|Home | About | Journals | Submit | Contact Us | Français|
Mutations in myelin protein zero (MPZ) cause Charcot–Marie–Tooth disease type 1B. Many dominant MPZ mutations, including R98C, present as infantile onset dysmyelinating neuropathies. We have generated an R98C ‘knock-in’ mouse model of Charcot–Marie–Tooth type 1B, where a mutation encoding R98C was targeted to the mouse Mpz gene. Both heterozygous (R98C/+) and homozygous (R98C/R98C) mice develop weakness, abnormal nerve conduction velocities and morphologically abnormal myelin; R98C/R98C mice are more severely affected. MpzR98C is retained in the endoplasmic reticulum of Schwann cells and provokes a transitory, canonical unfolded protein response. Ablation of Chop, a mediator of the protein kinase RNA-like endoplasmic reticulum kinase unfolded protein response pathway restores compound muscle action potential amplitudes of R98C/+ mice but does not alter the reduced conduction velocities, reduced axonal diameters or clinical behaviour of these animals. R98C/R98C Schwann cells are developmentally arrested in the promyelinating stage, whereas development is delayed in R98C/+ mice. The proportion of cells expressing c-Jun, an inhibitor of myelination, is elevated in mutant nerves, whereas the proportion of cells expressing the promyelinating transcription factor Krox-20 is decreased, particularly in R98C/R98C mice. Our results provide a potential link between the accumulation of MpzR98C in the endoplasmic reticulum and a developmental delay in myelination. These mice provide a model by which we can begin to understand the early onset dysmyelination seen in patients with R98C and similar mutations.
Myelin protein zero (MPZ) is the major myelin protein expressed by Schwann cells and mediates myelin compaction in the PNS (Lemke and Axel, 1985; Shapiro et al., 1996). MPZ is a 248 amino acid transmembrane protein and a member of the immunoglobulin supergene family. It constitutes one immunoglobulin-like extracellular domain, one transmembrane domain, a basic cytoplasmic domain and a 29 amino acid signal peptide, which is cleaved in the endoplasmic reticulum (Lemke and Axel, 1985). MPZ, like other members of the immunoglobulin superfamily, is a homophilic adhesion molecule (Filbin et al., 1990). Crystallographic data suggest that MPZ forms a lattice of homotetramers that interact between opposing membranes to mediate myelin compaction (Shapiro et al., 1996). Mpz-deficient mice produce poorly compacted myelin sheaths (Giese et al., 1992). Interestingly, overexpression of MPZ in mice also disrupts myelination by inhibiting Schwann cell wrapping of axons, which is also consistent with an adhesive function for the protein (Yin et al., 2000). Taken together, these data demonstrate that MPZ plays an essential role in myelination, probably by holding together adjacent wraps of myelin membrane through MPZ-mediated homotypic interactions.
Mutations in the MPZ gene cause Charcot–Marie–Tooth disease type 1B (CMT1B) (Hayasaka et al., 1993). To date, >120 disease-causing MPZ mutations have been reported (http://www.molgen.ua.ac.be/CMTMutations/default.cfm). We have systematically examined the phenotypic presentation in CMT1B and most patients can be clustered into two phenotypic groups; one with onset of symptoms in infancy and a second with onset of symptoms in adulthood (Shy et al., 2004). Early-onset patients have severely slowed nerve conduction velocities typically <10m/s. In contrast, late-onset patients demonstrate axonal loss with nerve conduction velocities >30m/s.
The R98C mutation [also called R69C with a different numbering system that does not count the 29 amino acid signal peptide (Warner et al., 1996)] causes a severe, early-onset form of CMT1B. Patients do not begin walking until ~3-years of age, lose the ability to ambulate independently prior to the end of the second decade and have nerve conduction velocities <10m/s in the upper extremities (Shy et al., 2004; Bai et al., 2006). To investigate how this and similar acting mutations disrupt myelination, we have generated R98C heterozygous and homozygous ‘knock-in’ mice by inserting the R98C mutation in the Mpz locus. We then analysed the clinical, physiological, morphological and molecular features of these animals.
All experiments performed on mice were conducted in accordance with experimental protocols approved by the Institutional Animal Care and Use Committees of San Raffaele Scientific Institute and Wayne State University, and the Italian Ministry of Health. The mutation encoding MPZR98C was targeted to a single Mpz allele by homologous recombination. The mutation was introduced into Mpz exon 3 of a 129S2 genomic clone by site-directed mutagenesis and confirmed by sequence analysis. Fragments of this clone were ligated into a construct containing the neomycin resistance gene flanked by loxP sites (Nodari et al., 2008) such that neoR was placed in Mpz intron 3 (Fig. 1). The construct R98CneoLP and a control (WTneoLP) were electroporated into TBV2 (129S2 strain) embryonic stem cells as described (Nodari et al., 2008). After confirming homologous recombination by Southern blot analysis with probes recognizing sequence flanking either the long or the short arms of the construct, positive embryonic stem cell clones were injected into blastocysts of host wild-type animals to obtain chimeras. Chimeras were outcrossed to obtain germline transmission of the mutant or WTneoLP alleles. The neoR-cassette, used for the selection of positive embryonic stem cell clones, is flanked by two loxP sites (LP) and was excised by breeding with CMVCre animals (Fig. 1A) (Nodari et al., 2008). Genotype analysis was performed by multiplex PCR analysis for neoR-containing mice (standard conditions; primer sequences available on request) and simplex PCR analysis for R98C mutant or wild-type alleles on genomic DNA prepared from tail samples. The simplex primers MpzEx3F (5′-CGATGAGGTGGGGGCCTTCAA-3′) and MpzEx3R (5′-ATAGAGCGTGACCTGAGAGG-3′) generate a 169bp amplimer. After digestion with HhaI, the wild-type allele migrates at 84bp, whereas the R98C allele migrates at 169bp. For wild-type loxP-only, the simplex primers MpzInt3F (5′-TCAAAGAGGGTGTCAGGGAG-3′) and MpzInt3R (5′-GTGGCCCAGATTGGTCTTTA-3′) generate a 355 or 305bp amplimer for wild-type loxP or wild-type, respectively. Two independently transmitted lines from one embryonic stem cell clone were obtained and bred congenic on either the FVB/N or C57BL/6N backgrounds. There were obvious behavioural and morphological phenotypes in R98C homozygous mice in either background. One R98C line on the FVB/N background was analysed fully in this study. RNA was prepared from sciatic nerves and reverse transcribed and Mpz complementary DNA was sequenced to validate that the mutation was present as expected. The designation of the line is FVB/N.129S2-MpzTMR98CLWLF1, hereafter called R98C.
