Targeted opening of the blood-brain barrier by proteins released outside cells

Primary cerebral microvascular endothelial cell cultures were prepared from female and male C57bl/6 mouse brains (Cell Biologics, Chicago, IL; catalog no. C57-6023) and grown in the recommended culture medium (Cell Biologics, catalog no. M1168), split at a ratio of 1:2, and used at passage 2. Experimental series were designed using different batches of cells in order to provide biological replicates.

Mouse brain endothelial cell barrier function was determined by measuring the cell-cell adhesion barrier resistance to electric current using an electric cell-substrate impedance sensor (ECIS) system (Model ZƟ; Applied BioPhysics Inc., Troy, NY) as previously described [24]. Briefly, 250 μL of cell suspension (5 × 10⁵ cells per mL) were seeded to each well of two 8-well-ECIS cultureware arrays (8W1E+) following the manufacturer’s instructions. Resistance was recorded in real time with the following settings: alternating current (1 V), frequency (4 kHz) at 7-s intervals. Data were normalized to baseline measurements just prior to the onset of treatment (t = 0) and readings were acquired continuously for 24 h. Both data acquisition and processing were performed using the ECIS data analysis software supplied by Applied Biophysics. Treatment with ZVAD-fmk (100 μM), a cell permeable broad caspase inhibitor (Tocris Bioscience; catalog no. 2163/1), oleylamine (250 μM; Sigma-Millipore; catalog no. 909831) and cholesterol sulfate (250 μM; Cayman; catalog no.15106) was performed for 1 h prior to adding vehicle or histones. The data were represented as normalized resistance, which is the resistance measured after treatment over the resistance acquired before treatment introduction.

Since every cell type has its characteristic adhesion and growth, the ability of mouse brain endothelial cells to form a tight monolayer was assessed using ECIS. In order to confirm the formation of a mature BBB, cells were grown until a constant baseline resistance was obtained. Resistance values increased over time (growth phase) until they reached to stable transendothelial electrical resistance (TEER) values (plateau phase) indicating that cells formed matured cell-cell junctions approximately 5 days (∼ 125 h) later. Thus, cells were used for different in vitro experimental series (ECIS assays, transwell permeability, or immunofluorescence) 5 days after being plated. All TEER experiments were done in triplicate with at least 3 independent biological replicates. Data show mean ± S.E.M.

Endothelial monolayer permeability experiments were conducted using brain endothelial cells grown at a density of 2 × 10⁵ cells on 0.33 cm² gelatin-coated inserts of 0.4 μm pore size (Corning Inc., Tewksbury, MA; catalog no. 3401). Endothelial cells were cultured on the inserts for 5 days to allow the cells to form a monolayer and mature TJs (as previously assessed by TEER measurements, see below). On the day of the experiment, vehicle (saline) or histones (10, 25, 50, and 100 μg/mL) were added to the top chambers followed by NaFl or Texas Red 3-kDa dextran to a final concentration of 1 mg/mL. The diffuse flux of the tracers was measured by sampling media (100 μL) from the bottom chamber at 3 h after adding the molecular tracers and treatments. The concentration of each tracer diffusing from the top to the bottom chamber was determined using a SpectraMax M3 plate reader. The amount of dye was calculated by extrapolation from a standard curve of known NaFl (with excitation at 490 nm and emission at 520 nm) concentrations using linear regression. Apparent permeability coefficients (P_app) of mouse brain endothelial monolayers were calculated as follows:

P_app = [C]/t × 1/a × V/[L], where [C] is the concentration of the tracer in the bottom chamber, t is time (in seconds), a is the cross-sectional area of the membrane (0.33 cm²), V is the volume of the bottom chamber (670 μL), and [L] is the initial tracer concentration in the side of the chamber loaded with the tracer (top chamber). Permeability coefficient was reported in units of cm s^–1. Statistical analysis (Kruskal–Wallis and Dunnett’s post-hoc test) was performed using GraphPad Prism 8.0 software (San Diego, CA).

In a different set of experiments, the baseline flux of NaFl, 1-kDa, and Texas Red 3-kDa dextran was determined by taking aliquots from the bottom chamber at different time points and replacing them with equal amount of fresh warmed media. Flux was calculated as the slope of the best-fit line of accumulated tracer (μg/cm²) versus time (min), using the least-squares method. The amount of tracer accumulated in the bottom of the chamber increased linearly with time (steady-state phase). The extrapolation of the linear portion of the curve to the x-axis yielded the lag time (time required to reach steady-state).

