Introduction

Guillain–Barré syndrome (GBS) is the most frequent cause of acute flaccid paralysis worldwide. GBS is mainly divided into demyelinating and axonal subtypes, namely acute inflammatory demyelinating polyneuropathy (AIDP) and acute motor axonal neuropathy (AMAN) [1]. A common misconception is that GBS has a good prognosis; however, up to 20% of patients remain severely disabled and approximately 5% die despite immunotherapy such as plasma exchange and intravenous immunoglobulin therapy [2]. Interestingly, corticosteroids do not significantly hasten recovery [3]. Instead, oral corticosteroids might delay recovery in patients with GBS [4]. This is in contrast to chronic inflammatory demyelinating polyneuropathy where steroids have been demonstrated to be an effective immunomodulatory agent. In GBS, specifically AIDP, the action mechanism of corticosteroids has yet to be understood. Understanding its role in the pathophysiology of GBS will likely facilitate the development of better treatment for GBS.

The disease mechanism of AIDP has yet to be clarified; whereas, that of AMAN has been elucidated in patients who underwent autopsy in China [5] and in rabbits immunized with ganglioside GM1 [6] or GM1-like lipo-oligosaccharide of Campylobacter jejuni [7]: The Gram-negative bacterium is the most frequent cause of antecedent infection in AMAN; The bacterium bearing GM1-like lipo-oligosaccharide induces the production of IgG anti-GM1 antibodies; The autoantibodies bound to GM1 at the nodes of Ranvier in the peripheral motor nerves; This activates complement, leading to disruption of axolemma in the motor nerves; Macrophages are recruited to scavenge the damaged axons. We previously showed corticosteroids inhibited the clearance of injured axons by macrophages and thus delayed axonal regeneration and clinical recovery [8]. This partly explains the reason why patients with GBS may not benefit from corticosteroids.

The phagocytosis of normal-looking myelin by macrophages in an early report of AIDP [9] might be misleading in suggesting that in the pathophysiology of AIDP, macrophages play a direct role in myelin destruction. However, autopsy specimens from patients with early AIDP showed complement deposition-induced myelin vesiculation before macrophage infiltration, suggesting the process of demyelinating is independent of macrophages [10]. Galactocerebroside is located in the myelin sheath [11]. IgG anti-galactocerebroside antibodies were detected in some patients with GBS who had an antecedent respiratory tract infection with Mycoplasma pneumoniae [12], which has a galactocerebroside epitope [13]. Molecular mimicry between M. pneumoniae and galactocerebroside may cause AIDP, although its animal model has yet to be developed. However, there have been earlier studies on galactocerebroside-immunized rabbits in understanding the pathogenesis of AIDP [14]. In the current study, we investigated whether macrophages have a direct role in myelin destruction or whether they are scavengers in galactocerebroside-immunized rabbit models to further understand the immunopathogenesis of AIDP.

Materials and methods Immunization procedure and nerve conduction studies

A total of five male Japanese white rabbits, weight 2–2.5 kg, were purchased from Eastern Breeding Co., Ltd (Xuzhou, Jiangsu, China). Our study was approved by the Ethics Committee of Medical Science Research, Affiliated Hospital of Jining Medical University (Reference number, 2020B009). Immunization of galactocerebroside (Sigma, St. Louis, MO) was performed as previously reported [15]. Clinical signs of rabbits were checked daily, and disease onset was defined as a clinical score of 10 points or more [16]. IgG anti-galactocerebroside antibodies were tested as earlier documented [17]. Nerve conduction studies (NCSs) in the sciatic nerves were performed as previously described [15], and the reference values were obtained based on normal healthy rabbits (n = 19).

Histopathological and immunohistochemical studies

Histopathological studies of lumbosacral spinal roots were performed as reported elsewhere [18]. A part of the spinal root specimens was used for toluidine blue staining and ultrastructural examination. The ultrastructural changes in the spinal roots were observed under transmission electron microscopy (Talos L 120C, Thermo Scientific, Brno, Czech). Another part of the lumbosacral spinal roots was cut into 6 μm cryosection for immunostaining experiment according to the protocols used previously [19]. Images were captured with Axio Imager Z2 microscope with ApoTome2 (Zeiss, Jena, Germany). Antibodies used in this study are listed in Supplemental Table, Supplemental digital content 1, https://links.lww.com/WNR/A717.