Chop-null mice (Zinszner et al., 1998) were maintained in the B6C3HF1 genetic background (Southwood et al., 2002). The genotypes were determined as described (Southwood et al., 2002). No neurological phenotype was identified in the Chop-null animals, consistent with previous publications (Southwood, 2002; Gow and Wrabetz 2009).
Mice were evaluated at 6 weeks, 3 months and for some experiments, at 6- or 9-months of age. Mice underwent three training trials on an IITC Life Science Roto-Rod (Series 8) with a ramp speed from 2 to 20rpm in 300s. A 1h rest was given after each trial and it was considered valid if the animals ran forward on the rod for at least 10s. The next day the latency to fall was recorded three times following the above-mentioned protocol for each time point and mouse. The average was used as the outcome value.
Mice aged 6–8 weeks, 3 months and 6 months were anaesthetized with 2.5% 2,2,2-tribromoethanol 250mg/kg body weight and placed on a heating pad (constant 37°C) to maintain constant body temperature throughout the electrophysiological study (Neuromax, XLTEK). Sciatic nerve motor nerve conduction velocity was recorded stimulating a distal (ankle) and a proximal (sciatic notch) site along the nerve with platinum monopolar needle electrodes. Paw/thigh compound muscle action potentials were obtained with an active electrode inserted into the bulk of the muscle in the middle of the paw or the thigh and a reference needle placed subcutaneously.
Samples were obtained from post-natal Day 2 up to 9-months of age. Sciatic nerve samples were fixed in 2.5% glutaraldehyde overnight and processed as previously described (Saporta et al., 2009). Tissue blocks were sectioned at 1-µm thickness and stained with toluidine blue for light microscopic examination. These semi-thin sections were used to image the whole cross-section of the sciatic nerve at ×60 and ×90 magnification. The whole cross-section was reconstructed from the overlapping images and analysed to quantify the number of axons per field (axonal density). The diameters of at least 1000 axons per animal were calculated from the areas of the axon. Area measurements and manual tagging were carried out using ImagePro Plus software (Media Cybernetics). G-ratio (axonal diameter/fibre diameter) analysis was performed on randomized photos of ultrathin sciatic nerve cross-sections imaged with a Zeiss EM900 electron microscope.
This technique has been described in previous studies (Bai et al., 2006). Nerves were either briefly fixed in 4% paraformaldehyde for 30min or freshly frozen in O.C.T. medium. For teased fibre immunohistochemistry, a portion of each nerve was teased into individual fibres on glass slides after a 30-min fixation in 4% paraformaldehyde. The slides were dried for 1h, reacted with primary antibodies the following day and kept at 4°C overnight followed by 2-h incubation with secondary antibodies. The stained slides were examined under a Leica or Nikon fluorescence microscope. For studies on cryostat sections, sciatic nerves were embedded in O.C.T. medium. Tissue blocks were cryosectioned at 10µm. The slides were dried overnight, incubated with primary antibodies at 4°C overnight and then incubated for 1.5h with secondary antibodies. The stained slides were examined under a Leica fluorescence or Zeiss confocal microscope. All primary antibodies used are listed in Supplementary Table 1. After washing in PBS, the samples were incubated with secondary antibody (donkey anti-mouse IgG conjugated with fluorescein isothiocyanate, 1:100, Jackson Immuno, no. 715-095-151; donkey anti-chicken IgG conjugated to Texas Red™, 1:100, Jackson Immuno no. 703-075-155; goat anti-rat IgG –NL637, 1:200, R&D, no. NL014; goat anti-mouse IgG2a biotin conjugate, 1:100 Southern Biotech Associates, no. 1080-08; Cy5-conjugated streptavidin, 1:1000 Jackson Immuno no. 016-170-084) overnight at 4°C. Nuclei were labelled with 4′,6-diamidino-2-phenylindole (DAPI) 1:500.
