Trafficking of immune cells across the blood-brain barrier is modulated by neurofibrillary pathology in tauopathies

Tauopathies represent a heterogeneous group of neurodegenerative disorders characterized by abnormal deposition of the hyperphosphorylated microtubule-associated protein tau. Chronic neuroinflammation in tauopathies is driven by glial cells that potentially trigger the disruption of the blood-brain barrier (BBB). Pro-inflammatory signaling molecules such as cytokines, chemokines and adhesion molecules produced by glial cells, neurons and endothelial cells, in general, cooperate to determine the integrity of BBB by influencing vascular permeability, enhancing migration of immune cells and altering transport systems. We considered the effect of tau about vascular permeability of peripheral blood cells in vitro and in vivo using primary rat BBB model and transgenic rat model expressing misfolded truncated protein tau. Immunohistochemistry, electron microscopy and transcriptomic analysis were employed to characterize the structural and functional changes in BBB manifested by neurofibrillary pathology in a transgenic model. Our results show that misfolded protein tau ultimately modifies the endothelial properties of BBB, facilitating blood-to-brain cell transmigration. Our results suggest that the increased diapedesis of peripheral cells across the BBB, in response to tau protein, could be mediated by the increased expression of endothelial signaling molecules, namely ICAM-1, VCAM-1, and selectins. We suggest that the compensation of BBB in the diseased brain represents a crucial factor in neurodegeneration of human tauopathies.


Introduction
Neuroinflammation manifests before a significant loss of neural tissue in the process of neurodegeneration, suggesting that neuroinflammation promotes the progression of pathogenesis in a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 resulting in the transmigration of peripheral blood-borne cells into brain areas affected by neurofibrillary pathology in tauopathies. In this study, we used an in vitro BBB model and transgenic rat model for tauopathy (line SHR-72) to investigate the role of tau protein in inducing vascular changes in the BBB, and the mechanism involved in the tau-mediated deregulation of BBB dynamics.

Animals and brain tissues
All animals used in this study were from the in-house breeding colony (SPF like, monitored according to the Federation of European Laboratory Animal Science Associations). In this study, we used transgenic rat model (line SHR-72, 6 month old) stably expressing human tau protein truncated at amino acids 151-391 (aa 151-391/4R). The transgenic rats develop progressive age-dependent neurofibrillary degeneration in the brainstem [28]. All animals were housed under standard laboratory conditions with free access to water and food and were kept under diurnal lighting conditions. All experiments on animals were carried out according to the institutional animal care guide-lines conforming to international standards (Directive 2010/63/EU) and were approved by the State Veterinary and Food Committee of Slovak Republic (RO-1101/14-221C). In survival surgery experiments each animal was stored in a separate cage and was monitored daily (food consumption, water consumption, and wound healing). Humane endpoints based on body weight loss and body condition scoring were in place. For terminal experiments, animals were euthanized with CO2. No unexplained mortality occurred in these studies. Human Alzheimer's brain tissue was procured from Department of Psychiatry, University of Geneva School of Medicine (Geneva, Switzerland; University of Geneva brain collection; Dr. Enikö Kövari)-Female, Neuropathology: Alzheimer´s disease, Braak stage 5 [29]. All animal experiments and also experiments with human tissue were monitored by the Institutional ethics committee (Ethics Committee of Institute of Neuroimmunology)

Expression and purification of recombinant tau protein
Human truncated tau (aa 151-391/4R) was expressed in Escherichia coli strain BL21(DE3) from a pET-17 expression vector and purified from bacterial lysates by size-exclusion chromatography in phosphate buffer saline (PBS; 137 mM NaCl, 2.7 mM KCl, 10 mM Na 2 HPO 4 , 2 mM KH 2 PO 4 , pH 7.4). Purified tau protein was stored under argon in working aliquots at -70˚C. The purity of tau protein was verified by SDS gel electrophoresis and western blot analysis with monoclonal DC25 antibody (epitope aa 347-354 of the longest human tau isoform 2N4R; AXON Neuroscience R&D Services SE, Bratislava, Slovakia).

