|Home | About | Journals | Submit | Contact Us | Français|
Multiple sclerosis (MS) typically manifests in early to mid adulthood, but there is increasing recognition of pediatric-onset MS, aided by improvements in imaging techniques. The immunological mechanisms of disease are largely unexplored in pediatric-onset MS, in part because studies have historically focused on adult-onset disease. We investigated autoantibodies to myelin surface antigens in a large cohort of pediatric MS cases by flow cytometric labeling of transfectants that expressed different myelin proteins. While antibodies to native myelin oligodendrocyte glycoprotein (MOG) were uncommon among adult-onset patients, a subset of pediatric patients had serum antibodies that brightly labeled the MOG transfectant. Antibodies to two other myelin surface antigens were largely absent. Affinity purification of MOG antibodies as well as competition of binding with soluble MOG documented their binding specificity. The prevalence of such autoantibodies was highest among patients with a very early onset of MS: 38.7% of patients less than 10 years of age at disease onset had MOG antibodies, compared to 14.7% of patients in the 10–18 year age group. B cell autoimmunity to this myelin surface antigen is therefore most common in patients with a very early onset of MS.
Multiple Sclerosis (MS)3, a chronic inflammatory demyelinating disease of the CNS, is the most common cause of chronic neurological disability in young adults (1). There has been increasing recognition of pediatric MS over the past two decades, in part due to the advent of MRI techniques which enable sensitive detection of CNS white matter abnormalities (2). It is now estimated that 2–5% of cases of MS begin before an age of 16 (3, 4), but the true incidence of pediatric MS remains unknown. The average age of clinical onset in most pediatric MS cohorts is between 8 and 14 years, but the disease occurs even in children 2–5 years of age. In the majority of pediatric MS patients (>90%), the disease course is relapsing-remitting. Pediatric disease onset is associated with a younger age at disease progression milestones, such as the requirement for mobility aids, than adult-onset MS (3–6). A short interval (less than one year) between the first two demyelinating episodes, incomplete recovery after the first attack as well as a secondary-progressive disease course are unfavorable prognostic factors (7, 8). Cognitive impairment is common in pediatric-onset MS, and can seriously affect academic performance (9–11). Unlike adult-onset MS, which is more common in females and populations with European ancestry, pediatric MS occurs in many ethnic groups and does not have a strong gender bias before puberty (12, 13). An analysis of MS patients cared for in Toronto showed that parents were of non-European descent for a substantially larger percentage of pediatric MS patients (41.9% of mothers and 48.8% of fathers) than for adult-onset MS patients (8.1% of mothers and 7.1% of fathers), and represented ethnic groups from Asia, the Caribbean and the Middle East (13).
Early during the disease course, there can be substantial difficulties distinguishing between pediatric MS and acute disseminated encephalomyelitis (ADEM) (14). ADEM is characterized by an acute onset of widespread CNS inflammation and demyelination and is more common in children than adults (14–16). For a diagnosis of ADEM, new criteria established by the International Pediatric MS Study Group require a polysymptomatic clinical presentation that includes encephalopathy, which is defined as either a behavioral change (such as confusion, excessive irritability) or alteration in consciousness (lethargy, coma) (17). ADEM typically follows a monophasic course with partial or complete recovery. However, recurrent ADEM (defined as a new event of ADEM with a recurrence of the initial symptoms three or more months after the initial event) and multiphasic ADEM (two or more distinct ADEM episodes, each meeting clinical criteria for ADEM but involving new areas of the CNS, separated by at least three months) occur rarely (18). Because the prognosis and treatment regimens for MS and ADEM differ substantially, it is essential to understand the biological similarities and differences between these conditions.
The largest reported study of pediatric demyelinating diseases illustrates this issue. In this study, 296 patients were followed after an initial demyelinating episode; 168 patients (57%) experienced two or more episodes of demyelination and were given a diagnosis of MS (19). While MS was more commonly preceded by clinically isolated syndrome (CIS), 20% of the patients with a final diagnosis of MS were considered to have ADEM at clinical onset. Given these diagnostic difficulties, biomarkers that reflect differences in disease pathogenesis would be valuable. For example, antibodies to aquaporin-4, a water channel enriched on astrocytic foot processes at the blood-brain barrier, are commonly found in neuromyelitis optica (NMO), an inflammatory demyelinating disease that affects the optic nerves and the spinal cord (20). High antibody titers are associated with increased disease activity (21), and the presence of aquaporin antibodies has recently been incorporated into the diagnostic criteria for NMO (22).
