shows promissing potential for TBI treatment, eliminating the RNOS and inhibiting subsequent neuroinflammation [15,26,56,57]. Necessarily, exploiting new biocatalysts with high catalytic activities and favorable biocompatibility, and developing new delivery modes for the biocatalysts for the noninvasive intervention of TBI remain constant challenges in this field [15,43]. As more biocatalysts with distinguished catalytic activities against oxidative stress are under development [20,58], there will be more potential therapeutics in the use of noninvasive intervention against TBI.

Conclusions
In summary, we have reported a noninvasive therapeutic to treat brain trauma at the early stage with gold clusters of well-defined structure and high catalytic selectively. Compared with reported nanozymes, the gold clusters showed higher catalytic activities toward water splitting and reduction of O 2 and H 2 O 2 by introducing the Cu or Cd catalytic active site. The biological results reveal that clusters can modulate the oxidative stress and further alleviate neuroinflammation by noninvasive adminstration. Besides, these results conclude that Au 24 Cd 1 preferentially decreases IL-1β and IL-6, while Au 24 Cu 1 shows the tendency to decrease TNF-α, indicating their different selectivity for alleviating neuroinflammation. Behaviors and histology illustrated their favorable effects against TBI after noninvasive administration. In conclusion, our work shows the great potential of noninvasive early intervention for the treatment of TBI with enzymatic biocatalysts.

Materials and reagents
All chemicals and reagents were purchased from commercial sources and used without further purification. Gold chloride (HAuCl 4 ·3H 2 O) was purchased from Sigma-Aldrich. Copper nitrate (Cu(NO 3 ) 2 ), Cadmium nitrate (Cd(NO 3 ) 2 ), 3-Mercaptopropionic acid (MPA), Sodium borohydride (NaBH 4 ), Sodium hydroxide (NaOH), were purchased from Aladdin. Ultrapure water (18.2 MΩ*cm) was used for all the experiments. Kits and fluorescent probes were purchased from commercial sources and used as per the instructions.

Materials preparation, characterization and catalytic activities
The MPA-protected Au 25 , Au 24 Cu 1 and Au 24 Cd 1 clusters were prepared as per the previous reports [26,59,60]. Briefly, HAuCl 4 (aqueous, 20 mM, 0.25 mL) and MPA (aqueous, 5 mM, 2 mL) were added to water (2.35 mL) and stirred for 5 min at room temperature. Then, NaOH solution (aqueous, 1 M, 0.3 mL) was added to the reaction mixture, followed by the addition of 0.1 mL of NaBH 4 (43 mg of NaBH 4 powder in 10 mL of 0.2 M NaOH solution). Au 25 MPA 18 was collected after the final reaction mixture stirred at room temperature for 3 h in the dark and aged at 4 °C for 12 h. The Au 24 Cu 1 and Au 24 Cd 1 were synthesized based on the same method, except the Au atoms in HAuCl 4 (20 mM, 0.25 mL) were replaced by various nitrate metal ions (Cu 2+ , Cd 2+ ) at a 4% molar ratio. Ultrafiltration tubes of 3 and 10 K at 3500 rpm/min were used for ultrafiltration to remove smaller organic ligands and larger-sized clusters, and lyophilization was used to collect the purified product for further test and investigation.
A JEM-2100 F electron microscope (JEOL, Japan) was employed to acquire transmission electron microscopic (TEM) images. A Malvern Zetasizer Nano ZS90 (UK) was employed to measure dynamic light scattering (DLS) to test the hydrodynamic size and determine the zeta potential of clusters.
The XPS spectrum of the metal elements was performed using ESCALAB Xi+ spectrometer, with a monochromatic Al Kα X-ray source (Thermo Fisher Scientific). All XPS spectra are processed with XPSPEAK41 software and corrected with the peak of C1s (standard: 428.8 eV). ICP-MS (Agilent, US) was used to determine the content of each metal element in the clusters.
The clusters were deposited onto the surface of glassy carbon (GC) electrodes for electrochemical assay [14,40]. Briefly, as-prepared MPA-protected Au 25 , Au 24 Cu 1 and Au 24 Cd 1 clusters (20 µL, 0.5 mg/mL) were deposited onto the surface of GC electrodes and dried naturally, then Nafion solution (3 µL) was dropped onto the surface of GC electrodes and dried, the modified electrodes were used for catalytic activity tests.
Cyclic Voltammetry (CV, scan rate: HER, 10 mV s −1 ; OER, 100 mV s −1 ) and Linear Sweep Voltammetry (LSV, scan rate: HER, 1 mV s −1 ; OER, 5 mV s −1 ) measurements were taken electrochemical analyzer, CHI760E, Shanghai) to evaluate the performance of the clusters modified electrodes for HER and OER. HER was carried out in 0.5 M H 2 SO 4 , and a graphite rod and a saturated calomel electrode were used as the counter electrode and reference electrode, respectively. OER was carried out in 1 M KOH, and a platinum wire and an Ag/AgCl electrode were used as the counter electrode and reference electrode, respectively. CV (scan rate: 50 mV s −1 ) and LSV (scan rate: 5 mV s −1 ) measurements were taken (electrochemical analyzer, CHI760E, Shanghai) to evaluate the catalytic activities for O 2 and H 2 O 2 reduction. A three-electrode cell was adopted for both O 2 and H 2 O 2 reduction, a platinum wire and a saturated calomel electrode were used as the counter electrode and reference electrode, respectively. Oxygen reduction reaction (ORR) was carried out in O 2 -saturated in 0.01 M PBS (pH 7.4), and the H 2 O 2 reduction was performed in the presence of 9.8 mM H 2 O 2 in N 2 -saturated 0.01 M PBS (pH 7.4).
Cell viability and Immunofluorescence. HT22 was used in all the cellular experiments. Cells were seeded into the 96-well plates and grew in Dulbecco's modified Eagle's medium (DMEM) at 37 °C with 5% CO 2 . After being stimulated with 100 µM H 2 O 2 for 6 h, the culture medium was replaced by Au 25 , Au 24 Cu 1 and Au 24 Cd 1 dissolved in the DMEM at different doses, and then cells were incubated for another 24 h. HT22 cells (1 × 10 5 cells per well) were seeded in 6-well plates for 12 h and stimulated with 100 µM H 2 O 2 for 6 h before being treated with the clusters (12 ng/µL). Cell survival was determined at the MTT concentration of 5 mg/mL for 2.5 h and detected at optical density (OD) 490 nm. Fluorescent staining was carried out to evaluate the intracellular oxidative stress levels with different probes such as DHE for O 2 ·− and DCFH-DA for ROS, and cell images were captured by a fluorescence microscope.

