Article Text

Download PDFPDF

Research paper
Autocrine MMP-2/9 secretion increases the BBB permeability in neuromyelitis optica
  1. Ayako Tasaki1,
  2. Fumitaka Shimizu1,
  3. Yasuteru Sano1,
  4. Miwako Fujisawa1,
  5. Toshiyuki Takahashi2,
  6. Hiroyo Haruki1,
  7. Masaaki Abe1,
  8. Michiaki Koga1,
  9. Takashi Kanda1
  1. 1Department of Neurology and Clinical Neuroscience, Yamaguchi University Graduate School of Medicine, Ube, Japan
  2. 2Department of Neurology, Tohoku University Graduate School of Medicine, Miyagi, Japan
  1. Correspondence to Dr Takashi Kanda, Department of Neurology and Clinical Neuroscience, Yamaguchi University Graduate School of Medicine, 1-1-1, Minamikogushi, Ube, Yamaguchi 7558505, Japan; tkanda{at}yamaguchi-u.ac.jp

Abstract

Objective Pathological breakdown of the blood-brain barrier (BBB) is thought to constitute the beginning of the disease process in neuromyelitis optica (NMO). In the current study, we investigated possible molecular mechanisms responsible for the breakdown of BBB using NMO sera.

Methods We analysed the effects of sera obtained from anti-aquaporin 4 (AQP4) antibody-positive NMO spectrum disorder (NMOSD) patients, multiple sclerosis (MS) patients and control subjects on the production of claudin-5, matrix-metalloproteinases (MMPs)-2/9, and vascular cell adhesion protein-1 (VCAM-1) in human brain microvascular endothelial cells (BMECs). We also examined whether immunoglobulin G (IgG) purified from NMOSD sera influences the claudin-5 or VCAM-1 protein expression.

Results The disturbance of BBB properties in BMECs following exposure to NMOSD sera was restored after adding the MMP inhibitor, GM6001. The secretion of MMP-2/9 by BMECs significantly increased after applying the NMOSD sera. The sera from NMOSD patients also increased both the MMP-2/9 secretion and the VCAM-1 protein level by BMECs. The IgG purified from NMOSD sera did not influence the BBB properties or the amount of MMP-2/9 proteins, although it did increase the amount of VCAM-1 proteins in BMECs. Reduction in anti-AQP4 antibody titre was not correlated with a reduction in VCAM-1 expression.

Conclusions The autocrine secretion of MMP-2/9 by BMECs induced by humoral factors, other than IgG, in sera obtained from NMOSD patients potentially increases BBB permeability. IgG obtained from NMOSD sera, apart from anti-AQP4 antibodies, affect the BBB by upregulating VCAM, thereby facilitating the entry of inflammatory cells into the central nervous system.

  • Blood-Brain Barrier
  • Multiple Sclerosis

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Introduction

Neuromyelitis optica (NMO) is a severe relapsing inflammatory disorder of the central nervous system (CNS), characterised by the development of optic neuritis and longitudinally extensive transverse myelitis (LETM), and is presumed to be a different disease entity from multiple sclerosis (MS).1 The groundbreaking discovery of an autoantibody against aquaporin 4 (AQP4), which is densely expressed in astrocytic foot processes,2 with a high sensitivity and specificity for the disease has enhanced our understanding of NMO. Several in vitro and in vivo studies have suggested that these antibodies contribute to the pathogenesis of disease and are crucial for the development of NMO.3–10 However, high serum levels of anti-AQP4 antibodies are not always accompanied by clinical relapse, indicating that the presence of serum anti-AQP4 antibodies alone is insufficient to induce a clinical relapse and that other factors, including inflammatory mediator(s), are required.11–13

The distinctive histological features of NMO are predominantly observed in perivascular lesions, where the blood-brain barrier (BBB) is located.5 Circulating anti-AQP4 antibodies must penetrate the BBB, which is composed of the tight junctions, and enter the CNS space in order to bind to the AQP4 expressed on astrocytic endfeet. The intravenous or intraperitoneal administration of this antibody in normal animals has, thus far, not successfully induced the development of NMO,9 ,10 because, under normal conditions, the BBB does not allow the entry of circulating anti-AQP4 antibodies into the CNS space. Some studies have demonstrated the presence of BBB damage, thus suggesting increased entry of anti-AQP4 antibodies and cytokines into the CNS space, which appears to be associated with the development of NMO14 ,15; however, the molecular mechanism(s) underlying the breakdown of the BBB in patients with NMO are not fully understood.

