Se@SiO2 nanocomposites suppress microglia-mediated reactive oxygen species during spinal cord injury in rats

Selenium (Se) is an essential trace element with strong antioxidant activity, showing a great prospect in the treatment of spinal cord injury (SCI). However, the narrow gap between the beneficial and toxic effects has limited its further clinical application. In this experiment, we used porous Se@SiO2 nanocomposites (Se@SiO2) modified by nanotechnology as a new means of release control to investigate the anti-oxidative effect in SCI. In vitro Se@SiO2 toxicity, anti-oxidative and anti-inflammatory effects on microglia were assayed. In vivo we investigated the protective effect of Se@SiO2 to SCI rats. Neurological function was evaluated by Basso, Beattie and Bresnahan (BBB). The histopathological analysis, microglia activation, oxidative stress, inflammatory factors (TNF-α, IL-1β and IL-6) and apoptosis were detected at 3 and 14 days after SCI. The favorable biocompatibility of Se@SiO2 suppressed microglia activation, which is known to be associated with oxidative stress and inflammation in vivo and in vitro. In addition, Se@SiO2 improved the rat neurological function and reduced apoptosis via caspase-3, Bax and Bcl-2 pathways in SCI. Se@SiO2 was able to treat SCI and reduce oxidative stress, inflammation and apoptosis induced by microglia activation, which may provide a novel and safe strategy for clinical application.


Introduction
Spinal cord injury (SCI) is a serious clinical trauma with a high disability rate. 1 With changes in social economy and lifestyle, the incidence of SCI is increasing annually. 2 Patients with SCI oen present with severe limb disorders and poor prognosis, resulting in severe economic and social burdens. The pathological mechanism of SCI is complex and remains poorly understood. 3 A number of factors are involved in the development of SCI, including oxidative stress, 4,5 inammation 6 and apoptosis. 7 A large number of studies have shown that oxidative stress is an important factor in SCI. Primary traumatic SCI induces the release of reactive oxygen species (ROS) and inammatory cytokines, resulting in the activation of microglia. 6 Microglia are a type of glial cells located throughout the central nervous system (CNS) and have diverse important functions in response to environmental changes in the spinal cord. 8 Excessive activation of microglia may induce the release of large amounts of ROS and inammatory cytokines, leading to more serious secondary SCI. 9,10 Oxidative stress is closely related to ROS. 11 ROS can produce by cell lysis and oxidative burst, 12 the presence of an excess of free transition metals 13 and normal cellular respiration. 14 There are free radical defense mechanisms in the CNS, mainly including superoxide dismutase (SOD) and other antioxidants, and chelation of transition metal catalysts. 15 There is a balance between the production and clearance of free radicals. Microglia activation at the injured site of the spinal cord can produce large amounts of ROS, leading to imbalance of oxidative stress, which may further lead to lipid peroxidation of the cell membrane, degeneration of intracellular proteins and enzymes, DNA damage and neuronal death or apoptosis. 16 Selenium (Se) is one of the essential micronutrients for human beings. 17 It is also the critical catalyst for glutathione peroxidase (GPx) 18,19 and mammalian thioredoxin reductase, 20 two important anti-oxidant enzymes that can suppress oxidative stress. Se deciency is associated with cancer, viral infections, and a variety of cardiovascular diseases, such as Keshan disease. 21 Many studies have shown that daily intake of 200 mg Se can reduce the incidence of cancer remarkably. 22 Se can reduce myocardial injury induced by oxidative stress, reduce the infarct size and improve ventricular remodeling aer infarction by increasing the expression and activity of GPx in serum. 23,24 Studies have shown that Se has a good neuroprotective effect by inhibiting ROS-mediated apoptosis in traumatic brain injury. 25,26 Therefore, the use of antioxidant Se for the treatment of SCI has a promising prospect. Compared with other antioxidants, Se is more stable and economical. However, the direct use of pure Se (nano-scale or marco-scale) has many limitations in clinical treatment due to the narrow margin between the benecial and toxic effects, which limited Se further application. 27,28 To solve this problem, an effective way is to directly deliver Se to target organization through nanocarriers, which is expected to achieve sustained drugs release, depress the toxic side effects and simultaneously enhance the therapeutic efficiency. 29 Various nanocarriers have been widely studied, such as liposomes, 30 nanocapsules, 31 dendrimers, 32 gold nanostructures, 33 silica nanostructures, 34,35 etc. However, silica nanostructures have unprecedented advantages such as good compatibility, adjustable pore structure, and simple, lowcost and controllable fabrication. Recently, Liu et al. 36 reported the use of porous Se@SiO 2 nanocomposites as a new strategy of nano-Se controlled release, which shown good biocompatibility for normal cells and could be used as drug carriers. Nano-modied Se could not only reduce the risk of Se toxicity 37,38 but showed high biological activity both in vivo and in vitro. 39,40 Some recent studies have proven that porous Se@SiO 2 nanocomposites shown good safety and antioxidant effects. Zhu et al. found that porous Se@SiO 2 nanospheres can be used to treat paraquat-induced acute lung injury by resisting oxidative stress. 41 Deng et al. found that porous Se@SiO 2 nanospheres can protect steroid-induced osteonecrosis of the femoral head 42 and protect the femoral head from methylprednisolone-induced osteonecrosis by resisting oxidative stress. 43 The use of porous silica nanocarriers to control the release of nano-Se could achieve a long-term effective concentration of Se, thus improve the therapeutic outcome. 42 Porous Se@SiO 2 nanocomposites are a new type of material with a good anti-oxidative effect and nano-Se controlled release, which represents a promising, safe and effective way for SCI treatments. However, its effect on SCI and related mechanisms have not been reported. The aim of this study was to research whether porous Se@SiO 2 nanospheres could signicantly alleviate oxidative stress and the inammatory response of microglia-mediated during SCI in rats, and explored the underlying mechanism in vitro and in vivo.

