Abstract
Charcot-Marie-Tooth disease type 4C is the most common recessively inherited demyelinating neuropathy that results from loss of function mutations in the SH3TC2 gene. Sh3tc2−/− mice represent a well characterized disease model developing early onset progressive peripheral neuropathy with hypo- and demyelination, slowing of nerve conduction velocities and disturbed nodal architecture. The aim of this project was to develop a gene replacement therapy for treating Charcot-Marie-Tooth disease type 4C to rescue the phenotype of the Sh3tc2−/− mouse model. We generated a lentiviral vector LV-Mpz.SH3TC2.myc to drive expression of the human SH3TC2 cDNA under the control of the Mpz promoter specifically in myelinating Schwann cells. The vector was delivered into 3-week-old Sh3tc2−/− mice by lumbar intrathecal injection and gene expression was assessed 4–8 weeks after injection. Immunofluorescence analysis showed presence of myc-tagged human SH3TC2 in sciatic nerves and lumbar roots in the perinuclear cytoplasm of a subset of Schwann cells, in a dotted pattern co-localizing with physiologically interacting protein Rab11. Quantitative PCR analysis confirmed SH3TC2 mRNA expression in different peripheral nervous system tissues. A treatment trial was initiated in 3 weeks old randomized Sh3tc2−/− littermate mice which received either the full or mock (LV-Mpz.Egfp) vector. Behavioural analysis 8 weeks after injection showed improved motor performance in rotarod and foot grip tests in treated Sh3tc2−/− mice compared to mock vector-treated animals. Moreover, motor nerve conduction velocities were increased in treated Sh3tc2−/− mice. On a structural level, morphological analysis revealed significant improvement in g-ratios, myelin thickness, and ratios of demyelinated fibres in lumbar roots and sciatic nerves of treated Sh3tc2−/− mice. Finally, treated mice also showed improved nodal molecular architecture and reduction of blood neurofilament light levels, a clinically relevant biomarker for axonal injury/degeneration. This study provides a proof of principle for viral gene replacement therapy targeted to Schwann cells to treat Charcot-Marie-Tooth disease type 4C and potentially other similar demyelinating inherited neuropathies.
Introduction
Charcot-Marie-Tooth disease type 4 (CMT4) includes several recessively inherited demyelinating neuropathy forms caused by mutations in different genes, which lead to loss of function of the encoded proteins. CMT4C appears to be the most prevalent among this overall rare group of neuropathies, accounting for almost half of all CMT4 cases (Fridman et al., 2015). Patients with CMT4C usually present in the first decade of life with delayed walking, progressive distal muscle atrophy and weakness, areflexia and sensory loss. Almost all patients develop foot deformities and scoliosis, often requiring surgery (Kessali et al., 1997; Gabreels-Festen et al., 1999; Azzedine et al., 2006). Cranial nerve involvement with severe hearing loss, slow pupillary light reflexes, and lingual fasciculations are common with phenotypic variations within families (Gooding et al., 2005; Colomer et al., 2006; Varley et al., 2015). A combination of proprioceptive loss and vestibular neuropathy may cause disabling imbalance early in disease evolution (Perez-Garrigues et al., 2014). Electrophysiological studies in CMT4C patients confirm the demyelinating process with mean median motor nerve conduction velocity (MNCV) of 22.6 m/s. Nerve biopsies were characterized by an increase of basement membranes around myelinated, demyelinated, and unmyelinated axons, relatively few onion bulbs, and, most typically, large cytoplasmic extensions of Schwann cells (Kessali et al., 1997; Gabreels-Festen et al., 1999; Senderek et al., 2003).
Linkage analysis studies and homozygosity mapping (LeGuern et al., 1996) identified the CMT4C disease locus on chromosome 5q32, with subsequent discovery of several mutations in the SH3TC2/KIAA1985 gene (Senderek et al., 2003). At least 28 different SH3TC2 mutations have been described to date, mostly truncating but also missense, with higher frequency among certain ethnic groups (Lassuthova et al., 2011) likely due to founder effects (Gooding et al., 2005). The gene covers 62 kb of genomic sequence and consists of at least 18 variably spliced exons. SH3TC2 encodes a protein of 1288 amino acids containing two Src homology 3 (SH3) and 10 tetratricopeptide repeat (TPR) domains sharing no overall significant similarity to any other human protein with known function. The presence of SH3 and TPR domains suggests that SH3TC2 could act as a scaffold protein (Senderek et al., 2003). SH3TC2 is well conserved among vertebrate species, whereas no non-vertebrate orthologues were identified. SH3TC2 is present in several components of the endocytic pathway including early and late endosomes, and clathrin-coated vesicles close to the trans-Golgi network and in the plasma membrane. This localization is altered in CMT4C mutants examined in vitro (Lupo et al., 2009).
A mouse model of CMT4C has been generated and studied in detail by replacing exon 1 of the Sh3tc2 gene with an enhanced GFP (eGFP)-Neo cassette (Arnaud et al., 2009). Sh3tc2-null mice developed an early onset progressive peripheral neuropathy with decreased motor and sensory nerve conduction velocity and hypomyelination. Onset of myelin pathology in sciatic nerves of Sh3tc2−/− mice was somewhere between postnatal Days 0 and 5, since at 5 days of age there was already a difference between the mutants and controls (Arnaud et al., 2009; Gouttenoire et al., 2013). This phenotype was progressive as demonstrated by increasing myelin pathology and more difference in nerve conduction velocities in mutant compared to control mice at 1 year of age as compared to 1–2 months of age (Arnaud et al., 2009).
Murine SH3TC2 is specifically expressed in Schwann cells and localized to the plasma membrane and to the perinuclear endocytic recycling compartment, suggesting a possible function in myelination and/or in regions of axoglial interactions. Ultrastructural analysis of myelin in the peripheral nerve of mutant mice showed abnormal organization of the node of Ranvier, a phenotype that was confirmed in CMT4C patient nerve biopsies. These findings suggested a role for the SH3TC2 gene product in myelination and in the integrity of the node of Ranvier (Arnaud et al., 2009). SH3TC2 interacts with ERBB2 and plays a role in the regulation of ERBB2 intracellular trafficking from the plasma membrane upon NRG1 activation while dysregulated NRG1/ERBB signalling was found in SH3TC2-deficient Schwann cells that may account for the disturbed axonal size sensing and the hypomyelination present in CMT4C patients (Gouttenoire et al., 2013). Additionally, SH3TC2 was found to be an effector of the small GTPase Rab11, a key regulator of recycling endosome functions. Neuropathy-causing missense mutations in SH3TC2 disrupted this interaction (Stendel et al., 2010), providing a likely mechanism for the progressive nature of CMT4C neuropathy.
