Axonal Pathology Precedes Demyelination in a Mouse Model of X-Linked Demyelinating/ Type I Charcot-Marie Tooth (CMT1X) Neuropathy

Abstract

X-linked Charcot-Marie-Tooth disease (CMT1X) is an inherited peripheral neuropathy caused by mutations in GJB1, the gene that encodes the gap junction protein connexin32 (Cx32). Cx32 is expressed by myelinating Schwann cells and forms gap junctions in non-compact myelin areas but axonal involvement is more prominent in X-linked compared to other forms of demyelinating Charcot-Marie-Tooth disease. To clarify the cellular and molecular mechanisms of axonal pathology in CMT1X, we studied Gjb1-null mice at early stages (i.e. 2- to 4-month-old) of the neuropathy, when there is minimal or no demyelination. The diameters of large myelinated axons were progressively reduced in Gjb1-null mice compared to those in wild type littermates. Furthermore, neurofilaments were relatively more dephosphorylated and more densely packed starting at 2 months of age. Increased expression of β-amyloid precursor protein, a marker of axonal damage, was also detected in Gjb1-null nerves. Finally, fast axonal transport, assayed by sciatic nerve ligation experiments, was slower in distal axons of Gjb1-null vs. wild type animals with reduced accumulation of synaptic vesicle-associated proteins. These findings demonstrate that axonal abnormalities including impaired cytoskeletal organization and defects in axonal transport precede demyelination in this mouse model of CMT1-X.

INTRODUCTION

Since the first report of Bergoffen et al (1), over 300 different GJB1 mutations have been found to cause the X-linked demyelinating/type 1 form of Charcot-Marie-Tooth disease (CMT1X) (http://www.molgen.ua.ac.be/CMTMutations/default.cfm). GJB1 encodes the gap junction (GJ) protein connexin32 (Cx32), which is thought to form GJs between the layers of non-compact myelin in peripheral nervous system (PNS) myelin sheaths (2–4). Patients with CMT1X develop a progressive length-dependent axonal loss that results in the clinical picture of progressive muscle atrophy, weakness, and sensory loss in distal extremities (5). The electrophysiological and pathological findings in CMT1X patients suggest that there is less demyelination and more axonal loss compared to patients with other types of CMT1 (6–10). In CMT1X, as in other forms of CMT1, axonal degeneration and not demyelination appears to be the main cause of neurological disability (11).

Several animal models of CMT1 have provided insight into how demyelinating neuropathies evolve. In Gjb1-null mice, demyelination begins at around post-natal day 90 (P90), with subsequent axonal loss (12, 13). In Pmp22-null (14), Trembler (15) and TremblerJ (16), as well as Mpz-null mice (17–18), demyelination has an even earlier onset, and distally accentuated axonal loss worsens over time. In demyelinated axons of Trembler mice, neurofilaments (NFs) are more densely packed and less phosphorylated and axonal transport is slower (19). Some of these abnormalities have also been found in nerves from patients with CMT1 (20). Transplantation of a segment of Trembler (21, 22) or CMT1A (23) nerve into normalnerve also produces similar changes in axons that have regenerated through the nerve graft but not in the surrounding nerve, demonstratingthat this effect results from interactions with genetically abnormal Schwanncells. Finally, mutant Schwann cells from CMT1X patients transplanted into sciatic nerves of nude mice induced increased density of axonal neurofilaments, depletion of microtubules, and increased density of vesicles and mitochondria without demyelination (24).

How Cx32 mutations lead to early axonal dysfunction and degeneration in the absence of demyelination remains poorly understood. The aim of this study was to examine the onset and mechanisms of early axonal pathology in Gjb1-null mice. We chose this model of CMT1X neuropathy because our recent studies demonstrated that loss of function is the main mechanism of GJB1 mutations (25). Furthermore, several clinical reports have shown similar disease severity in CMT1X patients with point mutations compared to complete deletion of the GJB1 gene (26–28). We show here that before demyelination occurs, Gjb1-null mice have altered NFs, disturbed axonal transport, and increased expression of amyloid precursor protein (APP), a marker of axonal injury (29). These results provide evidence that the loss of Cx32 function in PNS myelin sheaths affects axons independently of demyelination.

MATERIALS AND METHODS

Transgenic Mice

Gjb1-null mice (C57BL/6x129) were obtained from the European Mouse Mutant Archive, Monterotondo, Italy (originally generated by Prof. Klaus Willecke, University of Bonn, Germany) (30). In these mice, the neo^r gene was inserted in frame into the exon 2 of Gjb1 gene, which contains the open reading frame. Genotyping was performed using PCR screening with primers specific for the wild type (WT) Gjb1 mouse gene as well as for the neo^r gene (Gjb1-null), as previously described (25). Gjb1-null mice were compared to their WT littermates at the ages of 2 and 4 months. Studies were also performed in 8-month-old mice to assess progression of axonal pathology.