Brain endothelial cell monolayers plated on Lab-Tek® chamber slides (Thermofisher Scientific; catalog no. 1777399PK) were treated with vehicle or histones, washed in PBS containing 2 mM CaCl₂ and 2 mM MgCl₂, and fixed in 2% paraformaldehyde (PFA) in PBS for 20 min. Cells were rinsed in PBS, permeabilized for 10 min (in 0.1% Triton-X/PBS), and blocked (10% normal goat serum in 3% BSA/PBS) for 1 h at room temperature. Slides were incubated with primary antibodies ZO-1 (1:100; Thermofisher; catalog no. 40-2200) and claudin-5 (1:50; Thermofisher; catalog no. 35-2500) overnight at 4 °C, washed 7 times, and incubated with appropriate secondary Alexa Fluor® antibodies (diluted at 1:500). After washing, slides were then cover slipped using ProLong™ diamond antifade mountant with DAPI and confocal microscopy imaging was performed using a Leica DMi8 STED confocal microscope. Z-stack of images were acquired using a HC PLAPO 63×/1.4 NA objective. Images were reconstructed using Leica Application Suite X software (LAS X).

Cytotoxicity was evaluated with propidium iodide (PI; Invitrogen; catalog no. P3566) as a cell viability indicator for dead (PI permeable) and viable (PI impermeable) cells. In brief, mouse brain endothelial cells were grown to 90% confluence in 6-well plates and treated with vehicle (saline) or histones (10, 25, 50, 100, and 250 μg/mL) in culture media. After 3 h of incubation, the medium was removed to a fresh tube, and the remaining cells were trypsinized and transferred to the same tube. The cells were pelleted and resuspended in 1 mL PBS containing 1 mg/mL PI. Events (5 × 10⁴) were acquired on a BD LSR II flow cytometer (BD Biosciences, San Jose, CA), and data were analyzed using FlowJo 7.6.4. software. Cell survival rates were normalized by vehicle-treated cells (designated as 100%).

Following euthanasia, the brain was carefully removed and placed in ice-cold phosphate-buffered saline solution (PSS) containing (in mM): 119 NaCl, 4.7 KCl, 20 NaHCO₃, 1.1 KH₂PO₄, 1.2 MgSO₄, 1.6 CaCl₂, and 10 glucose (pH 7.4). Posterior cerebral arteries isolated from saline- or histone-injected mice were carefully dissected out of surrounding tissue and cut into ~ 2 mm segments. Arterial segments were cannulated in an arteriograph chamber (Living Systems Instrumentation, St Albans, VT) and superfused with warm (37 °C) PSS. Before each experiment, arteries (internal diameter of ~ 100 μm when pressurized at 10 mm Hg) were allowed to equilibrate for a period of 10 min at 10 mmHg intravascular pressure and contractile responsiveness was assessed by exposure to isotonic hyperpolarizing solution (60 mM KCl-PSS made by equimolar substitution of KCl for NaCl in PSS). Following equilibration, intravascular pressure was slowly raised from 10 to 60 mmHg (in vivo pressure). Pressurized arteries were mounted on an inverted light microscope equipped with a CCD camera and edge-detection software for continuous monitoring of internal diameter (IonOptix, Milton, MA). Only cerebral arteries that exhibited myogenic tone were used. Maximal diameter was obtained at the end of each experiment by superfusing Ca²⁺-free PSS containing a calcium channel blocker (100 μM, diltiazem). Vascular (myogenic) tone was measured in pressurized vessels in vitro as the difference in vessel diameter under active conditions (presence of calcium in PSS) vs. passive relaxed conditions (calcium-free PSS) obtained at the same intravascular pressure and calculated as a percentage of the Ca²⁺-free diameter to normalize for differences in diameter between arteries. Endothelial cell function was determined by measuring the dilatory response to activation of endothelial SK/IK channels with NS309 (Tocris Biosciences; catalog no. 3895) [25, 26].