Synthesis and characterization of porous Se@SiO 2 nanocomposites
Porous Se@SiO 2 nanocomposites were prepared according to the previous method 36 and provided by Shanghai University of Engineering Science (Shanghai, China). The surface characteristics of the prepared material were examined using D/max-2550 PC X-ray diffractometry (XRD; Rigaku, Cu-Ka radiation) and transmission electron microscopy (TEM; JEM-2100F). The cumulative release kinetics of Se elements in the porous Se@SiO 2 and nanospheres were detected in PBS (0.01 M, pH 7.4). They were then dissolved in deionized water for subsequent experiments.

Animals
Sprague-Dawley (SD) rats (female, 200-220 g) and suckling SD rats born within 24 h were purchased from the experimental animal center of the Second Military Medical University (Shanghai, China). Before the experiment, rats were housed in an experimental animal center at 20-22 C without specic pathogens for 1-2 weeks to adapt to the environment, with free access to food and water. All animal experiments were approved by the animal ethics committee of the Second Military Medical University. All experiments were performed in compliance with the Regulations for the Administration of Affairs Concerning Experimental Animals in Shanghai.

Culture and identication of primary microglia
Based on classical glial cell culture reported by McCarthy et al., 44 primary microglia were extracted using a modied shaking method. Briey, the 24 h newborn SD rat was submerged in alcohol for 10 min for disinfection. The skull was stripped and placed in an ice-PBS buffer. The pia mater, blood vessels, hippocampus and midbrain were removed under a dissecting microscope. The isolated cortex was placed in trypsin, cut into small pieces and digested for 5 min. Aer digestion, the cells were repeatedly washed, dispersed into a single cell suspension and ltered through a 400 mesh membrane to prepare a whole brain single cell suspension. The single cell suspensions was resuspended in DMEM/F12 medium and seeded in polylysinecoated T25 cells at 4 Â 10 8 per L cells. The cells were incubated in a 37 C, 5% CO 2 and saturated humidity incubator and shaken at 260 rpm at 37 C for 2 h. The shaken-down cell suspension was collected and inoculated into a T25 ask. Aer 1 h conventional culture, the untreated cells were discarded to obtain puried microglia. Purity of the microglia was tested by Iba-1 immunouorescence staining.

