Abstract

Anti-myelin-associated glycoprotein (MAG) neuropathy is an antibody-mediated polyneuropathy. We correlated clinical features, immunoglobulin (Ig) M blood levels, IgM deposition and axonal degeneration in skin biopsies of anti-MAG neuropathy patients. By confocal microscopy, IgM deposits were found exclusively within perineurium-enclosed nerves; they were not found on single, non-perineurium-ensheathed myelinated axons. There was a linear correlation between IgM accumulation in nerve fascicles with IgM blood levels but not with anti-MAG antibody titer or disease duration. Axons with specific IgM deposits had signs of axonal damage, including neurofilament disintegration. Nodal structures were intact even at sites where the axons showed pathologic changes. Ultrastructural analysis revealed degeneration of myelinating Schwann cells. Taken together, these findings suggest that inanti-MAG neuropathy patients, IgM deposits are entrapped within cutaneous perineurium-ensheathed nerve bundles where they accumulate in the endoneurial space. High local IgM levels in the endoneurium may be required for IgM deposition on myelin and subsequent axonal injury and degeneration. This study underlines theimportance of early, effective anti-B-cell treatments for preventing progression of this neuropathy.

Introduction

The anti-myelin-associated glycoprotein (MAG) polyneuropathy is an immune-mediated sensory motor neuropathy characterized by distal demyelination (1, 2); nerve conduction studies typically show a pattern of pronounced distal slowing (3-5). Postulated pathogenetic roles for anti-MAG autoantibodies in this condition are supported by observations that immunoglobulin (Ig) M antibodies are found in myelin sheaths of affected nerves, that passive transfer of antibodies causes disease, and that therapeutic reduction of antibody titer is associated with clinical improvement (6, 7). The morphologic findings in nerve biopsies include IgM deposits at sites of MAG expression, widening of myelin lamellae, axonal changes, loss of MAG, granular deposits in Schwann cell cytoplasm, impairment of remyelination, and abnormal Schwann cell/axon interactions (1, 8-13).

Skin biopsies are currently used to study small-fiber sensory neuropathies, and methods have been developed to evaluate C-fibers in the dermis (14). The value of repeated skin biopsy sampling for the prediction of therapeutic success has been demonstrated (15). Only very recently have immunohistochemical methods been applied to study myelinated nerve endings in glabrous and hairy skin from different body sites (16). We previously showed that IgM deposits were present on myelinated nerve fibers in skin biopsies of anti-MAG polyneuropathy patients (17). Additionally, we found a reduction in the density of small nonmyelinated nerve fibers in the epidermal skin layer (i.e. intraepidermal nerve fibers), suggesting that there is axonal loss in the underlying dermis (17). Clinical and electrophysiologic features also suggest that nerve fiber endings are particularly affected in the anti-MAG polyneuropathy.

In this study, we analyzed skin biopsies of anti-MAG polyneuropathy patients to characterize the IgM-associated abnormalities specific to cutaneous nerves. In particular, we focused on possible mechanisms by which the presence of anti-MAG IgM antibodies might cause axonal degeneration. We determined the IgM deposition patterns in skin nerve fibers and analyzed the integrity of the axonal cytoskeleton and the molecular composition of the node of Ranvier at sites of IgM deposition. We demonstrate that IgM deposits are specific to nerve bundles but may be less restricted to sites of MAG expression in cutaneous nerve fibers, as has been suggested in sural nerves (8). Nodes of Ranvier seem to maintain their molecular architecture despite the presence of anti-MAG IgM antibodies and axonal damage with disintegration and interruption of the neurofilament network. Ultrastructural studies did not show evidence of myelin widening but revealed a generalized degenerative process with axonal degeneration, Schwann cell pathology, and perineurial changes in affected cutaneous nerves.

Material and Methods

Patients

Skin nerve biopsies were performed in 12 patients (10 men and 2 women) who were 50 to 75 years old (mean, 62 years; Table). All were diagnosed with anti-MAG-associated polyneuropathy. At the time of the biopsy, all but 1 patient (patient 1) had a monoclonal gammopathy of unknown significance or a low-grade non-Hodgkin B-cell lymphoma that did not necessitate treatment other than for the neuropathy. Most patients had been treated with or were on rituximab; 1 patient (patient 7) referred from another institution had received additional cyclophosphamide and fludarabine because of severe progressive motor involvement. Four patients (patients 8, 9, 10, and 12) had not been treated. Electrophysiologic studies were performed either at the time of diagnosis or referral on right ulnar nerves; care was taken to keep limb temperature at 34°C. Antibody titers and serum IgM levels are determined at or within 3 months of the biopsy. At the time of the biopsies, serum anti-MAG antibody titers were determined by an ELISA assay (Bühlmann Laboratories, Schönenbuch, Switzerland).

TABLE.

Clinical and Laboratory Features of Anti-MAG Polyneuropathy Patients

All patients were examined at the University Hospital of Basel (Basel, Switzerland). The skin biopsies of 17 patients with a noninflammatory neuropathy were included as controls. Ages of control patients ranged from 41 to 76 years, with a mean age of 59.5 years. All patients gave their written informed consent to participate in the study. This study was approved by the ethics committee of the University Hospital of Basel.

Skin Biopsies

After local anesthesia with 2% lidocaine, all patients underwent 3-mm punch skin biopsies at the proximal thigh and distal leg. A third biopsy was taken at the hand in the first interdigital space. Specimens were immediately fixed in 2% paraformaldehyde-lysine/periodate solution for 24 hours at 4°C, cryoprotected overnight at 4°C, and serially cut with a cryostat microtome into 50-nm sections (14). Sections were either used for immunohistochemistry or stored at −20°C in a cryoprotectant solution following published protocols (18). Tissue specimens for electron microscopic analysis were immediately fixed in ice-cold 2% glutaraldehyde in 0.1 mol/L cacodylate buffer pH 7.4 and routinely processed (19). Ultrathin 70-nm-thick Epon tissue sections were examined using a Morgani 268D transmission electron microscope (FEI; Philips, Amsterdam, The Netherlands).

