Skin biopsies in myelin-related neuropathies: bringing molecular pathology to the bedside

Transcription

1 doi: /brain/awh483 Brain (2005), 128, Skin biopsies in myelin-related neuropathies: bringing molecular pathology to the bedside Jun Li, 1 Yunhong Bai, 1 Khaled Ghandour, 1 Pu Qin, 1 Marina Grandis, 1 Anna Trostinskaia, 1 Emilia Ianakova, 1 Xingyao Wu, 1 Angelo Schenone, 4 Jean-Michel Vallat, 5 William J. Kupsky, 2 James Hatfield 3 and Michael E. Shy 1 1 Department of Neurology and 2 Department of Pathology, Wayne State University School of Medicine, 3 Department of Pathology, John D. Dingell VA Hospital, Detroit, MI, USA, 4 Department of Neurological Sciences and Rehabilitation, University of Genova, Genova, Italy and 5 Service de Neurologie, CHU Dupuytren, Limoges, France Summary Skin biopsy is a minimally invasive procedure and has been used in the evaluation of non-myelinated, but not myelinated nerve fibres, in sensory neuropathies. We therefore evaluated myelinated nerves in skin biopsies from normal controls and patients with Charcot Marie Tooth (CMT) disease caused by mutations in myelin proteins. Light microscopy, electron microscopy and immunohistochemistry routinely identified myelinated dermal nerves in glabrous skin that appeared similar to myelinated fibres in sural and sciatic nerve. Myelin abnormalities were observed in all patients with CMT. Moreover, skin biopsies detected potential pathogenic Correspondence to: Dr Jun Li, Department of Neurology, Wayne State University, 4201 St Antoine, UHC-8D, Detroit, MI 48201, USA junli@med.wayne.edu abnormalities in the axolemmal molecular architecture previously undetected in human neuropathies. Finally, myelin gene expression at both mrna and protein levels was evaluated by real-time PCR and immunoelectron microscopy. Peripheral myelin protein 22 (PMP22) was increased in CMT1A (PMP22 duplication) and decreased in patients with hereditary neuropathy with liability to pressure palsies (PMP22 deletion). Taken together, our data suggest that skin biopsy may in certain circumstances replace the more invasive sural nerve biopsy in the morphological and molecular evaluation of inherited and other demyelinating neuropathies. Keywords: skin biopsy; myelinated nerve; Schwann cell; molecular architecture; real time PCR; inherited neuropathy Abbreviations: CMT = Charcot Marie Tooth disease; EM = electron microscope/microscopy; HNPP = hereditary neuropathy with liability to pressure palsies; MAG = myelin-associated glycoprotein precursor; MBP = myelin basic protein; MPZ = myelin protein zero; PLP1 = proteolipid protein 1; PMP22 = peripheral myelin protein 22 Received October 10, Revised January 19, Accepted February 21, Advance Access publication March 17, 2005 Introduction Skin biopsy has recently been used to study small-fibre sensory neuropathy (Kennedy and Wendelschafer-Crabb, 1993; Holland et al., 1997). This technique has successfully revealed decreased numbers of epidermal unmyelinated nerve fibres in patients with diabetes (Kennedy et al., 1996; Polydefkis et al., 2001), HIV (Polydefkis et al., 2002) and idiopathic small-fibre sensory neuropathies (Holland et al., 1998). Skin biopsy is a minimally invasive procedure in which only 3 5 mm punch biopsies are taken from selected sites of the patient s skin. Unlike invasive sural nerve biopsies, skin biopsies can be repeated in longitudinal studies (Polydefkis et al., 2002) and can be performed at various sites along the body, permitting evaluation of both proximal and distal sensory nerves. At present, skin biopsies have not been used to investigate myelinated nerves, in part because hairy skin, where the procedure is usually performed, possesses few myelinated nerve fibres. However, Nolano and colleagues have recently demonstrated that myelinated fibres can be labelled by immunohistochemistry in glabrous skin (Nolano et al., 2003). Moreover, de Groote and colleagues used electron microscopy (EM) from a skin biopsy to demonstrate hypomyelination and onion bulb formation, also seen in sural nerve biopsy, in a patient with Charcot Marie Tooth disease (CMT) (Ceuterick-de Groote et al., 2001). These results suggest that skin biopsies might be useful to evaluate myelinated nerves in many patients with demyelinating or dysmyelinating # The Author (2005). Published by Oxford University Press on behalf of the Guarantors of Brain. All rights reserved. For Permissions, please journals.permissions@oupjournals.org