Immunoblot Analysis

Sciatic nerves were dissected into the proximal (lumbar spine to the sciatic notch) and distal (sciatic notch to the knee) segments and lysed in ice-cold radioimmunoprecipitation assay (RIPA) buffer (10 mM sodium phosphate pH 7.0, 150 mM NaCl, 2 mM EDTA, 50 mM sodium fluoride, 1% NP-40, 1% sodium deoxycholate, and 0.1% SDS), containing a cocktail of protease inhibitors (Roche, Basel, Switzerland). For immunoblot analysis of NFs, samples were additionally treated with a cocktail of phosphatase inhibitors (Roche) to ensure preservation of phosphorylation status. No difference in the yield of either SMI31 or SMI32 was observed when the same samples were processed with Laemmli (62.5 mM Tris pH6.8, 2% SDS) as opposed to RIPA buffer, and no significant amounts of NFs were retained in the RIPA-insoluble fraction (Supplemental Fig. 1). Protein lysates (30 μg) were fractionated by 8% SDS-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to a Hybond-C extra membrane (Amersham, Piscataway, NJ) using a semi-dry transfer unit (Amersham). The membranes were blocked with 5% non-fat milk in Tris-buffered saline containing 0.1% Tween 20 (TBS-T) for 1 hour at room temperature (RT), then incubated overnight at 4°C with the following primary antibodies in 5% milk-TBS: mouse monoclonal antibodies against phosphorylated (SMI31, Abcam, Cambridge, UK, 1:25,000), dephosphorylated (SMI32, Abcam, 1:3,000) and total (phosphorylated and unphosphorylated- N52, Sigma Aldrich, St. Louis, MO; 1:2,000) NF-heavy (NF-H), as well as against β-tubulin (E7; Developmental Studies Hybridoma Bank, DSHB, University of Iowa , Iowa City, IA, 1:3,000), tau (DSHB, 1:1,000), glyceraldehyde 3-phosphate dehydrogenase (GAPDH; Santa Cruz Biotechnology, Santa Cruz, CA, 1:3,000) or myelin basic protein (MBP) (Abcam, 1:5,000). After washing, immunoblots were incubated for 1 hour with appropriate HRP-conjugated secondary antisera (Jackson ImmunoResearch, West Grove, PA, diluted 1:5,000–1:10,000) in 5% milk-TBS-T. The bound antibody was visualized by enhanced chemiluminescence system (ECL Plus, Amersham). The intensity of the bands was measured with Tinascan version 2.07d and normalized for loading with GAPDH bands. Genotypes were compared with the paired t-test (95% confidence interval; 0.05 threshold).

Immunohistochemistry

Mice were anesthetized with Avertin according to institutionally approved protocols, then transcardially perfused with phosphate-buffered saline (PBS) followed by fresh 4% paraformaldehyde in 0.1 M PBS. Sciatic nerves were post-fixed for 30 minutes in the same fixative. Teased nerve fibers were prepared from both fixed and unfixed nerves, dried on SuperFrost Plus glass slides overnight at RT and stored at −20°C. Teased fibers were permeabilized in acetone (−20°C for 10 minutes) and incubated at RT with blocking solution of 5% bovine serumalbumin containing 0.5% Triton-X for 1 hour. The primary antibodies diluted in blocking solution were incubated overnight at 4°C with mouse monoclonal antibodies against contactin-associated protein (Caspr, 1:50; gift of Dr. Elior Peles, Weizmann Institute of Science, Israel), MBP (Abcam, 1:500), PanNav (Sigma, 1:50), Cx32 (1:50; Zymed-Invitrogen, Carlsbad, CA), SMI31 (Abcam; 1:5000), SMI32 (Abcam; 1:500), synaptotagmin (DSHB; 1:50), dynein (Santa Cruz, 1:100), kinesin- (KIF3a, BD Biosciences, San Jose, CA, 1:50) and tau (DSHB, 1:300), as well as rabbit antisera against MBP (1:100; Sigma), Nav1.6 (1:100; Alomone Labs, Jerusalem, Israel), Nav1.8 (Alomone; 1:100), Kv1.2 (1:200; Alomone), APP (Chemicon; Millipore, Temecula, CA, 1:300).

