Objective To ascertain the hypothesis that the phenotypic differences between Bickerstaff's brainstem encephalitis (BBE) and Miller Fisher syndrome (MFS) are derived from the differences in the effects of sera on blood–brain barrier (BBB) and blood–nerve barrier.
Background Antibodies against GQ1b are frequently detected in BBE and MFS, and these two disorders may share the same pathogenesis, but the clinical phenotypes of BBE and MFS are substantially different.
Methods The effects of sera obtained from BBE patients, MFS patients and control subjects were evaluated with regard to the expression of tight junction proteins and transendothelial electrical resistance in human brain microvascular endothelial cells (BMECs) and human peripheral nerve microvascular endothelial cells.
Results The sera obtained from BBE patients decreased the transendothelial electrical resistance values and claudin-5 protein expression in BMECs, although the sera obtained from MFS patients had no effect on BMECs or peripheral nerve microvascular endothelial cells. This effect was reversed after the application of matrix metalloproteinase (MMP) inhibitor, GM6001. The presence or absence of anti-GQ1b antibodies did not significantly influence the results. MMP-9 secreted by BMECs was significantly increased after exposure to the sera obtained from BBE patients, whereas it was not changed after exposure to the sera obtained from MFS patients.
Conclusions Only the sera obtained from BBE patients destroyed BBB and it might explain the phenotypical differences between BBE and MFS. BBE sera disrupted BBB, possibly via the autocrine secretion of MMP-9 from BBB-composing endothelial cells.
- GUILLAIN-BARRE SYNDROME
- BLOOD-BRAIN BARRIER
Statistics from Altmetric.com
Bickerstaff described the entity known as Bickerstaff's brainstem encephalitis (BBE) with progressive ophthalmoplegia, ataxia and with features suggesting a central nervous system (CNS) disturbance, such as disturbed consciousness or hyperreflexia.1 ,2 Miller Fisher syndrome (MFS) is a variant of Guillain-Barré syndrome (GBS) which is characterised by the triad of ataxia, areflexia and ophthalmoplegia,3 these three syndromes form a continuous spectrum of the same disease process that is triggered by certain microbial infections.4 Anti-GQ1b IgG antibody is a diagnostic marker of MFS5 and BBE,6 and considered to be a pathogenic role in the development of these disorders.
The blood–brain barrier (BBB) and the blood–nerve barrier (BNB) protect the nervous system in the CNS and the nerve fibres in the peripheral nervous system (PNS) from systemic inflammatory reactions and immune responses. Several evidences have demonstrated that disruption of the BNB, causing the leakage of macromolecules like immunoglobulin and cytokines, is a key step in the disease process of GBS.7 ,8
BBE is a rare disorder probably affecting no more than one individual per 1 000 000.9 An autopsy case of a BBE patient demonstrated perivascular lymphocytic infiltration with perivascular oedema in the brainstem, suggesting that disruption of BBB may be pathogenic in the development of BBE.10 In contrast, there is no report describing the pathology of MFS, because it usually shows a benign course11 and tissue samples from the motor nerves are extremely rare. An approach from electrophysiological studies showed that the most consistent findings in MFS are peripheral sensory conduction abnormalities. The distal nerve terminals and nerve roots, where the blood–nerve barrier is anatomically deficient12 may be preferentially affected by immune attacks in acute inflammatory demyelinating polyradiculopathy (AIDP), a subtype of GBS and in case of MFS. It is unknown whether the disruption of BNB is involved in the development of MFS; however, a leaky BBB or BNB, which allows the intrusion of circulating anti-GQ1b IgG antibodies, may play a critical role in the development of BBE or MFS. The phenotypical dissimilarity between BBE and MFS may be derived from the differences in BBB and/or BNB breakdown.
The aim of the present study was to demonstrate the effects of BBE and MFS sera on the impairment of BBB and BNB function, and to clarify the roles of humoral factors, especially matrix metalloproteinase (MMP) and tumour necrosis factor (TNF)-α in the destruction of BBB and BNB.
