Brain, Vol. 122, No. 3, 449-460,
March 1999
© 1999 Oxford University Press
Article |
The distribution of ganglioside-like moieties in peripheral nerves
1 Departments of Neurology and 2 Neuroscience, Johns Hopkins University School of Medicine, Baltimore, Maryland, 3 National Center for Microscopy and Imaging Research at San Diego and 4 Department of Neurosciences, University of California San Diego, La Jolla, California, USA
Correspondence to:
Dr Kazim Sheikh, Department of Neurology, Johns Hopkins University, Pathology Building 509, 600 N. Wolfe Street, Baltimore, MD 21287, USA E-mail: ksheikh{at}welchlink.welch.jhu.edu
 |
Abstract |
|---|
 Top Abstract IntroductionMaterial and methodsResultsDiscussionReferences |
|---|
GM1 ganglioside has been implicated as a target of immune attack in some diseases of the peripheral nervous system. Anti-GM1 ganglioside antibodies are associated with certain acquired immune-mediated neuropathies. It is not clear how anti-GM1 antibodies cause nerve dysfunction and injury; however, sodium and/or potassium ion channel dysfunction at the node of Ranvier has been implicated. To gain insight into the pathogenesis of these neuropathies, we examined the distribution of GM1 ganglioside and Gal(ß1–3)GalNAc moieties in nerve fibres and their relationship to voltage-gated sodium and potassium (Kv1.1, 1.5) channels at the nodes of Ranvier in peripheral nerves from human, rat and dystrophic mice. Gal(ß1–3)GalNAc moieties were localized via the binding of cholera toxin and peanut agglutinin. As a control for the specificity of these findings, we compared the distribution of GM1 moieties to that of the ganglioside GT1b. Our study provides definitive evidence for the presence of Gal(ß1–3)GalNAc bearing moieties on the axolemmal surface of mature myelinated fibres and on Schwann cells. Gal(ß1–3)GalNAc binding sites did not have an obligatory co-localization with voltage-gated sodium channels or the potassium ion channels Kv1.1 and Kv1.5 and are thus not likely carried by these ion channels. In contrast with Gal(ß1–3)GalNAc, GT1b-like moieties are restricted to the axolemma.
Gal(ß1–3)GalNAc moieties; ani-GMI antibodies; voltage-gated sodium and potassium channels; acquired immune neuropathies; cholera toxin
CT = cholera toxin B subunit; DAB = 3,3 diaminobenzidine tetrahydrochloride; FITC = fluoroisothiocyanate; PNA = peanut agglutinin; TTC = tetanus toxin fragment C
|
Introduction |
|---|
|
TopAbstractIntroductionMaterial and methodsResultsDiscussionReferences |
|---|
The biology of gangliosides in nerve fibres has taken on a new relevance for neurologists as the list of disorders associated with elevated titres of antiganglioside antibodies lengthens and the controversy regarding the pathogenetic significance of these antibodies intensifies. For example, Gal(ß1–3)GalNAc, a moiety present in GM1, asialo GM1, and some glycoproteins, is the epitope recognized by anti-GM1 antibodies which are seen in a number of clinical settings. IgM anti-GM1 antibodies are present in the majority of sera from patients with multifocal motor neuropathy (Pestronk and Choski, 1997). Anti-GM1 antibodies, particularly of the IgG class, are present in some patients with either axonal or demyelinating forms of Guillain–Barré syndrome (Yuki et al., 1990
; Ho et al., 1995; Rees et al., 1995; Visser et al., 1995). In Guillain–Barré syndrome, anti-GM1 antibodies are frequently associated with severe cases and a poor prognosis, and with cases following Campylobacter jejuni infection (Yuki et al., 1990; Rees et al., 1995; Visser et al., 1995). Conduction abnormalities in multifocal motor neuropathy and some cases of Guillain–Barré syndrome indicate dysfunction at the nodes of Ranvier, and pathological studies have confirmed damage to the nodal and paranodal structures (Thomas et al., 1991; Kaji et al., 1993; Chaudhry et al., 1994; Griffin et al., 1996; Hafer-Macko et al., 1996).
The possible role of anti-GM1 antibodies in the pathophysiology of these peripheral neuropathies is not well understood. A recent study using voltage clamp techniques found that raised titres of anti-GM1 antibodies influence sodium and potassium currents at nodes of Ranvier (Takigawa et al., 1995), suggesting that these antibodies may modulate the function of sodium and potassium channels present at the node. Consistent with this site of action, GM1 has been localized at the nodes of Ranvier in myelinated peripheral nerve fibres (Ganser et al., 1983; Corbo et al., 1993; Kusunoki et al., 1993; Molander et al., 1997). However, these studies employed relatively low resolution microscopic techniques and could not determine whether Gal(ß1–3)GalNAc moieties were present on axons or surrounding structures.
In interpreting the physiological and clinical implications of anti-GM1 antibodies in peripheral neuropathies, a thorough understanding of the localization of this ganglioside in nerve fibres is essential. In this study, we used high resolution light and electron microscopic techniques to localize Gal(ß1–3)GalNAc moieties in peripheral nerves and examined the relationship of these moieties to voltage-gated sodium and potassium (Kv1.1, 1.5) channels. In addition, we compared normal myelinated fibres with those from dystrophic mice. These animals have large amyelinated fibres in their spinal roots with randomly distributed patches of sodium channels in the absence of nodes of Ranvier (Deerinck et al., 1997). This preparation allowed us to ask whether or not sodium channels have an obligatory co-localization with Gal(ß1–3)GalNAc. To determine the specificity of these findings, we compared the results with those obtained using tetanus toxin to localize GT1b. GT1b was chosen for its biological interest; it is a receptor for tetanus toxin (Rogers and Snyder, 1981; Angstrom et al., 1994).
