Abstract
purpose. To examine the distribution of anti–GQ1b, –GT1a, and –GD1b antibody binding in human extraocular muscles (EOMs), axial and limb muscles, and muscle spindles and thereby test the hypothesis that their distinctive ganglioside composition provides the molecular basis for selective involvement of EOMs and muscle spindles in Miller Fisher syndrome.
methods. Muscle samples from adult human EOMs, vastus lateralis, biceps brachii, lumbrical, psoas, and deep muscles of the neck were processed for immunohistochemistry, with monoclonal antibodies against ganglioside GQ1b, GT1a, and GD1b. Neuromuscular junctions (NMJs) were detected by α-bungarotoxin binding and by acetyl cholinesterase reaction.
results. Most motor endplates of human EOMs richly bound anti–GQ1b, –GT1a, and –GD1b ganglioside antibodies. Anti–GQ1b, –GT1a, and –GD1b ganglioside antibody bindings to NMJs in human limb and axial muscle were scarce, but the nerve terminals inside muscle spindles and in direct contact with intrafusal fibers were labeled with anti– GQ1b, –GT1a, and –GD1b ganglioside antibodies.
conclusions. The abundant and synaptic-specific binding of anti–GQ1b, –GT1a, and –GD1b ganglioside antibodies and the rich capillary supply in the human EOMs may partly explain the selective paralysis of these muscles in Miller Fisher syndrome.
Miller Fisher syndrome (MFS) is a variant form of the paralytic neuropathy Guillain-Barré syndrome, characterized by impairment of eye movements (ophthalmoplegia), limb incoordination (ataxia), and loss of tendon reflexes (areflexia).
1 MFS is most usually a postinfectious syndrome whose pathophysiology involves inflammatory disruption of peripheral nerve function and the presence of autoantibodies against gangliosides.
2 3 Gangliosides are a large family of sialylated glycosphingolipids highly enriched in neuronal and glial plasma membranes and important for the development, function, and maintenance of the nervous system.
4 It has been demonstrated that one mechanism by which anti–ganglioside antibodies arise is through molecular mimicry, with structurally similar bacterial lipopolysaccharides on the surface of the preceding infectious agent, most notably demonstrated for
Campylobacter jejuni.
5 Antibodies to ganglioside GQ1b are detected in the acute-phase sera in more than 90% of cases of MFS, but not in other relevant disease groups, and they therefore act as a sensitive and specific serum marker for MFS.
6 Anti–GQ1b antibodies in MFS almost invariably cross-react with the structurally similar ganglioside GT1a, though rare exceptions may exist.
7 8 Up to 50% of MFS sera also demonstrate reactivity with other gangliosides containing a disialosyl epitope, such as GD3 and GD1b (for a review, see Ref.
9 ). The presence of autoantibodies against gangliosides is associated with ophthalmoplegia in MFS,
10 Guillain-Barré syndrome,
11 and even Bickerstaff encephalitis.
12
The human extraocular muscles (EOMs) are among the most complex muscles in the body, and their unique features
13 14 15 include the presence of multiply innervated fibers in addition to single-innervated fibers normally seen in other muscles
16 and an atypical content of acetylcholine receptor (AchR) isoforms. The EOMs express gene transcripts specific for adult ε and fetal γ isoforms of AchR, unlike the other mature muscles which contain only adult ε AchR isoform.
17 18
One of the major manifestations of MFS is EOM paralysis, and it has been predicted that the affected anatomic sites in MFS should contain more GQ1b and GQ1b-like gangliosides than unaffected sites. Chiba et al.
7 reported that the extraocular cranial nerves had high levels of immunoreactivity for GQ1b at nodes of Ranvier and that the nerve trunks supplying the human extraocular muscles have a relatively high content of GQ1b than other cranial or spinal nerves. Because the blood-nerve barrier partially protects the nerve trunks and the nodes of Ranvier from circulating antibodies, consideration has been given to the possibility that the neuromuscular junction (NMJ) may be a possible target in MFS because it lies outside the blood-nerve barrier. Furthermore, extraocular muscle nerve terminals are a site of action for botulinum toxins that are known to bind disialylated gangliosides to gain access through the presynaptic membrane to the exocytic apparatus.
19
In addition to ophthalmoplegia, typical MFS also involves abnormal coordination and a generalized absence of reflexes because of the loss of proprioceptive input from peripheral sense organs.
20 The sites at which this defect arises may include the cerebellum, structures in dorsal root ganglia, and muscle spindles.
1 9 20 21 22 Kuwabara et al.
