MHC Class II-Regulated Central Nervous System Autoaggression and T Cell Responses in Peripheral Lymphoid Tissues Are Dissociated in Myelin Oligodendrocyte Glycoprotein-Induced Experimental Autoimmune Encephalomyelitis¹

Experimental autoimmune encephalomyelitis (EAE),⁴ a model for multiple sclerosis (MS), can be experimentally induced in several species (guinea pig, mouse, rat, marmoset) by immunization with components derived from the myelin sheath such as myelin basic protein (MBP), proteolipid protein, and myelin oligodendrocyte glycoprotein (MOG) (1). We have previously shown that active immunization of susceptible rat strains with the extracellular part of MOG leads to an ascending paralysis and a histopathology faithfully mimicking the hallmarks of MS in human CNS, i.e., demyelination and axonal loss (2, 3, 4). This is in contrast to MBP-induced EAE in rats, which is a monophasic disease without the histopathological hallmarks of MS (1). Moreover, we reported that both MHC genes as well as non-MHC genes have a substantial impact on susceptibility to MOG-induced EAE in rats and on the histopathology, whereby MHC class II effects are relative to and can be overcome by effects of non-MHC genes, other MHC genes like the MHC class I gene products, and environmental influences (2, 5).

MOG is a 218-aa-long glycoprotein exposed on the outer surface of the myelin sheath (6). The physiological function of MOG is still unknown (6). Interestingly, MOG composes only ∼0.05% of the myelin sheath and is thought to be CNS specific. Especially the extracellular Ig-like domain of MOG comprising aa 1–125 has gained most interest for neurobiologists and immunologists because of its availability for pathogenic Abs (6, 7). It is encoded within the MHC in humans, mice, and rats (8). There are reports in humans indicating that MOG-directed immunity could cause lesion development in MS (9, 10, 11, 12, 13, 14, 15). This is supported by findings in marmosets, rats, and mice (2, 3, 4, 16, 17, 18). The dissection of genetic and environmental factors leading to detrimental CNS-directed MOG-specific immune attack is a matter of broad interest. Successful manipulations of MOG-directed immunity in a therapeutic sense could potentially benefit MS patients (19). As a prerequisite for successful therapy, target structures for therapy need to be defined that potentially differ in affected individuals depending on genetic allelic variations.

In this study, we systematically investigated the molecular mechanisms for the MHC haplotype influence on MOG EAE using MHC congenic Lewis (LEW) rats and a set of 18 meric overlapping peptides covering the extracellular part of rat MOG (rMOG). We first focused on the rat MHC class II molecules and assessed how the affinity of the peptide-MHC interaction correlated to disease susceptibility. Subsequently, we mapped the immunodominant and cryptic MOG T cell determinants, assessed linear B cell determinants, and tested for encephalitogenicity of defined determinants in the respective MHC congenic LEW rat strains to elucidate the molecular requirements for induction of EAE.

Although the MHC of rat (RT1)^l and RT1^u haplotypes remained insensitive to MOG-peptide disease induction, in both MOG-EAE-susceptible RT1^av1 and RT1ⁿ haplotypes disease could equally be induced by MOG-derived peptides. Typically, immunization with these peptides resulted in disease characterized by progressive paralysis and/or ataxia and a histopathology similar to recombinant rat MOG (rrMOG_1–125) induced EAE. Surprisingly, in the LEW.1N rMOG_91–108, the disease-inducing peptide, was not the immunodominant epitope as mapped by proliferation or IFN-γ enzyme-linked immunospot (ELISPOT) in peripheral lymphoid tissue. Neither could we demonstrate a T2 biased cytokine response in peripheral lymphoid tissue. Although the absolute need for MOG Abs in demyelination is still controversial (7, 20, 21), we could readily detect an Ab response to MOG_91–108 in all MHC congenic LEW rat strains immunized with rrMOG, pointing to a role for this stretch as a T and B cell determinant involved in lesion development. Elution of infiltrating cells from the CNS revealed high mRNA expression for IFN-γ, TNF-α, and IL-4 in LEW.1N rats as assessed by quantitative real time PCR, suggesting a localization of peptide-specific reactivated T cells to this compartment.

We demonstrate a powerful MHC class II regulation of MOG autoaggression in which MHC class II peptide binding and peripheral T cell immunodominance fail to predict which autoantigenic peptides may be relevant for an autoaggressive response. Instead, target organ responses may be decisive and should be further explored, as well as the role of overlapping pathogenic T and B cell epitopes and MHC tetramer technology.

