Background
Multiple sclerosis (MS) is a chronic inflammatory disease characterized by demyelination and axonal damage leading to chronic neurological deficits. A role of B cells and autoantibodies in disease pathogenesis is supported by numerous observations derived from a wide range of clinical, pathological, and experimental studies. In particular, oligoclonal bands (OCB) are found in the cerebrospinal fluid (CSF) in the majority of MS patients, implying involvement of intrathecal synthesis of immunoglobulin and B cells [
1] even though the antigenic specificity of this response remains unclear [
2]; plasma exchange can lead to disease amelioration in some patients [
3]; histopathological studies demonstrate that many lesions are associated with immunoglobulin and complement deposition [
4]; ectopic B cell follicles are present in the meninges of patients with secondary progressive MS and are implicated in driving disease activity in the underlying parenchyma [
5,
6]; demyelinating and axopathic autoantibodies can be identified in patient sera; and finally, phase II studies with anti-CD20 antibodies targeting B cells, such as Rituximab, have a profound effect on disease activity in patients with relapsing-remitting disease [
7].
Axonal injury and loss are now recognized to be the cause of permanent neurological disability in MS [
8,
9], but the molecular mechanisms responsible for this axonal pathology remain unclear. Evidence is now available implicating contributions from invading inflammatory cells [
10], complement [
11], glutamate excitotoxicity [
12], loss of trophic support provided by myelinating oligodendrocytes [
13], and axopathic autoantibodies [
14]. The relative contribution of these pathomechanisms in MS remains controversial but using a proteomics-based approach, Mathey et al. identified neurofascin as a potential target for an axopathic autoantibody response [
14]. Anti-neurofascin antibody titers were more pronounced in secondary progressive MS than in relapsing-remitting MS, and adoptive transfer of neurofascin-specific antibodies caused rapid worsening of experimental autoimmune encephalomyelitis (EAE), an animal model for MS. This clinical effect was associated with deposition of neurofascin-specific antibody and complement at nodes of Ranvier resulting in inhibition of neurotransmission and axonal damage [
14].
Neurofascin is just one among the increasing number of myelin and non-myelin antigens now implicated in the pathogenesis of MS, but the mechanisms by which these autoimmune responses are initiated are still elusive. Attempts at resolving this problem are confounded by epitope spreading, the ability of the initial pathogenic response to one autoantigen to initiate secondary responses to other tissue-specific targets. This phenomenon has been described extensively in EAE, in which epitope spreading is associated with induction of clinical relapses [
15]. In the case of neurofascin, we hypothesized epitope spreading to involve this autoantigen would be most likely in disease models in which there is extensive demyelination and axonal injury. This is not the case in the classical models of EAE induced by immunization with myelin basic protein (MBP) or MBP peptides, a protocol that results in a monophasic neuroinflammatory disease in rats that is associated with little or no demyelination [
16,
17]. In contrast, immunization with myelin oligodendrocyte glycoprotein (MOG) leads to a relapsing disease with extensive demyelination [
18,
19], a situation predicted to enhance availability of neurofascin epitopes to elicit a secondary response. We therefore investigated if there was a difference in epitope spreading to neurofascin in these two models of EAE in DA rat and if there was any correlation with disease severity. We report that the induction of EAE using recombinant MOG is associated with the development of a neurofascin-specific autoantibody response, which is not observed in animals with MBP peptide 63–88 induced EAE. Screening for this response in a (DAxPVG.1AV1) x DA (hereafter referred as DA backcross) population immunized with MOG revealed that neurofascin-specific antibody levels which correlate with disease severity and that this antibody response was dependent on non-MHC genes. These observations are of great importance in view of the current attempts to extend the study of MS genetics to explore mechanisms determining disease severity.
Methods
Ethics statement
All experiments in this study were approved and performed in accordance with the guidelines from the Swedish National Board for Laboratory Animals and the European Community Council Directive (86/609/EEC) under the ethical permits N338/09, N298/11, and N478/12 entitled “Genetic regulation, pathogenesis and therapy of EAE, an animal model for multiple sclerosis”, which was approved by the North Stockholm Animal Ethics Committee (Stockholms Norra djurförsöksetiska nämnd). The rats were tested according to a health-monitoring program at the National Veterinary Institute (Statens Veterinärmedicinska Anstalt, SVA) in Uppsala, Sweden.