Immunohistochemistry and Confocal Colocalization Studies

The floating tissue sections were washed and incubated in 15% ethanol overnight at room temperature for lipid extraction. Triple immunofluorescent colocalization studies were performed on multiple sections for each biopsy. Primary antibodies used were polyclonal rabbit anti-pan-neurofilament antibody cocktail (1:100; Biomol International, Plymouth Meeting, PA), monoclonal mouse anti-neurofilament 200 kD (phos. and non-phos. clone N52, 1:100, Sigma, St. Louis, MO), monoclonal rat anti-myelin basic protein (1:100, MAB386; Chemicon, Temecula, CA), monoclonal mouse anti-contactin-associated protein (CASPR; 1:100, 275-κ gift from H. Peles, the Weizmann Institute, Rehovot, Israel), polyclonal rabbit anti-neurofascin (1:100, NFC1 [155/186]; gift from P. Brophy, Edinburgh, Scotland, UK), monoclonal mouse anti-epithelial membrane antigen (1:100; Serotec, Oxford, UK), polyclonal rabbit anti-protein gene product 9.5 (PGP9.5; 1:500; Biogenesis, Poole, UK), monoclonal mouse anti-myelin-associated glycoprotein (MAG; clone D3A2G5, 1:100) (20).

Secondary antibodies (all from Jackson Immunoresearch Laboratories, West Grove, PA) included carbocyanine (Cy) 2/3-conjugated donkey anti-mouse IgG antibody (1:250), Cy2/3-conjugated donkey anti-rabbit IgG antibody (1:250), (Cy2/5-conjugated donkey anti-rat IgG antibody (1:250), and (Cy5-conjugated goat anti-μ-chain-specific human IgM antibody (1:250).

A confocal microscope (Zeiss LSM 510 Meta; Zeiss, Oberkochen, Germany) equipped with an argon laser (488 nm), 2 He/Ne lasers (543 and 633 nm, respectively), and a 63×/1.25 oil immersion objective was used for 3-channel scanning. Carbocyanine 2, Cy3, and Cy5 labels were recorded in green, red, and blue channels, respectively, by extracting the emission ranges of 510 to 531 nm for Cy2, 563 to 617 nm for Cy3, and 660 to 799 nm for Cy5. Settings were kept identical for all patients for intensity comparison in sets stained in parallel.

When applicable, 1.5-μm-thick optical slices were acquired in serial image stacks (25 μm of total thickness; 0.3 × 0.3 × 0.75-μm voxel resolution) were used for 3-dimensional reconstruction (Imaris 4.5; Bitplane AG, Zürich, Switzerland).

For IgM intensity range measurements, we prepared picture series of each nerve fascicle found in the skin biopsy of an individual patient, with increasing detector sensitivity (steps of 30 U, total range from 0 to 900 U), while keeping all other parameters constant. On these pictures, we measured the average fluorescence intensity of the IgM staining within each nerve fascicle. Medium detector sensitivity was evaluated at mean pixel intensity (128 = 1/2 of maximum pixel value 255) for each nerve. Statistical analysis was done using mean ± standard deviation. To compare IgM levels or anti-MAG blood titer with IgM intensity, a linear regression analysis (with correlation coefficient r2 and probability p) was performed.

Paranodal Length Measurements

Contactin-associated protein (CASPR) clusters were analyzed on triple-color images as outlined below. Images were captured with a 63 × 1.25 oil-immersion objective and processed with least squares matching image measurement tools. We measured CASPR cluster length in nerves of healthy controls and anti-MAG neuropathy patients. Both sides of the node were measured independently, and in each group, between 16 and 22 nodes from different patients were analyzed. Cluster length of neurofascin 155 was analyzed in a comparable manner (data not shown). In anti-MAG neuropathy patients, only nodes with heavy IgM deposits were analyzed. Data are represented as mean ± standard deviation. Groups were compared using the Student t-test.

Results

Innervation of Hairy Skin

We used immunohistochemical staining to visualize nerve bundles and identify myelinated nerve endings in skin biopsies. The staining specificity and low background allowed reliable detection of skin nerves and nerve endings. In contrast to glabrous skin, mechanoreceptors such as Meissner corpuscles and Merkel complexes and Ruffini and Pacinian corpuscles in deeper dermis are very rare in hairy skin; therefore, vertically oriented intrapapillary myelinated endings were absent. Nerve bundles in hairy skin biopsies were found throughout the dermis but seemed to follow collagen-rich bands that also contained vessels (Figs. 1A, B). Independent of their size, nerve bundles were surrounded by a perineurial sheath, as indicated by epithelial membrane antigen stain, which is commonly used to visualize this structure in routine diagnostic histopathology (21) (Fig. 1A). To date, skin biopsies have mainly been used in diagnostic studies of small-fiber neuropathies in which the intraepidermal fiber density is counted and correlated with the length of the basal membrane (Fig. 1B) (14). An antibody against PGP9.5 is used routinely for both research and diagnostic purposes in the analysis of cutaneous nerves fibers (14, 16). The expression patterns of PGP9.5 (Figs. 2A, C) and neurofilament (Figs. 2B, C) were, in most cases, very similar. Neurofilaments labeled with a pan-neurofilament antibody were found mainly in myelinated axons, whereas PGP9.5 preferentially stained small, nonmyelinated axons, including free sensory endings of the epidermis (not shown) and fiber innervating glands (Fig. 2D-G). Myelin basic protein-immunostained myelinated fibers colocalized with vessels and hair follicles (Figs. 2F-H) (16). Although the upper layer of the dermis was well innervated, only a few nerve bundles contained myelinated axons. The myelinated axons did not reach the subepidermal plexus and were only found in dermal layers more than 20 μm from the epithelium. Myelinated axons in bundles were equally identifiable using PGP9.5 or neurofilament antibodies. Most of the epidermal nerve bundles, independent of their size, contained a perineurial cell layer (Figs. 1A, 2H). Different nerve bundles ran alongside each other either ensheathed in the perineurium or as free axons (Fig. 2H). There were thin nerve bundles that contained unmyelinated axons with variable numbers of myelinated axons in the dermis of hairy skin (Fig. 2G; arrow), but most of the nerve bundles did not contain myelinated axons.