All studies were performed in accordance with the University of South Florida Institutional Animal Care and Use Committee based on the National Institute of Health guidelines for care and usage of laboratory animals. Female and male C57bl/6 mice (12–16 weeks of age) were bred and housed in cages on a 12-h light cycle with ad libitum access to water and a standard laboratory diet. Mice were anesthetized with isoflurane (2.5%) and then administered a single intravenous injection, through the retro-orbital venous sinus, of histones at 45 mg/Kg (highest non-lethal dose of histones that corresponds to circulating histone levels of ~ 80 μg/mL [27]) or the same volume of sterile saline. Mixture of unfractionated histones (consisting in 7.5% H1, 20.8% H2A, 32.5% H2B, 10.2% H3, and 28.9% H4 as previously reported [26]) isolated from calf thymus was used in this study (dissolved in sterile saline at 10 mg/mL; molecular weights of histones are 11- to 21-kDa depending on the fraction; Roche® distributed by Millipore-Sigma; catalog no. 10223565001). Mice were euthanized at 24 h, 3, 7, or 14 days after histone or saline injection depending on the experimental series as indicated in the results and figure legends.

The paracellular permeability of BBB to different molecular size tracers was determined by ratiometric assays of amount of dye/mg of tissue. Two groups of mice (saline or histones) were intravenously injected through the retro-orbital venous sinus with sodium fluorescein (NaFl; 5 μL/g of a 100 mg/mL NaFl solution; 376 Da; Stokes radius 0.45 nm; Sigma-Millipore; catalog no. 6377), Alexa Fluor 555–cadaverine (5 μg/g of a 1 mg/mL solution; 1-kDa; Invitrogen; catalog no. A30677), Alexa Fluor 648–3-kDa dextran (5 μg/g of a 1 mg/mL solution; Invitrogen; catalog no. D34681↗), and 1-kDa infrared dye (IRdye; 5 μg/g of a 1 mg/mL solution; Sigma-Millipore; catalog no. SCJ4600056) at 24 h post-saline or histones injection. Tracers were allowed to distribute in vivo for 30 min (NaFl) or 1 h (Alexa Fluor 555–cadaverine, 3-kDa dextran and 1-kDa cadaverine) in freely moving mice. Then, mice were anesthetized with urethane (1.75 mg/Kg; intraperitoneally) and transcardially perfused with ice-cold 0.1 M phosphate buffer (PBS; pH 7.4). After dissection, brains were harvested, weighed, and homogenized with 20% trichloroacetic acid and then diluted with equal volumes of borate buffer 0.05 M (for NaFl) or 1% Triton X-100 in PBS (for 1-kDa cadaverine and 3-kDa dextran). Tissue homogenates were centrifuged at 12,000×g for 20 min at 4 °C and proteins were precipitated with ethanol. The fluorescence of a 100-μL aliquot of the brain samples along with 100 μL of blank or standards was measured on a fluorescence microplate reader (SpectraMax M3; Molecular Devices, Sunnyvale, CA). The concentrations of the samples were within the linear range of the standard curve. Amount of dye (ng) was normalized per brain weight (mg).

The site-specific disruption of the BBB was determined in a subset of brains isolated from animals injected with 1-kDa IRdye. Ventral and dorsal brains were scanned by an ex vivo near-infrared (NIR) imaging system (Odyssey CLx; Li-Cor, Lincoln, NE) using 700- and 800-nm excitation. Brains were excised and sectioned into 1 mm slices from rostral to caudal using a coronal Adult Mouse Brain Slicer Matrix (Leica). Subregional analysis of tracer distribution was measured in coronal sections by measuring mean fluorescence intensity of nine coronal sections covering + 3 to − 5 mm from anterior to posterior relative to bregma. Mean fluorescence intensity for each coronal section was plotted against the distance to bregma (mm) to evaluate the progression of the leakage throughout the brain.