Female rats, 10–14 wk of age, were used in all experiments. All strains have been described (2, 22). ACI rats were originally obtained from Harlan Sprague-Dawley (Indianapolis, IN), PVG-RT1^a rats from Harlan U.K. Limited (Blackthorn, U.K.), DA, LEW, LEW.1A, LEW.1AV1, and LEW.1W from the Zentralinstitut für Versuchstierzucht (Hannover, Germany), and LEW.1N, LEW.1AR1, LEW.1AR2, LEW.1WR1, and LEW.1WR2 from H. Hedrich (Medizinische Hochschule, Hannover, Germany). Subsequently, they were locally bred in filter boxes and routinely tested for specific pathogens. Breeding pairs were checked for homozygosity by examination of a microsatellite marker located within the RT1 region.

The synthetic peptides (Table I) were synthesized by (¹H-benzotriazol-1-yl)-1,1,3,3-tetramethyluronium hexafluorophosphate strategy (Å. Engström, Department of Medical and Physiological Chemistry, University of Uppsala, Uppsala, Sweden). Peptides were purified by reversed-phase chromatography and, subsequently, analyzed by plasma desorption mass spectroscopy. The degree of purity of the used peptides was >99%. The reference peptides for the peptide binding assays were N-terminally biotinylated.

rrMOG, corresponding to the N-terminal sequence of rMOG (aa 1–125) was expressed in Escherichia coli and purified to homogeneity by chelate chromatography (2, 17). The purified protein in 6 M urea was then dialyzed against PBS to obtain a preparation that was stored at −20°C. rrMOG was used in all cell cultures at a concentration of 3 μg/ml. These Ag concentrations had given optimal stimulations in preceding titration experiments. Con A was purchased from Sigma (St. Louis, MO) and used in all cell cultures at a concentration of 1 μg/ml.

The rats were anesthetized by inhalation anesthesia with methoxyflurane (Metofane; Pitman-Moore, Mundelein, IL) and injected intradermally at the base of the tail with a total volume of 200 μl inoculum, containing 50 μg rrMOG in saline or, alternatively, 100 μg of peptide emulsified (1:1) with CFA (Sigma) containing 200 μg or, alternatively, 500 μg heat-inactivated Mycobacterium tuberculosis (strain H 37 RA; Difco, Detroit, MI).

Rats were scored for clinical signs of EAE and weighed daily up to 40 days postimmunization (p.i.) by two alternating investigators. The signs were scored as follows: grade 1, tail weakness or tail paralysis; grade 2, hind leg paraparesis or hemiparesis; grade 3, hind leg paralysis or hemiparalysis; grade 4, complete paralysis (tetraplegia), moribund state, or death.

After short anesthesia, rats were given in each nostril 60 μg of peptide in 60 μl of PBS on days 11, 10, 9, 8, 7, and 6 before induction of active EAE with rrMOG. Controls received PBS without peptide.

Histological evaluation was performed on paraformaldehyde-fixed, paraffin-embedded sections of brains and spinal cords (2, 3, 4). Paraffin sections were stained with hematoxylin and eosin, Luxol fast blue, and Bielschowsky silver impregnation to assess inflammation, demyelination, and axonal pathology, respectively. An inflammatory index was calculated from the number of perivascular inflammatory infiltrates of each rat on an average of 15 complete cross-sections of spinal cord. The degree of demyelination was evaluated for brain and spinal cord sections separately and semiquantitatively described and scored (2, 3, 4). In adjacent serial sections, immunohistochemistry was performed with Abs against the following targets: macrophages/activated microglia (ED1; Serotec, Oxford, U.K.), T cells (W3/13; Seralab, Sussex, U.K.), C9, rat Ig (biotinylated α-rat, Amersham, Little Chalfont, Buckinghamshire, U.K.), and glial fibrillary acidic protein (GFAP; Boehringer- Mannheim, Mannheim, Germany). Bound primary Ab was detected with a biotin-avidin technique. Control sections were incubated in the absence of primary Ab or with nonimmune rabbit serum. The procedures were described (2, 3, 4).

Draining inguinal LN were dissected out under deep anesthesia. LN were disrupted and MNC washed twice in DMEM (Life Technologies, Paisley, U.K.), resuspended in complete medium (CM) containing DMEM supplemented with 1% rat serum, 1% penicillin/streptomycin (Life Technologies), 1% glutamine (Life Technologies), and 50 μM 2-ME (Life Technologies) and flushed through a 70-μm plastic strainer (Falcon; BD Biosciences, Franklin Lakes, NJ). MNC from spleen were prepared in the same way as from LN with the difference that RBC were lysed with lysis buffer consisting of 0.15 M NH₄Cl, 1 mM KHCO₃, and 0.1 mM Na₂ EDTA adjusted to pH 7.4.

