This review summarises recent evidence supporting the involvement of the specialised nodal and perinodal domains (the paranode and juxtaparanode) of myelinated axons in the pathology of acquired, inflammatory, peripheral neuropathies.
The identification of new target antigens in the inflammatory neuropathies heralds a revolution in diagnosis, and has already begun to inform increasingly targeted and individualised therapies. Rapid progress in our basic understanding of the highly specialised nodal regions of peripheral nerves serves to strengthen the links between their unique microstructural identities, functions and pathologies. In this context, the detection of autoantibodies directed against nodal and perinodal targets is likely to be of increasing clinical importance. Antiganglioside antibodies have long been used in clinical practice as diagnostic serum biomarkers, and associate with specific clinical variants but not to the common forms of either acute or chronic demyelinating autoimmune neuropathy. It is now apparent that antibodies directed against several region-specific cell adhesion molecules, including neurofascin, contactin and contactin-associated protein, can be linked to phenotypically distinct peripheral neuropathies. Importantly, the immunological characteristics of these antibodies facilitate the prediction of treatment responsiveness.
- inflammatory neuropathy
Statistics from Altmetric.com
Acquired peripheral neuropathies have a wide range of causes, including some that can result in devastating neurological deficits, including paralysis, sensory loss, autonomic disturbances and respiratory failure. Guillain-Barré syndrome (GBS) and chronic inflammatory demyelinating polyneuropathy (CIDP) account for the majority of immune-mediated neuropathies. Within these broad groups, diverse subtypes are increasingly recognised. Because some of these have atypical features and respond poorly to standard therapies, their detection is of clear clinical importance.
Highly specialised regions of the peripheral nerve, in particular the node and paranode, are currently in the spotlight, as their dysfunction following antibody-mediated attack is thought to be critical to the pathogenesis of a significant proportion of these potentially treatable neuropathies. This notion is embodied by the terms peripheral ‘nodopathy’ and ‘nodo-paranodopathy,’ initially coined by Uncini et al,1 and now adopted to represent neuropathies of any aetiology thought to involve dysfunction of the node, and/or paranode (table 1). This review summarises the functional anatomy of the nodal region, and describes this emerging concept, also including a discussion of antibody-mediated pathology affecting the juxtaparanode. As the construction ‘nodo-paranode-juxtaparanodopathy’ is somewhat unwieldy, hereafter, for simplicity, we use the term ‘nodo-paranodopathy’ to encompass neuropathies resulting from pathology affecting either the node, paranode or juxtaparanode.
Organisation and function of myelinated axon domains
During development of vertebrate peripheral nerves, Schwann cells myelinate axons that are larger than ~1 µm in diameter. Each myelin internode is flanked by nodes of Ranvier—the sites where axolemma is most exposed to the extracellular fluid, and where outward and inward currents of salutatory conduction take place. Complex axoglial interactions shape the node, as well as the flanking paranodal and juxtaparanodal regions (figure 1). The axonal and glial components of each domain contain distinct molecular components. Our understanding of these molecules, their contribution to the molecular organisation of myelinated axons and the deleterious effects of their dysfunction is rapidly expanding.2 3
Node (voltage-gated sodium channels, gliomedin, neurofascin-186 and NrCAM)
Peripheral nervous system (PNS) and central nervous system (CNS) nodes are formed by interactions between axons and glial cells, Schwann cells and oligodendrocytes/astrocytes, respectively. Nodal enrichment of voltage-gated sodium channels (VGSC) serves to generate inward current of the action potential, and the internodal compact myelin sheath reduces the internodal capacitance, enabling saltatory conduction. Nav1.2 predominates in developing nodes, later switching to Nav1.6 during myelination.
In the PNS, the nodal localisation of VGSCs is governed by two types of axoglial interactions. The first depends on a complicated set of molecular interactions that are depicted in figure 1. Gliomedin (GLDN)—a cell adhesion molecule (CAM) secreted by Schwann cell microvilli into the extracellular matrix—binds to neurofascin-186 (NF186) in the axolemma. NF186, in turn, recruits the nodal isoform of ankyrin G, which is the key adaptor protein that mediates the binding of NF186, NrCAM and VGSCs to the nodal cytoskeletal protein βIV-spectrin.3–5 The fact that NF186 null mice demonstrate disrupted axonal conduction highlights the role of NF186 in supporting the node.6
The second type of axoglial interaction results from the apposition of two adjacent paranodes during development.7 As internodes elongate during development, the paranodes act as fence, preventing the lateral diffusion of NF186, NrCAM and VGSCs from under the myelin sheath, so that as adjacent internodes meet, the two opposing heminodes form one node.