Cortex and hippocampus from saline- and histone-injected mice were macroscopically dissected, and all visible white matter was discarded. Tissue was homogenized in 1 mL RIPA lysis buffer (Millipore; catalog no. 20-188) containing protease and phosphatase inhibitors (Roche; catalog no. 11697498001 and 04906845001). The supernatant of a 12,000×g spin was used for immunoblotting. Homogenates containing 20 μg of protein (Pierce™ BCA kit) were run on a 4–20% Tris-Glycine gel and transferred onto a nitrocellulose membrane (BioRad). The membrane was blocked in Li-Cor blocking buffer for 1 h at room temperature, probed with primary antibodies against occludin (1:500; BD Biosciences, catalog no. 611091), ZO-1 (1:500; Thermofisher; catalog no. 40-2200), VE-cadherin (1:1000; Abcam; catalog no. ab205336), GFAP (1:1000; Sigma-Millipore; Catalog no. MAB360), iBA1 (1:500; Abcam; catalog no. ab178846), and beta-actin (1:1000; Li-Cor; catalog no. 926-42212 and 926-42210). Membranes were washed with PBS-T (0.1% Tween 20) and probed with Li-Cor IRDye secondary antibodies in blocking buffer containing 0.1% Tween-20 + 0.05% SDS for 1 h at room temperature. The washed membranes were then imaged on a Li-Cor Odyssey CLx imaging system and blots were analyzed using the Image Studio Lite 5.2.5 software. Relative protein abundance was obtained by densitometry analysis. The signal intensity of all protein bands was normalized to the beta-actin loading control band (relative abundance).

Mice were anesthetized at 24 h after saline- or histone injection and fixed by transcardial perfusion at an intravascular pressure of 60 mmHg with a solution containing 2% glutaraldehyde in 0.1 M sodium cacodylate buffer, pH 7.4 (Electron Microscopy Sciences; catalog no. 16536-15). Tissue samples from the relevant areas of the brain (cerebral cortex and hippocampus) were dissected from intact brains and post-fixed in 2% glutaraldehyde in 0.1 M sodium cacodylate for an additional 24 h at 4 °C. Samples were rinsed in 0.1 M sodium cacodylate (Electron microscopy sciences; catalog no. 11652) for 6 h and subsequently immersed in 2% aqueous osmium tetroxide with 1.5% potassium ferrocyanide buffer for 4 h at room temperature. After 3 rinses in 0.1 M cacodylate buffer for a total time of 15 min, dehydration steps followed using a series of graded ethanol (35, 50, 70, 95, and 100%) and a final change into acetone prior to resin infiltration (Embed810 eponate). Infiltration steps included immersion of samples in intermediate solutions of 1:1 and 2:1 resin to acetone for 1 h each, prior to final immersion in pure fresh resin overnight followed by embedding in silicon flat molds and polymerization in a 60 °C oven. Ultrathin (90–100 nm) sections were obtained using an ultracut microtome without additional contrast enhancement. A series of non-overlapping images capturing the entire capillary circumference were obtained with a JEOL1400 transmission electron microscope equipped with a side mounted Gatan Orius digital camera.

Anesthetized mice were transcardially perfused through the left ventricle with ice-cold PBS. Mouse brains were cut through the hippocampus (~ 1.46 mm caudal to bregma), embedded in Tissue-Tek® O.C.T. compound (Sakura Finetek USA Inc., Torrance, CA), and snap-frozen (“fresh frozen tissue”). Coronal sections were sectioned at 20 μm on a Leica CM1950 cryostat and stored at − 80 °C until use. Cryosections were thawed and immediately fixed in either ice-cold methanol (claudin-5, ZO-1) or 2% PFA (VE-cadherin) for 15 min.

For glial fibrillary acidic protein (GFAP) and ionized-calcium binding adapter molecule-1 (iBa1) immunostaining, animals were transcardially perfused via the left ventricle with 4% PFA in PBS. After 1 day of post-fixation in the same fixative, brains were embedded in paraffin and then sectioned (4 μm) using a microtome.