RNA prepared from mouse sciatic nerve was fractionated in 1% agarose/2.2M formaldehyde gel (10µg/lane), transferred to nylon membranes in 6× saline-sodium citrate (SSC) and UV cross-linked (Southwood et al., 2002). The following complementary DNAs were used as probes: a 1.5kb fragment of rat myelin basic protein, a 1.8kb fragment of rat MPZ, a 1.4 fragment of mouse proteolipid protein, an 800bp fragment of rat peripheral myelin protein 22, a full length complementary DNA (2.4kb) of rat myelin-associated glycoprotein, a 600bp fragment of mouse Chop (gift from David Ron) and a 763bp fragment of murine binding immunoglobulin protein (gift from David Ron). Hybridization followed standard protocols (Kamholz et al., 2000; Menichella et al., 2001).
Wild-type and knock-in mice were intraperitoneally injected with 100μg/g body weight of 150mM NaCl solution containing 10mg/ml bromodeoxyuridine (Sigma). One hour later, mice were sacrificed and sciatic nerves were dissected and snap frozen in liquid nitrogen. Cross-sections 10-μm thick were processed as previously described (Pennuto et al., 2008). Terminal deoxynucleotidyl transferase dUTP-biotin nick end labelling (TUNEL) analysis was performed as previously described (Feltri et al., 2002). For each genotype, three animals were sacrificed and for each animal three sciatic nerve sections were selected. A total of three fields were randomly selected from each nerve cross-section yielding a total of 27 fields per genotype for quantification of bromodeoxyuridine and TUNEL-positive nuclei.
Sciatic nerves were dissected, sealed in quartz capillary tubes in physiological saline (154mM NaCl, 5mM Tris buffer, pH 7.4) and analysed within 30min as described previously (Avila et al., 2005). Diffraction experiments used a fine-line source on a 3.0kW Rigaku X-ray generator operated at 40kV by 14–22 mA and a linear, position-sensitive detector (Molecular Metrology) (Avila et al., 2005). Patterns were recorded for 10min. Seven hours later another pattern was recorded for 2h to detect any swelling occurring under physiological conditions. Wild-type nerves showed no swelling under these conditions. Measurement of the positions of the reflections in the myelin diffraction pattern revealed the myelin period and quantitation of the intensities above background permitted calculation of membrane profiles. From the membrane profiles, we measured the widths of the extracellular and cytoplasmic appositions and the thickness (distance between the lipid polar head groups) of the membrane bilayer. These structural parameters are determined as the distances between the middles of the head-group layers across the intermembrane (extracellular and cytoplasmic) spaces and within a single bilayer. The disorder in the stacking of membranes in multilamellar myelin was determined by plotting the integral widths w2 as a function of h4, where the intercept on the ordinate axis is inversely related to the number of repeating units N (the coherent domain size), and the slope is proportional to the fluctuation in period (lattice or stacking disorder) (Inouye and Kirschner, 1989). The relative amount of multilamellar myelin among the samples was determined by measuring the total integrated intensity (M) above background (B; after excluding the small angle region of the pattern around the beam stop as well as the wide-angle region of the pattern) for samples having equivalent exposure times. We have shown previously that a scatter plot of the fraction of total scattered integrated intensity that is attributable to myelin [M/(M+B)] versus myelin period (d) permits distinction of different demyelinating phenotypes (Avila et al., 2005).
Total RNA was extracted using RNeasy® Micro Kit (50) (QIAGEN). Complementary DNA was made from the RNA with SuperScript® III First Strand Synthesis SuperMix for QRT-PCR (11752-050, Invitrogen); 100ng RNA/ea was used in 20µl reaction buffer. TaqMan® Gene Expression Master Mix (no. 4369016) and Power SYBR® Green PCR Master Mix (no. 4367659), both from Applied Biosystems were used in quantitative PCR analysis. Quantitative real-time PCR was carried out on an MJ Research Opticon (Bio-Rad) detection system. All samples were normalized to GADPH. Primers and probes are listed in Supplementary Table 2.
Frozen sciatic nerves dissected from wild-type and mutant animals were pulverized, dissolved in lysis buffer (RIPA Sigma R0278 and proteinase inhibitors), kept on ice for at least 30min and centrifuged at 14000rpm for 10min at 4°C. The protein content of the supernatant was determined using a bovine serum albumin standard curve. Equal amounts of the protein were loaded on 6–12% SDS-PAGE gels and electroblotted onto polyvinylidene difluoride membranes. The results were normalized to β-actin (1:10000) and to untreated wild-type samples. All westerns were repeated at least three times.
The following antibodies were used: chicken anti-MPZ (Novus Biologicals, MB100-1607, 1:2000) mouse anti-BiP/GRP18 (BD Transduction Laboratories, 610978, 1:500), rabbit anti-ATF6 (Alexis Biochemicals, 70B1413.1, 1:500), CHOP (gift from Alex Gow, B1259), mouse anti-PSAPK/JNK (Cell Signalling, T183/Y185, 81E11), rabbit anti-c-Jun (Santa Cruz, H79, sc1694, 1:500). Horseradish peroxidase-conjugated secondary antibodies (1:5000–20000 dilution; Sigma) were detected using ECL reagents (Bio-Rad) with autoradiography film (Kodak Scientific Imaging Film, Blue XB).
COS-7 cells were co-transfected as described (Pennuto et al., 2008) with MpzS63del, MpzR98C or MpzWt, each with the haemagglutinin epitope tag fused to the carboxy terminus (Tinelli et al., manuscript in preparation) and pED-mXBP1deltaC(un)-d2EGFP (Back et al., 2006). X-box binding protein 1 (Xbp1) splicing was assayed as the proportion of DAPI nuclei that also contained green fluorescence after correction for transfection efficiency (proportion of haemagglutinin-positive cells in each culture dish; Tinelli et al., manuscript in preparation).
