A New Mouse Model of Autoimmune Ocular Myasthenia Gravis

Abstract

purpose. To establish a novel model of autoimmune ocular myasthenia gravis (oMG) in mice and study the pathogenic mechanisms of oMG.

methods. oMG was induced in HLA-DQ8 transgenic, HLA-DR3 transgenic, major histocompatibility complex (MHC) class II–deficient, C57BL/6, and C57BL/10 mice by immunization with an Escherichia coli plasmid expressing the recombinant human acetylcholine receptor (AChR) alpha subunit.

results. All strains of immunized mice developed ocular myasthenia gravis with varying disease incidence and severity. HLA-DQ8 transgenic mice were highly susceptible to oMG. Mice with oMG had serum autoantibodies to the mouse extraocular AChR, pathologic deposits of IgG, C3, and C5b-C9 in their extraocular and limb neuromuscular junctions, and droopiness of eyelids. HLA-DR3 transgenic and MHC class II–deficient mice were relatively resistant to oMG induced by AChR alpha subunit immunization and had minimal ocular abnormalities.

conclusions. These findings suggest that oMG pathogenesis could be triggered by immunity to the human AChR alpha subunit and that MHC class II molecule is required for human AChR alpha subunit presentation and CD4 cell–mediated anti-AChR antibody class switching. Differential oMG susceptibility observed in DQ8 and DR3 transgenic mice correlated with the intensity of lymphocytes to respond to the human AChR alpha subunit. This new model of oMG will be a valuable tool for studying the mechanism of oMG and gMG pathogenesis in humans and for preclinical therapeutic analysis.

Myasthenia gravis (MG), one of the best-characterized autoimmune diseases, is caused by autoimmune attack to the nicotinic acetylcholine receptor (AChR) at the neuromuscular junction (NMJ). Weakness of the extraocular muscles (EOMs), resulting in diplopia, and of the levator palpebrae, causing droopiness and weakness of lid closure, is the initial sign of MG in most patients, ultimately occurring in nearly all patients. ¹ Approximately 20% of MG patients display only ocular symptoms (oMG), ² but 50% to 70% of oMG patients subsequently develop generalized MG (gMG). It is unclear how oMG develops and progresses to gMG. There may be different genetic, antigenic, physiologic, and immunopathological manifestations between oMG and gMG. gMG-like disease is inducible in C57BL/6 (B6) and HLA class II transgenic mice by immunization with the AChR purified from Torpedo californica or from the human muscle cell line TE671. ^{3 4 5 6} Ocular symptoms and signs of weakness are uncommon in this model but are often a feature in human oMG and gMG. HLA-DQ8 gene polymorphism has been associated with MG, ⁷ and HLA-DQ8 transgenic mice immunized with human AChR (H-AChR) develop predominantly gMG without ocular symptoms. ⁶ We do not have a good animal model of oMG or a model in which both oMG and gMG coexist. 

The H-AChRα subunit exists as two isoforms, isoforms 1 and 2. Isoform 2 has 25 more amino acids (GDMVDLPRPSCVTLGVPLFSHLQNE) inserted at amino acids 58/59 (mature peptide sequence) than isoform 1. The two variants arise from alternative RNA splicing with and without the P3A exon. ⁸ Isoform 1 is differentially expressed only in the skeletal muscle, whereas isoform 2 is constitutively expressed in skeletal muscle, brain, heart, kidney, liver, lung, and thymus. ⁹  

To study the potential for the H-AChRα subunit to induce MG, we immunized and boosted HLA-DQ8 transgenic, HLA-DR3 transgenic, and B6 mice with a full-length recombinant H-AChRα subunit in complete Freund adjuvant (CFA). HLA-DQ8 and DR3 alleles are associated with MG, and HLA-DQ8 and HLA-DR3 transgenic mice are susceptible to experimental autoimmune myasthenia gravis (EAMG) induced by human AChR. ⁶ B6 and C57BL/10 (B10) mice are highly susceptible to EAMG when immunized with Torpedo californica AChR. Therefore, we selected HLA-DQ8, HLA-DR3, B6, and B10 mice to study EAMG pathogenesis after H-AChRα subunit immunization. H-AChRα subunit immunity triggered clinical and pathologic features of oMG in HLA-DQ8 transgenic B6 and B10 mice. HLA-DQ8 transgenic mice immunized with P3A⁺ H-AChRα subunit developed more severe oMG than did those immunized with the P3A⁻ isoform. Moreover, HLA-DR3 transgenic mice were relatively resistant to oMG induction with the H-AChRα subunit than were HLA-DQ8 transgenic mice. 

Materials and Methods

Mice

We used 7- to 8-week-old HLA-DQ8 (Aβ⁰.DQ8) and HLA-DR3 (Aβ ⁰ .DR3) transgenic mice (both transgenic mouse strains are deficient in mouse major histocompatibility complex (MHC) class II and bear only a human HLA-DQ or DR molecule). Transgenic mice express functional HLA-DQ8 or DR3 genes in the B10 background. The transgenic mice were obtained after the tenth backcross generation to B10 mice and were brother/sister mated for more than five generations at the University of Texas Medical Branch (UTMB). In all the transgenic mice, expression of HLA class II and absence of endogenous murine MHC class II molecules were verified by FACS analysis, PCR, or both. ^{6 10} The immune system in HLA transgenic mice developed normally, with appropriate expression and functionality of HLA class II and CD4 molecules. ^{6 10} Eight-week-old C57BL/6 (B6), B10, and MHC class–II deficient (Aβ ⁰ ) mice (in the B6 background) were purchased from the Jackson Laboratory (Bar Harbor, ME). All animal experiments were approved by the UTMB Animal Care and Use Committee and were conducted in adherence with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