2 neuropathies. To test this hypothesis, we chose to evaluate skin from a series of patients with demyelinating forms of CMT in which genetic abnormalities and phenotypic presentations are clearly defined. Thus, the consequence of genetic alterations can be effectively correlated with nerve morphological and molecular alterations. CMT refers to inherited peripheral neuropathies. At present, mutations in over 30 different genes causing CMT have been identified (Belgian CMT database: The identification of precise genetic causes of many forms of CMT has revolutionized the diagnosis of these disorders and has also clarified several confusing issues about CMT phenotypic variation (Li et al., 2003). Although exciting and clinically important, this wealth of new information has not yet led to a treatment for CMT, in part because little is known about the pathogenic mechanisms by which the various mutations cause neuropathy. This lack of knowledge is largely due to the difficulties of obtaining human nerve material to study. Since sural nerve biopsies are not usually necessary for diagnosis, they are not therefore performed on most patients with CMT. Thus, there is a need to develop a better method to evaluate nerves morphologically in patients with CMT. Accordingly, we have performed skin biopsies on a group of normal human controls without neuropathy as well as a series of patients with the most common, demyelinating forms of CMT: CMT1A, CMTX1 and hereditary neuropathy with liability to pressure palsies (HNPP). We found that dermal myelinated nerve fibres could be routinely identified in skin biopsies from glabrous skin of normal fingers and forearms. Morphological features and the molecular organization of the myelinated nerve fibres appeared similar to that reported in human sural nerve and in rodent sciatic nerve, except that myelin internodes were shorter. Dermal nerves from CMT1A and CMTX patients had evidence of thin myelin, axonal loss, and adaxonal myelin decompaction. Immunohistochemical analysis revealed alterations of the axolemmal molecular architecture. Immuno-EM studies demonstrated the predicted increase in protein levels of peripheral myelin protein 22 (PMP22) in CMT1A nerves and decreased PMP22 levels in HNPP nerves. These results demonstrate that dermal myelinated nerves not only show abnormalities previously detected in sural nerve biopsies but also detect abnormal features not previously reported in sural nerve morphological studies of CMT that may provide important pathogenic clues in these disorders. Methods Subjects Eight normal controls (two women and six men) and nine patients (four women and five men) with CMT1A (n = 5), HNPP (n = 3) and CMTX (n = 1) were studied. Ages ranged from 30 to 57 years in control subjects and from 14 to 83 years in patients. Appropriate consent was obtained from participants and the study was approved by the Wayne State University Human Investigation Committee. Skin biopsies in myelin-related neuropathies 1169 Skin biopsy We biopsied the lateral aspects of the fingers and ventral aspects of the forearm (Fig. 1). Skin surface was sterilized with ethanol swabs and anaesthetized with 2% lidocaine. The finger skin biopsies were performed between the first and second interphalangeal joints or between the second interphalangeal and palmophalangeal joints. To obtain Meissner s receptors and their innervating myelinated nerves, biopsies were done 2 3 mm from the border of the hairy skin of the finger. We used small skin punches 2 mm in diameter and with tip length 8 mm (Miltex Instrument Co.) (Fig. 1). The skin punch was pushed perpendicularly to the surface of skin, rotated in a clockwise direction, and advanced until reaching 5 mm depth (a little over half-way to the punch tip). Finger skin is rich in small vessels, so the biopsy wound required at least 15 min of compression to stop bleeding. A bandage was applied over the wound. Patients were instructed to change a fresh bandage daily until the wound had healed. Digit arteries and nerves are located deeply, close to the phalangeal bone. The biopsy tip was still far from these structures even after advancing 5 mm. Thus, it has been a safe procedure and we have encountered no significant complications. For the forearm skin biopsy, an identical technique was used. The biopsy was taken 7 8 cm proximal to the wrist crease. Bleeding was much less and only 2 min of compression was required to stop bleeding in most cases. Light microscopy and electron microscopy As soon as the dermal punch had been removed from the skin, it was washed with phosphate-buffered saline (PBS) to remove the blood and immediately transferred into 2.5% glutaraldehyde cacodylate for 24 h of fixation. Tissues were cut into four quadrants, osmicated in 1% osmium solution for 1.5 h, dehydrated, and embedded in Epon. Tissue blocks were sectioned vertically with 1 mm thickness and stained with toluidine blue or methylene blue. Nerve bundles with myelinated fibres were identified under the light microscope. The blocks were then trimmed and cut into ultrathin sections (about 90 nm). These ultrathin sections were contrasted with lead citrate and uranyl acetate and examined under an EM (Zeiss EM 900). Immunohistochemistry A separate skin punch from an adjacent area was taken, washed briefly with PBS buffer, and transferred into 4% paraformaldehyde for 15 min of fixation. Tissues were then embedded in OCT medium in a plastic mould and gradually frozen in a methylbutane container that was placed in a dry-ice bucket. Once the tissues were hardened (usually within 3 5 min), they were transferred into a cryostat, stabilized for about 20 min, cut into 20 mm sections, and mounted onto Superfrost Plus glass slides (Fisher Scientific). Slides were dried overnight at room temperature and stained next morning or stored in a freezer at 20 C for future use. These slides were then permeabilized with 100% acetone for 10 min, blocked with 5% fish gelatin, incubated with primary antibodies overnight, reacted with secondary antibodies for 2 h next day, washed, dried, and coverslipped for fluorescence microscope examination. The following primary antibodies were used in this study: rat monoclonal antibodies to myelin basic protein (MBP; Chemicon; 1 : 500); mouse monoclonal antibodies to myelin protein zero (MPZ) (Archelos et al., 1993); rabbit monoclonal antibodies to PMP22 (Labvision, 1 : 200); rabbit polyclonal antibodies to protein gene product 9.5 (PGP9.5) (Biogenesis; 1 : 1000); mouse monoclonal

