John D. Pollard, Nerve Research Foundation, Brain and Mind Research Institute, The University of Sydney, Level 7, 94 Mallett Street, Camperdown NSW 2050, Australia. Tel: +61 2 9351 0730; Fax: +61 2 9351 0653; E-mail: firstname.lastname@example.org
Early pathological studies in patients with acute and chronic inflammatory demyelinating neuropathies, and the animal model experimental autoimmune neuritis (EAN) showed similarities in the process of demyelination. These studies focused on compact myelin proteins and peptides as targets of immune attack in Guillain-Barré syndrome (GBS), chronic inflammatory demyelinating polyradiculoneuropathy (CIDP), and EAN. However, serological studies in patients with subsets of GBS highlighted the importance of gangliosides – glycolipids enriched in non-compact Schwann cell regions and the node, paranodal, and internodal axolemma. In the acute motor axonal neuropathy (AMAN) rabbit model, antibodies to the ganglioside GM1 bind in the nodal region, impair Na channel clustering and disturb Schwann cell/axon organisation. Schwann cell neurobiological studies now highlight the importance of adhesion molecules, including neurofascins, gliomedin, contactins, and NrCAM to Schwann cell/axon integrity. Changes to nodal fine structure by immune responses against such molecules may provide a mechanism for reversible conduction failure or block. Recovery of patients with CIDP or multifocal motor neuropathy (MMN) following treatment may sometimes be better explained by reversal of conduction failure than remyelination or regeneration. This review considers the importance of the intricate molecular arrangements at the nodal and paranodal regions in inflammatory neuropathies such as CIDP. Early images of compact myelin stripping and phagocytosis, may have diverted the research focus away from these vital non-compact myelin Schwann cell areas.
Recurrent or relapsing inflammatory neuropathy was described on a number of occasions (Eichorst, 1874; Targowla, 1894; Hoesterman, 1914) prior to the study by Guillain, Barré, Strohl, and Landry (Guillain et al., 1916) which established the clinical syndrome of acute idiopathic polyneuropathy. It is interesting that a number of early reports associated these recurrent neuropathies with nerve hypertrophy (Nattrass, 1921; Harris and Newcombe, 1929), a finding which tended to be forgotten until recent magnetic resonance imaging (MRI) studies showed that proximal hypertrophy occurs in at least 50% of cases followed-up over the long term (Duggins et al., 1999). Progressive hypertrophic neuropathy without family history but showing nerve inflammation was also reported early by Roussy and Cornil (1919). Austin (1958) reviewed 32 cases of chronic relapsing and hypertrophic neuropathies from the literature. Interest in the possible treatment of these neuropathies was stimulated particularly by his description of a patient in whom improvement was documented on numerous occasions following treatment with corticosteroids.
Subsequent studies drew attention to the similarities between these chronic relapsing neuropathies and the Guillain-Barré syndrome (GBS) (Hinman and Magee, 1967; Thomas et al., 1969) and further evidence for the efficacy of corticosteroid accumulated (Thomas et al., 1969; DeVivo and Engel, 1970; Matthews et al., 1970). However, a plethora of descriptive terms reflecting certain aspects of these neuropathies – relapsing, subacute, progressive, hypertrophic, steroid-responsive – showed the lack of a clear definition until Dyck et al. (1975) introduced the term chronic inflammatory polyradiculoneuropathy to include all cases of chronic inflammatory demyelinating neuropathy, whatever their time course. The same group (Dyck et al., 1982) modified the descriptive title to chronic inflammatory demyelinating polyradiculoneuropathy (CIDP), a term which emphasises the pathology, a multifocal demyelination which predominantly affects proximal regions of the peripheral nervous system (PNS), the spinal roots, nerve trunks, and major plexuses (Harris and Newcombe, 1929; Thomas et al., 1969; Dyck et al., 1975; Prineas and McLeod, 1976).