Induction and Clinical Evaluation of gMG and oMG

All mice were anesthetized and immunized with human AChR alpha subunit emulsified in CFA (Difco, Detroit, MI) or immunized subcutaneously with CFA at four sites (two hind footpads and shoulders) on day 0. Then they were boosted twice (boosting days for each experiment are specified in Results) subcutaneously at four sites on the back. Mice were terminated 28 days after the third immunization. For clinical examination, mice were left for 3 minutes on a flat platform and were observed for signs of gMG. Clinical muscle weakness was graded as follows: grade 0, normal posture, muscle strength, and mobility; grade 1, normal at rest, with muscle weakness characteristically shown by a hunchback posture, restricted mobility, and difficulty raising the head after exercise, consisting of 20 to 30 paw grips on cage top grid; grade 2, grade 1 symptoms without exercise during observation period on flat platform; grade 3, dehydrated and moribund with grade 2 weakness; and grade 4, dead. oMG was evaluated by the documentation of partial or full closure of one or both eyes. oMG severity was determined by the following scores: 25% closure of one eye, grade 0.5; 50% closure of one eye, grade 1; 75% closure of one eye, grade 1.5; and 100% closure of one eye, grade 2. If both eyes were involved, the scores were multiplied by 2. 

Recombinant Escherichia coli Expressing H-AChRα Subunit

Escherichia coli BL21 (DE3) cells (Novagen, Darmstadt, Germany) were transformed with vector pET-19b with or without cloned H-AChRαP3A⁺ or P3A⁻ cDNA. The transformed colonies were selected on LB medium containing amoxicillin (50 μg/mL) and were grown overnight in a shaker at 37°C. The E. coli transformants were selected on LB_amp (50 μg/mL) agar plates. A single colony was picked to grow overnight in 5 mL LB_amp, and this culture was subsequently used to seed 500 mL LB_amp. After the culture grew to an A₆₀₀ of 0.8 to 1.0, 1 mM isopropyl β-D-1-thiogalactopyranoside was added to induce the synthesis of protein from the lac promoter and were grown at 37°C for another 1.5 hours. Cells were harvested by centrifugation at 8000g for 15 minutes. The cell pellet were washed (3×) in PBS (20 mM Na/Na phosphate buffer containing 150 mM NaCl) at pH 7.3. Cells were mixed with 6 M urea containing phenylmethylsulfonyl fluoride (1 mM), sonicated (10 seconds ×3), and incubated in a shaking water bath at 4°C for 24 hours. Protein concentration was determined by the method of Bradford, using bovine serum albumin as a standard. ¹¹ The cell extract used to immunize mice was checked for the presence of recombinant protein by Western blot analysis using both anti-his tag and anti-AChRα mAbs. 

Western Blot Assay

SDS-PAGE analysis was performed (Mini-Protein gel; Bio-Rad, Hercules, CA) as outlined by the procedure of Laemmli ¹² with some modifications. Samples were mixed with 2× sample loading buffer (6% SDS, 15% 2-mercaptoethanol, 30% glycerol, and 0.3 mg/mL bromophenol blue in 188 mM Tris-HCI, pH 6.8), heated at 90°C for 10 minutes, and separated by 10% SDS-PAGE in a Tris/glycine buffer (25 mM Tris, 192 mM glycine containing 0.1% SDS, pH 8.3). Separated proteins in the gels were electrophoretically transferred onto nitrocellulose membrane at 380 mA for 1 hour, and the membranes were blocked with 5% skimmed milk in Tris buffer saline (20 mM Tris/HCl, 150 mM NaCl, 0.05% Tween 20, pH 7.5) for 2 hours. After they were washed with TBS, the membranes were treated overnight at 4°C with anti-his tag mAb (Novagen) or anti-H-AChRα mAb (Chemicon International Temecula, CA) 3 μg/mL, diluted in TBS containing 1% BSA. Bound antibodies were detected by horseradish peroxidase-conjugated anti-mouse immunoglobulin (Amersham Bioscience, Piscataway, NJ) using the ECL detection system (Amersham) according to the manufacturer’s instructions. The recombinant H-AChRα subunit was overexpressed as histidine-tagged proteins in E. coli, with six histidine residues at the N-terminal. The size of the H-AChRα, both in isoform 2 (P3A⁺) and isoform 1 (P3A⁻), was 53 to 54 kDa (Fig. 1) . The values are close to a calculated molecular mass of mature peptides of isoforms 1 and 2 (437 amino acids and 462 amino acids, respectively) with an additional six histidine residues (51.5 and 53.2 kDa, respectively). E. coli cell lysate (extract) used for immunization contained 0.5% of the purified recombinant H-AChRα, as calculated from purified fraction from nickel column. 

Radioimmunoassay for Anti-AChR Antibody

Serum samples were collected after immunizations with AChR in CFA. The antibody response to H-AChRα, H-AChR, crude extracts of mouse muscle (M-AChR), ocular mouse muscle AChR (O-AChR), and limb mouse muscle AChR (L-AChR) was measured by radioimmunoassay according to a previously described method. ¹³ Anti-AChR antibody levels were expressed in nanomoles of α-bungarotoxin (BTx) binding sites bound per liter of serum for individual mice. 