Little is currently known about disease mechanisms in pediatric MS, in particular with respect to similarities and differences to adult-onset MS. Recent clinical trials with Rituximab in adult MS patients have shown that B cell depletion results in a pronounced reduction in new demyelinating lesions (23), proving that B cells play a central role in the disease. Oligoclonal IgG antibodies are present in the cerebrospinal fluid (CSF) of the majority of adult-onset MS patients, and CSF analysis of 136 patients with a disease onset before age 16 identified oligoclonal IgG in 92% of cases (24), but the specificity of these antibodies remains unknown. Pathogenic autoantibodies need to be able to bind to structures on the surface of the myelin or oligodendrocytes, but the abundant myelin antigens myelin basic protein and proteolipid protein are inaccessible to antibodies in the intact myelin sheath. MOG, a CNS specific myelin and oligodendrocyte surface protein, is one of the relevant candidate antigens, and a monoclonal antibody (mAb 8–18C5) against MOG induces severe demyelination in mice or rats with mild experimental autoimmune encephalomyelitis (25, 26). While antibodies able to bind to the native conformation of the protein cause demyelination, antibodies to MOG peptides that do not bind the folded protein are not pathogenic (27, 28). It is therefore critical to differentiate between antibodies to folded MOG and denatured or linear epitopes, a task for which traditional assays such as western blots and ELISAs are poorly suited.
We recently reported a radioimmunoassay (RIA) with a tetrameric form of MOG for the detection of MOG autoantibodies in patients with inflammatory demyelinating CNS diseases and found antibodies to MOG in a subset of patients with ADEM (13/69 patients, 18.8%) (29). Interestingly, one of 19 pediatric MS sera gave a strong signal in this RIA, while only one of 109 adult-onset MS sera was positive. This pediatric MS serum sample also labeled a MOG-GFP transfectant, indicating that autoantibodies in this patient indeed bound native MOG. We therefore decided to examine a larger cohort of pediatric and adult-onset MS patients in order to define the relationship between the presence of such autoantibodies and age at disease onset.
Pediatric and adult samples were collected by international neurological centers that care for adults and/or children with MS and other demyelinating diseases. Pediatric MS was defined using the McDonald Criteria (30, 31). None of the cases defined as pediatric MS met the criteria for multiphasic or recurrent ADEM (17). The majority of pediatric-onset MS and pediatric control serum samples were obtained from centers participating in the Wadsworth International Pediatric MS Consortium, which did not include MRI data. Adult-onset MS was defined using the Poser or McDonald criteria (30–32). Most adult-onset MS (including RRMS, SPMS, and PPMS), CIS, and control serum and CSF samples were collected at the Partners MS Center at Brigham and Women’s Hospital, Boston. Serum samples from patients with neuromyelitis optica were from the Hôpital Civil, Strasbourg, France, and viral encephalitis samples were provided by the New York State Department of Health. Each site collected samples using a protocol approved by their Institutional Review Board, and informed consent was obtained from all subjects. Samples were aliquoted and stored at −80°C.
The cloning of human MOG into the pEGFP-N1 vector (Clontech) and transfection of Jurkat cells was previously described (29). Transfected cells were grown in DMEM (Invitrogen) supplemented with 10% FBS (Atlanta Biologicals), 100 IU/mL penicillin, 100µg/mL streptomycin, 35mM HEPES, and 2mM L-glutamine (Mediatech) with 1mg/mL G418 (Invitrogen) to maintain selection. MOG-GFP and GFP clones were generated by sorting single GFPhi cells on a FACS Aria (Becton-Dickinson). After expansion, MOG clones were stained with an antibody to MOG (clone 8–18C5), and MOG-GFP and GFP clones with a matching level of GFP brightness were selected.
Cells were harvested, washed in PBS containing 1% bovine serum albumin (Fisher) (PBS/BSA), and resuspended at a density of 1×106/mL. 50,000 cells (50µL) were incubated with serum at a 1:50 dilution (or other, as indicated) in V-bottom plates (Corning) for 1 hour at 4°C. For CSF stains, 25,000 cells at a density of 2×106/mL were incubated with 50µL of CSF. Cells were washed twice with 200µL of PBS/BSA, and incubated with biotinylated secondary antibodies to human IgG (pan-IgG clone HP-6017, Sigma, or subclass-specific IgG1 clone HP6069, IgG2 clone HP6002, IgG3 clone HP6047, and IgG4 clone HP6023, Calbiochem) at a 1:1000 dilution for 30 minutes at 4°C. Cells were washed twice and incubated with streptavidin-PE (Invitrogen) at a 1:1000 dilution for 20 minutes, washed once with PBS alone and fixed with 1% formaldehyde in PBS at 4°C prior to analysis. 10,000 events per well were recorded on an LSRII instrument equipped with a high-throughput sampler (Becton-Dickinson). Data analysis was performed in FlowJo (Treestar) and Excel (Microsoft). Nonparametric statistical tests were performed in SISA Tables (Quantitative Skills).