Previous studies have demonstrated that the serum and cerebral spinal fluid (CSF) concentrations of matrix-metalloproteinase-9 (MMP-9) and adhesion molecules in NMO patients are significantly higher than those observed in patients with MS and healthy controls and are correlated with the clinical and radiological severity of the disease.16 ,17 Accumulating evidence also suggests that the BBB disruption induced by MMPs-2/9 is an important step in the development of some inflammatory CNS diseases, including MS18 and experimental allergic encephalomyelitis (EAE).19 We thus hypothesised that MMP-2/9 is a candidate molecule causing the breakdown of the BBB in NMO patients. In the current study, we investigated the contributions of humoral factors, particularly MMP-2/9, in the sera obtained from patients with NMO to malfunction of the BBB.

Materials and methods

Sera

This study was conducted in accordance with the Declaration of Helsinki, as amended in Somerset West in 1996, and approved by the ethics committee of the Medical Faculty, Yamaguchi University. We collected sera from 14 patients with NMO spectrum disorders (NMOSD) (all female; mean age, 52 years), including 10 patients in the acute phase and four patients with stable disease, who were hospitalised at our institution. Anti-AQP4 antibody assays were conducted according to the methods described in the previous report,6 and all patients tested positive for anti-AQP4 antibodies and fulfilled the clinical criteria for NMOSD.20 ,21 Seven patients with definitive NMO (patients nos. 1, 2, 4, 6, 7, 9 and 10) and three patients with partial NMO, defined as isolated LETM (patients nos. 3, 5 and 8), in the acute phase were enrolled in this study; no patients with isolated optic neuritis were included. Blood samples were obtained within 25 days of the initial attack (mean time from symptom onset to serum sample collection: 13.2 days (SD=5.4)). At the time of sampling, one of 10 patients in the acute phase was being treated with methylprednisolone pulse therapy, although none were taking immunosuppressive drugs, including corticosteroids or azathioprine. Serum samples were also collected from four patients with NMOSD in the stable phase who were being treated with corticosteroids or had been in clinical remission for at least 6 months (patients nos. 5, 8, 9 and 10). Seven of the 10 NMOSD patients were positive for anti-brain microvascular endothelial cell (BMEC) antibodies (patients nos. 2, 4, 5, 6, 8, 9 and 10), while the other three were negative for anti-BMECs antibodies (patients nos. 1, 3 and 7). The method used to detect anti-BMECs antibodies has been previously described.15 Serum samples were also obtained from 10 patients with conventional MS during the acute phase who fulfilled the revised McDonald criteria22 (four men, six women; mean age, 35.7 years) and 10 healthy volunteers for comparison. Blood samples were obtained from six patients before treatment with high-dose intravenous methylprednisolone (patients nos. 2, 3, 4, 8, 9 and 10) and from four patients after such treatment (patients nos. 1, 5, 6 and 7) within 28 days of the first appearance of symptoms. Six of the 10 MS patients and one of the 10 healthy controls were positive for anti-BMECs antibodies. All samples were immediately stored at −80°C until the analysis in order to avoid repeated freeze/thaw cycles. We inactivated all samples at 56°C for 30 min immediately prior to the analysis.

Cell culture and treatment

We used the conditionally immortalised human cell lines previously described by our group, including microvascular endothelial cells of the brain (BMECs), termed ‘TY08,’,23 astrocytes, termed ‘hAST-AQP4,’24 and human blood-nerve-barrier-derived endothelial cells, termed ‘FH-BNBs.’.25 The cell cultures were maintained in fresh medium containing 10% patient or healthy control sera, and 10% fetal bovine serum (FBS) and applied as controls at 37°C in 5% (vol/vol) CO2/air. The cells were cultured for one day before total mRNA was extracted and the transendothelial electrical resistance (TEER) value was measured. Total proteins were extracted 2 days later.

Reagents

The culture medium was Dulbecco's modified Eagle's medium (DMEM) (Sigma, St Louis, Missouri, USA) containing antibiotics and 10% FBS (Sigma). 23 The polyclonal anti-MMP-9, anti-MMP-2, anti-actin, and anti-vascular endothelial growth factor (VEGF) antibodies were purchased from Santa Cruz (Santa Cruz, California, USA). The polyclonal anti-claudin-5 antibodies were obtained from Invitrogen (Carlsbad, California, USA), and the polyclonal anti-vascular cell adhesion molecule-1 (VCAM-1) antibodies were obtained from R&D systems (Minneapolis, Minnesota, USA). The GM6001 was obtained from Chemicon (Temecula, California, USA), and the MMP-2 and MMP-9 inhibitors were obtained from Santa Cruz.