Animals
The DA rats were purchased from Harlan Laboratories (Harlan Laboratories, Venray, The Netherlands). The animals were housed in polystyrene cages containing aspen wood shavings and had free access to standard rodent chow and water. They were immunized with either MOG or MBP
63–88 for induction of EAE and used for T cell proliferation assays ( see the “
Thymidine incorporation assay” for description). Additionally, a DA backcross population was established between DA/Kini and MHC-identical PVG.1AV1 strains, as described previously [
20]. To create the F1 generation, four breeding pairs with DA female founders were established. The N2 generation was bred from eight breeding pairs, with DA females or males crossed to F1 males and females, respectively. Four N2 litters, 421 backcross rats (213 females and 208 males) were subjected to MOG-EAE induction. Assessment of antibodies against neurofascin at day 35 after MOG immunization was performed in 333 (174 females and 159 males) DA backcross rats as 88 of the DA backcross rats either died during EAE or sacrificed according to the ethical considerations.
Induction and determination of disease phenotypes
DA female age-matched rats between 11 and 12 weeks of age were anesthetized with isoflurane (Forene; Abbott Laboratories, Chicago, IL, USA) and immunized by a single subcutaneous injection in the dorsal base of the tail with 200 μl inoculum containing either one of the following five preparations: (a) 20 μg rat recombinant MOG (aa 1–125 from the N terminus, expressed in Escherichia coli and purified to homogeneity by chelate chromatography, hereafter denoted MOG) in phosphate-buffered saline (PBS) (Life Technologies, Paisley, Scotland) emulsified (1:1) with incomplete Freund’s adjuvant (IFA) (Sigma-Aldrich, St Louis, MO, USA), (b) 100 μg guinea pig MBP peptide (aa 63–88 from the N terminus with sequence AARTTHYGSLPQKSQRSQDENPWHF, hereafter denoted MBP63–88) purchased from GL Biochem (Shanghai) Ltd. (Shanghai, China) in PBS emulsified (1:1) with complete Freund’s adjuvant (CFA) (Sigma-Aldrich, St Louis, MO, USA) containing 0.5 mg Mycobacterium tuberculosis (MTB) (DIFCO Laboratories, Detroit, Michigan, USA), (c) 50 μg recombinant rat neurofascin 155 (hereafter denoted rrNF) (R&D systems, Minneapolis, MN, USA) in PBS emulsified (1:1) with CFA containing 1 mg MTB, (d) IFA, or (e) CFA containing 0.5 mg M. tuberculosis.
In another experiment, female and male DA backcross rats between 8 and 16 weeks of age were anesthetized with isoflurane and each animal received a 200-μl inoculum containing MOG (females 12.5–40 μg and males 25–80 μg) in PBS emulsified (1:1) with IFA.
The rats were weighed and monitored daily for clinical signs of EAE, from day 7 until the day of sacrifice at day 56 post immunization (p.i.) for the DA rats and day 35 p.i. for the DA backcross rats. The clinical scoring scale was as follows: 0, no clinical signs of EAE; 1, tail weakness or tail paralysis; 2, hind leg paraparesis or hemiparesis; 3, hind leg paralysis or hemiparalysis; 4, tetraplegia or moribund; and 5, death. The following clinical parameters were used: EAE incidence, clinical signs for more than 1 day; onset of EAE, the first day clinical signs were observed; maximum EAE score, the highest clinical score observed during disease; cumulative EAE score, the sum of all daily clinical scores; duration of EAE, the number of days with EAE; and weight loss (WL0), which is a quantitative trait considered to correlate well with a clinical EAE course and that represents (weight at day 0 p.i.—minimum weight during the experiment)/weight at day 0 p.i.
