New BBB Model Reveals That IL-6 Blockade Suppressed the BBB Disorder, Preventing Onset of NMOSD

The Institutional Review Boards of Yamaguchi University Graduate School of Medicine and Chugai Pharmaceutical Co., Ltd, approved all study protocols, and signed informed consent was obtained from each blood donor.

NMO-IgG was the IgG fraction isolated from pooled serum collected from 10 patients with NMOSD, and control IgG isolated from pooled serum collected from healthy volunteers at the Yamaguchi University Graduate School of Medicine. Both NMO-IgG and control IgG were purified by protein G affinity and adjusted for assays by extensive dialysis. Anti-AQP4 and anti-GRP78 antibodies were previously detected in NMO-IgG but not in control IgG. Satralizumab was prepared at Chugai Pharmaceutical Co., Ltd. We used NMO-IgG, control IgG, and satralizumab at a final concentration of 100 µg/mL.

hECs used were adult human brain microvascular ECs transfected. hECs, hAST and hPCT and immortalized with a plasmid encoding ts-SV40-LT as previously described.^28,29,33,34 hECs and hASTs were cultured as previously described.³⁰ hPCTs were maintained in Dulbecco's Modified Eagle's Medium (Gibco BRL) supplemented with 10% (v/v) heat-inactivated fetal bovine serum and antibiotics (100 UI/mL penicillin G sodium, 100 µg/mL streptomycin sulfate). Astrocyte medium was used as the coculture medium.

hPCTs and hASTs were cocultured on Transwell insert membranes having 3-µm pores (Corning Life Sciences, Tewksbury, MA), with hPCTs on the luminal side and hASTs on the abluminal side. hECs were cultured in Nunc dishes with an UpCell surface (Thermo Fisher Scientific, Waltham, MA), which afford sheet-like detachment of confluent cells and extracellular matrix when the temperature is shifted to 20°C. The sheet of confluent hECs was detached and transferred onto the hPCTs cocultured with hASTs on the insert. The polymers of the UpCell surface are slightly hydrophobic at 37°C but become hydrophilic at 20°C to form an aqueous film between cells and polymers resulting in sheet-like detachment of cells including their surrounding native extracellular matrix structures.

The triple cocultured inserts were transferred to an automated cell monitoring system (cellZscope; CellSeed Inc., Tokyo, Japan). After addition of satralizumab and NMO-IgG or control IgG to the vascular (hEC) side or the brain parenchymal (hAST) side or to both sides, the TEER values were measured using the cellZscope device, which can record the TEER every minute for 96 hours.

The isolation of peripheral blood mononuclear cells (PBMCs) and flow-based transmigration assays were performed in a 3D BioFlux flow chamber device (Fluxion Bioscience, San Diego, CA) as previously described.³⁰ The migrated cells were enumerated by a hemocytometer and then normalized to migrated cell numbers determined by flow cytometry. After collection, cells were fixed for 10 minutes in 1% paraformaldehyde at room temperature and washed in phosphate buffered saline + 0.1 mM ethylenediaminetetraacetic acid, followed by blocking in mouse IgG. Cells were labeled with anti-human CD45 eFluor450, CD8a APC-eFluor780 (eBioscience, San Diego, CA), CD3Alexa Fluor 647 (BioLegend, San Diego, CA), CD19 BV711, and CD4 PE-CF594 (BD Biosciences). Data were acquired using BD FACSCanto II (BD Biosciences) and analyzed by FlowJo software (v.10.4.1; Treestar, Ashland, OR).

Control IgG and satralizumab were labeled with IRDye 800CW protein (IRDye 800CW Protein Labeling Kit; LI-COR, Lincoln, NE) following the manufacturer's protocol. After exposing hECs in triple cocultured inserts to labeled control IgG or satralizumab, the microvolumes that translocated to the lower chamber were detected by an Odyssey Infrared Imaging System (LI-COR), and the apparent BBB permeability coefficient (Papp; mm/seconds) was calculated as previously described.³⁰

NMO-IgG or control IgG was added to hECs in triple cocultured inserts and incubated for 12 hours. Following the manual of the Easy‐Titer Antibody Assay Kit (Thermo), anti-human IgG coated beads were added to the lower well and incubated with the sample for 60 minutes. Accumulated IgG in the lower well was measured by a spectrophotometer, following the manufacturer's instructions. The total amount of accumulated IgG was normalized to 1 using control IgG and reported as IgG accumulation.

After exposing hECs in triple cocultured inserts to satralizumab plus NMO-IgG or satralizumab plus control IgG for 24 hours, the concentration of satralizumab in the lower chamber was measured by using enzyme-linked immunosorbent assay (ELISA) with anti-satralizumab antibody. The samples were run according to the manufacture's protocol. The total amount of accumulated satralizumab was normalized to 1 using control IgG + satralizumab and reported as satralizumab accumulation.