Quantitative real-time PCR analysis

For real-time RT-PCR, total RNA synthesised from PBS-washed cells and single-stranded cDNA was prepared from 40 ng of total RNA. The sequences of human primers for MMP-2, MMP-9 and G3PDH have been previously described.26 A Stratagene Mx3005P instrument (STRATAGENE, Cedar Greek, Texas, USA) was used to perform the quantitative real-time PCR analyses, and the relative quantity of each molecule was calculated according to the Rv=RGene/RGAPDH using a software program as previously described.26

Western blot analysis

The protein samples (10–20 μg) were fractionated in a 10% gel and electrophoretically transferred onto polyvinylidene difluoride membranes (Amersham, Chalfont, UK), as previously described.23 The membranes were treated with the primary antibody in PBS-T and 5% milk (dilution 1:100) for 2 h, followed by incubation with the secondary antibody (dilution 1:2000) for 1 h. The bands were visualised with an enhanced chemiluminescence kit (ECL-prime, Amersham, UK). The relative density of each band was measured using the Quantity One software program (Bio-Rad, Hercules, California, USA).

TEER studies

A Millicell electrical resistance apparatus (Endohm-6 and EVOM, World Precision Instruments, Sarasota, Florida, USA) was used to measure the TEER values of the cell layers, as previously described.23 The BMECs were seeded at 1×106 cells/insert on the collagen-coated culture inserts and cultured with each flesh medium (the conditioned medium contained 10% patient or healthy control sera) for 24 h.

Permeability studies

Permeability studies with sodium fluorescein were used to determine the degree of paracellular flux across confluent BMEC monolayers, as previously described.27 The confluent monolayers were prepared on 24-well culture plates. Next, we added 500 μl of culture medium containing sodium fluorescein (10 μg/ml; molecular mass of 400 kDa) on the upper chamber of each well. The lower chamber was sampled after incubation, and the amount of fluorescence that passed through the cell-covered inserts was measured using a MX3000P instrument (Stratagene).

Treatment with MMP inhibitors

GM6001, an MMP-2 inhibitor and an MMP-9 inhibitor were used to inhibit MMP-2 or MMP-9. The serum samples obtained from the patients were preincubated with 25 μM of GM6001, 5 μM of the MMP-2 inhibitor or 5 μM of the MMP-9 inhibitor for 12 h at 37°C. The cells were cultured with the conditioned medium containing sera obtained from either the NMOSD or MS patients or healthy controls; all samples contained an MMP inhibitor.

ELISA

The concentrations of total MMP-2 and MMP-9 observed prior to incubation at 56°C were measured using ELISA (R&D systems). The samples were run in triplicate according to the manufacturer's protocol.

IgG purification and exposure of the cells to purified IgG

Affinity chromatography using a Melon Gel IgG Spin Purification Kit (Thermo Scientific, Rockford, Illinois, USA) was performed to obtain purified immunoglobulin G (IgG) fractions from the sera of five patients with anti-BMEC antibody-positive NMOSD and five healthy individuals. The cells were cultured in medium containing purified IgG (final concentration 400 μg/mL) obtained from FBS as a control (Sigma), the patients or the healthy volunteers.

Absorption of anti-AQP4 antibodies

The methods used to study the absorption of anti-AQP4 antibodies have been previously explained.15 Sera obtained from two different NMOSD patients (NMOSD1 and NMOSD2) were added to the hAST-AQP4 cells for 150 min and used for the subsequent analyses. In both cases, the titres of anti-AQP4 antibodies were decreased by at least 50% after the 150 min incubation period (anti-AQP4 antibody titres: NMOSD1 before treatment, 1:8 and after 150 min of treatment, 1:4; NMOSD2 before treatment, 1:2048 and after 150 min of treatment, 1:512).15

Data analysis

Average values are reported as the mean±SEM. The indicated statistical tests (unpaired, two-tailed Student t tests) were performed assuming significance for p<0.05.

Results

The NMOSD sera decreased the barrier function of the BMECs

Figure 1 shows the effects of the NMOSD sera, including those obtained from seven patients with definitive NMO and three patients with isolated LETM, on the expression of tight junctional and adhesion molecules in the BMECs using a western blot analysis. The sera obtained from the NMOSD patients significantly decreased the amount of claudin-5 in the BMECs, whereas the sera obtained from the MS patients and healthy controls did not change the amount of this protein (figure 1A–D). By contrast, the sera obtained from patients with either NMOSD or MS significantly increased the amount of VCAM-1 expressed by the BMECs (figure 1A–C,E), whereas the sera obtained from the healthy controls did not affect the level of this protein. Treatment with the sera obtained from the NMOSD patients significantly showed lower TEER values and greater NaF permeability of BMECs, while the sera obtained from the MS patients and healthy controls did not have this effect (figure 1H,I). There were no significant differences between NMO and isolated LETM samples in terms of the degree of BBB damage following exposure to the sera (figure 1F,G).