Statistical analyses

Data not normally distributed were presented as median with data range. For the immunostaining experiment, at least three different nerve roots from the lumbosacral segment were stained for analysis. Mann–Whitney U test was used to compare fluorescence intensity of myelin basic protein (MBP), number of voltage-gated sodium (Nav) channel clusters and contactin-associated protein 1 (Caspr1) per field between AIDP and normal control rabbits (n = 3). All data analysis was performed using GraphPad Prism Version 8.0 software (GraphPad Software Inc., San Diego, CA, USA). A two-sided P-value of <0.05 was considered to be significant.

Results Electrophysiological and serological findings

Two of the five showed clinical symptoms (AIDP1 and AIDP2). After the fifth immunization, AIDP1 developed muscle weakness in the neck and four limbs (maximal score 13 on day 1), then improved by 3 on day 5. Another (AIDP2) had a clinical score of 8 after the seventh immunization, and then improved by 4 on day 3 (Supplemental figure 1, Supplemental digital content 1, https://links.lww.com/WNR/A717). However, the score of 8 did not fulfill our criterion for disease development. The remaining three displayed well until the endpoint. After the fifth immunization, all of the five were detected with the elevation of serum anti-galactocerebroside IgG antibodies.

On day 1, the NCSs revealed markedly decreased compound muscle action potentials (CMAPs) amplitude (1.6 mV, reference value more than 5.7) and temporally dispersed F waves (Fig. 1). Prolonged minimal latency of F waves was also recorded (14.9 ms, reference range less than 12.2). On day 21, CMAPs significantly recovered in amplitude, but distal motor latency was delayed (3.4 ms, reference range less than 2.7). These findings were compatible with the features of acute de- and re-myelination process. In AIDP2, NCSs done on days 1 and 4 showed prolonged distal latencies at 3.2 and 2.9 ms, respectively. Distal CMAPs were slightly dispersed temporally, whereas the amplitude of CMAPs recorded on each day remained within the normal range (10.0 and 7.9 mV, respectively). No electrophysiological abnormalities were observed in the other three rabbits.

Fig. 1:

Electrophysiological features of the AIDP rabbit. Serial nerve conduction studies in AIDP rabbit (AIDP1) before and after onset of symptom. The tibial nerve was stimulated at the medial malleolus and popliteal fossa, and compound muscle action potentials (CMAPs) were recorded from the medial plantar muscle. F waves were also recorded from the medial plantar muscle after the tibial nerve stimulation at medial malleolus. Before immunization, CMAP amplitudes after medial malleolus and popliteal fossa stimulations were 11.1 mV (reference value, > 5.7) and 8.3 mV (> 4.9), respectively. The distal motor latency was 2.1 ms (< 2.7). One day after onset, distal CMAP amplitude was significantly decreased by 1.6 mV. The distal motor latency remained within normal range and motor conduction velocity between the medial malleolus and popliteal fossa was normal at 38.9 m/s (33.4–100.2). Three days after onset, the amplitude of CMAPs was remained reduced and the distal latency was prolonged at 2.9 ms. On day 6, CMAPs showed significant abnormal temporal dispersion (arrows). On day 21, distal CMAP amplitude was returned to normal range (7.0 mV), whereas distal motor latency (3.4 ms) and CMAP duration (2.9 ms (1.4–2.0) after stimulation at medial malleolus) were remained prolonged. The minimum latency of F waves recorded before immunization was 10.5 ms. On days 1, 3 and 6, the F wave latency was prolonged and F waves were dispersed. On day 21, being along with recovery of the clinical symptoms, the latency of F waves was shortened, but was still abnormal.

Pathological findings

On toluidine blue staining, significant multifocal demyelinating plaques were present in both the ventral and dorsal roots in the AIDP rabbit (Fig. 2a and b). In the demyelinating plaques, individual nerve fibers had thinning myelin in various degrees and some naked axons were observed. Macrophages were present in or around the demyelinating plaques and some accumulated around the vessels as clusters as previously described [20]. On electron microscopy, various views of demyelination and remyelination were seen. Some fibers demonstrated the vesiculation of outermost layer of myelin sheath without immune cellular invasion. The myelin sheets became loose and partially dissociated from certain locations, and vesicular dissolution of the myelin sheath was seen in the spaces between the myelin sheets (Fig. 2c and d). Some nerve fibers were completely demyelinated, leaving intact naked axons (Fig. 2e). On both toluidine blue staining and ultrastructural examination in AIDP2 and the other three rabbits, no obvious demyelinating changes were observed.