Quantitation of APP+ Teased Sciatic Nerve Fibers

Quantitation of APP+ teased sciatic nerve fibers was carried out after staining with a rabbit antiserum against APP, combined with monoclonal against Caspr or MBP. At least 75 randomly selected fibers per animal from at least 3 different animals were examined in each genotype and age group (2, 4, and 8 months). Fibers from both genotypes were teased and stained at the same time and scored for APP+ staining by 2 observers (N.V. and K.A.K.) who were blinded to the genotype. Fibers were considered to have positive staining for APP when there was strong and continuous APP immunoreactivity (IR) extending to more than 3 internodes. Counts were reproducible between observers and on repeated evaluation of the slides (performed twice). We compared the proportion of APP+ fibers in each genotype with the unpaired t-test.

Electron Microscopy

Deeply anesthetized 2-, 4- and 8-month-old mice were transcardially perfused with 0.9% saline followed by 2.5% glutaraldehyde in 0.1M PBS (pH 7.2). The femoral and sciatic nerves were dissected and further fixed overnight at 4°C and then osmicated, dehydrated, and embedded in Araldite resin. Transverse semi-thin sections (1 μm) were stained with alkaline toluidine blue. Ultrathin sections (80–100 nm) were counterstained with lead citrate and uranyl acetate, and examined in a JEOL JEM-1010 transmission electron microscope (JEOL Ltd, Tokyo, Japan).

Analysis of Myelination

Transverse semi-thin sections of the sciatic nerves were examined. All demyelinated, remyelinated and normally myelinated axons were counted using the following criteria: axons larger than 1 μm without a myelin sheath were considered demyelinated; axons with myelin sheaths that were <10% of the axonal diameter and/or axons that were surrounded by “onion bulbs” (circumferentially arranged Schwann cell processes and extracellular matrix) were considered remyelinated; the other myelinated axons were considered normally myelinated. All measurements were performed blindly with respect to the genotype. Proportions of abnormally myelinated fibers were compared with the Mann-Whitney U test (significance level: p = 0.05).

Analysis of Axon Profiles

Axon calibers were assessed in 2- and 4-month-old WT and Gjb1-null mice (n = 3 from each genotype) using the ImagePro Analyzer 6.3 (Media Cybernetics, MD). Digital images of femoral nerves were captured with the 20x objective. The total area of axons (sum of all axon transverse areas) within each nerve and the total number of axons were measured. Axons were further categorized and quantified according to their mean diameter into 8 classes (i.e. 1–2, 2–3, 3–4, 4–6, 6–8, 8–10, 10–12 and 12–14 μm). All results were compared with the t-test.

Ultrastructural Analysis

Ultrathin transverse and longitudinal sections of femoral motor nerves from at least 3 different WT and Gjb1-null mice at the age of 2 and 4 months were examined. Only normally myelinated large axons (6- to 8-μm-diameter) were analyzed. To count mitochondria, the entire axon cross section was photographed at x15,000 magnification. To measure the cytoskeleton elements, a series of x30,000 images were then obtained from at least 10 different axons in each nerve. NFs and microtubules were counted automatically from 3 to 5 different randomly selected 1 μm² areas in each axon using the ImagePro Plus analyzer. All measurements were performed blindly in regards to the genotype. The average numbers of NFs, microtubules and mitochondria per 1 μm² were obtained in each axon, mouse and genotype and their densities were compared between genotypes with the t-test.

Sciatic Nerve Ligations

Four month- old Gjb1-null mice and their WT littermates (n = 4 per genotype) were deeply anesthetized. The right sciatic nerve was exposed and 2 5/0 silk ligatures (SilkamR, Braun, Tuttlingen, Germany), approximately 5 mm apart, were placed immediately distal to the sciatic notch. The mice were killed 3 hours later and equal nerve segments (~1-cm-long) proximal and distal to the ligature were collected. The corresponding segments were also collected from the non-ligated contralateral sciatic along with samples of the lumbar spinal cord. Tissues were further processed for immunoblot analysis as above. The segments between the ligatures were collected separately, fixed for 30 minutes in 4% paraformaldehyde and processed for immunohistochemistry for several proteins undergoing axonal transport as above.