Toxicity of porous Se@SiO 2 nanocomposites on microglia in vitro
To investigate the effect of different concentrations of porous Se@SiO 2 nanocomposites on microglia toxicity, primary microglia 5 Â 10 3 cells per well and 100 ml complete medium were inoculated into 96-well plates. Aer adherence, they were replaced with different intervention media: 0, 10, 20, 40, 80, 160 and 320 mg ml À1 porous Se@SiO 2 nanocomposites and cultured for 24 h. Cell counting kit 8 (CCK-8, Biyuntian, China) was used in strict accordance with the test procedure instructions. Cytotoxicity of different concentrations of porous Se@SiO 2 nanocomposites on microglia was examined using a microplate reader (Bio Tek, USA) at 450 nm. The experiment was repeated 3 times with 2 replicates per condition.
Effects of Se@SiO 2 on oxidative stress injury of microglia mediated by H 2 O 2 in vitro First, the median lethal dose of H 2 O 2 against microglia (median, lethal, dose, LD 50 ) was detected, and this concentration was used in subsequent trials. Microglia 5 Â 10 3 cells per well in 100 ml complete medium were inoculated into 96well plates and then incubated in DF-12 medium with out serum containing 0, 100, 200, 400, 600, 800 and 1000 mM H 2 O 2 for 24 h. Cytotoxicity was measured using CCK-8. 400 mM H 2 O 2 was selected as the treatment condition of microglia to induce oxidative stress. Based on the effect of different concentrations on the viability of microglia, 10 and 40 mg ml À1 Se@SiO 2 were used to study the Se@SiO 2 dose effect in this experiment. The primary microglia were cultured in 6well plates at a density of 1 Â 10 5 per well for 12 h. Aer complete adherence of the microglia, the 6-well plates were divided into a blank, control, 10 mg ml À1 Se, and 40 mg ml À1 Se groups. The blank group was incubated for 24 h in a complete culture medium containing 0 mg ml À1 of porous Se@SiO 2 nanocomposites. The control group was incubated for 24 h in a complete medium containing 400 mM H 2 O 2 . While the 10 mg ml À1 Se and 40 mg ml À1 Se groups were incubated for 24 h using 400 mM H 2 O 2 and complete culture medium containing 10, 40 mg ml À1 porous Se@SiO 2 nanocomposites, respectively. Aer 24 h, the supernatant and adherent cells were collected. CCK-8 assay was used to detect microglia viability. Malondialdehyde (MDA) and SOD kits were used to detect intracellular oxidative stress levels. TNFa, IL-1b and IL-6 ELISA kits were used to detect cell supernatant inammatory factors. The experiment was repeated 3 times with 2 replicates per condition.

Establishment of the SCI rat model and implantation of the intrathecal catheter
The rat was anesthetized by pentobarbital sodium (50 mg kg À1 , intraperitoneal injection) and xed in a bayonet-type rat xator. The intrathecal catheter implantation rat model was established as Yaksh and Rudy described. 45 Aer successful anesthesia, the lamina was cut off from the lumbar region, and the PE-10 catheter was inserted. Effusion of cerebrospinal uid (CSF) along the catheter was regarded as successful catheterization. Each rat was housed in a single cage. Aer full awakening, rats with motor dysfunction were removed and 10 ml 2% lidocaine was injected along the catheter. The presence of movement dysfunction in both lower extremities of the rat 30 seconds aer injection indicates successful catheterization. Aer 1 day successful establishment of the intrathecal catheter implantation rat model, improved Allen SCI model of contusion was used with modications as previously described. 46,47 A 10 g weighted hammer fell from a height of 50 mm and impinged on rat thoracic 9-11 vertebrae. For the rst 3 d aer surgery, 50 000 units conventional penicillin was delivered via intramuscular injection to prevent infection. The rats received abdominal massages every 6 h aer the procedure to help them urinate until they were capable of urinating by themselves.
In vivo experimental design and sample collection 135 rats were randomly divided into a sham, control and Se@SiO 2 group, with an average of 45 rats per group. Aer successful establishment of the model, porous Se@SiO 2 nanocomposites (1 mg kg À1 , 10 ml) or normal saline (NS, 10 ml) was injected intrathecally into Se@SiO 2 and control groups. The concentration of porous Se@SiO 2 nanocomposites was selected according to our previous research. 41,42 Se@SiO 2 (1 mg kg À1 , 10 ml) or 10 ml NS was administered gently into the restrained conscious rats by using a microinjector (Gaoge, Shanghai, China) attached to the intrathecal catheter within 2 min, followed by a ush with 20 ml NS. Se@SiO 2 and the vehicle were administered immediately following SCI and daily aer surgery. 15 rats in each group were sacriced with the overdose anesthesia aer 3 and 14 days SCI respectively. The 3 mm around the center of the damaged spinal cord was sampled and collected. 5 rats in each group were sacriced for western blotting analysis. 5 rats in each group were examined for MDA, SOD and inammatory factors (TNF-a, IL-1b and IL-6). The remaining 5 rats in each group were subjected to histological and immu-nouorescence assays.