20 demonstrated that the ataxia, as well as areflexia, was caused by the involvement of afferents in muscle spindles. Muscle spindles are sensory mechanoreceptors that relay information about muscle length, and changes in muscle length, to the central nervous system.
23 They consist of a bundle of encapsulated small-diameter muscle fibers, intrafusal fibers, which receive sensory and motor innervation.
23 24 The sensory innervation is distributed along the equatorial and juxtaequatorial regions, whereas the motor endings innervate the polar region toward or beyond the end of the spindle capsule.
23
In the present study, the distribution of GQ1b and GQ1b-like ganglioside epitopes in human EOMs and limb muscles was investigated with the aim of substantiating the hypothesis that the nerve terminal might be one potential target in MFS and that differential ganglioside composition/content between EOM and limb endplates might account for the site-specific clinical manifestations of anti–GQ1b antibody syndromes.
Twenty-seven EOM samples (from 6 men [age range, 32–56 years] and 1 woman [age 26 years]), 17 samples from vastus lateralis, biceps brachii, deep muscles of the neck, the first lumbrical and psoas muscles (from 12 males and 5 females [age range, 13–55 years]), and the lower limb muscles from one fetus at 20 weeks of gestation were collected with the approval of the Ethical Committee of the Medical Faculty, Umeå University, according to the ethical recommendations of the Swedish Transplantation Law and the tenets of the Declaration of Helsinki.
The muscle samples were mounted, rapidly frozen in propane chilled with liquid nitrogen, and stored at −80°C. Frozen specimens were serially sectioned, 5 μm thick, at −23°C using a cryostat (Reichert Jung; Leica, Nussloch, Germany).
Neuromuscular junctions were detected using the acetylcholine-esterase reaction
25 and by rhodamine-conjugated α-bungarotoxin labeling (Molecular Probes Inc., Eugene, OR). Immunohistochemistry was performed on air-dried tissue sections, as previously described,
26 with a battery of antibodies. Two monoclonal antibodies against gangliosides were selected from a panel of anti–ganglioside antibodies that have been previously described.
27 28 EM6 (1:100) binds gangliosides GQ1b and GT1a, whereas MOG1 (1:50) binds ganglioside GD1b. In addition, mouse anti–AChR γ (GTX74890; 1:50; Gentex, Landskorna, Sweden) and rat anti–AChR ε (mAb168; 1:10; gift from Socrates J. Tzartos, University of Patras, Rio, Greece)
29 were used to detect the fetal γ AChR and the adult ε AChR subunits, respectively. Monoclonal antibody ALD 19 against slow-tonic myosin heavy chain (MyHC) isoform (1:400; gift from Donald A. Fischman, Cornell University, New York, NY) was used to identify the slow tonic multiply innervated muscle fibers in EOM and the bag fibers in muscle spindles. Monoclonal antibody BF35 against slow- and fast-twitch IIa MyHC isoforms (1:1000; gift from Stefano Schiaffino, University of Padova, Padova, Italy) was used to label the remaining muscle fibers.
Control sections were treated as described, except that the primary antibody was either omitted or substituted by nonimmune serum. No staining was observed in control sections. To determine the percentage of NMJs stained with anti–ganglioside antibodies in EOMs, sections in which numerous NMJs were readily seen by α-bungarotoxin labeling were randomly selected, and the number of NMJs was counted.
The sections were examined with a microscope (Eclipse E800; Nikon, Melville, NY) equipped with a color camera (Spot RT; Diagnostic Instruments Inc., Sterling Heights, MI). Computer-generated images were processed using commercial software (Photoshop; Adobe Systems Inc., Mountain View, CA).
The muscle spindles were analyzed in two regions: the A region, including the equator and the juxtaequatorial region containing the periaxial fluid space; and the B region, extending from the end of the periaxial fluid space to the end of the spindle capsule. Intrafusal fibers were classified into nuclear bag and nuclear chain fibers according to their patterns of immunoreactivity.
30
There was a massive immunoreactivity with mAbs EM6 and MOG1 in the human EOMs, in contrast to practically absent or scarce reactivity in limb muscle samples
(Fig. 1) . As previously reported,
31 anti–GQ1b/GT1a (EM6) and anti–GD1b (MOG1) mAbs bind to nerve axons, blood vessels, and NMJs in muscle tissue sections. This study describes the distribution of these gangliosides in nerve terminals at NMJs (their association with skeletal muscle fibers outside the blood-nerve barrier).