Mouse brain sections were permeabilized and blocked for 1 h at room temperature in blocking buffer (0.3% Triton X-100 and 10% normal goat serum in 3% BSA/PBS) and incubated overnight at 4 °C with the following primary antibodies: GFAP (1:500; Genetex; catalog no. GTX108711), iBA1 (1:250; Abcam; catalog no. ab178847), claudin-5 (1:50; Thermofisher; catalog no. 35-2500), VE-cadherin/CD144 (1:100; BD Pharmingen; catalog no. 555289), and ZO-1 (1:100; Thermofisher; catalog no. 40-2200) diluted in blocking buffer at constant agitation. To visualize brain microvessels, sections were incubated with Dylight 488-conjugated Lycopersicum esculentum lectin (1:100; Vector laboratories, Burlingame, CA; catalog no. DL-1174) during the primary antibody incubation. After the washing steps, sections were incubated with fluorescent secondary antibodies (1:500; Alexa Fluor® dyes; Invitrogen) diluted in blocking buffer for 1 h at room temperature. Coronal brain sections were subsequently washed for 1 h with washing buffer, mounted in ProLong™ diamond antifade mountant with DAPI (Life Technologies; catalog no. P36966↗) and imaged using a Leica DMi8 STED confocal microscope using a 20× HC PLAPO, N.A. 0.75 objective (Leica Microsystems, Rueil-Malmaison, France). Images were acquired as z-stack projections and pseudo-coloring was performed to better visualize brain microvessels using Leica Application Suite X software (LAS X). Images of the confocal microscopy observations (512 by 512 pixels) were acquired sequentially, between stacks, to eliminate spectral crosstalk from the different channels. For each brain slice, one random field per animal was selected from the cortex (40x magnification for TJ and AJ proteins) and hippocampus (× 20 magnification). Six to nine separate vessels from each image were analyzed, and the data averaged and expressed as percentage. The area of ZO-1, claudin-5, or VE-cadherin-positive staining was normalized to the total area of lectin-positive microvessels using the Fiji hand free drawing tool to trace the outline of individual microvessels and the FIJI area measurement tool. The n value represents the number of animals in the group, with at least 5–6 mice analyzed. Overlays and maximum-intensity projections of 20 single slices from z-stack images were reconstructed post-acquisition by using ImageJ package FIJI, version 2.0 (fiji.sc/Fiji) (National Institutes of Health, Bethesda, MD) to obtain the presented images. Optimal antibody concentrations were first assessed by titration to determine working concentrations used for experiments. Gain, digital offset, and laser intensity were kept standardized.

The brain water content was determined from differences in the wet and dry weight of the whole brain as follows: [(wet weight − dry weight)/wet weight] ×100%.

Histones were conjugated with a fluorescent NHS ester dye following the manufacturer’s instructions. Briefly, 10 mM stock solution of Alexa Fluor 594–NHS succinimidyl ester (Invitrogen; catalog no. A20009) prepared in anhydrous DMSO was added to a 10 mg/mL solution of calf thymus histones prepared in 50 mM borate buffer (pH 9.0) while gently stirring at room temperature. The reaction was quenched by adding 100 μL of 50 mM Tris-glycine after 1 h. The protein-dye conjugate was purified to remove excess of unbound dye using centrifugal Amicon® ultra-15 filters with a 3-kDa molecular weight cutoff (Sigma-Millipore; catalog no. UFC900308). The protein concentration was measured using a Pierce™ BCA protein assay kit (Pierce/Thermo Fisher; catalog no. 23227). Finally, the histone-dye conjugate was aliquoted and stored at − 20 °C.

Data are presented as means ± standard error of the mean (S.E.M.). For animal experiments, female and male mice were randomly selected and used as biological replicates. For arteriography experiments, one artery per animal was used. Normality of the data was tested using the Shapiro–Wilk test for normality. For data that were not normally distributed (nonparametric data), the Mann–Whitney test (two groups, one variable) or Kruskal–Wallis test followed by Dunn’s correction (> 2 variables) was used. Data with normal distribution were analyzed by a two-tailed unpaired Student’s t-test (one variable) or one-way ANOVA with Dunnett’s or Tukey’s correction (> 2 variables). When two factors were analyzed, data were analyzed using two-way ANOVA with Tukey’s correction. Statistical test used to analyze the data is described in the figure legends. In all tests, a 95% confidence interval was used, for which P < 0.05 was considered a significant difference. Statistical analyses and graphs were performed in Prism 7.0 software (GraphPad, La Jolla, CA).

High electrical resistance is a defining characteristic of the BBB and is commonly used to evaluate endothelial barrier integrity (“endothelial tightness”) [24, 28, 29]. Measurements of TEER and transwell permeability assays were used to assess functional endothelial barrier integrity alongside immunocytochemistry to test whether changes in endothelial permeability were accompanied by a reorganization of TJs. First, we used ECIS to monitor barrier formation of mouse brain cells (Supplemental Fig. 1). The tightness of the cell–cell contacts of primary mouse brain endothelial cells obtained as raw TEER readings was ∼ 2500 Ω (35–50 Ωcm²; Supplemental Fig. 1). Measurements showed that TEER values plateaued by day 5 indicating the formation of a tight endothelial barrier. High TEER values were maintained for another 4 days after the permeability barrier was achieved (Supplemental Fig. 1). NaFl and 1-KDa tracers were used to verify the impact of the molecular size on diffusion across brain endothelial cell monolayers. The accumulation of NaF and 1-KDa tracers increased with time showing higher flux and shorter lag time for NaF compared to the larger size 1-kDa tracer (Supplementary Fig. 2). However, Texas Red 3-kDa dextran did not permeate across the brain endothelial cell monolayer (data not shown). These results indicate that brain endothelial cells exhibit selective permeability to small molecular size tracers.