Fig. 2:

Histopathological findings of the spinal roots of the AIDP rabbit (AIDP1). (a and b) In the demyelinating plaques, the thinly myelinated fibers (asterisk) and naked axons (star) were observed especially around the blood vessels (empty triangle). Macrophages infiltration was tended to be associated with the fibers with demyelination. Some macrophages presented around the vessels as clusters (arrow head). (a) Indicates the ventral root, and (b) the dorsal root. Toluidine blue staining, scale bar = 50 µm. (c and d) Electron microscopy showed that nerve fiber undergoing demyelination at the early stage with myelin vesiculation and detached outer myelin lamellae (five-star) and myelin sheets were separated (arrow head). No macrophages were observed around. Scale bars = 2 µm. (e) Some nerve fibers were completely demyelinated and bare intact axons were left. A, axon; SC, Schwann cell; M, macrophage containing debris. Scale bars = 2 µm.

In AIDP1, significant IgG binding was demonstrated along the myelin surface (Fig. 3a). The demyelinating plaques were indicated by the lack of immunoreaction for MBP. The immunoreaction for βIII-Tubulin, expressed on the axolemma, was relatively preserved in the demyelinating plaques. The fluorescence intensity of MBP staining in AIDP1 was significantly less when compared to the normal controls (relative fluorescence intensity, median 9.9 [range 8.7–14.4] vs. median 25.6 [range 18.7–32.5]; P < 0.001) (Supplemental figure 2, Supplemental digital content 1, https://links.lww.com/WNR/A717). Co-localization of IgG, C3c and myelin debris (indicated by agglomerated MBP staining) was observed in and around the demyelinating plaques (Fig. 3b). Moreover, colocalization of RAM11-positive-macrophages and C3c were found (Fig. 4a). Some of myelin debris was engulfed by the macrophages (Fig. 4b). For AIDP2, there was significantly less fluorescence intensity of MBP staining than the normal controls (median 33.2 [range 26.5–39.5] vs. median 45.5 [range 35.5–50.8]; P < 0.001). No significant immunoreaction for IgG, C3c nor RAM11 was observed in AIDP2 and the normal control.

Fig. 3:

Localization of IgG binding, activated complement deposition and myelin debris in the spinal roots of the AIDP rabbit. (a) IgG deposition and demyelination in the spinal roots of the AIDP rabbit (AIDP1). Immunostaining showed significant IgG deposition (purple) and demyelination (indicated by lack of immunoreaction for MBP, green) in the AIDP rabbit. In demyelinating plaques (white dotted frame), most axons lacked myelin wrapping, but relatively preserved immunoreaction for βIII-tubulin. In control rabbit, no apparent staining of IgG was observed. Scale bars = 50 µm. (b) On immunostaining, broad deposition of IgG (purple) and C3c (green) was observed on the myelin surface of the AIDP rabbit. MBP staining was lacked in demyelinating plaques and agglomerations of MBP (myelin debris) were found. Co-existence of IgG binding, activated complement deposition and myelin debris was observed focally (arrow head). No specific staining of C3c and myelin debris were observed in roots in control. Scale bars = 20 µm.

Fig. 4:

Localization of complement deposition, myelin debris and macrophages in the spinal roots of the AIDP rabbit. (a) Immunostaining for macrophages (RAM11, red) and C3c (green) were shown. C3c deposition was broad, whereas macrophages and C3c colocalized focally in the nerve roots of AIDP rabbit (AIDP1) (arrow head). No staining of macrophages and C3c was observed in control. Scale bars = 20 µm. (b) Immunostaining showing reduced MBP (red), scattered myelin debris (indicated by agglomerated MBP staining, arrow head) and macrophages (RAM11, green). Most myelin debris were engulfed by macrophages (arrow) and a few were left to be scavenged (arrow head). Scale bars = 20 µm.