Paranode (neurofascin isoform 155, contactin-associated protein and contactin)
The structural integrity, and thus function, of paranodes rely on septate-like junctions, which are unique structures comprising neurofascin isoform 155 (NF155) on myelin loops, and heterodimers of contactin-1 (CNTN1) and contactin-associated protein (Caspr) on the axolemma (figure 1). Septate-like junctions are absent in mice that lack CNTN1, Caspr or NF155, such that juxtaparanodal voltage-gated potassium (K+) channels (VGKC) are not excluded from the paranodal regions, allowing these K+ channels to interfere with saltatory conduction. Nodes form normally in these mutant mice, but become altered as mice age; these changes, as well as axonal pathologies, result in severe neurological deficits.8–10 Paranodal disruption, largely through juxtaparanodal VGKC invasion, has also been observed in cerebroside sulfotransferase null mice (CST-/-) , highlighting the essential role of sulfatide in stabilising the paranodal junctions.11 Further, in the mice lacking gangliosides (GalNAcT-/-), Susuki et al demonstrated axolemmal detachment of paranodal myelin loops, resulting in a reduction in Caspr and NF155 staining in both PNS and CNS paranodes and, again, mislocalisation of juxtaparanodal potassium channels to the paranode.12
Juxtaparanode (VGKC, Caspr2, contactin-2)
At the juxtaparanode, a complex of Caspr2 (on the axon), contactin-2 (CNTN2, also known as transient axonal glycoprotein 1; TAG1, expressed on both the axonal and glial membranes13), PSD93/95 and 4.1B link the VGKCs to the axonal spectrin cytoskeleton.14 These VGKCs mostly comprised Kv1.1 and Kv1.2 subtypes, and are involved in the repolarisation phase of action potentials.
Overview of ‘nodo-paranodopathies’
Inflammatory neuropathies are phenotypically heterogeneous. With an evolving understanding of the underlying pathophysiology, newly defined subtypes, overlapping and evolving in their clinical, serological and electrophysiological features, are emerging.
Antibodies have been implicated in the pathogenesis of the inflammatory neuropathies for decades. Originally, as one might expect, traditional myelin antigens, such as myelin-associated glycoprotein and myelin protein zero (P0), were considered to be the likely targets of autoantibodies in demyelinating disorders.15 16 Despite many years of study, however, there has been limited success in showing that such antibodies were widely present in, or pathogenically relevant to, the common types of inflammatory-demyelinating neuropathy.17 18 Attention thus turned to the nodes of Ranvier and surrounding regions. Initial hints to the likely pathological importance of the node came from histological studies of GBS autopsy material,19 later supported by electrophysiology and experimental evidence of nodal targeting by neuropathy-associated antibodies. In the past 5 years, it has become apparent that physiological and/or structural damage to these regions may, at least in part, be responsible for the varied but defined clinical phenotypes in what were previously considered demyelinating or axonal disorders.
Antiganglioside antibodies were the first serum antibodies to be linked with GBS.20 There is a particularly strong association with Miller Fisher syndrome (MFS) subtype, in which antibodies against the GQ1b ganglioside are detected in 95%–99% of cases.21 Although testing for these antibodies is used in clinical practice, the diagnosis of GBS continues to principally depend on the recognition of its clinical and electrophysiological features. More recently, tissue-based assays have been used to identify autoantibodies targeting the node and paranode in up to 30% of patients with immune-mediated neuropathies, including CIDP and the acute inflammatory demyelinating polyneuropathy (AIDP) form of GBS.22 Subsequent studies, largely based on transfected cell-based assays, have implicated NF155, NF186, GLDN, CNTN1 and Caspr as the specific targets for a proportion of these antibodies. Others are undoubtedly still to be identified.
These developments are prompting a revised classification scheme, that is, ‘nodo-paranodopathy,’ which recognises new electrophysiological categories and incorporates these serological biomarkers. It is hoped that these revisions will more accurately reflect shared mechanisms of injury, disease course and potential treatment strategies. Pathological involvement of this region was at first largely inferred from electrophysiological studies. Although serological and pathological studies have strengthened this paradigm, the utility and validity of the terminology remain to be fully established. In particular, the potentially bidirectional relationship between nodal/paranodal disruption and demyelination, and the accuracy of a term that implies exclusively nodal pathological localisation in the presence of antibodies directed against more ubiquitously expressed antigens (notably gangliosides), endure as unresolved issues.
Traditionally, GBS has been classified as ‘demyelinating’ or ‘axonal’ mostly on the basis of electrophysiological studies (figure 2).23 Conduction block (CB) or conduction velocity slowing were taken to indicate demyelination, with axonal degeneration characterised by reduced compound muscle action potential (CMAP) amplitudes, and associated with poor prognosis.