All comparisons between the three genotypes of the MpzR98C mice and between MpzR98C mice and those crossed with Chop-null animals were evaluated by one-way ANOVA followed by Tukey’s post hoc test. Prism 4 (GraphPad) was used to perform the statistical analysis. Unless otherwise stated in the text, all data points were evaluated by at least three animals or six nerves per genotype.
In order to study the pathogenetic mechanisms in early onset CMT1B, we generated a knock-in mouse model by inserting the R98C mutation into Mpz by homologous recombination (Fig. 1A). After crossing with a mouse expressing ubiquitous Cre (CMVCre) one LoxP site remains in intron 3 (R98C mice). To control for potential effects of this remaining loxP, we also generated a control mouse carrying only the LoxP site (indicated as WTLP). To examine the relative expression of the mutant Mpz allele compared with the wild-type allele, sciatic nerve messenger RNA from heterozygous mice (R98C/B6; WTLP/B6 or 129S2/B6) was reverse transcribed and amplified by PCR with alphaP32 dCTP using primers that recognize Mpz complementary DNA and flank a DpnII polymorphism present only in the wild-type allele. This allows a semi-quantitative estimate of the abundance of wild-type versus mutant messages (Wrabetz et al., 2000). One representative PCR cycle (of three) from five independent reverse transcriptase PCR experiments is shown in Fig. 1B. Quantification shows (mean±SEM for n=5) that loxP augments slightly the ratio of mutant to wild-type allele (WTLP compared with control 129/B6F1 hybrid), and that R98C is increased slightly above loxP by 1.25-fold (P=0.05 by ANOVA).
Homozygous R98C/R98C mice had a slow, unsteady gait that was readily apparent by the time of weaning at ~21 days after birth. Homozygous knock-in mice also had a persistent tremor that was exacerbated during walking (Supplementary Fig. 1). Heterozygous R98C/+ had no consistent observable clinical abnormalities up to at least 1 year of age. However, both heterozygous and homozygous mice had abnormalities on tests of motor performance. R98C/R98C mice were unable to maintain their balance on the rotating rod, and R98C/+ mice performed significantly worse than wild-type mice (Fig. 1C).
Neurophysiological studies revealed abnormalities in both heterozygous and homozygous mice. At 6–8 weeks of age, wild-type mice had normal nerve conduction velocities of ~40m/s and R98C/+ mice had slowed nerve conduction velocities of ~15m/s, similar to what is seen in heterozygous patients with the R98C MPZ mutation. R98C/R98C mice had severely slowed nerve conduction velocities of ~4m/s (Fig. 1D). Evoked compound muscle action potential amplitudes were markedly reduced in R98C/R98C mice and reduced to a lesser extent in R98C/+ mice compared with wild-type animals (Fig. 1D). Results were similar at 3- and 6-months of age (data not shown).
At 6 weeks of age, sciatic nerves from wild-type mice had numerous large and small calibre myelinated axons with an average G-ratio of 0.67. Heterozygous mice, consistent with their slow nerve conduction velocities, demonstrated markedly thinner myelin and an increased average G-ratio of 0.77. Homozygous R98C nerves exhibit very thin and poorly compacted myelin when it is present at all (inserts in Fig. 2M–O) and because of this the calculation of G-ratios was not possible.
To determine the onset and progression of the above abnormalities we repeated the ultrastructural analysis in both younger and older animals. In post-natal Day 2 mice, our earliest time point, early myelin profiles were detected in wild-type and heterozygous knock-in mice. Virtually no myelin was present in post-natal Day 2 homozygous knock-in mice, suggesting that myelination was delayed even at this time in the R98C/R98C animals (Fig. 2A–C). Abnormalities in homozygous mice were comparable between 2- and 6-weeks old and 9-month old mice (Fig. 2); thus there was no obvious progression of myelin abnormalities after 2-weeks old.
Small diameter axons were less affected by the R98C mutation than large diameter axons. In wild-type nerves, myelin thickness ranges from 0.1 to 1.2µm and increases in proportion to axon diameter. For small diameter axons (<2.0µm), the distribution of myelin thickness in R98C/+ nerves was similar to that of wild-type nerves. However, myelin thickness was significantly reduced for large diameter axons, never getting thicker than 0.5µm (Fig. 3A). In addition, the number of myelinated axons was reduced in both R98C/+ and R98C/R98C mice compared with wild-type animals (Fig. 3C), and again large diameter axons were disproportionately affected, shifting the distribution towards smaller diameter axons (Fig. 3B).
To determine whether the mutation caused abnormalities in myelin period, membrane packing or regularity, we next performed X-ray diffraction analysis on sciatic nerve segments from each of the three genotypes (wild-type, R98C/+ and R98C/R98C). A decreased strength of scattering intensity from myelin was seen in R98C/+ and, to a greater extent, in R98C/R98C sciatic nerves, confirming the decreasing relative amounts of myelin in heterozygous (70%) and homozygous (90%) animals.