Post-mortem studies have shown that although the major sites of involvement in the PNS are the spinal roots, proximal nerve trunks, and major plexuses, lesions may occur at any level including the autonomic nerves and the distal ends of motor nerves (Dyck et al., 1975). These changes consist of patchy demyelination, oedema, and variable inflammatory infiltrates (Hyland and Russell, 1930; Asbury et al., 1969; Thomas et al., 1969; Dyck et al., 1975). At nerve biopsy, usually a distal sensory nerve such as the sural nerve, any of the following may be seen – no abnormality, endoneurial and/or subperineurial oedema, demyelinated or thinly myelinated axons, onion-bulb formation, macrophage-mediated demyelination, axonal degeneration, or inflammatory infiltrates, either perivascular or endoneurial (Prineas, 1971; Dyck et al., 1975; Prineas and McLeod, 1976; Pollard, 1994). Prineas (1971), Bonnaud et al. (1974), and Prineas and McLeod (1976) studied actively demyelinating lesions in CIDP and found the same macrophage-mediated pattern of demyelination which had previously been described in the animal model experimental autoimmune neuritis (EAN) – a disorder induced by injection of peripheral nerve homogenate including the compact myelin component of the Schwann cell – and in GBS (Lampert, 1969; Prineas, 1972). Prineas noted macrophage stripping and phagocytosis of compact myelin lamellae following displacement of normal-appearing Schwann cell outer mesaxon from the compact myelin membrane. These striking illustrations (Fig. 1) and the strong association with EAN led to the popular view that in the inflammatory demyelinating neuropathies, subacute and chronic, the responsible neuritogen resided in the compact myelin proteins. This assumption was further reinforced in our own laboratory with the development of an animal model of CIDP established in rabbits. Outbred rabbits injected with a peripheral nerve myelin (PNM) homogenate developed either a progressive or relapsing/progressive disease similar in clinical, neurophysiological, and pathological features to CIDP (Fig. 2) (Harvey et al., 1987). In that model, the level of anti-myelin antibody correlated to some extent with the disease course, and disease stabilisation was achieved by plasma exchange or infusion which resulted in a reduction of the antibody level (Harvey et al., 1988; 1989a; 1989b). It is significant, however, that studies by Saida et al. regarding the pathogenicity of rabbit EAN serum showed that the demyelinating activity of rabbit EAN serum is due largely to an antibody to a Schwann cell, lipid, galactocerebroside (Saida et al., 1978; 1981). The general assumption by most researchers in the field for many years was that the target antigens in inflammatory demyelinating neuropathies most likely resided within the protein components of the compact myelin of the Schwann cell and numerous studies on EAN were performed with particular attention paid to compact myelin antigens. A similar situation existed in the field of multiple sclerosis (MS) where for many years immunopathogenic studies focused more on the animal model experimental autoimmune encephalitis (EAE) and central nervous system (CNS) compact myelin antigens of the oligodendrocytes such as myelin basic protein (MBP) and myelin oligodendrocyte glycoprotein (MOG) (Martini et al., 2010) than studies on the MS tissue. Although EAN may be induced in several animal species, following injection with compact myelin protein antigens such as P2 or P0 or peptides derived from these proteins, none of these has been established as the major target antigen in inflammatory neuropathy in man.
Evidence from Patients
Meaningful advances in our understanding of inflammatory neuropathy have however occurred more from studies in patients than from animal models. The first of these advances occurred in the field of GBS with the recognition of homogeneous subtypes of the disease and that some subtypes were associated with antibodies to gangliosides. Studies particularly from the Johns Hopkins Group performed initially in Northern China and from Yuki and colleagues from Japan established an association between acute motor axonal neuropathy (AMAN), a GBS subtype, and antibodies to GM1 and GD1a gangliosides (Yuki et al., 1990; 1992; McKhann et al., 1993; Jacobs et al., 1996; Ho et al., 1999). Interestingly, following immunisation of European patients with bovine gangliosides, Illa et al. (1995) reported that this caused a similar neuropathy in a small proportion of patients. In Miller Fisher syndrome, antibodies to GQ1b gangliosides are present in the majority of patients (Yuki et al., 1993; Willison and Veitch, 1994), and ophthalmoplegia in GBS and Bickerstaff's encephalitis also correlates with the presence of these antibodies. Ligand-binding studies using ganglioside-binding bacterial toxins such as cholera toxin have shown that GM1, GD1b and polysialosylated gangliosides are enriched in the paranodal loops of the Schwann cell while GQ1b is enriched at the nodes of Ranvier in oculomotor nerves. Gangliosides have also been identified on the paranodal and internodal axolemma (Willison et al., 2005). Serum from patients with AMAN binds to the node of Ranvier and the paranodal and internodal axolemma (Griffin et al., 1995; 1996). Willison and colleagues (Plomp et al., 1999; Willison, 2005) have shown that sera from Miller Fisher patients and human monoclonal anti-GQ1b immunoglobulin M (IgM) antibodies bind not only to the neuromuscular junction but also to the perisynaptic unmyelinated Schwann cells of this region. This is a region of the peripheral nerve devoid of compact myelin. Thus, in these inflammatory neuropathies in which antibody-mediated pathogenic mechanisms have clearly emerged, the target of these antibodies is not the compact myelinated region of the Schwann cell.