Female C57BL/6J mice (7 weeks old; Charles River Laboratories Japan, Inc., Kanagawa, Japan) were used. EAE was induced in mice by subcutaneous immunization (on day 0) with 50 µg of the myelin oligodendrocyte glycoprotein 35–55 peptide (MOG35-55; Peptide International, Louisville, KY) emulsified in complete Freund adjuvant (Difco Laboratories, Detroit, MI) supplemented with Mycobacterium tuberculosis extract H37Ra (Difco Laboratories). In addition, mice received 250 ng pertussis toxin (List Biological Laboratories, Campbell, CA) IV on day 0 and intraperitoneally on day 2. Control mice were treated with complete Freund adjuvant and saline alone. Anti-IL-6 receptor antibody (MR16-1) was prepared using a hybridoma established in Chugai Pharmaceutical's laboratories.³⁸ The EAE mice were intraperitoneally administered MR16-1 (8 mg/mouse) on day 7 after MOG35-55 immunization. On day 15 or 16 after MOG35-55 immunization, spinal cords, spleens, and sera were harvested for immunohistochemistry, flow cytometry, and TEER studies. EAE mice were sequentially scored for clinical signs of EAE according to the following scale: 0, no apparent disease; 1, limp tail; 2, hind limb weakness; 3, hind limb paresis; 4, hind limb paralysis; 5, hind limb and forelimb paralysis; and 6, moribundity and death.

Mice were anesthetized with isoflurane, and transcardial perfusion was performed with 20 mL of cold phosphate buffered saline. The L3–L5 segment of the lumbar spinal cord was removed, fixed in 4% paraformaldehyde, and placed in a 30% sucrose solution overnight. Samples were embedded in optimal cutting temperature compound, and frozen slices of spinal cord (10 µm thick) were obtained with a cryostat. Spinal cord slices were stained by using the following primary antibodies: goat anti-albumin antibody (1:200, A90-134A; Bethyl Laboratories, Inc., Montgomery, TX), biotin-conjugated donkey anti-mouse IgG antibody (1:200, 715-066-151; Jackson ImmunoResearch, West Grove, PA), and rat anti-CD4 antibody (1:100, 550280; BD Pharmingen Inc., San Diego, CA). After overnight incubation with primary antibodies at 4°C, spinal cord sections were incubated with secondary antibody Alexa Fluor 488-conjugated donkey anti-goat IgG (1:200, 705-546-147; Jackson ImmunoResearch) and Alexa Fluor 488-conjugated streptavidin (2 µg/mL, 016-540-084; Jackson ImmunoResearch). For CD4 staining, biotin-conjugated donkey anti-rat IgG (1:200, 712-066-153; Jackson ImmunoResearch) and Alexa Fluor 488-conjugated streptavidin (2 µg/mL, 016-540-084; Jackson ImmunoResearch) were used. Slides were mounted using Vectashield Antifade Mounting Medium (H-1200; Vector Laboratories, Burlingame, CA). Spinal cord slices were randomly selected from each mouse and observed under a BZ-9000 Fluorescence Microscope (Keyence, Osaka, Japan). Positive staining areas were calculated using the BZ-II analyzer (Keyence), and CD4-positive T cells were counted with imaging analysis software (WinROOF Version 6.3.1; Mitani Corporation, Fukui, Japan).

Spleens collected from control mice and EAE mice were homogenized and passed through a 100- and 40-μm cell strainer to isolate mononuclear cells. Red blood cells were hemolyzed with ammonium-chloride-potassium lysing buffer (Gibco, Carlsbad, CA). Mononuclear cells were incubated with Mouse BD Fc Block (BD Pharmingen Inc.) before staining. The cells were initially stained with fluorescein isothiocyanate-conjugated anti-CD4 antibody (100510; BioLegend) and then intracellularly stained using PE-conjugated anti-IL-17A (506904; BioLegend) and APC-conjugated anti-interferon-γ (505810; BioLegend) antibodies; staining was performed with the Fixation/Permeabilization Solution Kit with BD GolgiPlug (BD Biosciences) according to the manufacturer's protocol. For analysis of Treg cells, Fc-blocked cells were initially stained with fluorescein isothiocyanate-conjugated anti-CD4 and BV421-conjugated anti-CD25 (102034; BioLegend) antibodies and then intracellularly stained using the APC-conjugated anti-Foxp3 antibody (17-5773-80B; Invitrogen, Carlsbad, CA); staining was performed with a Foxp3 Transcription Factor Staining Buffer Set (Invitrogen) according to the manufacturer's protocol. Data were acquired using BD FACSCanto II (BD Biosciences) and analyzed using FlowJo 10.4.1 (Treestar).