In summary, both the clinical phenotype and molecular basis of CMT4C suggest loss of function of SH3TC2, which results in early onset and slowly progressive hypomyelination in the peripheral nerves. Sh3tc2−/− mice recapitulate all major aspects of the disease. We have therefore developed a gene replacement strategy for CMT4C using a viral vector delivered intrathecally for Schwann cell-specific expression of human SH3TC2 (hSH3TC2). Treatment of Sh3tc2−/− mice at the age of 3 weeks resulted in significant improvement of several pathological features indicating that this approach could be further developed to treat CMT4C patients.
Materials and methods
Cloning of lentiviral vectors
SH3TC2 cDNA was cloned into the pcDNA3 vector under the CMV promoter (gift from Dr Vincenzo Lupo, University of Valencia). We added the myc-tag downstream of SH3TC2 by PCR amplification (Supplementary material). Correct localization of SH3TC2.myc tag was confirmed by cell transfection into HeLa cells (Supplementary Fig. 1). The lentiviral vector backbone originating from pCCLsin.PPT.hPGK.GFP.pre, in which hPGK promoter was replaced by the myelin-specific rat myelin protein zero (Mpz) promoter (Sargiannidou et al., 2015; Kagiava et al., 2016) was used to clone the SH3TC2.myc construct. The Gibson Assembly master mix (NEB) was used to ligate the two fragments together. The full vector (LV-Mpz.SH3TC2.myc) was confirmed by direct sequencing analysis (Fig. 1A). The LV-Mpz.EGFP vector, already cloned in our lab, was used as a mock vector (Fig. 1B).
Expression analysis of LV-Mpz.SH3TC2.myc vector. Map of the LV-Mpz.SH3TC2.myc (A) and LV-Mpz.Egfp (B) transfer plasmids used for the full and mock lentiviral vector production, respectively. Longitudinal sections of lumbar roots (C and D) and sciatic nerves (E and F) from Sh3tc2−/− mice 4 weeks following intrathecal injection with the full or mock vector, as indicated, were stained with anti-myc (green) and anti-hSH3TC2 (red) antibodies. Sh3tc2−/− mice injected with full vector show hSH3TC2 expression in the perinuclear cytoplasm of a subset of Schwann cells co-localizing with myc immunoreactivity in lumbar roots (D) and sciatic nerve teased fibres (F). Human SH3TC2 expression was absent from mock vector-injected tissues (C and E). Higher magnification images and separate channels of expressing Schwann cells are shown. Cell nuclei were stained with DAPI (blue, asterisks). Scale bars = 20 μm. (G) Counts of myc-positive cells in sciatic nerves and lumbar roots of Sh3tc2−/− mice 4 weeks following intrathecal injection of full vector were on average 38 ± 11.11% in lumbar roots (n = 3) and 17 ± 2.88% in sciatic nerves (n = 5 mice). (H) Vector copy numbers following intrathecal injection obtained from n = 4 Sh3tc2−/− mice 4 weeks after injection were on average 0.65 ± 0.21 in lumbar roots and 0.25 ± 0.15 in sciatic nerves. (I) Quantitative real-time PCR analysis of hSH3TC2 expression in lumbar roots and sciatic nerves following intrathecal injection with full vector (n = 3 mice). Fold expression represents relative human SH3TC2 mRNA expression levels of full vector-injected compared to non-injected Sh3tc2−/− mice used as a baseline control. A human nerve sample h1355N (sural nerve biopsy) served as positive control. All samples were normalized to endogenous control Gapdh.
Vector production
Lentiviral vectors were produced by transient co-transfection of HEK 293T cells with transfer plasmid and helper plasmids using the calcium phosphate co-precipitation method as described previously (Sargiannidou et al., 2015) (Supplementary material).
CMT4C mouse model
The generation and characterization of the Sh3tc2−/− mouse model has been described previously (Arnaud et al., 2009). This line was kept under specific pathogen free (SPF), standard controlled conditions of temperature (21–23°C), humidity, air exchange and light cycle (12/12 h light/dark) and provided with standardized mouse diet and drinking water ad libitum. All experimental procedures were conducted in accordance with animal care protocols approved by the Cyprus Government’s Chief Veterinary Officer (project license CY/EXP/PR.L3/2017) according to national law and European guidelines (EC Directive 86/609/EEC).
Human sural nerve biopsy material
As a positive control for hSH3TC2 expression derived from the viral vector, we used anonymized material from a diagnostic sural nerve biopsy tissue following informed consent by the patient under the Protocol ‘Translational Research-Neurogenetics Department’ in accordance with the Declaration of Helsinki, and approved by the National Bioethics Committee of Cyprus (EEBK/EP/2013/28).
Intraneural and intrathecal vector delivery
Vector delivery was performed under anaesthesia either by direct intraneural injection (10 μl) into surgically exposed mid-sciatic nerves of 2-month-old mice (Sargiannidou et al., 2015), or intrathecally (30 μl) into L5–L6 intervertebral space of 3-week-old mice (Kagiava et al., 2016, 2018; Kagiava and Kleopa, 2018) (Supplementary material).
Vector copy number determination
The average vector copy number per cell in DNA extracted from lumbar root and sciatic nerves was calculated as the total vector copy number divided by the total cell number (Geraerts et al., 2006; Christodoulou et al., 2016) (Supplementary material).
Immunofluorescence
Mice were anaesthetized with avertin according to institutionally approved protocols, and then transcardially perfused with phosphate-buffered saline (PBS) followed by fresh 4% paraformaldehyde. The lumbar spinal cord with all roots attached as well as sciatic nerves were dissected and frozen for cryosections. Sciatic nerves were also teased into fibres under a stereoscope. Teased fibres or sections were permeabilized in cold acetone and incubated at room temperature with a blocking solution of 5% bovine serum albumin (BSA) containing 0.5% Triton™ X for 1 h. Primary antibodies (Supplementary material) were incubated at 4°C overnight followed by appropriate secondary antibodies at room temperature. Slides were mounted with fluorescent mounting medium (DAKO) and images photographed under a fluorescence microscope with a digital camera using AxioVision software (Carl Zeiss MicroImaging).
Quantification of SH3TC2/myc expression was performed at 4 weeks post-injection in ventral root sections (n = 3 mice) and in sciatic nerve teased fibres (n = 5 mice) from five different areas from each slide. The number of myc-positive Schwann cells as well as the total number of cells in each picture was counted to determine the average expression ratio.