Materials and methods
The acute phase sera of the illness were obtained from 11 BBE, 10 MFS and 10 GBS patients who were diagnosed at general and teaching hospitals throughout Japan within 3 weeks after the first appearance of symptoms. BBE and MFS patients met the clinical criteria based on the report by Odaka et al,10 and GBS patients met the clinical criteria by Hughes et al.11 Eight of the 11 BBE and 9 of the 10 MFS patients were positive for anti-GQ1b IgG antibodies and 5 of 10 GBS patients were positive for antiganglioside antibodies. The serum samples were collected from one BBE patient during the acute and recovery phase of the illness, after 1½ months from the first symptoms. Samples were stored at −80°C until use, and were incubated at 56°C for 30 min just prior to use. Eight healthy individuals’ sera were served as normal controls. The use of patients’ sera was approved by the ethics committee of Kinki University and Yamaguchi University following the principles of the Declaration of Helsinki. All patients consented to participate in this study.
Cell culture and treatment
The immortalised human microvascular endothelial cells of brain (BMECs) and peripheral microvascular endothelial cells (PnMECs) were generated as described previously.13 ,14 BMECs and PnMECs were treated with culture medium containing 10% serum in a humidified atmosphere of 5% CO2/air. They were also treated with culture medium containing 10% fetal bovine serum (FBS; Sigma, St. Louis, Missouri, USA) which was used as control.
The culture medium for BMECs and PnMECs consisted of Dulbecco's modified Eagle's medium (DMEM; Sigma) supplemented with 10% FBS, 100 U/ml penicillin (Sigma), 100 μg/ml streptomycin (Sigma), 25 ng/ml amphotericin B (Invitrogen, Grand Island, New York, USA.). We purchased polyclonal rabbit anticlaudin-5 and antioccludin antibodies from Zymed (San Francisco, California, USA) and polyclonal rabbit anti-β-tubulin, anti-MMP-2, and anti-MMP-9 antibodies from Santa Cruz (Santa Cruz, California, USA). We also purchased a broad-spectrum MMP inhibitor, GM6001, from Chemicon (Temecula, California, USA) and a neutralising antibody against TNF-α from R&D Systems (Minneapolis, Minnesota, USA).
Transendothelial electrical resistance
The transendothelial electrical resistance (TEER) values of the cell layers were measured with a Millicell electrical resistance apparatus (Endohm-6 and EVOM, World Precision Instruments, Sarasota, Florida, USA) as described previously.15 Measurement of TEER values, serum samples from BBE or MFS patients were randomly assigned before measuring them and the examiner was not informed which samples belonged to which patients. Changes in TEER values were repeatedly measured on different days using the same samples in triplicate.
Western blot analysis
Ten micrograms of protein samples were separated by electrophoresis on 10–12.5% SDS-PAGE (Bio-Rad) gels, and transferred to nitrocellulose membranes (Amersham, Chalfont, UK). Membranes were blocked at room temperature for 2 h with blocking buffer (5% powdered skimmed milk in 25 mM Tris-HCl pH 7.6, 125 nM NaCl and 0.05% Tween 20 (phosphate buffered saline (PBS)-T)) and incubated for 1 h with the relevant primary antibodies (dilution at 1:100). Next, the membrane was incubated with a secondary antibody in PBS-T and 5% milk (dilution 1:1000) at room temperature for 1 h. Membranes were extensively washed in PBS-T and visualised by enhanced chemiluminescence detection (ECL-advance, Amersham, UK). A densitometric analysis was performed using the Quantity one software program (BIO-Rad, Hercules, California). The expression of β-tubulin was used as an internal standard.