MNC were cultured at a concentration of 2 × 10⁶ cells/ml in either 96-well round-bottom microtiter plates (Nunc, Roskilde, Denmark) with 100 μl of cell suspension per well or 24-well flat-bottom plates (Falcon; BD Biosciences) with 1000 μl of cell suspension per well at 37°C in a humidified atmosphere containing 5% CO₂.

All proliferative experiments were performed in triplicate in 96-well round-bottom microtiter plates. MNC (2 × 10⁵/well) in 100 μl CM were cultured with or without the relevant Ag for 60 h and subsequently pulsed with 0.5 mCi [³H]TdR (Amersham Pharmacia Biotech, Uppsala, Sweden) per well for 12 h. DNA was collected on glass fiber filters (Skatron, Sterling, VA), and [³H]TdR incorporation was measured in a beta counter (Beckman Coulter, Fullerton, CA).

To enumerate T cells secreting IFN-γ after Ag exposure, an ELISPOT method was used (2, 23). Nitrocellulose-bottomed 96-well plates (MAHA; Millipore, Molsheim, France) were coated with the mouse mAb DB1 (a generous gift of Peter van der Meide, TNO Primate Center, Rijswijk, The Netherlands), which reacts with rat IFN-γ. Following washing with PBS, the plates were blocked with DMEM containing 5% FCS (Life Technologies). MNC (4 × 10⁵ per well) in 200 μl CM were added to the plates and incubated for 48 h at 37°C in a humidified atmosphere containing 5% CO₂. For each Ag, triplicate determinations were performed. Cells were then discarded, and plates were washed four times with PBS. Secreted and bound IFN-γ was visualized with biotinylated DB12 (also a generous gift of Peter van der Meide), avidin-biotin peroxidase (Vector Laboratories, Burlingame, CA), and subsequently by staining with carbazole (Sigma).

Rats were perfused with cold PBS, and brains and spinal cords were dissected out at day 12 p.i. Subsequently, brains and spinal cords were homogenized in 10 ml 50% Percoll/0.1% BSA/1% glucose (Amersham Pharmacia Biotech) containing 500 U DNase type I (Life Technologies) using a B pistil and holder. Ten milliliters of 50% Percoll were added to each sample after homogenization. A discontinuous Percoll gradient was obtained by adding 7 ml of 63% Percoll below and 20 ml of 30% Percoll above the sample. Samples were centrifuged for 40 min at 1000 × g at 4°C. Lymphocytes were collected from the 63/50% Percoll interface. The cells were subsequently washed twice in 15–25 ml PBS with centrifugation at 600 × g for 15 min at 4°C. One fraction of the cells was immediately used for RNA extraction and the other fraction was cultured for 6 h in DMEM/5% FCS/penicillin/streptomycin/glutamine in the presence of Ag.

Blood samples for Ab measurements were taken at days 12 and 40 p.i. ELISA plates (96-well; Nunc) were coated with 2.5 μg/ml (100 μl/well) rrMOG or 10 μg/ml of peptide (100 μl/well) overnight at 4°C. Plates were washed with PBS/0,05% Tween 20 and blocked for 1 h at room temperature. After washing, diluted serum samples were added and plates were incubated for 1 h at room temperature. Then, plates were washed and rabbit anti-rat antiserum (Nordic, Tilburg, The Netherlands) was added and incubated for 1 h at room temperature. Unbound Abs were removed by washing before the addition of peroxidase-conjugated goat anti-rabbit antiserum (Nordic) diluted in PBS/0.05% Tween 20 (1/10,000). After a 30-min incubation, plates were washed and bound Abs were detected by addition of 3,3′,5′5′-tetramethylbenzidine (TMB, Sigma). The enzymatic reaction was stopped with 1 M HCl after a 15-min incubation in the darkness, and the optical density was read at 450 nm.

ELISA kits for detection of IL-4 and IL-10 (BioSource International, Camarillo, CA) were used with supernatants from MNC that had been incubated at a concentration of 2 × 10⁶ cells/ml with or without the relevant Ag or Con A according to the recommended procedures.

Total RNA was extracted from LN and CNS cells using a RNeasy Mini Kit (Qiagen, Hilden, Germany). To avoid amplification/detection of contaminating genomic DNA, extracted RNA was treated with RNase-free DNase (Promega, Madison, WI). Subsequently, cDNA was synthesized by reverse transcription with Moloney murine leukemia virus reverse transcriptase and random pdN6 primers in the presence of RNase inhibitor (Promega). Amplification was performed on an Applied Biosystems Prism 7700 Sequence Detection System (Applied Biosystems, Foster City, CA) using the SYBR-green method with a two-step PCR protocol (95°C for 10 min followed by 40 cycles of 95°C for 15 s and 60°C for 1 min). All primers were constructed over exon/exon borders (Table II). Relative quantity of mRNA levels was performed using the standard curve method. The amount of mRNA in each sample was calculated as the ratio between the amount of cytokine and the amount of GAPDH in this sample. For Ag-restimulated cells, the cytokine/GAPDH ratio of the control sample (without Ag) was set to 1 and the ratio of other samples was expressed relative to the control.

