JNK signaling pathways

Biomaterials 121 (2017) 64e82

Contents lists available at ScienceDirect

Biomaterials journal homepage: www.elsevier.com/locate/biomaterials

Blood-brain barrier dysfunction induced by silica NPs in vitro and in vivo: Involvement of oxidative stress and Rho-kinase/JNK signaling pathways Xin Liu, Baiyan Sui, Jiao Sun* Shanghai Biomaterials Research & Testing Center, Shanghai Key Laboratory of Stomatology, Ninth People's Hospital, Shanghai Jiaotong University School of Medicine, Shanghai 200023, China

a r t i c l e i n f o

a b s t r a c t

Article history: Received 4 January 2016 Received in revised form 3 January 2017 Accepted 3 January 2017 Available online 4 January 2017

Silica nanoparticles (SiO2-NPs) has been extensively exploited in biomedical fields and mostly designed to enter the circulatory system, however, few studies focused on the potential adverse effects of SiO2-NPs exposure on the blood-brain barrier (BBB) that serves as a critical barrier between the central nervous system (CNS) and the peripheral circulation. This study attempts to provide an understanding of whether and how SiO2-NPs disrupts the BBB in vitro and in vivo. Through a human BBB model, we found that SiO2NPs could induce tight junction loss and cytoskeleton arrangement, and increase inflammatory response and the release of vascular endothelial growth factor (VEGF) of brain microvessel endothelial cells (BMECs), which further activates astrocytes to amplify the generation of VEGF and increase the aquaporin-4 expression, and thus causing BBB disruption through a complex immunoregulatory loop between BMECs and astrocytes under SiO2-NPs exposure. Additionally, our data show that inhibition of reactive oxygen species (ROS) and Rho-kinase (ROCK) could effectively protect the SiO2-NPs-induced BBB dysfunction. In vivo studies further confirmed that SiO2-NPs could cause the BBB paracellular opening, oxidative stress and astrocyte activation in brains of SpragueeDawley (SD) rats. These findings demonstrate that SiO2-NPs could disturb BBB structure and function and induce BBB inflammation, and suggest that these effects may occur through ROS and ROCK-mediated pathways, which not only improve neurotoxicity evaluation for SiO2-NPs but also provide useful information in development of SiO2-NPs in neuro-therapeutics and nanodiagnostics. © 2017 Published by Elsevier Ltd.

Keywords: Silica nanoparticles Blood-brain barrier Tight junction Inflammation Oxidative stress Rho-kinase

1. Introduction Silica nanoparticles (SiO2-NPs), due to its unique characteristics such as large surface area, high structural stability, easy surface functionalization, low cost of production, excellent biocompatibility and protracted circulation properties, has garnered increasing attention in the biomedical field including drug delivery, imaging, cell tracking and photothermal therapy [1]. For example, a cancerselective SiO2-NPs was recently approved by the US Food and Drug Administration (FDA) for a first-in-human clinical trial [2]. However, these unique features such as small size and high surface activity enabled NPs to negotiate various biological barriers and have easier access to the organ and tissue in the body [3]. Such

* Corresponding author. No. 427, Ju-men Road, Shanghai 200023, PR China. E-mail address: [email protected] (J. Sun). http://dx.doi.org/10.1016/j.biomaterials.2017.01.006 0142-9612/© 2017 Published by Elsevier Ltd.

barrier penetrability of SiO2-NPs has been shown to occur across the placental barrier and enter fetal liver and brain [4]. The brain tissue, which has very limited regenerative capacity, is generally protected from exogenous insults by the blood-brain barrier (BBB). Once engineered NPs enter into the circulatory system as drug/gene carrier and disrupt the BBB function, it may unintentionally reach the brain and give rise to neurotoxic effects. Therefore, a fundamental understanding of the biological impact of SiO2-NPs on BBB is of great importance for nanosafety evaluation. The BBB not only serves as a natural barrier between the central nervous system (CNS) and the peripheral circulation, but also plays a critical role in maintaining neuronal microenvironment and brain homoeostasis. The protective properties of BBB are mainly conferred by the intricate multi-cellular vascular structure of brain microvessel endothelial cells (BMECs) coupled with astrocytes and pericytes in the brain. BMECs, the core element of the BBB, have continuous intercellular tight junctions, minimal pinocytosis

X. Liu et al. / Biomaterials 121 (2017) 64e82

activity and multiple specific transport systems, which greatly restrict the transport of cells and molecules in and out of the brain. Many studies have reported that some metallic NPs (e.g. titanium dioxide, aluminum oxide, gold and silver NPs) could cause significant pro-inflammatory response in BMECs [5e8] and increase endothelial paracellular permeability by altering endothelial tight junctions [9e11]. Relatively recent data have shown that SiO2-NPs could pass through the in vitro BBB models and the permeability was mainly dependent on particle's size [12e14], and transcellular trafficking was thought to be a mechanism for the SiO2-NPs crossing of BBB [14]. However, in vivo reports on this topic provide considerable controversial results. Several studies demonstrated that SiO2-NPs administered by cerebral perfusion or dermal administration was able to pass through the BBB and reach the brain tissue [15,16], while in another study, it has been reported that dermal and oral administration of SiO2-NPs did not enter into rats' brain by compromising the BBB [17]. Moreover, only a few studies focused on the toxicology effects of SiO2-NPs on BBB and revealed that combination of SiO2-NPs with several adverse factors (e.g., hypertension with stress, environmental toxicants) could aggravate brain pathology and induce cerebrovascular toxicity by enhancing pro-inflammatory responses and disrupting the BBB integrity [18,19]. As a result, currently significant efforts have been directed towards investigating the penetrated ability of SiO2-NPs in the BBB for transport of therapeutic or diagnostic agents, few studies focused on the possible adverse effects of SiO2-NPs on BBB, particularly the detailed mechanisms of how SiO2-NPs influences BBB has been still unclear. Oxidative stress has been proven to play a pivotal role in BBB damage after a variety of insults such as hypoxia, drug, trauma or neurodegenerative disorders [20,21]. Likewise, inflammation is closely associated with BBB dysfunction. Several pro-inflammatory mediators (reactive oxygen species (ROS), cell adhesion molecules (CAMs), cytokines, chemokines) were reported to not only regulate the magnitude of leukocyte extravasation into brain parenchyma, but also directly cause tight junctions proteins loosening, edema formation, and leakiness of BBB in vitro and in vivo [20,22,23]. Moreover, many previous literature including ours suggested that SiO2-NPs mediated-nanotoxicity including pulmonary toxicity, genotoxicity, cardiovascular toxicity, hepatic and brain injury was related to oxidative stress and pro-inflammatory gene activation [24e28]. To date, although oxidative stress and inflammation have been extensively studied in both BBB damage and nanotoxicity's mechanism, insufficient data is available about their roles in NPs induced increase of BBB permeability. In our previous work, we have found that SiO2-NPs could induce pro-inflammatory response and apoptosis of human umbilical vein endothelial cells (HUVECs) through oxidative stress via Mitogenactivated protein kinases (MAPKs) and NF-kB pathways [29,30], and other previous studies have revealed that both RhoA/Rhokinase (ROCK) and MAPKs pathways were involved in the regulation of BBB integrity [23,31]. Depending on the context, we hypothesized that SiO2-NPs might induce BBB inflammation and alter the BBB structure and function by oxidative stress, MAPK and ROCK signaling pathways. To test this hypothesis, we firstly developed an in vitro BBB model consisting of both primary human BMECs (HBMECs) and astrocytes that closely mimics in vivo conditions. Due to the fact that intrinsic characteristics of NPs are related to their toxicity, comparative studies of nanoscale and microscale materials are essential. Thus, utilizing this model, we focused on understanding the BBB permeability, BBB structure and function changes, and evaluation of BBB inflammation after exposure to SiO2-NPs and silica microparticles (SiO2-MPs) by a battery of systematic investigations including cell viability, cellular uptake, transendothelial electrical resistance (TEER), cytoskeleton

65

arrangement, tight junctions protein expression, CAMs and aquaporin 4 (AQP4) expression. Cytokines secretion [vascular endothelial growth factor (VEGF), interleukin-6(IL-6), IL-8, IL-1b], ROS generation and Damage-associated molecular patterns (DAMPs) as well as the MAPK and ROCK signaling pathways were also investigated to elucidate the detailed mechanism of the BBB dysfunction regulated by SiO2-NPs. To confirm the in vitro results, the BBB permeability, oxidative stress and astrocyte activation in brain of SpragueeDawley (SD) rats were further evaluated by Evans blue, inductively Coupled Plasma Techniques (ICP), immunohistochemistry (IHC) and immunofluorescence (IHF) analysis. Additionally, hematological parameters were also monitored to indicate potential system toxicities. 2. Materials and methods 2.1. Preparation and characterization of silica particles The SiO2-NPs were synthesized by a sol-gel method according to our previous published procedure [32]. The SiO2-MPs [Cat. No: S5631] were purchased from Sigma-Aldrich (Sigma, St. Louis, MO, USA). The size and shape of these particles were examined under TEM (JEOL-2010, JEOL Ltd., Tokyo, Japan) and SEM (JSM-6700F, JEOL Ltd., Tokyo, Japan). Size distribution and average particle diameters were obtained from TEM images by measurement at least 150 particles using ImageJ (NIH free software). The surface area of these samples was analyzed by the BrunauereEmmetteTeller (BET) method using a Surface Area Analyzer (ASAP2020, Micromeritics). Hydrodynamic diameter and zeta-potential of these particles in endothelial cell complete medium (ECM) (Sciencell, Carlsbad, CA, USA) were measured with a Malvern Zetasizer Nano ZS and Mastersizer Micro instrument (Malvern Instruments, Worcestershire, UK). Prior to inoculation into the in vitro and in vivo systems, both silica particles were sterilized by ethylene oxide according to previous published literature [33], and the endotoxin content of the samples was negative at the level of 0.25 EU/mL. 2.2. In vitro studies 2.2.1. Cell culture and in vitro blood-brain-barrier model Primary HBMECs (Cat. No: cAP-0002) isolated from normal human brain tissue were purchased from Angio-proteomie (Boston, MA, USA) and grown in ECM. Primary human astrocytes (Cat. No: 1800) isolated from human cerebral cortex were purchased from Sciencell Research Laboratories (Carlsbad, CA, USA) and grown in astrocyte complete medium (Sciencell, Carlsbad, CA, USA). The BBB in vitro model was constructed according to the modified procedure from previous studies [34]. In brief, HBMEC with a density of 100,000 cells/cm2 at third passage were cultured on rat tail collagen- (Sigma, St. Louis, MO, USA) coated Transwell inserts (1.0 mm pore size, Millipore, Darmstadt, Germany) in the 12-well culture plates for 2 h, and then the astrocytes were transferred at third passage to the bottom of the 12-well plates at a concentration of 50,000 cells/cm2, and the microplates were then incubated at 37
C in a humidified 5% CO2 atmosphere for 12e14 days to form the in vitro model. On day 12e14, the TEER value of the HMBECs was measured, and according to previously published papers [35], the following permeability experiments were performed when TEER values were >200 U cm2. To examine the effects of ROS inhibition and ROCK/JNK blockade on the BBB model after exposure to SiO2-NPs, HBMECs in the coculture or monoculture were respectively pretreated with 5 mM N-acetyl cysteine (NAC), or 20 mM SP600125 (JNK inhibitor) or 20 mM Y-27632 (ROCK inhibitor) (Sigma, St. Louis, MO, USA) for 1 h, and then 100 mg/mL SiO2-NPs was added and co-incubated for 24 h,

66

X. Liu et al. / Biomaterials 121 (2017) 64e82

subsequently the cells or supernatants were collected for a series of analysis according to the experiment schedule.