Immunoblot analysis
Immunoblots of lumbar root and sciatic nerve lysates collected 4 weeks post-injection (Supplementary material) were incubated with anti-myc (1:1000; Santa Cruz) and anti-GAPDH (1:4000; Santa Cruz Biotechnology) primary antibodies followed by HRP-conjugated anti-mouse or anti-rabbit secondary antibodies (Jackson ImmunoResearch, diluted 1:3000). The bound antibody was visualized by an enhanced chemiluminescence system (GE Healthcare Life Sciences).
Real-time PCR analysis of SH3TC2 expression
RNA was isolated with the Qiagen RNeasy® Lipid Tissue Mini Kit following the manufacturer’s protocol from snap-frozen lumbar roots and sciatic nerve tissues 4 weeks after intrathecal injection as well as from a frozen human sciatic nerve biopsy sample used as positive control and from non-injected Sh3tc2−/− mouse nerves used a baseline control. After DNase treatment, RNA was quantified by spectrophotometry and 0.5 μg of RNA was used to synthesize cDNA using TaqManTM reverse-transcription reagents. Human SH3TC2 probe (Hs00226194_m1, Applied Biosystems) was used.
Treatment trial in Sh3tc2−/− mice
For the gene therapy trial, we used two groups of Sh3tc2−/− mice. All animals were intrathecally injected at the age of 3 weeks. Littermate mice were randomized to either receiving the full vector LV-Mpz.SH3TC2.myc (treatment group) or the mock vector LV-Mpz.Egfp (lacking the Sh3tc2 gene) as control group. Randomization was based on animal numbering after tailing. We randomized 2:1 because more mice with full vector injection were needed to analyse SH3TC2.myc expression by immunostaining, immunoblot, and real-time PCR (above), therefore more fully-treated mice were available for behavioural and electrophysiological studies. However, an equal n number was used for the morphometric analysis groups (semithin sections). Mice were evaluated at the age of 11 weeks by behavioural analysis (n = 12 mock and n = 21 full vector) and electrophysiological testing (n = 10 mock and n = 17 full vector), and sacrificed afterwards for SH3TC2 expression analysis or for quantitative morphometric analysis (n = 9–10 per group) (Supplementary Fig. 8). All outcome evaluations were performed by examiners blinded to the treatment condition. Primary endpoint was considered the rescue of pathological changes in sciatic nerves. Secondary endpoints were the improvement in motor conduction velocities and in behavioural performance.
Behavioural analysis
Rotarod and foot grip tests were performed 8 weeks post-injection to assess motor balance, coordination and muscle strength as described previously (Kagiava et al., 2018) (Supplementary material).
Electrophysiological assessment
Bilateral sciatic nerves were stimulated in anaesthetized animals at the sciatic notch and distally at the ankle via bipolar electrodes with supramaximal square-wave pulses (5 V) of 0.05 ms. Compound muscle action potentials were recorded by a bipolar electrode inserted between digits 2 and 3 of the hind paw and latencies were measured from the stimulus artefact to the onset of the negative M-wave deflection based on previously described methodology (Zielasek et al., 1996). MNCVs were calculated by dividing the distance between the stimulating and recording electrodes by the result of subtracting the distal latency from the proximal latency.
Morphometric analysis of myelination in lumbar spinal roots and mid-sciatic nerves
Mice were transcardially perfused with 2.5% glutaraldehyde in 0.1 M phosphate buffer. The lumbar roots and sciatic nerves, were dissected and fixed overnight at 4°C, then osmicated, dehydrated, and embedded in araldite resin. Transverse semithin sections (1 μm) of the lumbar spinal cord with ventral roots and mid-sciatic nerves were obtained and stained with alkaline toluidine blue. Sections were visualized with 10×, 20×, 40×, and 63× objective lenses and captured with a Zeiss AxioCam HR camera. To examine the degree of abnormal myelination in both groups, images visualized at 63× were used.
The g-ratios for all myelinated fibres in the two treatment groups were calculated by Image-Pro software using a custom-made macro that detects the axons and their myelin sheath according to colour. This macro calculates g-ratio by dividing the average inner perimeter of the axon by the average outer perimeter of the axon, as well as the average myelin thickness for each myelinated fibre. This analysis allowed the calculation of g-ratios and myelin thickness. Because more pronounced myelination deficits were observed in fibres >4 μm in diameter in the original characterization of this model (Arnaud et al., 2009), we also analysed the subset of fibres >4 μm separately in addition to the analysis of all fibres >1 μm. In addition, completely demyelinated fibres (defined as axons >1 μm in diameter devoid of myelin sheath) in anterior lumbar roots and mid-sciatic nerves were manually counted and compared to the total fibre numbers, to obtain the ratio of demyelinated fibres.
Assessment of nodal pathology
Teased sciatic nerves from wild-type (n = 2), full vector-injected (n = 5) and mock vector-injected Sh3tc2−/− mice (n = 4) were used to measure the nodal length by directly measuring the width of Nav1.6 immunoreactive area using Image-Pro software. A total of 55 nodes from mock-treated and 89 nodes from fully-treated mice, along with 18 nodes from wild-type mice were measured and compared.
Assessment of transcriptional changes in treated Sh3tc2−/− mice
The expression of several genes previously reported to be dysregulated in younger Sh3tc2−/− mice (Arnaud et al., 2009) was analysed in sciatic nerves and lumbar roots first in untreated 11-week old Sh3tc2−/− compared to wild-type littermates (Supplementary material). Subsequently the expression of the most significantly dysregulated genes Scip, Cdgap, Cxcl24 and Mpz was compared between full and mock treatment groups, using quantitative PCR. Gapdh was used as a loading control.
Measurement of blood neurofilament light levels
Blood samples were obtained from 6- and 25-week-old wild-type (n = 2–3 per age group) and untreated Sh3tc2−/− mice (n = 3–4 per age group), as well as from 11-week-old full vector- or mock vector-treated Sh3tc2−/− mice (n = 5–7 per treatment group) using standard methods (Parasuraman et al., 2010). Blood samples were collected into EDTA-containing tubes and centrifuged at 20°C and 3500 rpm for 10 min. Plasma was aliquoted and stored at −80°C until testing. Plasma sample neurofilament light (NfL) concentration was measured at UCL using a commercially available NF-Light kit on a single molecule array (SimoaTM) HD-1 instrument (Quanterix).