NMJs were identified on the basis of α-bungarotoxin staining. α-Bungarotoxin binds to AChRs in the postsynaptic membrane and is therefore a good marker of the synaptic site of muscle fibers. In EOMs, NMJs were numerous in practically all sections examined, and three distinctive staining patterns were identified after labeling with α-bungarotoxin in cross-sections and longitudinal sections: (1) large chevron- or c-shaped endplates, lobulated with deep junctional folds
(Figs. 2A2 2B2 2C2 2I2) ; (2) linear endplates with or without shallow folds
(Figs. 2D2 2E2) ; (3) and single or strings of small endplates
(Figs. 2G2 2H2 2J2 2K2) . The first two were typical en plaque endplates whereas the third type were en grappe endplates, typical of multiply innervated muscle fibers.
16 NMJs with deep lobulated folds outnumbered the other two staining patterns in the samples studied. Endplates with lobulated deep folds and en grappe endplates were present only in EOMs, whereas endplates in vastus lateralis and biceps brachii muscles had shallow synaptic folds (
Figs. 3A2 3B2 3C2) .
Both anti–ganglioside antibodies stained the EOM motor endplates in close vicinity to the AChRs labeled by α-bungarotoxin. They stained only the NMJs and did not stain the extrasynaptic surfaces of the muscle fibers
(Figs. 2A 2B 2C 2D 2E 2F 2G 2H) . Most endplates identified with α-bungarotoxin were stained with anti–ganglioside GQ1b/GT1a (91.9%; 142 NMJs examined) and GD1b (90.7%; 123 NMJs examined) antibodies. In general, the anti–ganglioside antibodies showed a cometlike staining pattern over the nerve terminals, in addition to the lobulated staining at the synaptic sites
(Figs. 2A 2B 2C) . The size and staining intensity of the cometlike tail varied from endplate to endplate. Few motor endplates in the EOM had a linear strand appearance, and they generally had lower staining intensity with the two anti–ganglioside antibodies and lacked cometlike staining
(Figs. 2D 2E) . Anti–ganglioside staining was also found in the vicinity of α-bungarotoxin, suggesting that the respective epitopes were in, or were closely associated with, the synaptic membranes
(Figs. 2D 2E) .
Some exceptional endplates were wrapped around the whole surface of the muscle fiber, both in sections treated with α-bungarotoxin and in anti–ganglioside mAbs
(Fig. 2F) . Anti–ganglioside staining was also present inside these muscle fibers in a weblike pattern.
En grappe motor endplates were observed primarily in longitudinal sections. Anti–ganglioside epitopes were localized at the synaptic regions close to the en grappe motor endplates
(Figs. 2G 2H) . Staining with anti–ganglioside antibodies was less extensive at the en grappe than at the lobulated en plaque NMJs, whereas staining intensity was similar at both types of NMJs.
Association of Ganglioside Epitopes with Adult and Fetal AChR Subunits in Human EOMs
The specificity of antibodies GTX74890 to fetal AChR γ subunit and MA168 to adult AChR ε subunit was confirmed for human fetal and adult vastus lateralis muscles. At 20 weeks of gestation, anti–AChR γ antibody stained all the endplates visualized by α-bungarotoxin, whereas only sporadic endplates bound the anti–AChR ε antibody weakly (not shown). In contrast, staining intensity with anti–AChR ε was very strong, and anti–AChR γ was generally not detectable on normal adult vastus lateralis muscle fibers (not shown).
In adult EOMs, most en plaque endplates identified by α-bungarotoxin were strongly stained by anti–AChR ε antibody
(Fig. 2I) . The anti–AChR ε–positive endplates were seen primarily in fibers with large diameters and had a lobulated or liner cleft appearance. Anti–AChR γ antibody strongly stained the small en grappe endplates in muscle fibers with small diameters
(Figs. 2J 2K) .
The two anti–ganglioside antibodies labeled both endplates containing adult ε and fetal γ AChR subunits, and they stained en grappe endplates containing the fetal AChR γ subunit with less extent because of the small size of the endplates.