2.2.2. Cell viability assays Cell viability was measured by assessment of mitochondrial function using the CellTiter 96 ® AQueous One Solution Assay (Promega, Madison, WI, USA). In cocultures, the HBMECs on the upper side of a coated polyester transwell membrane was treated with differing concentrations of SiO2-NPs and SiO2-MPs in the 12well plates, likewise, the HBMECs monoculture were treated with the same concentrations of silica particles in the 96-well plates. After 24 h of exposure, the HBMECs and astrocytes from the coculture system and the HBMECs monoculture were respectively incubated with cell medium supplemented with 20% v/v of MTS (3(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4sulfophenyl)-2H-tetrazolium). After 4 h of incubation, the supernatants were transferred to new 96-well plates to ensure that light transmission was not disturbed by particles and absorbance at 490 nm were determined on a Microplate Reader (Multiskan GO, Thermo Scientific, MA, USA). Moreover, IC50 value was further calculated to quantify the concentration required inhibiting 50% of growth of HBMECs by GraphPad prism (version5) software.

2.2.3. Transendothelial electrical resistance measurement The BBB permeability was assayed by measuring the TEER across the cell monolayer. Generally, a decrease in TEER reflects an increase in permeability and a loss of barrier function. In this study, TEER of the BBB model was recorded in PBS at 0, 3, 6, 9, 12 and 24 h after 100 mg/mL of SiO2-NPs or SiO2-MPs exposure using the Millicell ERS-2 device (Merck Millipore, Billerica, MA, USA). The resistance value was multiplied by the surface area of the insert (1.12 cm2) and expressed as U$cm2. The TEER of each sample was corrected for background resistance without cells. The results were normalized to TEER measured before silica particles exposure and presented as absolute values.

2.2.4. Horseradish peroxidase (HRP) transmissivity measurement For the HRP flux measurement, 460 mL minimum essential medium (MEM) without phenol red (GIBCO, Scotland, UK) containing 500 ng HRP (MW 40 kD, Sigma-Aldrich) was added to the upper chamber in place of cell supernatants 60 min before the end of the experiments, and meanwhile the cell medium in the bottom well was replaced with 1140 mL MEM without phenol red. After 1 h, a total of 50 mL medium was collected from each bottom well and mixed with 100 mL peroxidase substrate for 5 min. The reaction was terminated by adding 50 mL sulphuric acid (1 mol/L). The optical density was measured at 450 nm and the HRP transmissivity was assayed from the standard curve according to the following equation: HRP% ¼ [(CHRP  V/CHRPi  Vi)  100%], where CHRP is the HRP concentration in the bottom well, CHRPi is the HRP concentration in the insert, V is the medium volume in the bottom well, and Vi is the medium volume in the insert.

2.3.1. Transmission electron microscopy and energy-dispersive Xray spectroscope Transmission electron microscopy (TEM) was used to visualize particles cellular uptake and tight junction changes in HBMECs exposed to silica particles. In brief, HBMECs were treated by two silica particles (100 mg/mL) for 24 h, and then fixed in 2.5% glutaraldehyde followed by 1.5% osumium tetraoxide. The fixed cells were dehydrated and embedded in EPON resin. Ultrathin sections were stained with lead citrate and observed under a transmission electron microscope (Philips CM120, Eindhoven, the Netherlands). Meanwhile, the elements distribution of particles in the sections was also characterized by energy-dispersive X-ray spectroscope (EDS)(OXFORD IET-250). 2.3.2. Filamentous (F)-actin staining HBMECs were cultured in the presence of two silica particles (100 mg/mL) for 24 h, and then stained with Rhodamine-phalloidin to examine the structure of F-actin. Briefly, washed cells were fixed with 4% polyoxymethylene, washed again, and permeabilized for 5 min with 0.1% Triton X-100. The cells were incubated with a 1% solution of bovine serum albumin (BSA) for 30 min at room temperature (RT), and stained with Rhodamine-phalloidin at 37
C for 1 h in dark conditions (0.20 mol/L, Sigma), and the nuclei were stained with 40 ,6-diamino-2-phenylindole dihydrochloride (DAPI, Sigma) at RT for another 3 min. Stained F-actin was visualized using a confocal laser scanning microscope (Nikon A1R, Japan). 2.3.3. Cytokines measurement For the assay of cytokines (VEGF, IL-1b, IL-6, IL-8) and damageassociated molecular patterns (DAMPs) (high mobility group box-1 (HMGB1), S100A8/S100A9 Heterodimer), the supernatants of BBB cocultures or HBMECs monocultures exposed to particles (100 mg/ mL) were respectively collected after 24 h, and then quantified by an enzyme linked immunosorbent assay (ELISA) kit (VEGF, abcam, Cambridge, UK; IL-1b, IL-6, eBioscienc, SanDiego, USA; IL-8 and S100A8/S100A9 Heterodimer, R&D Systems, Oxford, UK; HMGB1, Chondrex, Redmond, USA) according to the manufacturer's instructions.

2.3. Determination of transport of silica particles across the BBB in vitro

2.3.4. Flow cytometric analysis For the measurement of cell adhesion molecules, after exposure to 100 mg/mL of silica particles for 24 h, cells were collected and stained with the specific primary mAb for 30 min at 4
C in the darkness, washed once with PBS, and analyzed. The following mAbs were employed: ICAM-1 (CD54-PE), VCAM-1 (CD106-FITC), Eselectin (CD62E-APC) (MiltenyiBiotec, Bergisch Gladbach, Germany). The proportion of cells expressing ICAM-1, VCAM-1 and Eselectin in the different experimental conditions was measured. For the measurement of intracellular ROS, HBMECs in 6-well plates were washed twice with PBS and incubated in 2 mL working solution of 20 , 70 -dichlorofluorescein diacetate (DCFH-DA) (Applygen, Beijing, China) at 37
C in dark for 30 min. Then the cells were washed twice with cold PBS and resuspended in the PBS and the mean fluorescence intensity (MFI) of 104 cells was quantified for analysis of ROS. All flow cytometric analyses were performed by Guava easyCyte flow cytometer (Merck-Millipore, Billerica, MA, USA) and data were collected and analyzed by InCyte software (Merck-Millipore).

Briefly, after exposure to 100 mg/mL of silica particles for 24 h, the cells-free supernatants in the bottom well of the transwell cocultures were collected and digested, and then the concentration of Si was measured by Inductively Coupled Plasma-Optical Emission Spectrometry (ICP-OES, VISTAPRO, Agilent, USA) as according to previous published papers [29].

2.3.5. Western blot analysis HBMECs were treated by two silica particles (100 mg/mL) for 24 h, and then washed once with ice-cold PBS, and lysed in ice-cold lysis buffer [(50 mM Tris-HCl, 150 mM NaCl, 1% NP-40, 0.1%sodium dodecyl sulfate (SDS), Applygen, Beijing, China] containing 1 mM phenylmethylsulphonyl fluoride (PMSF) (Sigma, St. Louis, USA) and

X. Liu et al. / Biomaterials 121 (2017) 64e82

phosphatase inhibitor cocktail (Sigma, St. Louis, USA) for 30-min. After centrifuging the lysates at 12,000 rpm, 4
C for 10 min, the supernatants were collected and stored at 80
C until used. The protein concentrations of these extracts were determined by performing the bicinchoninic acid (BCA) protein assay (Pierce, Rockford, USA). Equal amounts of lysate proteins (40 mg) were then loaded onto SDS-polyacrylamide gels (8%e12% separation gels) and electrophoretically transferred to nitrocellulose (NC) membranes (Amersham Biosciences, US). After blocking with 5% nonfat milk in Tris-buffered saline (TBS) containing 0.05% Tween-20 (TBST) for 1h at room temperature, the membrane was respectively incubated with anti-Occludin and ZO-1 (1:500, abcam, Cambridge, UK), antip-ERK, ERK and AQP4 (1:1,000, Bioworld Technology, USA), anti-pp38, p-p-JNK, JNK and p-38 (1:1,000, CST, USA), b-actin (1:5,000, abcam, Cambridge, UK) at 4
C overnight, washed with TBST, and incubated with a horseradish peroxidase-conjugated anti-rabbit IgG/anti-mouse IgG secondary antibody at 37
C for 1 h. The antibody-bound proteins were detected using the ECL chemiluminescence reagent (Millipore, USA). 2.4. In vivo studies 2.4.1. Animals care and experiment design Male Sprague Dawley rats (SD rats) weighing of 180e200 g were purchased from SLACCAS Laboratory Animal Co, Ltd. (Shanghai, China). The animals were housed in stainless steel cages in an airconditioned controlled room (23 ± 2
C, 50e70% relative humidity, with a 12-h light/dark cycle). Rodent diet and water were available ad libitum. After a week acclimation, rats were randomly divided into experimental and control groups with each group having 15 rats. Two experiment groups had respectively received intraperitoneal injections of SiO2-NPs or SiO2-MPs daily at the doses of 50 mg/kg of body weight. The control group (saline group) was only given 0.9% saline intraperitoneally. On day 28, all animals were sacrificed by overdose of pentobarbital, and the blood from each animal were harvested and prepared for the following experiment analysis. All the animal studies were approved according to the Ethics Committee of Animal Care and Experimentation of the National Institute for Environmental Studies, China. 2.4.2. Evaluation of bloodebrain barrier (BBB) permeability To evaluate the potential effects of silica particles on BBB permeability, Evans blue (EB) dye was used as a marker of albumin extravasations according to a previous study [17]. In brief, the EB dye (2% in saline, 8 mL/kg) was injected via the tail vein and was allowed to remain in circulation for 1 h on day 28 after repeated intraperitoneal administrations. Subsequently, the thorax was opened under anesthesia. The rats were perfused with saline through the coronary artery until the fluid from the right atrium became colorless, and paraformaldehyde (4%) was used to perfuse the brain. After decapitation, the brain was removed and dissected, and the weight of each of these brain sections was measured. Each brain section was incubated with formamide at 55
C for 24 h. The supernatant was separated after centrifugation at 12,000 g for 20 min and the absorbance of EB was measured at 635 nm by a Spectrophotometer Reader (Multiskan, Thermo Scientific, MA, USA).