Statistical analysis
Behavioural, electrophysiological and morphometric analysis data, as well as plasma NfL levels obtained from full or mock vector-treated groups were compared using the Mann-Whitney U-test. Nodal width, real time PCR results, and baseline NfL levels were compared with two-tailed Student’s t-test (significance level for all comparisons, P < 0.05).
Data availability
Data supporting these findings are available upon request.
Results
The LV-Mpz.SH3TC2.myc vector induces widespread expression of SH3TC2 in Schwann cells
The LV-Mpz.SH3TC2.myc (full) and LV-Mpz.Egfp (mock) plasmids (Fig. 1A and B) were successfully cloned and verified by direct sequencing. Following lentiviral (LV) vector production, the vector titres obtained ranged from 5.3 × 109 to 1.5 × 1011 vector copies/ml as determined by quantitative PCR and ELISA. To assess expression and cell specificity based on the Mpz promoter, vectors were injected either intraneurally (10 μl) into exposed sciatic nerves (Sargiannidou et al., 2015), or intrathecally (30 μl) into the L5–L6 intervertebral space of anaesthetized mice (Kagiava et al., 2016, 2018). The expression and distribution of SH3TC2.myc was examined 4 and 8 weeks after vector injection of the full vector into Sh3tc2−/− mice, compared to non-injected or mock vector-injected animals.
Immunostaining of sciatic nerve teased fibres from mice injected intraneurally with the LV-Mpz.SH3TC2.myc vector showed myc immunoreactivity in the perinuclear cytoplasm of a subset of myelinating Schwann cells in different nerve fascicles obtained from both proximal and distal sections of the nerve. In contrast, no myc immunoreactivity was detected in non-injected sciatic nerve fibres (Supplementary Fig. 2). We similarly examined the expression of myc-tagged SH3TC2 in sciatic nerve teased fibres from intrathecally-injected mice. We found perinuclear myc immunoreactivity in a subset of myelinating Schwann cells in spinal roots (Fig. 1C and D) and in peripheral nerves from full vector injected but not from mock vector injected mice, including sciatic nerve (Fig. 1E and F) and femoral motor nerve fibres (data not shown), in keeping with our previous findings showing widespread gene expression following intrathecal injection (Kagiava et al., 2016). In addition to immunostaining with anti-myc antibodies, we tested several commercially available anti-SH3TC2 antibodies to confirm expression of the lentiviral vector and confirmed that a rabbit antiserum (ab204334) provided specific SH3TC2 immunoreactivity as demonstrated by immunostaining of human sural nerve biopsy sections (Supplementary Fig. 3).
Double immunostaining for hSH3TC2 and myc tag of lumbar root sections and sciatic nerve fibres from LV-Mpz.SH3TC2.myc intrathecally-injected Sh3tc2−/− mice confirmed the co-localization of virally expressed hSH3TC2 in the perinuclear Schwann cell cytoplasm, typically in a granular pattern, in a subset of myelinating Schwann cells, while SH3TC2 immunoreactivity was absent in tissues from mock-injected Sh3tc2−/− mice (Fig. 1C–F and Supplementary Fig. 4). To clarify the cell specificity of hSH3TC2 expression in peripheral nerve tissues further, we also stained sciatic nerve cross sections from LV-Mpz.SH3TC2.myc vector intrathecally-injected Sh3tc2−/− mice with the combination of anti-SH3TC2 with anti-vimentin (fibroblast marker) or with myelin basic protein (MBP, a myelinating Schwann cell marker) antibodies. This analysis showed that virally delivered SH3TC2 was expressed specifically in MBP-positive myelinating Schwann cells but not in vimentin-positive fibroblasts or non-myelinating Schwann cells (Neuberger and Cornbrooks, 1989; Walko et al., 2013) in the peripheral nerves of injected mice (Supplementary Fig. 5).
We used immunostaining to examine the expression of virally-delivered hSH3TC2 in relation to the recycling endosome GTPase Rab11, which is known to associate with SH3TC2 (Roberts et al., 2010; Stendel et al., 2010). Immunostaining for Rab11 in untreated Sh3tc2−/− mice showed characteristic perinuclear expression, which did not differ from the Rab11 expression pattern in wild-type mouse nerve fibres. In LV-Mpz.SH3TC2.myc-injected Sh3tc2−/− mice, virally expressed hSH3TC2 (labelled with anti-myc) showed co-localization with Rab11 in Schwann cells (Supplementary Fig. 6), indicating an interaction with endogenous Rab11, in keeping with the physiological role of SH3TC2.
Quantification of SH3TC2 expression rates revealed by immunostaining in Sh3tc2−/− mice 4 weeks following intrathecal injection of LV-Mpz-SH3TC2.myc showed that 17 ± 2.9% of myelinating Schwann cells in sciatic nerves (n = 5 mice counted) and 38.1 ± 11.1% in ventral lumbar roots (n = 3 mice counted) expressed SH3TC2 (Fig. 1G). These expression rates corresponded to a proximal to distal gradient of vector biodistribution following intrathecal injection as assessed by higher vector copy numbers in lumbar roots (0.65 ± 0.21) compared to sciatic nerve (0.25 ± 0.15) samples (Fig. 1H).
Expression of myc-tagged hSH3TC2 was also shown by immunoblot analysis using the anti-myc antibody in sciatic nerve and lumbar root lysates obtained 4 weeks following intrathecal injection of the LV-Mpz.SH3TC2.myc vector. The predicted protein band at 144 kDa was detectable in sciatic nerves as well as in lumbar roots in two of the three mice examined compared to a negative control (Supplementary Fig. 7). Finally, hSH3TC2 expression was also assessed by quantitative PCR analysis. The SH3TC2 transcript was detected in most sciatic nerve samples from either intraneurally (Supplementary Fig. 2E) or intrathecally-injected mice, as well as in lumbar root samples from intrathecally-injected animals (Fig. 1I), at levels that were below that of an adult human sural nerve.
Gene replacement treatment in Sh3tc2−/− mice leads to improved motor performance and nerve function
After establishing an efficient gene replacement method for SH3TC2, we proceeded with a randomized, mock vector-controlled, blinded treatment trial by intrathecal injection in groups of Sh3tc2−/− littermate mice at 3 weeks of age, at early stages of neuropathy. Following injections, mice were observed for 8 weeks and then examined for possible phenotype rescue using behavioural, electrophysiological and histological methods at 11 weeks of age (Supplementary Fig. 8). Results were then compared between groups treated with the full (LV-Mpz.SH3TC2.myc) as opposed to the mock (LV-Mpz.Egfp) vectors. All behavioural and electrophysiological studies were performed by examiners blinded to the treatment status.