Staining patterns with both anti–ganglioside antibodies were similar in vastus lateralis, biceps brachii, and psoas muscles. In general, these tissue sections were unstained with the two anti–ganglioside antibodies. However, staining was occasionally observed either in interstitial spaces confined to a group of fibers or around the entire or part of a muscle fiber surface, with or without the concomitant presence of NMJs
(Fig. 3) . Because of the low density of motor endplates in limb and trunk muscles, only 10 motor endplates were encountered in the biceps brachii, 11 in the psoas, and 40 in the vastus lateralis muscle samples examined. Motor endplates labeled by α-bungarotoxin in biceps and psoas muscles were in general unstained with these anti–ganglioside antibodies; similarly, in vastus lateralis, only 7% and 6% of the endplates were stained with anti–GQ1b/GT1a and anti–GD1b antibody, respectively. However, all endplates labeled with the anti–ganglioside mAbs were found in areas with staining of the fiber contours or of the interstitial compartments
(Fig. 3C) . Furthermore, the level of ganglioside staining observed at these NMJs was not increased compared with that of their surroundings. The concurrence of positive staining at NMJs and in interstitial compartments or around fiber surface with anti–GQ1b/GT1a and –GD1b antibodies indicated that the distribution of anti–ganglioside antibody staining observed in limb muscles was not solely related to the presence of NMJs.
Thirty-five muscle spindles encountered in the biceps brachii, the first lumbrical, and the deep muscles of the neck were examined: 21 muscle spindles were encountered in the juxtaequatorial region, 11 in the B region, two in the equator, and one outside the capsule sleeve. In the equatorial region, where abundant group Ia sensory nerve terminals encircle intrafusal fibers, the entire surface of each individual intrafusal fiber was strongly stained with anti–ganglioside antibodies
(Figs. 4A 4B 4C) . In the juxtaequatorial region, where intrafusal fibers are mainly innervated by group II afferents, the gangliosides were also present
(Figs. 4D 4E 4F 4G) . Muscle spindles were generally unstained in the B region, though positive stainings were found occasionally in the space between intrafusal fibers
(Figs. 4H 4I 4J 4K) .
The anti–GQ1b/GT1a antibody did not stain the spindle capsules in the examined spindles, whereas anti–GD1b stained mostly the outer layer of capsules in 60% of the examined spindles.
To the best of our knowledge, this is the first systematic immunohistochemical study to examine the distribution of anti–GQ1b/GT1a and –GD1b antibody binding in human EOMs, axial and limb muscles, and muscle spindles. The major findings of the study are that the NMJs of human EOMs are richly bound by GQ1b, GT1a, and GD1b ganglioside antibodies; there is a paucity of anti–GQ1b, –GT1a, and –GD1b ganglioside antibody binding to NMJs in human limb and axial muscle; the presence of GQ1b-like gangliosides at the NMJs in human EOMs was not correlated with the presence of the different AChR subunits; the nerve terminals inside muscle spindles and in direct contact with intrafusal fibers were stained with anti–GQ1b, –GT1a, and –GD1b ganglioside antibodies.
By definition, immunohistochemical studies using anti–ganglioside antibodies do not demonstrate the presence of a specific or a particular ganglioside but more broadly indicate that a specific antibody binds to a presumed epitope. Ideally, we would want to combine immunohistochemistry with biochemical demonstration of the ganglioside profile of motor nerve terminals. However, their minute size and sparse distribution makes this practically impossible. Immunolocalization of gangliosides using specific antibodies is long recognized as notoriously complex.
32 33 Many gangliosides are cryptic in tissue and subject to a range of fixation variables, and many anti–ganglioside antibodies that are useful in immunohistochemistry have specificity for more than one ganglioside, as in this study. In pilot studies we screened a range of different antibodies to polysialylated gangliosides that showed no reactivity with motor endplates in any site, presumably because of subtle steric effects concerning the antibody-ganglioside interaction. The current data thus must be interpreted with caution because absolute certainty regarding the presence of absence of a specific ganglioside cannot be concluded from immunohistochemistry alone.
Initially, the restriction of the paralytic effects of MFS to a limited group of muscles was considered solely to be due to the different patterns of ganglioside distribution noted between cranial and somatic nerve trunks. Chiba et al.
7 first noted, by using immunohistochemistry, that the extraocular cranial nerves expressed high levels of GQ1b at nodes of Ranvier. Later the binding of antibodies specifically reactive to GQ1b and GT1a gangliosides at NMJs was reported.
9 31 34 35 However, GQ1b can be detected biochemically in peripheral nerve in significant amounts at sites unaffected by MFS
36 37 ; thus, it has been proposed that their differences in tissue distribution of GQ1b gangliosides cannot solely or sufficiently explain the regional localization of the clinical pathologic features because target gangliosides are also present at sites unaffected by MFS.