The mechanism by which GM1 antibodies cause conduction impairment in AMAN has been further investigated by Yuki and colleagues in their rabbit AMAN model (Susuki et al., 2007b). The group showed that anti-GM1 antibodies caused a complement-mediated disruption of the sodium channel clustering by disturbing the paranodal axoglial junctions, the Schwann cell microvilli along the nodes of Ranvier as well as the nodal cytoskeleton of the axon, all of which contribute to the stabilisation of sodium channel clusters. In immunohistochemical studies, it was shown that where there was marked nodal deposition of complement and membrane attack complex, the sodium channel clustering was disturbed in association with disruption of the molecules which contribute to the unique fine structural arrangement of Schwann cell/axonal relationships at the node (Fig. 3). To further investigate the role of gangliosides in the nodal region, Susuki et al. (2007a) examined the nodal region in mice lacking GM1 and GD1a. They showed that in these animals some paranodal loops failed to attach to the axolemma, potassium channels at the juxtaparanodes were mislocated to the paranodes and nodal sodium channel clusters were scattered. These changes were more prevalent in ventral than dorsal roots. These researchers concluded that gangliosides were important lipid raft components which maintained the stability of the Schwann cell/axon architecture at the paranodes. To further investigate the role of ganglioside antibodies in the pathogenesis of neuropathy, recent studies using passive transfer and intraneural injection of monoclonal antiganglioside antibodies have been instructive (David et al., submitted). This study showed that GM1 and GD1a antibodies delivered by either method caused a reversible conduction block in the rat sciatic nerve. This block occurred in the absence of demyelination. In the presence of activated T cells to open the blood-nerve barrier, the block associated with anti-GD1a antibodies was profound. Moreover, a dose-dependent axonal degeneration was observed. The study did not examine the fine structural changes at the nodes associated with this block but the mechanism was presumed to involve reversible changes in the ion channel function and structure as described by Susuki et al. (2007b). These findings are of particular interest in view of the fact that whereas some patients with AMAN develop a severe axonal neuropathy with poor recovery, others recover rapidly and completely from significant degrees of paralysis.
Recent Advances in Schwann Cell/Axonal Neurobiology
Rapid impulse (salutatory) conduction is dependent on the unique structural adaptations of myelinated Schwann cells. Myelinated Schwann cells are now recognised to be comprised of several functional regions or domains; nodes of Ranvier, paranodes, juxtaparanodes, and the internodal regions. Voltage-gated sodium channels are clustered at the nodes whereas Shaker-type voltage-gated potassium channels, Kv1.1 and Kv1.2, are clustered at the juxtaparanodes (Poliak et al., 2003; Arroyo and Scherer, 2007). Key adhesion molecules are involved in these specialised arrangements. The binding of gliomedin within the Schwann cell microvilli to axonal neurofascin 186 and NrCAM plays a crucial role in anchoring the nodal sodium channel clusters to the cytoskeletal proteins, ankyrin G, and spectrin. In contrast, neurofascin 155 within the Schwann cell paranodal loops interacts with and binds to axonal contactin and Caspr, which in turn are attached to spectrin and ankryin B, thereby forming septate-like junctions separating nodal sodium channels from the juxtanodal potassium channels (Fig. 3) (Eshed et al., 2005; Schafer et al., 2006; Arroyo and Scherer, 2007; Salzer et al., 2008). It is significant that Contactin or Caspr-null mice, as may be expected, lack septate-like junctions and Kv1.1 and Kv1.2 channels are mislocated from the juxtaparanode to the paranode, thus directly opposing nodal sodium channels and consequently slowing the conduction. In such mice, sodium channels are less well confined to the node (Rios et al., 2003).
Thus, the more recent evidence from certain subtypes of GBS has now turned our attention away from the protein components of Schwann cell compact myelin as the likely source of immune targets in inflammatory neuropathy and shown that ganglioside antibodies present in specific forms of these neuropathies bind to the nodal region of myelinated fibres. Moreover, new data from neurobiology has now revealed a number of as yet unstudied molecules which may well be expected to provide new targets in this region.