This view has been revised in light of new electrophysiological and experimental data that demonstrate ‘reversible conduction failure’ (RCF) of motor nerves, which can rapidly recover, in an increasing number of cases of ‘axonal’ GBS.24 RCF (sometimes termed axonal CB) is characterised by reversible conduction slowing or block, mimicking demyelination, but without temporal dispersion. It can result from disorders that disrupt the nodal axolemma, that is, ‘nodo-paranodopathies,’ and has been suggested to be the neurophysiological correlate of antibody-mediated attack at the node or paranode (figure 3).24–26 Thus, initial neurophysiology showing CB could represent either nodo-paranodopathy with RCF or demyelination. Furthermore, it should be appreciated that CB can only be demonstrated electrophysiologically if the affected nerve can be stimulated both proximally and distally with respect to the site of injury. As the distal nerve terminals and proximal roots lack a blood–nerve barrier, however, these regions are particularly susceptible to injury by circulating antibodies.26 In keeping with this theory, McGonigal et al further demonstrated a gradient of increasing vulnerability of nodes of Ranvier to antibody (anti-GD1a in this case) and complement mediated injury within the distal motor nerve terminal, from the protected proximal nodes to more exposed distal nodes. Even if the pathology at proximal or distal segments is axonal CB or RCF, standard electrophysiology will only be able to show prolonged distal motor latencies (DMLs - with distal pathology), or uniformly reduced CMAPs and/or delayed/absent F-waves (with proximal pathology). Currently accepted electrodiagnostic criteria, which consider CB, prolonged DMLs and delayed F-waves as purely demyelinating phenomena, likely underestimate the proportion of axonal GBS variants, particularly when neurophysiological studies are performed early in the disease course.25 27 It has therefore been proposed that individual patients should undergo serial studies to reduce misclassification.
As an alternative to serial neurophysiological studies, criteria have been proposed to optimise the reliability of a single study, thereby increasing the diagnostic sensitivity of demyelination as well as differentiating axonal subtypes within the first week.25 Specifically, retrospective analysis of nerve conduction studies (NCS) in a large cohort of patients with GBS, comparing the use of existing and modified criteria, revealed a significant shift in classification from AIDP (72% to 56%) and equivocal (10% to 8%) to axonal GBS (18% to 35%), which was further subdivided into acute motor axonal neuropathy (AMAN) and acute motor-sensory axonal neuropathy. These numbers closely resemble the diagnostic shift seen in patients who undergo serial NCS,27 although they have yet to be validated with cohorts using serial studies.
Based on these updated findings, a new study found that by retrospectively applying modified ‘early RCF criteria’ and ‘de/remyelination criteria’ in patients with GBS who underwent serial NCS, it was possible to define two distinct groups of patients.28 However, limitations, including inadequate specificity of electrophysiological cut-off values to categorise all patients, and the lack of clinical and prognostic correlation, should motivate subsequent studies to validate and strengthen these conclusions.
Patterns of electrophysiological abnormalities associated with antibody-mediated attack at the node and paranode have been described,24 and include either or a combination of axonal loss with universally reduced CMAP amplitudes (axonal degeneration), and acute motor CB with or without RCF. Of course, CB could be a manifestation of demyelination, and indeed the distinction between types of CB may be aided by the presence or absence of certain antibodies.28 29 In GBS, far fewer electrophysiological features of demyelination, in conjunction with rapid resolution of reduced CMAPs and conduction slowing, were observed in IgG anti-GM1 antibody positive, compared with seronegative, patients.29 Here, the CB was considered secondary to antibody-mediated loss of VGSCs, reducing conduction at the node, and would account for the rapid recovery observed in some patients with AMAN. More generally, mechanisms leading to ‘axonal CB’ are likely to be broad, and include (1) myelin detachment at the paranode, (2) nodal lengthening (which might both be considered limited forms of demyelination), (3) disruption of nodal VGSCs and generation of the action potential, and (4) disorganised polarisation leading to an ‘inexcitable’ axolemma (figure 2D).30
Ultimately, distinguishing transitory CB from axons irreversibly committed to degeneration has clear implications for the reversibility of peripheral nerve injury. The factors influencing whether axolemmal block will spontaneously resolve, transition to axonal degeneration or indeed whether it can progress to segmental demyelination are unknown and the subject of ongoing research.30 In the context of spinal cord contusion, recent investigations support the concept of transiently damaged myelinated axons existing in a ‘metastable state,’ in which they maintain the capacity for spontaneous recovery, but beyond which permanent damage would occur.31 It has been demonstrated in vivo that this sublethal state is characterised by intra-axonal accumulation of calcium, and that the application of a timely intervention, in this case sealing of calcium-permeable plasma-membrane pores, can favourably alter axonal fate. With an expanding understanding of factors determining progression of CB to axonal degeneration in peripheral neurons, it should follow that similar principles could be applied.