RT1.B and RT1.D molecules were purified from MHC congenic LEW rat (LEW, LEW.1A, LEW.1N, LEW.1W) thymic and splenic tissues by affinity chromatography using OX-6 (specific for RT1.B molecules) and OX-17 (specific for RT1.D molecules) Abs as described (24). Briefly, tissues were lysed in PBS containing 1% Nonidet P-40 (Boehringer Mannheim, Mannheim, Germany) in the presence of protease inhibitors. The lysates were cleared of nuclei and debris by centrifugation at 40,000 × g for 60 min and passage over a 45-μm filter. The cleared lysates were cycled over OX-6- and OX-17-coupled cyanogen bromide-activated Sepharose-4B (Pharmacia) columns. The columns were washed with 20 column volumes of PBS/0.1% SDS/0.5% Nonidet P-40, 3 column volumes of PBS/0.05% Nonidet P-40, and 3 column volumes of PBS/1% octyl β-d-glucopyranoside (Sigma). Bound MHC molecules were eluted with 0.05 M diethylamine pH 11/0.15 M NaCl/0.1% octyl β-d-glucopyranoside. After neutralization with 2 M Tris-HCl pH 6.3, the purity of the eluted proteins was assessed by SDS-PAGE and subsequent silver staining. The presence of stable MHC class II complexes for each of the haplotypes was confirmed by running the proteins in SDS-PAGE without denaturation through boiling. The protein content was measured with the BCA protein assay (Pierce, Rockford, IL) using BSA as a standard.

Relative affinities of MOG peptides for purified RT1.B and RT1.D molecules were measured by an inhibition ELISA based on a dissociation-enhanced lanthanide fluoroimmunoassay (Wallac, Turku, Finland) (24). Initially, biotinylated tracer peptides were used in a direct binding assay to establish optimal binding conditions for each of the purified RT1.B and RT1.D molecules. In the inhibition ELISA, RT1.B and RT1.D (50–100 nM) molecules were incubated with fixed amounts of their respective tracer peptides (10–50 nM) used in our preceding study (24) in the presence of a range of dilutions of the unlabeled MOG peptides (10-fold dilutions between 1 nM and 100 μM). pH 5 was the optimal pH for binding of the tracer to most of the purified MHC molecules. The binding buffer consisted of a carbonate buffer titrated to pH 5 containing 2 mM EDTA, 0.01% azide, 0.1 mM PMSF, and 0.1% Nonidet P-40 (Boehringer Mannheim). After an incubation of 48 h at 37°C, the peptide-MHC complexes were transferred to Ab-coated (OX-6 or OX-17) ELISA plates (FluoroNunc; Nunc) to remove the excess of nonbound peptides. Europium-labeled streptavidin (Wallac) was added to the plates and incubated for 1 h at room temperature. Finally, the plates were treated with an enhancement solution (Wallac), which releases chelated europium from streptavidin and forms a highly fluorescent solution that can be measured by using a dissociation-enhanced lanthanide fluoroimmunoassay fluorometer (Wallac). The IC₅₀ was determined by plotting the percentage of inhibition vs the concentration of added MOG peptide. Peptides were tested in two to three independent experiments.

To investigate how the affinity between peptide and the restricting MHC molecule might contribute to the immunogenic and encephalitogenic potential of certain peptides in the LEW congenic rat strains, we measured relative affinities of the 18 meric rMOG peptides spanning the extracellular part of rMOG for purified rat RT1.B and RT1.D molecules. RT1.B is supposed to be the rat homolog for HLA-DQ or I-A and RT1.D for HLA-DR or I-E. Both the RT1.B^l and RT1.B^u molecules, which are associated with the MOG-resistant haplotype, mainly bound peptides derived from the C-terminal part of rMOG. In contrast, the RT1.B^a and the RT1.Bⁿ molecules displayed affinity for rMOG peptides covering the whole extracellular part of rMOG (Fig. 1). In general, RT1.D molecules bound a broader range of rMOG peptides and with higher relative affinities than the RT1.B molecules (Fig. 1). Although all the alleles studied were capable of binding several rMOG-derived peptides, the RT1.Bⁿ and RT1.Dⁿ molecules were extremely promiscuous binders with almost all of the rMOG peptides binding to both of the molecules.