67

weightDry weight/Wet weight  100%. 2.4.4. Oxidative stress and silicon content assay SD rats were sacrificed on 28 days after repeated intraperitoneal injection of either saline or SiO2-NPs or SiO2-MPs, and the brain tissues were extracted, divided into left brain and right brain. For the oxidative stress assay, the five left brain tissues per group were weighed and lysed in ice-cold lysis buffer, the mixtures were homogenized by an ultrasonic cell disruptor (Sonics vibra cell, VCX105) five times, each for 5 s, at 4
C. The solution was then centrifuged at 14000 g at 4
C for 15 min. The supernatants were collected and the oxidative biomarkers malondialdehyde (MDA) was analyzed by the commercial colorimetric assay kit (Beyotime Institute of Biotechnology, Jiangsu, China). For the Si content assay, the five right brain tissues per group were weighed and digested, and then the concentration of Si was measured by ICP-OES. 2.4.5. Blood collection and analysis 28 days after exposure, the SD rats were anesthetized, and blood samples were collected from femoral artery into centrifuge tubes with 3.8% sodium citrate pre-added under anesthesia. Hematological parameters were determined using a hematological autoanalyzer (Coulter T540 hematology system; Beckman Coulter, Inc). 2.4.6. Histopathological analysis and immunohistochemistry/ immunofluorescence (IHF) Following the exposure regimen described above, SD rats were anesthetized with pentobarbital and perfused through the heart with saline followed by 4% paraformaldehyde. The brains were removed, postfixed, and embedded in paraffin according to standard histological techniques. Five-micrometer thick coronal sections were cut and then mounted onto glass slides. The sections were stained with hematoxilin and eosin (H&E) and subsequently processed for histopathological examination. For the phosphorylation of myosin light chain (p-MLC) measurement, the brain sections were immunohistochemically stained using antibodies for the p-MLC (abcam, UK), and universal-labeled streptavidin biotin kit (Dako, Denmark) was used as the detection system. After staining, the sections were evaluated and photos were taken using an optical microscope (Leica, Tokyo, Japan). For the glial fibrillary acidic protein (GFAP) and AQP4 detection, the brain section were processed for immunofluorescence using primary antibodies for the GFAP and AQP4 (abcam, UK), and Cy3-conjugated goat anti-rabbit secondary antibodies with nuclear stain DAPI (boster biological technology, Wuhan, China) were used as the detection system. Sections were examined and photographed using fluorescence microscopy (Olympus BX53, Japan). For further quantify the immunofluorescence stained areas. Sections stained for GFAP and AQP4 were photographed using constant exposure settings. Single-channel immunofluorescence images were converted to black and white and thresholded and the amount of stained area measured in different tissue compartments using NIH ImageJ software. Total GFAP and AQP4-positive area was expressed as a percentage of total area in each field according to previous published paper [37]. 2.5. Data analysis

2.4.3. Brain water content assay Brains were separated on day 28 after repeated intraperitoneal administrations and divided into two hemispheres, and cerebella were abandoned. The brain samples were weighed to acquire the wet weights. The samples were dried in an electric oven at 100
C for 24 h for dry weights. Water contents are represented as percentages of wet weights in the following formula [36]: Wet

The data are expressed as means ± SD or means ± SEM. Statistical analyses were performed with SPSS 20.0 software, and statistical comparisons were analyzed using the t-test and one way ANOVA followed by Tukey's HSD post hoc test. Differences were considered statistically significant when the p-value was less than 0.05.

68

X. Liu et al. / Biomaterials 121 (2017) 64e82

3. Results and discussion 3.1. Particle characterization Physicochemical characteristics of NPs including size, shape, chemical composition, physiochemical stability and surface area, generally influence their biological behavior [38]. As shown in Fig. 1, TEM indicated that the amorphous SiO2-NPs were well-dispersed, approximately spherical in shape and had an average size of 20 ± 6.34 nm (Fig. 1A), while SEM showed that SiO2-MPs were severely aggregated with an average particle size of 1.7 ± 0.62 mm (Fig. 1B). To obtain a more accurate particles size under natural flow state in vivo, hydrodynamic diameters of the two particles were further verified by dynamic light scattering (DLS). DLS analysis of SiO2-NPs revealed a single peak at ~190 nm within the size range from 100 up to 300 nm, and the single peak in size distribution suggested that SiO2-NPs were monodispersed (Fig. 1C). In contrast, SiO2-MPs had a wider particle size distribution ranging from 2.5 up to 95 mm with the largest peak at~20 mm (Fig. 1D). Additionally, both SiO2-NPs and SiO2-MPs had similar Zeta potentials within the range of 10 to 25 mV in the ECM which implies that the two silica dispersion is similarly electrically stabilized. The DLS data indicated that SiO2-NPs had smaller size, better dispersion and lower agglomeration than SiO2-MPs in ECM. As the particle size becomes small, the specific surface area is increased. The nitrogen

adsorption analysis confirmed that the specific surface area was (147.8 ± 1.66) m2/g for SiO2-NPs and (5.01 ± 0.1) m2/g for SiO2-MPs (Fig. 1E) and showed a good correlation between the results of size and surface area of the two particles. Many studies have found that smaller NPs bind significantly higher numbers of microarray proteins at 10% (v/v) plasma than larger NPs with the same surface chemistry [39e41]. Therefore, the increase in surface area determines the potential number of reactive groups on the surface, suggesting the SiO2-NPs have more surface activity and enhanced adsorption properties, which might cause unusual biological effects compared to the conventional and commercial counterparts. 3.2. Establishment of BBB in vitro model To investigate the biological impact of SiO2-NPs on the BBB and the possible mechanisms involved in these processes, an in vitro BBB model should be first established. The principal components of the BBB are the endothelial cells, astrocytes and pericytes. Several in vitro animal BBB models using various species of primary BMECs and endothelial cell line (e.g. porcine, bovine, rat, or mouse cells), monoculture or co-culture with pericytes/astrocytes, have been recently developed for NPs BBB permeability assessment [8,12,13,34,42,43]. However, few studies attempt to co-culture human primary BMECs and human primary astrocytes as BBB models, which make it difficult to draw firm conclusions concerning the

Fig. 1. Characterization of two silica particles. A): TEM image of SiO2-NPs; B): SEM image of SiO2-MPs; C): Hydrodynamic size analysis of SiO2-NPs; D) Hydrodynamic size analysis of SiO2-MPs; E): Particle size, hydrodynamic diameter, surface area and zeta-potential. Inset: histogram showing size distribution; About 150 NPs were considered for sample to obtain the size distribution histogram.

X. Liu et al. / Biomaterials 121 (2017) 64e82

crossing of the human BBB by NPs. In the present study, to mimic as closely as possible the normal BBB micro-environment in vivo and demonstrate the relevance of our findings in humans, we established an in vitro human BBB co-culture model. As shown in Fig. 2A, the BBB model contained two human primary cell types: HBMECs grown on a microporous filter membrane, and astrocytes in the bottom well. The BBB models were generally characterized by measurements of transendothelial electrical resistance (TEER). After 14 days of coculture of HBMECs and human astrocytes, we observed that HBMECs formed a dense cell layer. Subsequently, TEER measurement was carried out to monitor the tightness of the HBMEC monolayer. The TEER value in the HBMECs monoculture was 152 ± 6 U cm2, while co-culture with astrocytes caused a significant increase of TEER to the range of 219 ± 14 U cm2, consistent with the data showing TEER values of human BBB model which were in the range of 150e300 U cm2 [44e46]. Compared to in vitro animal BBB model or in vivo conditions [43], the TEER value of in vitro human BBB model was considerably lower. However, the use of primary human origin brain endothelial cells could help to avoid interspecies genetic variation. Therefore, our HBMECs/ astrocyte co-culture model was suitable for investigating the mechanistic insights into BBB permeability of NPs. 3.3. Cytotoxic effects of the silica particles on the BBB coculture and single cell culture To clarify if silica NPs or MPs exhibit cytotoxic effects on HBMECs, cells were exposed to different concentrations (0, 25, 50, 100, 200, 400, 800 mg/mL) of particles for 24 h and cell viability was measured using the colourimetric MTS assay, a well-established photometric method to determine the cell mitochondrial function. In HBMECs, SiO2-NPs showed higher cytotoxic effect than

69

SiO2-MPs across all the concentration in a dose-dependent manner. In detail, at the low concentration of 50 mg/mL SiO2-NPs, a slight decrease in cell viability of HBMECs was observed which reduced the percentage of viable cells from 100% to 87%, followed by a moderate decrease to 51.4% at the concentration of 400 mg/mL, and a severe decrease to less than 35% at concentration of 800 mg/mL (Fig. 2B). However, no more than 50% decrease in the proportion of viable cells was observed even at the highest concentrations of SiO2-MPs (Fig. 2B). The IC50 value was 436 mg/mL (~136 mg/cm2 of cell culture) for the SiO2-NPs on HBMECs, whereas was more than 800 mg/mL for the SiO2-MPs. Our results were in agreement with those reported in the literature, showing a dose- and sizedependent reduction in cell viability after exposure to amorphous SiO2-NPs in human endothelial cells (EAHY926 cell line) and IC50 value was respectively 89 and 254 mg/cm2 of cell culture for 19-nm and 60-nm SiO2-NPs [47]. In addition, cell viability was further analyzed using a Resazurin fluorescence assay, which indicates the cellular metabolic activity. A dose dependent cytotoxic effect was also observed after exposure to SiO2-NPs or SiO2-MPs at concentrations ranging from 25 to 800 mg/mL. Moreover, the percentage of Alamar blue reduction in SiO2-NPs group was significantly lower than that in SiO2-MPs group at the higher dose of 400 and 800 mg/ mL (Fig. S1). The fluorimetric cytotoxic data were consistent with the MTS result, indicating that the interferences between the particles and dye might be rather low in our studies. In order to determine whether HBMECs and astrocytes were affected in the coculture system, the MTS assay was also used, which demonstrated similar cytotoxic patterns of dose-dependent responses in HBMECs mono- and co-cultures exposed to silica particles, and cocultivation of HBMECs with astrocytes had no negative impact on cell viability in both HBMECs and astrocytes (Fig. 2CeD). In most cases, a sub-cytotoxic dose might reveal a beneficial balance

Fig. 2. Cytotoxic effect of the silica particles on the BBB coculture and single cell cultures. A): Schematic view of the co-culture architecture. Human brain microvessel endothelial cells (HBMECs) were grown on a semi permeable membrane whereas astrocytes lay on the well bottom; Changes in cell viability in B) the HBMECs monocultures after 24 h of exposure to different concentrations of SiO2-NPs or SiO2-MPs as well as C) HBMECs or D) astrocytes from the BBB coculture. Results are presented in mean ± SEM of three independent experiments *p < 0.05; **p < 0.01 vs. control. #p < 0.05 significant difference as compared groups.