As eluded to previously, the molecular architecture of the nodal regions reflects their function to initiate, maintain and control conduction, and depends on specialised axoglial interactions. Pathological studies of molecular organisation of peripheral nerves could provide insights into how the nodal region is affected in patients with GBS and CIDP, but are infrequently performed due to the invasive nature of nerve biopsy. Skin biopsies, however, can give a morphological assessment of myelinated nerve fibres, and have provided evidence of significant nodal disruption, particularly in patients with CIDP.32 Specifically, in both skin and nerve33 biopsies, elongation of the nodes, and localisation of Caspr seem to be reliable diagnostic markers to differentiate between segmental demyelination and nodal/axonal pathology. Further, using VGSCs and Caspr as markers of the node and paranode, respectively, Cifuentes-Diaz et al were able to demonstrate altered structure at the node and paranode in peroneal nerve biopsies of 12 patients with CIDP,33 though this was not recapitulated in the dermal nerve fibres.
In axonal neuropathies, such as AMAN, several factors observed in human and rabbit models also point towards pathological involvement of the node.34 35 In the rabbit model of AMAN, Susuki et al described a specific mechanism of nodal disruption in peripheral motor nerves, where anti-GM1 antibodies incur complement-mediated disruption of sodium channel clusters in association with membrane attack complex pore formation in the acute phase. Contributing to this, the axonal cytoskeleton and Schwann cell microvilli, and crucially their axoglial interactions, are thought to be disrupted, further destabilising sodium channel clusters.34 In addition, nodal lengthening and paranodal myelin detachment were noted, again affecting depolarisation and manifesting electrophysiologically as CB.30 These findings have important therapeutic implications, as this process has been shown to be reversed by the addition of complement inhibitors.26
Neuropathies and the node
The node of Ranvier and paranode have emerged as targets for immune-mediated pathology in GBS, CIDP and multifocal motor neuropathy (MMN).36 Gangliosides, glycosphingolipids containing sialic acid, are concentrated at the node and paranode and exposed axonal membranes at nerve terminals, and antibodies directed against these molecules are commonly associated with ‘axonal’ forms of GBS.37 In contrast, nodal adhesion molecules, in particular NF155, NF186, GLDN and CNTN, are targeted in some neuropathies traditionally considered ‘demyelinating,’ such as CIDP, where their dysfunction leads to VGSC alteration and conduction defects.22
Since their original detection in chronic paraproteinaemic neuropathy in the 1980s,38 antibodies against gangliosides20 39 have been identified in a diverse variety of inflammatory neuropathies. Of note, some of these have been historically considered axonal, and others demyelinating. There is now evidence to suggest that this apparent dichotomy is resolved by reconsidering these conditions as nodopathies.
Gangliosides are widely distributed throughout cell membranes in both the PNS and CNS. The relative abundance of the different types varies across motor and sensory nerves,40 and between individual cranial nerves.41 Antiganglioside antibodies are reported to target the node (GM1,42 GD1a,26 GD1b/disialosyl43) or paranode (GQ1b).44 However, these binding patterns do not always correlate with the distribution of the antigen revealed by other methodological means,45 and may be further influenced by variations in the fine specificity of different antiganglioside antibodies.
IgG antibodies to GM1, a widely expressed ganglioside at the nodal axolemma and on Schwann cell microvilli, are most frequently associated with AMAN42 and MMN,46 serving as a pathophysiological link between these acute and chronic immune neuropathies. The mechanism by which anti-GM1 antibodies produce purely motor phenotypes is not well understood, however, as GM1 itself is known to be present in similar abundance in both motor and sensory nerves.40 The fact that the ability of some anti-GM1 antibodies to access their target antigen is influenced by other cis interacting gangliosides within cell membranes47 may be relevant. In a recent study, however, Harschnitz et al confirmed that human-induced pluripotent stem cell (hiPSC)-derived motor and sensory neurons both express GM1, and demonstrated targeting of each cell type by IgM GM1 antibodies. Although such antibodies activate the classical complement pathway in both settings, motor nerves prove far more susceptible to subsequent axonal injury, for as yet unclear reasons.48 Another line of evidence demonstrates that antiganglioside antibodies are rapidly internalised at certain sites, and that this internalisation attenuates complement-mediated injury.49 50 Region variations in this protective mechanism, or its localised failure, might also provide an explanation for the focality of antibody-mediated injury in the context of a more widely expressed target antigen.