|
Material and methods |
|---|
|
TopAbstractIntroductionMaterial and methodsResultsDiscussionReferences |
|---|
Cholera toxin B subunit (CT) conjugated to either HRP (horseradish peroxidase) (CT–HRP) or biotin (biotin–CT) (List Biologicals, Campbell, USA), peanut agglutinin (PNA) conjugated to HRP (PNA–HRP) (Dako, Carpinteria, USA) or biotin (biotin–PNA) (Vector Labs, Burlingame, USA) and tetanus toxin fragment C (TTC) conjugated to HRP (TTC–HRP) or fluoroisothiocyanate (FITC) (TTC–FITC) (List Biologicals) were used to determine the distribution of glycoconjugate binding sites. CT binds to the sialyated trisaccharide (Gal(ß1–3)-GalNAc(ß1–4)-Gal) where a sialic acid residue is bound to the internal galactose; such oligosaccharide moieties are carried by GM1 ganglioside. PNA binds to Gal(ß1–3)GalNAc, primarily on glycoproteins, and could cross-react weakly with GM1 (Lotan and Sharon, 1978
; Momoi et al., 1982). TTC binds to B-series gangliosides including GT1b (see Table 1). We used rabbit anti-sodium channel, rabbit anti-Kv1.1 and Kv1.5 antibodies that had been generated against specific peptides as previously described (Dugandzija-Novakovic et al., 1995; Mi et al., 1995). Anti-sodium channel antibodies were kindly provided by Dr R. Levinson (University of Colorado) and anti-potassium channel antibodies were a kind gift from Dr T. Schwartz (Stanford University, USA).
|
Rodent tissue preparation for teased fibres
In most of the studies we used male Lewis rats 10–12 weeks old. In specific studies we used mice, comparing 4–6-week-old dystrophic mice, strain 129/ReJ-lama2dy, with their control heterozygous litter mates. All animals were anaesthetized with Nembutal (1 ml/kg). The animals were perfused through the heart with Ringer's solution at 35°C followed by 2% fresh paraformaldehyde in 0.1 M PBS (phosphate-buffered saline) pH 7.4 for 5 min. The dorsal and ventral spinal roots and sciatic nerves were removed and fixed for an additional 30 min on ice. Nerves were washed in PBS and incubated in collagenase type IV in PBS (1 mg/ml) for 20 min at room temperature. Following a few washes in ice cold PBS, the nerves were desheathed and teased into small bundles of fibres under a stereomicroscope. These teased fibre preparations were immunostained as described below.
Human tissue preparation for teased fibres
Dorsal and ventral roots were obtained at autopsy from a patient with no history of peripheral nerve disease. Tissue was obtained ~16 h post-mortem. The roots were fixed in 2% paraformaldehyde for 1 h over ice and prepared as described above.
Tissue preparation for fresh frozen cryosections
Animals were anaesthetized as described above, and the sciatic nerves biopsied between the sciatic notch and mid-thigh. Normal human sural nerves from routine biopsy were also used. These biopsies were done at Johns Hopkins Hospital as a part of the diagnostic work-up in patients suspected of having peripheral nerve disease and were performed after obtaining informed consent. The biopsied nerves were snap frozen in isopentane at –70°C and cryosectioned. The 8–12 µm longitudinal and cross sections were fixed by air drying on polylysine-coated glass slides for glycolipid extraction and staining.
Localization of toxin and lectin binding in teased fibres
The teased fibres were incubated in working buffer (5% normal goat serum and 1% bovine serum albumin in PBS) for 1 h on ice to block non-specific binding. After blocking, they were incubated with CT–HRP (2–4 µg/ml), PNA–HRP (5–10 µg/ml) or TTC–HRP (5–10 µg/ml) in working buffer at 4°C overnight followed by five 10 min washes in PBS on ice. Binding of these ligands was detected by incubating the nerves with 0.05% 3,3 diaminobenzidine tetrahydrochloride (DAB) and 0.01% hydrogen peroxide in PBS for 5–10 min on ice. The reaction was terminated by washing with PBS. The human spinal roots were labelled with biotin conjugated CT and PNA and developed with streptavidin–FITC (Vector Laboratories) for confocal microscopy. The samples were further teased, mounted in glycerol, cover slipped, and sealed with nail polish for microscopy. Images were acquired using an Axiophot microscope (Zeiss AG) on rodent preparations. Confocal images were acquired on human teased fibres as described below.
Double fluorescence labelling forGal(ß1–3)GalNAc moieties and voltage gated Na+ and K+ channels
All buffers used in these preparations contained 0.1% Triton X-100. Teased fibres were incubated in working buffer for 1 h. After the blocking step the fibres were co-incubated with biotin-CT 10 µg/ml and either anti-sodium channel antibody, anti-Kv1.1 or anti-Kv1.5 in working buffer overnight at 4°C. Similarly, teased fibres were incubated with combinations of biotin–PNA 10 µg/ml and anti-sodium or potassium channel antibodies in working buffer overnight at 4°C. The concentration of all primary antibodies was 5–20 µg/ml. After overnight incubation, teased fibres were washed with working buffer five times (10 min each) on ice. After washing, the teased fibres were incubated with streptavidin-FITC 1 : 100 and rhodamine conjugated anti-rabbit IgG (Jackson ImmunoResearch) 1 : 100 in working buffer containing propidium iodide 5 µg/ml for 1 h, on ice, followed by several washes in PBS. Teased fibres were mounted in Gelvatol. Confocal microscopy was performed using an MRC-1024 system (BioRad) attached to an Axiovert 35M microscope (Zeiss AG). Excitation illumination was with 488 nm, 568 nm and 647 nm light from a krypton/argon laser. Individual images (1024 x 1024 pixels) were saved to optical disc (Pinnacle Micro), converted to PICT format and merged as pseudo-colour RGB images using Adobe Photoshop (Adobe Systems). Because of the overlap of sodium channel and PNA staining, the triple-label images were displayed as two dual fluorescence images side by side for clarity. Digital prints were from a Fujix pictography 3000 printer (Fuji).
To determine whether CT and PNA bind to the same site, nerve fibres were preincubated with either CT or PNA and then stained with PNA or CT as described above. In addition, teased fibre preparations were double stained with CT–FITC and biotin–PNA developed with streptavidin–rhodamine. The working buffer for these preparations did not contain Triton X-100.