The immunoglobulin domain of MOG and the membrane proximal immunoglobulin domain of CD80 were linked to a BirA biotinylation site and cloned into the pET22b vector (Promega). Proteins were expressed in E. coli, refolded from inclusion bodies, purified by HPLC, and biotinylated using the BirA enzyme using published protocols (33, 34). 100µg of biotinylated MOG or CD80 was incubated with 50µL of settled streptavidin-agarose beads (Sigma) in 200µL of PBS/BSA overnight at 4°C on an orbital rotator, and washed to remove unbound antigen. 10µL of settled beads were incubated with 5µL of serum and 200µL of PBS/BSA for 5 hours at 4°C. The supernatant was saved, and the beads were washed four times with 1mL PBS/BSA, followed by elution in 200µL of 0.1M Glycine, pH 2.5. Eluted proteins were separated from the beads by spin filtration (Spin-X 0.2 micron, Corning) and the pH was neutralized by addition of 2M Tris base. 60µL of unbound or eluted proteins were incubated with 50,000 cells in a volume of 125µL with or without 25µg of non-biotinylated recombinant MOG as a competitor, and bound antibodies were detected by FACS as described above.
Antibodies used for immunocytochemistry were purified from sera by protein G immunoprecipitation, eluted with 0.1M glycine pH 2.5, dialyzed into PBS, and biotinylated using sulfo-NHS-biotin (Pierce) at a 50-fold molar ratio of biotin to IgG. MOG and BSA were conjugated to cyanogen bromide-activated Sepharose beads (Amersham) per manufacturer’s instructions and blocked in PBS + 5% BSA overnight. 200µg of total biotinylated IgG was affinity purified on MOG and BSA beads in PBS + 2% BSA overnight, and unbound antibodies were retained prior to the first wash. Bound antibodies were eluted using 0.1M glycine pH 2.5 + 1% BSA as a carrier protein. Bound and unbound antibodies were concentrated to approximately 200µL using Microcon concentrators (Millipore).
Tissue sections from the CNS of six adult tissue donors without CNS disease were from the NeuroResource tissue bank, UCL Institute of Neurology, London, UK. Serial cryostat sections (10µm) were captured onto ply-L-lysine coated slides, acetone-fixed and incubated with blocking sera.
Sections were incubated for time with amount total biotinylated IgG or IgG further purified on MOG or BSA beads. Immunoperoxidase staining was performed with a Vectastain avidin-biotin peroxidase kit (Vector Laboratories) using diaminobenzidine and nickel (II) chloride to give black staining (35). Staining patterns observed with patient IgG were compared to adjacent sections labeled with myelin and oligodendrocyte-specific antibodies, the anti-MOG monoclonal antibody 8–18C5 and the oligodendrocyte-specific monoclonal antibody 14E (36). Primary antibodies were omitted for immunocytochemistry controls.
The pCI-neo vector (Promega) was modified to create a vector suitable for expression of N-terminally tagged type I transmembrane proteins with a fluorescent reporter under control of an internal ribosome entry site (IRES). Oligonucleotides were used to link the H2-Kb signal peptide and HA epitope tag (YPYDVPDYASL) to a unique multiple cloning site containing XcmI, Bsu36I, EspI, and XbaI recognition sequences, which was inserted at the NheI-XbaI restriction sites of pCI-neo. IRES and zsGreen sequences were amplified from the pHAGE-CMV-fullEF1α-IRES-ZsGreen vector (created by Jeng-Shin Lee and provided by the Dana-Farber/Harvard Cancer Center DNA Resource Core) and inserted at a NotI site downstream of the pCI-neo multiple cloning site.
Human MAG and OMG cDNA were obtained from Geneservice Ltd (www.geneservice.co.uk) and used to clone the extracellular, transmembrane, and cytoplasmic domains of each antigen by PCR using Phusion polymerase (New England Biolabs). These sequences, minus the endogenous signal peptide, were cloned into the modified vector with XcmI and XbaI enzymes (New England Biolabs). Jurkat cells were transfected by electroporation and stable transfectants selected with 2mg/mL G418. Cells expressing zsGreen were sorted and stained with a biotinylated antibody to the HA tag (clone 3F10, Roche) and streptavidin-PE (Invitrogen). Single PEhi, zsGreenhi cells were sorted as above, and the clones with highest HA expression were used for autoantibody detection.