Tau filament assembly and detection of tau oligomers
In vitro oligomerization of recombinant truncated tau protein was carried out using heparin (sodium salt from porcine intestinal mucosa, Grade I-A, cell culture tested, �140 USP units/ mg, powder, H3149-100KU, Mw approx. 6000 g/mol, Sigma-Aldrich, St. Louis, MO) as an inducer at a final concentration of 25 μM in PBS. The reaction was performed overnight (for at least 12 hours) at 37˚C. After incubation, tau oligomers were collected by centrifugation at 100 000 x g for 1 hour at room temperature, and the pellet was re-suspended in PBS and sonicated for 5 seconds at 20% power output using MS72 probe of Bandelin Sonopuls Sonifier (Bandelin, Berlin, Germany). 1 μM aliquots of aggregated protein were stored at -70˚C.
Oligomerization of tau protein was verified by gradient SDS electrophoresis (5-20% gel) and electron microscopy.

Isolation of sarkosyl-insoluble tau
Sarkosyl-insoluble tau from AD brain and transgenic rat brain were isolated as previously described [30]. Briefly, 1-5 g of gray matter from human AD brain-enriched in paired helical filament tau (PHF-tau) or 200 mg of the brainstem from rat brain (n = 2) were dissected, cleaned of blood vessels and meninges and used. After sarkosyl extraction and ultracentrifugation, pellets containing PHF-tau were resuspended in 2 ml PBS and sonicated, and boiled for 5 min. Then 0.5 ml aliquots were loaded onto a stepwise 1-2.5 M discontinuous sucrose gradient and spun for 16 hours at 4˚C in a Beckman ultracentrifuge using SW 32 Ti rotor (Beckman Coulter, Inc., Brea, CA, USA) at 175 000 x g. The band between 1.5 and 1.75 M sucrose interface was collected and centrifuged at 100 000 x g. The purity of the preparation was assessed by western blot analysis and electron microscopy.

Transmission electron microscopy
Animals under tiletamine-zolazepam/xylazine anesthesia (4/2mg/kg) (CT/TG-2/2) were perfused with PBS mixed with heparin (100 000 IU/l PBS, Sigma-Aldrich, St. Louis, MO) for 1 min and subsequently perfused with 4% glutaraldehyde (Sigma-Aldrich, St. Louis, MO) in 0.1 M cacodylate buffer (Sigma-Aldrich, St. Louis, MO). Brains were removed and fixed in the same buffer overnight. Blocks of brain tissue were post-fixed in 40 mM osmium tetroxide (Sigma-Aldrich, St. Louis, MO) in cacodylate buffer for 1 h. After rinsing in cacodylate buffer and dehydration in ethanol, samples were embedded in araldite resin. Ultrathin sections (60 nm thick) were cut using Leica EM UC6 ultramicrotome (Pragolab s.r.o, Czech Republic) and stained with uranyl acetate and lead citrate and analyzed by FEI Morgagni 268 electron microscope (Oregon, USA). For morphological examination of in vitro oligomerized tau and PHFtau by electron microscopy the samples collected by ultracentrifugation were dissolved in pure water and placed on carbon-coated 400 mesh copper grids (Christine Gröpl, Austria) for 2 min. Grids were washed with pure water and negatively stained with 2% uranyl acetate for 1 min. The stained grids were immediately analyzed by FEI Morgagni 268 electron microscope.

Isolation and cultivation of rat primary glial culture
Rat mixed glial culture was prepared from cerebral cortices of 1-2 day old Sprague Dawley rats (n = 8/isolation). The animals were euthanized by CO 2, and the cerebral cortices were dissected, stripped of the meninges and mechanically dissociated by repeated pipetting followed by passage through a 20 μm nylon mesh (BD Falcon, Franklin Lakes, USA). Cells were plated on 6-well plates pre-coated with poly-L-lysine (10 μg/ml, Sigma-Aldrich, St. Louis, MO) and cultivated in DMEM medium (PAA laboratories GmbH, Germany) containing 10% fetal calf serum (FCS, Thermo Fisher Scientific, Waltham, Massachusetts, USA) and 2 mM L-glutamine (Life Technologies Invitrogen, Carlsbad, CA) at 37˚C, 5% CO 2 in a water-saturated atmosphere.