3 1170 J. Li et al. Fig. 1 Dermal myelinated nerves are identifiable in the glabrous skin. Glabrous skin from the lateral aspect of the finger was taken from a control subject and stained with toluidine blue. The dermal myelinated nerve fibres within a small nerve bundle are shown (A). These small nerve bundles run parallel to the surface of skin. In B, a longitudinal section of myelinated nerve fibre is shown (arrowheads). These nerve fibres can also be identified with immunohistochemistry by using polyclonal antibodies to neuron-specific ubiquitin hydrolase (PGP9.5), a pan-axonal marker (black arrowheads in C and D). These nerves innervate dermal mechanical receptors. Meissner receptors are indicated by white arrowheads in C and D. Notice that these myelinated nerve fibres run in the vertical direction and often in a one-to-one relationship with individual Meissner receptors. Thus, they behave as naturally occurring teased nerve fibres. This is particularly suitable for studying the molecular architecture of myelinated nerve fibres. and rabbit polyclonal antibodies to contactin-associated protein (Caspr) (Peles et al., 1997); mouse monoclonal anti-kv1.2 (Upstate Biotechnology; 1 : 100); mouse monoclonal antibodies to voltagegated sodium channels (Sigma; 1 : 500). Immunoelectron microscopy This procedure was similar to the EM mentioned above except for the following. Tissues were fixed in 0.1 or 0.5% glutaraldehyde and 4% paraformaldehyde fixative. They were embedded in LR white and sectioned. The sections were incubated with primary antibodies and gold particle-conjugated secondary antibodies prior to being contrasted and imaged. The images were imported into ImagePro Plus software. Myelin regions were divided into small squares by the software. Gold particles were counted in any square where myelin was intact. The density of myelin proteins was derived as summated particles divided by summated areas of those squares. Real-time PCR The fresh skin was ground in liquid nitrogen. RNA was extracted in TriZol. Three micrograms of RNA treated with RNase-free DNase was subjected to the reverse transcription reaction with the cdna Cycle Kit (Invitrogen). Two-hundred nanograms of the cdna was used for each real-time PCR in SYBR Green PCR Master Mixture (Applied Biosystems, Warrington, UK). Total reaction volume was 25 ml. The reaction took place in a DNA Engine Opticon CFD-0200 (MJ Research, Waltham, MA, USA). Comparison was based on normalization with the reading of the 18S expression level. Reaction temperatures were 52 C for 2 min, 94 C for 10 min, 94 C for 15 s and 62 C for 1 min (40 cycles in total). The following primers were used: myelin-associated glycoprotein (MAG), upper stream (US) 5 0 -CAA CAg TgA ACg gga caa Tg-3 0 DS 5 0 -gct CgC TCT CgT AgA TgA CC-3 0 ; MBP, US 5 0 -ATC CAA gta CCT ggc CAC Ag-3 0 DS 5 0 -TgT gag TCC TTg CCA gag C-3 0 ; myelin proteolipid protein (PLP1) US 5 0 -ggc AgA TCT TTg gcg ACT AC-3 0 down stream (DS) 5 0 -CCT AgC CAT TTT CCC AAA CA-3 0 ; PMP22, US 5 0 -CTg gtc TgT gcg TgA TgA gt-3 0 DS 5 0 -AgA TgA CAC CgC TgA gaa gg-3 0 ; MPZ, US CAC TAT gcc AAg gga CAA CC-3 0 DS CAA gtg AAC gtg CCA TTg TC-3 0 ; Connexin-32 (CX32), US 5 0 -TCC CTg CAg CTC ATC CTA gt-3 0 DS 5 0 -gat gtg gac CTT gtg CCT CT 3 0. Results Dermal myelinated nerves identifiable in the glabrous skin To determine whether we could easily identify myelinated nerves from skin biopsies, we first performed biopsies from multiple sites in lower and upper extremities in normal human controls, without neuropathy. In most sites, myelinated fibres could be recognized in the deep dermal layer. However, these myelinated fibres were much less frequent and difficult to detect in hairy skin (data not shown). In glabrous skin of the upper extremity we could reproducibly identify small nerve bundles containing myelinated fibres (Fig. 1A) and