70

X. Liu et al. / Biomaterials 121 (2017) 64e82

between efficiency and toxicity and allow us to assess potential adverse effects at molecular levels other than only due to cell death. According to the standard of ISO 10993-5:2009 (Biological evaluation of medical devices- Part 5: Tests for in vitro cytotoxicity), the relative growth rate (RGR) more than 75% represents Grade 0e1, which are considered non-cytotoxic. In order to carry out the subsequent study on BBB function and inflammatory response, a low dose (100 mg/mL) which results in no more than 20% decrease in cell viability was chosen in the following experiments according to our previous studies [30]. 3.4. Structural and functional BBB disruption caused by SiO2-NPs in vitro 3.4.1. Increased BBB permeability by SiO2-NPs We further characterized the BBB permeability as an initial step to evaluate the BBB structural and functional changes after exposure to SiO2-NPs. The TEER of BMECs monolayer or co-culture was usually an indicator of BBB integrity. As shown in Fig. 3A, SiO2-NPs

led to a time-dependent decrease in TEER of BBB co-culture model from 3 h to 24 h, compared to the untreated control (p < 0.05). The decrease in TEER was even more dramatic at 12 h (approximately 75% of the initial TEER-value at 0 h). When treated with SiO2-NPs for 24 h, the TEER were a little higher than that in the 12 h group, but there was no statistical difference. In contrast, SiO2-MPs did not elicit significant drop in TEER. Under physiological conditions the BBB is impermeable to macromolecules in the blood stream [21]. To further determine whether the decrease in TEER caused by SiO2NPs is also reflected by an enhanced permeability of BBB co-culture model to large molecules, we measured HRP transmissivity as a permeability assay at 24 h after SiO2-NPs exposure. Consistent with the TEER values, the HRP permeability significantly increased only after SiO2-NPs exposure, compared to the untreated control and SiO2-MPs treatment (Fig. 3B). These results indicated that the BBB structural and functional impairment induced by SiO2-NPs was probably dependent on its small size and high surface area. The increased permeability may help neuro-therapeutic drugs cross the BBB, but meanwhile, it also may allow deleterious molecules or NPs

Fig. 3. Changes in structure and function in the BBB coculture after 24 h of exposure to silica particles. A) Time-dependent transendothelial electrical resistance (TEER) reduction by SiO2-NPs; B) HRP permeability increased at 24 h after SiO2-NPs exposure (100 mg/mL); C) Si concentrations of the supernatants from the bottom well in the transwell system; D) Western blot analysis of tight junctions (Occludin, ZO-1) of HBMECs stimulated by SiO2-NPs or SiO2-MPs; b-actin was used as an internal control to monitor for equal loading; E) TEM micrographs of HBMECs exposed for 24 h to SiO2-NPs or SiO2-MPs. (a,a1,a2): HBMECs without any treatment; (b,b1,b2): HBMECs with SiO2-NPs treatment; (c,c1,c2): HBMECs with SiO2-MPs treatment. (a,b,c): Overall cell morphology (scale bar: 2 mm for aec, a1ec1); (a2ec2): Higher magnification of part area in cells (scale bar: 500 nm); (N: the nucleus, mi: the mitochondria). Blue arrows denote NPs uptake by macropinocytotic vesicles; Red arrows denote tight junctions; F) Confocal laser scanning microscope image for HBMECs exposed for 24 h to SiO2-NPs or SiO2-MPs. G) Energy-dispersive X-ray spectroscope analysis of nano-size dots within vesicles. The red box indicates the nano-size dark dots within vesicles, the black box indicates the cytoplasm area. The red ellipse indicates a silicon (Si) peak. a) HBMECs with SiO2-NPs treatment; b) graph of energy-dispersive X-ray spectroscope analysis of the red box area in panel a; c) graph of energy-dispersive X-ray spectroscope analysis of the black box area in panel a; d) HBMECs with SiO2-MPs treatment; e) graph of energy-dispersive X-ray spectroscope analysis of the red box area in panel d; f) graph of energy-dispersive X-ray spectroscope analysis of the black box area in panel d; Data represent means ± SD; n ¼ 3.*p < 0.05, **p < 0.01 vs. control. #p < 0.05 ##p < 0.01 significant difference as compared groups. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

X. Liu et al. / Biomaterials 121 (2017) 64e82

71

into the brain and cause the brain parenchyma injury. In order to further clarify whether SiO2-NPs can traverse the in vitro BBB model, we performed the transport study of SiO2-NPs across the BBB by detecting the silicon content of the cell medium from basolateral sides of the BBB model. The ICP and the X-ray Fluorescence (XRF) measurement confirmed Si concentration of the cell supernatants from basolateral sides in SiO2-NPs-stimulated cocultures was significantly higher than the untreated control and SiO2-MPs treatment (Fig. 3C, Table S1), suggesting that SiO2-NPs had a greater ability to cross the BBB than SiO2-MPs. A recent study also demonstrated that the permeability of SiO2-NPs (30 nm) in the BBB model of BMECs and pericytes was significantly higher than those of the larger silica particles (100 nm, 400 nm) [12]. In the case of intact BBB barrier, it was at most a rare event for the translocation of NPs across the BBB and most NPs remained in the barrier and accumulated in the lysosomes [48]. Taken together the data from BBB permeability and ICP measurement, it can be inferred that the translocation of SiO2-NPs across the BBB was mainly dependent on the paracellular permeability, which due to BBB disruption after exposure to SiO2-NPs.

Moreover, we also observed that SiO2-MPs resulted in a few stress fiber and gaps formation. Our results were consistent with other studies in the literature demonstrating that SiO2-NPs could disrupt the cytoskeleton organization in endothelial cells [53]. This suggests that exposure to SiO2-NPs induces contractile changes in cytoskeletal elements in BMECs to increase barrier permeability. It has been viewed that endocytosis of inorganic particles has even been correlated with disruption of cytoskeletal networks [54]. Our TEM imaging further revealed that both SiO2-NPs and SiO2-MPs could be taken up by HBMECs and sequestered in macropinocytotic vesicles in cytoplasm (Fig. 3E). Nano-size dots were analyzed by EDS to verify whether it is the aggregated dye materials. Our result showed that a silicon (Si) peak appeared when we analyze the particle-like dots into the vesicles in the cells treated with SiO2-NPs or SiO2-MPs, while no Si peak was observed when we analyzed the other area into the cells, indicating that the nano-size dots into the vesicles were SiO2-NPs or SiO2-MPs (Fig. 3G).

3.4.2. Down-regulation of tight junctions by SiO2-NPs In the next step, we sought to identify the cellular mechanism for TEER decrease and increase of BBB permeability induced by SiO2-NPs. Tight junctions, as hallmarks for BBB integrity, constitute the first and the most important seal to limiting paracellular permeability, also participating in the regulation of BBB stability [49]. It is generally accepted that the tight junctions between BBB endothelial cells lead to high TEER and low paracellular permeability [50]. In order to clarify the mechanism for the decrease of BBB permeability after SiO2-NPs exposure, we next detected the ultrastructure of tight junctions by TEM. As indicated in Fig. 3E, the tight junctions was observable as an electron-dense zone between cellecell contact in untreated or SiO2-MPs treated HBMECs, while the structure of tight junctions became widened after SiO2-NPs administration for 24 h, indicating that the integrity of the HBMECs monolayer could be disrupted by SiO2-NPs. Structurally, tight junctions multiprotein complexes are constituted by several transmembrane proteins (e.g. Occludin, Claudin, junctional adhesion molecules) linked to the actin cytoskeleton via interaction with cytoplasmic accessory proteins (e.g. zonula occludens protein 1 (ZO-1)) [51]. To further confirm the tight junction changes at the molecular level, Occludin and ZO-1 as the marker proteins of tight junctions were further analyzed by Western blot assays. Our results showed that the expression levels of occludin and ZO-1 were markedly reduced when HBMECs were stimulated with SiO2-NPs for 24 h, while SiO2-MPs had no impact on the tight junctions proteins expression (Fig. 3D). The down-regulation of ZO-1 and occludin can make the tight junctions opened and the BBB permeability increased by the paracellular pathway [52]. Recent studies also reported that gold NPs exposure could reduce the expression of ZO-1, occludin and claudin-5 in HUVECs in vitro and enhanced BBB permeability in vivo [9].

3.5.1. Increased release of VEGF and IL-6 in HBMECs mono- and coculture by SiO2-NPs Based on our observations that SiO2-NPs could increase the BBB permeability and induce tight junction loss and cytoskeleton contraction, we next asked which factors play key roles in the paracellular BBB opening during exposure to SiO2-NPs. Generally, under the pathological conditions in vivo, a variety of factors are involved in increasing BBB permeability including the release of VEGF, inflammatory mediators cytokines and chemokines, and free radical [50]. As many of these factors can also be released by astrocytes, we should consider the role of astrocytes on increasing permeability of the brain endothelium by SiO2-NPs. Therefore, to attempt to clarify the prominent factors affecting the structure and function of BBB in response to silica particles, we detected and compared the levels of cytokines (VEGF, IL-6, IL-8, IL-1b) from HBMECs co-culture and monoculture. Among these cytokines, VEGF are key regulators of vascular permeability. Both in vitro and in vivo experimental evidences supported that VEGF could induce the loss of tight and adherens junctions, cytoskeletal changes and the formation of fenestrations in endothelial cells [55]. Our data showed that SiO2-NPs exposure could cause a marked increase (~4fold) in the release of VEGF from HBMECs, compared to the untreated HBMECs (Fig. 4A), suggesting that VEGF might be a key determinant for SiO2-NPs-enhanced BBB permeability. Something noteworthy is that when co-cultured with astrocytes in a noncontact manner, a mildly amplification of VEGF (~2-fold) secretion was observed after exposure to SiO2-NPs compared to HBMECs monoculture (Fig. 4A), indicating that there was an additional secretion of VEGF from astrocytes under indirect exposure of SiO2NPs. Likewise, IL-1b release was slightly induced only in the HBMECs/astrocytes co-culture model in response to SiO2-NPs (Fig. 4B). Previous studies have shown that non-stimulated human astrocytes do not constitutively produce IL-1b, only activated human astrocytes could release IL-1b [56]. It is therefore reasonable to conclude that a secondary activation of astrocytes was induced by the SiO2-NPs-activated HBMECs. Recently, there are compelling views showing that inflammatory stimuli can activate astrocytes to generate and release VEGF, which increases BBB permeability [57]. To determine whether SiO2-NPs exposure could cause an inflammatory microenvironment in the BBB model, we then detected the levels of pro-inflammatory cytokines in HBMECs co-culture and monoculture after exposure to silica particles. As shown in Fig. 4C, IL-6 release from HBMECs monoculture in response to SiO2-NPs is quite pronounced after 24 h, however, little amplified effects of IL-6

3.4.3. Actin cytoskeleton rearrangement of HBMECs induced by SiO2-NPs The actin cytoskeleton of BMECs also contributed to establish endothelial tight junction integrity. Therefore, we further evaluated the changes in filamentous actin (F-actin) staining with Rhodamine-conjugated phalloidin. As shown in Fig. 3F, in the untreated HBMECs, many F-actin bundles in cellular cytoplasm were neatly arranged with peripheral dense band of actin located at cell borders. Upon exposure to SiO2-NPs, the peripheral dense band disappeared, and there was increase in stress fiber formation and sizable paracellular gaps between endothelial cells developed.