From the positions of the X-ray reflections, myelin periods were calculated for wild-type, R98C/+ and R98C/R98C to be: 177.0±0.4Å (n=8), 178.4±0.5Å (n=7) and 193.1±4.2Å (n=8), respectively. Compared to the wild-type period, the <1% increase in period for the R98C/+ and the 9% increase for the R98C/R98C were significant (P<0.0001) (Table 1). In contrast, optic nerve myelin in the CNS (in which MPZ is not expressed) showed no significant differences in period between wild-type and R98C/+ or R98C/R98C animals (Table 1). Calculation of membrane pair profiles for the diffracted intensities showed that the wider period for R98C/R98C was accounted for by a ~20Å swelling at the extracellular apposition (Table 1).
The extent of membrane packing irregularity was investigated by analysing the peak widths in the diffraction patterns (Avila et al., 2005). The wild-type nerve segments yielded a mean coherence length for myelin of 1657±334Å (n=8; corresponding to 9.4±1.9 repeats) and a 2.2% distortion in period (0.022±0.0014) The R98C/+ nerve segments (n=7) yielded a coherence length that was reduced by ~30% from the wild-type value (P<0.005) and a period distortion (Δ/d) that was ~25% greater than the wild-type value (P<0.0001), indicating more irregularity. The R98C/R98C nerve segments (n=8) yielded a coherence length that was reduced by ~60% of the wild-type value (P<0.0001) and a distortion (Δ/d) that was twice as much as the wild-type value (P<0.0001), indicating even more irregularity. R98C/+ showed a narrower period and substantially less myelin than heterozygous Mpz null mice (Mpz+/−) (Supplementary Table 3). Data demonstrating that changes in myelin amount and compaction occur early in the disease and then stabilize are also provided in Supplementary Table 3.
We identified an increased number of cell nuclei in heterozygous (33%) and, even more so in homozygous (~75%, compared with wild-type sciatic nerves) R98C mice (Fig. 4A). To quantify Schwann cell proliferation in the animals, we performed bromodeoxyuridine analysis of the mutant nerves. Bromodeoxyuridine labelling was markedly elevated in homozygous knock-in mice demonstrating increased proliferation compared with wild-type animals. Bromodeoxyuridine labelling was only marginally increased in heterozygous mice (Fig. 4B). To determine whether Schwann cells were turning over because of increased cell death we also performed TUNEL assays on the nerves. There was no increase in apoptosis of R98C/+ mice or R98C/R98C mice compared with controls (Fig. 4C).
To determine the effect of MPZR98C on the coordinated expression of myelin specific genes, we performed northern blots on the sciatic nerve. Mpz, Mbp, Pmp-22 and Mag messenger RNA levels were reduced in heterozygous and to a greater extent in homozygous knock-in mice (Fig. 4D). Mpz protein expression was also reduced (Fig. 4E). These results are distinct from those of Mpz-null mice in which messenger RNA levels of these genes are similar in wild-type and heterozygous-null mice and dys-coordinately regulated in homozygous-null animals (Giese et al., 1992; Xu et al., 2000). Combined with the clinical, morphological studies and X-ray diffraction data shown above, these results suggest that the R98C mutation causes a toxic gain of function in myelinating Schwann cells that prevents them from progressing past an immature myelination stage when the mutation is homozygous and inhibits myelination when the mutation is present in a heterozygous state.
Promyelinating Schwann cells that have established a 1:1 relationship with an axon begin myelination by decreasing the expression of inhibitory transcription factors such as c-Jun and increasing the expression of activating transcription factors such as Krox-20 (EGR2) and Oct6/SCIP (Jessen and Mirsky, 2008; Woodhoo and Sommer, 2008). To investigate the developmental state of R98C Schwann cells we performed immunohistochemistry on teased fibres prepared from wild-type and mutant nerves at post-natal Day 13, near the peak of myelination. We found increased numbers of c-Jun (Fig. 5A–C) and decreased numbers of Krox-20 immunoreactive nuclei (Fig. 5D–F) in heterozygous and, to a greater extent, homozygous nerves. Occasional c-Jun positive Schwann cells were associated with MBP-positive internodes or hemi-nodes suggesting that these internodes were transitioning between a myelinating and pro-myelinating phenotype [Fig. 5B(1) and C(1)]. Most Krox-20 immunoreactive cells also stained positive for MBP. To quantitate these results we labelled adjacent cross-sections of post-natal Day 13 sciatic nerve with antibodies to c-Jun and Krox-20 and counted the percentage of immunoreactive nuclei. The percentage of c-Jun positive nuclei was progressively higher in R98C/+ and R98C/R98C mice (Fig. 5G). The percentage of Krox-20 labelled nuclei was similar between wild-type and R98C/+ nerves but significantly reduced in R98C/R98C nerves (Fig. 5H).
The distinct morphological, X-ray diffraction and gene expression findings between R98C mice and the Mpz knockout mice indicate that a toxic gain of function may be an important part of the pathophysiology of MpzR98C mutants. In vitro (Khajavi et al., 2005; Grandis et al., 2008) and in vivo (Pennuto et al., 2008) studies suggest that some mutant MPZ proteins are not transported correctly to the myelin sheath and may be retained in the endoplasmic reticulum. To determine whether this occurred with MpzR98C we first localized Mpz expression in mutant sciatic nerve by teased fibre immunohistochemistry. Mpz was localized in a perinuclear distribution in addition to its normal presence in myelin in R98C/+ nerve. In R98C/R98C nerves, Mpz was only detectable in a perinuclear distribution and was not detectable along axons (Fig. 6A). To further characterize the localization of MpzR98C, we performed immunohistochemistry with antibodies to Mpz and the endoplasmic reticulum chaperone BiP in cross-sections of sciatic nerves from wild-type and homozygous R98C mice. In wild-type nerves, MPZ localized to the myelin sheath and BiP stained the endoplasmic reticulum. In homozygous animals, however, both proteins co-localized to the endoplasmic reticulum, confirming that the expressed MpzR98C was retained in this organelle (Fig. 6B). Furthermore, immunohistochemistry and northern blot analysis demonstrated an increased expression of BiP in homozygous animals compared to wild-type (a marginal increase was also observed in heterozygous animals), suggesting that endoplasmic reticulum stress was present in R98C homozygotes, with consequent overexpression of endoplasmic reticulum chaperones (Fig. 6B).