Our findings provide support for the hypothesis that the restriction of paralysis to the EOMs might be attributed to the availability of antib-GQ1b, –GT1a, and –GD1b binding epitopes in EOMs compared with limb muscles. In contrast to biochemical studies
37 by which all gangliosides in one tissue are pooled together, the present study clearly visualized the precise localization of the ganglioside epitopes in each muscle fiber despite not identifying the biochemical nature of the ligand. The two ganglioside antibodies stained the synaptic, but not the extrasynaptic, areas of EOM muscle fibers, indicating that the presence of anti–ganglioside antibody epitopes in human EOM fibers is synapse specific. In limb muscles, however, the distribution of ganglioside epitopes was scarce and not restricted to the synaptic regions. Instead, ganglioside epitopes were present in the interstitial space adjacent to plasma membranes of muscle fibers, consistent with the known wide distribution of gangliosides in the human body.
Our results are indirectly supported by single-fiber electromyography (SFEMG) data showing that in patients with MFS whose sera were positive to anti–GQ1b antibody, neuromuscular transmission in their limb muscles was normal, indicating that anti–GQ1b antibody has no effect on the NMJs in human limb muscles,
38 though other studies have suggested that NMJs might be affected.
39 What these EMG studies were unable to show was the presence of neuromuscular dysfunction in extraocular muscles because this is technically impracticable. However, increased jitter values in SFEMG of orbicularis oculi muscle has sometimes been identified in anti–GQ1b-positive MFS.
40 At present, the human SFEMG data have not conclusively or consistently shown evidence of NMJ dysfunction in human MFS, and this remains an area for further studies.
41 In addition to possible paralytic effects from NMJ dysfunction, our results do not preclude the involvement of other sites, such as the nodes of Ranvier in extraocular nerves, as has been supported by previous immunohistochemical studies.
7
Additional factors may be of importance for the selective involvement of the EOM in MFS. Human EOMs have the highest capillary supply reported thus far for all muscles in the body—1170 ± 180 capillaries/mm
2 in the orbital layer and on average 1050 ± 190 capillaries/mm
2,
42 compared with 440 ± 180 capillaries/mm
2 in biceps brachii.
43 A rich blood supply may be of utmost importance for the access of the circulating anti–ganglioside antibodies and of the complement cascade elements to the target NMJs. Furthermore, differences in capillary permeability and functional properties between the EOMs and the limb muscles cannot be excluded.
Most MFS cases involve areflexia and ataxia. The site or sites at which these defects arise is debated, with the cerebellum, dorsal root ganglia, and muscle spindles possible candidates.
1 9 20 21 22 Data indicate that the dorsal root ganglia are possible target sites for antibodies against gangliosides because of the properties of their blood-nerve barrier.
21 Kuwabara et al.
20 reported that patients with MFS showed the same postural body sway peak as did patients with sensory ataxia but that it was distinct from that of patients with cerebellar ataxia. Thus, they proposed that ataxia and areflexia were caused by the involvement of afferents in the muscle spindles. Our study further supports a role for muscle spindles in MFS because nerve endings on the intrafusal fibers were strongly stained with anti–ganglioside antibodies at the equatorial region and showed a staining pattern that clearly outlined the surface of each individual fiber. Although we cannot be sure of the afferent or efferent nature of nerve contacts labeled in the present study, the results suggest that the group Ia afferents in muscle spindles may contain high levels of GQ1b, GT1a, GD3, and GD1b gangliosides because the type Ia afferents typically encircle intrafusal fibers in the equator and form annular terminals.
23 At the juxtaequatorial region, the intrafusal fibers showed a punctate staining pattern, suggesting that group IIa afferents also may contain ganglioside molecules because these afferents form patchy or flower spray endings rather than encircling the intrafusal fibers.
23 Dorsal root ganglia, in humans
21 44 and rabbits,
45 may be another possible target site for anti–ganglioside antibodies related to ataxia because their blood-nerve barrier is more permeable than that of nerve root vessels.
46
In summary, in addition to the staining seen in nerves and blood vessels, the abundant synaptic staining by anti–GQ1b, –GT1a, and –GD1b ganglioside antibodies in the human EOMs and muscle spindles compared with limb muscles is reported here. Together with other factors, including the absence of a blood-nerve barrier at this site, the high capillary supply unique for EOMs, the clinical similarity with botulism, and the murine experimental data, our findings support the possibility that the NMJs of EOMs might be a target for anti–disialylated ganglioside antibodies in MFS and related neuropathies.
The authors thank Socrates J. Tzartos and Stefano Schiaffino for kindly providing antibodies, and Margaretha Enerstedt and Anna-Karin Olofsson for excellent technical assistance.