Histopathological observation of the damaged spinal cord segment
The spinal cord was xed in 4% PFA, stained with hematoxylineosin (HE), sliced into 5 mm sections, and observed for morphological changes under a microscope aer 3 and 14 days SCI respectively. The quantication percentage of cavity area were calculated 14 days aer SCI (percentage of cavity area ¼ cavity area/total area Â 100%, n ¼ 5).

Detection of oxidative stress by SOD and MDA kits
To conrm the change in oxidative stress as represented by the content of SOD and MDA aer Se@SiO 2 treatment, the injured spinal cord tissue and Se@SiO 2 -treated microglia were collected and tested by the SOD and MDA kits, knowing that MDA is an indicator of lipid oxidation and SOD is an indicator of antioxidant ability. Then, adherent cells or 100 mg spinal tissue was homogenized immediately in 1 ml NS, and the homogenate was centrifuged at 2000 g for 15 min at 4 C. The protein concentration in the supernatant was determined by BCA assay, and the SOD activity (U m À1 ) and MDA concentration (nmol mg À1 protein) were measured using SOD and MDA kits (Nanjing Jiancheng Bioengineering Institute, Nanjing, China).

Detection of inammation by ELISA
To conrm changes in inammatory factors (TNF-a, IL-1b and IL-6) aer Se@SiO 2 treatment, the injured spinal cord tissue and Se@SiO 2 -treated microglia were collected and tested by ELISA kits (R&D Systems, Minneapolis, MN, USA). 100 mg spinal tissue or microglia cell supernatant was homogenized immediately in 1 ml NS, and the homogenate was centrifuged at 2000 g for 15 min at 4 C. The protein concentration in the supernatant was determined by BCA assay. ELISA kit manual operation tested in strict accordance to standard protocols and used the microplate reader (ELx800, BioTek) at 450 nm to calculate the content of the sample.

Western blot
3 and 14 days aer SCI, 3 mm spinal cord in the center of the lesion was collected, frozen and stored (n ¼ 5). The samples were dissolved in the cell lysate and then centrifuged at 14 000 g at 4 C for 15 min, and the protein concentration in the sample was determined by BCA. The protein concentration in each sample was adjusted to 50 mg. The sample was processed according to the regulations strictly. The primary antibody Iba-1 (1 : 500, Abcam ab5076, Cambridge, UK), caspase-3 (1 : 500, cell signaling #9662, USA), Bax (1 : 1000, Abcam ab32503, Cambridge, UK) and Bcl-2 (1 : 500, Abcam ab59348, Cambridge, UK) antibody were incubated overnight and then labeled with the corresponding secondary antibody (Jackson 1: 2000). The ratio of optical density of band to optical density of GAPDH band was calculated as the relative expression level of the sample protein level.
Immunohistochemistry 3 and 14 days aer SCI, the injured spinal card was harvested for immunohistochemistry to detect the expressions of Ibl-1 and neurolament 200 (NF200). The rats were killed by overdose anesthesia and infused with pre-cooling PBS for 10 min rstly and then perfused with 4% PFA for 20 min. The harvested spinal cord was xed in 4% PFA for 4 h, dehydrated with 20% sucrose for 24 h, and then dehydrated with 30% sucrose for 48 h. Aer tissue deposition, the tissue was xed with OTC, sliced into 10 mm sections with a microtome, immunohistochemically stained, washed with PBS for 3 times, incubated overnight with primary antibodies anti-Iba-1 (1 : 200, Abcam, Cambridge, UK) and mouse anti-NF200 (1 : 200, Abcam, Cambridge, UK) at room temperature, washed again with PBS for 3 times, incubated with the corresponding secondary antibody for 2 h, dyed with dapi, and sealed with 50% PBS glycerol. Finally, images were captured with a uorescence microscope (Olympus, Tokyo, Japan) by randomly selecting 10 elds at the spinal tissue. For quantication, the percentage of positive cells was calculated as (number of positive cells)/(total number of cells in the eld) Â 100%.
Terminal deoxynucleotidyl transferase dUTP nick-end labeling (TUNEL) staining Aer TUNEL staining 3 and 14 days aer SCI, the number of apoptotic cells in the injured spinal card was quantied with a uorescence detection kit (Roche Molecular Biochemicals, Indianapolis, IN, USA). The sections were de-paraffinized, rehydrated at 60 C, incubated in a 20 mg ml À1 proteinase K working solution for 15 min at room temperature following standard protocols, and observed under a microscope (Carl Zeiss Jena, Oberkochen, Germany). The number of apoptotic cells and the total number of cells were determined using Image-Pro Plus 6.0 (Media Cybernetics, Silver. Spring, USA). Ten random elds per section were analyzed and the apoptosis index was calculated (n ¼ 5).
Hind-limb motor function aer SCI modeling in rats 45 rats were placed on the platform before and 1, 3, 7, 14, 21 and 28 d aer modeling, and scored by the Basso, Beattie and Bresnahan (BBB) scale. 48 The activities of the hip, knee, ankle and trunk movement and coordination were recorded by two technicians who were blind to the experimental grouping. The scores of two individuals were compared; if they were found to be of different, a joint evaluation was required (n ¼ 15).