Rotarod analysis performed as described recently (Kagiava et al., 2018) 8 weeks after injection showed that at a speed of 20 rpm, fully-treated mice (n = 21) remained on the rotarod significantly longer (408.8 ± 37.3 s) than mock-injected mice (253.9 ± 65 s; P = 0.024; n = 12). Likewise, at 32 rpm, fully-treated mice remained on the rotarod significantly longer than mock-treated littermates (full: 207.7 ± 33.4 s; mock: 91.3 ± 17.5 s; P = 0.01; Mann-Whitney U-test) (Fig. 2A and B). Furthermore, the foot grip test, which evaluates hind limb strength by using an equipment that automatically measures the grams of force required to pry the mouse from the handle, revealed that fully-treated mice (n = 23) applied significantly greater force than mock-treated mice (n = 12) (full: 91.4 ± 17 g; compared to mock: 62.2 ± 21 g; P = 0.0067; Mann-Whitney U-test) (Fig. 2C). However, the motor performance of treated Sh3tc2−/− remained below the indicative levels of wild-type littermate mice.
Behavioural and electrophysiological analysis of treated Sh3tc2−/− mice. Improved rotarod motor performance was observed at two different speeds tested, 20 rpm (A) and 32 rpm (B), in full vector-injected Sh3tc2−/− mice (n = 21) compared to mock vector-treated mice (n = 12), as well as an increase in foot grip force generated in full (n = 23) compared to mock (n = 12) groups (C). Results of MNCV of mock and full vector-treated mice 8 weeks after vector injection are shown in D. Full vector-treated mice (n = 17) show significantly increased MNCV compared to the mock group (n = 10) (Mann-Whitney U-test used for all comparisons). Results from behavioural testing and MNCV measurements in wild-type (WT) littermate mice (n = 2–3) are included in all diagrams, indicating that response to treatment in Sh3tc2−/− mice was partial.
MNCV were measured in sciatic nerves according to previously published methods (Zielasek et al., 1996). The mean MNCV in the fully-treated Sh3tc2−/− mice (32.43 ± 4.36 m/s; n = 17 mice) was significantly faster than that in the mock-treated group (25.72 ± 4.29 m/s; n = 10 mice) (P = 0.0067; Mann-Whitney U-test), but remained slower than that of wild-type mice (41.7 ± 2.89 m/s; n = 3 mice) (Fig. 2D). Thus, both motor behavioural tests, as well as nerve conduction studies indicate a significant, albeit partial, therapeutic benefit in Sh3tc2−/− mice treated with the LV-Mpz.SH3TC2.myc than mock vector-treated littermates.
Improvement of peripheral myelination in treated Sh3tc2−/− mice
To examine whether SH3TC2 gene replacement can improve the hypo- and demyelination that characterizes Sh3tc2−/− mice further, we performed morphological analysis of anterior lumbar roots and mid-sciatic nerves comparing treated (n = 9) and mock-treated (n = 10) Sh3tc2−/− mice 8 weeks after vector injection, at 11 weeks of age. All morphometric analysis was performed blinded to the treatment status.
For quantitative analysis of myelination, we measured the g-ratios as well as the average myelin thickness of all myelinated fibres with diameter >1 μm in toluidine blue stained transverse semithin sections (1 μm) using Image-Pro software. Because more pronounced myelination deficits were observed in fibres >4 μm in diameter in this model (Arnaud et al., 2009), we also analysed the subset of fibres >4 μm separately in addition to the analysis of all fibres. Our morphological analysis confirmed a significant improvement in g-ratios, myelin thickness and ratios of demyelinated fibres in anterior lumbar roots and mid-sciatic nerves of fully-treated compared to mock-treated Sh3tc2−/− mice. In ventral lumbar roots (Fig. 3), g-ratios of all myelinated fibres in mock-treated mice were 0.703 ± 0.028 as compared to 0.637 ± 0.027 in fully-treated mice (P = 0.00052; Mann-Whitney U-test). In the subset of fibres >4 μm in diameter, the mock-treated group had a g-ratio of 0.762 ± 0.042 and the treated group 0.705 ± 0.038 (P = 0.015). Average myelin thickness in mock-treated group was 0.766 ± 0.16 μm and in fully-treated 1.053 ± 0.28 μm (P = 0.03), while in the subset of fibres >4 μm the myelin thickness was 0.758 ± 0.14 μm in the mock group and 1.185 ± 0.28 μm in the treated group (P = 0.0048).
Morphometric analysis of anterior lumbar roots. Representative low (A and B) and higher magnification (C and D) images of toluidine blue-stained semithin sections from anterior lumbar spinal roots of 11-week-old mock- (A and C) and full vector- (B and D) treated Sh3tc2−/− mice. Mock-injected mice show thin myelin sheaths in most of the axons as well as some completely demyelinated fibres (asterisks in C and D), whereas full vector-injected mice show improvement of this myelin pathology. The scatter plots below display as indicated g-ratios (E) and myelin thickness (G) of individual axons versus axonal diameter (red points: full treatment group; light blue points: mock-treated group). Each point corresponds to one fibre. Comparison of g-ratios from all fibres, as well as from the subset of fibres >4 μm (F) shows significant reduction in fully-treated (n = 9) compared to mock treated mice (n = 10), while the corresponding myelin thickness (H) is increased. The ratio of completely demyelinated fibres is also reduced in fully-treated compared to mock-treated Sh3tc2−/− mice (I) (P-values obtained with the Mann-Whitney U-test). Scale bars = 20 μm (A and B), 2 μm (C and D).
In the sciatic nerves (Fig. 4), the average g-ratio of all myelinated fibres in the mock-treated group was 0.673 ± 0.034 and in the fully-treated group 0.593 ± 0.033 (P = 0.0009), whereas for fibres >4 μm g-ratio was 0.725 ± 0.051 in the mock and 0.638 ± 0.046 in the treated group (P = 0.0048). Myelin thickness of all size fibres was increased in treated mice (1.234 ± 0.407 μm) compared to the mock group (0.899 ± 0.311 μm), an increase that did not reach statistical significance (P > 0.05). However, in the larger diameter fibres (>4 μm) of fully-treated mice there was a significant increase in myelin thickness (1.410 ± 0.376 μm) compared to the mock group (0.986 ± 0.344 μm) (P = 0.03).