In the setting of endoplasmic reticulum stress, a group of signalling pathways termed the unfolded protein response (UPR) is activated. UPR activation aims to reduce the load of unfolded proteins through upregulation of chaperones and global attenuation of protein synthesis, and can also lead to programmed cell death when endoplasmic reticulum stress is overwhelming. A pathogenic UPR was found in mice carrying a demyelinating CMT1B mutation, MpzS63del (Wrabetz et al., 2006; Pennuto et al., 2008). The UPR is mediated initially by three molecules located in the endoplasmic reticulum membrane: IRE1, ATF6 and PERK. Three parameters are usually used to detect UPR activation: XBP1 splicing as an indicator of IRE1 pathway activation, ATF6 cleavage and increase in the levels of the transcription factor CHOP and its translocation to the nucleus, as an indicator of PERK pathway activation (Fig. 6E). Around the peak of myelination (post-natal Day 13) real-time PCR revealed a trend towards increased Xbp1 splicing in homozygous R98C mice sciatic nerves. However, this did not reach statistical significance (P=0.26). XBP1 splicing was also detected in a cell culture reporter system using COS-7 cells transfected with a construct expressing a messenger RNA encoding XBP1 fused out of frame to GFP. When XBP1 is spliced, the GFP open reading frame is moved into the correct translation frame and GFP is expressed in cell nuclei (Back et al., 2006). Transfection of COS-7 cells with MpzR98C led to a greater increase in nuclear GFP than that seen with MPZwt (Fig. 6C and Supplementary Fig. 2). We also noted increased expression of JNK, a potential activator of c-Jun transcription (Angel et al., 1988), in homozygous R98C nerves (Fig. 7F). Increased cleaved ATF6 was detected in both heterozygous and homozygous sciatic nerves by western blot analysis (not shown). Finally, an increase in the levels of Chop, a downstream mediator of the PERK pathway, was demonstrated by northern blot analysis in sciatic nerve of R98C mice at post-natal Day 13 (Fig. 6D). An increased number of Chop-positive nuclei was also observed in post-natal Day 13 R98C homozygous sciatic nerves, confirming that Chop translocated to the nucleus where it exerts its effects (Fig. 6D). Interestingly, although CHOP activation has been previously linked to increased apoptosis in other cell models, we did not detect an increase in programmed cell death in our R98C mice by TUNEL assay. Therefore, CHOP activation did not lead to the apoptosis of R98C expressing Schwann cells. Taken together, these data indicate that a canonical UPR is active in R98C nerves. Interestingly, although increased levels of Chop were detected by real-time-PCR and western blot analysis in later time points (6 weeks and 3 months), there was no increase in CHOP-positive nuclei in homozygous mice outside the post-natal Day 13 time point, suggesting that UPR activation was limited to the normal period of active myelination and was coincident with the period when we observed most of the changes in Schwann cell development in R98C nerves. A schematic representation of the potential effects of the UPR on R98C nerves is illustrated in Fig. 6E.
MpzS63del also causes CMT1B although the morphological phenotype of S63del and R98C nerves is dissimilar. S63del nerves manifest demyelination, and less evidence of the hypomyelination found in R98C nerves. In addition, myelin debris, onion bulbs and segmental demyelination are detectable in S63del samples that are not present in R98C nerves. UPR activation has also been demonstrated in S63del Mpz mice (Wrabetz et al., 2006). Deleting Chop led to clinical improvement, improved F-wave latencies and reduced numbers of demyelinated axons at 6-months of age (Pennuto et al., 2008). Of note, hypomyelination was not improved in S63del/Chop-null mice. To determine whether we would obtain similar benefits by ablating Chop, we crossed the R98C mutant mice into the Chop-null background (Fig. 7A). However, we observed no improvement in Rotarod function, no changes in nerve conduction velocities or F-wave latencies and no morphological changes by electron microscopy in myelin in R98C mutant mice in a Chop-null background (Fig. 7B and C). We did find that compound muscle action potential amplitudes increased towards normal in R98C/+/Chop−/− mice compared with R98C/+/Chop+/+ animals (Fig. 7D). However, there was no change in the size or distribution of myelinated fibres in mice without Chop (Fig. 7E) compared with mice with Chop (Fig. 3B).
We also compared Jnk in R98C mice with and without Chop by real-time PCR. Jnk expression was increased in both R98C/+/Chop−/− and R98C/R98C/Chop−/− mice similar to what we observed in the corresponding R98C mice with normal levels of Chop (Fig. 7F). To examine PERK arm activation in the absence of Chop, we measured eIF2α phosphorylation levels by western blot (Back et al., 2009). eIF2α phosphorylation levels were increased in R98C/Chop-null mice as were levels of ATF6 and cleaved ATF6 (Fig. 7G). Finally, c-Jun levels were increased in R98C mutant mice without Chop as they were in mice expressing Chop (Fig. 7G). Taken together these results demonstrate that eliminating Chop as a downstream transcription factor from the PERK arm of the UPR has minimal effects on the R98C knock-in mouse model of CMT1B.