To investigate the node/paranode lesions, Nav channel clusters, highly concentrated at the nodes of Ranvier, and Caspr1, expressed on the paranode, were stained and compared between AIDP and the control rabbits. There was significant disruption of Nav channel clusters in AIDP1 than those in controls (number of Nav channel clusters staining per field, median 1.0 [range 0–1.0] vs. median 1.0 [range 1.0–3.0]; P < 0.001; Fig. 5). The immunostaining for Caspr1 was not impaired in AIDP1 or in the control rabbits. There was no difference in frequency of Caspr1 per field in the spinal roots between AIDP1 and controls (number of Caspr1 staining per field, median 2.0 [range 1.0–5.0] vs. median 2.0 [range 1.0–3.0]; P = 0.18). In AIDP2, disruption of Nav channel clusters was observed in few nodes of Ranvier with normal staining of Caspr1. There was no significant difference in the frequency of Nav channel clusters (median 2.0 [range 0–5.0] vs. median 3.0 [range 2.0–4.0]; P = 0.16) and Caspr1 (median 9.0 [range 5.0–14.0] vs. median 11.0 [range 6.5–17.0]; P = 0.25) in AIDP2 and the control rabbit. It should be noted that only ventral roots were immunostained for analysis.

Fig. 5:

Disruption of voltage-gated sodium (Nav) channel clusters in the spinal roots of the AIDP rabbit. Immunostaining for βIII-tubulin (the marker of axon) was used to location of Nav channel clusters (green, in a) and Caspr1 (green, in b) in spinal roots of the AIDP (AIDP1) and control rabbits. (a) Immunostaining fluorescence of Nav channel clusters in spinal roots of the AIDP rabbit is incomplete and weak. There is significantly reduced frequency of Nav channel clusters in the AIDP rabbit than in control rabbit (P < 0.001). Scale bars = 20 µm. (b) Immunostaining for Caspr1 in both AIDP and control rabbits was intact without difference in frequency (P = 0.18). Scale bars = 20 µm.

Discussion

In the present study, the galactocerebroside-immunized rabbit displayed acute tetraparesis with typical electrophysiological and pathological features of peripheral nerve demyelination, similar to human AIDP [10]. The demyelinating lesions displayed features of early as well as recovery stages of demyelination process.

The role of macrophages in the pathogenesis of AIDP has been debated. In an early report of GBS, evidence from biopsy samples showed that macrophages invaded the basement membrane and myelin lamellae were stripped away by the cytoplasmic processes of macrophages which contained the myelin debris [9]. ‘Macrophage-associated’ demyelination was a characteristic feature in the sural nerves from AIDP patients [21,22]. These studies may be misleading in implicating macrophages in the demyelinating process. In Chinese patients with AIDP, myelin vesiculation occurred earlier as a result of complement activation, preceding macrophage infiltration [10]. The autopsy study of Japanese patients with GBS showed that macrophage infiltration in the nerve roots was more prominent in patients with longer disease duration in both demyelinating and axonal subtypes [23]. These studies suggest that macrophages are unlikely to play a role in the initial demyelination process, but instead act as scavengers that remove nerve debris, facilitating nerve recovery.

In AMAN rabbits immunized by the GM1 or lipo-oligosaccharide of C. jejuni, IgG and the activated complement bound to the axolemma at the nodes of Ranvier, resulting in the axonal degeneration [7,19]. Macrophages were recruited to scavenge the degenerated axons. In the current study, the AIDP rabbit had nerve fibers with detached outer myelin lamella, followed by the vesicular demyelination in the absence of any macrophages. IgG and C3c were diffusely deposited on the myelin surface, whereas the macrophages were present at local sites engulfing the myelin debris. Our findings suggest that macrophages act as scavengers of nerve debris and are not involved in the initial demyelination process. As previously demonstrated in the AMAN model [6,7], studies investigating the mechanism of action of corticosteroids in the AIDP model will be important. Corticosteroids inhibit macrophage activation which may inadvertently result in reduced clearance of myelin debris, thus delaying the remyelination process and clinical recovery.

In addition, we found an unexpected result in the AIDP rabbit. Although IgG and complement were deposited on the myelin surface, Nav channel clusters were reduced at the nodes of Ranvier in AIDP1. In human AIDP, only a few demyelinating lesions could be observed in the most severely affected patient who died 3 days after onset [10], which indicates that conduction failure in the peripheral nerves may precede the demyelination process in AIDP. In a study of teased fiber staining using sera from Japanese GBS patients, serum IgG from about one-third of patients who were clinically diagnosed with AIDP bound at the nodes or paranodes, which serve as the initiating site of demyelination [24]. The sural nerve biopsy study showed that four of 11 patients had macrophage infiltration preferentially in the nodes [21]. In AIDP, where the presumed target antigen is the myelin, proposing a new concept of pathophysiology that involves the node/paranode region requires further study. Gliomedin or neuronal cell adhesion molecules, which are expressed on the microvilli of Schwann cells play an important role in inducing Nav channel clusters on the axolemma at the nodes of Ranvier [25]. Secondary depletion of such nodal molecules due to demyelination might contribute to this phenomenon. The disruption of Nav channel clusters would then reduce the safety factor of the saltatory conduction via reducing inward sodium ion current at the nodes, resulting in clinical muscle weakness.