Our previous FACS experiments utilized a MOG-GFP construct in which GFP was fused to the cytoplasmic domain of full length human MOG so that the brightness of GFP fluorescence reported on the level of MOG expression. A stably transfected Jurkat cell line had been generated, containing cells with different levels of MOG-GFP expression. FACS analysis of serum samples from ADEM patients showed a diagonal staining pattern for IgG binding versus GFP fluorescence (29), indicating that the labeling intensity closely correlated with the density of the antigen at the cell surface. This finding suggested that the sensitivity of the assay could be improved by single cell cloning of the cells that expressed the highest level of MOG-GFP. We therefore sorted single cells based on GFP fluorescence from MOG-GFP or control GFP transfectants and selected MOG-GFP (abbreviated as MOG) and GFP control clones with similar GFP expression levels for the experiments described here.
To enable sensitive detection of autoantibodies, a three-step staining procedure was used in which 50,000 MOG or control GFP cells were incubated with serum at a 1:50 dilution and bound antibodies were detected using biotinylated anti-human IgG and streptavidin-phycoerythrin (PE). Specific autoantibody binding was expressed as the ratio of the PE mean fluorescence intensity (MFI) of the MOG versus the GFP clone (binding ratio), and ratios greater than 5 were considered to be positive. This threshold is three standard deviations above the mean binding ratio of all non-autoimmune and non-neurological controls (rounded off from the actual value of 5.03). Samples with binding ratios greater than 3 were analyzed for a total of 2 to 5 measurements per sample, depending on the volume of serum available. Sera with average binding ratios greater than 5 were reproducibly positive (Supplemental Figure 1).
Figure 1A shows that labeling of the MOG clone by a number of pediatric MS sera was very bright (binding ratios of >50) and illustrates the range of staining intensities for positive pediatric MS sera. All positive sera labeled the MOG transfectant at a 1:400 dilution and several sera even at a 1:800 dilution (Figure 1B). Comparison of the MOG and control GFP clones for each sample was important because the sera differed in the level of non-specific binding to the Jurkat cells (Figure 1A). The MOG clone gave a higher fluorescence signal than the MOG line, and binding ratios that were typically 5 to 10 fold higher than with the lines (Supplemental Figure 2). Because of their superior properties, these GFP and MOG clones were used in all subsequent analyses. Compared to the tetramer RIA, the major advantages of this flow cytometric assay are that it enables higher throughput and avoids the use of radioactivity.
Sera from patients with pediatric or adult-onset MS as well as control donors from an international group of clinics (Table 1) were tested for antibodies to MOG at a 1:50 dilution. Antibodies to MOG were detected in 28 of 131 pediatric-onset MS samples (21.3%) but rarely found in controls (Figure 2A). In particular, all samples from patients with Juvenile diabetes (n=28) as well as patients with non-neurological diseases (n=37) were negative. Among patients with other neurological diseases (n=34), one (with a developmental delay) was clearly positive while the other was marginally positive.
By comparison, MOG antibodies were only detected in a small subset of adult-onset MS patients (11/254 sera, 4.3%) and the brightness of labeling with these serum samples was substantially lower than with the positive pediatric MS samples. The difference in anti-MOG frequency between all adult- and pediatric-onset MS cases was statistically significant (p=4.48×10−7). All anti-MOG positive pediatric MS sera were from individuals with relapsing-remitting MS, the most common MS subtype in our pediatric group (Table 1). Antibodies to MOG were found in all major clinical subtypes of adult-onset MS, including 6/171 relapsing-remitting, 3/49 primary progressive, and 2/34 secondary progressive patients. Antibodies to MOG were not detected in adults with neuromyelitis optica or viral encephalitis, and were rarely found at low levels in adults with a clinically isolated syndrome (Figure 2A).
Matched serum and CSF samples were available from 13 pediatric; difficulties in obtaining cerebrospinal fluid prevented analysis of more pediatric MS samples. Antibodies to MOG were detectable in the CSF of one pediatric-onset MS patient (binding ratio of 8.63 for CSF, ratio of 23.7 for paired serum) who had a first clinical event at 13.3 years of age. In two other pediatric MS cases with paired samples, MOG-specific antibodies were detectable in the serum (binding ratios of 20.4 and 5.8) but not in paired CSF. It is important to consider that the IgG concentration is ~1000-fold higher in the serum compared to the CSF and that detection in the serum may therefore be more sensitive. Also, it is possible that antibodies in CSF are present as immune complexes and therefore not able to bind MOG in our assay or that a fraction of MOG antibodies are absorbed by their target antigen on oligodendrocytes/myelin.