Serum collection
To determine antibody levels, blood was collected for serum at day 12, day 26, day 41, and day 56 p.i. in the DA rats and at day 12 and 35 p.i. in the DA backcross rats. The rats were anesthetized under isoflurane, and the blood was collected by tail bleeding; at the day of sacrifice, the rats were euthanized under CO2 and the blood was collected through cardiac puncture. The blood was centrifuged at 3600 rpm at 4 °C for 20 min, and serum was collected and preserved at −20 °C until ELISA analysis.
Anti-neurofascin IgG and IgG isotypes determination
ELISA was used to determine anti-rrNF IgG and isotype-specific for IgG1 and IgG2b, anti-MOG IgG, and anti-MBP63–88 IgG. Furthermore, to assess specificity of anti-rrNF IgG antibodies, we measured anti-human neurofascin 155 (hNF155), anti-human neurofascin 186 (hNF186), anti-C-reactive protein (CRP) IgG and anti-interleukin 9 receptor (IL-9R) IgG by ELISA. To investigate cross-reactivity between anti-rrNF IgG and anti-contactin-2 (TAG-1), anti-TAG1 IgG antibodies were assessed by the same method.
Ninety-six-well ELISA plates (Sigma-Aldrich, Roskilde, Denmark) were coated with 10 μg/ml of either of the following: rrNF (R&D systems), hNF155, hNF186 (both produced as described in Ng et al. Neurology 2012), recombinant rat MOG, guinea pig MBP63–88, and recombinant rat TAG-1 (R&D systems), each one of them diluted in 0.1 M NaHCO3 (pH 8.2) (100 μl/well). The coated plates were stored overnight at 4 °C. The plates were washed twice with PBS/0.05 % Tween-20, and free binding sites were blocked with 5 % fat-free milk in PBS/0.05 % Tween-20 for 1 h at room temperature (RT). The rat sera was diluted 1:200 (for anti-rrNF IgG, IgG1, and IgG2b isotypes, for anti-MOG IgG in MBP63–88-EAE rats, for anti-MBP63–88 IgG in MOG-EAE rats, for anti-hNF155 IgG, for anti-hNF186 IgG, for anti-CRP IgG, and for anti-IL-9R IgG), 1:1000 (for anti-MOG IgG in MOG-EAE rats and for anti-MBP63–88 IgG in MBP63–88-EAE rats) in PBS/1 % milk and incubated in duplicates in plates for 1 h at RT. The plates were washed and incubated for 1 h at RT with rabbit anti-rat IgG (1:2000), IgG1 (1:1000), and IgG2b (1:2000) (Nordic, Tilburg, The Netherlands). Unbound antibodies were removed by washing prior to addition of a peroxidase-conjugated goat anti-rabbit antiserum (Nordic) diluted in PBS/0.05 % Tween-20 (1:10,000). After 30 min of incubation, the plates were washed extensively and bound antibodies were visualized with 3,3′,5,5′-tetramethylbenzidine (Sigma). The reaction was stopped by addition of 1 M HCl after 5–10 min in darkness, and the optical density (OD) was read at 450 nm using an Emax Microplate Reader (Molecular Devices, Sunnyvale, USA). Seven twofold dilutions of serum samples from rats with high levels of each antibody detected in a pilot study were included on each plate to create a standard curve and to permit interplate comparisons. The serum samples from the animals immunized with IFA, CFA, or rrNF were also included in each ELISA plate as negative (serum from IFA- and CFA-immunized rats) and positive controls (serum from rrNF-immunized rats).
In the MOG- and MBP63–88-immunized DA rats, we considered the serum specimens to be antibody-positive when the optical density exceeded the cut-off value, which was set at 5 standard deviations (SD) above the mean optical density in serum specimens from IFA- or CFA-immunized DA rats, respectively.
In the DA backcross population, due to the absence of IFA-immunized control animals, we arbitrarily chose to use a more stringent cut-off for antibody positivity. We considered serum specimens to be antibody-positive when OD exceeded the cut-off value, which was set at 15 % of the serum specimens exhibited the highest antibody values. The stringent threshold was also based on testing different cut-off values, e.g., when the serum specimens were considered positive if they exceeded 10 or 20 % of specimens with the highest OD values, and the outcome in our analysis did not differ. Furthermore, OD values lower than the cut-off of 15 % that we set in the analysis clearly represented a background signal.