Morphometric analysis of mid-sciatic nerves. Representative low (A and B) and higher magnification (C and D) semithin sections from mid-sciatic nerves of 11-week-old mock- (A and C) and full vector- (B and D) treated Sh3tc2−/− mice. Mock-injected mice show thin myelin sheaths in most of the axons as well as some completely demyelinated fibres (asterisks in C and D), and these abnormalities are ameliorated in full vector injected mice. The scatter plots below display as indicated g-ratios (E) and myelin thickness (G) of individual axons versus axonal diameter (red points: full mice; light blue points: mock treated mice). Average g-ratios from all fibres, as well as from the subset of fibres >4 μm are shown in F, and corresponding average myelin thickness in H, demonstrating significant improvement of both parameters in full vector-treated (n = 9) compared to mock-treated mice (n = 10). There is also significant reduction of the ratio of completely demyelinated fibres in full vector-treated animals (I) (all P-values obtained with the Mann Whitney U-test). Scale bars = 20 μm (A and B), 2 μm (C and D).
Finally, the ratios of completely demyelinated fibres in both the anterior lumbar roots and in the sciatic nerves were significantly decreased in the fully-treated compared to the mock-treated mice (Figs 3I and 4I). Similar to the original observation (Arnaud et al., 2009), we did not observe any axonal loss or any inflammatory cells in PNS tissues of Sh3tc2−/− mice at 11 weeks of age, regardless of treatment condition. Moreover, we could not find any myelination defects in lumbar spinal cord fibres (data not shown) in keeping with lack of any CNS myelination defects in this model based on previous examination of the optic nerve (Arnaud et al., 2009). Thus, LV-Mpz.SH3TC2.myc-treated mice show significant improvement of myelination in peripheral nervous system tissues compared to mock-treated littermates.
Improvement of nodal pathological changes in treated Sh3tc2−/− mice
Since nodal abnormalities consisting of nodal widening have been detected as early as 1 month of age in Sh3tc2−/− mice and in biopsied nerves from CMT4C patients (Arnaud et al., 2009), we also examined nodal pathology to clarify whether gene replacement could reverse some of these early onset pathological features of CMT4C. Morphometric analysis was performed in preparations of sciatic nerve teased fibres that were immunostained for nodal marker Nav1.6, as well as for paranodal Caspr (CNTNAP1) and juxtaparanodal Kv1.1 (KCNA1) or Caspr2 (CNTNAP2). We found that the characteristic nodal widening present in nerve fibres from mock-treated animals was reduced and approached the wild-type appearance in nerve fibres of fully-treated Sh3tc2−/− mice with SH3TC2 expression (Fig. 5A–E and Supplementary Fig. 9). We assessed the nodal length by directly measuring the width of Nav1.6 immunoreactive area. Quantification of 55 nodes from n = 4 mock-treated and 89 nodes from n = 5 fully-treated Sh3tc2−/− mice, along with 18 nodes from n = 2 wild-type mice confirmed these observations (Fig. 5F). Thus, this early characteristic nodal abnormality that occurs in the absence of SH3TC2 can also be rescued by the gene therapy.
Improvement of nodal phenotype in treated Sh3tc2−/− mice. Immunostaining of sciatic nerve teased fibres from mock (A) or full vector (B) treated Sh3tc2−/− mice shows expression of myc-tagged SH3TC2 (green) only in treated fibres (open arrowheads in B) around the Schwann cell nucleus (asterisks) and the normal appearance of Nav1.6 (red)-labelled adjacent node in the fibre from fully-treated mouse (B) in contrast to the elongated Nav1.6-labelled node in the fibre from mock-treated mouse (A) (nodal areas indicated by arrows). (C and D) High magnification images of longitudinal sciatic nerve semithin sections from mock-injected (C) or full-vector injected (D) Sh3tc2−/− mice demonstrate the nodal elongation (C), which improves after treatment (D). (E) Staining of sciatic nerve fibres from Sh3tc2+/+ (wild-type, WT), as well as mock-treated and full vector-treated Sh3tc2−/− mice with Nav1.6 (red) nodal marker shows the typical elongation of nodal width (indicated by white lines) in mock-treated Sh3tc2−/− mice, which is improved in full vector-treated Sh3tc2−/− mice nearing the appearance of Sh3tc2+/+ nodal areas, consistent with improved nodal molecular architecture. (F) Results of quantification of 55 nodes from n = 4 mock-treated and 89 nodes from n = 5 full vector-treated Sh3tc2−/− mice, along with 18 nodes from n = 2 wild-type mice, confirm significant improvement of nodal elongation measured by Nav1.6 staining (Mann-Whitney U-test). Scale bars = 2 μm (C and D), 10 μm (E).
Transcriptional changes in Sh3tc2−/− mice
Several genes involved in myelination have been found to be downregulated in 4-week-old Sh3tc2−/− mice (Arnaud et al., 2009) with simultaneous upregulation of transcription factors and other genes implicated in cell signalling and adhesion processes. These findings indicate that multiple genes and cellular pathways may play a role in CMT4C pathogenesis. Therefore, we examined whether SH3TC2 gene replacement could rescue some of these gene expression abnormalities. First, we examined the expression levels in lumbar roots as well as in sciatic nerves of selected dysregulated genes comparing 11-week-old Sh3tc2−/− mice with wild-type littermates. Several of these genes were upregulated in the sciatic nerves more than in the lumbar roots of Sh3tc2−/− mice compared to wild-type littermates, although mostly not at the same magnitude as reported at 4 weeks of age. Among the genes examined, the most prominent upregulation was found for Scip/Oct-6 (Pou3f1) expression, while we also found that Mpz expression was downregulated (Supplementary Fig. 10A). We then selected some of the most dysregulated genes at this age in Sh3tc2−/− mice, and performed a comparison of fully-treated versus mock-treated animals (n = 4 mice per group). This analysis showed a trend for ameliorated overexpression of dysregulated genes in both lumbar roots and sciatic nerves from treated mice, including Pou3f1/Scip, Cdgap, and Cxcl14, but only Pou3f1/Scip downregulation in the sciatic nerves was statistically significant (Supplementary Fig. 10B and C).