3.5. Contributing factors of paracellular BBB opening during exposure to SiO2-NPs in vitro

72

X. Liu et al. / Biomaterials 121 (2017) 64e82

Fig. 4. Cytokine production induced by silica particles in mono- and co-cultures. A) VEGF release; B) IL-1b release; C) IL-6 release; D) IL-8 release; E) HMGB-1 release; F) Human S100A8/S100A9 Heterodimer. Normal HBMECs served as the negative control. Data represent mean ± SD, n ¼ 3. *p < 0.05, **p < 0.01 vs. control. #p < 0.05 significant difference as compared groups. (EC: HBMECs; AS: astrocytes; SiO2-NPs: silica nanoparticles; SiO2-MPs: silica microparticles).

release was seen in HBMECs when co-cultured with astrocytes (Fig. 4C). Perhaps not surprising was that SiO2-MPs had no significant effects on VEGF and IL-6 release, implying nanoscale silica particles had a greater potential for triggering pro-inflammatory activation of HBMECs and astrocytes than micro-scale silica. Numerous studies including ours have shown that SiO2-NPs alone could directly activate HUVECs to produce a variety of proinflammatory mediators such as IL-6, IL-8 and ROS [24,29,30,58]. IL-6, as a pleiotropic cytokine in the CNS injury, has been reported to induce VEGF release from astrocytes as reported by previous studies [59]. Herein, our data suggest that IL-6 released from SiO2NPs-activated HBMECs is a potent paracrine modulator of astrocytes activation/function in the BBB model. In addition to these cytokines, DAMPs, also known molecules released by stressed cells undergoing necrosis that act as endogenous danger signals to promote and exacerbate the inflammatory response. The best known DAMPs are high mobility group box-1 (HMGB1), S100A8 and S100A9 [60,61]. To clarify the role of DAMPs in the BBB dysfunction induced by SiO2-NPs, we also further measured the release of DAMPs in the cells-free supernatants of HBMECs with or without NPs treatment by ELISA assay Our data showed that no significant increase in DAMPs release in cell-free supernatants of HBMECs after SiO2-NPs or SiO2-MPs treatment, compared to that in the negative control (cells without any particle treatment). (Fig. 4EeF), indicating the increase of BBB permeability was not dependent on the DAMP effects and the unspecific cytotoxicity of SiO2-NPs. 3.5.2. Inflammatory activation and oxidative stress in HBMECs induced by SiO2-NPs To further confirm the pro-inflammatory effects of SiO2-NPs on

the brain endothelial cells, the three CAMs [E-selectin(CD62E), ICAM-1(CD54), and VCAM-1(CD106)] were measured in HBMECs after exposure to silica particles for 24 h. Flow cytometry analysis showed that stimulation of HBMECs with 100 mg/mL SiO2-NPs induced a significant upregulation of constitutive ICAM-1 expression (92.5 ± 0.57% of positive cells) compared to cells exposed to control (80.7 ± 2.2% of positive cells). Amazingly, E-selectin cell surface levels were found to be dramatically higher (22.4 ± 3.5% of positive cells) in HBMECs after SiO2-NPs exposure versus the negative control (0.19 ± 0.035% of positive cells). In the SiO2-MPs exposure group, there was no evidence of increased adhesion molecule expressions in HBMECs (Fig. 5AeB). These results are in accordance with many previous studies showing that SiO2-NPs could elevate the ICAM-1 and E-selectin expression of endothelial cells and cause a significant increase in human monocytes adhesion to endothelial cells [29,62]. The up-regulation of ICAM-l and Eselectin are generally considered hallmarks of pro-inflammatory endothelial phenotypes, especially E-selectin, expressed solely on the activated endothelial cells [63], which are taken as conclusive evidences of HBMECs activation by SiO2-NPs. Increased expression of ICAM-1 and E-selectin are thought to not only facilitate the adhesion and transmigration of leukocytes across BBB [63], but also directly contribute to reduce endothelial cells barrier function by affecting cell junctions and the cytoskeleton [64]. BBB leukocyte infiltration contributes to the development of many neurological diseases such as multiple sclerosis, epilepsy, Alzheimer's disease and stroke [65]. Therefore, SiO2-NPs exposure is more likely to induce the pathological state of BBB dysfunction and increase the risks of neurological diseases than SiO2-MPs. Our previous published studies have illustrated that ROS plays a key role in the SiO2-NPs-induced endothelial cells inflammatory

X. Liu et al. / Biomaterials 121 (2017) 64e82

73

Fig. 5. Inflammatory activation and Oxidative stress in HBMECs and aquaporin4 (AQP4) expression in astrocytes induced by silica particles. A) Dot-plot of flow cytometry analysis showing the expressions of CD54, CD106, and CD62E on HBMECs exposed to SiO2-NPs or SiO2-MPs for 24 h; B) Bar graph shows the percentage of CD54, CD106, and CD62E positive cells in particles-stimulated HBMECs; C) ROS generation by silica particles in HBMECs; D) The AQP4 expression in astrocytes in BBB coculture by western blot analysis. Normal HBMECs served as the negative control. Data represent mean ± SD, n ¼ 3. **p < 0.01 vs. control.

74

X. Liu et al. / Biomaterials 121 (2017) 64e82

Fig. 6. The activation of MAPK signaling pathways involved in HBMECs exposed to silica particles. A) The activation of JNK, p38 and ERK induced by silica particles. b-actin was used as an internal control to monitor for equal loading; B) The relative density of band by gray value analysis.

activation [29]. Besides the pro-inflammatory ability of ROS to induce endothelial cells activation, recent evidence showed that ROS could alter BBB integrity by cytoskeleton rearrangements and disappearance of tight junction proteins claudin-5 and occludin [23]. Moreover, exposure to iron NPs has been reported to induce an increase in endothelial cell permeability through the ROS generation [66]. Considering the essential role of ROS on the BBB dysfunction, we here asked, if SiO2-NPs exposure can directly cause ROS production in HBMECs to induce BBB dysfunction. Consistent with our previous ROS results from SiO2-NPs-stimulated HUVECs [29], SiO2-NPs exposure could also cause a significant increase of intracellular ROS levels (~1.5 fold) in HBMECs compared to the negative control, while SiO2-MPs exposure did not induce the increase of ROS in HBMECs (Fig. 5C), suggesting that nano-sized SiO2 induce greater ROS as compared to their larger counterparts, which might due to their smaller size and higher surface to volume ratio. 3.5.3. AQP4 expression of astrocytes in the BBB co-cultures induced by SiO2-NPs In addition to endothelial cells, astrocytes have been demonstrated to play a critical role in maintaining BBB structure and function and in regulating the brain homeostasis [57,67]. AQP4, a water channel protein expressed mainly in the astrocytes is considered a specific marker of vascular permeability and thought to play an important role in brain edema [68]. We then detected the AQP4 expression of astrocytes in the BBB co-cultures after exposure to silica particles. Something noteworthy is that a dramatically increase of AQP4 expression of astrocytes was observed in HBMECs/ astrocytes co-culture system after SiO2-NPs exposure for 24 h. The up-regulation of AQP4 suggests that SiO2-NPs might cause BBB homeostasis dysfunction. It has been reported that oxidative stress mechanisms might be involved in AQP4 expression regulation [68].

Thus, the increased AQP4 expression of astrocytes in the BBB coculture model might be due to the ROS release from SiO2-NPsactivated BBB coculture. 3.6. Signaling pathways involved in SiO2-NPs induced-BBB dysfunction in vitro It was found that SiO2-NPs could compromise the BBB permeability, disrupt the BBB structure integrity and induce the BBB inflammatory activation and oxidative stress. We then further investigated the signaling pathway involved in these processes. Among different types of intracellular signaling pathways implicated in regulating endothelial permeability and inflammation, the activation of MAPKs pathways appear to be crucial signaling cascade respond to extracellular stimuli such as stress and growth factors [31]. Herein, three major kinases of MAPKs including extracellular signal-regulated kinase (ERK), c-Jun N-terminal kinases (JNK) and p38 MAPK were determined in HBMECs after exposure to SiO2-NPs for 24 h. As depicted in Fig. 6, among the three MAPKs, only JNK was dramatically activated in HBMECs after exposure to SiO2-NPs for 24 h, while both ERK and p38 phosphorylation was not altered by SiO2-NPs, which is in accordance with our previous studies [29,30]. Surprisingly, SiO2-MPs exposure also induced JNK phosphorylation in HBMECs (Fig. 6A). JNK, an important stress responsive kinase, has been reported to be a critical regulator involved in neuro-inflammation and bloodebrain barrier disruption by various forms of insults [69]. However, it seemed that SiO2-NPs-induced BBB dysfunction was not entirely dependent on the JNK pathway, since SiO2-MPs could also induce JNK activation but had no apparent adverse effects on BBB. Possibly, JNK pathway was just an “intermediate pathway”, and the question is as follows: what are the upstream or downstream components of

X. Liu et al. / Biomaterials 121 (2017) 64e82

75

Fig. 7. The protective effects of antioxidant (NAC), JNK inhibitor (SP600125) and ROCK inhibitor (Y-27632) on SiO2-NPs-induced A) TEER values in BBB; B) HRP permeability; C) Occludin expression in HBMECs; D) the release of VEGF and IL-6. Data represent means ± SEM; n ¼ 3.*p < 0.05, **p < 0.01 vs. control. #p < 0.05 ##p < 0.01 significant difference as compared groups.

such signaling? Recently, significant attention has been focused on Rho/ROCK as major regulators of BBB dysfunction and it was found that ROCK could modulate the downstream JNK activation in microvascular endothelial cells during inflammation [31,70]. Moreover, several factors including ROS, pro-inflammatory cytokines, microtubule destabilization and VEGF are generally reported to be as upstream activators of the ROCK and MAPK pathways for regulating endothelial permeability [31,50]. To further confirm whether and how these signaling molecules regulate BBB structure and function, we then examined the protective effect of ROS, JNK and ROCK inhibition on BBB permeability, integrity and inflammation after exposure to SiO2-NPs using the ROS inhibitor NAC, JNK inhibitor SP600125 and ROCK inhibitor Y-27632, respectively. Suppression of JNK and ROCK could significantly increase the TEER values to the normal levels in BBB co-culture model starting at early exposure of SiO2-NPs for 3 h, while suppression of ROS did not elevate the TEER value until exposure of SiO2-NPs for 9 h, suggesting ROS signaling might lag behind ROCK/JNK signaling in HBMECs after exposure to SiO2-NPs. At 24 h, suppression of ROS and ROCK could prevent the BBB permeability induced by SiO2-NPs, while JNK inhibition could not fully recover the TEER values to normal levels (Fig. 7A). Consistent with the TEER results, the blockade of ROS and ROCK could dramatically decrease BBB permeability, while sole inhibition of JNK had no marked effects on HRP permeability (Fig. 7B), indicating ROS and ROCK might be the key signaling molecules implicated in SiO2-NPs-induced BBB permeability increase. Both ROS and ROCK signaling pathways are reported to mediate BBB permeability by inducing cytoskeletal contractile responses and affecting tight junction proteins phosphorylation [50,71]. Interestingly, our results showed that only NAC (ROS inhibitor) could significantly protect the down-regulation of

tight junction occludin expression in SiO2-NPs-stimulated HBMECs, while ROCK and JNK inhibitors did not have enough effects on the tight junctions up-regulation, suggesting ROS was the key signaling molecules for decreasing the expression of tight junctions in HBMECs exposed to SiO2-NPs (Fig. 7C). On the other hand, suppression of ROS did not alleviate the release of VEGF and proinflammatory cytokines (IL-6) from SiO2-NPs-activated BBB coculture, while suppression of ROCK and JNK could dramatically reduce the BBB inflammatory response and VEGF generation (Fig. 7DeE). These results suggest that both ROS and ROCK respectively mediated two distinct and independent signaling pathways involved in the BBB dysfunction induced by SiO2-NPs, where the ROS signaling pathway was more prone to regulating the tight junction expression, and ROCK/JNK signaling pathway was required for the release of VEGF and IL-6. Based on the abovementioned results, we delineated a schematic representation of the mechanisms underlying SiO2-NPsinduced BBB dysfunction (Fig. 8): SiO2-NPs exposure could induce oxidative stress and thus result in down-regulation of the tight junctions (ZO-1 and Occludin); meanwhile it might cause microtubule destabilization and further activate the ROCK/JNK signaling to induce the release of VEGF and inflammatory activation of brain endothelial cells (e.g. upregulation of ICAM-1and E-selectin, release of IL-6); when the astrocytes were added, some pro-inflammatory mediators (e.g. IL-6, ROS) from the endothelium subsequently activated astrocytes to amplify the release of VEGF and increase the expression of AQP4, which in turn enhanced the inflammatory response and permeability in BBB, and thus finally resulted in BBB disruption by a complex immunoregulatory loop between BMECs and astrocytes under SiO2-NPs exposure.