Antibodies used for immunoblot analysis in nerve ligation experiments included mouse monoclonal against synaptic proteins SV2 (developed by Kathleen M. Buckley, DSHB; diluted: 1:500), synaptotagmin (mAb48; developed by Louis Reichardt, DSHB, diluted: 1:1,000) and synapsin (DSHB; 1:250, developed by Erich Buchner), and against Dynein intermediate chain (Santa-Cruz, 1:1,000), MBP (Abcam; 1:5,000), and GAPDH (Santa Cruz; 1: 3,000). Blots were incubated with appropriate secondary antibodies and developed as above. For quantification, the intensity of the bands was measured and synaptic vesicle protein bands were normalized for loading according to GAPDH bands. We then calculated the ratios of synaptic protein levels in proximal and distal segments from ligated compared to contralateral unligated nerves and analyzed the results from Gjb1-null and WT mice with paired t-tests.

RESULTS

Development of Demyelination in Gjb1-Null Nerves

The femoral nerves of Gjb1-null mice showed progressive, predominantly motor demyelinating neuropathy starting after 3 months of age (12, 13), with ~7% abnormally myelinated fibers at 4 months (25). To assess the degree of demyelination in the sciatic nerve we analyzed epoxy sections at 2, 4 and 8 months of age. Compared to femoral motor nerves, sciatic nerves had fewer abnormally myelinated fibers: 0% at 2 months and 3% at 4 months (Fig. 1). To determine whether the molecular architecture of myelinated axons was affected, we immunostained teased fibers of 2- and 4-month old Gjb1-null sciatic nerves. All myelinated fibers had normal appearing, MBP+ myelin sheaths and all nodes of Ranvier were Nav1.6+ (Supplemental Fig. 2E, F). We did not detect nodal Nav1.8 staining (Supplemental Fig. 2G, H), which has been found in some abnormally myelinated fibers in Tr^J mice (31), whereas small-diameter dorsal root ganglion neurons were Nav1.8+ (data not shown). As previously reported (32), Shaker-type Kv1.1/1.2 potassium channels and Caspr2 were normally expressed at juxtaparanodes and Caspr was normally localized at paranodes (Supplemental Fig. 2A–D). Moreover, the internodal lengths were not significantly different between 4-month-old Gjb1-null mice and their WT littermates (Supplemental Fig. 3), further supporting the contention that demyelinated and remyelinated axons are scarce at this age.

Figure 1.

Sciatic nerve demyelination in Gjb1-null mice is minimal up to 4 months of age. (A–F) Semithin sections from 2- (A, B), 4- (C, D) and 8-month-old (E, F) wild type (WT) (A, C, E) and Gjb1-null (B, D, F) mice. In Gjb1-null sciatic nerves there are no demyelinated or remyelinated axons at 2 months of age. Abnormally myelinated axons are rare at 4 months (D), and become more numerous at 8 months (F), including demyelinated (arrows) and remyelinated (arrowheads). Bar = 10 μm. (G) Quantitative analysis of sciatic nerve myelination in WT and Gjb1-null mice at 2, 4, and 8 months of age (n = 4 per genotype and age group) shows very mild demyelination in Gjb1-null mice at 4 months with 3.4 ± 1.2% (average ± SD) of fibers being abnormally myelinated vs. none in WT, (p = 0.01, Mann-Whitney U test); by 8 months there is progression (15.7 ± 7.6% of fibers abnormally myelinated vs. 0.4 ± 0.5% in WT; p = 0.03).

Increased Packing Density and Progressive Dephosphorylation of NFs in Gjb1-Null Nerves

As a first step in determining whether axons were affected in Gjb1-null mice prior to the onset of demyelination, we performed a quantitative analysis of the axonal calibers in femoral motor nerves of 2- and 4-month-old littermates. Even at 2 months, the numbers of large (>8 μm diameter) axons were reduced in Gjb1-null nerves; this reduction was more prominent at 4 months (Fig. 2). In contrast, Gjb1-null nerves showed higher percentages of medium caliber fibers (4–8 μm diameter), suggesting a shift towards smaller axon diameter in the Gjb1-null mice without axonal loss. Consistent with this observation, total axonal area and total number of axons did not significantly change at these ages.

Figure 2.

Reduced axonal diameters in Gjb1-null mice. (A–D) Representative semithin sections of femoral motor nerves from 2- (A, B) and 4-month-old (C, D) wild type (WT) (A, C) and Gjb1-null (B, D) mice. A few axons are demyelinated or remyelinated at 4 months of age in Gjb1-null nerves. Scale bar = 20 μm. (E–J) Quantitative analysis of axon diameters (E, F), total axonal area (G, I), and total numbers of axons (H, J) in WT and Gjb1-null mice (n = 3 per age group and genotype) at 2 months (E, G, H) and 4 months (F, I, J). At both ages the percentages of large myelinated fibers (>8μm) is reduced in Gjb1-null vs. WT nerves. In contrast, Gjb1-null nerves show higher percentages of medium sized fibers (4 to 6 μm at 2 months; 4 to 8 μm at 4 months). The total axonal area and the total number of axons are not significantly changed in Gjb1-null nerves, suggesting a shift to smaller diameters without axonal loss. *Indicates significant differences (2-tailed t-test).