ELISA for Anti–AChR IgG Isotypes

One microgram H-AChRα (P3A⁺ or P3A⁻), purified H-AChR, or purified M-AChR was coated onto a 96-well microtiter plate (Dynatech Immulon 2; Dynatech Laboratories, Chantilly, VA) with 0.1 M carbonate bicarbonate buffer (pH 9.6) overnight at 4°C. The assay was then performed according to previously published methods. ^{6 14}  

Detection of IgG, C3, and MAC Deposits at the NMJ by Immunofluorescence Microscopy

Immunofluorescence studies were performed on acetone-fixed 7 μm–thick sections obtained from forelimb and EOM samples of mice, as published previously. ¹⁴ Muscle sections were fixed with acetone and NMJ IgG; C3 and MAC deposits were detected using goat anti-mouse IgG (Chemicon International), goat anti-mouse complement C3 (ICN-Cappel, Aurora, OH), or rabbit anti-human C5b-9 (MAC; Calbiochem, San Diego, CA), respectively. The sections were viewed under a fluorescence microscope (IX-70; Olympus, Tokyo, Japan) with a digital camera (DP-11; Olympus). 

Lymphocyte Proliferation in Response to H-AChRα P3A⁺ Protein

Draining lymph nodes were harvested, and single-cell suspensions were prepared from DQ8, DR3, B6, and MHC^−/− (Aβ⁰) mice 7 days after immunization with 20 μg H-AChRα P3A isoform. A total of 3 × 10⁵ cells/well were seeded into round-bottom 96-well plates in RPMI 1640 medium supplemented with 10% FBS, 25 mM HEPES buffer, 3 × 10⁻⁵ M 2-ME, 100 U/mL penicillin, and 100 μg/mL streptomycin and were stimulated with purified HAChRα P3A isoform (1–3 μg/mL). Cells were incubated at 37°C in 5% CO₂ for 4 days with the addition of [3H] thymidine (1 μCi/well) for the final 18 hours of culture. The ³H incorporation was determined in a beta scintillation counter (Beckman Coulter Inc., Fullerton, CA). 

Statistical Analysis

To determine the significance of the observed results, three statistical tests were used. Incidences of clinical oMG and gMG were compared using the Fisher exact test, clinical scores were compared using Mann–Whitney U test, and all other parameters were compared using ANOVA and Tukey post hoc test. 

Results

HLA-DQ8 Transgenic Mice Are Highly Susceptible to oMG

HLA-DQ8, DR3, B6, B10, and Aβ ⁰ mice were immunized with 10 μg H-AChRαP3A⁺ protein in CFA on day 0 and were boosted twice (month apart). Cumulative incidences of oMG and oMG severity scores for all the strains are shown in Table 1 . The highest incidence (89%) of oMG was seen in HLA-DQ8 mice. Differences in the incidences between HLA-DQ8 and HLA-DR3 transgenic mice and HLA-DR3 transgenic and B6, B10, and Aβ ⁰ mice were significant, as evaluated by Fisher exact test (P < 0.0001 for HLA-DQ8 vs. HLA-DR3 and P < 0.01 for B6, B10, and Aβ ⁰ vs. HLA-DR3). As opposed to nonimmunized mice with completely raised eyelids (Fig. 2A) , DQ8 mice with oMG had either partial (Fig. 2B)or severely (completely) closed eyelids (Fig. 2C) . After administration of anticholinesterase, they partially or fully opened their eyes. 

Because HLA-DQ8 transgenic mice were highly susceptible to oMG, we wanted to establish this model. First, we tested various immunizing doses of H-AChRαP3A⁺ protein for the induction of oMG in HLA-DQ8 transgenic mice. HLA-DQ8 transgenic mice (n = 14) were divided into three groups and immunized with 1 μg (four mice), 5 μg (five mice), or 20 μg (five mice) H-AChRαP3A⁺ protein in CFA on days 0, 30, and 60. After the second boost with H-AChRαP3A⁺ protein, 60% of mice exhibited ptosis of one or both eyelids. One of four from the 1-μg group, two of five from the 5-μg group, and three of five from the 20-μg group of H-AChRαP3A⁺–immunized mice developed oMG. Mice with partially or fully closed eyelids opened their eyes after the administration of neostigmine bromide (anticholinesterase) and atropine sulfate, suggesting that ptosis in these mice is a symptom of MG. Some of the sera obtained from the H-AChRαP3A⁺ protein–immunized mice reacted with H-AChRαP3A⁺ and H-AChR proteins (derived from TE671 cells) in the α-BTx radioimmunoassay (Fig. 3) . Reactivity to the self-mouse whole carcass muscle AChR (M-AChR) was minimal. However, the sera reacted effectively to crude extracts of mouse EOM-AChR (O-AChR) and to mouse L-AChR. No significant difference was observed between the serum antibody levels of mice immunized with different amounts of antigens. MG patients with ocular symptoms often have defective eye movements. We expected mice with ptosis to have a similar problem. However, because of the small sizes of the mice, we were unable to evaluate their ocular movements. We assessed NMJ deposits in the EOM of mice to observe possible pathologic involvement of muscles responsible for the movement of eyes. Through immunofluorescence microscopy, IgG, C3, and C5b-C9 (membrane attack complex; MAC) deposits were observed in the NMJ EOM of H-AChRα P3A⁺ protein–immunized mice (Fig. 4) . 