4 individual myelinated fibres innervating Meissner s receptors (Fig. 1C, D), as has been described in previous reports (Nolano et al., 2003). These Meissner s receptors and the nerves innervating them were readily revealed with antibodies to the pan-axonal marker PGP9.5 (Fig. 1C, D). Myelin ensheathing the fibres was visualized in semithin sections (Fig. 1B). The small nerve bundles containing myelinated fibres were usually located in the deep portion of the dermis and travelled in parallel to the surface of the skin. These bundles were readily detected in skin from both the forearm and the fingers. The number of small bundles was variable from one biopsy to another; however, the number of myelinated nerve fibres within the bundles appeared similar among biopsies taken from the same site (forearm versus finger) as long as the bundle size was roughly comparable. We also evaluated the dermal nerves by EM, which demonstrated myelin compaction with distances of nm between wraps, typical of what is reported for sural nerve (Dyck et al., 1993) (Fig. 2). These data suggest that dermal myelinated nerves are detectable and possess morphological features similar to those of other peripheral nerves. However, unlike sural nerve fibres, which are of large and small diameter, dermal myelinated fibres were of small calibre (usually less than 7 mm in diameter). Molecular architecture of dermal myelinated nerves is similar to that of mouse sciatic nerve Although undulations of the myelinated nerves made it difficult to observe the entire length of most internodes in one plane of the biopsy, internodes could be readily visualized with immunohistochemistry in most thick sections (20 30 mm). Thus, interpretation of the molecular architecture of the node, paranode, internodes and underlying axolemma was possible. We performed immunohistochemistry to confirm the expression of MPZ and MBP in myelin internodes of dermal nerves (Fig. 3). We then evaluated the molecular architecture of nodes, paranodes and juxtaparanodes in these myelinated fibres. Discrete bands of voltagegated sodium channels were identifiable at nodes of Ranvier (Fig. 3B). Paranodes were labelled by Caspr staining (Fig. 3A). Juxtaparanodes were identifiable by staining with voltage-gated Kv1.2 potassium channels (Fig. 3C). Their spiralling distribution was also clearly shown on the axolemma opposing the border of inner mesaxon of the internode. These findings were consistent with the results in rodent sural or sciatic nerve (Arroyo and Scherer, 2000; Scherer and Arroyo, 2002). Myelin protein gene expression in dermal nerves is similar to that in other myelinated nerves in the PNS PNS axonal signals induce myelinating Schwann cells to express a group of myelin-specific genes and their proteins in a coordinate pattern. Thus, MPZ, PMP22, MBP, MAG and PLP1 mrnas were all detectable by real-time PCR Skin biopsies in myelin-related neuropathies 1171 Fig. 2 Dermal myelinated nerve compaction is similar to that of myelinated sural or sciatic nerve. The figure demonstrates the ultrastructure of myelination (A and B) and the Schwann cell (A) in dermal myelinated nerve fibres. Myelin periodicity is nm (B), similar to what has been described for other PNS myelinated fibres (Dyck et al., 1993). performed on human sural nerve, with MPZ mrna levels expressed at much higher levels than the other myelin specific mrnas (Fig. 4). Similar expression patterns were obtained with real-time PCR performed with samples from glabrous skin (Fig. 4). PMP22 mrna levels in the skin were lower than those in the sural nerve. We believe that this is probably the result of the small amount of PMP22 mrnas present in the skin, making precise comparison with sural nerve levels difficult, as is probably the case for other myelinspecific mrnas present in small amounts. Increased PLP1 mrna expression in skin compared with sural nerve probably results from the known PLP1 expression in fibroblasts in addition to Schwann cells (Wight et al., 1997). To obtain both spatial and quantitative information about myelin proteins, we performed immuno-em on skin biopsies taken from human subjects without neuropathy. MPZ, MBP and PMP22 were all detected in compact myelin labelled with antibodies conjugated with gold particles (Fig. 5). Consistent with mrna studies, gold particles labelled with antibodies

5 1172 J. Li et al. Fig. 3 The molecular architecture of dermal myelinated nerves. An immunohistochemistry study was performed on a skin biopsy from the forearm of a normal subject. Double-staining with PGP9.5 (green) and Caspr (red) antibodies shows that discrete bands of Caspr in the paranodal regions flank the node of Ranvier bilaterally (arrows). Spiralling distribution of Caspr on the axolemma opposing inner mesaxon can also be visualized (A, inset). Voltage-gated sodium channels are restricted at the nodes of Ranvier (arrows in B). Potassium channels are located in the juxtaparanodes (arrowheads) and spiral along the axolemma opposing the inner mesaxon of the internode (arrows in C). Myelin basic protein (blue) is also stained in these nerves (arrows in D). These data suggest that skin myelinated nerves have a molecular architecture that is similar to that of other PNS myelinated nerves. Fig. 4 Gene expression of dermal myelin proteins. Real-time PCR was used to detect myelin-specific mrnas from Schwann cells in skin biopsies. We extracted RNA from a 2 mm skin punch taken from finger skin from a normal volunteer and a human sural nerve, and performed real-time PCR on the samples using primer pairs which were specific for multiple myelin protein genes. The results demonstrate the presence of similar profiles of myelin protein mrna in both human skin and in sural nerve, myelin protein zero being expressed at the highest levels. Proteolipid protein (PLP) levels appear higher in skin because PLP mrna is also expressed by fibroblasts. 18S RNA was used as a control.