76

X. Liu et al. / Biomaterials 121 (2017) 64e82

Fig. 8. Schematic conclusion of biological pathways of SiO2-NPs-mediated BBB dysfunction. SiO2-NPs exposure could induce oxidative stress and thus result in down-regulation of the tight junctions (ZO-1 and Occludin); meanwhile it might cause microtubule destabilization and further activate the ROCK/JNK signaling to induce the release of VEGF and inflammatory activation of brain endothelial cells (e.g. upregulation of ICAM-1and E-selectin, release of IL-6); when the astrocytes were added, some pro-inflammatory mediators (e.g. IL-6, ROS) from the endothelium subsequently activated astrocytes to amplify the release of VEGF and increase the expression of AQP4, which in turn enhanced the inflammatory response and permeability in BBB, and thus finally resulted in BBB disruption by a complex immunoregulatory loop between BMECs and astrocytes under SiO2-NPs exposure.

3.7. Assessment of BBB opening and brain edema after SiO2-NPs exposure in vivo in SD rats Due to our surprising observation that SiO2-NPs exposure could cause paracellular BBB opening in the BBB in vitro co-culture model, we next validated if SiO2-NPs exposure could affect the BBB permeability in vivo. Firstly, SD rats were treated by intraperitoneal injection of silica particles for 28 consecutive days, and then the cerebral micro-vasculature permeability was examined using an Evans blue extravasation assay at days 28 post injection. Because there are no available guidelines for conducting NPs in vivo toxicity assessment, considering SiO2-NPs as medical materials are mainly used for long-term body contact, we thus selected intraperitoneal injection as the route of exposure and chose a dose of 50 mg/kg according to ISO 10993-11 and previous literature for studying systemic toxicity of SiO2-NPs [72]. It is well known that Evans blue dye can bind albumin of the plasma and act as a classic indicator of

BBB leakage in vivo. In this study, the gross view of the brains of rats in experimental groups is shown in Fig. 9A. Evans blue extravasation was greater in the SiO2-NPs group than in the negative control, and no significant influence of SiO2-MPs exposure was observed on this finding in these animals (Fig. 9B), indicating SiO2-NPs exposure could elevate BBB permeability and cause a paracellular BBB opening. To investigate whether SiO2-NPs enter into the brain parenchyma through the BBB, the Si concentrations in brain were measured by ICP. Interestingly, the Si brain concentrations tend to increase in rats after exposure of SiO2-NPs in comparison with the SiO2-MPs and control groups (Fig. 9D), indicating SiO2-NPs might more easily compromise the BBB and enter into the brain than SiO2MPs. Similar results were also observed in a previous study showing that SiO2-NPs (70 nm, 0.5 mg/day) could enter not only the skin but also the cerebral cortex and the hippocampus of brain tissue in mice after the 28-day dermal application [16]. On the other hand, another study reported that no deposition of SiO2-NPs

X. Liu et al. / Biomaterials 121 (2017) 64e82

77

Fig. 9. Effects of silica particles on Evans blue extravasation, brain water content and oxidative stress in vivo. A) Representative photographs of the gross view of brain. Evans blue extravasation was observed in whole brain as dark blue areas; B) Quantitative analysis of Evans blue contents in brain tissues; C) Brain water content in brain tissue after silica particles exposure; D) The Si concentrations in brain tissue after silica particles exposure; E) The MDA levels in brain tissue after silica particles exposure. Data shown are mean ± SEM. (n ¼ 5). *p < 0.05, **p < 0.01 vs. the negative control (0.9% saline). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

(20 nm and 100 nm) was observed in hippocampus, striatum, and cerebellum regions of rats brain after 90 days of repeated dermal and oral administrations [17]. Moreover, the initial biodistribution data, including ours, also showed that neither 30 days of intravenous injection nor 7 days of intraperitoneal application of SiO2-NPs (20 nm, 10e20 mg/kg) did not cause the significant increases in the SiO2-NPs accumulation in the brain of mice [32,73]. The discrepancy between these brain biodistribution data from systemic uptake of SiO2-NPs might be explained in terms of particles size and shape, exposure routes, exposure dose and dosing frequency. For example, intraperitoneal application of SiO2-NPs may improve the NPs deposition within the brain when compared to intravenous injection. Due to a slower uptake of NPs into the bloodstream when intraperitoneal injection is used, NPs could not be dismantled as quickly by the liver and were thus distributed into the brain. In addition, other parameters such as smaller particle size, appropriate dose without NPs aggregation and increasing dose frequency may promote the transposition of NPs to the brain. To further

determine whether SiO2-NPs exposure induced brain edema following the BBB opening, the brain water content was measured with wet-dry weights method. No measureable changes were found in the cerebral water content in SiO2-NPs group, compared to the negative and SiO2-MPs controls (Fig. 9C). Evaluation of the brain sections stained with H&E revealed a few neuronophagia and perivascular edema formed within the cerebral cortex in rats treated with SiO2-NPs (Fig. 10B, b). Satellitosis which composed of oligodendrocytes (with few astrocytes) clustered around a possibly degenerate neuron was also observed in the SiO2-NPs treated group (Fig. 10B, b), while no significant neuronal degeneration was observed in hippocampus region (Fig. 10). 3.8. In vivo oxidative stress and the expression of p-MLC, GFAP and AQP4 in brain induced by SiO2-NPs As mentioned above, the evidence from in vitro BBB model showed that SiO2-NPs exposure induced BBB dysfunction mainly by

78

X. Liu et al. / Biomaterials 121 (2017) 64e82

Fig. 10. Alterations in histopathology in the brain tissue following treatment with silica particles. AeC, aec) Cerebral cortex; DeF, def) Hippocampus. A, a, D, d are examples from a 0.9% saline treated control rat. B, b, E and e are examples from a SiO2-NPs treated rat. C, c F and f are examples from a SiO2-MPs treated rat. The black arrow in b points to neuronophagia and satellitosis. The black arrow in b points to perivascular edema. aec (400) are magnified views of regions from AeC(100) showing cortex region; def (200) are magnified views of regions from DeF (40) showing hippocampus region. Data shown are representative of five separate experiments.

the direct action of oxidative stress and cytoskeletal contractile via ROCK pathway in endothelial cells and the indirect effects of astrocytes activation. Therefore, to assess the extent of oxidative stress, ROCK and astrocytes activation, the MDA level and the endothelial myosin light chain (MLC) phosphorylation, the GFAP and AQP4 expression were measured in the brain of SD rats exposed to silica particles. Consistent with the in vitro ROS results, intraperitoneal exposure of the 20 nm SiO2-NPs for 28 days resulted in a significant increase in the MDA production in the brain, while no apparent changes of MDA was observed in the SiO2-MPs and the negative control group (Fig. 9E). MDA is the end product of lipid peroxidation induced by excessive generation of intracellular ROS, so our results indicate that SiO2-NPs could cause oxidative injury to the brain, which is in accordance with previous studies [28,74]. Our in vitro studies have shown that besides a down-regulation of tight junctions by oxidative stress, another possible explanation for the SiO2-NPs -induced opening of the BBB could be the activation of the endothelial cell contractile machinery by ROCK pathway. ROCK promotes phosphorylation of the regulatory light-chain of myosin (MLC), which leads to activation of the endothelial contractile elements [50]. Our results are in accordance with these in vitro data, and showed that phosphorylation MLC staining in brain microvessels was enhanced by SiO2-NPs exposure in the cerebral cortex

and hippocampus region of rat brain, compared to the negative and SiO2-MPs controls (Fig. 11). To investigate the role of astrocyte in the SiO2-NPs-mediated BBB opening in vivo, the GFAP and AQP4 expression were further measured in brain of SD rats. The immunostain of cytoskeletal protein GFAP demonstrated a marked up-regulation of GFAP immunoreactivity in the cerebral cortex and hippocampus regions of brains in animals exposed to SiO2-NPs, and in GFAP stained sections, more reactive astrocytes were observed in the SiO2-NPs group, in comparison to the SiO2-MPs group and the control animals (Fig. 12, Fig. S2), suggesting that SiO2-NPs exposure significantly induced astrocytes activation in vivo. This result was also consistent with a recent study showing that silver nanoparticle exposure could increase the GFAP expression in astrocytes in the rat brain [75]. AQP4, a water channel protein, is largely expressed in the circumvascular astrocyte end-feet around the surface of capillaries, which considered a specific marker of vascular permeability and thought to play an important role in brain edema. Our in vitro evidence has pointed to an increase of AQP4 by SiO2-NPs in the BBB model. Immunostaining further showed that AQP4-positive cell was widely expressed on astrocytes around capillaries in the cerebral cortex and hippocampus regions after SiO2-NPs exposure (Fig. 12, Fig. S3). Evidence demonstrated that both

X. Liu et al. / Biomaterials 121 (2017) 64e82

79

Fig. 11. Phosphorylation myosin light chain (p-MLC) immunostaining of the brain tissue following treatment with silica particles. AeC, aec) Cerebral cortex; DeF, def) Hippocampus. A, a, D, d are examples from a 0.9% saline treated control rat. B, b, E and e are examples from a SiO2-NPs treated rat. C, c F and f are examples from a SiO2-MPs treated rat. aec (400) are magnified views of regions from AeC(100X) showing cortex region; def (400) are magnified views of regions from DeF (100X) showing hippocampus region. Data shown are representative of five separate experiments.

neuroinflammation characterized with glial cell activation and proinflammatory cytokines was related to AQP4 expression [36]. Therefore, it could be inferred the increased AQP4 in rats brain by SiO2-NPs induction might due to the activation of astrocytes.

inflammatory environment can lead to the generation of oxygenand nitrogen-free radicals as well as pro-inflammatory cytokines which contribute to the development of BBB dysfunction. 4. Conclusion

3.9. Systemic inflammation profile of sub-acute exposure to SiO2NPs Besides local response such as inflammatory response and oxidative stress, systemic response was also thought to be intimately linked to the CNS injury. Therefore, we then performed the hematology examination to assess the potential systemic inflammatory potential of SiO2-NPs. Complete blood count sampling in the SiO2-NPs treated group showed a significant increase in the number of white blood cells with increase in the percentage of granulocytes and decrease in the lymphocytes percentage in the SiO2-NPs group, in comparison to the control and to the SiO2-MPs group, while the monocytes percentage was not significantly altered (Table 1). Meanwhile, exposure to SiO2-NPs caused a significant decrease in circulating platelets compared to the control and SiO2-MPs group (Table 1). These results indicate that SiO2-NPs could induce systemic inflammation, which was consistent with the data reported by other studies [24,25]. The systemic

In summary, our in vitro results demonstrated that a sub-toxic dose of SiO2-NPs (20 nm) could initially induce tight junctions loss and cytoskeleton arrangement, lead to inflammatory activation and increase the release of VEGF in BMECs, which further activate astrocytes to amplify the VEGF generation and increase the AQP4 expression, and thus finally cause BBB disruption by a complex immunoregulatory loop between BMECs and astrocytes under SiO2-NPs exposure. In addition to this, our data show that inhibition of ROS and ROCK/JNK signaling could effectively protect the SiO2NP-induced BBB dysfunction. In vivo studies further confirmed that SiO2-NPs could cause the BBB paracellular opening, oxidative stress and astrocyte activation in the brain. However, no obvious adverse effects were observed in the BBB structure and function in the SiO2MPs group. These data strongly suggested that we must distinguish NPs from existing microparticles, and address specialized neurotoxicity risk assessment for NPs. When taken together, these findings demonstrate that SiO2-NPs could disturb BBB structure and

80

X. Liu et al. / Biomaterials 121 (2017) 64e82

Fig. 12. Gial fibrillary acidic protein (GFAP) and Aquaporin-4(AQP4) immunostaining of the brain tissue following treatment with silica particles. AeC) The GFAP expression in cerebral cortex; DeF) The GFAP expression in hippocampus; G) Area occupied by GFAP-positive scar-forming astrocytes; HeJ) The AQP4 expression in cerebral cortex; KeM) The AQP4 expression in hippocampus; N) Area occupied by AQP4-positive cells; A, D, H, K are examples from a 0.9% saline treated control rat. B, E, I and L are examples from a SiO2-NPs treated rat. C, F, J and M are examples from a SiO2-MPs treated rat. (200X). n ¼ 5.*p < 0.05, **p < 0.01 vs. control. #p < 0.05 ##p < 0.01 significant difference as compared groups.