FIGURE 1.

Innervation of hairy skin. (A) A 3-mm punch biopsy of hairy skin of a control patient. A general overview of hairy skin innervation is shown by the staining for individual fibers using antibodies against neurofilament (green) and for perineurial cells using an antibody against epithelial membrane protein (red). The background is enhanced to visualize the biopsy outline; this accentuates the edge effect of the neurofilament staining in the upper left and lower right. (B) Schematic drawing of the structures within this skin punch biopsy. Myelinated nerves were only found in the dermis, whereas unmyelinated fibers are localized in the epidermis. (C) Schematic drawing of a myelinated nerve with its main structures highlighted as they relate to this study.

FIGURE 1.

Innervation of hairy skin. (A) A 3-mm punch biopsy of hairy skin of a control patient. A general overview of hairy skin innervation is shown by the staining for individual fibers using antibodies against neurofilament (green) and for perineurial cells using an antibody against epithelial membrane protein (red). The background is enhanced to visualize the biopsy outline; this accentuates the edge effect of the neurofilament staining in the upper left and lower right. (B) Schematic drawing of the structures within this skin punch biopsy. Myelinated nerves were only found in the dermis, whereas unmyelinated fibers are localized in the epidermis. (C) Schematic drawing of a myelinated nerve with its main structures highlighted as they relate to this study.

FIGURE 2.

Axon markers of skin nerves in control patients. (A-C) Neurofilament (NF) and protein gene product 9.5 (PGP9.5) staining of a nerve bundle. Protein gene product 9.5 (green, A), NF (red, B), and overlay (C, yellow). (D-G) Glandular innervation triple stained using anti-PGP9.5 (D, green), anti-NF (E, red), and anti-myelin basic protein (F, blue). Double-positive nerve fibers are yellow in (G). Arrow indicates a nerve bundle containing some myelinated fibers. Round structures (red) are nonspecifically stained. (H) A typical skin nerve is ensheathed in a perineurial cell layer (epithelial membrane antigen, green) and contains nonmyelinated (NF, red) and myelinated (MBP, blue) axons. Scale bar = 10 μm.

FIGURE 2.

Axon markers of skin nerves in control patients. (A-C) Neurofilament (NF) and protein gene product 9.5 (PGP9.5) staining of a nerve bundle. Protein gene product 9.5 (green, A), NF (red, B), and overlay (C, yellow). (D-G) Glandular innervation triple stained using anti-PGP9.5 (D, green), anti-NF (E, red), and anti-myelin basic protein (F, blue). Double-positive nerve fibers are yellow in (G). Arrow indicates a nerve bundle containing some myelinated fibers. Round structures (red) are nonspecifically stained. (H) A typical skin nerve is ensheathed in a perineurial cell layer (epithelial membrane antigen, green) and contains nonmyelinated (NF, red) and myelinated (MBP, blue) axons. Scale bar = 10 μm.

IgM Deposition in Skin of Anti-MAG Neuropathy Patients

Immunoglobulin M deposits detected in the skin nerves of anti-MAG neuropathy patients were always restricted to ensheathed nerve bundles. Figure 3A shows typical abundant IgM deposits throughout a perineurium-ensheathed nerve bundle (arrows); the deposits often seemed concentrated around individual myelinated nerve fibers in these bundles (arrowhead). Single myelinated nerve fibers not associated with an ensheathing perineurium and nonmyelinated fibers had no IgM deposits (Fig. 3B). A more detailed analysis of IgM-positive nerve bundles revealed different types of IgM deposits (Figs. 3C-F). The most frequently observed IgM deposits were diffusely distributed throughout the endoneurium without a clear association with either nerves fibers or perineurial sheath (Fig. 3C). In a number of nerve bundles, however, the highest IgM accumulations were along the perineurial cell layer (Fig. 3D; arrows), along individual (myelinated) nerve fibers (Fig. 3E; arrow), or at sites of MAG expression such as Schmidt-Lanterman structures or nodes of Ranvier (Fig. 3F; arrows). Overall, IgM was barely detectable in nonneural tissue components.

FIGURE 3.

Immunoglobulin M (IgM) deposits in cutaneous nerves of anti-myelin-associated glycoprotein (MAG) neuropathy patients. (A) Myelinated fiber (arrowhead) running in parallel with nonmyelinated fibers (red, arrows) in a nerve bundle with IgM deposition. IgM (blue), neurofilament (red), and myelin (green). Sites of triple colocalization (arrowhead) are white. (B) A single branching myelinated fiber of the same patient does not show IgM deposits using the same microscope settings: neurofilament (red); myelin (green); colocalization, yellow (C-F) Examples of different types of IgM deposits (C). Nerve delineated by a perineurium (green) filled with diffuse IgM deposits (blue) and neurofilament (red). (D) Nerve (red) in which IgM deposits (blue) concentrate around the perineurium (arrows). (E) Fiber-specific IgM (blue, arrow) is also found on an individual myelinated nerve fiber in a bundle of fibers (red). (F) IgM deposits are also at nodes shown with an antibody against MAG (green, arrows). Scale bar = 10 μm.

FIGURE 3.

Immunoglobulin M (IgM) deposits in cutaneous nerves of anti-myelin-associated glycoprotein (MAG) neuropathy patients. (A) Myelinated fiber (arrowhead) running in parallel with nonmyelinated fibers (red, arrows) in a nerve bundle with IgM deposition. IgM (blue), neurofilament (red), and myelin (green). Sites of triple colocalization (arrowhead) are white. (B) A single branching myelinated fiber of the same patient does not show IgM deposits using the same microscope settings: neurofilament (red); myelin (green); colocalization, yellow (C-F) Examples of different types of IgM deposits (C). Nerve delineated by a perineurium (green) filled with diffuse IgM deposits (blue) and neurofilament (red). (D) Nerve (red) in which IgM deposits (blue) concentrate around the perineurium (arrows). (E) Fiber-specific IgM (blue, arrow) is also found on an individual myelinated nerve fiber in a bundle of fibers (red). (F) IgM deposits are also at nodes shown with an antibody against MAG (green, arrows). Scale bar = 10 μm.