Isolation and cultivation of primary rat brain endothelial cellsin vitro BBB model
Isolation of primary brain endothelial cells was performed as previously described [31]. Briefly, rats (200-250 grams, 6 month old; n = 4/isolation) were euthanized by CO 2, and their brains were removed. Under sterile conditions, the brainstem and cerebellum were dissected, and white matter from the midbrain and the choroid plexus were removed. The cortical tissues were cleansed from meninges and were homogenized on ice in DMEM-F12 medium (PAA laboratories GmbH, Germany) with 0.1% bovine serum albumin (BSA, Sigma-Aldrich, St. Louis, MO). The homogenate was centrifuged at 800 x g for 10 min at 4˚C. The supernatant was aspirated and the pellet resuspended in pre-warmed digestion mix containing 1 mg/ml collagenase/dispase (Roche Diagnostics, Indianapolis, USA) and 10 μg/ml DNase I (Roche Diagnostics, Indianapolis, USA).
The homogenates were incubated with a pre-prepared digestion mix at 37˚C for 30 min with gentle shaking. The preparation was centrifuged at 800 x g for 10 min at 4˚C, and the pellets were resuspended in 20% BSA in the medium. The tissues were centrifuged at 1500 x g for 15 min at 4˚C to obtain pellet containing microvessels with a fraction of myelin and BSA on the top which was centrifuged again. The microvessels were pooled and re-suspended in prewarmed digestion mix and incubated for 15 min at 37˚C. The pellet was centrifuged at 800 x g for 10 min at 4˚C and washed with serum containing DMEM-F12 culture medium.

Endothelial permeability
The Transwell inserts (in a 12-well format, containing an endothelial monolayer or without cells) were transferred into 12-well plates containing 1.5 ml of Ringer-HEPES solution (150 mM NaCl, 5.2 mM KCl, 2.2 mM CaCl 2 , 0.2 mM MgCl 2 , 6 mM NaHCO 3 , 5 mM HEPES, 2.8 mM glucose; pH 7.4) per well. The cell culture medium was removed from the inserts, and 0.5 ml of Ringer-HEPES solution containing 10 μM Lucifer Yellow (LY) (Sigma-Aldrich, St. Louise, Missouri, United States) was added to the upper (luminal) compartment. All incubations were performed at 37˚C. After 20, 40 and 60 min, 200 μl aliquots from each lower compartment was quantified for fluorescence intensity (Fluoroscan Ascent FL, Labsystems; excitation wavelength: 428 nm; emission wavelength: 536 nm). The endothelial permeability coefficient (Pe) of LY was calculated in cm/s -1 . The permeability values of the inserts (PSf, for inserts with a coating only) and the inserts plus endothelium (PSt, for inserts with a coating and cells) were taken into consideration by applying the following equation: 1/PSe = 1/PSt − 1/ PSf. To obtain endothelial permeability coefficient (Pe, in cm/s -1 ), the permeability value corresponding to the endothelium alone was then divided by the insert's porous membrane surface area. The permeability value of LY in our BBB model was around 11 x 10 −6 cm/s -1 , below the acceptable maximum limit of 15 x 10 −6 cm/s -1 . The TEER values of our in vitro BBB model was determined by Ohm meter "EVOM" (World Precision Instrument, EVOM Sarasota, FL, USA). The TEER value of the model was approximately 300 ± 20 O cm 2 .

Isolation of rat peripheral blood monocyte-derived macrophages (PB-MoM)
Rat peripheral blood monocyte-derived macrophages were obtained from peripheral blood of healthy animals anesthetized by tiletamine-zolazepam/xylazine anesthesia (4/2mg/kg) (n = 2/ isolation). 4 ml of blood was diluted with sterile PBS and carefully layered on 8 ml of Ficoll-Paque Plus (GE Healthcare, UK) and centrifuged at 800 x g for 30 min at 25˚C. Using sterile Pasteur pipette, the cell layer was transferred into a clean centrifuge tube and re-suspended in PBS. The cells were centrifuged at 800 x g for an additional 15 min at 25˚C. Isolated cells were cultivated in RPMI 1640 medium (Thermo Fisher Scientific, Waltham, Massachusetts, USA) containing 20% FCS and 2 mM L-glutamine.

Transendothelial migration assay of PB-MoM across in vitro BBB model
Rat primary endothelial cells were seeded on the upper chamber of Transwell inserts with 1 μm pore size (Becton Dickinson, New Jersey, USA) in 12-well Costar plates (Corning Incorporated, New York, USA) and grown to confluence. Three weeks old astrocytes were cultivated as an adherent monolayer and incubated with oligomeric tau (at final concentration 1μM) or PHF-tau for 48 hours. After 48 hours, the culture media were removed from glial cells and added to the bottoms of 12-well Costar plates with endothelial cells. Subsequently, 1 x 10 5 of PB-MoM cells were loaded into the apical chamber of the inserts. After 24 hours, the PB-MoM that transmigrated into the basolateral chamber of inserts were stained with monoclonal mouse anti-rat CD45 antibody (1:300; Bio-Rad Laboratories, CA, USA) and examined by LSM 710 confocal microscope (Zeiss, Jena, Germany). In a control experiment, endothelial cells were cultivated with oligomeric tau for 48 hours. The oligomeric tau was added directly to the bottom chamber of in vitro BBB model. For quantification, the PB-MoM from five random fields per each insert was counted. To delineate the functional role of adhesion molecules ICAM-1 and VCAM-1 in PB-MoM binding, corresponding monoclonal antibodies (at final concentration 0.25 μg/cm 2 , all from Abcam, Cambridge, UK) were incubated with endothelial cell monolayers for 30 min prior to the addition of PB-MoM. PB-MoM that transmigrated into the basolateral chamber of inserts were stained with monoclonal mouse anti-rat CD45 antibody.