Table 1 Hematology data of main groups in the 28-day study of silica particles. Parameter

Control

SiO2-NPs

SiO2-MPs

WBC (109/L) Lymphocyte (%) Monocyte (%) Granulocyte (%) RBC (1012/L) HGB (g/L) HCT (%) MCV (fL) PLT (109/L)

12.44 ± 1.51 66.14 ± 4.95 2.84 ± 0.27 31.12 ± 5.18 6.91 ± 0.15 133.2 ± 2.88 40.16 ± 0.95 58.2 ± 0.84 594.8 ± 105.36

25.8 ± 1.44** 44.93 ± 6.02** 3.06 ± 0.76 52 ± 5.27** 6.33 ± 0.44 124.3 ± 5.11 36.1 ± 0.93* 57.1 ± 2.45 275.67 ± 55.11**

10.7 ± 1 60.9 ± 5.32 2.46 ± 0.23 37.12 ± 4.18 7.09 ± 0.4 134 ± 8 40.08 ± 1.66 51.67 ± 0.76 598.8 ± 58.24

Lymphocytes, monocytes and granulocytes expressed in %; WBC, white blood cells; RBC, red blood cells; HGB, Haemoglobin; HCT, hematocrit; MCV, mean cell volume; PLT, platelet; Data are given as mean ± standard deviation (n ¼ 5); Significantly different from negative control (*P < 0.05, **P < 0.01).

function and induce BBB inflammation, and these effects may occur through ROS and ROCK-mediated pathways, which not only improve neurotoxicity evaluation of SiO2-NPs but also provide useful information in the development of SiO2-NPs in neurotherapeutics and nanodiagnostics.

Acknowledgements This work was supported by grants from National Natural Science Foundation of China (no. 81201191, no. 81271700), and Shanghai Sci-Tech Committee Foundation (15DZ2290100).

Appendix A. Supplementary data Supplementary data related to this article can be found at http:// dx.doi.org/10.1016/j.biomaterials.2017.01.006. References [1] L. Tang, J. Cheng, Nonporous silica nanoparticles for nanomedicine application, Nano Today 8 (2013) 290e312. [2] M. Benezra, O. Penate-Medina, P.B. Zanzonico, D. Schaer, H. Ow, A. Burns, et al., Multimodal silica nanoparticles are effective cancer-targeted probes in a model of human melanoma, J. Clin. Investig. 121 (2011) 2768e2780. [3] S. Barua, S. Mitragotri, Challenges associated with penetration of nanoparticles across cell and tissue barriers: a review of current status and future prospects, Nano Today 9 (2014) 223e243. [4] K. Yamashita, Y. Yoshioka, K. Higashisaka, K. Mimura, Y. Morishita, M. Nozaki, et al., Silica and titanium dioxide nanoparticles cause pregnancy complications in mice, Nat. Nanotechnol. 6 (2011) 321e328. [5] C. Disdier, J. Devoy, A. Cosnefroy, M. Chalansonnet, N. Herlin-Boime, E. Brun, et al., Tissue biodistribution of intravenously administrated titanium dioxide nanoparticles revealed blood-brain barrier clearance and brain inflammation in rat, Part Fibre Toxicol. 12 (2015) 27. [6] W.J. Trickler, S.M. Lantz-McPeak, B.L. Robinson, M.G. Paule, W. Slikker Jr., A.S. Biris, et al., Porcine brain microvessel endothelial cells show proinflammatory response to the size and composition of metallic nanoparticles, Drug Metab. Rev. 46 (2014) 224e231. [7] W.J. Trickler, S.M. Lantz, R.C. Murdock, A.M. Schrand, B.L. Robinson, G.D. Newport, et al., Silver nanoparticle induced blood-brain barrier inflammation and increased permeability in primary rat brain microvessel endothelial cells, Toxicol. Sci. 118 (2010) 160e170. [8] W.J. Trickler, S.M. Lantz, R.C. Murdock, A.M. Schrand, B.L. Robinson, G.D. Newport, et al., Brain microvessel endothelial cells responses to gold nanoparticles: in vitro pro-inflammatory mediators and permeability, Nanotoxicology 5 (2011) 479e492. [9] C.H. Li, C. Jhan, Y.W. Cheng, C.H. Tsai, C.W. Liu, C.C. Lee, et al., Gold

X. Liu et al. / Biomaterials 121 (2017) 64e82

[10]

[11]

[12]

[13]

[14]

[15]

[16]

[17]

[18]

[19]

[20]

[21]

[22]

[23]

[24]

[25] [26]

[27]

[28]

[29]

[30]

[31]

[32]

[33]

[34]

[35]

nanoparticles increase endothelial paracellular permeability by altering components of endothelial tight junctions, and increase blood-brain barrier permeability in mice, Toxicol. Sci. 148 (2015) 192e203. L. Chen, R.A. Yokel, B. Hennig, M. Toborek, Manufactured aluminum oxide nanoparticles decrease expression of tight junction proteins in brain vasculature, J. Neuroimmune Pharmacol. 3 (2008) 286e295. L. Xu, M. Dan, A. Shao, X. Cheng, C. Zhang, R.A. Yokel, et al., Silver nanoparticles induce tight junction disruption and astrocyte neurotoxicity in a rat blood-brain barrier primary triple coculture model, Int. J. Nanomed. 10 (2015) 6105e6119. S. Hanada, K. Fujioka, Y. Inoue, F. Kanaya, Y. Manome, K. Yamamoto, Cellbased in vitro blood-brain barrier model can rapidly evaluate nanoparticles' brain permeability in association with particle size and surface modification, Int. J. Mol. Sci. 15 (2014) 1812e1825. D. Liu, B. Lin, W. Shao, Z. Zhu, T. Ji, C. Yang, In vitro and in vivo studies on the transport of PEGylated silica nanoparticles across the blood-brain barrier, ACS Appl. Mater. Interfaces 6 (2014) 2131e2136. D. Ye, S. Anguissola, T. O'Neill, K.A. Dawson, Immunogold labeling reveals subcellular localisation of silica nanoparticles in a human blood-brain barrier model, Nanoscale 7 (2015) 10050e10058. J. Jampilek, K. Zaruba, M. Oravec, M. Kunes, P. Babula, P. Ulbrich, et al., Preparation of silica nanoparticles loaded with nootropics and their in vivo permeation through blood-brain barrier, Biomed. Res. Int. (2015) 812673. H. Nabeshi, T. Yoshikawa, K. Matsuyama, Y. Nakazato, K. Matsuo, A. Arimori, et al., Systemic distribution, nuclear entry and cytotoxicity of amorphous nanosilica following topical application, Biomaterials 32 (2011) 2713e2724. K.H. Shim, K.H. Jeong, S.O. Bae, M.O. Kang, E.H. Maeng, C.S. Choi, et al., Assessment of ZnO and SiO2 nanoparticle permeability through and toxicity to the blood-brain barrier using evans blue and TEM, Int. J. Nanomed. 9 (Suppl 2) (2014) 225e233. B. Zhang, L. Chen, J.J. Choi, B. Hennig, M. Toborek, Cerebrovascular toxicity of PCB153 is enhanced by binding to silica nanoparticles, J. Neuroimmune Pharmacol. 7 (2012) 991e1001. H.S. Sharma, D.F. Muresanu, R. Patnaik, A. Sharma, Exacerbation of brain pathology after partial restraint in hypertensive rats following SiO(2) nanoparticles exposure at high ambient temperature, Mol. Neurobiol. 48 (2013) 368e379. A.K. Shetty, V. Mishra, M. Kodali, B. Hattiangady, Blood brain barrier dysfunction and delayed neurological deficits in mild traumatic brain injury induced by blast shock waves, Front. Cell Neurosci. 8 (2014) 232. C.M. Zehendner, L. Librizzi, J. Hedrich, N.M. Bauer, E.A. Angamo, M. de Curtis, et al., Moderate hypoxia followed by reoxygenation results in blood-brain barrier breakdown via oxidative stress-dependent tight-junction protein disruption, PLoS One 8 (2013) e82823. S.M. Stamatovic, O.B. Dimitrijevic, R.F. Keep, A.V. Andjelkovic, Inflammation and brain edema: new insights into the role of chemokines and their receptors, Acta Neurochir. Suppl. 96 (2006) 444e450. G. Schreibelt, G. Kooij, A. Reijerkerk, R. van Doorn, S.I. Gringhuis, S. van der Pol, et al., Reactive oxygen species alter brain endothelial tight junction dynamics via RhoA, PI3 kinase, and PKB signaling, FASEB J. 21 (2007) 3666e3676. A. Nemmar, S. Albarwani, S. Beegam, P. Yuvaraju, J. Yasin, S. Attoub, et al., Amorphous silica nanoparticles impair vascular homeostasis and induce systemic inflammation, Int. J. Nanomed. 9 (2014) 2779e2789. Q. Chen, Y. Xue, J. Sun, Kupffer cell-mediated hepatic injury induced by silica nanoparticles in vitro and in vivo, Int. J. Nanomed. 8 (2013) 1129e1140. Z. Du, D. Zhao, L. Jing, G. Cui, M. Jin, Y. Li, et al., Cardiovascular toxicity of different sizes amorphous silica nanoparticles in rats after intratracheal instillation, Cardiovasc Toxicol. 13 (2013) 194e207. D. Huh, B.D. Matthews, A. Mammoto, M. Montoya-Zavala, H.Y. Hsin, D.E. Ingber, Reconstituting organ-level lung functions on a chip, Science 328 (2010) 1662e1668. J. Wu, C. Wang, J. Sun, Y. Xue, Neurotoxicity of silica nanoparticles: brain localization and dopaminergic neurons damage pathways, ACS Nano 5 (2011) 4476e4489. X. Liu, J. Sun, Endothelial cells dysfunction induced by silica nanoparticles through oxidative stress via JNK/P53 and NF-kappaB pathways, Biomaterials 31 (2010) 8198e8209. X. Liu, Y. Xue, T. Ding, J. Sun, Enhancement of proinflammatory and procoagulant responses to silica particles by monocyte-endothelial cell interactions, Part Fibre Toxicol. 9 (2012) 36. L. Gonzalez-Mariscal, R. Tapia, D. Chamorro, Crosstalk of tight junction components with signaling pathways, Biochim. Biophys. Acta 1778 (2008) 729e756. Xie G, Sun J, Zhong G, Shi L, Zhang D, Biodistribution and toxicity of intravenously administered silica nanoparticles in mice, Arch. Toxicol., 84: 183e190. A. Franca, B. Pelaz, M. Moros, C. Sanchez-Espinel, A. Hernandez, C. FernandezLopez, et al., Sterilization matters: consequences of different sterilization techniques on gold nanoparticles, Small 6 (2010) 89e95. E. Garcia-Garcia, S. Gil, K. Andrieux, D. Desmaele, V. Nicolas, F. Taran, et al., A relevant in vitro rat model for the evaluation of blood-brain barrier translocation of nanoparticles, Cell Mol. Life Sci. 62 (2005) 1400e1408. m, Y. Kataoka, M. Niwa, Permeability studies on in vitro M.A. Deli, C.S. Abraha bloodebrain barrier models: physiology, pathology, and pharmacology, Cell