Electron microscopy
Fixed, teased nerve fibres were stained with CT–HRP, PNA–HRP and TTC–HRP as above. After developing in DAB, the fibres were washed with PBS, post-fixed in 1% osmium tetroxide in PBS for 1 h, and washed in distilled water. Following dehydration in graded ethanol, teased fibres were embedded in Durcupan ACM resin (Electron Microscopy Sciences) and polymerized at 60°C for 24 h. Three dimensional light microscopy of peroxidase-labelled epoxy-embedded teased fibres was performed and thin sections were cut from stained areas for electron microscopy. The ultrathin sections (100 nm) were cut using a diamond knife (Diatome) and mounted on uncoated copper grids. These sections were imaged at 80 kev. (kito-electron volts) using either a 100CX or 2000FX electron microscope (JEOL).
Chloroform : methanol extraction of glycolipids
Cryosections were treated with chloroform : methanol (C : M) 1 : 1 (volume : volume) for 30 min at room temperature. The C : M treated and adjacent untreated sections were stained with CT-HRP (4–8 µg/ml) or PNA–HRP (10 µg/ml) or TTC–HRP (10 µg/ml) for 1 h at room temperature. These sections were developed with 0.05% DAB and 0.01% hydrogen peroxide in PBS for 5–10 min at room temperature and cover slipped for light microscopy. The images were acquired using an axiophot (Zeiss AG).
|
Results |
|---|
|
TopAbstractIntroductionMaterial and methodsResultsDiscussionReferences |
|---|
The human tissue did not yield clearly interpretable results for double labelling and electron microscopy studies, most likely due to artifactual changes resulting from the prolonged post-mortem interval, therefore these results are not included. Sciatic nerves, dorsal and ventral roots had similar staining patterns and intensity and the results presented below are representative of all these preparations. A summary of the toxin/lectin binding results is shown in Table 2
.
|
Staining with CT
The staining of myelinated teased fibres in normal animals showed CT binding to the nodal gap and the paranodal region of the Schwann cell (Fig. 1A
). At the light microscopic level it was not possible to determine whether CT bound to the nodal axolemma. In these preparations CT frequently stained Schmidt–Lanterman incisures. On fresh frozen cross sections, CT stained the axons and possibly the internodal outer (abaxonal) Schwann cell surface (Fig. 1D). Although compact myelin seemed to have some staining with CT when compared with PNA and TTC stained fresh frozen sections, the intensity of staining does not preclude non-specific binding (for comparison see Fig. 1D–F). CT binding to a human teased fibre is shown in Fig. 2A.
|
|
Electron microscope preparations of myelinated nerve fibres confirmed that binding of CT was to the nodal axolemma (Fig. 3A
). In these preparations CT binding was restricted to the axolemma, with no binding to Schwann cells. In some fibres, presumably those with better penetration of CT into the periaxonal space of the internodes, the paranodal and internodal axolemma were also stained (Fig. 3B and C).
|
In the unensheathed and amyelinated axons from the spinal roots of dystrophic mice, CT binding was diffuse and continuous along the axolemma, (Fig. 3D
). In these preparations the axolemmae of adjacent axons are apposed against each other in the absence of an intervening ensheathing Schwann cell plasmalemma. The myelinated fibres in the same animals appeared similar to myelinated fibres in normal animals. CT binding was extractable with C : M treatment, suggesting that the CT binding sites are carried by glycolipids.
Staining with PNA
In teased fibre preparations, PNA binding was concentrated at the nodal gap with some binding to the outside of Schwann cells (Figs 1C and 2B). In fresh frozen cross sections PNA stained the abaxonal Schwann cell surface and axolemma. There was no PNA binding to compact myelin on cryosections (Fig. 1F).
In electron microscope preparations of myelinated nerve fibres from normal and from dystrophic animals, PNA staining was present on both Schwann cells and axons (Fig. 3E). PNA binding was present on the abaxonal surface and microvilli of the terminal myelin loops of Schwann cells (Fig. 3E). Staining was also seen on the nodal axolemma. In some preparations where the Schwann cell had retracted from the axons, this staining extended to involve the paranodal axolemma. These results were in agreement with the light microscopic observations. In electron microscope preparations of amyelinated axons from the spinal roots of dystrophic mice, PNA binding was diffuse and continuous along the axolemma (Fig. 3E inset).
Unlike CT binding, PNA binding was not affected by C : M treatment, suggesting that at least some of the PNA binding sites are on glycoproteins.
Staining with TTC
Teased fibres (Fig. 1B), cryosections (Fig. 1E) and electron microscope preparations (Fig. 3F) showed TTC binding restricted to the nodal and internodal axolemma, with no staining of Schwann cells or myelin (Fig. 1B and E). In electron microscope preparations of amyelinated axons from the spinal roots of dystrophic mice, TTC binding was diffuse and continuous along the axolemma (Fig. 3F inset).
As with CT binding, TTC binding was extractable with C : M treatment, suggesting that TTC binding sites are carried by glycolipids.
Double fluorescence labelling
Ion channels and ganglioside-like moieties
These preparations in general confirmed the localizations already described for CT and PNA binding in myelinated fibres (Fig. 4A and B), with the exception that CT binding to the nodal gap was not observed most likely due to the use of Triton X-100 (see below). The myelinated fibres in the dystrophic animals showed staining patterns similar to normal animals, whereas the amyelinated large axons had diffuse and continuous CT and PNA staining along the outer surface of the axolemma (Fig. 4C and D). Some patches of more intense CT binding were observed (Fig. 4C). The amyelinated regions in the nerve roots of dystrophic animals do not have any Schwann cells and the axolemmae of adjacent axons abut against each other. The absence of Schwann cells in these amyelinated regions is confirmed by the absence of propidium iodide staining (Fig. 4C and D).
|
In myelinated fibres of normal and dystrophic animals the anti-sodium channel antibodies stained the nodal region (Fig. 4A
) as described previously (Ritchie et al., 1990), and this staining co-localized with PNA and CT staining at the nodal gap (Fig. 4A and B). The antibodies to the potassium channel Kv1.1 stained the juxtaparanodal region of myelinated fibres (Fig. 4E and F). This staining was adjacent to CT staining at the paranodes (Fig. 4E). The staining for potassium channel Kv1.5 was present in the paranodal region (Schwann cell canaliculi) as described previously (Mi et al., 1995). Although Kv1.5 and CT both stained the paranodal region, the two staining patterns did not completely overlap (Fig. 4G).