Human Acetylcholine Receptor Alpha Subunit Isoform 2 Immunity Potentiates oMG in HLA-DQ8 Transgenic Mice

Next, we compared the ocular myasthenogenicity of H-AChRαP3A⁺ protein with that of H-AChRαP3A⁻ protein. HLA-DQ8 mice were immunized with 10 μg H-AChRαP3A⁺ (8 mice) or H-AChRαP3A⁻ (10 mice) protein in CFA on days 0, 29, and 51. The onset of oMG symptoms began on day 3 in P3A⁺ protein–immunized mice; mean day of onset was 27.7 ± 19.6 days after secondary immunization. Compared with H-AChRαP3A⁺ protein–immunized mice, the onset of oMG symptoms was delayed (42.7 ± 14.7 days) and was less severe (grade 1 or less in one eye) in H-AChRαP3A⁻ protein–immunized mice (Figs. 5B 5D) . The less severe and significantly delayed onset (P = 0.03) of oMG associated with H-AChRαP3A⁻ protein immunized mice implicates the importance of the 25 amino acid insert present in the P3A⁺ isoform in accelerating autoimmune oMG in HLA-DQ8 transgenic mice. 

oMG Precedes gMG

Mild generalized muscle weakness (grade 1) was observed in significantly more of H-AChRαP3A⁺ protein–immunized mice than in H-AChRαP3A⁻ protein–immunized mice (6/9 compared with 2/10 respectively; P = 0.0085; Figs. 5A 5B ). The onset of mild generalized muscle weakness was also significantly delayed (57.4 ± 3.6 days) in H-AChRαP3A⁻ protein–immunized mice compared with H-AChRαP3A⁺ protein–immunized mice (40.2 ± 21.1 days; P = 0.01; Figs. 5A 5B 5C 5D ). Therefore, oMG develops before generalized muscle weakness in HLA-DQ8 mice immunized with the H-AChRα subunit. Moreover, H-AChRαP3A⁺ protein immunization induces more severe oMG and more frequent gMG than H-AChRαP3A⁻ protein immunization. Ironically, the mice with oMG in both eyes in the P3A⁺ group (three mice) and the P3A⁻ group (two mice) invariably developed milder gMG. Three of eight (37%) H-AChRαP3A⁺ protein–immunized and 5/7 (71%) H-AChRαP3A⁻ protein–immunized HLA-DQ8 mice did not develop gMG and had purely oMG. These data implicate that the probability of developing gMG depends on the severity of oMG in H-AChRα subunit–immunized mice. Further, immune responses to specific antigenic determinants (e.g., GDMVDLPRPSCVTLGVPLFSHLQNE), in the H-AChRα subunit P3A⁺ isoform could determine the outcome of oMG and gMG development and severity. 

Anti–H-AChRα Subunit Antibodies Cross-React with M-AChR

Sera collected on day 58 after second immunization of H-AChRα were assayed for antibodies against the H-AChRαP3A⁺ or P3A⁻ protein, H-AChR, O-AChR, and L-AChR. In the α-BTx assay (Fig. 6) , sera from 4 of 9 (44%) H-AChRαP3A⁺ protein–immunized mice bound H-AChRαP3A⁺ protein, and 7 of 9 (77%) mice sera bound both H-AChR and M-AChR. This significant affinity for the M-AChR indicates the presence of cross-reactive antibodies (autoantibodies) because the H-AChRαP3A⁺ protein has 95.4% identity with the amino acid sequence of skeletal M-AChR (except for the additional 25 amino acids in the P3A⁺ protein). These autoantibodies are presumably directed against the α-subunit of M-AChR. ¹⁵ Sera from 4 of 8 H-AChRαP3A⁻ protein–immunized mice bound to H-AChRαP3A⁻ protein, sera from 5 of 8 of the serum samples bound to H-AChR, and sera from 4 of 8 bound to M-AChR. Therefore, P3A⁻ protein immunization also induced cross-reactive autoantibodies to M-AChR. Given that the EOM-AChR has elevated α subunit expression compared with L-AChR, we wanted to evaluate the existence of any potential differences in P3A⁺ and P3A⁻ immune sera binding to O-AChR compared with L-AChR. Three of nine H-AChRαP3A⁺ protein-immune sera bound O-AChR and L-AChR, whereas 5 of 8 H-AChRαP3A⁻ protein-immune sera bound O-AChR, and 3 of 8 mouse sera bound L-AChR (Fig. 6) . The increased binding of sera obtained from P3A⁻ protein–immunized mice to M-AChR and O-AChR could be attributed to extensive cross-reactivity of the H-AChRαP3A⁻ protein to the M-AChRα subunit (95.4% identity between H-AChRαP3A⁻ protein with M-AChRα subunit). However, the level of serum antibodies to M-AChR, O-AChR, or L-AChR did not correlate with the severity of oMG symptoms. 

Although the α-BTx assay is a very sensitive quantitative assay, autoantibodies directed to the α-BTx binding sites could not be detected because α-BTx occupies that site. Therefore, we also performed ELISA to detect all antibody binding to H-AChRα, H-AChR, and M-AChR and to determine their antibody isotypes. In ELISA, we could detect high levels of serum IgM antibodies to the H-AChRα, H-AChR, and M-AChR in isoform-immunized mice. Most of the HLA-DQ8 transgenic mice immunized with H-AChRαP3A⁺ and the P3A⁻ isoform had antibodies of all isotypes (IgM, IgG, IgG₁, IgG_2b, and IgG_2c) that bound to H-AChRα, H-AChR, and M-AChR (Fig. 7) . Generally, there was no difference in the level of serum anti–H-AChRα, H-AChR, or M-AChR antibody of any isotypes between H-AChRαP3A⁺ and H-AChRαP3A⁻ protein–immunized mice. However, a significant difference (P < 0.05) in anti–H-AChRα IgM response was observed between H-AChRαP3A⁺ and H-AChRαP3A⁻ protein–immunized mice. Moreover, serum levels of anti–H-AChRα IgG antibodies were correlated (P = 0.03, R = 0.59, Spearman nonparametric correlation test) with gMG scores. 