We next analyzed NF density by electron microscopy in large myelinated axons in femoral motor nerves (>6 μm), excluding axons sectioned near the nodal region as well as axons that were demyelinated or remyelinated. NF density at 2 months was significantly greater in the Gjb1-null (n = 4, 116.4 ± 6.1 NF/μm²) vs. WT mice (n = 3, 92.9 ± 1.6 NF/μm²) (p = 0.0016, 2-tailed t-test). NF density was further increased at 4 months (144 ± 6.3 NF/μm² in Gjb1-null (n = 5) vs. 89.9±5.6 NF/μm² in WT mice (n = 4) (p < 0.001). There were no significant differences in the density of microtubules or mitochondria (Fig. 3).

Figure 3.

Increased density of axonal neurofilaments in Gjb1-null femoral nerves. (A–F) Electron micrographs from femoral motor nerves of 4-month-old wild type (WT) (A, C, E) and Gjb1-null (B, D, F) mice. Low-magnification images (A, B) show that axons are normally myelinated in Gjb1-null nerve. Higher magnification images of longitudinal (C, D) and transverse (E, F) sections of normally myelinated axons illustrate the higher density of neurofilaments (NFs) in the Gjb1-null vs. the WT axons. At higher magnification of longitudinal (insets, C, D) and transverse sections (insets, E, F), NFs (open arrowheads) appear more tightly packed in Gjb1-null axons; microtubules (arrowheads) do not show clear differences in density. Areas devoid of NFs seen in the WT (arrows in C and E) are absent in the Gjb1-null (D, F). M = mitochondria. (G–I) Average density per μm² of NFs (G), microtubules (H) and mitochondria (I) obtained in at least 10 normally myelinated axons (6-to 8-μm diameter) from 2- and 4-month-old mice in each genotype (n = 3–5 mice per genotype and age group). The density of NFs is higher in Gjb1-null mice vs. WT starting at 2 months and is further increased at 4 months. The densities of microtubules and mitochondria are not significantly different at any age (*Indicates significant results).

Because dephosphorylation is associated with tighter packing of NFs (33–35), we immunoblotted sciatic nerve lysates for dephosphorylated (SMI32) and phosphorylated (SMI31) NF-H. At least 3 Gjb1-null and 3 WT littermates were analyzed at 2 and 4 months of age. Nerve lysates from single animals were immunoblotted at least 3 times and the signal intensity relative to that of GAPDH was quantified. Gjb1-null nerves had significantly higher SMI32 signals vs. their WT littermates starting at 2 months (189.9 ± 47.5% of WT at 2 months, p = 0.04; and 150.2 ± 11.8% at 4 months, p = 0.009), demonstrating increased dephosphorylation of NF-H, with a simultaneous reduction of the SMI31 signals at 4 months (90.8 ± 23.9% of WT at 2 months, p = 0.28; 73.9 ± 7.9% at 4 months, p = 0.014) (Fig. 4A–D). Gjb1-null nerves also had reduced phosphorylation of NF-medium (NF-M) (Supplemental Fig. 4A). In line with these results, total NF levels did not significantly differ between genotypes (Supplemental Fig. 4B), but the proportion of phosphorylated NFs was significantly lower at both ages (at 2 months 72.1 ± 4.1% of total NF was phosphorylated in WT mice, and 55.3 ± 4.2% in Gjb1-null littermates, p = 0.00033. At 4 months, 75.6 ± 4.1% of total NF was phosphorylated in WT mice and 60.5 ± 3.7% in Gjb1-null littermates, p = 0.015). Dynein and kinesin levels were not significantly changed (data not shown), whereas tubulin levels were minimally reduced in Gjb1-null nerves at 4 months (94.2 ± 2.2% of WT; p = 0.02). MBP levels were similar between genotypes even at 4 months, in keeping with minimal demyelination at this age.

Figure 4.