In regions containing amyelinated axons the anti-sodium channel antibodies stained clusters of sodium channels randomly distributed along the axolemma as described previously. These patches of sodium channels did not co-localize with either CT or PNA staining (Fig. 4C and D).
Neither CT nor PNA blocked binding of the other. In the double labelling studies (Fig. 4H) the binding patterns of CT and PNA could easily be differentiated. In these double labelling studies, nodal CT binding was preserved by the exclusion of detergent from the working buffer.
|
Discussion |
|---|
|
TopAbstractIntroductionMaterial and methodsResultsDiscussionReferences |
|---|
The present study demonstrates the distribution of Gal(ß1–3)GalNAc and TTC binding sites in mature myelinated fibres in three species: human, rat and mouse. The distribution of these binding sites proved to be similar in all three species. Our study definitively demonstrates for the first time the presence of Gal(ß1–3)GalNAc epitopes on the axolemmal surface of mature myelinated fibres, and confirms the presence of these epitopes on the Schwann cell (Ganser et al., 1983
; Corbo et al., 1993; Molander et al., 1997). Further, we found that these Gal(ß1–3)GalNAc binding sites do not have an obligatory co-localization with voltage-gated sodium channels or the potassium ion channels Kv1.1 and Kv1.5 and are likely not carried by these ion channels. In contrast with Gal(ß1–3)GalNAc, TTC-binding moieties are restricted to the axolemma.
Distribution of Gal(ß1–3)GalNAc binding sites in nerve fibres
The results obtained in this study suggest that CT and PNA have similar yet distinct staining patterns. Further, the experiments using C : M extraction suggest that CT binding sites are present on a glycolipid, most likely GM1, and PNA binding sites are likely carried on glycoprotein(s). The peripheral nerve glycoproteins that are known to bind PNA include N-CAM (neural cell adhesion molecule), cytotactin, cytotactin-binding proteoglycan, P0 glycoprotein, OMgp and Versican. Among these, all but P0 glycoprotein have been reported to be concentrated at the nodes of Ranvier (Shuman et al., 1983; Rieger et al., 1986; Hoffman et al., 1988; Crossin et al., 1989; Apostolski et al., 1994).
The electron microscopic localization results convincingly demonstrate that Gal(ß1–3)GalNAc binding sites are carried on the axolemma. CT binding to the nodal axolemma was suggested by previous light microscopic teased fibre studies (Ganser et al., 1983; Corbo et al., 1993; Kusunoki et al., 1993), and these observations were confirmed by electron microscopy with both CT and PNA. It was possible to determine from these ultrastructural studies that the CT and PNA binding was present on the nodal axolemma and not on the Schwann cell basement membrane overlying the nodal gap. These studies thus provide evidence that the internodal axolemma of myelinated nerve fibres may also carry Gal(ß1–3)GalNAc moieties. Further evidence that Gal(ß1–3)GalNAc binding sites are present on the axolemma was provided by the electron microscopic studies performed on the large amyelinated axons in the spinal roots of dystrophic mice where continuous staining of the axolemma with PNA and CT was seen. These results are consistent with biochemical studies that have shown the presence of GM1 ganglioside in axonal fractions of nerve (Svennerholm et al., 1994).
Localization of Gal(ß1–3)GalNAc binding sites and ion channels around the nodes of Ranvier
The distribution of sodium and Kv1.1 and Kv1.5 potassium channels around nodes of Ranvier has been determined previously (Mi et al., 1995). Since anti-GM1 antibodies have been reported to affect the physiological function of sodium and potassium channels (Takigawa et al., 1995), the distribution of Gal(ß1–3)GalNAc binding sites in relation to these ion channels is of interest. Although sodium channel immunostaining co-localized with CT and PNA staining at the nodal gap, the results of the double labelling experiments on spinal roots from dystrophic mice demonstrate that Gal(ß1–3)GalNAc binding sites do not have obligatory co-localization with sodium channels. Immunostaining for Kv1.1 potassium channels, shown by previous studies to be present on the juxtaparanodal axolemma (Mi et al., 1995), did not co-localize with either CT or PNA binding. The potassium channel Kv1.5 was localized to the paranodal region of the Schwann cell surface but the pattern of staining was different from that of CT binding at the paranode. Thus, Gal(ß1–3)GalNAc binding sites had regional co-localization with sodium and potassium channels in and around nodes of Ranvier but there was no obligate co-localization between these binding sites and ion channels. These sites are neither present as part of the glycosylated tail of these channels, nor necessarily associated with them.
Implications for immune mediated peripheral nerve damage
The presence of Gal(ß1–3)GalNAc binding sites in and around nodes of Ranvier may be relevant to the pathogenesis of some acquired immune mediated neuropathies. Elevated titres of antibodies against Gal(ß1–3)GalNAc epitopes are found in the majority of patients with multifocal motor neuropathy (Pestronk et al., 1988; Pestronk and Choski, 1997), which is a chronic, demyelinating neuropathy selectively affecting motor fibres and associated with persistent conduction block. Increased titres of anti-GM1 antibodies are also associated with both demyelinating and axonal variants of Guillain–Barré syndrome (Yuki et al., 1990; Walsh et al., 1991; Kornberg et al., 1994; Ho et al., 1995; Rees et al., 1995). In light of the present results, it can be speculated that antibodies directed against Gal(ß1–3)GalNAc epitopes might injure paranodal Schwann cells and cause demyelination. In addition, the demonstration of Gal(ß1–3)GalNAc on axonal surfaces provides a potential explanation for axonal Guillain–Barré syndrome. Thus, it is possible that anti-GM1 antibodies could result in injury to both axons and Schwann cells.