Isolation of brain microvessels
Animals (CT/TG -12/12) were euthanized by CO 2 overdose, and the brainstem was removed. The tissues were homogenized on ice in DMEM-F12 with 0.1% BSA. The homogenate was centrifuged at 800 x g for 10 min at 4˚C. To remove myelin, the pellets were resuspended in 25 ml of BSA (20% w/v), and centrifuge at 1500 x g for 20 min at 4˚C without brake. The pellets were resuspended and re-centrifuged under same conditions. The resulting pellets were resuspended in DMEM-F12 with 0.1% BSA and filtrated through 100 μm pore nylon filter (BD Falcon, Franklin Lakes, USA). The microvessels were collected by additional filtration through 20 μm nylon filters (BD Falcon, Franklin Lakes, USA). The microvessels were centrifuged at 800 x g for 10 min at 4˚C. Finally, the microvessels were lysed in RLT lysis buffer and stored.

RNA extraction and quantification of endothelial signaling molecules by quantitative real-time PCR
Rat primary endothelial cells on 6-well plates were briefly washed with 1 ml pre-warmed 1x PBS and lysed in RLT buffer (1 ml per well). Total RNA was isolated using RNeasy Mini Kit according to the manufacturer's recommendations (Qiagen, Hilden, Germany). The genomic DNA was removed by DNase I digestion during the RNA purification. RNA was eluted into 40 μl RNase-free water. The integrity of isolated total RNA samples was determined with an Agilent 2100 Bioanalyzer using an RNA 6000 Nano Labchip kit (Agilent Technologies, Waldbronn, Germany). For transcriptomic analysis, high-quality RNA samples were used (RNA integrity number = 8.0 to 9.5). Gene expression of endothelial cells signaling molecules was performed using Rat Endothelial cell biology PCR arrays profiling 84 genes (Qiagen, Germany). Briefly, total RNA was inversely transcribed into cDNA by RT2 first strand kit (Qiagen, Germany) and 40 ng of resulting cDNA was used as a template for each qPCR reaction. Components of 25 μl qPCR reaction were as follows: 12.5 μl 2x RT2 SYBRGreen/ROX mastermix; 11.7 μl RNase-free water and 0.8 μl of cDNA (50ng/μl). Cycling conditions included an initial 95˚C denaturation for 10 min, and 40 cycles of 95˚C for 15 s together with amplification at 60˚C for 1 min. PCR specificity was checked by melting curve analysis. Fold change of target genes was calculated using ddCt method with beta-actin (Actb) and ribosomal large protein P1 (Rplp1) as reference genes. Data in Table 1 and S1 Table represent the average of three different experiments.

Immunohistochemistry
Transgenic rats and age-matched controls (CT/TG-4/4) were deeply anesthetized with tiletamine-zolazepam/xylazine anesthesia (4/2mg/kg) and perfused intracardially with PBS. The brainstem was removed and embedded in cryostat embedding medium (Leica, Wetzlar, Germany) and frozen above the surface of liquid nitrogen. 10-μm-thick brain sections were cut on a cryomicrotome (Leica CM 1850, Leica, Wetzlar, Germany), affixed onto poly-L-lysine coated slides and left to dry at room temperature for 1 hour. Sections were fixed for 10 min in icecold acetone/ethanol solution and blocked for 60 min in blocking solution (DAKO, Mississauga, Ontario, Canada). Sections were incubated in primary antibodies: monoclonal mouse anti-rat CD11b (1:300; Bio-Rad Laboratories, CA, USA), monoclonal mouse anti-rat CD45