Early dephosphorylation of axonal neurofilaments in Gjb1-null nerves without demyelination. (A–D) Immunoblot analysis of cytoskeletal components in sciatic nerves from 3 pairs of 2-month-old (A, B) and 4 month-old (C, D) wild (WT) and Gjb1-null littermates. (A, C) Representative blots are shown for antibodies to phosphorylated (SMI31) and non-phosphorylated (SMI32) NF-heavy (NF-H; 200 kDa) and β-tubulin (55 kDa) and glyceraldehyde 3-phosphate dehydrogenase (GAPDH) (37 kDa, loading control) and myelin basic protein (MBP, 20 kDa). In all pairs, SMI32 levels are greater in Gjb1-null vs. WT littermates. At 2 months SMI31 levels are not different, but at 4 months, SMI31 levels appear reduced, along with minimal reduction of β-tubulin; MBP levels are unchanged. (B, D) Results of quantitative analysis (average band intensity values for cytoskeletal proteins from at least 3 pairs of WT and Gjb1-null mice each run at least 3 times in SDS-PAGE and electrotransferred onto nitrocellulose for immunolabeling, normalized for loading to GAPDH). Non-phosphorylated NF-H levels are significantly greater in Gjb1-null nerves while phosphorylated NF-H levels are decreased at 4 months. β-tubulin levels are mildly reduced in Gjb1-null nerves at 4 months. *Indicates significant results (p < 0.05, paired t-test). (E–H) Immunostaining of sciatic nerve teased fibers for SMI32 (green) and MBP (red); cell nuclei are stained with DAPI (blue). Axons from 2-month-old Gjb1-null mice (F) have more SMI32-immunoreactivity (IR) (green arrows) than those from WT littermates (E), with inhomogeneous pattern along the entire axons. At 4 months, SMI32-IR completely fills some axons (H) of Gjb1-null mice. The SMI32+ axons have normal-appearing, MBP+ myelin sheaths (red open arrowheads). Scale bar in H: 10 μm.

To corroborate these results we immunostained teased fibers from 2-, 4-, and 8-month-old Gjb1-null mice and their WT littermates for SMI32 and MBP. Compared to WT nerves, SMI32-IR was clearly increased in 2-month-old Gjb1-null nerves and appeared inhomogeneous and extended to several internodes of each axon. At 4 and 8 months of age, SMI32-IR was more intense, appearing to fill the entire internodes of many axons, especially the large ones, whereas thin axons showed lower SMI32-IR (Fig. 4E-H and data not shown). SMI32-IR fibers had normal-appearing MBP+ myelin sheaths at all ages, thereby demonstrating that these were not demyelinated.

Axonal Degeneration Occurs at Early Stages of Neuropathy in Gjb1-Null Mice

Because intra-axonal accumulation of APP is a marker of axonal degeneration (29), we next immunostained teased fibers from sciatic nerves of 2-, 4- and 8-month-old mice for APP. Gjb1-null mice had more APP+ axons than their WT littermates and the proportion of APP+ axons increased with age (Fig. 5). At 2 months,11.6 ± 5.2% in WT and 26.2 ± 2.0% of fibers in Gjb1-null were APP+ (p = 0.027, 2-tailed t-test for unpaired samples); at 4 months, 12.9 ± 6.9% in WT and 42.1 ± 11% in Gjb1-null were APP+ (p = 0.022); and at 8 months, 12.4 ± 5.5% in WT and 63.6 ± 8.0% in Gjb1-null were APP+ (p = 0.014). The pattern of MBP-IR surrounding the APP+ axons looked normal, indicating that myelination was normal. APP-IR was most pronounced in the paranodes and juxtaparanodes bordering the nodes of Ranvier, as shown by co-staining for the paranodal protein Caspr (Fig. 5 and data not shown).

Figure 5.

Amyloid precursor protein (APP)+ axons in Gjb1-null sciatic nerves. (A–D) Representative images of teased fibers from 4-month-old wild type (WT) (A, B) and Gjb1-null (C, D) mice, double labeled with antibodies to APP (red) and either myelin basic protein (MBP, green, A, C) or contactin-associated protein (Caspr, green, B, D). Many Gjb1-null axons (red open arrowheads), but not WT axons, are APP+; these are surrounded by normal-appearing MBP+ myelin sheaths (green open arrowheads) and have normal-appearing Caspr+ paranodes (green arrowheads). Scale bar: 10 μm. (E) Quantitative analysis of APP+ axons in 2-, 4-, and 8-month old mice (n = 3 or 4 in each genotype and age group). The percentile of APP+ axons increases with age and, at every age, is significantly higher in Gjb1-null vs. WT mice.