Blood neurofilament light levels as biomarkers
Finally, in a small group of full or mock vector-treated Sh3tc2−/− mice (n = 5–7 mice per group), we measured blood NfL levels, an established biomarker for axonal injury/degeneration found to be elevated in sera from CMT patients including two patients with CMT4C (Sandelius et al., 2018). We first confirmed that serum NfL levels were elevated in Sh3tc2−/− compared to Sh3tc2+/+ littermate mice starting at 6 weeks of age and showing progressive elevation at 25 weeks of age (Fig. 6A). In treatment groups, mock-treated Sh3tc2−/− mice at the age of 11 weeks showed an average NfL concentration of 491 ± 87 pg/ml (range: 348–814), while NfL concentration in full vector-treated Sh3tc2−/−mice was significantly lower 289 ± 18 pg/ml (range: 240–340; P = 0.022; Mann-Whitney U-test) (Fig. 6B). Thus, this emerging blood biomarker shows a treatment response in the CMT4C model.
Assessment of serum NfL levels in Sh3tc2−/− mice at baseline and after treatment. (A) Comparison of serum NfL levels between untreated Sh3tc2−/− and Sh3tc2+/+ littermate mice showed significant elevation in the CMT4C model both at 6 weeks of age (P = 0.0047; n = 3 mice per group), as well as at 25 weeks of age (P = 0.0024; n = 4 Sh3tc2−/− and n = 2 Sh3tc2+/+ mice). The elevation of NfL concentration in Sh3tc2−/− mice increased significantly from 6 to 25 weeks of age (P = 0.0011), indicating progressive axonal dysfunction in this model of demyelinating neuropathy over time. In contrast, no significant progression was found when comparing 6- to 25-week old Sh3tc2+/+ mice (P > 0.05) (Student’s t-test for all baseline comparisons). (B) Comparison of NfL levels in 11-week old full vector-treated (FULL, n = 7) compared to mock vector-treated (MOCK, n = 5) Sh3tc2−/− mice showed a significant reduction of NfL concentration (P = 0.022, Mann-Whitney U-test).
Discussion
In this study, we demonstrate that a viral vector allowing for Schwann cell-targeted expression of hSH3TC2 could provide a gene therapy approach for CMT4C. Delivery of this vector into Sh3tc2−/− mice, an authentic genetic model of CMT4C, achieved Schwann cell-specific expression that was widespread after intrathecal injection. An early treatment trial comparing LV-Mpz.SH3TC2myc-treated with mock vector-treated Sh3tc2−/− mice demonstrated significant phenotype rescue, as indicated by improved motor behavioural tests, increased motor conduction velocities, reduction of myelin pathology and restoration of lengthened nodal architecture in Sh3tc2−/− mice. Thus, our study provides a first proof of principle for viral gene replacement therapy targeted to Schwann cells to treat CMT4C.
Like other demyelinating neuropathies, CMT4C results from mutations in a Schwann cell-specific gene. Indeed the initial characterization of endogenous SH3TC2 expression in mice revealed high expression specifically in sciatic nerves, localized to Schwann cells (Arnaud et al., 2009), indicating a cell-autonomous loss of function mechanism underlying CMT4C neuropathy. Therefore, a targeted replacement of SH3TC2 in myelinating Schwann cells will be ultimately necessary to treat CMT4C. Based on our previously-tested lentiviral vector with demonstrated Schwann cell tropism showing high infection rates in Schwann cells, as well as the rat Mpz promoter, which drives gene expression restricted to myelinating Schwann cells (Sargiannidou et al., 2015; Kagiava et al., 2016), we developed a novel vector to deliver the human SH3TC2 gene into Schwann cells. Cell-specific expression of SH3TC2 was shown with immunostaining in combination with cell markers, as well as by immunoblot analysis and by detection of the transcript at the RNA level. Moreover, we could detect the characteristic vesicular pattern of SH3TC2 immunoreactivity at the plasma membrane and in perinuclear Schwann cell cytoplasm characteristic of the endocytic recycling compartment, as well as co-localization with the interacting protein Rab11 (Ullrich et al., 1996; Arnaud et al., 2009; Stendel et al., 2010), suggesting that it assumes the functional role of the endogenous mouse SH3TC2.
The overall vector copy numbers and gene expression rates and levels achieved were not high, as only a subset of Schwann cells was found to be positive for SH3TC2/myc immunoreactivity. This is in line with the fact that we only achieved a partial phenotype rescue as demonstrated by all outcome measures. In other studies we also found that similar expression rates and lower than physiological expression levels of the neuropathy-associated gene such as GJB1 (connexin 32) were still adequate to provide a significant therapeutic benefit in disease models, although not a complete rescue (Kagiava et al., 2016, 2018). Moreover, lower correction rates in vivo were shown to provide therapeutic benefit in gene editing studies for muscular dystrophy (Long et al., 2014), overall indicating that a therapeutic benefit can be achieved even with suboptimal gene replacement.
The Sh3tc2−/− mouse provides a relevant model of CMT4C as it recapitulates all major phenotypic features of the disease. Although motor behavioural studies were not previously performed at baseline in this model, we used both rotarod analysis and foot grip tests, which are sensitive for peripheral nerve dysfunction as shown in our previous studies (Kagiava et al., 2016; Kagiava et al., 2018) and in other models of demyelinating neuropathy (Zhao et al., 2018). Both of these tests were sensitive enough to demonstrate a significant improvement in the treated compared to the mock-treated group. In keeping with these motor behavioural improvements, we also found that motor nerve conduction velocities were increased in treated Sh3tc2−/− mice. In the initial characterization of this model, at the age of 10–12 weeks, corresponding to the age of our treatment outcome evaluation, average MNCV were ∼10 m/s slower compared to the wild-type average (Arnaud et al., 2009). In our groups, treated Sh3tc2−/− mice were ∼9.3 m/s slower than the wild-type average, but 6.7 m/s faster and significantly improved than mock-treated littermates.
Detailed morphometric analysis of myelinated fibres in spinal motor roots and mid-sciatic nerves demonstrated an improvement in myelination assessed by g-ratio calculation as well as by direct average myelin thickness measurements. Hypomyelination was documented in Sh3tc2−/− mice as early as postnatal Day 5, becoming progressively worse at later ages (Arnaud et al., 2009). This indicates that the neuropathy in this model and in CMT4C results from a combination of developmental and progressive hypomyelination. This is in line with the putative function of SH3TC2 in myelin biogenesis and maintenance, based on the regulation of cargo transport, including myelin proteins, through the recycling endosome (Ang et al., 2004; Trajkovic et al., 2006; Winterstein et al., 2008). We analysed mice at the age of 11 weeks at which a marked myelination defect was apparent and found a significant reduction of g-ratios along with increased myelin thickness in treated mice. The morphological improvements following gene therapy were only partial, likely reflecting the fact that only a subset of myelinating Schwann cells expressed SH3TC2. However, it is notable that even early onset pathological changes can be improved following treatment. Since we treated mice already at 3 weeks of age, it remains to be shown whether treatment at a later time point, could also provide a therapeutic benefit.