In vivo cell tracking
Animals under anesthesia induced by tiletamine-zolazepam/xylazine anesthesia (4/2mg/kg) (6 month old transgenic rats and age-matched non-transgenic controls, CT/TG-4/4) were injected by carboxyfluorescein diacetate tracer (CFDA, Invitrogen Life Technologies, Carlsbad, CA) in 200 μl dimethyl sulfoxide (DMSO, Sigma-Aldrich, St. Louis, MO) with 10 μl heparin (Sigma-Aldrich, St. Louis, MO). Heparin was used as a precaution to avoid clotting. We injected 0.2 mg of CFDA solution per animal directly into the spleen. After 48 and 72 hours, rats were deeply anesthetized with a ketamine-xylazine cocktail and transcardially perfused with PBS. The brain was removed and embedded in cryostat embedding medium (Leica, Wetzlar, Germany) and frozen above the surface of liquid nitrogen. Table 1. Inflammatory mediators associated with gene signaling in endothelial cells and brain capillaries exposed to tau proteins. Transcriptomic analysis of endothelial cells from in vitro BBB model and isolated capillaries from brainstem of transgenic rats (SHR72) and control animals. RT-PCR reactions were run in triplicate with Actb and Rplp1 used as the reference genes. Minimum fold change was set at � 2, � -2.

Data analysis and statistics
All experiments were repeated at least three times. Data are presented as mean ± standard deviation (SD). Statistical analysis was performed using Prism (Version 5.0, GraphPad, Inc., SanDiego, CA). Differences between means were analyzed using two-tailed Student´s t-test. For correlation, Pearson's correlation coefficient was used. Differences at p<0.05 were accepted as statistically significant. � p<0.05, �� p<0.01, ��� p<0.001 are used to denote statistical significance.

Image analysis (ImageJ)
Relative staining pattern and intensity of projections from transgenic and controls rats were visualized by confocal microscopy and evaluated. ImageJ (public domain ImageJ software) was used for the evaluation and quantification of immunohistochemistry slides. We quantified 10 slices from each transgenic and control brains. For semiquantitative analysis, the color pictures were converted to grayscale 8-bit TIFF file format and regions of interest were analyzed. The grayscale 8-bit images were converted to 1-bit images, on which the number of immunolabelled structures localized in the area of interest was measured. The average intensity/pixel values of each area were then calculated, and the average intensity/pixel values representing the background intensity were subtracted from those of the immunolabelled areas. Cumulative data from every image was subjected to statistically analysis.

Active infiltration of PB-MoM cells into the area with neurofibrillary pathology
In order to confirm that peripheral cells actively migrate into the areas affected by tau neurofibrillary pathology, we used the CFDA cell labeling. After 2-and 3-days post-injection, we quantified the numbers of CFDA-labeled cells in the brain of control and transgenic animals. After 48 hours, we did not detect CFDA-labelled cells in control animals; however, we observed the presence of these cells in transgenic rat brain, mainly in the brainstem and midbrain regions (Fig 4A and 4B, CT 0 vs. TG 14.5 ±3.695, p = 0.0028; n = 4), the areas with neurofibrillary pathology. After 72 hours, we rarely observed CFDA-labeled cells in control tissues; however, significantly higher levels of CFDA labeled cells was observed at 72 hours in transgenic rodent brains (Fig 4C and 4D, CT 1.833 ± 0.7923 vs. TG 40.5 ± 6.761, p = 0.0002; n = 4). Interestingly, the number of CFDA labeled cells increased after 72 hours when compared to 48 hours (Fig 4E, p = 0.0071). Double staining with pan-laminin demonstrated that CFDAlabeled cells were found in both brain parenchyma (Fig 4F, white arrows) and perivascular space ( Fig 4G). Moreover, further analysis revealed that CFDA positive cells colocalized with CD11b staining (Fig 5A and 5B), confirming the infiltration of CD11b cells in the brain, from the periphery. Interestingly, the infiltration of CFDA positive cells was observed in areas with increased immunostaining for ICAM-1 (Fig 5C and 5D). This suggests that PB-MoM cells infiltrated tissue with neurofibrillary pathology associated with inflammatory changes of endothelial cells.
Interestingly, the distribution of the ICAM-1, VCAM-1, and Sele strongly correlated with the presence of neurofibrillary pathology (pThr212 staining) (Fig 7A-7C); however, we did not observe any correlation between Selp and pThr212 staining in the brainstem of transgenic rats (Fig 7D).