The mechanism of conduction block in anti-GM1-associated neuropathies is still debated (Thomas et al., 1991; Santoro et al., 1992; Uncini et al., 1993; Harvey et al., 1995; Roberts et al., 1995; Wirguin et al., 1995; Hirota et al., 1997). Alteration of sodium and potassium ion channel function and injury to the structures around the nodes of Ranvier have been proposed as the underlying mechanisms for conduction block at the nodes of Ranvier (Takigawa et al., 1995; Hirota et al., 1997). There are at least three observations which implicate complex gangliosides in the modulation of ion channel function in the nervous system. The study of Takigawa et al. (1995) using a voltage clamp technique on isolated single myelinated nerve fibres, demonstrated that anti-GM1 antibodies, in the absence of complement, increased the rate of rise and the amplitude of the K+ current. These antibodies in the presence of complement decreased the Na+ current, and caused non-specific current leakage (Takigawa et al., 1995). Our demonstration that GM1 ganglioside is distributed along the axolemma where Na+ and K+ (Kv1.1, Kv1.5) channels are localized provides a potential explanation for these results. Alternatively, ion channel dysfunction at the nodes could possibly be mediated by complement activation. Anti-GM1 antibodies are usually of IgM and IgG1 and IgG3 subtypes (Willison and Veitch, 1994); these immunoglobulins can activate complement. Our previous studies have shown deposition of activated complement at the nodes of Ranvier in the spinal roots of patients with acute motor neuropathy (Hafer-Macko et al., 1996). The activated complement cascade can in turn activate phospholipase A2 (Panesar et al., 1997) which releases arachidonic acid, and this free fatty acid has been shown to alter the function of sodium and potassium (Kv1.1 and 1.5) channels (Honore et al., 1994; Gubitosi-Klug et al., 1995; Bendahhou et al., 1997). Roberts et al. (1995) have shown that anti-GM1 antibodies can inhibit motor nerve terminals at neuromuscular junctions in phrenic nerve diaphragm preparations in a complement independent fashion. Motor nerve terminals are known to contain GM1-like epitopes (O'Hanlon et al., 1996; K. Sheikh and J. Griffin, unpublished observations). These two observations suggest that an immune attack against GM1 ganglioside could alter nodal and synaptic ion channel function in the peripheral nervous system. The decreased neural conduction velocity (despite normal central neural morphology) in GM2/GD2 synthase knock-out mice, which lack the capability to synthesize complex gangliosides, including GM1 (Takamiya et al., 1996), further supports the possibility that complex gangliosides have a modulatory effect on normal ion channel functioning. Which specific ganglioside is responsible, by its absence, for the neural conduction delay in these mice is not known, but it is attractive to speculate that it involves alteration of ion channel function as a result of the absence of GM1 at synapses and/or nodes of Ranvier. The results of the present study showing clustering of GM1 at and around the nodes of Ranvier would support such a role.
That the functional consequences of anti-GM1 immune attack may require nodal structural abnormalities as a mechanism of conduction block is suggested by a recent study by Hirota et al. (1997). These workers could not demonstrate conduction block in spinal roots after application of anti-GM1 sera from humans (patients with neuropathy) and animals over a short period of time. They proposed that persistent conduction block in patients with multifocal neuropathy could be secondary to chronic immunoglobulin deposits on demyelinated paranodal axons. These axon-bound immunoglobulins may interfere with Schwann cell recognition of axonal antigens and thus remyelination, causing persistent paranodal demyelination and conduction block. Whatever the mechanism of injury and dysfunction, the presence of target antigens, in this case GM1, on axons is a prerequisite. The demonstration of GM1-like epitopes on the axolemma makes these hypotheses plausible.
TTC binding and specificity issues
All localization techniques used in this study demonstrated that TTC binding sites are restricted to the axolemma. This TTC binding is completely abolished by C : M treatment, which argues that the TTC binding sites in nerve fibres are of a glycolipid nature. TTC binds to B-series gangliosides, with strongest affinity to GT1b and GD1b and possibly with comparable affinity to GQ1b (Rogers and Snyder, 1981; Critchley et al., 1986; Angstrom et al. 1994). Among other major brain gangliosides, GM1 and GD1a also bind TTC but they have approximately tenfold lower affinities than GT1b and GD1b (Rogers and Snyder, 1981). Our localization studies do not identify the TTC binding glycolipids on the axolemma. It seems unlikely that they are GD1b and GQ1b since previous immunohistological studies using monoclonal antibodies against these gangliosides showed them to be restricted to paranodal myelin (Chiba et al., 1993; Kusunoki et al., 1993). In the present study, the localization of TTC binding was very different from that of CT. The CT staining pattern is likely to reflect GM1 binding since a similar distribution has been obtained by using monoclonal anti-GM1 antibodies (Molander et al., 1997). This suggests that TTC does not bind to GM1. An analytical study by Ogawa-Goto et al. (1992) did not demonstrate the presence of GT1b in myelin fractions extracted from the human peripheral nervous system suggesting that it is present on axons. In view of the above findings it is likely that among major brain gangliosides, GT1b is the TTC binding glycolipid on the axolemma.
Technical considerations
The multiple localization techniques used in this study yielded slightly different but complementary results. CT appears to stain along the abaxonal Schwann cell plasmalemma in cryosections, whereas in teased fibre preparations CT predominantly stained the paranodal region of Schwann cell. The different staining patterns in these preparations are probably due both to inherent variation in the degree of GM1 ganglioside expression in different regions of the Schwann cell and to the use of a fixative in teased fibres, which could potentially decrease CT binding along the internodal abaxonal Schwann cell plasmalemma. We have noticed a decrease in ganglioside staining in peripheral nerves with use of fixatives like paraformaldehyde; other investigators have made similar observations (G. O'Hanlon and H. Willison, personal communication). In contrast, in the preparations for electron microscopy, CT only stained axons and not Schwann cells. Although GM1 ganglioside is present in myelin fractions as indicated by chemical studies (Ogawa-Goto et al., 1992; Svennerholm et al., 1994), our immunocytochemical results suggest, but do not conclusively demonstrate, the presence of staining for GM1 in compact myelin. This apparent discrepancy could be due to intricate interactions between glycolipids and myelin proteins which could potentially mask the CT binding sites on GM1 ganglioside. The absence of nodal staining with CT in the confocal microscopy preparations represents the differential effect of detergent (Triton X-100) on the nodal axolemma (single membrane) versus the paranodal region of Schwann cell (which consists of multiple uncompacted myelin/membrane loops). The differential effects of detergent on the nodal axolemma seen in our study is consistent with the results of Molander et al. (1997), who used a monoclonal anti-GM1 antibody to stain rat spinal roots. Spinal root floating section preparations containing Triton showed an absence of nodal staining and a pattern of paranodal staining very similar to our results with CT. No results were available from preparations without Triton in the Molander study as the IgM anti-GM1 antibody used required permeablization.