Our study has several limitations. First, only one rabbit (AIDP1) developed pathologically proven demyelination. Regardless of the clinical symptom, nerve conduction abnormalities and scarce immunoreaction for MBP in AIDP2, the conventional pathological observation did not show morphological changes and the disruption of nodal Nav channel clusters was not significant in the spinal roots. Demyelination in the galactocerebroside-immunized rabbits occurs in a multifocal manner, and the conduction abnormalities can precede the morphological changes as shown in AIDP1. The possible explanations for the paucity of obvious demyelination in AIDP2 are the restricted distribution and minimal morphological changes due to mild disease conditions. Second, the current study did not provide sufficient evidence for the cause of Nav channel cluster disruption because the involvement of the node/paranode sites in AIDP was not anticipated. Extensive pathological analysis will be required to clarify the pathophysiology of early symptoms and correlated molecular changes in nodes and paranodes in AIDP.

In conclusion, we demonstrated that in the AIDP rabbit, initial vesiculation of myelin occurred without inflammatory cells and macrophages served as scavengers of nerve debris. Nav channel cluster disruption at the nodes of Ranvier might be the underlying pathogenesis of progressive muscle weakness seen in the acute stages of AIDP. Our findings deserved further validation by studies using more rabbits.

Acknowledgements

We thank Professor Nortina Shahrizaila from the Department of Medicine, Faculty of Medicine, University of Malaya, Kuala Lumpur, Malaysia, for her critical reading and edits. We thank Min Wang for her technical assistance with the nerve conduction studies. We thank Di Qi and Wenchao Li from Medical Research Centre, Affiliated Hospital of Jining Medical University for their help with preparation of sampling and imaging for electron microscope.

This work was supported by the Taishan Scholars Program of Shandong Province (tsqn202103189), the National Natural Science Foundation of China (81771298 and 82101420), and the Key Research and Development Plan of Jining City (2020YXNS025).

Y. Wang and N. Yuki developed the concept and design of the article. J. Xu and Y. Wang draft the manuscript. J. Xu, F. Gao, F. Shan, and Q. Shi performed the immunization and daily check. J. Xu and F. Gao did the electrophysiological test, sampling and toluidine blue staining. J. Xu did the immunostaining, imaging and data analysis. N. Kokubun and N. Yuki performed the critical reading and revision. Y. Wang finalized the manuscript.

Conflicts of interest

There are no conflicts of interest.