Basal membrane deformities are observed in the microvessels in the transgenic model of tauopathy
We performed electron microscopy to investigate if there are any structural changes in the structure of BBB in transgenic rodents. Electron imaging of brainstem from controls (Fig 8A-8D) showed normal ultrastructural appearance consisting of the smooth luminal surface of endothelial cells, surrounded by continuous and even basement membrane. Pericytes with well-defined nuclei were observed. Tight junctions (TJs) were intact, and surrounding astrocytes and myelinated axons showed normal morphology.
On the other hand, the capillaries from transgenic animals showed abnormal features (Fig 8E-8H). Numerous wavy protrusions on the luminal surface of the endothelial cells and hemidesmosome-like structures on its basal side were observed. The basal membrane was significantly uneven, and the TJs were distorted or deformed. Moreover, clear and dense oedematous astrocytes were associated with the vessels in the brainstem of the transgenic rat model.

Misfolded tau-induced transmigration of cells across BBB is glia dependent
We were intrigued whether misfolded tau induces deregulation of the BBB and facilitate transmigration of cells across the BBB. For this purpose, we employed in vitro BBB model and subjected to treatment with in vitro prepared oligomeric tau (truncated tau aa 151-391/4R, 25 Immune cells trafficking in tauopathies kDa), and insoluble PHF-tau isolated from transgenic rat model and human AD brain. Characterization of oligomeric tau using Western blot analysis showed the presence of high

Fig 6. Expression of adhesion molecules is localized in the brain area affected by neurofibrillary pathology.
Immunostaining of adhesion proteins ICAM-1, VCAM-1 and selectins in transgenic and control rats. Confocal microscopy showed that brain capillaries stained with ICAM-1, VCAM-1 and selectins antibodies (red color) were distributed throughout the brain affected by neurofibrillary lesions immunolabelled with pThr212 and pSer202/pThr205 (AT8, green color). Low or no signal was detected in the brainstem of control rats and frontal cortex of control and transgenic rats. Morphological examination of oligomeric tau by electron microscopy (EM) revealed the presence of small circular oligomers with diameters ranging from 10 to 30 nm and short filaments up to 120 nm long with a width of 10-15 nm (S1B Fig). Also, imaging of insoluble PHFtau from rodent and AD brain preparations showed the presence of longer filamentous and twisted structures of up to 200 nm (S1D and S1F Fig).
The in vitro BBB model (S2 Fig) was utilized to analyze transmigration of PB-MoM across endothelial monolayer in the presence of oligomeric tau species (Fig 9A-9D). Direct abluminal exposure of brain endothelial cells alone to aggregated tau proteins did not evoke any significant responses. However, when glial cells were present, we detected increased transmigration of PB-MoM (visualized by CD45 immunostaining and counted) after 24 hours of addition of aggregated tau. In control conditions only, low amount of monocyte-input was able to transmigrate to the opposite side of Transwell inserts. In contrary, we observed accelerated transmigration of PB-MoMs with oligomeric tau by 319.0% (p < 0.0001, n = 30), rat PHF-tau by 300.6% (p < 0.0001, n = 30), and human PHF-tau to 252% (p < 0.0001, n = 30; Fig 9E) in the presence of glial cells. This suggests a scenario that interaction of oligomeric tau proteins with glial cells and their following activation augment migration of PB-MoM across the endothelial monolayer.

Involvement of ICAM-1 and VCAM-1 in aggregated tau-mediated transendothelial migration of PB-MoM
We examined if the transmigration of PB-MoM across in vitro BBB model is mediated by adhesion molecules ICAM-1 and VCAM-1. Endothelial cells grown on Transwell inserts were preincubated with monoclonal antibodies against ICAM-1 and VCAM-1 for 30 min before the addition of PB-MoM. The pre-treatment of endothelial cells with monoclonal antibodies significantly reduced the transmigration of PB-MoM across in vitro BBB model (Fig 10). The addition of an antibody against VCAM-1 caused a decrease in transmigration from 100% to 62% (p = 0.0187, n = 30). Monoclonal anti-ICAM-1 antibody reduced the transmigration from 100% to 53% (p = 0.0194, n = 30). These results demonstrate that the adhesion molecules play a role in regulating the flux of blood cells across the in vitro BBB model.