The pediatric MS samples came from MS centers in six different countries and MOG positive cases were identified for each center (Figure 2B), with some differences in the frequency of positive cases. Among adult-onset MS patients, the frequency of antibodies to MOG differed substantially between centers (0–8.3% of positive sera). While 12 of 43 pediatric MS samples from Canada contained anti-MOG, none of the 53 Canadian adult MS samples were MOG-positive (Figure 2B). Since all of these samples were analyzed by one individual with the same assay, differences in the frequency of MOG-positive adult-onset MS cases reported by different groups may, at least in part, reflect differences in the studied patient populations.
A number of possible factors may be responsible for the presence of MOG antibodies in a subset of pediatric MS patients. No significant association between anti-MOG and gender or ethnicity was found (Supplementary Table 1). However, the presence of antibodies significantly correlated with a younger age at disease onset (p=0.009): 38.7% of individuals with a first clinical event before age 10 had MOG antibodies, compared to 14.7% of patients with a disease onset between 10 to 18 years (Figure 3A). As stated above, the frequency of MS patients with MOG antibodies was even lower in the adult-onset population (4.3%). In adult-onset MS, 3 of the 6 MOG-positive patients for whom data were available had their first clinical event at an age of over 40 years (Figure 3B), and there was no correlation between antibodies and age at first clinical event (p=0.1835). Antibodies to MOG were found in pediatric MS patients with recent disease onset (<1 year) and also those with a longer disease course (>5 years) (Figure 3C), indicating that MOG can be a target in early stages of MS, and that the antibodies may persist, at least for some time. Most patients were either untreated or had received interferons at the time of sample collection (Supplementary Table 2), and the usage of specific immunomodulatory drugs was not significantly different in the MOG positive and negative groups (p= 0.636904).
Based on our previous findings of MOG antibodies in ADEM and the fact that this disease is most common in children, we investigated the relationship between antibodies to MOG and clinical presentation at the first demyelinating event. A significant association was found between anti-MOG and a first clinical event diagnosed as ADEM (p=0.0049): 30.8% of MOG-positive MS patients had an initial diagnosis of ADEM, compared to 8% of MOG-negative cases. Also, 50% of pediatric MS cases with an initial ADEM-like first event had MOG antibodies (Supplementary Table 1).
Antibodies to MOG may cause demyelination and oligodendrocyte death by either complement activation or Fc receptor-dependent mechanisms. IgG subtypes differ in the ability to fix complement and induce antibody-dependent cellular cytotoxicity (ADCC), and we therefore sub-classified MOG-specific IgG antibodies by FACS using a panel of secondary antibodies specific for defined IgG subtypes (IgG1–IgG4). The majority of sera contained only MOG-specific IgG1 (7 of 12 sera), one sample had both MOG-specific IgG1 and IgG3 and another was dominated by IgG2 (Figure 4). The lower sensitivity of the subclass-specific secondary antibodies compared to the pan-IgG antibody prevented identification of the dominant IgG subtype in three samples. Human IgG1 is efficient in fixing complement and inducing antibody-dependent cell-mediated cytotoxicity.
The murine 8–18C5 monoclonal antibody to MOG is well known to bind oligodendrocytes in culture and induce demyelination in vivo. We tested whether antibodies from pediatric MS patients they could compete with 8–18C5 for binding to MOG-GFP cells. Because the secondary antibody used to detect bound IgG recognizes human but not mouse proteins, this assay was not confounded by detection of 8–18C5 on the cells. Addition of 8–18C5 decreased the binding of patient serum antibodies to MOG-GFP cells in a dose-dependent manner for all samples tested (Figure 5A). This result shows that the antibodies from pediatric MS patients bind to surfaces on MOG that significantly overlap with the 8–18C5 epitope.
To further verify the specificity of these antibodies, we affinity purified polyclonal antibodies from sera using a MOG protein preparation representing the extracellular domain that had been refolded from E. coli inclusion bodies. The protein carried a C-terminal BirA tag for site-specific biotinylation, enabling capture onto streptavidin beads. The MOG extracellular domain contains a single Ig domain and we used a related protein of similar size (membrane proximal Ig domain of CD80) as a control. Biotinylated MOG or CD80 were bound to streptavidin-coated agarose beads and incubated with pediatric MS sera. Both bound and unbound antibodies were used to stain MOG and GFP clones. Antibodies capable of staining the MOG clone did not bind to control beads, but were isolated from all four sera using MOG beads (Figure 5B). Addition of soluble recombinant MOG protein as a competitor to the staining reaction substantially reduced the level of specific binding to the MOG clone. The recombinant MOG protein utilized for affinity purification and competition of antibody binding was not glycosylated, indicating that at least a subset of MOG-specific antibodies in pediatric MS sera do not require the N-linked glycan for binding. These experiments firmly establish that the antibodies detected with our flow cytometric assay are indeed specific for MOG.