Figure 1

Serum samples were collected from seven patients with definite neuromyelitis optica (NMO) (patients nos. 1, 2, 4, 6, 7, 9 and 10) and three patients with partial NMO, defined as isolated longitudinally extensive transverse myelitis (LETM) (patients nos. 3, 5 and 8). (A–C) The changes in the amount of claudin-5 and VCAM-1 in BMECs were determined after exposure to the sera from patients with NMOSD, multiple sclerosis (MS) or from healthy controls by a western blot analysis. (D, E) The bar graph reflects the combined densitometry data from each independent experiment (mean±SEM, n=10). The amount of claudin-5 was significantly decreased after the exposure of BMECs to the sera from patients with NMOSD, whereas it was not affected by the sera from patients with MS or from healthy controls. The amount of VCAM-1 expressed by the BMECs was significantly increased after exposure to NMOSD or MS sera, although it was not influenced by the sera from healthy controls. (F, G) No significant differences were observed between the patients with isolated LETM and definite NMO in terms of the amount of claudin-5 and VCAM-1 in the BMECs after exposure to the patients’ sera. (H, I) The transendothelial electrical resistance value of BMECs was significantly decreased (H), and the NaF permeability of BMECs was significantly increased (I), after exposure to NMOSD sera, but these parameters were not influenced by the exposure to sera from patients with MS or healthy controls (mean±SEM, n=10, *p<0.01). Control: non-conditioned DMEM containing 20% fetal bovine serum (FBS); NMOSD: conditioned medium with 10% serum from an NMOSD patient diluted with non-conditioned DMEM containing 10% FBS; MS: conditioned medium with a 10% concentration of serum from an MS patient diluted with non-conditioned DMEM containing 10% FBS; normal: conditioned medium with 10% serum from a healthy control diluted with non-conditioned DMEM containing 10% FBS; LETM: conditioned medium with 10% serum from an isolated LETM patient diluted with non-conditioned DMEM containing 10% FBS; definite NMO: conditioned medium with 10% serum from a definite NMO patient diluted with non-conditioned DMEM containing 10% FBS.

The MMP-2/9 inhibitor reversed the BBB damage caused by the NMOSD sera

Figure 2 shows the contribution of MMP-2/9 to the malfunction of the BBB observed in the NMOSD patients. The amount of claudin-5 proteins in the BMECs was significantly increased following exposure to the NMOSD sera pretreated with GM6001, a broad-spectrum MMP inhibitor, regardless of the presence of anti-BMECs antibodies, compared with that observed in cells not exposed to sera pretreated with GM6001 (figure 2A,D). However, the expression of claudin-5 was unchanged following exposure to the sera obtained from the MS patients or healthy controls, regardless of whether the samples were pretreated with GM6001 (figure 2B,C,E,F). Additionally, the TEER values of the BMECs were significantly higher and the NaF permeability was significantly lower following exposure to the NMOSD sera pretreated with GM6001 compared to that in the cells exposed to sera not pretreated with GM6001 (figure 2G,H). The TEER values and NaF permeability were unchanged following exposure to sera obtained from the MS patients or healthy controls, regardless of the presence or absence of GM6001 pretreatment (figure 2G,H). The decreases in the claudin-5 protein levels and TEER values observed in the BMECs were restored by the application of a selective MMP-2 or MMP-9 inhibitor (figure 2I–K), although no significant differences were noted in the degree of BBB disruption following the inhibition of either MMP-2 or MMP-9 alone (figure 2I–K).