6 Skin biopsies in myelin-related neuropathies 1173 Fig. 5 Spatial and quantitative expression of myelin specific proteins. Immunoelectron microscopy studies were performed in skin biopsies from normal controls. MPZ (A), MBP (B) and PMP22 (C) were labelled with antibodies conjugated with gold particles. These particles were quantified with ImagePro Plus software (D). Studies were subsequently performed in patients with CMT1A and HNPP. PMP22 levels were increased in patients with CMT1A (PMP22 duplication) and decreased in patients with HNPP (PMP22 deletion) (D), as has been described previously in sural nerve (Vallat et al., 1996). to MPZ were far more numerous than those labelled with antibodies to PMP22 (Fig. 5A, C). Taken together, the results from the above studies demonstrate that the structure and organization of myelinated nerves in glabrous skin is similar to that elsewhere in the PNS. Myelin abnormalities in dermal nerves from patients with inherited neuropathies To investigate changes in dermal nerves in patients with inherited neuropathies we evaluated a series of patients with CMT1A, HNPP or CMTX. Abnormalities were detected in all patients. In patients with CMT1A (n = 3), the number of myelinated nerve fibres within fascicles was less than in control subjects (similar to Fig. 6A, data not shown), although morphometric quantification was not performed. One patient, with a severe phenotype, showed total absence of myelinated nerve fibres in the small nerve bundle in skin from his finger. Some nerve fibres appeared thinly myelinated (Fig. 6B). Interestingly, regenerating clusters (or pseudo-onion bulbs) were only occasionally observed by EM (Fig. 6C). No onion bulbs were found. The most common abnormality, detected by EM, was decompaction of inner myelin wraps (Fig. 6D), which was present in most of the myelinated nerve fibres in all patients. Skin biopsies from our CMTX1 patient were also abnormal. The nerve bundle (Fig. 6A) also had a reduction in

7 1174 J. Li et al. myelinated fibres compared with control bundles, consistent with electrophysiological studies and reports from sural nerve biopsies demonstrating axonal loss in CMTX1 (Hahn et al., 2001). EM again demonstrated decompaction in the adaxonal myelin. Re-organization of paranodes and juxtaparanodes in CMT1 and CMTX To investigate the molecular architecture of regions critical for communication between Schwann cells and axons, we performed immunohistochemistry on proteins known to play a role in axo-glial interactions at nodes, paranodes, juxtaparanodes and internodes. CMT1A biopsies contained nerves in which Caspr bands, a paranodal marker, were abnormally distributed, with abnormal spreading into the internodes (Fig. 6F). Possible segmental demyelination was suggested by fibres with shortened internodes in which MBP staining was absent (Fig. 6F). In some nerves from CMTX1, voltagegated potassium channels were distributed in patches along the axolemma rather than in discrete locations (Fig. 6G). Finally, myelinated dermal nerves from HNPP patients showed a paranodal enlargement or tomacula, as has been described in sural nerves from patients with HNPP (Fig. 6H) Changes in PMP22 expression quantifiable by immunoelectron microscopy We wished to determine whether altered expression of PMP22 in dermal myelinated nerve could be detected, as has previously been reported, in sural nerves from patients with CMT1A and HNPP (Vallat et al., 1996). Accordingly we performed immuno-em for PMP22 on four patients with CMT1A and two patients with HNPP and compared the number of gold particles/mm 2 with results from two healthy controls. The number of gold particles/mm 2 was evenly distributed in all the fibres of the normal control ( ). Expression of PMP22 was significantly higher (P < 0.02) in the CMT1A patients ( ) and lower in the HNPP patients ( , P < ) than in the controls. While CMT1A PMP22 levels were higher than controls, there was variability in the number of PMP22 gold particles among patients. Conceivably, this could contribute to the known phenotypic variability in CMT1A (Krajewski et al., 2000). Discussion We have demonstrated that bundles of myelinated nerves can be routinely identified and evaluated in biopsies taken from glabrous skin in the finger and forearm. Individual myelinated nerve fibres innervating dermal mechanical receptors can be readily visualized. For the most part, these dermal nerves appeared identical to other myelinated nerve fibres in the PNS. EM revealed the same 9 12 nm periodicity as that described in sural nerve myelin. The molecular organization of internodes, nodes, paranodes and juxtaparanodes was similar to what has been described in rodent sciatic nerve (Scherer and Arroyo, 2002; Arroyo and Scherer, 2000). MPZ is the predominant protein in dermal myelin, which also contains the myelin-specific proteins PMP22 and MBP. We detected only two differences between the sensory myelinated axons in glabrous skin and those in sural nerve. First, dermal nerves