In this regard, the observed improvement in the abnormalities of nodal architecture is particularly interesting. The elongation of nodes of Ranvier observed in CMT4C nerve biopsies as well as in the Sh3tc2−/− mouse present already at early developmental stages (Arnaud et al., 2009) likely contributes to the functional impairment and progression of neuropathy. The cause of this abnormality is unclear. A dual mechanism drives the formation of the nodes of Ranvier and clustering of sodium channels during development, consisting of the nodal adhesion system of neurofascin186/NrCAM and gliomedin, as well as of the paranodal axon-glial junctions formed by Caspr/Contactin-NF155 that restrict the lateral diffusion of nodal sodium channels (Feinberg et al., 2010; Amor et al., 2014). Given that Nav1.6 channels remain clustered within the elongated nodes (Arnaud et al., 2009) and our own observations), the elongation of the nodes in CMT4C is likely to result from defects in lateral growth of Schwann cells during development, making the second mechanism less efficient. Interestingly, these defects appear to be reversible, since we observed a significant remodelling of nodal architecture approaching the structure of wild-type nodes in treated compared to mock-treated mice. This may actually be a more responsive therapeutic outcome compared to the myelin thickness at the internode, accounting for the functional improvements observed in our study.
Dysregulation of several genes associated with various cellular functions, such as transcription factors regulating Schwann cell maturation, signalling, adhesion, immunity, extracellular matrix and myelination have been reported in 4-week-old Sh3tc2−/− mice (Arnaud et al., 2009). We confirmed that several of these genes were upregulated in the sciatic nerves more that in the lumbar roots of Sh3tc2−/− mice compared to wild-type littermates at 11 weeks of age, although to a lesser degree that reported at 4 weeks, suggesting that transcriptional abnormalities are more pronounced during development and persist in adulthood. We found significant improvement only of Scip/Oct-6 (Pou3f1) dysregulation in sciatic nerves of treated mice, while other genes showed a non-significant trend for normalization. This result may be explained by the fact that we only achieved a partial expression and pathology rescue based on the other outcome measures used, as well as due to smaller degree of transcriptional changes at this age compared to 4 weeks of age, making a therapeutic response more difficult to demonstrate. Scip/Oct-6 plays a crucial role in the regulation of promyelinating Schwann cell maturation before axonal wrapping (Bermingham et al., 1996) and appears to be a sensitive marker in hypomyelinating and demyelinating neuropathies, correlating with the overall reduction of promyelinating Schwann cells after SH3TC2 replacement. In a recent treatment trial in models of CMT1A, Scip/Oct-6 was also most prominently upregulated at baseline and was the most responsive to the therapeutic intervention (Zhao et al., 2018).
Lastly, we demonstrate for the first time that a clinically relevant biomarker, plasma NfL concentration, previously shown to be elevated in CMT patients including two cases with CMT4C, correlating with clinical severity (Sandelius et al., 2018), is also progressively elevated in Sh3tc2−/− mice and shows a significant improvement following gene replacement therapy. Thus, plasma NfL level appears to be a clinically relevant and treatment-responsive biomarker in the CMT4C model that can be useful in future preclinical and clinical studies.
Several challenges remain in the path to developing a clinically translatable gene therapy for CMT4C. In the current study, we tested the efficacy of intrathecal genes as early as possible, at 3 weeks of age. However, we also need to demonstrate that treatment at later stages after the onset of neuropathy can provide a similar therapeutic benefit. This question is relevant to treating CMT4C patients with a long disease course and likely ensuing secondary axonal loss, which is not present in the mouse model. Nevertheless, given that this is an early onset hypomyelinating neuropathy, even early intervention provides a proof of principle that abnormalities of myelin biogenesis and maintenance may be reversible, as shown by the rescue of the early nodal elongation and hypomyelination.
Furthermore, better vector biodistribution and higher gene expression levels are likely to provide a more robust therapeutic response. Alternative vectors such as AAV9 or AAVrh10 (Foust et al., 2009; Tanguy et al., 2015; Gurda et al., 2016) could be tried if they can target Schwann cells as well as the lentiviral vector. They offer the advantage of improved safety with in vivo delivery because they do not integrate into the host genome and show low immunogenicity (Calcedo and Wilson, 2013). However, the limitation of smaller transgene capacity needs to be overcome, for example by using minimal Schwann cell-specific promoters. Confirmation of adequate viral vector biodistribution after intrathecal injection will also be needed in large animal models, while the alternative route of intravenous vector delivery should be considered (Bevan et al., 2011; Tanguy et al., 2015). This may be also relevant since the intrathecal injection could be difficult in CMT4C patients with prominent scoliosis (Kessali et al., 1997; Gabreels-Festen et al., 1999; Azzedine et al., 2006).
In conclusion, this study provides evidence that SH3TC2 gene replacement using a Schwann cell targeted viral vector can provide a therapeutic benefit in a model of CMT4C demyelinating neuropathy, leading to partial rescue of functional, electrophysiological and morphological abnormalities. Further studies are needed to optimize vector safety and biodistribution, gene expression levels, as well as to demonstrate effectiveness of this approach after the onset of neuropathy, in order to facilitate the clinical translation for CMT4C patients.
Abbreviations
- CMT4
Charcot-Marie-Tooth disease type 4
- MNCV
motor nerve conduction velocity
- NfL
neurofilament light
Acknowledgements
We would like to thank Dr Vincenzo Lupo, University of Valencia, for providing us with the SH3TC2 cDNA construct, as well as Dr John Svaren for helpful discussions in the initial planning of the project. We also thank Martha Foiani for help with measuring NfL samples, Dr Kyriaki Markoullis for help with the real time PCR analysis as well as Christos Karaiskos and Dr George Lapathitis for assistance with mouse maintenance and behavioural testing.
Funding
This study was funded by the Charcot-Marie-Tooth Association (USA) (grant to K.A.K.) and by the French Muscular Dystrophy Association (AFM-Téléthon Grant 19719 to K.A.K.). H.Z. is a Wallenberg Academy Fellow supported by grants from the Swedish Research Council, the European Research Council, the Knut and Alice Wallenberg Foundation, the Leonard Wolfson Experimental Neurology Centre, the UK Dementia Research Institute at UCL and the Wellcome Trust.
Competing interests
The authors report no competing interests.