Figure 2

(A–C) The effects of the broad-spectrum matrix-metalloproteinase (MMP) inhibitor, GM6001, on the expression of claudin-5 in brain microvascular endothelial cells (BMECs) after exposure to the sera from 10 NMO spectrum disorder (NMOSD) patients were examined by a western blot analysis. Exposure to sera from NMOSD patients preincubated with GM6001 led to an increase in the amount of claudin-5 protein in the BMECs in comparison with the cells exposed to NMOSD sera without GM6001 pretreatment. (D–F) Each column reflects the combined densitometry data from each independent experiment (mean±SEM, NMOSD n=10, multiple sclerosis (MS) n=5, normal n=5, *p<0.01). (G) (H) The transendothelial electrical resistance (TEER) values of BMECs was significantly increased (G) and the NaF permeability in the BMECs was significantly decreased (H) following the incubation with NMOSD sera that had been pretreated with GM6001 compared to that in the cells exposed to sera without GM6001 pretreatment (mean±SEM, NMOSD n=10, MS n=5, normal n=5, *p<0.01). (I) The effects of the selective MMP-2 or MMP-9 inhibitor on the expression of claudin-5 in BMECs after exposure to the sera from NMOSD patients were investigated by a western blot analysis. Preincubation with a selective MMP-2 or MMP-9 inhibitor led to an increase in the amount of claudin-5 protein in BMECs compared to cells treated with sera without the inhibitor treatment. (J) Each bar graph reflects the combined densitometry data from each independent experiment (mean±SEM, n=5, MMP-2: p<0.05, MMP-9: p<0.05). There were no significant differences in the amount of claudin-5 following the inhibition of MMP-2 alone or MMP-9 alone. (K) The TEER value of BMECs was significantly increased after the incubation with sera from NMOSD patients pretreated with a selective MMP-2 or MMP-9 inhibitor compared to that in the cells without pretreatment with the MMP-2 or MMP-9 inhibitor (mean±SEM, n=5, MMP-2: p<0.05, MMP-9: p<0.01). No significant differences in the TEER values were observed following the inhibition of MMP-2 alone or MMP-9 alone. NMOSD: conditioned medium with 10% NMOSD patient serum diluted with DMEM containing 10% fetal bovine serum (FBS); NMOSD+GM6001: conditioned medium with 10% NMOSD sera pretreated with GM6001; MS: conditioned medium with 10% MS patient serum diluted with DMEM containing 10% FBS; MS+GM6001: conditioned medium with 10% MS sera pretreated with GM6001; normal: conditioned medium with 10% healthy individual serum diluted with DMEM containing 10% FBS; Normal+GM6001: conditioned medium with 10% healthy individual sera pretreated with a GM6001. NMOSD+MMP-2 inhibitor: conditioned medium with 10% NMOSD sera pretreated with an MMP-2 inhibitor; NMOSD+MMP-9 inhibitor: conditioned medium with 10% NMOSD sera pretreated with an MMP-9 inhibitor.

Autocrine MMP-2/9 secretion and damage to the BBB caused by the NMOSD sera

Figure 3 shows the contribution of the autocrine secretion of MMP-2/9 by BMECs following exposure to NMOSD sera. No significant differences were observed regarding the serum concentrations of MMP-2 and MMP-9 between the serum samples obtained from the NMOSD and MS patients or healthy controls, as determined using the ELISA method (figure 3A,B). The induction of MMP-2 and MMP-9 mRNA in the BMECs was significantly increased following exposure to the NMOSD sera, although it remained unchanged following the exposure of the cells to sera obtained from the MS patients and healthy controls (figure 3C,D). We next examined whether the sera obtained from the NMOSD and MS patients would increase the MMP-2 and MMP-9 protein expression levels in the BMECs (figure 3E–G). The sera obtained from the acute-phase NMOSD patients significantly increased the amount of MMP-2 and MMP-9 secreted by the BMECs, whereas the sera obtained from the MS patients or healthy controls did not have this effect (figure 3H,I). The presence of anti-BMECs antibodies did not influence the MMP-2/9 protein secretion by the BMECs (figure 3J,K). Meanwhile, the amount of claudin-5, MMP-2 and MMP-9 proteins in the BMECs was not affected by incubation with the stable-phase NMOSD sera (figure 3L,M), and the levels of these proteins in the FH-blood-nerve barriers (BNBs) were not influenced by exposure to the acute-phase NMOSD sera (figure 3N,O).