Pathologic IgG, C3, and MAC Deposits in EOM NMJ

The NMJ (α-BTx deposits) of extraocular and limb muscles of H-AChRα–immunized mice exhibited IgG, C3, and MAC deposits (Fig. 8) , suggesting that anti-AChR IgG–mediated C3 and MAC activation plays a pivotal role in NMJ destruction in this oMG model in HLA-DQ8 transgenic mice. Although H-AChRαP3A⁻ protein–immunized mice had less EOM MAC, there was no significant difference between the IgG, C3, and MAC deposits of mice immunized with H-AChRαP3A⁺ or P3A⁻ protein. However, EOM sections contained more C3 (P = 0.005) and MAC (P = 0.02) deposits than limb muscles, possibly because of poorer complement-inhibiting mechanisms of EOM. ¹⁶ Generally, IgG, C3, or MAC deposits of H-AChRα subunit–immunized mice are three to four times lower than expected from the limb muscle samples of Torpedo AChR–immunized B6 mice (our unpublished observations, 2006). It is possible that in human MG, initial autoimmunity against the H-AChRα induces ocular, mild, generalized MG with few deposits at the NMJ. After production of autoantibodies against other subunits (because of AChR destruction and exposure of all T and B cell epitopes for recognition by T and B cells), the clinical course of MG advances to a more severe and generalized status with full-scale inflammation with IgG and complement deposition in skeletal muscle NMJ. NMJ deposit (IgG, C3, MAC) values were compared with the clinical gMG and oMG scores of individual mice using Spearman nonparametric correlation, and no significant correlation could be found between the amounts of deposits and the severity of clinical disease. 

E. coli Extract without the H-AChRα Subunit Insert (E. coli Plasmid) Fails to Induce oMG in HLA-DQ8 Transgenic Mice

H-AChRαP3A⁺ protein is expressed in E. coli plasmid; therefore, the antigen solution used for immunizations contained E. coli proteins other than H-AChRα subunit (as observed by gel electrophoresis; data not shown). We wondered whether E. coli with plasmid could be potentiating the ptosis-inducing effect of H-AChRα. Even Mycobacterium tuberculosis found in CFA could have a similar effect. To determine whether E. coli extract with plasmid or CFA could be at least partially responsible for the ocular signs, HLA-DQ8 transgenic mice were immunized with 20 μg H-AChRαP3A⁺ protein in CFA (10 mice), E. coli plasmid in CFA (6 mice), or CFA only (6 mice) (Fig. 9) . Eighty-nine percent of HLA-DQ8 mice immunized with H-AChRαP3A⁺ protein in CFA developed partial (variable rates) or full closure (ptosis) of one or both eyelids. Interestingly, E. coli plasmid–immunized HLA-DQ8 transgenic mice had low levels of serum anti–M-AChR Ab and NMJ deposits (Fig. 9) . These results suggest that immunization with E. coli with plasmid induces the production of small amounts of antibodies against mouse AChR and NMJ deposits. However, E. coli plasmid or CFA immunity did not induce ocular symptoms of oMG, suggesting that an H-AChRα insert in the plasmid is required for significant pathologic deposits of IgG, C3, and MAC and for the development of oMG symptoms. 

HLA-DR3 Transgenic Mice Are Resistant to oMG

B6 (n = 10) and HLA-DR3 transgenic (n = 8) mice were immunized with H-AChRαP3A⁺ protein in CFA. HLA-DR3 transgenic mice were highly resistant to oMG induction (Table 1 ; Fig. 10 ). Compared with HLA-DR3 transgenic mice, HLA-DQ8 transgenic mice and B6 mice also had significantly higher serum anti–M-AChR Ab levels (Figs. 9 10)and increased amounts of NMJ C3, IgG, and MAC deposits at their EOM and limb muscles (Figs. 9 10) . 

MHC Class II Molecule in oMG

To assess the contribution of MHC class II molecules in oMG pathogenesis, B6, B10, and Aβ0 mice in the B6 background (10 each; Jackson Laboratory) were immunized with recombinant H-AChR αP3A⁺ protein (20 μg) in CFA and in IFA on days 30 and 60. Serum anti–M-AChR antibodies were measured in these strains by ELISA, on day 45 after immunization with H-AChRα subunit. Serum anti–M-AChR antibodies of IgG, IgG2b, IgG2c, and IgG1 isotypes were significantly reduced or absent in Aβ0 mice compared with wild-type B6 and B10 mice (Fig. 11A) . However, Aβ0 mice had higher levels of anti-AChR IgM antibodies than did B6 mice because of defective class switching of the antibody isotype. This defective class switching was caused by a deficiency of CD4⁺ helper T cells in the Aβ0 mice. The data implicate that MHC class II is mandatory for anti-AChR antibody isotype switching after immunization with H-AChRα subunit immunization. B6 and B10 mice had more severe oMG than did Aβ⁰ mice (Fig. 11B) . MHC class II–deficient mice were not completely resistant to clinical symptoms of oMG induced by the H-AChRα subunit, suggesting that other genes may also contribute to the clinical symptoms of oMG. 