Next, we evaluated whether histones affect the barrier integrity of endothelial cells by measuring TEER. Histones at a concentration of 100 μg/mL (~ 7 μM) induced a 35% reduction in TEER showing a peak TEER value at 3 h after the addition of histones (Fig. 1a). Furthermore, we used transwell permeability assays to assess the permeability apparent (P_app) of the brain endothelial cell monolayers to NaFl following histone treatment. Only the highest concentration of histones (100 μg/mL) yielded a statistically significant increase in permeability of brain endothelial cells to NaFl (Fig. 1b).

Permeability data were reinforced by immunostaining of junctional proteins ZO-1 and claudin-5, which revealed a reorganization of TJs in cells exposed to histones. Cells treated with histones at 100 μg/mL showed discontinuous and more diffuse staining of ZO-1 and claudin-5 compared to the distinctly pattern localized at the cell borders in cells treated with vehicle (Fig. 1c, d).

Previous studies have shown that calf thymus histones are toxic to culture cells including pulmonary endothelial cells [4, 5, 30], intestinal epithelial cells [30], smooth muscle cells [31, 32], and freshly isolated erythrocytes [33]. Next, we investigated whether histones induce cell death in brain endothelial cells. Cytotoxic effects of histones were evaluated by flow cytometry using PI. Histones (10, 25, 50, 100, and 250 μg/mL) added to brain endothelial cells for 3 h did not affect cell death compared to vehicle-treated cells (Fig. 2a). Moreover, histone-induced decrease in TEER was not inhibited by the pan caspase inhibitor ZVAD-fmk (100 μM), suggesting that the drop in TEER was not the result of caspase-dependent apoptosis (Fig. 2b). Additionally, the extent of endothelial cytotoxicity in vivo was also determined. The effect of circulating histones on the native cerebrovascular endothelium was studied on intact mouse pressurized-cerebral arteries isolated from saline- and histone-injected mice. Circulating histones did not alter endothelial reactivity of myogenically active mouse cerebral arteries as evidenced by intact endothelial-dependent dilations to the vasodilator NS309 (Fig. 2c, d). These data demonstrate that histones are not toxic to endothelial cells of the cerebral vasculature.

To assess how circulating histones affect the paracellular permeability of the BBB, we used a histone injection mouse model in which a single non-lethal dose of histone mixture is administrated intravenously. We studied brain uptake of NaFl in mice at 24 h after saline or histone injection. In histone-treated mice, we found a regional variation of BBB permeability where histones significantly increased the uptake of NaFl in the hippocampus compared to cerebral cortex (Fig. 3a).

Next, we studied the size selectivity of the BBB opening induced by histones using different molecular size tracers. The uptake of NaFl and 1-kDa cadaverine showed higher total brain accumulation in histone-treated mice compared with the control group, further suggesting that histones make the brain vasculature leaky (Fig. 3b, left; Supplemental Fig. 3a). However, 3-kDa dextran did not show any signs of extravasation into the brain parenchyma (Fig. 3b, right; Supplemental Fig. 3b,c), indicating that histones mediate a size-selective diffusion of solutes across the paracellular pathway of the BBB.

In a separate set of experiments, we injected 1-kDa IRDye and quantified its accumulation in the brain using ex vivo NIR fluorescence imaging. Tracer distribution within the brain was analyzed in serial coronal sections at the level of + 3 mm anterior and − 5 mm posterior to the bregma. Representation of tracer accumulation on each coronal section versus distance from bregma showed the progression of the leakage as distribution of the tracer within the brain from caudal to rostral (Fig. 3c). Brain water content was not different between groups (77 ± 0.2% and 76 ± 0.1%, saline- and histone-injected mice, respectively; n = 6 each). Collectively, these results indicate that circulating extracellular histones increase the paracellular permeability of the BBB in a size- and region-selective manner.