Statistical analysis
SPSS 21 soware (SPSS, Chicago, IL, USA) was used for data statistical analysis. Data were expressed as mean AE standard deviation (SD). The hind-limb movement and neurological function scores were analyzed by ANOVA and Fisher's LSD post hoc test. The other indexes were evaluated by ANOVA. P < 0.05 were considered statistically signicant.

Properties of porous Se@SiO 2 nanocomposites
A series of tests on XRD (Fig. 1a), cumulative release kinetics (Fig. 1b) and TEM ( Fig. 1c and d) conrmed that the nanocomposites used in this test were porous Se@SiO 2 nanocomposites with good Se controlled release. The structure of the nanocomposites was detected by XRD spectra, showing (100), (011), (110) and (012) clear characteristic peaks, which were the same as those of the hexagonal system referenced by the standard Se (JCPDS, card, no. 65-1876). The amorphous silica Se@SiO 2 nanocomposites exhibited a stable increase in the lowangle region in the XRD spectra. TEM showed that the diameter of the homogeneous nanocomposites was about 50 nm. The dispersed nanoparticles had an interplanar spacing about 0.218 nm, which matched to the spacing of the standard hexagonal Se (110) crystal planes, further conrming that the small nanoparticles dispersed in the silica were Se nanometer crystals. The previous Brunauer-Emmett-Teller (BET) characterization showed that Se@SiO 2 nanocomposites were porous. 36 It was found that the pyrrolidinylquinoxaline (PVP) penetrated into the silica shell and formed a channel in the Se@SiO 2 nanocomposite, through which very small nanoparticles were able to release from the porous Se@SiO 2 nanocomposite material. The Se element in Se@SiO 2 nanocomposites could achieve controlled release at 37 C 0.01 M PBS (Fig. 1b).

Morphology and purity of the primary microglia
Aer 24 h mixed culture, adherent cells showed light transmitting microglia, astrocytes, neurons and some red blood cells. Cultured glial cells began stratifying aer 7-9 days, with astrocytes and neurons in the lower layer, and semi-adherent, refractive and round microglia and a small amount of oligodendrocytes in the upper layer. The microglia separated by constant temperature shaking method were round, refractive and uniformly sized. Aer 24 h inoculation of the puried microglia, cells adhered to the wall and became irregular, spindle shaped, branched or spider-like, without obvious proliferation. Immunouorescence staining with microglia  specic antibody Iba-1 showed that the positive rate of the primary microglia was over 95%, indicating that the purity of the primary microglia was high (Fig. 2).

Se@SiO 2 biocompatibility test and the median lethal H 2 O 2 concentration in primary microglia in vitro
As shown in Fig. 2c, the viability of the primary microglia was insignicantly affected when the concentration of Se@SiO 2 was lower than 80 mg ml À1 . When the solubility was higher than 80 mg ml À1 , obvious toxicity could be observed, indicating that the porous Se@SiO 2 nanocomposites had good biocompatibility. Cell viability was detected by CCK-8 aer application of different concentrations of H 2 O 2 (0, 100, 200, 400, 600, 800 and 1000 mM) to the primary microglia for 24 h (Fig. 2). The viability of the 7 groups were signicantly different, and cytotoxicity was positively correlated with the H 2 O 2 concentration. When the concentration of H 2 O 2 was 400 mM, the viability of the microglia decreased by about 50%, which was the half lethal dose of microglia and used as the toxicity condition for microglia in the following test (Fig. 2).