MHC Class II Molecule in Lymphocyte Proliferative Response Is H-AChRαP3A

Lymphocytes of H-AChRα subunit–immunized B6 and DQ8 transgenic mice proliferated well when challenged in vitro with purified H-AChRαP3A protein (Fig. 12) . However, H-AChRαP3A subunit–immunized Aβ0 or DR3 mice failed to proliferate well when challenged with purified H-AChRαP3A protein. Therefore, MHC class II alleles influence the H-AChRα–specific lymphocyte response, and the MHC class II molecule is required for the presentation of H-AChRαP3A protein to B6 and DQ8 T cells. 

Discussion

We have established a novel small animal model for human oMG using HLA-DQ8 transgenic and B6 mice immunized with the recombinant H-AChRα subunit expressed in E. coli. Although the H-AChRαP3A⁻ isoform induced oMG, H-AChRαP3A⁺ isoform immunization induced a more rapid and severe form of oMG in HLA-DQ8 transgenic mice. Thus, the H-AChRαP3A⁺ isoform with the additional 25 amino acids (GDMVDLPRPSCVTLGVPLFSHLQNE) inserted at amino acid position 58/59 of the H-AChRαP3A⁻ isoform has an oMG-potentiating effect. Recently, we showed that mouse EOM and limb muscles express AChRα P3A⁺ proteins (isoform 2). ¹⁷ The EAMG-potentiating effect of AChRα P3A⁺ protein may also be attributed to different folding of the protein as a result of the additional amino acid residues. The highly immunogenic H-AChRαP3A⁺ protein induced the production of a significant amount of cross-reactive mouse AChR antibodies capable of inducing oMG. 

It is intriguing why immunization of HLA-DQ8 mice with the human AChR (derived from TE671 cell line) induced predominantly gMG without any ocular symptoms. ⁶ One hypothesis is that whole H-AChR (derived from TE671 cell lines containing the P3A⁺ isoform) immunization induces immune responses to numerous AChR T and B cell epitopes but fails to induce a significant autoimmune response directed against the pathogenic oMG-potentiating epitopes of the H-AChRα subunit. H-AChR immunization in HLA-DQ8 mice fails to present oMG pathogenic H-AChRα subunit epitope(s) to B cells. However, the H-AChRα subunit immunization presents oMG pathogenic epitope(s) to B cells. In addition, H-AChR protein derived from the TE671 cell line lacks the E. coli protein involved in the triggering of autoimmune ocular abnormalities in HLA-DQ8 transgenic mice. 

oMG precedes gMG in a significant number of mice immunized with H-AChRαP3A⁺ and P3A⁻ protein. Similarly, in MG in humans, oMG precedes gMG. Furthermore, 37% of H-AChRαP3A⁺ protein–immunized and 71% of H-AChRαP3A⁻ protein–immunized mice developed only oMG. The disease does not progress beyond oMG in approximately 20% of oMG patients. 

We hypothesize that autoimmune response to H-AChRα subunit or H-AChRα subunit cross-reactive (e.g., gamma or other subunit) epitopes induces oMG in humans. Exposure to a molecular mimic of AChRα subunit epitopes in a susceptible person could trigger T and B cell autoimmune responses to AChR subunit pathogenic epitopes and could culminate in EOM disease. The initial selective lesion in the EOM in oMG could be caused by an autoimmune response to the γ subunit of EOM AChR, which is not expressed in the skeletal muscle, ¹⁸ the increased expression of AChRα subunit in the EOM compared with skeletal muscles, ¹⁸ the physiological differences (e.g., decreased safety factor in neuromuscular transmission in EOM) between EOM and skeletal muscles, ¹⁸ the differential expression of immune-related genes in EOM, ¹⁹ or the decreased expression of a complement regulator protein decay accelerating factor (DAF) in EOM compared with skeletal muscles. ²⁰ Antibody and complement-mediated destruction of EOM AChR in oMG provides the immune system with all the pathogenic B and T cell epitopes from all the subunits of AChR. Generalized MG ensues when pathogenic autoimmune responses mediated by dominant B and T cell epitopes are present in the entire molecule. Our data suggest that native conformation of the H-AChRα subunit is not required for the induction of oMG because nonnative conformation of the recombinant H-AChRα subunit derived from the E. coli expression system is sufficient to potentiate oMG abnormalities in HLA-DQ8 transgenic mice. 