To test the duration of BBB disruption (to NaFl), four groups of mice were used at 1, 3, 7, and 14 days after saline or histone injection. The uptake of NaFl increased significantly on days 1 through 7 after starting histone injection (Fig. 3d). Importantly, the peak time of BBB disruption was reached at 3 days, whereas the reversibility of the BBB opening was observed on day 14 after histone injection (Fig. 3d).

The BBB is characterized by tightly sealed endothelial cells that restrict the paracellular transport of molecules into the brain parenchyma through the expression of TJs and AJs [18]. Next, we investigated whether histone-induced BBB disruption correlates with a decreased expression of TJ proteins (occludin, ZO-1) and an AJ protein (VE-cadherin). As shown in Fig. 4a, b, protein expression levels of ZO-1, occludin, and VE-cadherin in the hippocampus were reduced by 65%, 42%, and 46%, respectively, in the histone-treated group at 24 h post-treatment compared with the saline group (all P < 0.01). However, there was no significant attenuation of the protein levels of ZO-1, occludin, and VE-cadherin in the histone group in the cerebral cortex (Fig. 4. For gel source data, see Supplementary Fig. 4).

Furthermore, ultrastructural changes in TJ integrity at the endothelial cell-cell contact regions was investigated by electron microscopy. As shown in Fig. 4c, the cell–cell junctions between endothelial cells remained intact in the saline group and appeared as sealing intercellular clefts in brain capillaries from both cerebral cortex and hippocampus. The basal membrane was also continuous and integrated. However, as shown in Fig. 4d, circulating histones affected TJ structure in the hippocampus, while no obvious effect of histones on the structure of TJs was observed in the cerebral cortex.

In addition, the immunostaining of TJ proteins (claudin-5 and ZO-1) and the AJ protein (VE-cadherin) was also assessed. Consistent with BBB breakdown, unlike the cerebral cortex, the microvessels in the hippocampus from histones-treated mice showed a reduction in length of claudin-5 and ZO-1, and VE-cadherin of 58%, 43%, and 37.5%, respectively, compared to those from saline-treated mice (Fig. 5).

Astrocyte and microglia activation may occur concomitantly in response to direct insults inflicted to the brain or during systemic inflammation through a range of toxic circulating molecules that can access the brain microenvironment through a leaky BBB [34, 35]. The next series of experiments aimed to better understand whether histone-induced BBB disruption can result in activation of astrocytes and/or microglia in vivo. We used Western blotting to quantify the protein expression of a molecular marker of astrocytes (GFAP) and microglia (iBA1) in both cerebral cortex and hippocampus. We found no difference in levels of GFAP or iBA1 expression between groups (Fig. 6a, c. For gel source data, see Supplementary Figs. 5 and 6). As expected, we observed that GFAP staining was more conspicuous in the white matter than in the gray matter while iBA1 was uniformly distributed. Moreover, immunofluorescence staining showed no differences in GFAP- or iBA1-positive astrocytes or microglia, respectively at 1 day after histone exposure in either the hippocampus or cerebral cortex (Fig. 6b, d. For gel source data, see Supplementary Figs. 5 and 6). Together, these results indicate that histone-induced BBB dysfunction is not dependent on astrocytes or microglia.

Based on previous reports and given the anionic nature of the plasma membrane, positive charges on histone tails may facilitate histone anchorage to cell membranes [5, 36]. Next, we tested the effect of exogenous manipulation of membrane cationicity on barrier function integrity [31]. We investigated the effect of cholesterol sulfate and oleylamine to either decrease or increase, respectively, membrane cationicity on histone-induced barrier disruption. Cholesterol sulfate enhanced histone-mediated decrease in TEER suggesting an increased histone-membrane interaction. The addition of positive charges to the plasma membrane with oleylamine inhibited histone-mediated decrease in TEER (Fig. 7b, c). Barrier integrity measurements were further confirmed with claudin-5 immunostaining. In the presence of cholesterol sulfate, histone-treated cells showed decreased claudin-5 immunostaining and increased histone molecules attached to cell membranes. Oleylamine did not alter claudin-5 immunostaining and showed a lack of histone-membrane interaction (Fig. 7c). These results indicate that histone-induced barrier disruption is due to a histone-membrane interaction that can be modulated by changes in membrane surface charges.

The data that support the findings of this study are available from the corresponding author upon reasonable request.