Se@SiO 2 reduces H 2 O 2 -induced microglia cytotoxicity, oxidative stress injury and release of inammatory factors in vitro
Based on the median lethal H 2 O 2 concentration of the primary microglia, 400 mM was used as a toxicity concentration. The results of CCK-8 showed that Se@SiO 2 could reduce the cytotoxicity of H 2 O 2 to primary microglia markedly, and the survival rate of 40 mg ml À1 group was signicantly higher than that of 10 mg ml À1 group (Fig. 3a, P < 0.05), indicating that the higher the concentration of Se@SiO 2 , the stronger the ability to reduce cytotoxicity. MDA and SOD assays showed that Se@SiO 2 could signicantly reduce oxidative stress in microglia aer H 2 O 2 toxicity, and the level of oxidative stress in 40 mg ml À1 group was signicantly lower than that in 10 mg ml À1 group (Fig. 3b and c, P < 0.05). The results of ELISA showed that the concentration of TNF-a, IL-1b and IL-6 in the supernatant of microglia demonstrated that Se@SiO 2 could signicantly reduce the intensity of inammatory reaction of microglia aer H 2 O 2 toxicity, and the level of inammatory reaction in 40 mg ml À1 group was significantly lower than that in 10 mg ml À1 group (Fig. 3d-f, P < 0.05). This journal is © The Royal Society of Chemistry 2018 Se@SiO 2 enhanced axonal regeneration and improved rat hind-limb motor function aer SCI BBB scores and axonal regeneration were used to evaluate the effect of Se@SiO 2 on functional recovery. It was found that the BBB scores increased progressively in both SCI and Se@SiO 2 groups over time, indicating natural nerve recovery aer SCI in rats. Rats receiving Se@SiO 2 had signicantly higher BBB scores than those in SCI group (Fig. 4e, P < 0.05, n ¼ 15). NF200 was used to identify the axonal regeneration in the spinal cord lesion 14 days aer SCI. NF200 is the marker of the medium and large neuron, which can indirectly reect the degree of axonal injury and repair. The NF200 positive axons were distributed in the vicinity of the injury center, and the number was much higher aer SCI. The number of NF200 positive axons in the Se@SiO 2 group was signicantly higher than that of the SCI group and sham group (Fig. 4a-d, P < 0.05, n ¼ 5).

Histopathological analysis
A clear boundary between the gray matter and the white matter of the spinal cord tissue was observed in sham group. The outline of nerve cells was clear, the cytoplasm was homogeneous and deeply stained, and the whole tissue structure was intact and clear. The boundary became obscure 3 days aer SCI. A large number of necrotic nerve cells, atrophic cytoplasm, degeneration and vacuolar degeneration of cytoplasm, and disorganized arrangement of nerve bers were observed. There was hemorrhage, liquefaction and inammatory cell inltration in the middle of the spinal cord lesion. Cell death, hemorrhage and inammatory cell inltration at the spinal cord lesion were attenuated markedly aer administration Se@SiO 2 . The neuronal structure was complete; some neuronal cell vacuolar degeneration, a small amount of inammatory cell inltration, and nerve bers arranged in an orderly manner. Fourteen days aer SCI, formation of voids was observed in the spinal cord of SCI group. The boundary between the gray matter and the white matter of the spinal cord was unclear, and inammatory cell inltration was observed. In contrast, the spinal cavity became smaller and inammatory cell inltration was reduced in Se@SiO 2 group. The quantication percentage of cavity area in the Se@SiO 2 group was much lower than that in the SCI group 14 days aer SCI (Fig. 5, P < 0.05, n ¼ 5).