81

Mol. Neurobiol. 25 (2005) 59e127. [36] S. Cao, P. Zhu, X. Yu, J. Chen, J. Li, F. Yan, et al., Hydrogen sulfide attenuates brain edema in early brain injury after subarachnoid hemorrhage in rats: possible involvement of MMP-9 induced blood-brain barrier disruption and AQP4 expression, Neurosci. Lett. 621 (2016) 88e97. [37] M.A. Anderson, J.E. Burda, Y. Ren, Y. Ao, T.M. O'Shea, R. Kawaguchi, et al., Astrocyte scar formation aids central nervous system axon regeneration, Nature 532 (7598) (2016) 195e200. [38] M.A. Gatoo, S. Naseem, M.Y. Arfat, A.M. Dar, K. Qasim, S. Zubair, Physicochemical properties of nanomaterials: implication in associated toxic manifestations, Biomed. Res. Int. (2014) 498420. [39] M.P. Monopoli, C. Aberg, A. Salvati, K.A. Dawson, Biomolecular coronas provide the biological identity of nanosized materials, Nat. Nanotechnol. 7 (2012) 779e786. [40] A. Lesniak, F. Fenaroli, M.P. Monopoli, C. Aberg, K.A. Dawson, A. Salvati, Effects of the presence or absence of a protein corona on silica nanoparticle uptake and impact on cells, ACS Nano 6 (7) (2012) 5845e5857. [41] D.J. O'Connell, F.B. Bombelli, A.S. Pitek, M.P. Monopoli, D.J. Cahill, K.A. Dawson, Characterization of the bionano interface and mapping extrinsic interactions of the corona of nanomaterials, Nanoscale 7 (2015) 15268e15276. [42] E. Brun, M. Carriere, A. Mabondzo, In vitro evidence of dysregulation of bloodbrain barrier function after acute and repeated/long-term exposure to TiO(2) nanoparticles, Biomaterials 33 (2012) 886e896. [43] R. Qiao, Q. Jia, S. Huwel, R. Xia, T. Liu, F. Gao, et al., Receptor-mediated delivery of magnetic nanoparticles across the blood-brain barrier, ACS Nano 6 (2012) 3304e3310. [44] G.D. Kanmogne, K. Schall, J. Leibhart, B. Knipe, H.E. Gendelman, Y. Persidsky, HIV-1 gp120 compromises blood-brain barrier integrity and enhances monocyte migration across blood-brain barrier: implication for viral neuropathogenesis, J. Cereb. Blood Flow. Metab. 27 (2007) 123e134. [45] I. Megard, A. Garrigues, S. Orlowski, S. Jorajuria, P. Clayette, E. Ezan, et al., A coculture-based model of human blood-brain barrier: application to active transport of indinavir and in vivo-in vitro correlation, Brain Res. 927 (2002) 153e167. [46] R.C. Winger, J.E. Koblinski, T. Kanda, R.M. Ransohoff, W.A. Muller, Rapid remodeling of tight junctions during paracellular diapedesis in a human model of the blood-brain barrier, J. Immunol. 193 (2014) 2427e2437. [47] D. Napierska, L.C. Thomassen, V. Rabolli, D. Lison, L. Gonzalez, M. KirschVolders, et al., Size-dependent cytotoxicity of monodisperse silica nanoparticles in human endothelial cells, Small 5 (2009) 846e853. [48] M. Bramini, D. Ye, A. Hallerbach, M. Nic Raghnaill, A. Salvati, C. Aberg, et al., Imaging approach to mechanistic study of nanoparticle interactions with the blood-brain barrier, ACS Nano 8 (2014) 4304e4312. [49] M. Tajes, E. Ramos-Fernandez, X. Weng-Jiang, M. Bosch-Morato, B. Guivernau, A. Eraso-Pichot, et al., The blood-brain barrier: structure, function and therapeutic approaches to cross it, Mol. Membr. Biol. 31 (2014) 152e167. [50] S.M. Stamatovic, R.F. Keep, A.V. Andjelkovic, Brain endothelial cell-cell junctions: how to “open” the blood brain barrier, Curr. Neuropharmacol. 6 (2008) 179e192. [51] W.Y. Liu, Z.B. Wang, L.C. Zhang, X. Wei, L. Li, Tight junction in blood-brain barrier: an overview of structure, regulation, and regulator substances, CNS Neurosci. Ther. 18 (2012) 609e615. [52] L. Chen, W. Liu, P. Wang, Y. Xue, Q. Su, C. Zeng, et al., Endophilin-1 regulates blood-brain barrier permeability via EGFR-JNK signaling pathway, Brain Res. 1606 (2015), 44e1653. [53] J. Duan, Y. Yu, Y. Li, P. Huang, X. Zhou, S. Peng, et al., Silica nanoparticles enhance autophagic activity, disturb endothelial cell homeostasis and impair angiogenesis, Part Fibre Toxicol. 11 (2014) 50. [54] M. Ehrenberg, J.L. McGrath, Binding between particles and proteins in extracts: implications for microrheology and toxicity, Acta Biomater. 1 (2005) 305e315. [55] D.O. Bates, Vascular endothelial growth factors and vascular permeability, Cardiovasc Res. 87 (2014) 262e271. [56] S.S. Choi, H.J. Lee, I. Lim, J. Satoh, S.U. Kim, Human astrocytes: secretome profiles of cytokines and chemokines, PLoS One 9 (2014) e92325. [57] M.V. Sofroniew, Astrocyte barriers to neurotoxic inflammation, Nat. Rev. Neurosci. 16 (2015) 249e263. [58] D. Napierska, L.C. Thomassen, B. Vanaudenaerde, K. Luyts, D. Lison, J.A. Martens, et al., Cytokine production by co-cultures exposed to monodisperse amorphous silica nanoparticles: the role of size and surface area, Toxicol. Lett. 211 (2012) 98e104. [59] S. Loeffler, B. Fayard, J. Weis, J. Weissenberger, Interleukin-6 induces transcriptional activation of vascular endothelial growth factor (VEGF) in astrocytes in vivo and regulates VEGF promoter activity in glioblastoma cells via direct interaction between STAT3 and Sp1, Int. J. Cancer 115 (2005) 202e213. [60] L. Wang, H. Luo, X. Chen, Y. Jiang, Q. Huang, Functional characterization of S100A8 and S100A9 in altering monolayer permeability of human umbilical endothelial cells, PLoS One (2014) e90472. [61] M.E. Bianchi, DAMPs, PAMPs and alarmins: all we need to know about danger, J. Leukoc. Biol. 81 (2007) 1e5. [62] D. Napierska, R. Quarck, L.C. Thomassen, D. Lison, J.A. Martens, M. Delcroix, et al., Amorphous silica nanoparticles promote monocyte adhesion to human endothelial cells: size-dependent effect, Small 9 (2013) 430e438. [63] S.J. Lee, E.N. Benveniste, Adhesion molecule expression and regulation on cells of the central nervous system, J. Neuroimmunol. 98 (1999) 77e88.

82

X. Liu et al. / Biomaterials 121 (2017) 64e82

[64] P.R. Clark, T.D. Manes, J.S. Pober, M.S. Kluger, Increased ICAM-1 expression causes endothelial cell leakiness, cytoskeletal reorganization and junctional alterations, J. Investig. Dermatol. 127 (2007) 762e774. [65] S.F. Rodrigues, D.N. Granger, Blood cells and endothelial barrier function, Tissue Barriers 3 (2015) e978720. [66] P.L. Apopa, Y. Qian, R. Shao, N.L. Guo, D. Schwegler-Berry, M. Pacurari, et al., Iron oxide nanoparticles induce human microvascular endothelial cell permeability through reactive oxygen species production and microtubule remodeling, Part Fibre Toxicol. 6 (2009) 1. [67] N.J. Abbott, L. Ronnback, E. Hansson, Astrocyte-endothelial interactions at the blood-brain barrier, Nat. Rev. Neurosci. 7 (2006) 41e53. [68] S. Ceccariglia, A. D'Altocolle, A. Del Fa, A. Silvestrini, M. Barba, F. Pizzolante, et al., Increased expression of Aquaporin 4 in the rat hippocampus and cortex during trimethyltin-induced neurodegeneration, Neuroscience 274 (2014) 273e288. [69] L.W. Wang, Y.F. Tu, C.C. Huang, C.J. Ho, JNK signaling is the shared pathway linking neuroinflammation, blood-brain barrier disruption, and oligodendroglial apoptosis in the white matter injury of the immature brain, J. Neuroinflamm. 9 (2012) 175.

[70] P.Y. Mong, C. Petrulio, H.L. Kaufman, Q. Wang, Activation of Rho kinase by TNF-alpha is required for JNK activation in human pulmonary microvascular endothelial cells, J. Immunol. 180 (2008) 550e558. [71] P.B. Pun, J. Lu, S. Moochhala, Involvement of ROS in BBB dysfunction, Free Radic. Res. 43 (2009) 348e364. [72] T. Liu, L. Li, C. Fu, H. Liu, D. Chen, F. Tang, Pathological mechanisms of liver injury caused by continuous intraperitoneal injection of silica nanoparticles, Biomaterials 33 (2012) 2399e2407. [73] M. Guo, X. Xu, X. Yan, S. Wang, S. Gao, S. Zhu, In vivo biodistribution and synergistic toxicity of silica nanoparticles and cadmium chloride in mice, J. Hazard Mater. 260 (2013) 780e788. [74] A. Parveen, S.H. Rizvi, Sushma, F. Mahdi, I. Ahmad, P.P. Singh, et al., Intranasal exposure to silica nanoparticles induce alterations in pro-inflammatory environment of rat brain: involvement of oxidative stress, Toxicol. Ind. Health (2015), http://dx.doi.org/10.1177/0748233715602985 pii: 0748233715602985 [Epub ahead of print]. [75] N. Yin, X. Yao, Q. Zhou, F. Faiola, G. Jiang, Vitamin E attenuates silver nanoparticle-induced effects on body weight and neurotoxicity in rats, Biochem. Biophys. Res. Commun. 458 (2015) 405e410.