It is also remarkable that HLA-DQ8 transgenic mice show increased susceptibility to oMG, whereas HLA-DR3 transgenic mice are highly resistant. Association of the DQ8⁺ HLA haplotype with MG has only once been reported, ²¹ but DR3 association with MG is more widely recognized. ^{7 21} Immunization of HLA-DR3 transgenic mice with the H-AChR ε subunit generated a high incidence of oMG (Bednov A et al., manuscript in preparation). Therefore, the outcome of oMG might depend on the autoimmune response induced by specific interactions between the individual HLA molecules and H-AChR subunits. MHC class II is required for H-AChR alpha subunit antigen presentation and anti-AChR antibody isotype switching after immunization with H-AChRα subunit. The differential oMG susceptibility observed in DQ8 and DR3 transgenic mice correlated with the intensity of lymphocyte to respond to the human AChR alpha subunit. The occurrence of ptosis in the H-AChRα subunit–immunized MHC class II–deficient mice could be attributed to a non-MHC gene influence or to an anti-AChR IgM–induced abnormality because Aβ⁰ mice had significantly more serum anti–M-AChR IgM antibodies, and IgM can fix complement effectively. The latter is not an ideal mechanism by which NMJ AChR is destroyed because IgM with a molecular weight of 900 kDa may be unable to enter the NMJ. However, in various autoimmune peripheral polyneuropathies, IgM molecules have been shown to have access to the neuromuscular junction, react with the target epitopes, induce alterations in the quantal release of acetylcholine, and ultimately block the NMJ functions through complement-dependent and -independent mechanisms. ^{22 23} These studies suggest that in the absence of substantial amounts of anti–AChR IgG, anti–AChR IgM might affect the NMJ functions and thus might induce ocular myasthenic symptoms. 

The ocular signs and pathology in EOM and limb muscles induced by H-AChRα subunit immunization make these HLA-class II transgenic and B6 mouse models ideal for studying the roles of various immune related cells, genes, and molecules in the development oMG and gMG. Moreover, our oMG model in HLA-DQ8 and B6 mice could be used as a preclinical model to evaluate the efficacy of therapeutic agents to control oMG or to prevent the progression from oMG to gMG. It has been suggested that oMG and gMG are two different disease entities. Our data, demonstrating the development of oMG and gMG in HLA-DQ8 transgenic mice after H-AChRα subunit immunization, suggest that oMG and gMG could be one disease induced by an autoimmune response to the H-AChRα subunit. oMG could be triggered by a cross-reactive or an autoimmune response to pathogenic AChRα subunit epitopes and not to an autoimmune response directed to the whole AChR molecule. The implication of sequence GDMVDLPRPSCVTLGVPLFSHLQNE within H-AChRαP3A⁺ protein in oMG pathogenesis suggests the possible involvement/importance of this peptide in combination with other dominant H-AChR T-cell epitopes ⁶ in the induction of antigen-specific tolerance for oMG. 

²  These authors contributed equally to the work presented here and should therefore be regarded as equivalent authors.

³  Present affiliation: Department of Neurology Xiang Ya Hospital, Central South University, Chang Sha, Hunan, People’s Republic of China;

⁵  Department of Psychiatry, John Peter Smith, Hospital of Tarrant County, Fort Worth, Texas;

⁶  Department of Neurology, University of Pennsylvania, Philadelphia, Pennsylvania.

 Supported by the Muscular Dystrophy Association, Myasthenia Gravis Foundation of America, and Association Française Myopathies. HY was an MG Foundation Osserman/Sosin/McLure postdoctoral fellow and an MDA Neuromuscular Disease Research Career Award recipient. ET was an MG Foundation Osserman/Sosin/McLure postdoctoral fellow, and MDA Neuromuscular Disease Research Career Award recipient. JL is an MG Foundation Osserman/Sosin/McLure postdoctoral fellow.

 Submitted for publication March 5, 2007; revised July 19, 2007; accepted September 20, 2007.

 Disclosure: H. Yang, None; B. Wu, None; E. Tüzün, None; S.S. Saini, None; J. Li, None; W. Allman, None; S. Higgs, None; T.L. Xiao, None; P. Christadoss, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

Western blot of E. coli protein extracts expressing recombinant H-AChRαP3A⁺ (1) and P3A⁻ (2) using anti–His tag mAb and anti–H-AChRα mAb.

Table 1.

 

Clinical oMG Incidence and Severity Scores of Various Mouse Strains Immunized with H-AChRαP3A Subunit

Mouse Strains	Mice (n)	Incidence (%)	P	Severity Score ± SE
HLA-DQ8	10	89	0.0001	1.82 ± 0.14
B10	10	78	0.01	0.21 ± 0.05
B6	10	75	0.01	0.25 ± 0.06
Aβ0	10	50	0.01	0.15 ± 0.09
HLA-DR3	8	12	—	0.12 ± 0.13

  P values (Fisher exact test) represent the significance of difference in statistical comparisons between the oMG incidences of HLA-DR3 transgenic mice and other strains.

Figure 2.

 

Clinical signs of oMG in HLA-DQ8 transgenic mice immunized with the H-AChRαP3A⁺ isoform. Nonimmunized HLA-DQ8 transgenic mouse (A), partial closure (ptosis) of an eye (B), and full closure of an eye (C) in H-AChRαP3A⁺ protein immunized mice.

Figure 3.

 

Serum antibodies to human and mouse muscle AChR in H-AChRαP3A⁺ isoform–immunized HLA-DQ8 mice. Sera collected on day 58 from H-AChRαP3A⁺ isoform (1 μg •, 5 μg □, 20 μg ♦) immunized mice were tested for antibody binding to H-AChRα, H-AChR, M-AChR, O-AChR, and L-AChR by α-BTx radioimmunoassay. Nonimmunized HLA-DQ8 mice (○) did not have serum antibodies to any of the tested antigens. Results are expressed as nanomoles of α-BTx binding sites precipitated per liter of serum. Differences between antibody levels were not statistically significant.

Figure 4.

 

IgG, C3, and MAC deposits in the NMJ of H-AChRα protein immunized and nonimmunized HLA-DQ8 transgenic mice. EOM sections obtained from mice immunized with H-AChRα (top) were stained for mouse IgG, C3, and MAC (right, green). NMJs were localized by α-BTx (left, red) (magnification for all, ×100). Images of stained or nonstained areas (right) were taken from the same location with α-BTx binding sites (left) under green fluorescent light.