Nodal Changes in EAN
As described earlier, EAN has been the accepted animal model of GBS or acute inflammatory demyelinating polyneuropathy (AIDP) and is classically produced by immunisation with heterologous whole PNM or peptides derived from compact myelin. Disease induced by peptides such as those derived from P2 or P0, results mainly in a T-cell response and affected animals show little demyelination pathologically whereas whole myelin induces a B- and T-cell response with abundant demyelination seen histologically (Taylor and Pollard, 2001). A recent study in EAN by Lonigro and Devaux (2009) in which EAN was induced by both PNM (EAN–PNM) or neuritogenic P2 peptide (EAN–P2) showed that in animals in which disease was induced by PNM but not in those with P2 peptide-induced disease, there was disruption of sodium channel clusters at the nodes which was associated with loss of the adhesion molecules gliomedin and neurofascin at the nodal region. Reduction in the density of these important adhesion molecules preceded loss of sodium channel clusters and paranodal demyelination and was accompanied by the presence of antibodies to gliomedin and neurofascin. Paranodal junctions were also affected and Kv1 potassium channels normally located at the juxtaparanode were dispersed to the paranode and nodal region. Electrophysiological studies in these animals showed conduction slowing and dispersion, more pronounced in the EAN–PNM group. Widening of the nodes was seen in both EAN groups but was greater in the EAN–PNM group. As the accepted animal model of AIDP, these findings again highlight the possibility that in humans inflammatory demyelinating neuropathy, changes in the nodal region precede and are of greater significance than changes in compact myelin.
Evidence from Response to Therapy
Figure 4 shows the response to therapy of a patient with CIDP treated with intravenous immunoglobulin (IVIg) with the later addition of corticosteroid and azathioprine. The power of ankle dorsiflexion in this patient was measured daily in each leg for 2 years by her husband, a mechanical engineer, using an instrument which measured her capacity to resist a series of weights. It may be noted that over the early months of treatment following the administration of IVIg alone, the strength improved over a few days, remained at an improved level for about 10 days and then declined again such that at the end of the month, her strength was similar to the pre-treatment level. This finding mirrors a common clinical experience wherein treatment needs to be given on a regularly recurring basis. It is of interest following the later addition of the immunosuppressive agents that this biphasic response was gradually superseded by a progressive and maintained increase in strength. Such reversible changes in strength over a short timeframe are not compatible with demyelination and remyelination. A more likely pathophysiological scenario would be a reversible conduction block due to nodal/paranodal changes such as those associated with ganglioside antibodies as discussed earlier. A second example of rapid response to therapy in the field of inflammatory neuropathy, which again cannot be due to remyelination, is seen in patients with multifocal motor neuropathy (MMN). Following treatment with IVIg, patients with MMN may show remarkable improvement in strength, which may well be associated with reversal of physiological conduction block (Fig. 5). Similarly to patients with CIDP, those with MMN need continuing treatment with IVIg, sometimes as frequently as every week or two. Taylor and colleagues (Taylor et al., 2004), studied proximal nerve biopsied from patients with MMN at the site of conduction block and showed multifocal fibre loss with changes consistent with axonal degeneration and regeneration with no evidence of demyelination or remyelination. Clearly in this setting, recovery following IVIg cannot be due to nerve regeneration and in the absence of demyelination, conduction block and its reversal is more likely consistent with changes at the node/paranode as discussed earlier.
Antigenic Targets in CIDP
Antigenic targets in CIDP have been recently reviewed (Pollard, 2007; Hughes et al., 2010). The therapeutic response of patients with CIDP to plasma exchange strongly suggests that humoral mediators such as antibodies play a role in the pathogenesis of this disease. Immunoglobulin and complement binding to myelinated fibres has been reported in some patients (Dalakas and Engel, 1980; Pollard, 2007) with binding seen to either the compact or non-compact myelin of the Schwann cell (Figs. 6a and 6b). However, antibodies to defined antigens have been shown in a relatively small number of cases. These antigens include gangliosides, other glycolipids and several compact myelin protein antigens (Hughes et al., 2006a; Hughes, 2010). Yan et al. (2001) showed pathogenic antibodies to P0 in a small percentage of patients with CIDP who were highly responsive to plasma exchange but the antigenic targets remain unknown in the majority of patients. Because CIDP, like GBS, is now recognised to consist of several subtypes (Joint Task Force of the EFNS and the PNS, 2005; Hughes et al., 2006b), it is likely that differing pathogenic mechanisms involve different immune targets. Immunofluorescence studies showing binding to non-compact myelin (Fig. 6) indicate the need to consider molecules within this Schwann cell region such as myelin associated glycoprotein (MAG), connexin 29, E-cadherin, and so on; however, the main thrust of this article is that the non-compact nodal and paranodal specialisations of the Schwann cell of myelinated fibres provide a prime target for pathogenic antibodies in inflammatory neuropathies such as CIDP. Current studies in our group are examining sera from patients with inflammatory neuropathy for antibodies to molecules located in this region. Elevated levels of antibodies to neurofascin have already been shown in a significantly high percentage of patients with CIDP and GBS (Yan et al., 2010). However, the pathogenic nature of these antibodies remains to be established and many more molecules need to be studied. Recent advances in understanding the complexity of the Schwann cell/axonal relationship have now highlighted a number of molecules which could be considered as potential antigenic targets in CIDP and other inflammatory neuropathies. No doubt there are more to be defined.