We next sought to determine if MOG-specific antibodies from pediatric MS patients can bind MOG in the central nervous system (CNS) and therefore performed immunocytochemistry on acetone-fixed human brain sections containing both white and grey matter using biotinylated antibodies detected with avidin-peroxidase. As a positive control, myelin was stained with the anti-MOG monoclonal antibody 8–18C5 (Figure 6A and Supplemental Figure 3A) and antibodies directed against myelin basic protein and Wolfgram protein (not shown). Glial cells with a morphology of oligodendrocytes were immunostained on adjacent sections with the oligodendrocyte-specific antibodies 14E (Figure 6A and Supplemental Figure 3A), CNPase and carbonic anhydrase (not shown).
Total IgG was isolated from anti-MOG positive and negative sera, biotinylated and then further purified using beads with immobilized MOG or the control protein BSA. In samples from two MOG-positive pediatric MS cases (W52 and W24), antibodies affinity purified on MOG beads but not control BSA beads labeled CNS white matter, while no staining was observed with affinity purified antibodies from control donor D14 (Figure 6B). Antibodies from patient W52 also labeled cell bodies in the white matter similar in morphology to cells identified with the oligodendrocyte-specific antibody 14E.
Myelinated axon bundles in the subpial cortical grey matter were stained with the anti-MOG monoclonal antibody 8–18C5 and serum IgG from a previously studied ADEM patient (R4) with high-titer MOG antibodies (29) (Figure 6C). Similar small bundles of myelinated axons and glial satellite cells in the grey matter were stained with antibodies affinity purified on MOG beads from pediatric MS patient W52, but no staining was seen with antibodies eluted from BSA beads (Figure 6C). These glial cells were also 14E-immunopositive (not shown).
We also performed staining with total biotinylated serum IgG not purified on MOG or BSA beads (Supplemental Figure 3). IgG from ADEM patient R4 stained white matter in a myelin-like pattern (Supplemental Figure 3B). IgG from pediatric MS patient W52 labeled myelin and glial cells, while IgG from pediatric MS patient W24 primarily labeled cells with a glial morphology (Supplemental Figure 3B). Weak cell body stains were observed with IgG from a control donor (D14) and MOG negative pediatric MS and juvenile diabetes samples (data not shown). Overall, labeling with MOG-affinity purified antibodies was stronger because the specific antibodies were more concentrated. Also, affinity purification significantly reduced background staining.
The finding that affinity purified MOG-specific antibodies label CNS myelin, myelinated axons and glial cells with oligodendrocyte morphology demonstrates that these serum antibodies recognize their CNS target antigen.
To determine if patients with antibodies to MOG also have circulating antibodies to other myelin surface proteins, we created a set of Jurkat transfectants expressing myelin-associated glycoprotein (MAG), oligodendrocyte-myelin glycoprotein (OMG) or MOG. The proteins carried an N-terminal HA epitope tag to verify expression on the cell surface and clones expressing high levels of the HA epitope tag on the surface and a fluorescent reporter (zsGreen) in the cytosol (Figure 7A) were sorted and used to detect specific antibodies in pediatric MS sera by FACS. Examples of five anti-MOG+ sera are shown in Figure 7B. Antibodies to MAG and OMG were not found in most sera, but when detected only weakly stained the respective transfectant (Figure 7C). Sera from 25 anti-MOG negative pediatric MS patients and 25 pediatric controls did not contain antibodies to OMG or MAG (data not shown). MOG is therefore a more frequent target for autoantibodies in pediatric MS than MAG or OMG.