Acute and chronic ataxic neuropathies have been grouped by their association with antibodies directed against disialosyl gangliosides (which contain two sialic acid residues). In these conditions, a clear clinical-serological-pathological phenotype has emerged, characterised by sensory ataxia with relatively preserved motor function, antidisialosyl ganglioside antibodies and common pathophysiological effects centred on the node.51
Akin to the situation with GM1, above, anti-GD1b antibodies are found in both acute and chronic ataxic neuropathies, with IgG and IgM isotype predominance, respectively. In acute ataxic neuropathies, all of which could be considered part of the same clinical spectrum,43 anti-GD1b IgG antibodies have been linked to acute sensory and ataxic neuropathy, considered a GBS variant.52 These antibodies may cross react with GQ1b/GT1a, targets more commonly associated with MFS,21 ataxic GBS53 and the pharyngeal-cervical-brachial variant of GBS. GD1b reactivity seems to be specific for Romberg-positive sensory ataxia, in contrast to anti-GQ1b antibodies, which more closely correlate with the Romberg-negative, ‘cerebellar-like’ ataxia seen clinically in MFS or ataxic GBS.43 The latter pathology is considered secondary to involvement of muscle spindle afferent fibres to the spinocerebellar system,51 rather than pathology of the cerebellum itself. Crucially, the presence of GQ1b antibodies correlates with prompt and complete recovery, both clinically and electrophysiologically. In the absence of demyelinating features, this rapid reversibility is felt to be indicative of immune-mediated RCF, localising the pathology to the node/paranode42 and/or the nerve terminal. Evidence of ganglioside antibody and complement-mediated nodal alteration in GBS rabbit models,1 corroborated in one ataxic GBS anti-GQ1b antibody-positive patient where antibodies to GQ1b and GD1b specifically stained the nodes of sensory fibres,54 support this theory and suggest a final common pathway involving nodal VGSC and axoglial junction disruption causing RCF. Unchecked, this would be expected to eventually lead to irreversible axonal degeneration.55
IgM class antidisialosyl antibodies associated with chronic ataxic neuropathies, such as CANOMAD (chronic ataxic neuropathy, ophthalmoplegia, monoclonal IgM paraprotein, cold agglutinins and disialosyl antibodies), have also been shown to target the nodal axolemma, most recently in a myelinating culture system using hiPSC-derived sensory neurons. Interestingly, such reactivity could subsequently lead to either axonal degeneration or demyelination, determined in this model by the availability or otherwise of a source of complement.56 This is consistent with previous pathological studies, which have revealed a mixture of axonal and demyelinating features (reviewed in ref 51), and suggests that initial CB may rapidly reverse, or be a prelude to either axonal degeneration or ‘classical’ demyelination (figure 3). The pathological outcome is likely to be influenced by the duration of antibody exposure as well as the activity of the complement system.
Nodal protein antibodies
Experimental allergic neuritis (EAN) is presented as a model of demyelinating GBS/AIDP. In a paradigm using peripheral myelin as an immunogen, however, pathological effects were attributable to the disruption of nodal adhesion molecules and paranodal demyelination. Specifically, antibodies against NF186 and GLDN resulted in dispersion of VGSCs and altered conduction, although, again, regions of segmental demyelination were also observed.57 IgG antibodies with these specificities were also detected in the sera 62% of patients with MMN, in whom they were speculatively linked with motor nerve CB. Again, as these CAMs are found at all nodes,58 the mechanism by which they cause the motor-specific and localised injury required to produce isolated motor CB remains to be determined.
Recently, Devaux et al observed IgG antibodies targeting the nodal regions of rodent teased-nerve fibres in 44% of patients with AIDP, 42% AMAN and 30% CIDP, with specificity for the CAMs at the axoglial interface. Antibodies against NF186 showed a significant correlation with AMAN, whereas antibodies reactive to the glial protein GLDN were more commonly found in patients with AIDP. The fact that specific antigen targets were not detected in over half of these patients indicates there are more to be found. Interestingly, all AMAN sera with antibodies against nodal adhesion molecules (NF186, GLDN or CNTN), also contained anti-GM1 IgG antibodies, raising the possibility that pathology in some patients with GBS may be driven by more than one antibody.