contained primarily smallcalibre myelinated axons, whereas sural nerve contained both small and large diameter fibres. Second, myelin internodes from glabrous skin were shorter (around 70 mm) than those in sciatic or sural nerve ( mm). Skin biopsies were well tolerated by both patients and controls. Thus they were less invasive than sural nerve biopsies, which require surgery to obtain, result in sensory loss and subsequent dysaesthesias, and cannot be repeated multiple times. Glabrous skin biopsies proved effective in detecting many of the known morphological abnormalities in inherited demyelinating neuropathies, previously described in sural nerve biopsies. Thinly myelinated axons were observed from both CMT1A and CMTX1 patients, as well as axonal loss from both groups of patients. Biopsies from HNPP patients showed tomacula. Immuno-EM studies successfully revealed myelin proteins, such as P0, PMP22 and MBP, expressed in myelin. Immuno-EM of myelin protein PMP22 demonstrated the predicted increment of this protein in CMT1A and decrement of PMP22 in HNPP. Interestingly, typical onion bulb formation was not observed in CMT1A skin biopsies, though pseudo-onion bulbs were observed as regenerating axonal clusters. The lack of onion bulbs is not because dermal nerves are incapable of forming them, as onion bulbs were detected in most fascicles in a skin biopsy taken from a severely affected infant with neuropathy caused by a PMP22 missense mutation (Ceuterick-de Groote et al., 2001). The lack of onion bulbs is also probably not due to abnormalities in dermal nerve and are not severe enough to cause onion bulb formation, as onion bulbs are an early feature of CMT1A (Gabreels-Festen and Wetering, 1999). We speculate that the lack of onion bulbs in glabrous skin of CMT1A patients is because dermal myelinated nerve fibres are small in calibre and onion bulb formation occurs primarily in large-diameter fibres in CMT1A. The sural nerve, on the other hand, contains many large- and small-diameter fibres (Dyck et al., 1993). With respect to the point mutation case cited above, we suspect that the onion bulbs occurred because smaller-diameter fibres were more affected in this severe neuropathy. Alternatively, this autopsy skin biopsy may have been taken from deeper skin with more large-diameter nerve fibres. Glabrous skin biopsies also provided insight into at least two potential pathogenic mechanisms in CMT that have not been characterized in human sural nerve studies. First, skin biopsies detected abnormalities in the molecular architecture of paranodes and juxtaparanodes similar to those that are hypothesized to cause axonal degeneration in several mouse myelin mutants. Septate junctions between paranodal loops

8 Skin biopsies in myelin-related neuropathies 1175 and the axolemma are absent in UDP-galactose ceramide galactosyltransferase (cgt) and cerebroside sulphotransferase (cst)-null mice (Marcus et al., 2000; Honke et al., 2002). In the absence of septate junctions there is a profound re-organization of the axonal membrane. Caspr and contactin become diffusely distributed in the internodal axolemma rather than being sequestered in the paranode (Dupree and Popko, 1999). Kv1.1 and Kv1.2 are mislocalized to the paranodal axonal membrane (Boyle et al., 2001). Sodium channels remain concentrated at the nodes. Alterations of

9 1176 J. Li et al. the molecular architecture of peripheral nerves have been also shown in connexin32 / (gjb / ) mice (Neuberg et al., 1999). Molecular reorganization of ion channels and the paranodes was also identified in skin biopsies of our CMT1A and CMTX1 patients. These results from skin biopsies suggest that the abnormal molecular organization of paranodal loops and the underlying axolemma may impair Schwann cell axon communication and contribute to axonal degeneration in patients with CMT1A and CMTX1. A second potential insight into the pathogenesis of CMT arose from the frequent decompaction of periaxonal myelin lamella observed in both CMT1A and CMTX1 skin biopsies. These were frequent in all patient biopsies but were less likely to be seen in controls, suggesting that periaxonal decompaction is a common, perhaps early, pathological feature of CMT1A and CMTX1. In some respects this situation appears similar to the dying back of terminal twigs in axonal neuropathies. Terminal twigs are the most distal portions of axons, far removed from the neuronal perikaryon. Retraction of these terminal twigs from neuromuscular junctions is one of the early morphological abnormalities detected in neuropathies (Yin et al., 2004), as well as in SOD1 mouse models of ALS (Fischer et al., 2004). These terminal retractions are hypothesized to contribute to the pathogenesis of both disorders. Skin biopsy results suggest that a similar process may be occurring in Schwann cells in CMT1. The inner mesaxon and periaxonal myelin wraps are the most distal regions of the myelinating Schwann cell, the furthest removed from the Schwann cell