These results show that circulating high-titer antibodies to MOG are present in a substantial subset of pediatric-onset MS patients, in particular children with a very early disease onset. Stringent criteria establish their antigen specificity: the antibodies label MOG but not control transfectants, they can be affinity purified with recombinant MOG but not a structurally related control protein, and addition of soluble MOG inhibits antibody binding. Antibodies to MOG in pediatric-onset MS patients bound myelin in normal human white matter and myelinated axons in subpial grey matter. The majority of MOG-specific antibodies have the IgG1 subtype that can fix complement and bind to Fc receptors, indicating that they have the potential to damage myelin or oligodendrocytes if they gain access to the CNS. Such antibodies were rarely detected in pediatric controls or patients with adult-onset demyelinating conditions, including MS. In the pediatric patient population, their presence strongly correlated with age at disease onset and an initial ADEM-like presentation, but not with gender or ethnicity. We have begun analyzing autoreactive B cell responses in children with demyelinating events in greater detail. As part of this ongoing study, we have obtained serial serum samples from x children, 5 of whom were positive for antibodies to MOG. In 4 of the 5 cases, serum anti-MOG was detectable in follow-up samples, and persisted for 4 to 28 months after the initial clinic visit. Future prospective studies will examine the relationship between such an antibody response and clinical disease course, MRI features, response to treatment and prognosis.
Previous studies of autoantibodies in adult-onset MS have yielded conflicting results. While some have found antibodies to MOG in patients with MS or CIS (37, 38), others reported no difference in the frequency of anti-MOG between MS and other neurological diseases (39–41) or healthy controls (42). This controversy surrounding the presence of MOG antibodies in MS patients is likely attributable to differences in methodologies, and our flow cytometric approach therefore emphasized detection of antibodies to native, properly folded MOG protein.
Three other groups also generated MOG-expressing cells to examine the presence of autoantibodies in adult-onset MS patients. Haase et al. generated a stable MOG transfectant in LTK3− cells and found that only one of 17 serum samples from adult MS patients labeled this transfectant, even though all patients as well as healthy control subjects had antibodies that detected linear MOG peptides in an ELISA (43). Lavile et al. compared serum antibody staining of a MOG transfected CHO cell line to non-transfected CHO cells and concluded that IgG antibodies specific for native MOG were most frequently found in the serum of patients with CIS and relapsing-remitting MS (44). However, the binding ratios of the MOG and control cells were less than two for all positive samples, even though high serum concentrations were used for staining (1:10 dilution, compared to the 1:50 dilution used here). We defined binding ratios of >5 as positive, and obtained binding ratios as high as 200.2 in pediatric MS samples.
In another study, Zhou et al. transduced a human glioblastoma cell line with a lentivirus encoding MOG and stained these cells at a serum dilution of 1:36 (45). The MOG-specific antibody response was calculated by subtracting median fluorescence intensities obtained with MOG and control transfectants. The authors reported that 32% of adult-onset MS patients and 4% of control subjects had detectable antibodies, but the difference in MFI between the MOG and control cells was rather small (<50) for most sera. By comparison, we observed differences in MFI of 516 to 22,102 for the six representative examples shown in Figure 1A. We prefer to express the data as a binding ratio between the MOG and control transfectants rather than as a difference in MFI because the level of background differs greatly between sera. A sample with a high background and a small MOG-specific increase in binding can have a large change in MFI (for example, MFI values of 600 and 500 for MOG and GFP transfectants, respectively), while a sample with little background and the same MOG binding ratio would have a much smaller change in MFI (for example, 60 and 50). Although both analysis methods permit the identification of samples with high levels of MOG-specific antibodies, using MFI differences can result in the classification of samples with a high background as positive.
The pronounced therapeutic response to B cell depletion with Rituximab shows that B cells play a central role in the pathogenesis of adult-onset MS. It is likely that B cells contribute to disease progression not only as a source of autoantibodies, but also by presenting myelin-derived antigens to autoreactive T cells and driving inflammation through production of cytokines and chemokines. Antigen-specific B cells are highly effective as antigen presenting cells (46, 47), and their elimination could substantially decrease T cell priming and activation. The relative importance of these different mechanisms to the pathogenesis of MS remains unresolved. Also, the specificities of the involved B cells and their antibody products remain largely unknown.
In this study, we have analyzed the largest collection of adult-onset MS sera to date (n=254), and the results show differences in the frequency of MOG-positive cases between samples from MS centers in Canada (0/53, 0%), the US (5/129, 3.9%), and Switzerland (6/72, 8.3%). These apparent differences may be caused by the overall low frequency of antibodies to MOG in adult-onset MS, but differences in patient selection or genetic and environmental factors cannot be excluded. Sample handling may also be a contributing factor, but appears unlikely because we observed that MOG antibodies remain detectable after multiple freeze-thaw cycles.