Next we studied the MHC haplotype-dependent selection of MOG-immunogenic T cell determinants upon immunization with the extracellular rrMOG. Proliferative responses were measured with [³H]TdR uptake and numbers of cells producing IFN-γ by ELISPOT in response to overlapping rMOG sequence-derived peptides (Fig. 2 A). Peptide-specific responses were in general more sensitively detected by the ELISPOT assay for IFN-γ than proliferation assay. The particular determinants found dominant in the different haplotypes were reproduced in at least four independent experiments. There were different T cell determinants in the four investigated strains, demonstrating the regulation of MHC-related gene products on determinant selection. LEW rats displayed T cell responses to the overlapping peptides MOG_37–54 and MOG_43–60, indicating a dominant determinant in the MOG_37–60 region. LEW.1AV1 rats displayed T cell responses to the MOG_73–90 peptide and to the overlapping MOG_91–108 and MOG_97–114 peptides. In LEW.1N rats the response to MOG peptide 19–36 was dominant. In contrast, LEW.1W rats did not mount clear-cut responses to any of the peptides.

The strikingly different responses in the RT1^av1 or RT1^a and RT1^u rats allowed mapping of the T cell responses within the MHC using RT1^a/RT1^u intra-MHC recombinant rats. The RT1^a- and RT1^av1-dependent T cell determinants were purely dependent on the MHC class II region (RT1.B/D) and not on other genes within the MHC (Table III). In addition, non-MHC genes did not affect the MHC-selected immunodominant responses because those were similar in four rat strains with the RT1^av1 haplotype but different non-MHC genes. There were differences on the qualitative level of proliferative responses and IFN-γ secretion dependent on the non-MHC genes with higher IFN-γ secretion in susceptible strains (data not shown).

We measured Abs of rrMOG-immunized rats to the overlapping MOG peptides to study any MHC haplotype influences on B cell epitope selection. Interestingly, there were reactivities in all strains to peptides MOG_1–18 and MOG_7–24, MOG_91–108, and MOG_109–126 without major MHC-guided predominance (Fig. 2 B).

Two sets of MOG peptide immunization experiments, jointly shown in Table IV, were performed. The first set aimed at studying whether the defined dominant determinants were encephalitogenic as well. The second set studied whether cryptic MOG epitopes could be encephalitogenic. In the first set of experiments we immunized groups of LEW rats with peptide MOG_37–54, LEW.1AV1 rats with peptides MOG_73–90 or MOG_91–108, and LEW.1N rats with peptide MOG_19–36. As shown in Table IV, only LEW.1AV1 rats immunized with MOG_91–108 induced severe disease and demyelinating lesions. Remarkably, this peptide also proved to be a linear B cell determinant (Fig. 2 B).

The second set of MOG peptide immunizations studied MHC haplotype regulation of peptide encephalitogenicity systematically, to investigate cryptic determinants as well. Resistance to rrMOG-induced disease in LEW and LEW.1W rats (2) could be due to lack of generation of a suitable encephalitogenic epitope upon Ag processing. To overcome this restriction, we used each of the 19 available rMOG peptides to immunize the different LEW MHC congenic rats (Table IV). Some of the rats were observed until 40 days p.i., whereas others were sacrificed for histopathological examination already on day 16 p.i. No or only mild disease ensued in LEW or LEW.1W rats upon immunization with the peptides. Only one of three LEW rats developed mild disease to peptide MOG_43–60 and one of eight rats showed a mild inflammation after immunization with the MOG_37–54. One of eight rats had mild inflammation in the spinal cord after immunization with the MOG_91–108 without clinical disease. One of three LEW.1W rats showed slight inflammation of the spinal cord after immunization with peptide MOG_61–78 without paresis. In contrast, the two MHC congenic strains susceptible to rrMOG, LEW.1AV1 and LEW.1N, displayed conspicuous EAE after immunization with certain peptides. However, these were not congruent with those being immunodominant in regard to T cell immunogenicity. Thus LEW.1AV1 and to our surprise LEW.1N rats showed severe clinical as well as histopathological signs of EAE mainly upon immunization with peptides MOG_91–108 and MOG_97–114, but some milder disease and lesions also with peptides MOG_79–96 and MOG_109–126. Furthermore, one of three LEW.1N rats had mild clinical EAE with MOG_31–48 and one of three rats with MOG_43–60, both without detectable infiltrates in the CNS. The clinical signs of EAE induced in diseased rats were rather atypical and included weight loss, severe balance disturbance, and front leg paralysis instead of the more classical flaccid tail and hind leg paralysis described in MBP-induced EAE (2, 3, 23). Disease typically developed between days 15 and 20 p. i., and usually animals recovered and then relapsed in 40% of investigated rats (data not shown). The relapse rate could be higher, but is problematic to assess due to the difficulties in scoring the lesion-associated milder clinical symptoms like optic neuritis, sensory disturbances, and mild bladder dysfunction. The CNS lesions were found with higher incidence in the brain than in the spinal cord. There was a high incidence of optic neuritis. Demyelinating lesions contained complement deposition, indicating Ab-triggered demyelination (Fig. 3). Topography as well as histology mimicked typical MS.