In quantitative confocal measurements, the average intensities of IgM staining were estimated from the mean pixel intensities as a measure of IgM concentration (Fig. 4). By taking the range at the middle intensity (Fig. 4A; red line), we calculated the interpatient and intrapatient intensity range (Fig. 4B). There was heterogeneity of individual nerve fascicles of 1 patient in the range of 3%, which was smaller than interpatient heterogeneity of 10% variability. Immunoglobulin M blood levels and anti-MAG titers were compared with the IgM deposition values. There was a linear correlation between IgM accumulation in nerve fascicles and IgM blood levels (Fig. 4C; correlation coefficient, r2 = 0.486, p = 0.1), but not with anti-MAG antibody titer.

FIGURE 4.

Quantification of immunoglobulin M (IgM) deposition and correlations with IgM levels. (A) Quantitative analysis of IgM accumulation levels in nerve bundles: Pixel intensities of IgM staining of individual patients are shown as a function of detector gain. High detector gain numbers indicate lower IgM accumulation. Red dashed line indicates middle intensity value of pixels at 128. Fluorescent image panels demonstrate the visibility increase by increasing detector gain value from 510 to 660. (B) Correlations of intensity values with IgM and anti-myelin-associated glycoprotein (MAG) titers (Table). Interpatient and intrapatient variabilities are shown in percent (last column). (C) Regression analysis demonstrates a linear dependence between the average IgM accumulation levels in nerves with IgM levels of patients (•, r2 = 0.486; p = 0.1); specific anti-MAG titer did not correlate (♦, r2 = 0.17)

FIGURE 4.

Quantification of immunoglobulin M (IgM) deposition and correlations with IgM levels. (A) Quantitative analysis of IgM accumulation levels in nerve bundles: Pixel intensities of IgM staining of individual patients are shown as a function of detector gain. High detector gain numbers indicate lower IgM accumulation. Red dashed line indicates middle intensity value of pixels at 128. Fluorescent image panels demonstrate the visibility increase by increasing detector gain value from 510 to 660. (B) Correlations of intensity values with IgM and anti-myelin-associated glycoprotein (MAG) titers (Table). Interpatient and intrapatient variabilities are shown in percent (last column). (C) Regression analysis demonstrates a linear dependence between the average IgM accumulation levels in nerves with IgM levels of patients (•, r2 = 0.486; p = 0.1); specific anti-MAG titer did not correlate (♦, r2 = 0.17)

Analysis of Paranodal Architecture

Histologic and ultrastructural analyses have demonstrated that anti-MAG IgM antibodies are found not only at the surface of the myelin sheath but also throughout the compact myelin layers (12, 22). This has been difficult to explain. The paranodal region could be an entry point for anti-MAG IgM antibodies that allows IgM molecules to penetrate between the myelin lamellae (8). We therefore analyzed the molecular organization of the paranodal region in nerves with heavy IgM deposits and compared it to that in normal paranodal structures in controls (Fig. 5). The paranodal loops of Schwann cells were identified using an antibody against neurofascin 155 (23). The axonal nodal and paranodal regions were identified using 2 different antibodies; one recognizes Neurofascin 186, which is localized within the node, and the other is against CASPR, which is localized in the paranodal region of the axon (Figs. 5A-C) (24). The controls demonstrated normal, compact clustering of neurofascin 155/186 and CASPR on myelinated nerve fibers (Figs. 5A-C; arrows). In anti-MAG neuropathy patients, the clustering of CASPR molecules could also be observed in nerve fibers with massive IgM deposits (Fig. 5E; arrows). In addition, nerve fibers with heavy IgM deposition and normal-appearing nodal structures lacked neurofilament staining in their axon (Figs. 5D, E; arrowheads), whereas neurofilament staining remained intact at the nodes of Ranvier (Fig. 5E, F; arrow). We used 3-dimensional reconstruction to visualize the association between neurofilament and the CASPR clusters (Fig. 5F). To demonstrate the stability of CASPR clusters, we correlated neurofilament staining and IgM depositions to identify sites where the axonal neurofilament network was lost or severely disrupted (Figs. 5D, E, arrowheads; see also Fig. 6). Because of technical limitations, analysis of axonal neurofascin expression in nerves of anti-MAG polyneuropathy patients had to be done independently but showed comparable compact clusters (data not shown).

FIGURE 5.

Paranodal integrity in skin nerves of anti-myelin-associated glycoprotein (MAG) neuropathy patients. (A-C) Paranodal and nodal (arrows) structures of control skin nerves are stained using antibodies against contactin associated protein (CASPR; red), neurofascin 155/186 (blue), and myelin basic protein (green). Colocalization sites of all 3 labels are white. Note the colocalization of neurofascin 155 with CASPR at the paranode (purple), whereas the node itself is only neurofascin 186-positive (blue). (D-F) Paranodal structures with heavy immunoglobulin M (IgM) deposits (arrows) of anti-MAG neuropathy patients identified using antibodies against IgM (blue), neurofilament (green), and CASPR (red). Colocalization of CASPR (red) and IgM (blue) is purple; triple-positive structures are white. Arrowheads highlight IgM deposits where neurofilament staining is lost. (F) Three-dimensional reconstruction of nodal area with heavy IgM deposits CASPR (red), neurofilament (green), IgM (blue). (G) The lengths of 46 CASPR-positive structures were measured in 22 anti-MAG neuropathy and 24 control samples. Data are represented as mean ± standard deviation; p = 0.002. Scale bar = 10 μm.

FIGURE 5.