Quantitative real-time PCR analysis
Total RNA was extracted from BMECs or PnMECs and single-stranded cDNA was synthesised from 40 ng of total RNA. The samples were subjected to a PCR analysis and the standard reaction curves were analysed as described.15 The sequences of the primers were as follows: forward primer (5′-ACCTGGATGCCGTCGTGGAC-3′), reverse primer (5′-TGTGGCAGCACCAGGGCAGC-3′) for MMP-2,16 forward primer (5′-GTTCCCGGAGTGAGTTGA-3′), reverse primer (5′-TTTACATGGCACTGCAAAGC-3′) for MMP-9,17 forward primer (5′-GTCAACGGATTTGGTCTGTATT-3′), reverse primer (5′-AGTCTTCTGGGTGGCAGTGAT-3′) for Glyceraldehyde-3-phosphate dehydrogenase (G3PDH).18 G3PDH was used as an internal standard.
Treatment with an MMP inhibitor, GM6001
A broad-spectrum MMP inhibitor, GM6001 was prepared for the inhibition study.19 ,20 BMECs were incubated with serum samples containing 25 μM of GM6001 for 12 h on 37°C. TEER values were measured 24 h later, and the total proteins were obtained the next day.
Quantitative analysis of MMP-9 and TNF-α by an ELISA
Serum concentration of MMP-9 and TNF-α and the concentration of MMP-9 in conditioned medium of BMECs were determined in triplicate by an ELISA using commercially available kits (R&D Systems, Minneapolis, Minnesota, USA) and the samples were prepared as described.21 Results were expressed as picograms of MMP-9 or TNF-α per millilitre (pg/ml), based on the standards provided with the available kits.
Treatment with a neutralising antibody against TNF-α
BMECs were incubated with serum samples containing 2.0 μg/ml of a neutralising antibody against TNF-α at 37°C. TEER values were measured, and the total proteins were obtained 24 h later.
IgG purification from serum
The IgG fractions were obtained from serum samples by affinity chromatography using the Melon Gel IgG Spin Purification Kit (Thermo Scientific, Rockford, Illinois, USA). Cells were treated with culture medium containing purified IgG (final concentration 400 μg/ml). TEER values were measured 24 h later, and the total proteins were obtained the next day. Normal FBS-IgG was used as a control antibody.
Statistical values of the TEER for group analysis were obtained using one-way analysis of valiance (ANOVA). Differences in the medians were examined by the Mann-Whitney U test. A two-sided p value of less than 0.05 was considered to be statistically significant.
BBE sera decreased the BBB function
To analyse whether BBE or MFS patients’ sera affect BBB or BNB, we first investigated the effects of BBE or MFS sera on BMECs or PnMECs by evaluating their TEER values. There were no significant changes in TEER values of PnMECs after sera exposure (figure 1A). TEER values of BMECs were significantly decreased after BBE sera exposure, compared with those other than BBE sera exposure (figure 1B). These results indicated that only BBE sera reduced the BBB function. Presence of anti-GQ1b IgG antibodies did not influence the TEER values of PnMECs or BMECs.
BBE sera reduced the amount of tight junction protein in BMECs
We next analysed the expression of tight junction proteins, such as claudin-5 and occludin in PnMECs and BMECs after BBE or MFS or healthy controls’ sera exposure by a western blot analysis. Claudin-5 and occludin levels showed no significant changes after sera exposure in PnMECs (figure 2A,B). In contrast, claudin-5 level in BMECs was significantly reduced after BBE sera application, but not after other than BBE sera exposure (figure 2C,D). Occludin level was not significantly changed after sera exposure in BMECs (figure 2C,D). We next determined the TEER values and claudin-5 level in BMECs after exposure to one BBE patient's serum (BBE 7) during the acute and recovery phase. TEER values and claudin-5 level in BMECs were significantly decreased after the acute phase of BBE serum exposure, although the recovery phase did not lead to any change in BMECs (figure 2E–G). We ascertained the reproduction of the experiment regarding whether BBE sera decrease claudin-5 level in BMECs by the western blot analysis of another four acute-phase BBE sera. Claudin-5 level was also decreased after the additional four BBE sera exposure (figure 2F,G). These data also indicated that only BBE sera disrupt the BBB.