Figure 3

(A)(B) The MMP-2/9 concentration was analysed in the sera from patients with NMO spectrum disorder (NMOSD) or multiple sclerosis (MS), or from healthy control subjects. The bars indicate the mean of each group. There were no significant differences among the three groups. (C)(D) The MMP-2/9 expression in brain microvascular endothelial cells (BMECs) after exposure to sera from NMOSD patients was determined by relative quantification with a real-time RT-PCR analysis. The expression level of MMP-2 (C) and MMP-9 (D) in the BMECs was significantly upregulated after the application of sera from NMOSD patients, although it was unchanged after incubation with the sera from MS patients or healthy controls (mean±SEM, n=10, *p<0.001). (E)(F)(G) The changes in the MMP-2 and MMP-9 protein level in BMECs after treatment with sera from NMOSD patients were determined by a western blot analysis. Treatment with the sera from NMOSD patients significantly increased the amount of MMP-2 and MMP-9 protein (mean±SEM, n=10) (E), although the sera from MS patients or healthy controls did not have these effects (mean±SEM, n=10, *p<0.01) (F, G). (H, I) The bar graph reflects the combined densitometry data from each independent experiment (mean±SEM, n=10, MMP-2: p<0.01, MMP-9: p<0.05). (J, K) The presence of anti-BMECs antibodies did not influence the MMP-2 or MMP-9 protein levels secreted by BMECs. (L) Sera collected during the stable phase of NMOSD (patients nos. 5, 8, 9 and 10) did not influence the amounts of claudin-5, MMP-2 and MMP-9 protein in the BMECs. (M) Each bar graph reflects the combined densitometry data from each independent experiment (mean±SEM, n=4). (N) The amounts of claudin-5, MMP-2 and MMP-9 protein in the human blood-nerve barrier (BNB)-derived endothelial cell lines were not changed after exposure to the acute-phase NMOSD sera. (O) Each bar graph reflects the combined densitometry data from each independent experiment (mean±SEM, n=10). Control: non-conditioned DMEM containing 20% FBS; NMOSD: conditioned medium with 10% serum from an NMOSD patient at the acute phase diluted with non-conditioned DMEM containing 10% FBS; MS: conditioned medium with a 10% concentration of serum from an MS patient diluted with non-conditioned DMEM containing 10% FBS; normal: conditioned medium with 10% serum from a healthy control diluted with non-conditioned DMEM containing 10% FBS; NMOSD with anti-BMECs Ab: conditioned medium with 10% serum from an NMOSD patient with anti-BMECs antibodies diluted with non-conditioned DMEM containing 10% FBS; NMOSD without anti-BMECs Ab: conditioned medium with 10% serum from an NMOSD patient without anti-BMECs antibodies diluted with non-conditioned DMEM containing 10% FBS; stable NMOSD: conditioned medium with 10% serum from an NMOSD patient at the stable phase diluted with non-conditioned DMEM containing 10% FBS. MMP, matrix-metalloproteinase.

Purified serum IgG increased the VCAM-1 expression

Figure 4A–E shows the effects of the purified serum IgG obtained from the NMOSD patients on malfunction of the BBB. We randomly selected five NMOSD patients (patients nos. 2, 4, 5, 6 and 9), including four patients with definitive NMO and one patient with isolated LETM, all of whom were positive for anti-BMECs antibodies and had not been treated at the time of serum sampling. The amount of claudin-5, MMP-2, MMP-9 and VEGF in the BMECs did not significantly change following a challenge with the IgG obtained from the five NMOSD serum samples with anti-BMECs antibodies, as determined using a western blot analysis (figure 4A,B). The TEER values and NaF permeability of the BMECs were also not affected following exposure to purified serum IgG fractions obtained from the patients with NMOSD (figures 4C,D). By contrast, the amount of VCAM-1 in the BMECs was significantly increased following the application of IgG obtained from the NMOSD sera with anti-BMECs antibodies, and did not change following exposure to purified IgG obtained from the sera of the healthy controls, as determined using a western blot analysis (figures 4A,B).

Figure 4

Five NMO spectrum disorder (NMOSD) patients (patients nos. 2, 4, 5, 6 and 9), including four patients with definite neuromyelitis optica (NMO) and one patient with isolated LETM were randomly selected. (A) The effects of the purified serum IgG from these NMOSD patients with anti-BMECs antibodies on the expression levels of VCAM-1 and claudin-5 in human BMECs was analysed by a western blot analysis. The amount of VCAM-1 protein in the BMECs was significantly increased after exposure to the purified IgG fractions of sera from NMOSD patients, whereas it was not affected by the purified IgG fractions from healthy controls, as determined by a western blot analysis. The amounts of claudin-5, MMP-2/9 and VEGF protein in BMECs was unchanged after the application of the purified IgG fractions of sera from NMOSD patients, as determined by a western blot analysis. (B) Each bar graph reflects the combined densitometry data from each independent experiment (mean±SEM, n=5, *p<0.05). (C, D) The transendothelial electrical resistance value (C) and the NaF permeability (D) of BMECs did not change after exposure to the purified IgG fraction from NMOSD patients. (E) The effects of reducing the anti-AQP4 antibody titre on the expression of the VCAM-1 protein in BMECs. The reduction of the anti-AQP4 antibody titre did not influence the expression of VCAM-1 in BMECs. (F) Each bar graph reflects the combined densitometry data from each independent experiment (mean±SEM, n=3). Control-IgG: conditioned medium containing purified IgG fractions obtained from fetal bovine serum (FBS); NMOSD-IgG: conditioned medium containing purified IgG fractions obtained from the sera of NMO patients; normal-IgG: conditioned medium containing purified IgG fractions obtained from the sera of healthy individuals; NMOSD: conditioned medium with 10% serum from an NMOSD patient diluted with DMEM containing 10% FBS; NMOSD after AST incubation: conditioned medium with 10% NMOSD sera after a 150 min incubation with astrocytes-expressing AQP4.