Paranodal integrity in skin nerves of anti-myelin-associated glycoprotein (MAG) neuropathy patients. (A-C) Paranodal and nodal (arrows) structures of control skin nerves are stained using antibodies against contactin associated protein (CASPR; red), neurofascin 155/186 (blue), and myelin basic protein (green). Colocalization sites of all 3 labels are white. Note the colocalization of neurofascin 155 with CASPR at the paranode (purple), whereas the node itself is only neurofascin 186-positive (blue). (D-F) Paranodal structures with heavy immunoglobulin M (IgM) deposits (arrows) of anti-MAG neuropathy patients identified using antibodies against IgM (blue), neurofilament (green), and CASPR (red). Colocalization of CASPR (red) and IgM (blue) is purple; triple-positive structures are white. Arrowheads highlight IgM deposits where neurofilament staining is lost. (F) Three-dimensional reconstruction of nodal area with heavy IgM deposits CASPR (red), neurofilament (green), IgM (blue). (G) The lengths of 46 CASPR-positive structures were measured in 22 anti-MAG neuropathy and 24 control samples. Data are represented as mean ± standard deviation; p = 0.002. Scale bar = 10 μm.

FIGURE 6.

Immunoglobulin M (IgM)-induced changes in the neurofilament network and myelin in skin nerves of anti-myelin-associated glycoprotein (MAG) neuropathy patients. (A-D) Four examples of disintegrated neurofilament (NF) in anti-MAG patients are shown. Neurofilament is green, whereas IgM is blue. Note the colocalization of IgM deposits with loss of NF staining (arrows). (E, F) Two types of observed myelin abnormalities, that is, focal demyelination (arrow) (E) and myelin disintegration and ovoid formation (F). Note the remains of axons in this structure. For both panels, NF is red, myelin is green, and IgM is blue. Scale bar = 10 μm. (G-N) In control skin nerves, NF staining is homogeneous in nerve fibers. Neurofilament correlates with protein gene product 9.5 (PGP9.5) expression. Neurofilament is red, PGP9.5 is cyan, and myelin basic protein (MBP; green). (O-R) In skin nerves of anti-MAG neuropathy patients, NF is discontinuous in myelinated fibers that have myelin-penetrating IgM deposits. NF is red, MBP is green, and IgM is blue. (S-V) In skin nerves of anti-MAG patients, discontinuous NF correlates with discontinuous or lost (arrows) PGP9.5 in axons with specific IgM deposits. NF is red, PGP9.5 is cyan, MBP is green, and IgM is blue.

FIGURE 6.

Immunoglobulin M (IgM)-induced changes in the neurofilament network and myelin in skin nerves of anti-myelin-associated glycoprotein (MAG) neuropathy patients. (A-D) Four examples of disintegrated neurofilament (NF) in anti-MAG patients are shown. Neurofilament is green, whereas IgM is blue. Note the colocalization of IgM deposits with loss of NF staining (arrows). (E, F) Two types of observed myelin abnormalities, that is, focal demyelination (arrow) (E) and myelin disintegration and ovoid formation (F). Note the remains of axons in this structure. For both panels, NF is red, myelin is green, and IgM is blue. Scale bar = 10 μm. (G-N) In control skin nerves, NF staining is homogeneous in nerve fibers. Neurofilament correlates with protein gene product 9.5 (PGP9.5) expression. Neurofilament is red, PGP9.5 is cyan, and myelin basic protein (MBP; green). (O-R) In skin nerves of anti-MAG neuropathy patients, NF is discontinuous in myelinated fibers that have myelin-penetrating IgM deposits. NF is red, MBP is green, and IgM is blue. (S-V) In skin nerves of anti-MAG patients, discontinuous NF correlates with discontinuous or lost (arrows) PGP9.5 in axons with specific IgM deposits. NF is red, PGP9.5 is cyan, MBP is green, and IgM is blue.

To evaluate paranode lengths, we measured the lengths of individual CASPR cluster paranodes and identified a statistically significant elongation in the length of CASPR clusters in the anti-MAG neuropathy patients compared with the controls (Fig. 5G).

IgM-Induced Axonal Changes in Skin Nerve Fibers

We previously reported the presence of IgM deposits in skin nerves of anti-MAG polyneuropathy patients, but the IgM-induced pathologic alterations were not analyzed systematically in that study (17). In particular, axonal changes induced by IgM antibodies have not been investigated either in skin or in nerve biopsies.

Because the integrity of the neurofilament system in the central nervous system is a sensitive indicator of the axon degeneration (25), we used an antibody cocktail against all neurofilament isoforms for this analysis. In controls, the neurofilament antibodies stained most of the axons of cutaneous nerve bundles homogenously (Figs. 2D-G, 6F). In contrast, in anti-MAG neuropathy patients, neurofilaments of fibers with massive IgM deposits seemed markedly disrupted (Figs. 6A-D; arrows). The axonal marker PGP9.5 also showed a patchy distribution pattern at sites of neurofilament disintegration or was even absent (Fig. 6T; arrows) in affected nerves (Figs. 6S-V). To exclude the possibility that the interrupted pattern of neurofilament staining is due to technical artifacts, we also stained skin of control patients in parallel. In the control samples, neurofilament staining in nerve fibers was always homogenous (Figs. 6G, K), and PGP9.5 was distributed homogeneously (Figs. 6K-N, arrows). In addition to axonal changes, we also detected myelin-specific changes at the light microscopic level. These included focal demyelination (Fig. 6E; arrows) and myelin ovoid formation (Fig. 6F).