Impairment of Axonal Transport in Gap Junction-Deficient Nerves

The accumulation of APP, which undergoes fast anterograde transport (36), suggested that axonal transport might be impaired in Gjb1-null nerves. To investigate this we ligated 1 sciatic nerve of 4-month-old Gjb1-null mice and their age-matched WT littermates; 2 ligatures were placed 5 mm apart and the mice were killed after 3 hours (Fig. 6A). We immunoblotted protein extracts of individual nerve segments immediately proximal and distal to the ligatures and the corresponding segments of the contralateral/unlesioned nerve for several proteins associated with synaptic vesicles that undergo both anterograde and retrograde transport (37–39). Therefore, they normally accumulate following nerve ligation, and this accumulation is diminished in nerves with impaired fast axonal transport (40, 41). The levels of synaptotagmin, SV2, and synapsin in the segment distal to the ligature were lower in Gjb1-null mice compared to their WT littermates, indicating that retrograde axonal transport was reduced (Fig. 6). We found similar results for nerves that had been ligated for 6 hours (not shown). In addition, the steady state levels of the same proteins (Fig. 6D), as well as of the microtubule associated protein tau (42) (Supplemental Fig. 6E), were higher in unlesioned nerves from Gjb1-null mice, consistent with slower axonal transport. Finally, the retrograde motor dynein accumulated in the segments distal to the ligation and was depleted from the segments proximal to the ligation in WT but not in the Gjb1-null nerves, further supporting the fact that axonal transport is impaired in mice deficient for Cx32 (Supplemental Fig. 5). Likewise, the anterograde motor kinesin (as well as tau) accumulated in segments proximal to the ligation in WT but much less in Gjb1-null nerves (Supplemental Fig. 6A–D). Taken together, these findings indicate that both anterograde and retrograde axonal transport is impaired in Gjb1-null nerves.

Figure 6.

Early impairment of axonal transport in Gjb1-null mice. (A) Schematic of the sciatic nerve ligation experiment. One sciatic nerve was doubly ligated; 3 hours later, segments proximal (LigP) and distal (LigD) to the ligation and the corresponding segments of the contralateral/unligated nerve (Un) were collected. (B) Immunostaining at the distal ligation site (arrows) shows more synaptotagmin accumulation in the wild type (WT) than in the Gjb1-null nerve. Bar = 40 μm. (C) Representative immunoblots of lysates from ligated (LigP and LigD) and unligated nerves, from WT and Gjb1-null mice. Membranes were blotted with monoclonal antibodies to SV2 (95 kDa), synapsin (70 kDa) and synaptotagmin (65 kDa), glyceraldehyde 3-phosphate dehydrogenase (GAPDH, loading control) and myelin basic protein (MBP) (20 kDa). The levels of SV2, synapsin and synaptotagmin in LigD (vs. same region of unligated nerves) are higher in WT than in Gjb1-null nerves; their steady state levels in unligated nerves are higher in the Gjb1-null. GADPH and MBP levels are similar in WT and Gjb1-null mice. (D, E) Quantitative results from 3 independent experiments. (D) Bars represent average percentage increase in synaptic protein levels (normalized for GAPDH signals) in unligated nerves from Gjb1-null vs. WT mice. (E) Bars represent average percentage increase in the synaptic protein levels in LigD and LigP vs. the contralateral unligated nerves, after normalization for GAPDH signals. SV2, synapsin, and synaptotagmin levels are significantly higher in WT vs. the Gjb1-null LigD segments (1-tailed paired t-test).

DISCUSSION

Here, we provide the first evidence of axonal pathology in early stages of neuropathy in Gjb1-null mice, a model of CMT1X. We find increased packing density and dephosphorylation of axonal NFs, accumulation of APP and synaptic vesicle-associated proteins, and diminished retrograde axonal transport before significant demyelination has occurred. These findings highlight novel aspects of CMT1X pathogenesis and emphasize the importance of Cx32 GJs in mediating Schwann cell-axonalinteractions.

Early Axonal Cytoskeleton Changes in Gjb1-Null Mice

NFs are the major cytoskeletal proteins in axons and are heteropolymers composed of 3 polypeptide subunits named according to their relative molecular weights, NF–H, NF–M and NF-light (NF-L) (43, 44). In normal myelinated axons, the number of NFs correlates with the axonal size (45–47), and the lack of axonal NFs in quail, mice, or humans markedly reduces axonal calibers and conduction velocities (45–50). In addition to the number of NFs, the phosphorylation of the KSP (Lys-Ser-Pro) repeats in NF-H and NF-M increases their negative charge and thereby augments their lateral spacing and the formation of cross-bridges (34, 35, 51). In the PNS (52) and the CNS (51, 53), increased spacing of NFs may contribute to the axonal diameter. However, eliminating phosphorylation of either NF-H (54, 55) or NF-M KSP repeats (56) did not affect NF spacing and acquisition of normal axonal caliber, suggesting that NF-H and NF-M may have partially overlapping functions (57). Thus, it remains to be shown whether dephosphorylation of both NF-M and NF-H, as shown here in Gjb1-null mice, affects axon calibers.