Reducing the anti-AQP4 antibody level did not affect the VCAM-1 expression

Figure 4E,F shows that the reduction of the anti-AQP4 antibody titres in the sera did not influence the expression level of VCAM in the BMECs. Following incubation with the hAST-AQP4 cells, a more than 50% decrease was observed in the anti-AQP4 antibody titres in the sera obtained from two different NMOSD patients (see Methods).13 The expression of VCAM-1 in the BMECs was unchanged following treatment with the serum samples with a diminished anti-AQP4 antibody titre (figure 4E,F) compared to that observed following exposure to the untreated samples, suggesting that a 50% or 70% decrease in the anti-AQP4 antibody titre in NMOSD sera is not correlated with a reduction in the VCAM-1 expression level in BMECs.

Discussion

Several lines of evidence suggest that most anti-AQP4 antibodies are produced not intrathecally, but rather peripherally, in patients with NMOSD.6 ,28 In a recent animal study, the systematic injection of serum IgG obtained from NMO patients was insufficient to induce the formation of NMO-like lesions in normal rats, although it was sufficient to cause the disease in animals with a pre-existing inflammatory state in the CNS, such as that which occurs in experimental autoimmune encephalomyelitis (EAE) models mediated by encephalogenic T-cells, or in mice following exposure to Freund's adjuvant.10 ,29 Saadoun et al30 also reported that only direct co-injection of IgG obtained from NMO patients with human complement into the murine brain can induce characteristic histological features of NMO. These findings indicate that BBB damage is required for circulating anti-AQP4 antibodies to enter the CNS and may be involved in the pathogenesis of NMO. This role of BBB damage is also suggested clinically by the finding of increased albumin CSF/serum ratios in anti-AQP4 antibody-positive NMOSD patients.31 The present study demonstrated that sera obtained from NMOSD patients, including both those with definitive NMO and isolated LETM, decreases the BBB function, as determined using our newly established human BMECs. Our previous study showed that sera obtained from definitive NMO patients reduce the BBB function by upregulating VEGF in BMECs.15 Because IgG against BMECs was found in the sera obtained from the NMO patients, we speculated that anti-BMEC antibodies derived from NMO patients cause BBB damage. However, another molecule other than anti-BMECs antibodies that induce BBB damage may be present in NMO sera, because the sera obtained from the NMO patients without anti-BMECs antibodies also decreased the BBB function.15

We hypothesise that the MMP-2 and/or MMP-9 present in NMOSD sera partly contribute to increasing permeability of the BBB. The present study demonstrated that BBB damage following exposure to NMOSD sera was prevented when the cells were pretreated with GM6001, a broad-spectrum MMP inhibitor. Further supporting this possibility, the mRNA expression and protein levels of MMP-2/9 were increased in the cells treated with NMOSD sera, regardless of whether anti-BMEC antibodies were present. However, there were no significant differences in the serum concentrations of MMP-2/9 between the NMOSD patients and healthy controls. These findings suggest that humoral factors other than anti-BMECs antibodies present in the sera obtained from NMOSD patients induce MMP-2/9 secretion by BMECs via an autocrine mechanism, thus inducing BBB damage due to autodegradation of tight junction proteins, including claudin-5. Therefore, even minimal secretion, which does not influence the serum concentration, may have a significant effect. Our previous report demonstrated that the sera obtained from patients with Bickerstaff’s brainstem encephalitis, but not Miller Fisher syndrome, induce BBB damage in the same in vitro BBB model.26 The present study also showed that sera obtained from stable-phase NMO patients did not influence the amount of claudin-5 or MMP-2/9 or the TEER values in this in vitro BBB model. Additionally, the sera obtained from the acute-phase NMO patients did not affect the amounts of these proteins in our in vitro BNB model. These results suggest that the BBB damage observed following exposure to NMO sera due to the autocrine secretion of MMP-2/9 may be a specific and essential event that occurs in the BBB only during the acute phase of NMO. The current study also demonstrated that GM6001 has therapeutic potential for restoring BBB disruption in NMOSD patients. Previous reports have suggested that GM6001 repairs BBB damage in setting of EAE and inhibits the development of clinical EAE.32 However, treatment with GM6001 is not recommended for clinical use at present, because the drug does not appear to be sufficiently selective, which may result in undesirable side effects, and more selective MMP inhibitors have been proposed.33 Novel approaches for restoring the BBB using more selective MMP inhibitors during the acute stage of the disease might achieve promising therapeutic benefits in patient with NMO.