To study the immunogenicity of overlapping MOG peptides and to investigate whether this had any relation to encephalitogenicity we investigated quality of peptide-specific T cell response in terms of proliferation as assessed by ³[H]TdR uptake and IFN-γ secretion as assessed by ELISPOT (Fig. 4,A). The different strains displayed high responses to the immunodominant peptides, defined after MOG immunization. However, in addition there were several cryptic determinants in most of the strains giving high T cell responses in the form of IFN-γ and proliferative indices. Although neither LEW nor LEW.1W rats developed disease upon peptide immunizations, both strains were clearly capable of raising T cell responses to several peptides (Fig. 4,A), which were partly combined with B cell responses to the peptide of immunization (Fig. 4,B). In LEW.1AV1 as well as LEW.1N rats there were Ab responses against MOG_91–108 (Fig. 4,B), which cross-reacted with rMOG_1–125 (data not shown), suggesting that pathogenic, potentially demyelinating Abs binding to surface-exposed full-length MOG in vivo could have a role in disease induced with MOG_91–108. Interestingly, there was no detectable Ag-induced T cell response to MOG peptide 91–108 in the LEW.1N rat strain in terms of IL-4 and IL-10 secretion as assessed by ELISA in peripheral lymphoid tissue (data not shown). Real time quantitative PCR for assessment of Ag-induced mRNA expression of IFN-γ, TNF-α, and IL-4 did not reveal T cell reactivity in LEW.1N rats but showed very high IFN-γ mRNA expression in LEW.1AV1 rats (data not shown). There was no Ag-induced TGF-β mRNA detected in LEW.1AV1 and LEW.1N rats (data not shown). Kinetic investigations on days 7, 9, 12, and 16 p.i. did not show T cell responses in LEW.1N rats (each time point n = 4). At the time point of immunization, we titrated peptide MOG_91–108 (1, 5, 10, 20, 50, and 100 μg MOG_91–108 in CFA at immunization, n = 4 rats each peptide dose) to assess the possibility that the height of the Ag dose could have an impact on MOG peptide 91–108-specific recall responses in peripheral lymphoid tissue in LEW.1N rats. Also, this did not result in detection of IFN-γ-secreting cells in peripheral lymphoid tissue (data not shown). Finally, we eluted infiltrating cells from CNS of LEW.1AV1 and LEW.1N rats, enriched these for lymphoid cells with Percoll gradients, and further analyzed them for IFN-γ, TNF-α, and IL-4 mRNA by quantitative real time PCR (Table V). There was mRNA message for IFN-γ, TNF-α, and IL-4 in LEW.1AV1 and LEW.1N rats. The height of the mRNA message was higher in LEW.1N rats compared with LEW.1AV1 rats. This underscores a difference of MHC class II-regulated T cell responses in peripheral lymphoid tissue compared with the target organ tissue in MOG-EAE. Interestingly, restimulation of eluted cells with Ag showed much higher IFN-γ mRNA expression in LEW.1AV1 compared with LEW.1N rats, indicating that T cells had been restimulated in vivo within the target organ to a higher degree in LEW.1N rats compared with LEW.1AV1 rats (Fig. 5).

After demonstrating that MOG peptide 91–108 immunization induced disease in the rats bearing RT1^av1 and RT1ⁿ haplotypes, we were interested in which regions within the MHC permitted disease and whether the peptide-induced disease was subject to non-MHC gene influences. A variety of inbred, MHC congenic, and intra-MHC congenic rat strains were immunized with MOG_91–108 (Table VI). All congenic and intra-MHC recombinant LEW rat strains with the RT1.B^a/D^a alleles in the MHC class II region developed disease and histopathological lesions, while rat strains with the RT1.B^u/D^u alleles in the MHC class II region remained unaffected. There were no apparent influences from the RT1.A or RT1.C region. Immunization of the rat strains with the susceptible RT1.B^a/D^a alleles in the MHC class II region, to study non-MHC gene regulation, demonstrated that all investigated dark Agouti rats developed disease, whereas all five PVG-RT1^a rats were protected and only one of four ACI rats had some mild CNS infiltrates. Thus, non-MHC genes regulate MOG peptide 91–108-induced EAE.