Adsorption of anti-rrNF IgG antibodies
To examine if there is cross-reactivity between anti-rrNF IgG and anti-TAG1 IgG antibodies, we depleted anti-rrNF antibodies in selected serum samples by performing consecutive adsorptions to plate-bound rrNF. Subsequently, we compared the response against TAG-1 between intact serum and anti-rrNF IgG antibody-depleted serum.
Tissue collection and preparation
To determine T cell response, 28 days after MOG immunization, the DA rats were euthanized under CO2 and the spleen and deep cervical lymph nodes were collected. The spleen and lymph nodes were placed in DMEM (Gibco-BRL, Grand Island, NY), enriched with 10 % fetal calf serum, 1 % glutamine, 1 % penicillin-streptomycin, and 1 % pyruvic acid (all from Life Technologies, Paisley, Scotland) before being mechanically separated by passing through a mesh screen with the bolus of a syringe. Erythrocytes were removed from the spleen samples by lysis with ACK buffer (Gibco, Invitrogen, Karlsruhe, Germany). The samples were resuspended in PBS, and cell numbers were assessed by counting with trypan blue as a dead cell exclusion dye.
Thymidine incorporation assay
To assess T cell proliferation, 400,000 splenocytes/well and 100,000 deep cervical lymph node cells/well were plated in 96-well round bottom plates (Nunc, Roskilde, Denmark) in triplicates. The cells were stimulated with either of the following: medium, 20 μg/ml MOG, 20 μg/ml rrNF (R&D systems), or 1 μg/ml concanavalin A (ConA) (Sigma-Aldrich) for 72 h at 37 °C and 5 % CO2. Eighteen hours before collection, 1 mCi of H3 radioactive thymidine (GE Halthcare, Bucks, UK) was added to each well. The cells were harvested using a Wallac Tomtec (Perkin, Elmer, Waltham, USA) and isotope incorporation was measured using a Wallac TriLux 1450 Microbeta (Perkin Elmer).
Genotype
Genotyping of 333 DA backcross rats was performed on DNA extracted from the tail tip/ear according to a standard protocol. Genotypes were determined by PCR amplification of 137 microsatellite markers. Fluorophore-conjugated primers were used (Applied Biosystems, Eurofins MWG, Operon, Foster City, CA, USA), and PCR products were size fractionated using an electrophoresis capillary sequencer (ABI3730, Applied Biosystems). The genotypes were analyzed using the GeneMapper software (v. 3.7, Applied Biosystems), and two independent observers manually evaluated all genotypes.
Statistical analysis
Spearman’s rank correlations between antibody levels and clinical EAE phenotypes or between antibody levels against other antigens were calculated in GraphPad Prism 5.0 (Graph Pad, San Diego, CA, USA). Differences in T cell response between medium and neurofascin stimulation were calculated in GraphPad Prism 5.0 using non-parametric Mann-Whitney U test. In the DA backcross population, antibody levels between healthy and sick groups and EAE clinical parameters between anti-neurofascin IgG(−) and anti-neurofascin IgG(+) groups were compared using non-parametric Mann-Whitney U test in GraphPad Prism 5.0.
Linkage analysis
The genetic map was defined using publicly available genome sequence (
http://oct2012.archive.ensembl.org/index.html). Linkage analysis was performed with the statistical software R/qtl version 2.11.1. [
21]. A single-quantitative trait loci (QTL) model analysis was performed using Haley-Knott regression model on transformed OD values, and “weight at day 0” and “sex” were used as additive covariates. Permutation tests (
N = 1000) were performed to determine the threshold levels for significant linkage. Differences in allelic effects of the QTL were analyzed with the non-parametric Mann-Whitney
U test using GraphPad Prism 5.0.