We have generated a mouse knock-in model of the human inherited neuropathy CMT1B that is caused by an R98C mutation in MPZ. This mutation causes a severe hypomyelinating neuropathy in which the onset of ambulation is delayed and in which impairment is severe prior to adulthood (Gabreels-Festen et al., 1996; Shy et al., 2004; Bai et al., 2006). The neuropathy is also severe in mice. R98C/+ mice are weak at an early age, have slow nerve conduction velocities and have reduced myelin in their PNS. R98C/R98C mice are extremely disabled in infancy, have minimal PNS myelin and have nerve conduction velocities of ~5m/s. It is interesting to compare nerve morphology from MpzR98C mice and MPZR98C patients. All reports of R98C sural nerve biopsies identified reduced numbers of myelinated axons and regions of segmental demyelination (Gabreels-Festen et al., 1996; Kirschner et al., 1996; Komiyama et al., 1997; Hattori et al., 2003; Bai et al., 2006), both of which we also observed in R98C/+ mice. However, there were differences between the mice and patients. We did not observe onion bulbs in the mice, whereas these were frequent in the patient we had observed (Bai et al., 2006). Similarly, we did not observe areas of non-compact myelin in R98C/+ mice whereas Gabreels-Festen et al. (1996) and Kirschner et al. (1996) detected uncompacted myelin in 23–68% of myelinated fibres, beginning at the major dense line in a sural nerve biopsy from a 14-month old. However, we did not detect non-compact myelin in either of the two biopsies performed on our adult patient with the R98C mutation (Bai et al., 2006). It may be that some morphological features are variable with patients having the same mutations, that certain findings differ depending on the specific nerve analysed or that there are differences between mice and humans related to onion bulb formation and thresholds for myelin decompaction.
Mutant MpzR98C is retained within the endoplasmic reticulum in mouse Schwann cells. In heterozygous animals there is Mpz in myelin as well as in the endoplasmic reticulum. We presume that the Mpz in myelin is predominantly from the wild-type allele because minimal Mpz and myelin are detectable around ensheathed axons from R98C/R98C mice. Taken together, these results suggest that MPZR98C is unable to be incorporated into compact myelin and that at least some normal MPZ remains able to be incorporated into myelin in the presence of one allele expressing MPZR98C.
Despite the paucity of Mpz in myelin, our data demonstrate that the R98C/+ neuropathy is not caused simply by a lack of Mpz. Heterozygous Mpz-null mice, in which there is haplo-insufficiency of Mpz, have no clinical phenotype or abnormalities of nerve conduction velocities until they are 4–6 months of age (Martini et al., 1995; Shy et al., 1997). Morphologically, Mpz+/− peripheral nerves also appear normal until several months of age when they develop asymmetric demyelination suggestive of chronic inflammatory demyelinating polyneuropathy (Shy et al., 1997) or immune-mediated neuropathies (Martini, 1999). Moreover, the coordinate programme of myelin gene expression during development is similar in wild-type and Mpz+/− mice (Shy et al., 1997; Menichella et al., 2001). R98C/+ mice differ in all of these aspects. They are abnormal clinically, have slow nerve conduction velocities, abnormal myelin and have abnormal myelin gene expression—all before 6-weeks of age.
Although homozygous Mpz null mice (Mpz−/−) also have severe neuropathy in infancy with nerve conduction velocities of ~5m/s, there are also significant differences between these animals and R98C/R98C knock-in mice. Mpz−/− mice have axons ensheathed by multiple layers of loosely compacted myelin (Giese et al., 1992) with a disorganized programme of myelin gene expression such that some genes like Mag and Plp are upregulated with others like Pmp22 downregulated (Giese et al., 1992; Xu et al., 2000). In contrast, R98C/R98C nerves have scarcely any wraps ensheathing axons and their programme of myelin gene expression is coordinately decreased compared to normal mice of the same age. Differences between Mpz knock out and R98C knock-in nerves were also detected by X-ray diffraction. Compared with the R98C/+, we found that sciatic nerve segments from the Mpz+/− nerves showed a greater relative amount of myelin and a wider period than R98C/+ nerves. Finally, R98Cneo/R98Cneo nerves (in which loxP-neoR-loxP has not been excised) express half of Mpz messenger RNA compared to R98C/R98C nerves, which would predict worse hypomyelination if MR98C acted through loss of function. Instead, R98Cneo/R98Cneo nerves manifest much milder hypomyelination, suggesting that MR98C provokes a dose-dependent toxic gain of function (C.F. and L.W., unpublished). Taken together, all of these data demonstrate that the neuropathy in R98C mutant mice occurs by abnormalities caused by the mutant Mpz rather than by the loss of Mpz itself. These differences also suggest important issues for therapy with CMT1B mutations that are similar to R98C. Since mutant R98C causes toxicity without being transported to myelin, therapeutic approaches might best be focused on detoxifying the mutant protein rather than on increasing levels of wild type Mpz in myelin.