All data are expressed as mean and standard error of mean (SEM). The statistical significance of differences was determined by using the unpaired t test, Tukey multiple comparison test with analysis of variance, or 2-way analysis of variance for comparison of time course data. Probability values of less than 0.05 were considered significant. Statistical analyses were performed using IBM SPSS Statistics (International Business Machines Corporation, Armonk, NY) or JMP version 11.2.1 software (SAS Institute, Cary, NC).

We used human brain microvascular ECs (hECs; TY10), hASTs with AQP4 expression, and hPCTs, each of which was conditionally immortalized by transfection with ts-SV40-LT (Figure 1A). These transfected cell lines proliferate at 33°C due to ts-SV40 LT's temperature sensitive inhibition activity of p53 and Rb. On the other hand, cells could differentiate into mature cells at 37°C because Ts-SV40 LT is inactivated and exhibit growth arrest^28,33,34 (Figure 1B). In these previous articles, we confirmed the expression markers of hECs, hASTs, and hPCTs (hECs; occludin, Claudin-5, Claudin-12, ZO-1, ZO-2, JAM-A, GLUT-1, CAT-1, LAT-1, 4F2hc, MCT-1, calreticulin, MDR1, ABCG2, MRP1, MRP2, MRP4, and MRP5, and hASTs; glial fibrillary acidic protein, EAAT2, and AQP4, hPCTs; platelet-derived growth factor receptor beta, Desmin, Kir 6.1, osteopontin, alfa smooth muscle actin, and NG2). To construct triple coculture systems of these human BBB cell lines, multiple steps were organized by including cell culture on Upcell dish with coated temperature-responsive polymer (eFigure 1, A and B, links.lww.com/NXI/A591). First, hASTs were cultured on abluminal side of insert membranes having 3-μm pores and incubated for 24 hours so that some astrocytic endfeet with AQP4 could protrude through the membrane pores (eFigure 1C and links.lww.com/NXI/A591). Second, after culturing of hPCTs on the luminal side of membranes (eFigure 1E, links.lww.com/NXI/A591), hASTs and hPCTs were cocultured at 33°C for 24 hours. Third, hECs were cultured on Upcell dish with coated temperature-responsive polymer, which can achieve sheet-like detachment of confluent cells and extracellular matrix by temperature shifting to 20°C (eFigure 1, F and G, links.lww.com/NXI/A591). Then, sheet-like detachment of confluent hECs was transferred onto the hPCTs. After cocultuting of these cell lines at 33°C for 24 hours, they differentiated into mature cells under the condition of 37°C. Confocal 3D analysis with living staining of each cell line showed that multicultured insert constituted the five-layer structures, which is consisted of hECs, hPCTs, astrocytic endfeet, membrane, and hAST (eFigure 1, H, links.lww.com/NXI/A591). Some endfeets protruded through the membrane pores. hPCTs and endfeet of hASTs were close to the hEC layer.

To evaluate the effect of satralizumab on NMO-IgG–induced transmigration of leukocytes, we constructed a flow-based dynamic BBB model incorporating hEC/hPCT/hAST triple coculture, which allows further investigation of leukocyte transmigration under flow (Figure 2, A–C). After exposing the EC side (vascular side) and the astrocyte side (brain parenchymal side) to satralizumab plus NMO-IgG or to NMO-IgG alone (Figure 2D), we counted the total numbers of all migrating cells and the numbers of phenotyped cells and compared their migrations relative to those with control IgG. Application of NMO-IgG increased the migrations of all PBMCs and CD4⁺, CD8⁺, and CD19⁺ cells relative to their migrations with control IgG, and the application of NMO-IgG plus satralizumab significantly suppressed this increase in the relative numbers of migrating PBMCs and CD4⁺ and CD8⁺ cells (Figure 2E).

Administration of MR16-1 (anti–IL-6 receptor antibody for mice) on day 7 after induction of EAE significantly and strongly prevented the onset of clinical signs in EAE mice (Figure 3A). We then confirmed the effect of MR16-1 on leukocyte transmigration in EAE mice. On day 15, the number of CD4⁺ T cells markedly increased in the spinal cords of EAE mice, whereas these cells were almost undetected in the spinal cords of control mice (control, 33.7 ± 21.5 [mean ± SEM] cells/spinal cord slice; EAE, 628.3 ± 196.5 cells/spinal cord slice) (Figure 3B). Administration of MR16-1 on day 7 significantly prevented the migration of CD4⁺ T cells into the spinal cord (EAE + MR16-1, 40.3 ± 17.7 cells/spinal cord slice) (Figure 3B). To exclude the possibility that the impact of MR16-1 on clinical signs and on leukocyte migration into the spinal cord was secondary to changes in the immune response, we evaluated the effect of MR16-1 on T-cell differentiation in EAE mice. In splenocytes, Th1 cells and FoxP3-positive regulatory T cells were significantly increased on day 16 in EAE mice (Figure 3, C and E). Th17 cells showed a tendency, but not significantly, to increase in EAE mice (Figure 3D). Administration of anti–IL-6 receptor antibody on day 7 did not affect the induction of these in EAE mice (Figure 3, C–E).