Our findings support the conclusion that antibodies to MOG are rather uncommon among adult-onset MS patients, while the anti-MOG reactivity identified in a subset of pediatric MS cases is among the strongest autoantibody responses observed so far in MS. Using multimeric forms of folded and glycosylated MOG protein, we previously detected similar antibodies to MOG in a subset of ADEM patients (29). In the current study of pediatric MS, the presence of MOG antibodies strongly correlated with an initial ADEM-like onset, although all children subsequently experienced two or more non-ADEM demyelinating events as required for a diagnosis of MS (17). 50% of pediatric MS patients with an initial diagnosis of ADEM had antibodies to MOG, while 92% of MOG negative pediatric MS patients did not have an ADEM-like initial event. Prospective studies are required to further define the relationship between MOG antibodies and ADEM as well as pediatric MS.
Why have we only detected antibodies to MOG, but not MAG or OMG, which are also myelin surface antigens with large extracellular domains? Chronic inflammation can result in immune responses to multiple self-antigens, a phenomenon referred to as epitope spreading (48). However, this should not be taken to mean that significant immune responses are mounted to every antigen in the target structure. Inherent properties of MOG may contribute to the generation of an antibody response in pediatric MS patients. MOG is a highly encephalitogenic protein in immunization-based animal models of MS and is the only known myelin component to induce pathogenic antibody and T cell responses (49, 50). MOG also has substantial sequence similarity with the milk protein butyrophilin, and antibody as well as T cell cross-reactivity between these antigens has been demonstrated (51, 52).
Why does the prevalence of these autoantibodies change with age at disease onset in pediatric populations? Several general and possibly interrelated factors may be involved: genetic susceptibility, the kinetics and magnitude of the autoimmune response, environmental factors, and the biology of myelination during childhood. In type 1 diabetes, children with particular combinations of MHC class II genes are more likely to become diabetic at a young age and thus appear to carry a higher genetic risk (53). Similarly, children who develop MS at a young age may carry combinations of genes that raise susceptibility to MS to a higher level than in individuals who develop MS later in life. The frequency and functionality of MOG-specific T cells could also affect production of MOG antibodies. The importance of T cell – B cell collaboration was highlighted by recent studies which showed that genetically engineered mice with a high frequency of both MOG-specific T cells and B cells spontaneously develop severe CNS inflammation, while mice with only a high frequency of MOG specific B cells remain healthy (54–56). Lastly, myelination is an ongoing process in children and adolescents, and local changes in the density and composition of myelin affect the distribution of MOG, which is exclusively found on mature myelin segments (57, 58).
Like adult-onset MS, pediatric MS appears to be a heterogeneous disease with complex biological mechanisms. Age at disease onset is an important factor in the specificity of the B cell response to myelin in pediatric MS patients, as MOG antibodies identify a subset of pediatric MS patients with a very early disease onset. Further investigation may define additional patient subgroups with other myelin antibody specificities.
The authors wish to acknowledge the contributions of the site investigators without whom the clinical samples would not have been obtained. Members of the Wadsworth Pediatric MS project include: Silvia Tenembaum, Hospital de Pediatria, Buenos Aires, Argentina; Anita Belman, SUNY, Stony Brook, USA; Alexei Boiko and Olga Bykova, Russian State Medical University, Moscow, Russia; Jayne Ness, Children’s Hospital of Alabama, Birmingham, USA; Jean Mah and Cristina Stoian, University of Calgary, Calgary, Canada; Marcelo Kremenchutzky, London Health Sciences, London, Canada; Mary Rensel, Cleveland Clinic Foundation, Cleveland, USA; Jin Hahn, Stanford University, Stanford, USA; Bianca Weinstock-Guttman and Ann Yeh, SUNY, Buffalo, USA; Kevin Farrell, University British Columbia, Vancouver, Canada; Mark Freedman, University of Ottawa, Ottawa, Canada; Emmanuelle Waubant, UCSF, San Francisco, USA; Martino Ruggieri, University of Catania, Catania, Italy; Matti Iivanainen, University of Helsinki, Helsinki, Finland; Virender Bhan, Dalhousie University, Halifax, Canada; Marie-Emmanuelle Dilenge, Montreal Children’s Hospital, Montreal, Canada.
This work was supported by grants from the National Institutes of Health (PO1 AI045757 to KWW), the National Multiple Sclerosis Society (RG 4122-A-6 to KWW), and the Wadsworth Foundation (to BB, A B-O).
3Abbreviations used in this paper: ADEM, acute disseminated encephalomyelitis; CIS, clinically isolated syndrome; CNS, central nervous system; CSF, cerebrospinal fluid; MAG, myelin-associated glycoprotein; MFI, mean fluorescence intensity; MOG, myelin oligodendrocyte glycoprotein; MS, multiple sclerosis; NMO, neuromyelitis optica; OMG, oligodendrocyte myelin glycoprotein