How myelinating glia increase the number of axonal neurofilaments and their phosphorylation is not known. Both axonalkinases and phosphatases regulate NF-M and NF-H phosphorylation(58). Multiple signal transduction cascades (triggered by growth factors, Ca²⁺ influx, or integrins) and many different kinases (e.g. ERK1/2, Cdk5, p38 MAP kinase/SAPK) lead to phosphorylation (59–62). In the CNS, axonal ensheathment by oligodendrocytes appears to be sufficient to increase axonal diameter, but not to the same extent as myelination (63). In the PNS, myelin-associated glycoprotein (MAG) appears to play a crucial role. MAG is localized on the adaxonal membrane (and non-compact myelin) of myelinating Schwann cells (64) and interacts with lipids on apposing axonal membrane (65). PNS axons in Mag-null mice have relatively dephosphorylated and tightly packed axonal NF, which is associated with reduced axonal calibers (66). However, because it is normally localized in Gjb1-null mice (13), MAG is unlikely to mediate the effects of the loss of Cx32.

Altered Axonal Transport in Gjb1-Null Mice

Axonal transport is required for axonal growth, function and viability (67–69) and is accomplished by 2 sets of motor proteins: kinesins for anterograde transport and dynactin complex for retrograde transport. APP is axonally transported (70, 71) and accumulates in a variety of demyelinating and neurodegenerative diseases in both rodents (72–75) and humans (76–78). Similarly, the microtubule associated tau undergoes slow axonal transport utilizing fast transport motors and is also known to accumulate in neurodegeneration (42). Thus, accumulation of APP, tau, and different synaptic vesicle-related proteins in unlesioned Gjb1-null nerves as shown here suggests impaired axonal transport before demyelination, which is further demonstrated by the results of nerve ligations.

It is possible that the perturbations of NFs we found diminish axonal transport. Mitochondria associate directly with NF sidearms and this interaction is mediated by NF phosphorylation as well as mitochondrial membrane potential (79, 80). NF cross-bridging, which depends on phosphorylation (35), competes with kinesin-dependent association of NFs with microtubules (81). This may be a mechanism by which C-terminal NF phosphorylation contributes to slowing in axonal transport of NFs and axon stability since disruption of NF-H increases NF motility (82). Furthermore, cultured neurons from Nefl-null mice show abnormal mitochondrial distribution and motility, suggesting that the NF network regulates the trafficking of these organelles (83–84). Similarly, dephosphorylation of axonal NFs is associated with mitochondrial depletion in multiple sclerosis (78).

Axonal Pathology Occurs Independently of Demyelination

Closer spacing of NFs, less phosphorylation, and smaller axonal calibers are also seen in other demyelinating mutants (13, 19, 22, 85–87). However, the cytoskeletal changes and defects in axonal transport in Gjb1-null mice occur in axons that appear to be normally myelinated, indicating dissociation of these axonal alterations from demyelination. This confirms and extends the list of axonal abnormalities that may occur independently of demyelination in other myelin mutants. For example, Plp1-null mice have impairment of fast axonal transport, despite being well myelinated (73). Degeneration of axons surrounded by normal appearing myelin sheaths has been found in Plp1-null (72) and Cnp1-null mice (88). In contrast, the CNS axons in shiverer mice, which are dysmyelinated due to lack of MBP, show no axonal pathology (89).

How the loss of GJs composed of Cx32 leads to axonal alterations remains unclear. Cx32 GJs provide a direct pathway of communication between the adaxonal and abaxonal Schwann cell cytoplasm (3). Perhaps the loss of a signal that depends on these GJs initiates the axonal alterations. The most well-developed possibilities are Ca²⁺ and IP3. The IP3-receptor-3 is also localized in the paranodal areas of Schwann cells (90). During neural activity, intracellular Ca²⁺ rises through the IP3 signalling cascade and undergoes ryanodin-dependent release from ER specifically at areas of non-compact myelin (91). In vitro studies showed that GJs allow the diffusion of Ca²⁺-mobilizing second messengers across coupled cells (92) and intracellular calcium concentration regulates Cx32 hemichannel opening (93). It remains to be determined whether Cx32 GJs in non-compact myelin areas of myelinated fibers serve as conduits for the rapid radial spread of these Ca²⁺ and IP3 signals required for axonal integrity.