Ultrastructural Changes in Cutaneous Nerves of Anti-MAG Neuropathy Patients

Ultrastructural analyses were performed on skin biopsy from 2 controls and 3 anti-MAG polyneuropathy patients (Fig. 7). Although myelin widening is a major feature of nerve biopsies in anti-MAG neuropathy patients (12, 26), myelin widening was not observed in any of 18 myelinated skin nerves we analyzed. In contrast, features of axonal degeneration with shrunken axon remnants (Fig. 7A; arrowhead), myelin collapse, vacuolization, disintegration, and fragmentation were common (Figs. 7A, B). There were also enlarged periaxonal spaces and some inappropriately thickened myelin sheaths (not shown). The cytoplasm of Schwann cells was often enlarged and filled with mitochondria (Fig. 7A; arrow); it also contained glycogen aggregates (Fig. 7B; arrowhead) and Reich bodies (Fig. 7B; arrow) and was often surrounded by ruffled membranes (Fig. 7A). Schwann cell clusters contained small axons at their periphery (Fig. 7C; arrowhead). These axons were not completely enwrapped by Schwann cell processes and abutted on endoneurial collagen (Fig. 7C; arrowhead). We did not find evidence for remyelination in the form of hypomyelinated axons, but there were bands of Büngner (Fig. 7C; arrows) that are formed by nonmyelinating Schwann cells. We also found hypertrophic formations that are formed by concentrically organized hyperactive Schwann cells that are encircled by a basal membrane (Fig. 7D; arrow); this basal membrane also surrounds electron lucent Schwann cells that contain unmyelinated axons (Fig. 7D; arrowhead). Overall, there seems to be a loss of myelinated fibers and an increase in endoneurial connective tissue (Fig. 7D), but no macrophages or histiocytes associated with nerve fibers were observed. This is consistent with the very chronic and indolent nature of this neuropathy.

FIGURE 7.

Ultrastructural analysis of dermal nerves of anti-myelin-associated glycoprotein (MAG) neuropathy patients. (A) Axonal degeneration (arrowhead) with collapse and fragmentation myelin. Schwann cell cytoplasm contains numerous mitochondria (arrow) and is surrounded by a ruffled cell membrane. (B) A degenerating myelinating Schwann cell contains myelin debris, large Reich granule (arrow), and glycogen aggregates (arrowhead). (C) Partial view of a nerve bundle with bands of Büngner characterized by Schwann cell hypertrophy within abundant leftover basal membranes (arrows). Axons are not completely enwrapped in Schwann cell processes (arrowhead). (D) Concentrically organized, probably hyperactive Schwann cells. Cells are ensheathed by a basement membrane and have focal pinocytotic vesicles (arrow). In the center is a Schwann cell cluster with lucent cytoplasm and some axon sprouts with one mesaxon (arrowhead).

FIGURE 7.

Ultrastructural analysis of dermal nerves of anti-myelin-associated glycoprotein (MAG) neuropathy patients. (A) Axonal degeneration (arrowhead) with collapse and fragmentation myelin. Schwann cell cytoplasm contains numerous mitochondria (arrow) and is surrounded by a ruffled cell membrane. (B) A degenerating myelinating Schwann cell contains myelin debris, large Reich granule (arrow), and glycogen aggregates (arrowhead). (C) Partial view of a nerve bundle with bands of Büngner characterized by Schwann cell hypertrophy within abundant leftover basal membranes (arrows). Axons are not completely enwrapped in Schwann cell processes (arrowhead). (D) Concentrically organized, probably hyperactive Schwann cells. Cells are ensheathed by a basement membrane and have focal pinocytotic vesicles (arrow). In the center is a Schwann cell cluster with lucent cytoplasm and some axon sprouts with one mesaxon (arrowhead).

Discussion

Anti-MAG polyneuropathy is predominantly characterized by sensory disturbances; neurophysiologic examination typically shows a marked delay of distal latencies, suggesting that nerve fiber endings are particularly affected (27, 28). We recently demonstrated IgM deposits and axonal loss in skin nerves of anti-MAG polyneuropathy patients (17), but the mechanisms of demyelination and axonal degeneration in this disorder are poorly understood. To date, IgM-induced changes have only been described in nerve biopsies (12, 29). In the present study, we found IgM accumulations only in nerve bundles and not within individual myelinated fibers in the skin of affected patients. Most fiber nerve bundles in the skin are ensheathed by a perineurial cell layer, and this tight barrier of epithelial cells may entrap IgM antibodies in the endoneurial space. Because anti-MAG IgM autoantibodies have low specific binding capacities and probably bind to their target structures only when they reach a critical local concentration, local entrapment and accumulation of IgM antibodies may be critical for the development of the neuropathy. Specific binding sites in nerve bundles include not only myelinated fibers but also the basement membrane of Schwann cells, nodes of Ranvier and, interestingly, the perineurial cells. Microheterogeneity and cross reactivity of anti-MAG antibody with HNK-1 and sulfate3-gluconuryl-paraglobuside-like glycoconjugates have been reported (30, 31), and these may be responsible for this broad recognition pattern. Individual nerve fascicles in sural nerve biopsies are affected differently by IgM deposits (22), and we also found in skin nerves considerable heterogeneity of IgM deposition among nerves in individual patient samples. This intrapatient heterogeneity was independent of serum IgM levels and/or duration of disease (Fig. 4B). Interestingly, there was a linear correlation of serum IgM levels with IgM deposition in nerve fascicles of the skin. Thus, local accumulation of IgM may increase binding of IgM to primary and secondary targets, leading to a degenerative response, including, but not limited to, the degeneration of myelinated axons.

Skin axons were identified using PGP9.5 (32), an axon-specific marker that has been widely used for diagnostic staining of C-fiber densities in the epidermis (14). We observed the neurofilament network was disrupted at sites of heavy IgM deposits, whereas IgM-negative axons showed homogeneous neurofilament staining. Disruption of the neurofilament network colocalized with the heaviest IgM deposits, indicating that the destructive degenerative process may at least partly be dependent on antibody-specific binding to myelin. Furthermore, PGP9.5 was also disrupted in affected axons. Axons with disrupted neurofilament and PGP9.5 may represent a late, rather than an initial, stage of degeneration.