To prove that MOG_91–108 contains the disease-inducing epitope in LEW.1N rats after immunization with rrMOG we nasally tolerized with MOG_19–36, MOG_91–108, or (as control) PBS and induced active EAE with rrMOG. Only LEW.1N rats nasally tolerized with MOG_91–108 showed amelioration of EAE, in contrast to LEW.1N rats tolerized with MOG_19–36, which developed the same severity of disease as PBS-treated controls (Fig. 6). This experiment strongly suggests that MOG_91–108 is the disease-promoting MOG determinant in LEW.1N rats immunized with rrMOG_1–125.

Based on this study we conclude that 1) the immune response against encephalitogenic sequences can differ dramatically in peripheral lymphoid tissue compared with the target organ tissue, the CNS (immunodominance in peripheral lymphoid tissue in MOG-EAE is not linked to disease-inducing potential). Immunization with MOG_19–36 failed to induce disease in LEW.1N, although the ex vivo response to this peptide was dominant in the RT1ⁿ haplotype. Rather, overt clinical disease accompanied by CNS lesions could be induced in this strain with MOG_91–108, a peptide to which no IFN-γ response or T2-biased cytokine response could be measured in peripheral lymphoid tissue but in the CNS); 2) MOG region 91–114 contains the autoaggressive or disease-promoting T cell determinants in LEW.1AV1 as well as LEW.1N rats; 3) the generation of a peptide-specific IFN-γ response or B cell reactivity was not sufficient to induce disease (for example, LEW rats did not develop disease after immunization with MOG_91–108, nor did LEW.1W rats develop disease after immunization with MOG_1–18, although in both cases immunization with these peptides led to an IFN-γ response and an Ab response that was cross-reactive with rrMOG (data not shown)). Possibly, this is due to lack of processing of these peptides in the CNS or to the presence of a regulatory subset of T cells, Ref. 25, 26, 27); and 4) although binding of MOG peptides is a prerequisite for immunogenicity, the predictive value of the MHC class II binding for encephalitogenicity is low. This is in line with earlier studies in proteolipid protein-induced EAE in mice (28). In contrast to this study we show also that peptides without a detectable ex vivo T cell response can induce EAE.

Immunization with MOG_91–108 not only led to different outcomes in the four tested LEW RT1 congenic rat strains. Non-MHC genes affected disease and immune responses as well: in four tested RT1^av1 congenic rat strains, all strains showed responses to this peptide by proliferation, but differences in IFN-γ secretion (data not shown) and disease susceptibility. These differences are the starting point of investigations on both MHC and non-MHC gene-mediated regulation of MOG_91–108-specific responses and susceptibility (5).

Immunodominance of T cell determinants was purely MHC class II molecule guided as we could show in epitope mapping studies after active immunization with rrMOG_1–125 of different MHC congenic, intra-MHC congenic, and inbred rats with different non-MHC genes (29). This observation points to the MHC molecule and its physicochemical properties as major determinator and ‘nursing’ structure for selection of peptides during Ag processing and subsequent presentation to T cells (30). The spectrum of MOG peptides binding to the different allelic variants of RT1.B molecules varied greatly, but to a lower degree for the RT1.D molecules. This is in line with our preceding work in MBP-induced EAE in different MHC congenic LEW rat strains (24). Certain haplotypes, like the RT1ⁿ haplotype, displayed a very promiscuous binding of MOG peptides. All immunogenic peptides bound with intermediate or high affinity to either one or both of the RT1.B or RT1.D molecules. But although several peptides bound well to the MHC molecules, many of these were nonimmunogenic or nonencephalitogenic. It demonstrates that the autoimmune responses to many parts of this protein are tolerized (31, 32, 33). Because, to our current knowledge, MOG is sequestered in the CNS, tolerization could be achieved by homologous proteins such as the butyrophylin and B7 family (8, 34) or host mimicking exogenous agents (23, 35, 36).

Our data argue for differential selection and/or tuning of MOG sequence 91–114- specific pathogenic and regulatory T cells in different MHC haplotypes during tolerance induction or peripheral activation (37). The origin of this haplotype-dependent effect might lie in the level of promiscuity of MOG-derived peptides binding to the haplotype-associated MHC class II molecules and the affinities of these interactions. The lack of a measurable T cell response to MOG_91–108 in LEW.1N rats in peripheral lymphoid tissue but the presence of such a response in the CNS target tissue argues for breakage of tolerance within the target tissue. The exact mechanism that leads to intra-CNS expansion of autoreactive T cells is presently unclear and the subject of ongoing investigations.