The anti-oxidative and anti-inammatory effect of Se@SiO 2 in vivo
The content of MDA and SOD was detected 3 and 14 days aer SCI to determine the level of spinal cord oxidative stress ( Fig. 6a  and b). Rats with SCI presented with signicantly reduced SOD activity and increased MDA content at the lesion site aer SCI, especially 3 days aer SCI. Compared with SCI group, the level of MDA decreased signicantly, and SOD activity increased signicantly in Se@SiO 2 group aer 3 and 14 days (P < 0.05). Se@SiO 2 intrathecal injection could reduce the level of oxidative stress of the damaged spinal cord markedly 3 and 14 days aer SCI. The contents of TNF-a, IL-1b and IL-6 were detected 3 and 14 days aer SCI by ELISA to determine SCI inammation (Fig. 6c-e). TNF-a, IL-1b and IL-6 were major indicators of inammation in the spinal cord. Compared with sham group, TNF-a, IL-1b and IL-6 levels in SCI group increased signicantly aer SCI (P < 0.05). However, the TNF-a, IL-1b and IL-6 contents reduced signicantly aer Se@SiO 2 treatment (P < 0.05). In addition, TNF-a, IL-1b and IL-6 levels decreased markedly in SCI group aer 14 days treatment compared with those aer 3 days treatment (P < 0.05).

Se@SiO 2 suppresses microglia activation in vivo
The activation of microglia played a critical role in the pathogenesis of SCI. To study the effect of Se@SiO 2 on ROS and inammation, microglia in the damaged spinal cord lesion were tested by immunouorescence ( Fig. 7a and b) and Western blot of Iba-1 protein (Fig. 7c and d). The number of microglia (Iba-1+) and Iba-1 protein increased signicantly 3 and 14 days aer SCI as compared with sham group (P < 0.05), while the number of microglia and Iba-1 protein in Se@SiO 2 group decreased signicantly compared with SCI group 3 and 14 days The anti-apoptotic effect of Se@SiO 2 in vivo TUNEL and western-blot showed that Se@SiO 2 exerted an antiapoptotic effect aer SCI. TUNEL staining was used to evaluate the cell apoptosis rate (Fig. 8a and b). The content of caspase-3, Bcl-2 and Bax in the spinal site as detected by western blot (Fig. 8c-e). TUNEL staining showed that the number of apoptosis cells increased signicantly in SCI group at 3 and 14 days aer SCI, but decreased markedly in Se@SiO 2 group compared with SCI group at 14 days (P < 0.05). Western blot assay shown that SCI increased the caspase-3 and Bax in the spinal lesion site as compared to that in sham group, but markedly decreased the levels of Bcl-2 (P < 0.05). Western blot assay also validated that the protein content of caspase-3 and Bax increased signicantly in SCI group compared with sham group, while the Bcl-2 decreased signicantly 14 days aer SCI (P < 0.05).