It has been suggested that the binding of anti-MAG autoantibodies may interfere with normal axon-Schwann cell signaling by penetrating into the adaxonal space via nodes of Ranvier and directly interfering with MAG signaling (8, 22). Nodes of Ranvier are known for their high levels of paranodal MAG expression and therefore may be major sites of anti-MAG antibody binding (8, 22). Interestingly, however, even at sites of heavy IgM deposition, paranodal architecture, as visualized by CASPR (i.e. representing the axonal site), most often retained their normal clustered appearance. Measurements of paranodal lengths (i.e. lengths of CASPR clusters), however, revealed subtle changes, including elongation of this cluster. This observation may indicate that there are axonal alterations at the node of Ranvier (33) and further supports the concept that the anti-MAG IgM antibodies induce alterations in axons and Schwann cells that are less dependent on structural loosening of Schwann cell loops and penetration of anti-MAG IgM into the periaxonal space than was previously assumed. At least in skin nerves, there seem to be a number of ways by which anti-MAG antibodies may interfere with the normal functions of myelinating Schwann cells (i.e. by disturbing their homeostasis and interfering with the axonal support). The finding that myelin widening is absent in skin nerves may indicate that the damage induced by anti-MAG IgM at nerve endings is not primarily caused by structural myelin damage. Moreover, myelin widening is observed only in 0% to 36% of nerve biopsies in anti-MAG neuropathy patients (10). Therefore, this very low occurrence of myelin widening argues against an IgM-induced pathology caused by direct penetration into the myelin sheath. In addition, widening seems to be associated with diseases without direct antibody involvement such as Guillain-Barré Syndrome and chronic inflammatory demyelinating polyneuropathy (34). High concentrations of IgM throughout the endoneurial space of skin nerves rather suggest that nerve damage also occurs via Schwann cell axon signaling and by physical factors involving pressure changes, metabolic imbalance, hypoxic changes, and edema.

Electron microscopic analysis demonstrated degenerating axons with disintegrating myelin. Axonal degeneration was also corroborated by the presence of denervated Schwann cell bands (Büngner bands). We did not observe hypomyelinated axons as evidence of remyelination. Nonmyelinating Schwann cells and axons have also been found to be sensitive in classical chronic inflammatory demyelinating polyneuropathy in which the primary response is directed against myelin or myelinating Schwann cells (35). It is therefore also likely that nonmyelinating Schwann cells participate in a reactive response to the presence of IgM in anti-MAG polyneuropathy. Because we observed Schwann cell hypertrophy even in patients with long disease courses, there may still be a potential for regeneration. In addition, hypertrophic changes of supporting cells such as perineurial cells and/or fibroblast-like cells could possibly result from antibody cross reactivity with sugar epitopes of the HNK-1 and sulfate3-gluconuryl-paraglobuside family, as well as pressure imbalance or edema.

In summary, we used skin biopsies of anti-MAG neuropathy patients to analyze the IgM-induced pathologic changes in nerves. We demonstrate that anti-MAG IgM antibodies specifically accumulate inside the nerve perineurium, where they are associated with a degenerative response of axons as indicated by the disintegration of cytoskeletal elements. The integrity of paranodal CASPR clusters indicates that axonal degeneration is probably not only a consequence of IgM penetration through the paranode and structural myelin damage but may also result from inappropriate signaling cascades and impaired homeostasis, thereby interfering with axonal maintenance. In addition, there is an extensive response of nonmyelinating Schwann cells, indicating a more generalized reactive response to the accumulation of IgM. The correlation of nerve IgM accumulation with IgM blood levels argues strongly for early therapeutic reduction in anti-MAG antibody titers by antibody-lowering treatments; this is supported by the success of this treatment in clinical studies (36, 37). Further studies are needed, however, to obtain a more quantitative and dynamic view of axonal changes in this neuropathy.

Acknowledgments

We thank Prof. K. Toyka (University Hospital Würzburg, Germany) for referral of patient, Dr. T. Hundsberger (Hospital St. Gallen, Switzerland) for referral of patient, and Prof. T. Kuntzer (CHUV, Lausanne, Switzerland) for referring patient.

References

1.
Smith
IS
Kahn
SN
Lacey
BW
.
Chronic demyelinating neuropathy associated with benign IgM paraproteinaemia
.
Brain
 
1983
;
106
:
169
95
2.
Nobile-Orazio
E
.
IgM paraproteinaemic neuropathies
.
Curr Opin Neurol
 
2004
;
17
:
599
605
3.
Radziwill
AJ
Steck
AJ
Renaud
S
.
Distal motor latency and residual latency as sensitive markers of anti-MAG polyneuropathy
.
J Neurol
 
2003
;
250
:
962
66
4.
Franssen
H
Notermans
NC
.
Length dependence in polyneuropathy associated with IgM gammopathy
.
Ann Neurol
 
2006
;
59
:
365
71
5.
Lupu
VD
Mora
CA
Dambrosia
J
.
Terminal latency index in neuropathy with antibodies against myelin-associated glycoproteins
.
Muscle Nerve
 
2007
;
35
:
196
202
6.
Steck
AJ
Murray
N
Justafre
JC
.
Passive transfer studies in demyelinating neuropathy with IgM monoclonal antibodies to myelin-associated glycoprotein
.
J Neurol Neurosurg Psychiatry
 
1985
;
48
:
927
9
7.
Renaud
S
Fuhr
P
Gregor
M
.
High-dose rituximab and anti-MAG-associated polyneuropathy
.
Neurology
 
2006
;
66
:
742
44
8.
Gabriel
JM
Erne
B
Bernasconi
L
.
Confocal microscopic localization of anti-myelin-associated glycoprotein autoantibodies in a patient with peripheral neuropathy initially lacking a detectable IgM gammopathy
.
Acta Neuropathol (Berl)
 
1998
;
95
:
540
46
9.
Gabriel
JM
Erne
B
Miescher
GC
.
Selective loss of myelin-associated glycoprotein from myelin correlates with anti-MAG antibody titre in demyelinating paraproteinaemic polyneuropathy
.
Brain
 
1996
;
119
:
775
87
10.
Mendell
JR
Sahenk
Z
Whitaker
JN
.
Polyneuropathy and IgM monoclonal gammopathy: Studies on the pathogenetic role of anti-myelin-associated glycoprotein antibody
.
Ann Neurol
 
1985
;
17
:
243
54
11.
Ellie
E
Vital
A
Steck
A
.
Neuropathy associated with “benign” anti-myelin-associated glycoprotein IgM gammopathy: Clinical, immunological, neurophysiological pathological findings and response to treatment in 33 cases
.
J Neurol
 