perikaryon. Thus, we hypothesize that a dying back process of the distal Schwann cell may be an early feature of both CMT1A and CMTX1. Moreover, this dying back process may contribute to the abnormalities in molecular organization described above. The complex cellular structures formed by myelinating Schwann cells and their axons are analogous in many respects to the neuromuscular junction formed between terminal axon twigs and muscle cells: both are highly ordered multicomponent systems formed by the interaction of two distinct cell types in order to carry out a specific biological function related to nerve transmission. Just as the molecular organization of the neuromuscular junction is disrupted by withdrawal of terminal twigs, so may the molecular organization of paranodes, juxtaparanodes and internodes be altered by alterations in distal Schwann cell myelin wraps. In summary, we believe that our data demonstrate that glabrous skin biopsies can be used to identify known and novel morphological abnormalities in inherited demyelinating neuropathies. Because skin biopsies can be repeated and performed at multiple sites, we believe they can be investigated as morphological markers of disease progression in both natural history studies and clinical trials. The abnormal findings from skin biopsies have been found to correlate with disability in the CMT1A rat model in a preliminary study (Horste et al., 2004). Moreover, our preliminary results demonstrate more severe morphological changes in patients more severely disabled by CMT1A. The data also support the use of glabrous skin biopsies in genotype phenotype correlations in CMT. At present, mutations in over 30 genes are known to cause CMT and in some cases over 250 mutations in an individual gene have been identified (Belgian CMT database). It is impossible to create knock-in mice for all these mutations. Glabrous skin biopsies should allow patients to serve as their own disease model for their disease in a far less invasive way than with sural nerve biopsies. We recognize that skin biopsies are not necessary to diagnose various forms of CMT, which can be achieved by genetic testing from blood. However, based on our data, we believe that they can become valuable research tools to understand disease mechanisms and develop treatments. Finally, there is no reason why glabrous skin biopsies cannot also be used to investigate non-heritable neuropathies, such as chronic inflammatory demyelinating polyradiculoneuropathy, Guillain Barré syndrome and diabetic neuropathy. Perhaps the most significant result of this study is that it demonstrates that myelinated nerves from patients can be obtained in a minimally invasive manner and then analysed by modern molecular techniques without causing harm to the patient. Fig. 6 Morphological and molecular alterations in skin biopsies of CMT patients. Skin biopsies were taken from patients with either a GJB1 mutation (CMTX1), chromosome 17q11.2 duplication (CMT1A) or chromosome 17q11.2 deletion (HNPP). Myelinated nerve fibres within a small nerve bundle were reduced in the skin biopsy of a patient with CMTX1 (A, original magnification 6003). Under EM, some nerve fibres from the same patient appeared thinly myelinated (B). A small regenerating cluster (pseudo-onion bulb) is shown from a patient with CMT1A (C). Myelin in most myelinated fibres was abnormal in all CMT1A and CMTX1 patients. Adaxonal myelin laminar decompaction was the most common pathological finding (D, from a CMT1A patient) and was present in most myelinated nerve fibres of all CMT1A and CMTX1 patients. Some decompaction was observed in control nerves but this was less frequent and never as severe as what was observed in patients. Immunohistochemistry showed evidence of reorganization of the molecular architecture in dermal myelinated nerves in both CMT1A and CMTX1 patients. CMT1A nerves contained fibres in which Caspr bands (paranodal marker) were abnormally distributed (F), as shown by triple staining with antibodies to Caspr (red), PGP9.5 (green) and MBP (blue). Two pairs of Caspr bands (arrowheads) flank a short segment of nerve axon (green), suggesting an evolving shortened internode, a region of possible segmental demyelination, or both. There was abnormal spreading of Caspr into the internode (arrow). These findings are in contrast to the discrete normal distribution of Caspr bands (E) seen in another nerve fibre from the same patient. In G, double staining with antibodies to voltagegated potassium channels (red) and PGP9.5 (green) was performed on the skin biopsy from a patient with CMTX1. Potassium channels were diffusely distributed in patches on the axolemma rather than in the discrete locations shown in normal controls (C). A tomacula observed in dermal nerves from an HNPP patient is shown in H. Myelin is labelled by MBP antibodies (green). The hemiparanode on the left, localized by Caspr staining (red), is enlarged. The paranodal axon, marked by Caspr, is eccentrically displaced by the tomacula.