Expression of pro-inflammatory molecules and proteins involved in diapedesis are deregulated in vitro endothelial cells and capillaries from transgenic tauopathy brain
We performed transcriptomic analysis of endothelial cells from in vitro BBB model and capillaries from brainstem of transgenic and control animals. At first, the transcriptional profile of genes that were up-or down-regulated at least 1.5fold in vitro conditions is shown in Table 1. Upon stimulation of in vitro BBB model with oligomeric tau, we found elevated Immune cells trafficking in tauopathies expression of genes related to inflammation, leukocyte influx, endocytosis, angiogenesis, blood coagulation, and vasoconstriction. We also observed an increase in expression of genes for proteins that allow the rolling, adhesion and extravasation of PB-MoM. In capillaries from transgenic tauopathy model, upregulation for Sele, Selp, Chemokine (C-X-C Motif) Ligand 1 (Cxcl1), Plasminogen Activator (Plau), Natriuretic Peptide B (Nppb), Chemokine (C-C Motif) Ligand 2 (Ccl2), Matrix Metallopeptidase 9 (Mmp9), Serpin Peptidase Inhibitor 1 (Serpin 1), Tumor Necrosis Factor (Tnf), Prostaglandin-Endoperoxide Synthase 2 (Ptgs 2) was observed. On the other hand, the expression of Kinase Insert Domain Receptor (Kdr) and Matrix Metallopeptidase 1 (Mmp 1) was decreased. Of the all 84 genes analyzed, 8 genes were upregulated in both in vitro and ex vivo conditions: Sele, Selp, Cxcl1, Ccl2, Mmp9, Serpine 1, Tnf, Ptgs2 (S1 Table).