The present study shows that CT binding is altered by differential tissue preparation and localization techniques and this may be due to the glycolipid nature of the molecules carrying CT binding sites. These results also suggest that multiple techniques may be required to determine the true distribution of these glycolipids.
The present study demonstrates definitively the distribution of CT and TTC binding moieties in mature myelinated nerve fibres. These localization studies provide insight into the potential role of these molecules in health and disease.
|
Acknowledgments |
|---|
This study was funded in part by NIH NS34846, RR04050 and NS14718.
|
References |
|---|
|
TopAbstractIntroductionMaterial and methodsResultsDiscussionReferences |
|---|
Angstrom J, Teneberg S, Karlsson KA. Delineation and comparison of ganglioside-binding epitopes for the toxins of Vibrio cholerae, Escherichia coli, and Clostridium tetani: evidence for overlapping epitopes. Proc Natl Acad Sci USA 1994; 91: 11859–63.
Apostolski S, Sadiq SA, Hays A, Corbo M, Suturkova-Milosevic L, Chaliff P, et al. Identification of Gal(beta 1–3)GalNAc bearing glycoproteins at the nodes of Ranvier in peripheral nerve. J Neurosci Res 1994; 38: 134–41.[Web of Science][Medline]
Bendahhou S, Cummins TR, Agnew WS. Mechanism of modulation of the voltage-gated skeletal and cardiac muscle sodium channels by fatty acids. Am J Physiol 1997; 272: C592–600.
Chaudhry V, Corse AM, Cornblath DR, Kuncl RW, Freimer ML, Griffin JW. Multifocal motor neuropathy: electrodiagnostic features. Muscle Nerve 1994; 17: 198–205.[Web of Science][Medline]
Chiba A, Kusunoki S, Obata H, Machinami R, Kanazawa I. Serum anti-GQ1b IgG antibody is associated with ophthalmoplegia in Miller Fisher syndrome and Guillain-Barre syndrome: clinical and immunohistochemical studies. Neurology 1993; 43: 1911–7.
Corbo M, Quattrini A, Latov N, Hays AP. Localization of GM1 and Gal(beta 1–3)GalNAc antigenic determinants in peripheral nerve. Neurology 1993; 43: 809–14.
Critchley DR, Habig WH, Fishman PH. Reevaluation of the role of gangliosides as receptors for tetanus toxin. J Neurochem 1986; 47: 213–22.[Web of Science][Medline]
Crossin KL, Hoffman S, Tan SS, Edelman GM. Cytotactin and its proteoglycan ligand mark structural and functional boundaries in somatosensory cortex of the early postnatal mouse. Dev Biol 1989; 136: 381–92.[Web of Science][Medline]
Deerinck TJ, Levinson SR, Bennett GV, Ellisman MH. Clustering of voltage-sensitive sodium channels on axons is independent of direct Schwann cell contact in the dystrophic mouse. J Neurosci 1997; 17: 5080–8.
Dugandzija-Novakovic S, Koszowski AG, Levinson SR, Shrager P. Clustering of Na+ channels and node of Ranvier formation in remyelinating axons. J Neurosci 1995; 15: 492–503.[Abstract]
Ganser AL, Kirschner DA, Willinger M. Ganglioside localization on myelinated nerve fibres by cholera toxin binding. J Neurocytol 1983; 12: 921–38.[Web of Science][Medline]
Griffin JW, Li CY, Macko C, Ho TW, Hsieh S-T, Xue P, et al. Early nodal changes in the acute motor axonal neuropathy pattern of the Guillain-Barre syndrome. J Neurocytol 1996; 25: 33–51.[Web of Science][Medline]
Gubitosi-Klug RA, Yu SP, Choi DW, Gross RW. Concomitant acceleration of the activation and inactivation kinetics of the human delayed rectifier K+ channel (Kv1.1) by Ca2+-independent phospholipase A2. J Biol Chem 1995; 270: 2885–8.
Hafer-Macko C, Hsieh S-T, Li CY, Ho TW, Sheikh K, Cornblath DR, et al. Acute motor axonal neuropathy: an antibody-mediated attack on axolemma. Ann Neurol 1996; 40: 635–44.[Web of Science][Medline]
Harvey GK, Toyka KV, Zielasek J, Kiefer R, Simonis C, Hartung H-P. Failure of anti-GM1 IgG or IgM to induce conduction block following intraneural transfer. Muscle Nerve 1995; 18: 388–94.[Web of Science][Medline]
Hirota N, Kaji R, Bostock H, Shindo K, Kawasaki T, Mizutani K, et al. The physiological effect of anti-GM1 antibodies on saltatory conduction and transmembrane currents in single motor axons. Brain 1997; 120: 2159–69.
Ho TW, Mishu B, Li CY, Gao CY, Cornblath DR, Griffin JW, et al. Guillain-Barre syndrome in northern China: relationship to Campylobacter jejuni infection and anti-glycolipid antibodies. Brain 1995; 118: 597–605.
Hoffman S, Crossin KL, Edelman GM. Molecular forms, binding functions, and developmental expression patterns of cytotactin and cytotactin-binding proteoglycan, an interactive pair of extracellular matrix molecules. J Cell Biol 1988; 106: 519–32.
Honore E, Barhanin J, Attali B, Lesage F, Lazdunski M. External blockade of the major cardiac delayed-rectifier K+ channel (Kv1.5) by polyunsaturated fatty acids. Proc Natl Acad Sci USA 1994; 91: 1937–41.