The abnormal ‘toxic’ gain of function caused by MR98C either arrests Schwann cell development in a promyelinating stage or slows their differentiation beyond this stage in a dose-dependent fashion. Myelinating Schwann cells in mice normally establish a 1:1 relationship with axons at the time of birth (Trapp et al., 1988). R98C/R98C Schwann cells have established this 1:1 relationship at birth, but never synthesize more than rudimentary myelin even by 6-months of age. Thus, electron micrographs at post-natal Day 2 resemble those at 6-months of age in the homozygous mice. R98C/+ Schwann cells do myelinate as the animals age but the myelin never reaches its normal thickness, particularly for large calibre axons and myelin gene expression always remains reduced compared with wild-type levels.
We do not completely understand how the MpzR98C arrests or slows myelination but it most likely involves regulation of the transcription factors c-Jun and Krox-20 as the proportion of cells expressing these is abnormally elevated and reduced, respectively, in R98C mice. c-Jun expression actively inhibits myelination when present in the nucleus of premyelinating Schwann cells (Parkinson et al., 2008). Additionally, Schwann cells with sustained heterologous expression of c-Jun are unable to myelinate in dorsal root ganglia co-cultures and in such cells Krox-20 expression is suppressed (Parkinson et al., 2008). Conversely, Krox-20 has been shown to inhibit c-Jun expression in Schwann cells. In cultured Schwann cells, enforced expression of Krox-20 is sufficient to decrease c-Jun expression (Parkinson et al., 2008). In R98C mice, c-Jun expression remains high in R98C/R98C Schwann cells, even at 6-months of age, suggesting that many Schwann cells are actively prevented from entering a myelinating state by their expression of c-Jun. c-Jun is also detected in R98C/+ Schwann cell nuclei but not in wild-type nuclei.
Schwann cells ensheathing axons in R98C mice can be separated into two groups based on c-Jun and MBP expression: c-Jun+/MBP− and c-Jun−/MBP+. Likewise, we can identify two groups of Schwann cells based on Krox-20 and MBP expression: Krox-20−/MBP− and Krox-20+/MBP+. We presume that those cells that express MBP (c-Jun− and Krox-20+) have committed to myelination, while those that don’t express MBP (c-Jun+ and Krox-20−) are in a premyelinating state. A recent in vitro study showed that Schwann cells can undergo multiple transitions between de-differentiated (premyelinating) and differentiated (myelinating) states without cell cycle re-entry. De-differentiation to a premyelinating state required JNK activity and the upregulation of c-Jun expression (Monje et al., 2010). In R98C mice we occasionally observe c-Jun+/MBP+ Schwann cells ensheathing an axon. This raises the intriguing possibility that these represent a transition between myelinating and premyelinating Schwann cells. We hypothesize that UPR activation in myelinating cells could drive this transition.
We have strong evidence for a canonical response of the UPR in nerves of R98C mice. One possibility is that the IRE1 arm of the UPR alters the regulation of c-Jun and Krox20. Endoplasmic reticulum stress is known to induce binding of TRAF2 to the cytoplasmic domain of IRE1 and in combination with signalling mediated by ASK1 to activate JNK (Nishitoh et al., 1998). JNK expression is increased in R98C Schwann cells. Activated JNK phosphorylates c-Jun and can stimulate and maintain its expression (Jessen and Mirsky, 2008). Induction of the UPR could thus promote the increased expression of c-Jun in R98C Schwann cells and drive their de-differentiation. Additional crossing of R98C or R98C-Chop null mice with Ire1-null animals could test the hypothesis. We would predict that less activation of the UPR would enable increased myelination to occur in these mice.
Activation of the UPR has also been detected in S63del mice. In S63del animals, ablation of Chop from the Perk arm of the UPR ameliorated the clinical phenotype, decreased F-wave latencies and reduced the numbers of demyelinated fibres in sciatic nerves (Pennuto et al., 2008). Conversely, ablation of Chop did not significantly alter the phenotype of the R98C mice. In patients, the R98C MPZ mutation causes a severe phenotype in which they do not walk independently until ~3 years of age and typically cannot walk independently in adulthood (Shy et al., 2004; Bai et al., 2006). In contrast, the S63del phenotype is milder with most patients walking by 1 year of age and ambulating independently as adults (Miller et al., in press). Myelination is developmentally delayed in R98C patients (congenital hypomyelination) where it is able to occur relatively normally in S63del patients. In S63del, demyelination at later stages appears to be the predominant mechanism causing neuropathy. Thus, endoplasmic reticulum stress and UPR activation are not limited to mutations that cause severe, early onset phenotypes. Rather, our results suggest that patients with CMT1B with various levels of clinical impairment may have UPR activation; however, the mechanisms through which the UPR affects myelination are likely to be distinct. Identifying which CMT mutations activate the UPR and the mechanisms by which it alters myelination will be likely to have therapeutic importance as treatments emerge that ameliorate endoplasmic reticulum stress and UPR activation.
NINDS (R01 NS41319A to M.E.S., R01 NS055256 to L.W. and R01 NS43783 to A.G.), Dr Saporta is currently a fellow of the INC RDCRC supported by the NINDS and ORD (U54NS065712-01).
Supplementary material is available at Brain online.
The authors thank Drs Sung Hoon Back and Randall Kaufman at the University of Michigan for the pED-mXBP1deltaC(un)-d2EGFP construct and Adrienne Luoma for assistance in the diffraction experiments and analysis.