GM6001 reversed BBB disruption caused by BBE sera in BMECs
To clarify the contribution of MMP to the BBB breakdown, we investigated the TEER values and claudin-5 protein level in BMECs after BBE, MFS and healthy individuals’ sera and FBS with or without GM6001 pretreatment by the western blot analysis. TEER values and claudin-5 level in BMECs after BBE sera exposure pretreatment with GM6001 were significantly increased compared with those without GM6001 (figure 3C, E, F), although they were not significantly changed in BMECs after MFS or healthy controls sera exposure pretreated with GM6001 compared with those without GM6001 (figure 3A, B, D–F). These results indicate that GM6001 restores the integrity of BBB only after BBE sera exposure.
BBE sera disrupted BBB through the upregulation of autocrine MMP-9 in BMECs
We next analysed the expression of MMP-2 and MMP-9 in PnMECs and BMECs after BBE or MFS sera exposure by a real-time PCR and the western blot analysis. The mRNA expression of MMP-9 and MMP-2 showed no significant changes after sera exposure in PnMECs (figure 4A). The mRNA expression of MMP-9 and MMP-2 in BMECs was significantly increased after BBE sera application, whereas it was not affected after MFS sera exposure (figure 4B). The amount of MMP-9 and MMP-2 protein in PnMECs showed no significant changes after sera exposure (figure 4C,D). In contrast, the amount of MMP-9 secreted by BMECs was significantly increased after BBE sera exposure, but not after MFS sera exposure (figure 4E,F). The amount of MMP-2 in BMECs was not affected after sera exposure (figure 4E,F). We assessed whether BBE sera increase the secretion of MMP-9 in conditioned medium of BMECs as determined by an ELISA method. However, the concentration of MMP-9 in the conditioned medium of BMECs after BBE sera exposure was under the level of determination (data not shown). Serum concentration of MMP-9 did not differ among BBE and MFS patients and healthy individuals by the ELISA method (figure 4G). There was no correlation between the TEER changes and the serum concentration of MMP-9 in the acute phase of BBE and MFS sera (data not shown).
TNF-α neutralising antibody reversed the BBB disruption caused by BBE sera in BMECs
To demonstrate the contribution of TNF-α to the BBB breakdown, we investigated the TEER values and claudin-5 protein level by the western blot analysis in BMECs after BBE, MFS and healthy controls’ sera pretreated with or without a neutralising antibody against TNF-α. TEER values and claudin-5 level in BMECs after BBE sera exposure pretreated with TNF-α neutralising antibody were significantly higher than those without TNF-α neutralising antibody (figure 5C,E,F). There were no significant changes in TEER values and claudin-5 level in BMECs after MFS or healthy controls’ sera exposure pretreated with TNF-α neutralising antibody, compared with those without TNF-α neutralising antibody (figure 5A,B,D–F). Serum concentration of TNF-α did not differ among BBE, MFS and GBS patients and healthy individuals by the ELISA method (figure 5G).
IgG purified from BBE and MFS sera did not affect the BBB function
We next evaluated the effects of IgG contained in BBE and MFS patients’ sera for the TEER values and claudin-5 protein level by the western blot analysis. There were no significant changes in the TEER values and claudin-5 level of PnMECs and BMECs after incubation with purified IgG from either type of sera. (figure 6A–F).