Kaji R, Oka N, Tsuji T, Mezaki T, Nishio T, Akiguchi I, et al. Pathological findings at the site of conduction block in multifocal motor neuropathy [see comments]. Ann Neurol 1993; 33: 152–8. Comment in: Ann Neurol 1994; 35: 246–7.[Web of Science][Medline]
Kornberg AJ, Pestronk A, Bieser K, Ho TW, McKhann GM, Wu HS, Jiang Z. The clinical correlates of high-titer IgG anti-GM1 antibodies [see comments]. Ann Neurol 1994; 35: 234–7. Comment in: Ann Neurol 1994; 36: 455.[Web of Science][Medline]
Kusunoki S, Chiba A, Tai T, Kanazawa I. Localization of GM1 and GD1b antigens in the human peripheral nervous system. Muscle Nerve 1993; 16: 752–6.[Web of Science][Medline]
Lotan R, Sharon N. Peanut (Arachis hypogaea) agglutinin. Methods Enzymol 1978; 50: 361–7.[Medline]
Mi H, Deerinck TJ, Ellisman MH, Schwarz TL. Differential distribution of closely related potassium channels in rat Schwann cells. J Neurosci 1995; 15: 3761–74.[Abstract]
Molander M, Berthold C-H, Persson H, Andersson K, Fredman P. Monosialoganglioside (GM1) immunofluorescence in rat spinal roots studied with a monoclonal antibody. J Neurocytol 1997; 26: 101–11.[Web of Science][Medline]
Momoi T, Tokunaga T, Nagai Y. Specific interaction of peanut agglutinin with the glycolipid asialo GM1. FEBS Lett 1982; 141: 6–10.[Web of Science][Medline]
O'Hanlon GM, Paterson GJ, Wilson G, Doyle D, McHardie P, Willison HJ. Anti-GM1 ganglioside antibodies cloned from autoimmune neuropathy patients show diverse binding patterns in the rodent nervous system. J Neuropathol Exp Neurol 1996; 55: 184–95.[Web of Science][Medline]
Ogawa-Goto K, Funamoto N, Ohta Y, Abe T, Nagashima K. Myelin gangliosides of human peripheral nervous system: an enrichment of GM1 in the motor nerve myelin isolated from cauda equina. J Neurochem 1992; 59: 1844–9.[Web of Science][Medline]
Panesar M, Papillon J, McTavish AJ, Cybulsky AV. Activation of phospholipase A2 by complement C5b-9 in glomerular epithelial cells. J Immunol 1997; 159: 3584–94.[Abstract]
Pestronk A, Choksi R. Multifocal motor neuropathy. Serum IgM anti-GM1 ganglioside antibodies in most patients detected using covalent linkage of GM1 to ELISA plates. Neurology 1997; 49: 1289–92.
Pestronk A, Cornblath DR, Ilyas AA, Baba H, Quarles RH, Griffin JW, et al. A treatable multifocal motor neuropathy with antibodies to GM1 ganglioside. Ann Neurol 1988; 24: 73–8.[Web of Science][Medline]
Rees JH, Gregson NA, Hughes RA. Anti-ganglioside GM1 antibodies in Guillain-Barre syndrome and their relationship to Campylobacter jejuni infection. Ann Neurol 1995; 38: 809–16.[Web of Science][Medline]
Rieger F, Daniloff JK, Pincon-Raymond M, Crossin KL, Grumet M, Edelman GM. Neuronal cell adhesion molecules and cytotactin are colocalized at the node of Ranvier. J Cell Biol 1986; 103: 379–91.
Ritchie JM, Black JA, Waxman SG, Angelides KJ. Sodium channels in the cytoplasm of Schwann cells. Proc Natl Acad Sci USA 1990; 87: 9290–4.
Roberts M, Willison HJ, Vincent A, Newsom-Davis J. Multifocal motor neuropathy human sera block distal motor nerve conduction in mice. Ann Neurol 1995; 38: 111–8.[Web of Science][Medline]
Rogers TB, Snyder SH. High affinity binding of tetanus toxin to mammalian brain membranes. J Biol Chem 1981; 256: 2402–7.
Santoro M, Uncini A, Corbo M, Staugaitis SM, Thomas FP, Hays AP, et al. Experimental conduction block induced by serum from a patient with anti-GM1 antibodies. Ann Neurol 1992; 31: 385–90.[Web of Science][Medline]
Shuman S, Hardy M, Pleasure D. Peripheral nervous system myelin and Schwann cell glycoproteins: identification by lectin binding and partial purification of a peripheral nervous system myelin-specific 170,000 molecular weight glycoprotein. J Neurochem 1983; 41: 1277–85.[Web of Science][Medline]
Svennerholm L, Bostrom K, Fredman P, Jungbjer B, Lekman A, Månsson J-E, et al. Gangliosides and allied glycosphingolipids in human peripheral nerve and spinal cord. Biochim Biophys Acta 1994; 1214: 115–23.[Medline]
Takamiya K, Yamamoto A, Furukawa K, Yamashiro S, Shin M, Okada M, et al. Mice with disrupted GM2/GD2 synthase gene lack complex gangliosides but exhibit only subtle defects in their nervous system. Proc Natl Acad Sci USA 1996; 93: 10662–7.
Takigawa T, Yasuda H, Kikkawa R, Shigeta Y, Saida T, Kitasato H. Antibodies against GM1 ganglioside affect K+ and Na+ currents in isolated rat myelinated nerve fibers [see comments]. Ann Neurol 1995; 37: 436–42. Comment in: Ann Neurol 1995; 37: 421–3.[Web of Science][Medline]
Thomas FP, Trojaborg W, Nagy C, Santoro M, Sadiq SA, Latov N, et al. Experimental autoimmune neuropathy with anti-GM1 antibodies and immunoglobulin deposits at the nodes of Ranvier. Acta Neuropathol (Berl) 1991; 82: 378–83.[Medline]
Uncini A, Santoro M, Corbo M, Lugaresi A, Latov N. Conduction abnormalities induced by sera of patients with multifocal motor neuropathy and anti-GM1 antibodies. Muscle Nerve 1993; 16: 610–5.[Web of Science][Medline]
Visser LH, van der Meché FG, Van Doorn PA, Meulstee J, Jacobs BC, Oomes PG, et al. Guillain-Barré syndrome without sensory loss (acute motor neuropathy). A subgroup with specific clinical, electrodiagnostic and laboratory features. Dutch Guillain-Barré Study Group. Brain 1995; 118: 841–7.