Immunohistochemical analysis on day 15 showed that leakage of albumin and IgG into the spinal cord was higher in EAE mice than in control mice (Figure 4, A and B). This leakage into the CNS indicates the increased permeability of the BBB. Treatment of EAE mice with MR16-1 significantly prevented these leakages into the spinal cord. As regards the inhibiting effect on barrier dysfunction with application of satralizumab on the vascular side, we showed that sera from EAE mice mildly reduced the barrier function of ECs in monoculture (eFigure 2, links.lww.com/NXI/A592) Serum from EAE mice at onset of EAE on day 16 after immunization significantly decreased the TEER value of a monolayer of mouse primary BMECs. Anti–IL-6 receptor antibody significantly prevented this reduction.

To evaluate whether NMO-IgG affects the barrier function of the BBB on the vascular side or the brain parenchymal side, we constructed the static in vitro BBB model, which allowed long-term measurement of real-time TEER. After addition of NMO-IgG or control IgG with either the vascular side, the brain parenchymal side, or both sides, the TEER values were measured in every minute for 5 consecutive days. Within 24 hours of application of NMO-IgG, TEER values in all groups had begun to decrease compared with control IgG, and at 72 hours, they had significantly decreased in all groups (Figure 5, A and B). Application of NMO-IgG to both the vascular and brain parenchymal sides resulted in the lowest TEER values of all groups. TEER values with brain parenchymal application of NMO-IgG were significantly lower than TEER values with vascular application of NMO-IgG.

To evaluate whether satralizumab affects the BBB dysfunction induced by NMO-IgG, we used the static BBB model incorporating hEC/hPCT/hAST triple coculture with either the vascular side or the brain parenchymal side or both sides exposed to NMO-IgG. After addition of satralizumab plus NMO-IgG or control IgG to either the vascular side or the brain parenchymal side or both sides at the same time, TEER values were measured as mentioned above. Ahead of these experiments, we confirmed that satralizumab did not affect the barrier function (eFigure 3, links.lww.com/NXI/A593). TEER values at 72 hours were significantly higher under conditions where both the vascular side and the brain parenchymal side were exposed to satralizumab plus NMO-IgG than under conditions of NMO-IgG alone (Figure 5C). The inhibiting effect of satralizumab on barrier dysfunction was almost the same when satralizumab was applied to the brain parenchymal side as it was when applied to the vascular side (Figure 5, D and E). Application of satralizumab to both the vascular and brain parenchymal sides had the highest inhibiting effect on barrier dysfunction among the 3 conditions (Figure 5, C and E).

To compare the microvolume translocation of satralizumab and NMO-IgG across the BBB, we constructed the in vitro BBB models in which measurement of microvolumes of satralizumab and IgG translocation through the BBB could be detected by Odyssey Infrared Imaging System and ELISA. We evaluated the apparent permeability coefficients of the BBB (Papp; mm/seconds) with respect to satralizumab and control IgG. Labeled satralizumab or IgG2 (control IgG) was added to the hEC (vascular) side. Then, the IgG that was translocated to the brain parenchymal side was detected by an infrared imaging system, and the apparent BBB permeability coefficients for satralizumab and control IgG were calculated. The Papp with respect to satralizumab was almost 3 times the Papp with respect to control IgG (Figure 6A).

After exposing the vascular side to NMO-IgG or control IgG, the translocated IgG was detected by the human IgG detection ELISA kit. The total amounts of accumulated IgG for NMO-IgG and control IgG were calculated, normalized with respect to control IgG, and reported as IgG accumulation. The intracerebral transferability of NMO-IgG was almost 1.5 times that of control IgG (Figure 6B).

In a previous study, we found the transfer rate of MR16-1 across the BBB in EAE mice to be almost 30 times that of normal mice.³² Thus, we explored whether the intracerebral transferability of satralizumab would be similarly increased in the presence of NMO-IgG. After exposing the vascular side of the triple coculture BBB model to satralizumab plus NMO-IgG or to satralizumab plus control IgG at the same time, the accumulated satralizumab was measured by ELISA with anti-satralizumab antibody. The total amount of satralizumab for NMO-IgG plus satralizumab was normalized with respect to control IgG plus satralizumab and reported as satralizumab accumulation. The relative satralizumab accumulation for NMO-IgG plus satralizumab was significantly increased to almost 3 times that for control IgG plus satralizumab (Figure 6C).