Discussion
Epitope spreading has been suggested to be responsible for relapses in EAE and also leads to increased disease severity and chronicity [
15]. While this concept has been generally discussed with respect to T cell reactivities, we hereby demonstrate that this extends to the B cell response in the MOG-induced EAE model, that this response correlates with disease severity, and that is genetically regulated. While we do not address the relapse aspect, our data strongly supports that the presence of an anti-neurofascin response correlates with heightened disease severity, implying that either anti-neurofascin response might contribute to pathology directly or that simply epitope spreading to neurofascin is a reflection of an exacerbated disease per se. While MS is characterized by both demyelination and direct damage to axons/neurons [
25], it is currently not completely clear what leads to axonal injury. Epitope spreading to axonal antigens, among other factors, might account for the neurodegenerative processes and axonal pathology seen in the late, progressive forms of the disease. Such a scenario is consistent with the previous observations of a more frequent presence of anti-neurofascin antibodies in progressive MS [
14].
Epitope spreading likely results from the loss of self-tolerance [
26] driven by tissue damage resulting in increased availability of autoantigens in an inflammatory context. Our data support this concept in that little or no spread to MOG or rrNF-specific B cell responses was observed in the animals with MBP
63–88 induced disease. The fact that disease in the MBP model is transient prevents us from drawing more extensive conclusions in this respect since chronic inflammation in the absence of demyelination might also result in epitope spreading. Nonetheless, it is attractive to consider that extensive demyelination in an inflammatory context such as seen in MOG-induced EAE is required to break/disrupt self-tolerance to neurofascin and other myelin-associated autoantigens.
Neurofascin exists in two isoforms: neurofascin 186 (NF186), a neuronal protein that clusters with voltage-gated Na + channels at the node of Ranvier and plays an important role in saltatory conduction; neurofascin 155 (NF155), an oligodendroglial/myelin protein that maintains the structure of the paranodal junctional complex [
27]. In the present study, we do not distinguish between these two isoforms although our data demonstrate that both are recognized by components of the rrNF-reactive repertoire. Previous studies demonstrate that such antibodies are able to mediate axonal injury by binding to NF186 exposed at the node of Ranvier [
6,
14], but whether this mechanism contributes significantly to enhance disease activity in seropositive animals in our study is an open question. Furthermore, we have observed that the antibody response to rrNF cross-reacts with another member of the Ig superfamily, TAG-1, a recently identified target in MS, which elicits an encephalitogenic T cell response selectively targeting gray matter tracts in experimental animals [
22]. If the immune response to these proteins does overlap, it might mean that together they constitute a molecular hot spot for the development of autoaggressive responses targeting the nodal domains of myelinated axons, but at the same time raises the question as to which might be the primary or dominant target.
It should also be noted that while our data in the DA backcross population demonstrate a significant association between the presence of rrNF-specific IgG/IgG2b responses and disease severity, this may simply reflect that epitope spreading to neurofascin is a consequence rather than a cause of exacerbated disease.
However, irrespective of the pathological significance of the rrNF-specific response in animals with MOG-EAE, our genome-wide linkage analysis demonstrates that this is at least in part controlled by a QTL on chromosome 3 that regulates the rrNF-specific IgG2b response. This technique does not allow fine-mapping down to the level of single genes and requires validation. However, this locus on chromosome 3 overlaps with a QTL we demonstrated previously to regulate EAE severity and the MOG-specific IgG response during the later phase of EAE in the same DA backcross population [
20,
28]. Similarly in a (DAxPVG) x PVG backcross study, this locus was linked to the MOG-specific IgG2c response and disease severity [
20,
28], while the same QTL showed a strong linkage to disease severity and MOG-specific IgG and IgG2b levels in the 10th generation of a DAxPVG advanced intercross line [
20,
28]. In rats, IgG2b and IgG2c are associated with the activation of Th1 dependent responses [
23] that actively participate in the induction of EAE, as opposed to IgG1 that is associated with Th2 reactivity [
23,
24]. These observations imply that the genetic polymorphisms in the locus identified on chromosome 3 may influence the development of Th1-associated autoantibody responses in general, rather than mediating a neurofascin-specific effect, and thereof regulate EAE severity.
Competing interests
The authors declare no conflict of interest.
Authors’ contributions
SF, AOGC, MJ, and TO conceived and interpreted the project. SF performed all the experiments, and MND assisted in the thymidine incorporation experiments. SF did analyses of the experiments. SF, AOGC, and TO wrote the manuscript. All authors edited the final manuscript. All authors read and approved the final manuscript.