Walsh FS, Cronin M, Koblar S, Doherty P, Winer J, Leon A, et al. Association between glycoconjugate antibodies and Campylobacter infection in patients with Guillain-Barre syndrome. J Neuroimmunol 1991; 34: 43–51.[Web of Science][Medline]
Willison HJ, Veitch J. Immunoglobulin subclass distribution and binding characteristics of anti-GQ1b antibodies in Miller Fisher syndrome. J Neuroimmunol 1994; 50: 159–65.[Web of Science][Medline]
Wirguin I, Rosoklija G, Trojaborg W, Hays AP, Latov N. Axonal degeneration accompanied by conduction block induced by toxin mediated immune reactivity to GM1 ganglioside in rat nerves. J Neurol Sci 1995; 130: 17–21.[Web of Science][Medline]
Yuki N, Yoshino H, Sato S, Miyatake T. Acute axonal polyneuropathy associated with anti-GM1 antibodies following Campylobacter enteritis [see comments]. Neurology 1990; 40: 1900–2. Comment in: Neurology 1991; 41: 1327–8, Comment in: Neurology 1991; 41: 1530–1, Comment in: Neurology 1993; 43: 1443–4.
Received July 13, 1998. Revised September 28, 1998. Accepted October 8, 1998.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  H. C. Lehmann, G. Meyer zu Horste, B. C. Kieseier, and H.-P. Hartung Review: Pathogenesis and treatment of immune-mediated neuropathies Therapeutic Advances in Neurological Disorders, July 1, 2009; 2(4): 261 - 281. [Abstract] [PDF]
|
|
|
|
 |
|
 K. Kaida, T. Ariga, and R. K Yu Antiganglioside antibodies and their pathophysiological effects on Guillain-Barre syndrome and related disorders--A review Glycobiology, July 1, 2009; 19(7): 676 - 692. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K. Kaida, M. Sonoo, G. Ogawa, K. Kamakura, M. Ueda-Sada, M. Arita, K. Motoyoshi, and S. Kusunoki GM1/GalNAc-GD1a complex: A target for pure motor Guillain-Barre syndrome Neurology, November 18, 2008; 71(21): 1683 - 1690. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 P. H. H. Lopez, G. Zhang, M. A. Bianchet, R. L. Schnaar, and K. A. Sheikh Structural requirements of anti-GD1a antibodies determine their target specificity Brain, July 1, 2008; 131(7): 1926 - 1939. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
E. Shavit, O. Beilin, A. D. Korczyn, C. Sylantiev, R. Aronovich, V. E. Drory, D. Gurwitz, I. Horresh, R. Bar-Shavit, E. Peles, et al. Thrombin receptor PAR-1 on myelin at the node of Ranvier: a new anatomy and physiology of conduction block Brain, April 1, 2008; 131(4): 1113 - 1122. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K. Susuki, M. N. Rasband, K. Tohyama, K. Koibuchi, S. Okamoto, K. Funakoshi, K. Hirata, H. Baba, and N. Yuki Anti-GM1 Antibodies Cause Complement-Mediated Disruption of Sodium Channel Clusters in Peripheral Motor Nerve Fibers J. Neurosci., April 11, 2007; 27(15): 3956 - 3967. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J T H Van Asseldonk, L H Van den Berg, S Kalmijn, R M Van den Berg-Vos, C H Polman, J H J Wokke, and H Franssen Axon loss is an important determinant of weakness in multifocal motor neuropathy. J. Neurol. Neurosurg. Psychiatry, June 1, 2006; 77(6): 743 - 747. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S Marconi, M Acler, L Lovato, L De Toni, E Tedeschi, E Anghileri, S Romito, C Cordioli, and B Bonetti Anti-GD2-like IgM autoreactivity in multiple sclerosis patients Multiple Sclerosis, June 1, 2006; 12(3): 302 - 308. [Abstract] [PDF]
|
|
|
|
 |
|
 N. Yuki, K. Susuki, M. Koga, Y. Nishimoto, M. Odaka, K. Hirata, K. Taguchi, T. Miyatake, K. Furukawa, T. Kobata, et al. Carbohydrate mimicry between human ganglioside GM1 and Campylobacter jejuni lipooligosaccharide causes Guillain-Barre syndrome PNAS, August 3, 2004; 101(31): 11404 - 11409. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
Y. Nishimoto, M. Koga, M. Kamijo, K. Hirata, and N. Yuki Immunoglobulin improves a model of acute motor axonal neuropathy by preventing axonal degeneration Neurology, June 8, 2004; 62(11): 1939 - 1944. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. C. Dalakas Intravenous Immunoglobulin in Autoimmune Neuromuscular Diseases JAMA, May 19, 2004; 291(19): 2367 - 2375. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K. Susuki, M. Odaka, M. Mori, K. Hirata, and N. Yuki Acute motor axonal neuropathy after Mycoplasma infection: Evidence of molecular mimicry Neurology, March 23, 2004; 62(6): 949 - 956. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K. Kaida, S. Kusunoki, K. Kamakura, K. Motoyoshi, and I. Kanazawa GalNAc-GD1a in human peripheral nerve: Target sites of anti-ganglioside antibody Neurology, August 26, 2003; 61(4): 465 - 470. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H. J. Willison and N. Yuki Peripheral neuropathies and anti-glycolipid antibodies Brain, December 1, 2002; 125(12): 2591 - 2625. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
Y. Gong, Y. Tagawa, M. P. T. Lunn, W. Laroy, M. Heffer-Lauc, C. Y. Li, J. W. Griffin, R. L. Schnaar, and K. A. Sheikh Localization of major gangliosides in the PNS: implications for immune neuropathies Brain, November 1, 2002; 125(11): 2491 - 2506. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R.D.M. Hadden, H. Karch, H.-P. Hartung, J. Zielasek, B. Weissbrich, J. Schubert, A. Weishaupt, D.R. Cornblath, A.V. Swan, R.A.C. Hughes, et al. Preceding infections, immune factors, and outcome in Guillain-Barre syndrome Neurology, March 27, 2001; 56(6): 758 - 765. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. K. Asbury New Concepts of Guillain-Barre Syndrome J Child Neurol, March 1, 2000; 15(3): 183 - 191. [Abstract] [PDF]
|
|
|
|
|
|
K. A. Sheikh, J. Sun, Y. Liu, H. Kawai, T. O. Crawford, R. L. Proia, J. W. Griffin, and R. L. Schnaar Mice lacking complex gangliosides develop Wallerian degeneration and myelination defects PNAS, June 22, 1999; 96(13): 7532 - 7537. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. Kaji and J. Kimura Facts and fallacies on anti-GM1 antibodies: physiology of motor neuropathies Brain, May 1, 1999; 122(5): 797 - 798. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||