1996
;
243
:
34
43
12.
Jacobs
JM
Scadding
JW
.
Morphological changes in IgM paraproteinaemic neuropathy
.
Acta Neuropathol (Berl)
 
1990
;
80
:
77
84
13.
Steck
AJ
Stalder
AK
Renaud
S
.
Anti-myelin-associated glycoprotein neuropathy
.
Curr Opin Neurol
 
2006
;
19
:
458
63
14.
Lauria
G
Cornblath
DR
Johansson
O
.
EFNS guidelines on the use of skin biopsy in the diagnosis of peripheral neuropathy
.
Eur J Neurol
 
2005
;
12
:
747
58
15.
Ebenezer
GJ
McArthur
JC
Thomas
D
.
Denervation of skin in neuropathies: The sequence of axonal and Schwann cell changes in skin biopsies
.
Brain
 
2007
;
130
:
2703
14
16.
Provitera
V
Nolano
M
Pagano
A
.
Myelinated nerve endings in human skin
.
Muscle Nerve
 
2007
;
35
:
767
75
17.
Lombardi
R
Erne
B
Lauria
G
.
IgM deposits on skin nerves in anti-myelin-associated glycoprotein neuropathy
.
Ann Neurol
 
2005
;
57
:
180
87
18.
McCarthy
BG
Hsieh
ST
Stocks
A
.
Cutaneous innervation in sensory neuropathies: Evaluation by skin biopsy
.
Neurology
 
1995
;
45
:
1848
55
19.
Erb
M
Flueck
B
Kern
F
.
Unraveling the differential expression of the two isoforms of myelin-associated glycoprotein in a mouse expressing GFP-tagged S-MAG specifically regulated and targeted into the different myelin compartments
.
Mol Cell Neurosci
 
2006
;
31
:
613
27
20.
Burger
D
Simon
M
Perruisseau
G
.
The epitope(s) recognized by HNK-1 antibody and IgM paraprotein in neuropathy is present on several N-linked oligosaccharide structures on human P0 and myelin-associated glycoprotein
.
J Neurochem
 
1990
;
54
:
1569
75
21.
Perentes
E
Nakagawa
Y
Ross
GW
.
Expression of epithelial membrane antigen in perineurial cells and their derivatives
.
An immunohistochemical study with multiple markers. Acta Neuropathol (Berl)
 
1987
;
75
:
160
65
22.
Ritz
MF
Erne
B
Ferracin
F
.
Anti-MAG IgM penetration into myelinated fibers correlates with the extent of myelin widening
.
Muscle Nerve
 
1999
;
22
:
1030
37
23.
Charles
P
Tait
S
Faivre-Sarrailh
C
.
Neurofascin is a glial receptor for the paranodin/Caspr-contactin axonal complex at the axoglial junction
.
Curr Biol
 
2002
;
12
:
217
20
24.
Poliak
S
Peles
E
.
The local differentiation of myelinated axons at nodes of Ranvier
.
Nat Rev Neurosci
 
2003
;
4
:
968
80
25.
Petzold
A
.
Neurofilament phosphoforms: Surrogate markers for axonal injury, degeneration and loss
.
J Neurol Sci
 
2005
;
233
:
183
98
26.
Vital
C
Vital
A
Deminiere
C
.
Myelin modifications in 8 cases of peripheral neuropathy with Waldenstrom's macroglobulinemia and anti-MAG activity
.
Ultrastruct Pathol
 
1997
;
21
:
509
16
27.
Kaku
DA
England
JD
Sumner
AJ
.
Distal accentuation of conduction slowing in polyneuropathy associated with antibodies to myelin-associated glycoprotein and sulphated glucuronyl paragloboside
.
Brain
 
1994
;
117
:
941
47
28.
Chassande
B
Leger
JM
Younes-Chennoufi
AB
.
Peripheral neuropathy associated with IgM monoclonal gammopathy: Correlations between M-protein antibody activity and clinical/electrophysiological features in 40 cases
.
Muscle Nerve
 
1998
;
21
:
55
62
29.
Mata
M
Kahn
SN
Fink
DJ
.
A direct electron microscopic immunocytochemical study of IgM paraproteinemic neuropathy
.
Arch Neurol
 
1988
;
45
:
693
97
30.
Fluri
F
Ferracin
F
Erne
B
.
Microheterogeneity of anti-myelin-associated glycoprotein antibodies
.
J Neurol Sci
 
2003
;
207
:
43
49
31.
Ilyas
AA
Gu
Y
Dalakas
MC
.
Induction of experimental ataxic sensory neuronopathy in cats by immunization with purified SGPG
.
J Neuroimmunol
 
2008
;
193
:
87
93
32.
Day
IN
.
Enolases and PGP9.5 as tissue-specific markers
.
Biochem Soc Trans
 
1992
;
20
:
637
42
33.
Howell
OW
Palser
A
Polito
A
.
Disruption of neurofascin localization reveals early changes preceding demyelination and remyelination in multiple sclerosis
.
Brain
 
2006
;
129
:
3173
85
34.
Ohnishi
A
Hirano
A
.
Uncompacted myelin lamellae in dysglobulinemic neuropathy
.
J Neurol Sci
 
1981
;
51
:
131
40
35.
Bosboom
WM
Van den Berg
LH
Dieks
HJ
.
Unmyelinated nerve fiber degeneration in chronic inflammatory demyelinating polyneuropathy
.
Acta Neuropathologica
 
2000
;
99
:
571
78
36.
Levine
TD
Pestronk
A
.
IgM antibody-related polyneuropathies: B-Cell depletion chemotherapy using Rituximab
.
Neurology
 
1999
;
52
:
1701
4
37.
Renaud
S
Gregor
M
Fuhr
P
.
Rituximab in the treatment of polyneuropathy associated with anti-MAG antibodies
.
Muscle Nerve
 
2003
;
27
:
611
15

Author notes

This work was supported by a grant from Habegger Ltd. Nicole Schaeren-Wiemers was supported by Grant No. 3100A0-112583 from the Swiss National Science Foundation.