Since circulating humoral factors, including anti-GQ1b IgG antibodies need to pass through BBB or BNB in order to reach the CNS or PNS parenchyma, we hypothesised that the phenotypical discrepancy between BBE and MFS may be derived from the differences in BBB and BNB breakdown. Namely, BBE sera can open BBB, inducing CNS involvement, whereas MFS sera disrupt BNB. In the present study, we used conditionally-immortalised human BBB-derived and BNB-derived endothelial cells13 ,14 to analyse the effects of BBE and MFS sera on the impairment of BBB or BNB function. We have previously reported that neuromyelitis optica sera disrupted BBB,15 and some growth factors, including vascular endothelial growth factor (VEGF), transforming growth factor (TGF)-β and basic fibroblast growth factor (bFGF) influence the BNB22 using our in vitro BBB and BNB models. We demonstrated that only acute phase of BBE sera opened BBB; TEER values and the expression of tight junction-related protein, claudin-5, now recognised as the most important component involved in maintaining BBB and BNB function,23 in BMECs were significantly decreased only after exposure to the acute phase of BBE sera. Treatment of BMECs with the recovery phase of BBE serum did not lead to any change in claudin-5 protein level or TEER values, thus indicating that soluble factors in BBE sera, which are present only in the acute phase, could disrupt the BBB. In contrast, MFS sera did not influence the BBB function. Previous study suggested that cerebrospinal fluid cells in BBE patients were more abundant than in MFS patients, because the disruption of BBB inducing the invasion of leukocytes into cerebrospinal fluid may be more severe in BBE patients than in MFS patients.10 Our results indicate that only BBE sera disrupt BBB, supporting the perivascular lymphocytic infiltration in brainstem present in autopsy cases of BBE.2 ,10 We also found that the presence of anti-GQ1b antibodies in patients’ sera did not influence the TEER values in BMECs or PnMECs. These results are consistent with our previous observation that only anti-GM1 antibody, not GQ1b antibody, can disrupt an endothelial monolayer of bovine endoneurial origin.24 Although anti-GQ1b antibody is still a candidate for the disruption of CNS/PNS barriers because endothelial cells forming BBB and BNB contain certain amount of GQ1b,25 ,26 we consider that the humoral factors in BBE patients’ sera other than anti-GQ1b antibody may be the key player(s) that upset BBB.
MFS sera did not influence BBB and BNB models in our study. Typical MFS shows abnormal coordination and an absence of deep tendon reflexes caused by interruption of proprioceptive inputs from peripheral sensory organs. Anti-GQ1b antibodies in MFS sera presumably attack responsible organs, including dorsal root ganglia and muscle spindles, which are abundant in GQ1b content.26 Importantly, these two organs are devoid of BNB and anti-GQ1b antibody can easily access the target organs. Another cardinal deficit of MFS is ophthalmoparesis, which is partially explained by the relative abundance of GQ1b ganglioside in oculomotor, trochlear and abducens nerves.27 ,28 Liu et al demonstrated that a monoclonal antibody against GQ1b bound the vast majority of motor end-plates of human oculomotor muscles;29 motor end-plates are also lacking in BNB. To develop MFS phenotypes, anti-GQ1b antibody may play a major role because all of the responsible sites are abundant in GQ1b ganglioside and devoid of BNB; hence, MFS sera are not required to breach BNB. Our results showed that two MFS patients’ sera decreased the TEER values of BMECs, as did the BBE sera. These patients were typical MFS patients showing the triad symptoms described above without features of CNS disturbance. These results might indicate that neurological disturbance could be caused by CNS as well as PNS origin in a very small but certain number of patients with MFS, although MFS sera did not significantly influence our BBB model. We do not have a clear explanation for these results; however, there may be factors other than BBB/BNB disruption that determine the phenotypic differences between BBE and MFS.
We demonstrated that the purified IgG from BBE and MFS sera did not cause the disruption of BBB or BNB. We speculate that soluble factors other than IgG that are present in BBE sera are key mediators of the BBB disruption. An MMP is one of the extracellular matrix components that construct the basement membrane.30 The MMP causing BBB and BNB breakdown may be implicated in the pathogenesis of experimental autoimmune encephalomyelitis (EAE), an animal model of multiple sclerosis31 and experimental autoimmune neuritis (EAN), an animal model of GBS.32 We hypothesised that circulating MMP-2 and/or MMP-9 in BBE or MFS sera were the candidate agents responsible for disrupting the BBB and/or BNB. Kieseier et al demonstrated that MMP-9 was localised around the blood vessels in EAE and in human nerve biopsies from GBS patients,31 ,32 indicating its potential role to disrupt BBB and BNB in these diseases. Créange et al reported that the circulating levels of MMP-9 increased and correlated with the severity in GBS patients, but not MMP-2, during the course of GBS.33 Asahi et al showed that MMP-9 gene knockout rescued mice from ischaemic injury by protecting the degradation of tight junctions in BBB34; however MMP-2 gene knockout did not significantly affect the lesion volumes after cerebral ischaemia, suggesting that MMP-9 plays a deleterious role in the progression of acute tissue damage in their mouse model of cerebral ischaemia.35 The mRNA expression and protein level of MMP-9 were increased, although, the mRNA expression of MMP-2 was increased but MMP-2 protein level was not significantly changed after BBE sera exposure in BMECs. These results indicated that MMP-9 is a key molecule responsible for disruption of BBB in BBE patients. This suggests that soluble factors in BBE sera stimulated MMP-9 secretion by endothelial cells forming BBB, thus causing the disruption of BBB via self-degradation of claudin-5. Serum concentration and the concentration in conditioned medium of BMECs of MMP-9 were not increased in BBE patients; the effect of MMP-9 was via autocrine mechanism and minimal secretion may lead to a significant effect.