Figure 5.

 

Kinetics of clinical ocular MG (oEAMG) and generalized MG (EAMG) in HLA-DQ8 transgenic mice immunized with H-AChRαP3A⁺ (A, C) or P3A⁻ isoform (B, D). EAMG incidence (A, B) (P = 0.0085 in Fisher exact test) and severity (C, D) (P = 0.02; Mann Whitney U test) between P3A⁺ and P3A⁻ immunized mice are significantly different. The delayed onset of EAMG in P3A⁻ immunized mice was significantly (P = 0.003) different from the onset of P3A⁺ immunized mice. Bars denote SEM.

Figure 6.

 

Serum antibodies to human and mouse muscle AChR in the H-AChRαP3A⁺ or P3A⁻ isoform immunized HLA-DQ8 mice. Sera collected on day 58 after second AChRα subunit immunization were tested for antibody binding to H-AChRαP3A⁺ or P3A⁻ isoform (H-AChRα), H-AChR, M-AChR, O-AChR, and L-AChR in α-BTx radioimmunoassay. Differences between antibody levels were not statistically significant.

Figure 7.

 

Serum anti-AChR antibody isotypes in H-AChRα P3A⁺ and and P3A⁻ isoform–immunized HLA-DQ8 mice. Sera were diluted 1:100 to detect IgM and 1:1000 to detect IgG and IgG isotype antibodies to H-AChRα (P3A⁺ or P3A⁻), H-AChR, and M-AChR. Only serum anti–AChRα IgM antibody response was significantly greater (*P < 0.05; one-way ANOVA and Tukey post hoc test) in P3A⁺ immunized mice than in P3A⁻ immunized mice.

Figure 8.

 

IgG, C3, and MAC deposits in the NMJ of H-AChRαP3A⁺ protein–immunized HLA-DQ8 transgenic mice. EOM or limb muscle sections obtained from mice immunized with H-AChRα (top) were stained for mouse IgG, C3, and MAC (right, green). NMJs were localized by α-BTx (left, red) (magnification for all, ×100). Images of stained or nonstained areas (right) were taken from the same location with α-BTx binding sites (left) under green fluorescent light. EOM samples obtained from naive HLA-DQ8 transgenic mice were used as negative controls.

Figure 9.

 

Clinical disease incidence, serum anti–M-AChR antibody levels (α-BTx-binding sites), and EOM and limb muscle NMJ C3, IgG, and MAC deposits of nonimmunized HLA-DQ8 transgenic mice and HLA-DQ8 transgenic mice immunized with H-AChRαP3A⁺ isoform (expressed in E. coli plasmid) in CFA, E. coli plasmid in CFA, or CFA only. Bars indicate standard errors. *P < 0.05, **P < 0.01; ***P < 0.001; ****P < 0.0001. Bars indicate standard errors. Results are representative of two independent experiments.

Figure 10.

 

Clinical oMG incidence, serum anti–M-AChR antibody levels, and EOM and limb muscle NMJ C3, IgG, and MAC deposits of nonimmunized and H-AChRαP3A⁺ isoform (in CFA)–immunized B6 and HLA-DR3 transgenic mice. HLA-DR3 transgenic mice were significantly resistant to oMG. **P < 0.01; ***P < 0.001; ****P < 0.0001. Bars indicate standard errors. Results are representative of two independent experiments.

Figure 11.

 

Serum anti–M-AChR antibody isotypes in H-AChRα P3A⁺ isoform immunized Aβ⁰, C57BL/10, and C57BL/6 mice (A). Sera were diluted 1:100 to detect isotypes. Serum anti–M-AChR IgM antibody response was significantly greater (*P < 0.05; one-way ANOVA and Tukey post hoc test) in Aβ⁰ mice than in C57BL/6 mice. Anti–M-AChR IgG, IgG1, IgG2b, and IgG2c antibody (**P < 0.01) was significantly lower in Aβ0 mice than in C57BL/6 or C57BL/10 mice. Sera of nonimmunized Aβ0, C57BL/10, and C57BL/6 mice were negative for all anti–AChR antibody isotypes. Kinetics of clinical oMG in Aβ^0, C57BL/10, and C57BL/6 mice immunized with H-AChRαP3A⁺ isoform (B). The oMG in Aβ⁰ mice was less severe than in C57BL/10 mice on day 15 after the third immunization (P < 0.05; one-way ANOVA and Tukey post hoc test).

Figure 12.

 

HLA class II allele influences lymphocyte proliferation in response to H-AChRα P3A⁺ protein. The intensity of P3A⁺ isoform–specific lymphocyte proliferation was statistically significant between HLA-DQ8 mice, HLA-DR3 mice, and wt C57BL/6 mice (P < 0.001) (A). Reduced H-AChRα P3A⁺ isoform–specific lymphocyte proliferation in MHC^−/− mice compared with B6 mice (P < 0.001) (B). Statistical significance was determined by two-way ANOVA. *P < 0.05, **P < 0.01, ***P < 0.001; results of Tukey post hoc test.

The authors thank Chella David for providing breeding pairs of HLA-DQ8 and DR3 transgenic mice and Ashok Chopra, Miles Cloyd, and Vivian Braciale for reviewing the manuscript. They also thank David Beeson for providing the AChRα subunit cDNA clone.