Discussion
In recent years, studies investigating the dysfunction and malfunction of brain barriers in the pathogenesis of tauopathies have gained prominence. In this study, we analyzed the mechanisms by which misfolded protein tau induce structural and functional changes of BBB, and facilitate the transmigration of blood-borne cells into the brain. Using transgenic rat model of tauopathy (line SHR-72) with progressive age-dependent neurofibrillary degeneration in the brainstem [28,32], and well established in vitro BBB model [31], we demonstrate that misfolded tau deregulates inflammatory, signaling and adhesion proteins in the BBB. Our results show that misfolded tau initiates signaling events leading to glial activation. Activated glial cells, in turn, aggravate transmigration of peripheral blood cells into the brain parenchyma.
According to previous studies, increased permeability of the BBB has been more related to cerebrovascular deposition of Aβ. Microvascular degeneration and various cerebrovascular abnormalities including endothelial and pericytic damage, reduced glucose transport, increased secretion of pro-inflammatory molecules as cytokines and chemokines by activated brain-resident immune cells were observed in the proximity of senile plaques [33,34].
Very little is known about the link between the role of misfolded tau protein in instigating functional and structural impairment of the BBB. Recent studies demonstrated that tau protein is sufficient to initiate BBB impairment in tau transgenic mouse models [35,36]. In human studies, the association between neurofibrillary pathology and progressive vascular alternations in PSP, Pick´s disease and Parkinsonism-dementia complex of Guam have been shown [4,21]. Also accumulation of activated immune cells has been documented in animal models and the human brain, indicating a contribution of neuroinflammation to the pathogenesis of neurodegenerative disorders [37]. Expression of truncated tau protein (aa 151-391/4R) induced upregulation of immune molecules such as CD11a, CD11b, CD18, CD4, CD45, and CD68. The number of immune-reactive brain resident cells progressively increased with the NFT load, suggesting that activated glial cells are involved in the immune response targeting tau neurofibrillary pathology [25].
Additionally, chronic neuroinflammation and an immune response are driven by microglia, and astrocytes trigger the structural and functional changes of BBB [38][39][40][41][42]. Neuroinflammation can promote changes in brain capillaries, such as disruption of tight junction proteins, atrophy of pericytes, thickening of basement membrane due to the accumulation of basal membrane proteins. These changes can exacerbate BBB integrity, alter transport systems or influence the role of BBB as a signaling interface. Neuroinflammation can affect BBB function by increasing vascular permeability to small molecules and plasma proteins, and finally enhancing migration of immune cells from periphery into the brain parenchyma. In the present study, we show an increase in levels of CD4 + and CD3 + T-cells in transgenic animals in the brain. Moreover, we observed transmigration of CD3 + and CD4 + T-cells into the brain stem, the area with extensive neurofibrillary pathology.
Although the migration of peripheral blood cells into brain in AD remains controversial [43,44], we were able to detect the presence of labeled CD11b + cells in the brain of tau transgenic animals, in proximity to tau pathology. This suggests that peripheral blood cells infiltrate the brain parenchyma when the BBB is compensated. The invasion is likely to be an active process relying on the interaction between endothelial adhesion molecules, and their ligands expressed on the surface of cells. Two main groups of adhesion molecules are associated with the activation of microvascular endothelium. Adhesion molecules ICAM-1 and VCAM-1 belong to the immunoglobulin superfamily and endothelial selectins such as Sele and Selp.
Increased expression of adhesion molecules and selectins initiate tethering, rolling, firm adhesion and finally extravasation of leukocytes to endothelial cells surface and enhance transmigration of immune cells from periphery into the brain [2,45]. In the transgenic model of tauopathy, we observed upregulation of both groups of endothelial adhesion molecules. Increased expression of ICAM-1, VCAM-1, Sele, and Selp was detected in the brainstem that is strongly affected by neurofibrillary pathology and invaded by T-cells. Overall, this proves the effect of tau-mediated activation of brain microvascular endothelium.
The specific mechanism of how tau affects the BBB in tauopathies remains unclear. In amyloid-burdened brain areas detrimental pro-inflammatory cytokines such as TNF-α, IFN-γ and IL-1β are likely to be the culprit. The Aβ plaques are decorated with inflammatory molecules as cytokines, chemokines, the proteins of the classical and alternative complement pathways, peroxisomal proliferators-activated receptors, proteoglycans, heat shock proteins, the metallomatrix proteinases and intercellular adhesion molecules that induced endothelial expression of adhesion molecules as ICAM-1 and VCAM-1 and enhance transmigration of immune cells from periphery into the brain [46,47].
To characterize the mechanism/s responsible for cell transmigration in tauopathies, we performed a set of in vitro experiments using double BBB model utilizing primary rat brain endothelial cells and primary mixed glial culture. Tau oligomers are significantly elevated in the patient´s brain, preceded the tangle formation and had contributed to the progression of tau neurodegeneration [31,32]. Therefore, we explored the possibility of oligomeric tau or insoluble PHF-tau derived from AD human brain and SHR-72 rat model to facilitate transmigration of peripheral blood cells. In in vitro BBB model, tau-induced glial activation led to release of inflammatory factors modifying endothelial properties. We observed an increase in the transmigration of peripheral blood cells across the endothelial monolayer as a result of dysfunction of endothelial cell-cell junction permeability and increased expression of inflammatory endothelial molecules. The migration of cells was blocked by an antibody to ICAM-1 and VCAM-1, indicating the role of trans-membranous proteins in the migration process. Our previous studies showed that tau proteins were not directly toxic to brain endothelial cells in vitro, but were able to activate glial cells [48]. We demonstrated that tau oligomers induce microglia activation and the increased expression of the inflammatory cytokines including IL-6, IL-1β, and TNF-α [25] that can promote structural and functional changes of BBB. Increased production of TNF-α and IL-1β regulates the expression of tight junction proteins: occludin-1, claudin-5, ZO-1 and ZO-2 [49,50]. Stimulation of endothelial cells with both TNF-α and IFN-γ increases expression levels of ICAM-1 , VCAM-1 , ALCAM , MCAM , Sele and Selp and various cytokine receptors by CNS vessels [51][52][53]. IFN-γ promoted transendothelial migration of CD4 + T-cells across the BBB [54].
Transcriptomic analysis showed dysregulation of genes related to inflammation, leukocyte influx, endocytosis, angiogenesis, blood coagulation, and vasoconstriction. Similarly, in isolated brain microvessels, we found dysregulation of genes related to inflammation, leukocyte influx, endocytosis, angiogenesis, blood coagulation, and vasoconstriction. This indicates that the same pathways are activated in vitro and in tau transgenic animals. Changes in genes related to inflammation are consistent with previous in vivo studies showing that AD microvessels release significantly higher levels of several inflammatory factors than non-AD microvessels [55]. Multiple studies have shown that the BBB and inflammation both play an important role in the pathology of tauopathies. Pathologically modified forms of tau protein that are released into the extracellular environment can activate microglia [56,57]. Another explanation for BBB changes is that neuronal tau accumulation triggers astrocytosis, causing these cells to detach from tight junctions [36]. We observed astrocyte swelling at EM, though the exact mechanism needs further investigation.
Overall, our results clearly demonstrated that misfolded truncated tau protein could activate glial cells producing a higher amount of detrimental inflammatory factors and modify endothelial properties and enhancing the migration of immune cells into the brain parenchyma. We suggest that increased diapedesis of monocytes across the BBB in response to tau protein may play a role in the pathophysiology of tau-related disorders and could contribute to an increased inflammatory burden in human tauopathies. Whether inflammatory processes modulating BBB permeability precede the process of neurodegeneration or are the consequence of disease´s pathology remains to be established.

Conclusions
The current study highlighted the role of tau protein in BBB changes observed in tauopathies. Neuroimmune events could be crucial components of neurofibrillary degeneration in the dis-ease´s pathology. Our data suggest that tau protein has a prominent role in regulating, perpetuating inflammation, and thus exacerbating the disease pathology.