In vivo treatment and behavioral experiment
TBI models: C57BL/6 mice at 21-23 g were employed to establish TBI models using an electromagnetically CCI injury device (eCCI-6.3, Custom Design & Fabrication, Inc), with an impactor of 5 m/s velocity, 0.61 mm depth, 150 ms duration, and 20º angle of dura mater on the vertical axis. The mice were divided into control, TBI, TBI+Au 25 , TBI+Au 24 Cu 1 , and TBI+ Au 24 Cd 1 group (n = 15) randomly. All mice were anesthetized with 10 % chloral hydrate (10 mg/kg) and the scalp was cut before placed on the stereotaxic frame. The craniotomy was carried out by drilling the skull in a circle of 2 mm in diameter. The scalp was sewn together carefully, and the clusters were added to the wound of TBI mice at a concentration of 50 mg/kg. The healing process was recorded photographically and the wound remaining was calculated after treatment.
Oxidative stress level: Brain tissues were taken out on days 7 and 14 post-treatment, then homogenized in 0.9% physiological saline and analyzed for SOD, GSH/GSSG, MDA, and H 2 O 2 using commercially available kits. All testing methods are carried out as per the instructions (Beyotime).
Morris water maze tests: Morris water maze (MWM) was conducted on days 13-17 and 26-30 post-treatment as described in the previously reported literature [12,61,62]. Briefly, the water maze was divided into four quadrants, and the platform was set in the center of quadrant I. Before spatial learning, visual discrimination learning was performed to determine whether the vision of mice was normal. In this procedure, each animal performed one trial where the platform was placed above the water to determine whether the vision of mice was normal. Animal with the visual problem would be excluded in the Morris water maze test. Each mouse was put into the pool to be trained and learned to search for the platform under the water in the order of quadrant II, III, IV and I with an inter-trial interval (ITI) of 60 min at almost the same time of each day for five days. The test was carried out without the platform on the fifth day, and each mouse was put into the pool at quadrant II and allowed for 60 s to track them.

Ex vivo verification
Western Blotting: The total protein in the brain tissue was extracted and the content of IL-1β (1:1000), IL-6 (1:1000) and TNF-α (1:1000) was analyzed. SDS-PAGE electrophoresis was performed before transferring to the membrane. Immune responses of the specific antibodies were carried out and the images were captured with autoradiography. All antibodies were purchased from Abcam.
ELISA analysis: Inflammatory cytokines including IL-6, IL-1β and TNF-α were determined by ELISA kits (Abcam, ab100712, ab197742, ab208348, respectively), and the assays were performed as per the instructions provided by the manufacture.
Tissue staining: Brain tissues were taken out at 30 days post-injury and fixed in 4% paraformaldehyde and embedded in paraffin. Immunofluorescent staining was performed with primary antibodies including anti-GFAP, IL-6, IL-1β antibodies as per the instructions of (Abcam). Then the slices were incubated with Alexa Fluor 488/594-conjugated goat secondary antibody for 1-1.5 h at room temperature under dark and counterstained with DAPI. Immunohistochemical staining for IL-6 and mice brain tissue was performed according to the instructions (Proteintech).
Toxicological studies: The biosafety of clusters was measured on male C57BL/6J mice at 7-8 weeks (21-23 g). Mice were treated with 200 µL of clusters at a dose of 50 mg/kg every other day. Hematology and blood biochemical indicators were tested on day 30.
The blood sample was obtained from the retroorbital sinus and stored in a test tube containing K2EDTA for testing. The blood sample used for biochemical analysis was allowed to stand for 30 min, and then centrifuged twice at 3500 rpm for 15 min each time, the supernatant was collected and tested. The main organs of the mice were collected and fixed in 4 % paraformaldehyde for 24 h, embedded in paraffin, and mounted on a glass slide (4 μm coronal section). The slices were stained with hematoxylin and eosin to observe the toxicity of clusters in major organs, including the heart, liver, spleen, lung, kidney and testis.

Statistic methods
Data are presented as mean ± standard deviation (SD) or standard error of the mean (SEM). For multiple comparisons, one-way analysis of variance (ANOVA) was performed using the SPSS 22 software to assess difference in means among groups and compared with the Con and TBI groups, analyzed by ANOVA.
Additional file 1: Figure S1. The content of metal elements in the clusters were quantified by ICP-MS. Figure S2. X-ray photoelectron spectroscopy of clusters. Figure S3. Cell survival of (a) HT22, (b) BV2 and (c) MA-c cells in the presence of various concentrations of clusters. Figure  S4. Hematology of mice treated with or without gold clusters on day 30. Figure S5. Blood biochemistry analysis of mice treated with or without gold clusters on day 30. Figure S6. Histology of major organs in mice.