Potentially very small numbers of MOG_91–108 autoreactive T cells could be present in peripheral lymphoid tissue that get activated by challenge with Ag but that are not detected ex vivo. They travel into the CNS where they get restimulated and expanded by Ag presented on local APCs. APCs in the CNS might have different stimulatory capacity compared to APC in peripheral lymphoid tissue. The MOG_91–108-specific T cells could also be partially tolerized and express a cytokine or chemokine that we did not investigate.

A discrepancy lies in the fact that, when mapping dominant determinants, immunization of LEW.1N rats with extracellular MOG systematically leads to large amounts of IFN-γ-secreting cells in peripheral lymphoid tissue, whereas none of the extracellular MOG peptides can equal this response. This is also in contrast with the RT1^l and RT1^av1 haplotypes, where the extracellular MOG response is generally lower than in the RT1ⁿ haplotype, but where the dominant determinants show a very clear-cut IFN-γ response (Fig. 2). Because both extracellular MOG and MOG_91–108 can induce severe EAE in LEW.1N rats, this demonstrates that IFN-γ response and ensuing disease do not always correlate. Because rMOG requires protein processing for the generation of an autoantigen-specific T cell response, the role of disease-associated IFN-γ might lie in the diversification of the immune response in forms of increased processing and presentation of different rMOG determinants through increased engagement of a variety of APC-like B cells (38, 39, 40). This might lead to the diversification of the MOG specific immune response by presentation of ‘hidden’ epitopes (41). Moreover, IFN-γ production of T cells might stimulate the production of autoantibodies to conformationally dependent or linear epitopes of MOG (42).

To obtain a pathogenic/encephalitogenic autoimmune response upon immunization with rrMOG_1–125, the response needs diversification toward a determinant within MOG sequence 91–114 because this peptide caused severe histopathological lesions and disease in LEW.1AV1 and LEW.1N rats, regardless of being immunodominant or cryptic. A constitutive level of presentation of this T cell determinant in the CNS is required, otherwise EAE would not develop after immunization with peptides MOG_91–108 and MOG_97–114 (43). Moreover, MOG_91–114-reactive T cells must pass the blood-brain barrier, what preferentially happens if T cells are activated and express the right set of accessory molecules like adhesion molecules (44, 45, 46). Our data indicate that in strains that do not show strong T cell reactivity toward these peptides in the peripheral lymphoid tissue, potentially a strong response at the target organ might exist. This response would be dependent on the local APC and their naturally processed and presented peptide determinants (43, 47).

Both susceptible LEW.1AV1 and LEW.1N strains, as well as the resistant LEW and LEW.1W strains, raised a B cell response upon immunization with MOG peptide 91–108, which was cross-reactive with rMOG (data not shown). Moreover, complement deposition was visible in most of the MOG_91–108- and MOG_97–114-immunized animals, suggesting Ab-triggered demyelination (3, 48, 49). Recently, Genain et al. were able to identify Abs against linear MOG epitopes specifically binding to disintegrating myelin around axons in lesions of acute MS (15). Taken together, these data strongly point toward a pathogenic potential of Abs against MOG_91–108 and MOG_97–114. If these Abs are involved in demyelination (7, 20), this would argue for a qualitative or quantitative difference in the Ab response raised against this peptide dictated by elements within the MHC, which map toward the RT1.B/D region. This might be due to differences in T cell help toward B cells. Alternatively, but not in line with our data, T effector cells might mediate B cell-independent demyelination (21).

In our study, only MOG sequence 91–114 reproducibly induced CNS inflammation and demyelination associated with severe clinical disease. This is in contrast to other studies where also MOG_35–55 induced demyelination and disease in LEW rats (50). The discrepancy might lie in the use of different substrains of LEW rats. Interestingly, MOG_35–55 is considered one of the main encephalitogenic regions of MOG because it is capable of inducing MS-like disease in several mouse strains (18, 51). These studies only tested MOG peptides predicted by computer programs that did not identify MOG_91–108 as a potential encephalitogenic T cell determinant. An exception is the work of Amor et al., who identified MOG_92–106 as encephalitogen in SJL mice by systematic immunization with overlapping peptides covering the extracellular part of MOG (17).

The data have implications for studies of potential pathogenic autoimmune T cell responses in humans in the sense that in this study neither peptide binding nor epitope mapping with proliferation and IFN-γ secretion, T2 cytokine ELISA, and mRNA expression for cytokines were capable of identifying all encephalitogenic determinants in peripheral lymphoid tissue. Current approaches for detection of disease inducing T cells in the periphery in humans might only reveal a limited and not relevant set of epitopes. The data underscore the need to investigate cellular responses within the target organ tissue.

We thank Anna Gustafsson for help with the peptide binding studies.