The downstream processes that occur after nodal-antibody binding seem to involve both humoral and cellular components. In a passive transfer model of anti-GLDN-mediated inflammatory neuropathy, early nodal disruption and paranodal demyelination was associated with IgG deposition and complement activation, whereas later demyelination was associated with prominent T-cell and macrophage infiltration, with little to no IgG deposition. Of further note, passive transfer of anti-GLDN IgG to P2 immunised EAN rats substantially increased the frequency of demyelination without a similar effect on axonal degeneration.59
Peripheral nodopathies can also be induced by toxic, ischaemic, nutritional and genetic mechanisms. These are worth mentioning to highlight the importance of the node in non-immune-mediated disease processes. As many as 70% of patients who are critically ill develop neuropathies, and rat models mimicking sepsis reveal reduced nodal VGSC activity and axonal depolarisation, again causing reversible CB.60 There is currently no recognised treatment for critical illness neuropathy, but a future ability to detect the early stages of this disease process could pave the way for disease-modifying therapies aimed at preventing subsequent axonal degeneration. Acute and chronic ischaemic neuropathies, often a result of a systemic vasculitis, have demonstrated reversible CB, presumably arising via an impact on the energy-dependent nodal ion channel pumps. However, this is thought more likely to lead to axonal degeneration than not.30
Neuropathies and the paranode
Axoglial adhesion molecules at the paranode, anchoring the terminal myelin loops to the axon, are the target of antibodies in some neuropathies. Most notably, around 10% of patients with CIDP have a distinct phenotype-serotype association with IgG antibodies against CNTN1 or NF155.61–63 Evidence of their specificity and pathogenicity continues to accumulate.64
NF has been proposed as a candidate nodal antigen in inflammatory demyelinating diseases of both the CNS65 and PNS.22 IgG4 antibodies against NF155 were originally detected in small cohorts of patients with inflammatory neuropathies, sharing some similar features but with varying response to treatment. Ng et al detected anti-NF155 IgG1 and/or IgG3 antibodies at low frequency in patients with AIDP (3/65) and CIDP (5/119).66 This was followed by the detection of a similar proportion of IgG4 anti-NF155 antibodies in a Spanish CIDP cohort.67 Overall, this antibody has been found between 4% and 18% of patients with CIDP.63 67 68 It is now apparent that the presence of anti-NF155 antibodies, specifically of the IgG4 subtype, is associated with a clinically homogenous CIDP phenotype, specifically being more aggressive, distal, motor predominant, and associated with ataxia, tremor and good response to rituximab but not IVIg.62–64 67 69 70 Further defining the features of anti-NF155 antibody-positive patients with CIDP, Ogata et al noted a younger age at onset, with a significantly higher cerebrospinal fluid protein, potentially reflecting prominent spinal root involvement.68 This typical phenotype may not apply to all patients, however, with previous reports of anti-NF155-positive patients also displaying CNS demyelination and good response to IVIg.65
Pathological studies suggest that these antibodies cause destabilisation of the transverse bands (also known as septate-like junctions), which link the paranodal myelin loops to the axon, with subsequent conduction slowing, likely secondary to nodal widening and paranodal demyelination.71 Expanding on these findings, Koike et al noted that the macrophage-mediated demyelinating process thought to characterise CIDP was not apparent in nerve biopsies from 10 predominantly anti-NF155 antibody-positive patients with CIDP, indicating an alternative pathogenic mechanism.72 The fact that IgG4 subclass antibodies do not effectively fix complement suggests that anti-NF155 antibodies are likely to act by blocking interactions with the Caspr/CNTN1 complex. The pathological importance of these antibodies is further supported by the observation that IgG4 depletion with rituximab correlates with clinical recovery in CIDP.69
Anti-CNTN1 IgG4 antibodies are detected in a lower proportion of patients with CIDP, who again share a relatively uniform clinical phenotype, characterised by older age, aggressive disease onset, motor predominance, early axonal loss and poor IVIg responsiveness.73 They potentially induce their pathological effects via a comparable mechanism to that proposed for NF155 antibodies. They have been associated with specific paranodal structural alterations in myelinated axons,74 and block axoglial interactions mediated by the Caspr-CNTN1-NF155 complex, without binding Fc receptors or triggering complement activation.75 This may be of relevance to the IVIg resistance observed in anti-CNTN1-positive patients. Interestingly, fine alterations or mutations in the CNTN1-specific N-linked carbohydrate residues, which are essential for glycoprotein recognition, may not cause gross structural abnormalities but can influence the selectivity of CNTN1 binding, potentially disturbing the paranodal complex and inducing demyelination.75 Of note, a homozygous mutation in the CNTN1 gene, thought to significantly reduce its expression at the neuromuscular junction, was detected in four neonates from a consanguineous family, with a lethal congenital myopathy syndrome.76