References 1. Yuki N, Hartung HP. Guillain-Barré syndrome. N Engl J Med 2012; 366:2294–2304. 2. Hughes RAC, Swan AV, Raphael JC, Annane D, van Koningsveld R, van Doorn P A. Immunotherapy for Guillain-Barré syndrome: a systematic review. Brain 2007; 130:2245–2257. 3. Hughes RAC, Brassington R, Gunn AA, van Doorn PA. Corticosteroids for Guillain-Barré syndrome. Cochrane Database Syst Rev 2016; 10:CD001446. 4. Hughes RAC, van der Meché FG. Corticosteroids for treating Guillain-Barré syndrome. Cochrane Database Syst Rev 2000:Cd001446. 5. Hafer-Macko C, Hsieh ST, Li CY, Ho TW, Sheikh K, Cornblath DR, et al. Acute motor axonal neuropathy: an antibody-mediated attack on axolemma. Ann Neurol 1996; 40:635–644. 6. Yuki N, Yamada M, Koga M, Odaka M, Susuki K, Tagawa Y, et al. Animal model of axonal Guillain-Barré syndrome induced by sensitization with GM1 ganglioside. Ann Neurol 2001; 49:712–720. 7. Yuki N, Susuki K, Koga M, Nishimoto Y, Odaka M, Hirata K, et al. Carbohydrate mimicry between human ganglioside GM1 and Campylobacter jejuni lipooligosaccharide causes Guillain-Barré syndrome. Proc Natl Acad Sci U S A 2004; 101:11404–11409. 8. Wang YZ, Lv H, Shi QG, Fan XT, Li L, Yi Wong AH, et al. Action mechanism of corticosteroids to aggravate Guillain-Barré syndrome. Sci Rep 2015; 5:13931. 9. Prineas JW. Acute idiopathic polyneuritis an electron microscope study. Lab Invest 1972; 26:133–147. 10. Hafer-Macko CE, Sheikh KA, Li CY, Ho TW, Cornblath DR, McKhann GM, et al. Immune attack on the Schwann cell surface in acute inflammatory demyelinating polyneuropathy. Ann Neurol 1996; 39:625–635. 11. Stoffel W, Bosio A. Myelin glycolipids and their functions. Curr Opin Neurobiol 1997; 7:654–661. 12. Meyer Sauteur PM, Huizinga R, Tio-Gillen AP, Roodbol J, Hoogenboezem T, Jacobs E, et al. Mycoplasma pneumoniae triggering the Guillain-Barré syndrome: a case-control study. Ann Neurol 2016; 80:566–580. 13. Kusunoki S, Chiba A, Hitoshi S, Takizawa H, Kanazawa I. Anti-Gal-C antibody in autoimmune neuropathies subsequent to mycoplasma infection. Muscle Nerve 1995; 18:409–413. 14. Saida T, Saida K, Dorfman SH, Silberberg DH, Sumner AJ, Manning MC, et al. Experimental allergic neuritis induced by sensitization with galactocerebroside. Science 1979; 204:1103–1106. 15. Susuki K, Nishimoto Y, Yamada M, Baba M, Ueda S, Hirata K, et al. Acute motor axonal neuropathy rabbit model: immune attack on nerve root axons. Ann Neurol 2003; 54:383–388. 16. Phongsisay V, Susuki K, Matsuno K, Yamahashi T, Okamoto S, Funakoshi K, et al. Complement inhibitor prevents disruption of sodium channel clusters in a rabbit model of Guillain-Barré syndrome. J Neuroimmunol 2008; 205:101–104. 17. Wang Y, Shi Q, Lv H, Hu M, Wang W, Wang Q, et al. IgG-degrading enzyme of Streptococcus pyogenes (IdeS) prevents disease progression and facilitates improvement in a rabbit model of Guillain-Barré syndrome. Exp Neurol 2017; 291:134–140. 18. Nishimoto Y, Koga M, Kamijo M, Hirata K, Yuki N. Immunoglobulin improves a model of acute motor axonal neuropathy by preventing axonal degeneration. Neurology 2004; 62:1939–1944. 19. Susuki K, Rasband MN, Tohyama K, Koibuchi K, Okamoto S, Funakoshi K, et al. Anti-GM1 antibodies cause complement-mediated disruption of sodium channel clusters in peripheral motor nerve fibers. J Neurosci 2007; 27:3956–3967. 20. Sommer C, Koch S, Lammens M, Gabreels-Festen A, Stoll G, Toyka KV. Macrophage clustering as a diagnostic marker in sural nerve biopsies of patients with CIDP. Neurology 2005; 65:1924–1929. 21. Koike H, Fukami Y, Nishi R, Kawagashira Y, Iijima M, Katsuno M, et al. Ultrastructural mechanisms of macrophage-induced demyelination in Guillain-Barré syndrome. J Neurol Neurosurg Psychiatry 2020; 91:650–659. 22. Vallat JM, Magy L, Corcia P, Boulesteix JM, Uncini A, Mathis S. Ultrastructural lesions of nodo-paranodopathies in peripheral neuropathies. J Neuropathol Exp Neurol 2020; 79:247–255. 23. Sobue G, Li M, Terao S, Aoki S, Ichimura M, Ieda T, et al. Axonal pathology in Japanese Guillain-Barré syndrome: a study of 15 autopsied cases. Neurology 1997; 48:1694–1700. 24. Devaux JJ, Odaka M, Yuki N. Nodal proteins are target antigens in Guillain-Barré syndrome. J Peripher Nerv Syst 2012; 17:62–71. 25. Feinberg K, Eshed-Eisenbach Y, Frechter S, Amor V, Salomon D, Sabanay H, et al. A glial signal consisting of gliomedin and NrCAM clusters axonal Na+ channels during the formation of nodes of Ranvier. Neuron 2010; 65:490–502.