Discussion
Oxidative stress damage caused by the imbalance of ROS production and the decrease in antioxidant capacity of the cell is thought to initiate and be the main mechanism of multiple neurological diseases such as Alzheimer's disease, 49 Parkinson's disease, 50 amyotrophic lateral sclerosis 51 and traumatic brain and SCI. 52 Se is one of the essential elements of the human body. At present, 25 kinds of selenoproteins have been found in human proteome, including the glutathione peroxidase family (GPxs), thiored oxidoreductase family (TrxRs), and deiodinase family. In addition to selenoprotein P, other selenoproteins contain a common active center of selenocysteine (Sec). Most Se proteins including GPxs and TrxRs are involved in the regulation of antioxidant defense and redox signaling. Se is also involved in the regulation of other special biological processes, including the biosynthesis of DNA, degradation of faulty folding proteins in the endoplasmic reticulum, regulation of redox transcription factors, apoptosis and immunoregulation, and transportation and storage of Se. 53 In this experiment, we investigated the anti-oxidative effect of porous Se@SiO 2 nanocomposites on microglia during SCI in vivo and in vitro. The newly prepared porous Se@SiO 2 nanocomposites are nano-modied in order to reduce the risk of nano-Se toxicity and improve the biological activity. Using porous, silica, nanocarriers to control the release of nano-Se can provide an effective concentration of Se for a long time and improve the treatment efficacy. Our in vitro experiment showed that porous Se@SiO 2 nanocomposites had favorable biocompatibility (Fig. 2). Knowing that exogenous H 2 O 2 can react with intracellular ions (iron or copper) to produce highly toxic ROS causing DNA damage and cell death, we used H 2 O 2 in this study to produce an oxidative stress injury model. 16 We found that porous Se@SiO 2 nanocomposites could reduce the ROS damage effectively against H 2 O 2 induced microglia activation and release of inammatory factors (Fig. 3). Microglia were characterized by high oxygen consumption, strong metabolism, easiness to produce oxygen free radicals, and susceptibility to ROS. 54 In vivo, we measured the content of microglia-specic protein Iba-1 by western blot and immunouorescence at different time points at the injury site of SCI in rats (Fig. 7) and found that Se could effectively reduce microglial hyperactivity in SCI-modeled rats. In addition, Se could markedly reduce the degree of oxidative stress and inammatory reaction (Fig. 6) of the injured segments in vivo. Porous Se@SiO 2 nanocomposites could signicantly suppress microglia activation related to oxidative stress and inammation aer SCI, improve the hindlimb neurological function (Fig. 4) and reduce apoptosis via the caspase-3 pathway (Fig. 8) in the rats aer SCI.
Se played a role in inhibiting microglial activation by signicantly reducing the production of ROS and the release of inammatory factors caused by H 2 O 2 . The specic molecular mechanism of Se in cells is the focus of current research. Studies have shown that Se can reduce the cellular oxidative stress level by increasing the expression or function of the intracellular GPxs and the TrxRs. 18 Studies have also shown that Se can inhibit the activation of 3 kinase/protein kinase B (PI3K/ AKT) and extracellular regulatory kinase (ERK) signaling pathways, and reduce the calcication of vascular smooth muscle cells induced by oxidative stress. Se can weaken the oxidative stress of activated endoplasmic reticulum and reduce apoptosis of smooth muscle cells. 55 Se can reduce the infarct size and the content of TNF-a aer ischemia by inhibiting the NF-kB pathway. 56 However, the specic molecular mechanisms involved in the inhibition of Se on microglial activation remain unclear, and need further studies.
Recent advances in nanoengineering have provided promising alternatives for the design of biocompatible micro-nano and nanocarriers for the controlled release of therapeutic agents to target CNS cells. 57-60 Some of the most promising nanostructures for delivery of therapeutic agents to promote SCI recovery include polymer nanoparticles, 61 surfacefunctionalized iron oxide nanomagnets, 62 carbon/metal nanotubes, 63 nanocomposites, 64 and micro or nanofabricated nerve devices. 65,66 The ability to precisely control dosing regimens and the implementation of specic targeting strategies may promote the recovery aer SCI and reduce toxic side effects. In order to solve the problem of the Se safety range, 27,28 this experiment used the porous Se@SiO 2 nano composites, 36 a new type of material that controls the release of Se. Some recent studies demonstrated that porous Se@SiO 2 nanocomposites could inhibit steroid-induced osteonecrosis of the femoral head by inhibiting oxidative stress. 42 The advantage of this material is that the nano-modied Se elements could not only signicantly reduce the toxicity risk of nano Se 37,38 but increase the biological activity of se. 39,40 The use of porous and silica nanocarriers to control the release of nano-Se could achieve a long-term effective concentration of Se and improve the treatment efficacy. 42 Compared with normal Se nanoparticles, Se in the Se@SiO 2 nanocomposites was limited by SiO2. Accompanying the entrance of PVP into an aqueous solution, trace Se can be released into the solution. 36 Our in vitro experiment showed that the porous silica nanocarrier was relatively stable, had good biocompatibility (Fig. 2), and the loaded Se elements could be continuously released to pH 7.4 0.01 M PBS (Fig. 1). In addition, the remarkable biocompatibility of porous Se@SiO 2 nanocomposites could reduce ROS damage effectively in vitro.
Se could inhibit apoptosis of the damaged segment induced by SCI by inhibiting the caspase-3 pathway in vivo (Fig. 8).
Experiments had shown that in situ injection of Se can inhibit ROS-mediated apoptotic neural precursor cell death in traumatic brain injury. 26 Our study showed that the inhibitory effect of intrathecal injection of porous Se@SiO 2 nanocomposites on apoptosis was especially pronounced 14 days aer SCI. On the one hand, Se may probably inhibit apoptosis in the spinal cord by inhibiting the intensity of oxidative stress and inammatory response associated with excessive activation of microglia. On the other hand, it may be because of the direct enhancement of neuronal cell resistance. The specic mechanism underlying the anti-apoptosis effect of Se needs further study.