Caspr has only recently been highlighted as a candidate target antigen in a study using serum from a cohort of 57 patients with GBS or CIDP, one from each group showing binding of Caspr at the paranodes of rodent teased fibres.77 Neuropathic pain was a prominent feature in the anti-Caspr group, which was not described in patients with other antiparanodal antibodies. The resolution of pain and functional recovery followed rituximab treatment in the patient with CIDP. In addition, the observed pathological features were not those of a small fibre neuropathy, the implication being that neuropathic pain was a direct consequence of paranodal antibody binding. This is important because it implicates the paranode as a primary site of pathology in the inflammatory neuropathies, and because (1) it has shown binding at the paranode in a patient with GBS and (2) it demonstrates the involvement of another culprit antibody subclass, IgG3, in the anti-Caspr patient with GBS. Combined with a previous study displaying an IgG3-predominant immunoreactivity in patients with acute, or GBS-like onset CIDP,74 this implies a conversion of IgG3 and complement-mediated pathological mechanisms in the acute, to IgG4 and complement-independent mechanisms in the chronic phase of these disorders. Clearly this has implications for future therapeutic approaches to these specific antibody-mediated conditions. Interestingly, NCS in the anti-Caspr-positive patient with CIDP showed evidence of temporal dispersion, thought to be an unequivocal indicator of demyelination/remyelination, yet nerve biopsy revealed axonal degeneration without demyelination, but with IgG deposition at the paranodes. This further highlights the difficulties in defining disease phenotypes using a single diagnostic modality, and suggests that the pathology in nodo-paranodopathies might not be as well circumscribed as initially envisaged.
One of the currently unanswered questions about these newly appreciated disorders is why antibodies directed against targets present at the paranodes of all myelinated nerves should produce specific patterns of injury. Furthermore, if anti-NF155, anti-CNTN1 and anti-Caspr antibodies all exert their pathological effects by disrupting the same axoglial complex, why do their associated disorders have distinct clinical features? It may be that subtle regional and/or fibre-type differences exist in the conformation and accessibility of paranodal adhesion molecules, although this has not yet been demonstrated.
Dismantling of the paranode has also been observed in the context of inherited neuropathies, namely Charcot-Marie-Tooth (CMT) types 1A and 1C, and congenital hypomyelinating neuropathy (CHN), a rare neonatal syndrome characterised by hypotonia and weakness. In three patients with CHN, mutations in CNTNAP1, the gene-encoding Caspr, were described in association with pathophysiological alterations to the paranode. Here, loss of the transverse bands resulted in paranodal myelin loop detachment and reduced conduction velocities.78 In mouse models mimicking CMT1A79 and 1C,80 paranodal structural changes, including loss of transverse bands and peeling back of myelin loops in the former, or myelin infolding in the latter, are shown to affect conduction and likely contribute to both the demyelination and axonal degeneration seen in these conditions.
Finally, although currently reported at low frequency, mutations in the genes encoding nodal/paranodal CAMs are emerging, and beginning to provide useful insights into their functions in humans. Mutation of the gene-encoding GLDN at the node, and Caspr (CNTNAP1) at the paranode, has been linked to a handful of patients with the recessively inherited lethal congenital contractural syndromes 11 and 7 (LCCS), respectively, the most severe type of widespread joint contractures in neonates.81 82 Interestingly, electron microscopy of the sciatic nerves from affected patients demonstrated significant nodal lengthening, and a reduction of myelinated nerve fibres. In LCCS7, this also corresponded with a marked reduction in motor nerve conduction velocities.
Neuropathies and the juxtaparanode
Antigenic targets at the juxtaparanode are also now emerging, but their associated peripheral neuropathies are less well characterised. Normal function here depends on the stability of the VGKC complex, in which VGKCs colocalise with CNTN2 and Caspr2 in myelinated peripheral neurons. VGKC-complex autoimmunity, originally linked to neuromyotonia, has become increasingly recognised in association with a broad spectrum of CNS more than PNS phenotypes, at least some manifesting as a result of neuronal hyperexcitability.83 Pathogenic IgG antibodies have now been shown to bind associated proteins such as LGI1 and Caspr2, rather than the ion channels themselves.84–86 The mechanisms of injury postulated include reduction of VGKC density, with impairment of repolarisation and neuronal hyperexcitability.87 Mutations in specific potassium channel genes, in particular Kv1.1 (KCNA1), are likewise known result in a syndrome characterised by episodic ataxia and myokymia.88–90
The strong expression of Caspr2 at the juxtaparanode, and its role in clustering VGKCs in myelinated neurons, make it a credible candidate target antigen in peripheral neuropathy. Caspr2 antibodies have been detected in large cohorts of patients with neurological disease, specifically manifesting with a combination of peripheral nerve hyperexcitability (eg, cramps and fasciculations), either in isolation or as part of a disorder of acquired neuromyotonia (Isaacs’ syndrome), as well as insomnia, limbic encephalitis, seizures and dysautonomia.86 Why Caspr2 antibodies selectively target the peripheral nerves in some cases, given the Caspr2 antigen is found in both the CNS and PNS, remains unclear.