Several studies have demonstrated that the administration of IgG obtained from the sera of NMO patients alone is insufficient to induce the formation of NMO lesions in the absence of inflammation and/or complement.8 ,10 ,27 We thus examined whether purified IgG obtained from NMOSD sera alone without complement has a direct influence on the properties of the BBB. In agreement with the findings of previous studies, the present study demonstrated that IgG derived from NMOSD sera alone does not influence the amount of claudin-5, MMP-2/9 or VEGF proteins, the TEER values or NaF permeability in the BBB. Taken together, these results indicate that unknown humoral factor(s), other than IgG, in the sera obtained from NMOSD patients may disrupt the BBB by inducing the autocrine secretion of MMP-2/9 and VEGF in BMECs; however, we were unable to identify which factor(s) in the NMOSD sera caused these effects, or to clarify whether IgG causes BBB damage only in the presence of complement in the present study.

The interaction of endothelial VCAM-1 with the VLA-4 expressed on leukocytes plays a key role in the transmigration of lymphocytes across the BBB, and is implicated in both the capture and strong adhesion of leukocytes to CNS microvessels.34 ,35 Our results demonstrated that the whole sera and purified IgG obtained from NMOSD patients increase the expression level of VCAM-1. We thus speculate that anti-AQP4 and/or anti-BMECs antibodies in the NMOSD sera are responsible for the upregulation of VCAM-1 in BMECs. A 50% or 75% reduction in the amount of anti-AQP4 antibodies in the sera obtained from NMOSD patients did not influence the ability of the sera to increase the VCAM-1 expression in the BMECs. These results support our hypothesis that anti-BMEC antibodies, other than anti-AQP4 antibodies, may contribute to the upregulation of VCAM-1 in the BBB, thereby causing the extravasation of inflammatory cells into the CNS parenchyma. However, anti-AQP4 antibodies remain a candidate causative factor involved in the disruption of the BBB, as anti-AQP4 antibodies could not be eliminated completely from NMO sera using the methods employed in this study, although they may only cause damage to the BBB in the presence of human complement. Natalizumab works primarily by blocking the interaction of α4-integrins with the VCAM-1 expressed on the endothelial cell surface, thereby preventing the transmigration of lymphocytes across the BBB.36 ,37 Our results suggest that the administration of natalizumab may be clinically effective against NMO. However, several authors have reported that treatment with natalizumab may exacerbate the disease and/or is ineffective in cases of NMO because the drug does not affect the entry of neutrophils into the CNS, which plays an important role in the development of NMO, and because it induces an increase in the number of circulating CD138 plasma cells and mature B cells via the redistribution of lymphocyte subsets in the periphery, causing an increase in the circulating anti-AQP4 antibody level.38–40 Novel approaches for inhibiting the upregulation of VCAM-1 in the BBB without increasing the secretion of circulating anti-AQP4 antibodies are needed to develop novel therapies for NMO.

In conclusion, our study demonstrated that humoral factors other than IgG that are present only in the sera of patients during the acute phase of NMO may decrease the barrier function of the BBB by increasing the autocrine secretion of MMP-2/9 by BMECs. Additionally, IgG, other than anti-AQP4 antibodies, obtained from NMO sera may upregulate the VCAM-1 expression in the BBB. These findings suggest that key molecules trigger the BBB breakdown observed in the pathogenesis of NMO. Increasing understanding of the molecular mechanism(s) responsible for BBB breakdown in patients with NMO may lead to the development of improved therapeutic strategies for treating this severe and currently treatment-refractory disease.

References

View Abstract

Footnotes

  • Contributors AT performed the experiments, analysed and interpreted the data and wrote the manuscript. FS performed the experiments, analysed the data, evaluated the data and edited the manuscript. YS, MF, TT, MA, HH and MA performed the experiments and analysed the data. MK evaluated the data and edited the manuscript. TK conducted and supervised the study, evaluated the data and wrote the manuscript.

  • Funding This work was supported by research grants (Nos. 24790886 and Nos. 22790821) from the Japan Society for the Promotion of Science, Tokyo, Japan and also by research grant (K2002528) from Health and Labor Sciences Research Grants for research on intractable diseases (Neuroimmunological Disease Research Committee) from the Ministry of Health, Labor and Welfare of Japan and also by the Translational Research Promotion Grant from Yamaguchi University Hospital.

  • Competing interests None.

  • Patient consent Obtained.

  • Ethic approval The study was approved by the ethics committee of Yamaguchi University.

  • Provenance and peer review Not commissioned; externally peer reviewed.