Our present studies demonstrated that GM6001 restored the BBB integrity in BBE. Some reports have suggested that the MMP inhibition has potential as therapeutic agent, at least in EAE and EAN, by restoring the impairment of BBB or BNB; EAE treatment with GM6001 could suppress the development of clinical EAE, or even reverse it, through reinstatement of the damaged BBB in the inflammatory phase of the disease.36 EAN treatment with a broad-spectrum MMP inhibitor, BB-1101, from the onset of symptoms could prevent the development of EAN and reduce the disease severity.37 Therapy directed specifically towards BBB repair using MMP inhibitors in the inflammatory phase of the disease might be a promising therapeutic strategy for BBE.
The presence of some circulating cytokines also appears to be linked to the pathogenesis of BBB breakdown in BBE patients. Several studies demonstrated that TNF-α indirectly destabilised BBB by inducing the production of MMP-9 by upregulating the expression of NF-κB. In EAE, combined inhibitors of MMP activity and TNF-α processing can reduce the disease severity.38 Our present studies demonstrated that the reduction of claudin-5 expression in BMECs after BBE sera exposure was restored after adding TNF-α neutralised antibody. These results indicate that TNF-α, which induces the expression of MMP-9 via NF-κB signalling, may be a key molecule responsible for disruption of the BBB in BBE patients. Some reports demonstrated that the serum concentration of TNF-α from AIDP patients or EAN were elevated and correlated directly with disease severity.11 ,39 Our study demonstrated that the serum concentration of TNF-α in BBE, MFS or GBS patients were not significantly increased compared with the healthy individuals. This result does not conflict with our hypothesis, in which endothelial cells that secrete TNF-α in BBE cases with a very small amount might be sufficient for the destruction of BBB, while being insufficient to increase the serum concentration. It is not surprising that the concentrations of TNF-α were not increased in our GBS patients, because Hadden et al reported that its increase was only seen in a very small number of GBS patients (7 of 148) who had an inexcitable result during a nerve conduction study.40
In conclusion, our study is the first to demonstrate that the sera obtained from BBE patients, but not MFS sera, disrupt BBB by increasing the autocrine secretion of MMP-9 in BMECs. Sera from BBE and MFS patients did not disrupt BNB. These data may provide novel pathological explanations concerning the phenotypical differences between BBE and MFS; the disruption of BBB by BBE sera is necessarily for the pathogenesis of BBE but not of MFS. A further analysis of the molecular mechanism(s) underlying the BBB breakdown in BBE should help in the development of therapies for this disabling disorder.
Contributors KS, FS, MK and TK conceived and designed the study and wrote the manuscript. KS, FS and AT analysed the data. All authors reviewed, amended and agreed on the final version of the manuscript.
Funding This work was supported by a research grant for Bickerstaff's brainstem encephalitis (to MK, AK, YN, and TK) from the Ministry of Health, Labour and Welfare of Japan and also by a research grant for neuroimmunological diseases (to SK, TK, and TK) from the Ministry of Health, Labour and Welfare of Japan.
Competing interests None.
Ethics approval This study was approved by the ethics committee of Kinki University and Yamaguchi University.
Provenance and peer review Not commissioned; externally peer reviewed.
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.