More recently, Caspr2-IgG has been linked to neuropathic pain.87 88 There are limited data to associate Caspr2 antibodies with a distinct neuropathy phenotype, and no in vivo pathogenicity studies to date. However, up to 17% of Caspr2 seropositive patients have evidence of a sensorimotor polyneuropathy.83 One patient was reported to present with an aggressive and treatment-unresponsive axonal GBS, in the presence of lung adenocarcinoma.91 In contrast, in two Caspr2-IgG-positive paediatric cases of GBS,92 and in a small cohort of older male patients mainly presenting with a full complement of hyperexcitable symptoms and Caspr2 antibodies, treatment with IVIg induced complete recovery over 3–6 months.93
In one patient with seizures, a painful peripheral neuropathy and neuromyotonia, a homozygous mutation in the gene-encoding Caspr2 (CNTNAP2) was detected, but its pathogenicity was not confirmed in knockout mice, implying there may be crucial involvement of associated, or perhaps unidentified proteins.86 Subsequent genetic studies have identified CNTNAP2 mutations in patients with complex epilepsy syndromes. Specifically, these were present in 13 individuals with cortical dysplasia-focal epilepsy syndrome,94 characterised by treatment-resistant epilepsy, gross motor deficits, mental retardation and reduced or absent deep tendon reflexes, and in three patients with Pitt-Hopkins-like syndrome 1,95 manifesting with seizures, severe congenital anomalies and mental retardation.
Although CNTN2/TAG-1 has not been implicated as an antigenic target in the pathogenesis of peripheral neuropathies, it is worth noting that genetic association studies have revealed the presence of specific single nucleotide polymorphism in the TAG-1 gene significantly associates with IVIg responsiveness in Japanese patients with CIDP.96 97 A further study of 24 Chinese, predominantly treatment-responsive patients with CIDP, however, did not find evidence of this, 98 although differences in methodology between studies limit our ability to make firm conclusions.
Clinical practice points—antibodies
IgG class GQ1b antibodies help identify complex ataxic/brainstem syndromes as inflammatory, providing diagnostic reassurance and support for the use of immunomodulatory therapy.
Testing for other antiganglioside antibodies in patients with GBS helps differentiate pathological subtypes but does not add to prognostication, and, as yet, does not support the use of non-standard therapies.
Paranodal protein antibodies
Antibodies against the nodal proteins NF155, CNTN1 and Caspr aid diagnosis of inflammatory neuropathies with atypical features.
Consider testing for;
rNF155 antibodies in the context of neuropathies with young age of onset, motor predominance, ataxia, tremor or additional central nervous systemCNS involvement.
CNTN1 antibodies in neuropathies with aggressive onset or motor predominance.
Caspr antibodies in patients with rapidly progressive, motor predominant neuropathies and/or prominent neuropathic pain
In patients with any IgG4 subclass paranodal-protein antibody, IVIg is often ineffective and rituximab should be considered.
Our pathophysiological understanding of the specialised nodal regions and their associated axoglial proteins is growing, and is amenable to intriguing hypotheses related to their role in the pathogenesis of immune-mediated attack on the peripheral myelinated axon. Recently, antibodies directed against a number of key adhesion molecules have been implicated, by way of their presence in high titres, in both acute and chronic inflammatory neuropathies. Such developments have revolutionised the distinction between axonal and demyelinating peripheral neuropathies, and are beginning to outline new disease classifications based on seropositivity, enhanced electrophysiological classification and identification of the presumptive underlying pathological targets and mechanisms. These classification schemes offer the promise of improved diagnosis and prognostication, and perhaps more importantly, should herald the use of targeted therapies directed against specifically determined disease processes. The challenge now is to ascertain the optimum methods for antibody detection, and for establishing the critical pathological mechanisms at play in individual patients.
Contributors JF wrote the initial draft of the manuscript, created the figures, revised content following comments from other authors, contributed to the development of its intellectual content and approved the final version. HJW and SSS edited and commented on drafts of the manuscript, contributed to the development of its intellectual content and approved the final version. SR conceived the manuscript, edited and commented on drafts of the manuscript, contributed to the development of its intellectual content and approved the final version.
Competing interests None declared.
Provenance and peer review Commissioned; externally peer reviewed.