10 Acknowledgements This study is supported by The Buuck Family Foundation, the Charcot Marie Tooth Association, MDA grants (J.L. and M.E.S.), and NIH grants NS (J.L.) and NS41319A (M.E.S.). References Archelos JJ, Roggenbuck K, Schneider-Schaulies J, Linington C, Toyka KV, Hartung HP. Production and characterization of monoclonal antibodies to the extracellular domain of P0. J Neurosci Res 1993; 35: Arroyo EJ, Scherer SS. On the molecular architecture of myelinated fibers. Histochem Cell Biol 2000; 113: Boyle ME, Berglund EO, Murai KK, Weber L, Peles E, Ranscht B. Contactin orchestrates assembly of the septate-like junctions at the paranode in myelinated peripheral nerve. Neuron 2001; 30: Ceuterick-de Groote C, De Jonghe P, Timmerman V, et al. Infantile demyelinating neuropathy associated with a de novo point mutation on Ser72 in PMP22 and basal lamina onion bulbs in skin biopsy. Pathol Res Pract 2001; 197: Dupree JL, Popko B. Genetic dissection of myelin galactolipid function. J Neurocytol 1999; 28: Dyck PJ, Griffin JW, Low PA, Poduslo JF. Peripheral neuropathy. W.B. Saunders; Fischer LR, Culver DG, Tennant P, et al. Amyotrophic lateral sclerosis is a distal axonopathy: evidence in mice and man. Exp Neurol 2004; 185: Gabreels-Festen A, Wetering RV. Human nerve pathology caused by different mutational mechanisms of the PMP22 gene. Ann NY Acad Sci 1999; 883: Hahn AF, Ainsworth PJ, Bolton CF, Bilbao JM, Vallat JM. Pathological findings in the x-linked form of Charcot-Marie-Tooth disease: a morphometric and ultrastructural analysis. Acta Neuropathol (Berl) 2001; 101: Holland NR, Crawford TO, Hauer P, Cornblath DR, Griffin JW, McArthur JC. Small-fiber sensory neuropathies: clinical course and neuropathology of idiopathic cases. Ann Neurol 1998; 44: Holland NR, Stocks A, Hauer P, Cornblath DR, Griffin JW, McArthur JC. Intraepidermal nerve fiber density in patients with painful sensory neuropathy. Neurology 1997; 48: Skin biopsies in myelin-related neuropathies 1177 Honke K, Hirahara Y, Dupree J, et al. Paranodal junction formation and spermatogenesis require sulfoglycolipids. Proc Natl Acad Sci USA 2002; 99: Horste MG, Nave KA, Sereda MW. PMP22 expression in skin biopsies serves as a marker for disease severity in a model of Charcot-Marie-Tooth disease 1A (CMT1A). In: First European and North American Charcot-Marie- Tooth Consortium Meeting, July 8 10, 2004, Antwerp. Kennedy WR, Wendelschafer-Crabb G. The innervation of human epidermis. J Neurol Sci 1993; 115: Kennedy WR, Wendelschafer-Crabb G, Johnson T. Quantitation of epidermal nerves in diabetic neuropathy. Neurology 1996; 47: Krajewski KM, Lewis RA, Fuerst DR, et al. Neurological dysfunction and axonal degeneration in Charcot-Marie-Tooth disease type 1A. Brain 2000; 123: Li J, Lewis R, Shy M. Charcot-Marie-Tooth disease. In: Aminoff M, Daroff R, editors. Encyclopedia of the neurological sciences. Academic Press: Los Angeles, USA; p Marcus J, Dupree JL, Popko B. Effects of galactolipid elimination on oligodendrocyte development and myelination. Glia 2000; 30: Neuberg DH, Sancho S, Suter U. Altered molecular architecture of peripheral nerves in mice lacking the peripheral myelin protein 22 or connexin32. J Neurosci Res 1999; 58: Nolano M, Provitera V, Crisci C, et al. Quantification of myelinated endings and mechanoreceptorsinhumandigitalskin. AnnNeurol2003; 54: Peles E, Nativ M, Lustig M, et al. Identification of a novel contactinassociated transmembrane receptor with multiple domains implicated in protein protein interactions. EMBO J 1997; 16: Polydefkis M, Hauer P, Griffin JW, McArthur JC. Skin biopsy as a tool to assess distal small fiber innervation in diabetic neuropathy. Diabetes Technol Ther 2001; 3: Polydefkis M, Yiannoutsos CT, Cohen BA, et al. Reduced intraepidermal nerve fiber density in HIV-associated sensory neuropathy. Neurology 2002; 58: Scherer SS, Arroyo EJ. Recent progress on the molecular organization of myelinated axons. J Peripher Nerv Syst 2002; 7: Vallat JM, Sindou P, Preux PM, et al. Ultrastructural PMP22 expression in inherited demyelinating neuropathies. Ann Neurol 1996; 39: Wight PA, Dobretsova A, Macklin WB. Regulation of murine myelin proteolipid protein gene expression. J Neurosci Res 1997; 50: Yin X, Kidd GJ, Pioro EP, et al. Dysmyelinated lower motor neurons retract and regenerate dysfunctional synaptic terminals. J Neurosci 2004; 24: