Patients with Charcot–Marie–Tooth neuropathy and gene targeting in mice revealed an essential role for the SH3TC2 gene in peripheral nerve myelination. SH3TC2 expression is restricted to Schwann cells in the peripheral nervous system, and the gene product, SH3TC2, localizes to the perinuclear recycling compartment. Here, we show that SH3TC2 interacts with the small guanosine triphosphatase Rab11, which is known to regulate the recycling of internalized membranes and receptors back to the cell surface. Results of protein binding studies and transferrin receptor trafficking are in line with a role of SH3TC2 as a Rab11 effector molecule. Consistent with a function of Rab11 in Schwann cell myelination, SH3TC2 mutations that cause neuropathy disrupt the SH3TC2/Rab11 interaction, and forced expression of dominant negative Rab11 strongly impairs myelin formation in vitro. Our data indicate that the SH3TC2/Rab11 interaction is relevant for peripheral nerve pathophysiology and place endosomal recycling on the list of cellular mechanisms involved in Schwann cell myelination.
Myelin is a specialized membranous sheath, made up of ∼70% lipids and 30% proteins. It is produced by two types of glia cells, oligodendrocytes in the central nervous system (CNS) and Schwann cells in the peripheral nervous system (PNS). Myelin surrounds nerve axons, allowing saltatory nerve conduction and ensuring maintenance of the axon at a long distance from the cell body (Griffiths et al., 1998; Lappe-Siefke et al., 2003; Nave and Trapp, 2008). The importance of myelin formation is illustrated by the severe neurological deficits seen in inherited and non-genetic demyelinating diseases of the CNS (e.g. multiple sclerosis and leukencephalopathies) and the PNS (e.g. autoimmune neuritis and hereditary neuropathies).
One condition associated with impaired myelination of the PNS is Charcot–Marie–Tooth disease type 4C (CMT4C) (LeGuern et al., 1996). CMT4C is an autosomal recessive form of hereditary motor and sensory neuropathies (Dyck et al., 1993) clinically presenting as distal muscle weakness and wasting, distal sensory deficits and prominent scoliosis (Kessali et al., 1997; Azzedine et al., 2006). Pathologically, CMT4C is characterized by layers of empty basal lamina encircling demyelinated and remyelinated axons, abnormal Schwann cell protrusions (Gabreels-Festen et al., 1999) and disorganization of the node of Ranvier (Arnaud et al., 2009).
We have previously shown that CMT4C is caused by mutations in the SH3TC2 (Src homology 3 domain and tetratricopeptide repeats 2)/KIAA1985 gene on chromosome 5q32 (Senderek et al., 2003). CMT4C is associated with both nonsense and missense mutations throughout the gene, and a strict genotype–phenotype correlation has not been established because of interfamilial and intrafamilial variability of clinical expression as well as marked allelic heterogeneity (Senderek et al., 2003; Azzedine et al., 2006; Colomer et al., 2006; Gosselin et al., 2008). SH3TC2 encodes SH3TC2/KIAA1985, a novel protein of unknown function containing several motifs potentially involved in protein–protein interactions. However, no protein binding to SH3TC2 has been identified so far. Very recently, we found that a mouse model without a functional copy of the Sh3tc2 gene developed a peripheral neuropathy that largely reproduced the human phenotype (Arnaud et al., 2009). Consistent with the tissue and the cell population involved in a demyelinating peripheral neuropathy, Sh3tc2 is exclusively expressed in Schwann cells in peripheral nerves. SH3TC2 is tethered to cellular membranes through an N-terminal myristic acid anchor and localizes to the plasma membrane and a perinuclear membranous structure (Arnaud et al., 2009; Lupo et al., 2009) that corresponds to the endocytic recycling compartment (Arnaud et al., 2009). However, as the physiological function of the gene product, SH3TC2, has not yet been identified, the pathomechanism causing impaired PNS myelination in CMT4C patients and Sh3tc2 knockout mice has so far remained unclear.
Here, we show that SH3TC2 is a novel effector of the small GTPase Rab11, a key regulator of recycling endosome functions. Neuropathy-causing missense mutations in SH3TC2 disrupt this interaction. These data, together with the demonstration of the role of Rab11 in myelination, support the importance of the SH3TC2/Rab11 interaction for normal myelination.
Materials and methods
All experiments with animals followed protocols approved by the veterinary office of the Canton of Zurich, Switzerland. The generation of Sh3tc2 knockout mice has been described previously (Arnaud et al., 2009). Genotypes were determined by polymerase chain reaction on genomic DNA derived from tail biopsies as reported earlier (Arnaud et al., 2009).
The generation of plasmids is outlined in the Supplementary material.
Cell culture and transfections
COS7, HEK293, HEK293T and RT4-D6P2T cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM; Invitrogen) supplemented with 10% foetal calf serum (Invitrogen) and 2 mM glutamine (Invitrogen). Culture medium for Flp-In T-Rex 293 cells (Invitrogen) additionally contained 100 µg/ml Zeocin™ (Invitrogen) and 15 µg/ml blasticidine (Invitrogen). All transfections in this study were performed using Lipofectamine 2000 (Invitrogen) according to the manufacturer’s instructions.
Flp-In T-Rex 293 cells that stably express inducible SH3TC2 fused to a C-terminal tandem affinity purification (TAP) tag (SH3TC2-C-TAP) or the TAP tag alone (C-TAP) were generated according to the manufacturer's instructions (Invitrogen). Hygromycin B (Invitrogen) at a concentration of 10 µg/ml was used for selection of stable cell clones. Cells were cultured with 15 µg/ml blasticidine and 10 µg/ml hygromycin B. In order to induce expression of SH3TC2-C-TAP or C-TAP, doxycycline (Sigma) was administered to the cells at a final concentration of 1 µg/ml for 20–24 h.
Generation of lentiviral stocks
For production of high-titre lentiviruses, HEK293T cells were transiently cotransfected with the pSicoR vector harbouring the complementary DNA of choice and the packaging constructs pMD2.G and psPAX2 (Addgene). Cell culture supernatant was collected after 48 and 72 h. The filtered supernatant was first centrifuged in an SW28 rotor (Beckmann Coulter) for 2 h at 21 000 r.p.m. at 11°C. Afterwards, the pellet was resuspended in DMEM/10% foetal calf serum and centrifuged again for 1 h at 16 000 r.p.m. at 4°C in a T60i rotor (Beckmann Coulter). The pellet was resuspended in 40 µl phosphate buffered saline (PBS), aliquoted and stored at −80°C.
Antibodies used in this study are described in the Supplementary material.
Tandem affinity purification
Flp-In T-Rex 293 cells stably expressing SH3TC2-C-TAP in a doxycycline-inducible fashion and a control cell line expressing the C-TAP tag alone (C-TAP) were used for purification of an SH3TC2 protein complex. The detailed protocol for tandem affinity purification has been described previously (Kleine et al., 2008).
Cultured cells were harvested, washed twice with PBS and lysed in lysis buffer (10 mM Tris–HCl, 5 mM EDTA, 150 mM NaCl, 1% Triton X-100) containing protease and phosphatase inhibitors (Sigma). Mouse and rat sciatic nerves were homogenized with a mortar and pestle in lysis buffer. Post-nuclear supernatants were boiled in sodium dodecyl sulphate (SDS) sample buffer (80 mM Tris pH 6.8, 10% glycerol, 2% sodium dodecyl sulphate, 0.002% bromphenol blue), resolved by sodium dodecyl sulphate polyacrylamide gel electrophoresis (SDS–PAGE) and electroblotted onto polyvinylidene fluoride membranes (Hybond-C; Amersham). Immunoblots were developed by incubation with appropriate antibodies followed by horse radish peroxidase- or alkaline phosphatase-chemiluminescence detection. Densitometry and quantification of protein levels were performed with Quantity One software (BioRad).
Transiently transfected HEK293 cells were harvested 24 h after transfection and the post-nuclear supernatant was pre-cleared with 30 µl of Protein G-Sepharose (GE Healthcare) at 4°C for 2 h. The supernatant was incubated with 30 µl of Protein G-Sepharose including 8 µg/ml of mouse anti-Myc, mouse anti-green fluorescent protein (GFP) or mouse anti-FLAG antibodies or mouse IgG. Immunoprecipitation was carried out at 4°C on a rotating wheel for 16 h. The precipitates were washed six times with cold lysis buffer and boiled in SDS sample buffer to elute protein complexes. The supernatants were processed using standard SDS–PAGE and western blotting procedures.
Glutathione S-transferase pulldown assays
In vitro translation of SH3TC2/Sh3tc2 and the generation of glutathione S-transferase (GST) or GST-Rab11 fusion proteins is described in the Supplementary material. In vitro translated SH3TC2/Sh3tc2 proteins (10 µl of standard reactions) were incubated for 1 h at 4°C with 50 µl glutathione-Sepharose beads coupled with GST or GST-Rab11 fusion proteins. The beads were washed four times with TNN (50 mM Tris pH 7.5, 250 mM NaCl, 5 mM EDTA, 0.5 % Nonidet-P40, 1 mM DTT) buffer before eluting proteins by boiling the beads in SDS sample buffer. Samples were resolved by SDS–PAGE followed by Coomassie Brilliant Blue staining and autoradiography.
Yeast two-hybrid assay
Protein interactions were assayed in yeast using a two-hybrid approach according to the manufacturer’s protocol (Dualsystems Biotech). pLexA-SH3TC2 or pLexA-Lamin C were transformed into the NMY32 reporter strain together with pACT2-Rab11a using the lithium acetate method. Cells were plated on plates lacking tryptophane and leucine (TRP-, LEU-) to select for transformants. After 3 days of growth, five medium-size colonies for each condition were replated on selection plates lacking tryptophane, leucine and histidine (TRP-, LEU-, HIS-). After 3 days protein interactions were determined by growth of colonies. The results given are representative of at least three trials.
Cells were fixed with 4% paraformaldehyde in PBS for 10 min at room temperature, washed in PBS and permeabilized with 0.1% Triton X-100 (Sigma). Fixed cells were blocked for 30 min with 10% goat serum in PBS containing 0.1% Triton X-100 before incubation in primary antibodies in blocking medium overnight at 4°C. Cells were washed in PBS and incubated with fluorescent secondary antibodies in blocking solution for 1 h at room temperature. After washing, cells were incubated with 4′,6′-diamidino-2-phenylindole dihydrochloride (DAPI) (Sigma) to visualize nuclei, washed again and mounted in Immu-Mount (Thermo Scientific). For Rab11, p230 and GM130 stainings, 0.05% saponin (Sigma) was used as detergent instead of Triton X-100. Images were acquired using either an AxioVert Observer D1 fluorescence microscope (Carl Zeiss) or a TCS SP1 laser scanning confocal microscope (Leica). Images were further processed using Photoshop software (Adobe).
Cells were harvested, washed twice in PBS and lysed in homogenization medium (0.25 M sucrose, 1 mM EDTA, 10 mM Hepes–NaOH, pH 7.4) including protease and phosphatase inhibitors. Iodixanol gradient solutions of 5, 10, 15, 20 and 25% were prepared from a 50% Optiprep (Gibco) solution. Solutions were layered in 950 µl fractions with 250 µl post-nuclear supernatant on the top in Ultra-Clear centrifuge tubes (Beckmann Coulter) and centrifuged in an SW55 rotor (Beckmann Coulter) at 35 000 rpm for 20 h at 4°C. After centrifugation, fractions were collected from the bottom by tube puncture and proteins were precipitated with trichloroacetic acid. Equal volumes of each fraction were analysed by SDS–PAGE and immunoblotting.
Rab11a transcript quantification
Analysis of Rab11a mRNA expression is described in the Supplementary material.
Transferrin receptor recycling assay
HEK293 cells were transfected with SH3TC2-GFP wild-type or SH3TC2-GFP harbouring CMT4C missense mutations. Cells transfected with GFP alone or GFP transfected cells additionally treated with 10 μM monensin [an inhibitor of transferrin receptor recycling (Stein et al., 1984)] were used as controls. Twenty-four hours after transfection, cells were starved for 4 h at 37°C in serum-free DMEM with 0.1% bovine serum albumin and subsequently incubated for 30 min in serum-free DMEM/0.1% bovine serum albumin containing 8.4 mg/ml transferrin-Alexa Fluor 647 (Invitrogen). Cells were washed with ice-cold PBS and chased in serum-free DMEM/0.1% bovine serum albumin containing 1 mg/ml unlabelled holo-transferrin (Sigma) at 37°C for different lengths of time. Cells were washed, acid-stripped (0.2 M Na2HPO4, 0.1 M citric acid) and trypsinized. Cells were fixed in paraformaldehyde for 20 min, pelleted and resuspended in PBS containing 2% foetal calf serum, 20 mM EDTA and 0.02 % NaN3. The fluorescence intensity of cell bound transferrin was measured for 2000 GFP positive cells and the average intensities of the cell populations were calculated. Data acquisition was done on a FACSCalibur flow cytometer using CellQuest software (BD Biosciences). The experiment was repeated three times.
Surface reconstruction of recycling endosomes
COS7 cells were transfected with wild-type SH3TC2-FLAG, SH3TC2_N881S-FLAG or GFP. Cells were fixed with paraformaldehyde 24 h after transfection and stained with rabbit polyclonal anti-Rab11a and mouse monoclonal anti-FLAG antibodies, followed by fluorescently labelled secondary antibodies. Z-stacks in 0.122 µm intervals were taken for 100 transfected cells per condition (identified by FLAG staining or GFP fluorescence) with equal settings for zoom factor, laser intensity and pinhole. To estimate the surface of recycling endosomes, Rab11a positive surface areas were quantified with Imaris software (Bitplane) and corrected for cell size. The surface module of Imaris software was used for 3D reconstructions of recycling endosomes.
In vitro myelination
Preparation of dorsal root ganglia for dorsal root ganglion explant cultures, culture conditions and assessment of myelination are described in the Supplementary material.
Preparation and culture conditions of rat Schwann cells and dorsal root ganglion neurons for Schwann cell-dorsal root ganglion neuron co-cultures are described in the Supplementary material. Four days prior to adding Schwann cells to dorsal root ganglion neurons, Schwann cells were transduced by overnight incubation with the lentivirus of interest. Infected Schwann cells (200 000 per coverslip) were added and co-cultures were kept in C-medium (MEM; Invitrogen), supplemented with 10% foetal calf serum, 4 g/l d-glucose, 2 mM l-glutamine (Invitrogen), 50 ng/ml neural growth factor (Harlan) and 1% penicillin/streptomycin for 3 days. Myelination was induced over 10 consecutive days with C-medium supplemented with 50 µg/ml ascorbic acid (Sigma). Subsequently cultures were fixed for 20 min in 4% paraformaldehyde and for additional 15 min in ice-cold methanol at −20°C and blocked for 20 min with PBS containing 5% bovine serum albumin, 0.1% goat serum and 0.2% Triton X-100. Coverslips were incubated overnight at 4°C with primary antibodies in blocking solution. After additional washing steps in PBS, cultures were incubated with fluorescent secondary antibodies for 1 h at room temperature. Coverslips were washed again, incubated with DAPI and mounted with Immu-Mount. Myelinated segments were stained with a rat monoclonal anti-myelin basic protein antibody, followed by a Cy3-conjugated goat anti-rat IgG antibody. Neurons were detected with a mouse monoclonal anti-neurofilament antibody and a goat anti-mouse IgG antibody conjugated to Cy5. Transduction efficiency was monitored by GFP fluorescence. Myelination was evaluated as follows: ten fields per coverslip were randomly acquired (magnification 10×) and mean Cy3 fluorescence as an estimate of the number of myelin basic protein positive segments was calculated using ImageJ software (National Institutes of Health). Experiments were done in triplicate, at least three coverslips per condition were analysed for each experiment.
The data show the mean ± SEM. Statistical significance was determined using a two-tailed Student’s t-test. Significance was set at *P < 0.05, **P < 0.01 or ***P < 0.001.
Based on the observation that SH3TC2 contains several protein motifs (Src homology 3 domain domains and tetratricopeptide repeat repeats, Fig. 1A) known to mediate the formation of protein complexes (Senderek et al., 2003), we hypothesized that we might be able to deduce the function of SH3TC2 from already known functions of interacting proteins. As no proteins interacting with SH3TC2 were known and as the putative protein binding motifs in SH3TC2 did not allow prediction of a bona fide interaction partner, we decided to perform an unbiased screen for interacting proteins. We purified an SH3TC2 protein complex by means of tandem affinity purification from HEK293 cells, analysed the complex on an SDS gel and subjected the resulting protein bands to mass spectrometry. Among other proteins, we identified the small GTPase Rab11 as a protein potentially interacting with SH3TC2 (Fig. 1B and Supplementary Fig. 1A). Rab11 is a key regulator of recycling pathways from endosomes to the plasma membrane (Ullrich et al., 1996) and consists of two isoforms, Rab11a and Rab11b, which are encoded by different genes and mainly differ in their C-termini. Rab11a and Rab11b are differentially expressed (Sakurada et al., 1991; Lai et al., 1994) and may also be functionally different to some extent (Lapierre et al., 2003).
In order to confirm the results of the interaction screen, we expressed SH3TC2-Myc and GFP-Rab11a fusion proteins in HEK293 cells and performed coimmunoprecipitation in both directions. In addition to the co-precipitation of Rab11a along with SH3TC2, immunoprecipitation in the other direction showed that SH3TC2 could be coprecipitated with GFP-Rab11a as well (Fig. 1C). Results from GST pulldown and yeast two-hybrid experiments provided further evidence for the interaction between SH3TC2 and both Rab11 isoforms and showed that SH3TC2 binds directly to Rab11 (Fig. 1D and E; Supplementary Fig. 1B). However, we were not able to identify the particular protein region of SH3TC2 that mediates the interaction with Rab11 (Supplementary Fig. 2).
We have recently shown that both overexpressed proteins, SH3TC2 and Rab11a, colocalize with γ-tubulin and internalized transferrin in the perinuclear recycling endosome (Arnaud et al., 2009). We now expand our earlier findings by studying the distribution of SH3TC2-FLAG and endogenous Rab11a in epithelial cell lines (HEK293 and COS7) and a Schwann cell line (RT4-D6P2T). Since there was no antibody available against endogenous Sh3tc2, RT4-D6P2T cells that express Sh3tc2 on the transcript level (data not shown) had to be transfected with the FLAG-construct as well. The localization of both proteins overlapped in the perinuclear region in all three cell lines (Fig. 2A). Moreover, strong colocalization was confirmed in consecutive optical sections along the microscope z-axis (Fig. 2B). Conversely, we observed no colocalization of SH3TC2 with marker proteins for other perinuclear organelles (Supplementary Fig. 3). In order to provide independent biochemical support for fluorescent microscopic findings, we subjected cultured cells overexpressing SH3TC2 to subcellular fractionation using Optiprep gradient centrifugation. Fractions were analysed for the presence of SH3TC2 by immunoblotting and the profile was compared with the distribution of Rab11a. The bands for Rab11a were shown to overlap with the signal for SH3TC2 (Fig. 2C).
As there is increasing evidence that the development and maintenance of Schwann cells and the myelin sheath require endosomal sorting (Trapp et al., 2004; Simons and Trotter, 2007), we decided to follow the SH3TC2/Rab11 interaction further. As there were no published data on Rab11 expression in the PNS, we studied Rab11a protein levels in Schwann cell lines, primary rat Schwann cell cultures and sciatic nerves of rat and mouse by western blotting (Fig. 3A). We found that Rab11a was prominently expressed in cultured Schwann cells and in the PNS of adult animals. Moreover, during the early stages of post-natal PNS development, Rab11a expression was strongly upregulated around post-natal day 10 (Fig. 3B), similar to the developmental expression profile of other proteins involved in myelination (Verheijen et al., 2003).
In order to explore whether binding of SH3TC2 to Rab11 is specific or whether SH3TC2 might promiscuously associate with different Rab GTPases, we performed GST pulldown assays with several proteins of the Rab family. No association was observed with Rab4a and Rab5a (associated with early endosomes) and Rab7a (associated with late endosomes) (Supplementary Fig. 4A). Having shown that wild-type SH3TC2 associates with Rab11 specifically we attempted to clarify whether the SH3TC2/Rab11 interaction is relevant for the pathogenesis of CMT4C. One obvious strategy to approach this intriguing question was to test CMT4C-causing SH3TC2 mutants (Fig. 1A) for their ability to associate with Rab11. We found that all tested amino acid substitutions related to missense mutations observed in patients with CMT4C disrupted the SH3TC2/Rab11a interaction (Fig. 4A). On the other hand, rat and mouse Sh3tc2 proteins bind to human Rab11a (Fig. 4B), although ∼20% of amino acid residues are not conserved between the human and the rodent proteins (Fig. 4C, a multiple sequence alignment of human, mouse and rat SH3TC3/Sh3tc2 is shown in the Supplementary material). These findings establish that the disruption of the SH3TC2/Rab11 complex is a specific effect of CMT4C-causing missense mutations.
Next, we asked whether the SH3TC2/Rab11 interaction is dependent on the activation state of Rab11. Like all GTPases, Rab proteins cycle between an inactive (guanosine diphosphate-bound) and an active (GTP-bound) conformation. Interconversion and accessibility of these two forms are temporally and spatially controlled by specific regulators (Stenmark, 2009). The GTP-bound, activated form of the GTPase associates with effector molecules through which it carries out its functions. We found that SH3TC2 interacts with the constitutive active GTP-bound Rab11_Q70L mutants, while there is no binding to the dominant negative guanosine diphosphate-bound Rab11_S25N mutants or to Rab11a_Q70L_I44E, a constitutively active mutant which has a second mutation in its effector domain abrogating effector binding (Wallace et al., 2002) (Fig. 5A and Supplementary Fig. 4B). Thus, the SH3TC2/Rab11 interaction is consistent with SH3TC2 functioning as a Rab11 effector.
Rab11 effectors have been shown to influence various functions of the recycling endosome, e.g. recycling of the transferrin receptor from endosomes back to the cell surface (Maxfield and McGraw, 2004). Assays for transferrin receptor recycling are commonly used as a paradigm to test recycling endosome functions (Sager et al., 1984). Such assays allow a quantitative assessment of the recycling rate of transferring receptor by measuring the amount of intact ligand that is retained in the cell (or has returned to the plasma membrane) upon time. Overexpression of wild-type SH3TC2 resulted in moderate slowing of transferrin recycling as compared with transferrin retention in cells transfected with CMT4C mutants or a GFP control vector (Fig. 5B and Supplementary Table 1). This suggests that wild-type SH3TC2 affects the rate of recycling along the receptor-mediated endocytosis pathway and this activity is largely absent in cells expressing SH3TC2 mutant proteins.
There is at least one known Rab11 effector, Rab11-FIP4, which causes a dramatic condensation of the recycling endosome (Wallace et al., 2002). Similarly, when SH3TC2 is expressed in transfected cells, not only does the endogenous Rab11a signal appear stronger, but the Rab11a-positive compartment appears to be more condensed in a perinuclear location (Fig. 2A and B). We quantified this effect by calculating the surface of Rab11a-positive endosomes in cells expressing wild-type SH3TC2 and the CMT4C-associated N881S mutant. The data we obtained confirmed that wild-type SH3TC2 condenses the perinuclear recycling compartment in a considerable proportion of transfected cells, while the N881S mutant had only a slight effect on recycling endosome morphology (Fig. 5C). This is in line with findings in teased fibres prepared from sciatic nerves of adult mice. The perinuclear Rab11 positive compartments are moderately less compact in Schwann cells lacking Sh3tc2 than in the wild-type situation (Fig. 5D).
Since a number of Rab11 effectors, including Rab11-FIP2 and Rab11-FIP4, have been shown to be capable of homodimerization (Lindsay and McCaffrey, 2002; Wallace et al., 2002), we were interested to determine if SH3TC2 displays similar properties. Using coimmunoprecipitation of full-length SH3TC2, we were able to demonstrate that SH3TC2 can self-interact (Supplementary Fig. 4C).
We finally compared the Rab11 expression levels in sciatic nerves of Sh3tc2 knockout mice and wild-type littermates, as it has been shown that Rab effectors can stabilize Rab-GTPases (Ganley et al., 2004). Indeed, we found that Rab11a protein levels were reduced in sciatic nerves of Sh3tc2 knockout mice (Fig. 5E). Conversely, levels of Rab11a mRNA were unchanged in Sh3tc2-deficient mice, suggesting that Rab11a expression is differently regulated at the protein level in knockout and wild-type animals.
Our results presented so far open the intriguing possibility that a new Rab11 effector, SH3TC2, regulates Schwann cell myelination. In order to explore a potential effect of Rab11 and endosomal recycling on the formation of myelin sheaths in the PNS directly, we made use of two different established in vitro myelination systems.
Dorsal root ganglia were isolated from embryonic day 13.5 mice and infected with a lentivirus encoding dominant negative (S25N) or constitutively active (Q70L) forms of a Rab11a-GFP fusion protein or GFP alone. After 8 days in culture, myelination was induced by adding ascorbic acid to the medium. After 12 subsequent days of culturing, myelinated segments were visualized by myelin basic protein staining. Infection of dorsal root ganglia explants with a lentivirus encoding the dominant negative mutant of Rab11a impaired myelination compared with the control infected with a virus for GFP alone. On the other hand, infection with a lentivirus containing constitutively active Rab11a resulted in an increased number of myelinated segments (Supplementary Fig. 5A and B).
However, as lentiviral infection of dorsal root ganglia explants targets both Schwann cells and neurons, our data left open the question whether overexpression of Rab11 mutants in Schwann cells or neurons, or in both populations, was responsible for altered myelination. In order to dissect Rab11 effects in Schwann cells and neurons, we used primary rat Schwann cell-dorsal root ganglion neuron co-cultures. Primary rat Schwann cells were infected with Rab11a variants or GFP control lentivirus and were added to cultured mouse dorsal root ganglia. Following induction of myelination by adding ascorbic acid, the amount of myelin production was assessed by myelin basic protein staining. Similar to dorsal root ganglion explant cultures, expression of the dominant negative mutant of Rab11a strongly impaired myelination, while expression of the constitutively active Rab11a resulted in moderately increased myelination (Fig. 6A and B). Altogether our data suggest that Rab11 controls Schwann cell myelination, most likely through regulation of SH3TC2 activity.
We have previously shown that mutations in the SH3TC2/Sh3tc2 gene cause demyelinating hereditary neuropathy in humans and mice (Senderek et al., 2003; Arnaud et al., 2009). The work presented here takes our understanding of the role of SH3TC2 in peripheral nerve demyelination to a new level: SH3TC2 acts as an effector of Rab11, a master regulator of recycling endosome functions (Maxfield and McGraw, 2004). In addition, we have shown that Rab11 itself is involved in the regulation of Schwann cell myelination in vitro, supporting the pathophysiological relevance of the observed SH3TC2/Rab11 interaction for peripheral nerves. Obviously, there is still the possibility that SH3TC2 can also influence myelination in a Rab11 independent manner. Potential candidates could be other proteins found in the SH3TC2 complex (Fig. 1A), whose role in the PNS and in Schwann cells has not yet been investigated.
Our findings directly lead to the question of how endosomal recycling might be involved in myelin sheath formation. Myelination requires dramatic changes in the cellular architecture of differentiating glia and a high degree of cell polarization that partitions the plasma membrane into distinct domains (Pfeiffer et al., 1993). Myelinating Schwann cells have at least five distinct membrane domains (Arroyo and Scherer, 2000; Simons and Trotter, 2007) that differ in their membrane lipid and protein composition. The outer (abaxonal) Schwann cell plasma membrane is specialized for extracellular interactions, while the inner (adaxonal) Schwann cell plasma membrane contains cell adhesion molecules (Previtali et al., 2001; Maurel et al., 2007; Spiegel et al.2007). Compact myelin in the PNS is largely composed of lipids (mainly cholesterol and sphingolipids) and exhibits narrow lamellar spacings, mediated by homophilic adhesion molecules (Filbin et al., 1990). Membranes of the paranodal loops are enriched for junctional proteins that permit intra-Schwann cell junctions and Schwann cell–axon junctions (Fannon et al., 1995; Boyle et al., 2001). Finally, the membranes of Schwann cell microvilli contain ligands for axonal cell adhesion molecules and contribute to sodium channel clustering at the nodes of Ranvier (Eshed et al., 2005).
One crucial mechanism underlying establishment and maintenance of cellular polarity is differential sorting of proteins to distinct plasma membrane domains along the secretory and endosomal pathways (Keller et al., 2001; Kreitzer et al., 2003). Several lines of evidence indicate that protein transport and targeting play an important role in myelin membrane assembly. During myelination, Schwann cells establish specialized microtubule networks, suggesting increased rates of vesicle transport (Trapp et al., 1995; Kidd et al., 1996). Sorting in the trans-Golgi network partitions myelin proteins into separate transport vesicles (Trapp et al., 1995) that are targeted to distinct Schwann cell compartments. Similarly, in polarized epithelial cells, transfected Schwann cell proteins are distinctly transported, potentially reflecting the segregation of compact and non-compact myelin components (Minuk and Braun, 1996; Kroepfl and Gardinier, 2001; Maier et al., 2006). Finally, myelin proteins undergo endocytic recycling in oligodendrocytes (Winterstein et al., 2008), which are the myelinating glia cells in the CNS. However, molecular pathways regulating vesicular transport during myelination have remained largely unknown. By demonstrating the SH3TC2/Rab11 interaction, we provide evidence for a distinct molecular mechanism, which probably involves the recycling of cargos that are critical for Schwann cell myelination. Sh3tc2-knockout mice may be instrumental for isolating such cargos by analysing the expression levels and transport routes of Schwann cell surface receptors that are known to undergo endocytic recycling and to be involved in the regulation of myelination.
While our immunoprecipitation, GST pull down and yeast-two-hybrid data clearly demonstrated the SH3TC2/Rab11 interaction, we were unable to define a shorter linear motif in the SH3TC2 protein which supports the interaction with Rab11 (Supplementary Fig. 2). Moreover, single amino acid substitutions disrupting the SH3TC2/Rab11 complex are spread throughout the protein. One potential explanation could be that mutations of certain residues result in misfolding of SH3TC2, ultimately leading to the loss of interacting or stabilizing regions. However, we did not observe degradation of SH3TC2 mutants when expressed in cells (data not shown). Therefore, a more likely explanation would be that the mutated residues are all found on the surface of the protein, forming an extended Rab11-interacting domain. In line with this hypothesis, SH3TC2 does not contain a known Rab11-binding motif found in several other Rab11-interacting proteins (Prekeris et al., 2001). Obviously, the presence of this motif is not a prerequisite to function as a Rab11 effector, as it does not occur in all Rab11-interactors (Wu et al., 2005; Westlake et al., 2007). An extended interaction domain consisting of several tetratricopeptide repeat repeats (which are contained in SH3TC2 as well) has been reported to enable p67phox to interact with Rac1, a small GTPase of the Rho family (Lapouge et al., 2000).
In summary, the molecular dissection of the function of SH3TC2, a protein mutant in a comparatively rare genetic form of peripheral neuropathies, suggests a link between PNS demyelination and a key biological mechanism, endocytic recycling. The identification of Schwann cell and myelin proteins that undergo recycling endosome-dependent transport will extend and improve our understanding of the pathogenesis of peripheral neuropathies and will eventually provide us with a list of new candidate genes and novel therapeutic targets. Assuming that endocytic recycling may assist morphogenesis of the myelin sheath by sorting and redirecting myelin components, it will be interesting to explore whether impaired endocytic recycling may be a common theme in demyelinating disorders of the PNS and maybe of the CNS as well.
During the preparation our manuscript, another article reporting the SH3TC2/Rab11 interaction and a potential effect of SH3TC2 on recycling endosome function was published (Roberts et al., 2010). While the data presented by Roberts et al. (2010) and by our group independently confirm that SH3TC2 is a new Rab11 effector molecule, we are going one step further and explore the role of Rab11 in Schwann cell myelination, demonstrating the relevance of the SH3TC2/Rab11 interaction for PNS pathology.
We thank Prof. M. McCaffrey (Department of Biochemistry, University College Cork, Ireland), Prof. L. A. Lapierre (Department of Surgery, Vanderbilt University Medical Center, Nashville, USA), Prof. B. Schlierf (Institut für Biochemie und Pathobiochemie, Friedrich-Alexander-Universität Erlangen-Nürnberg, Germany) and Prof. M. Zerial (Max Planck Institute for Molecular Cell Biology and Genetics, Dresden, Germany) for kindly providing us with Rab expression plasmids; and Prof. M. McCaffrey for her advice. J.S. is a Heisenberg fellow and C.S. a post-doctoral fellow of the Deutsche Forschungsgemeinschaft (DFG). A.R. received a PhD scholarship from RWTH Aachen University.
START program of RWTH Aachen University (to J.S.); the Interdisciplinary Centre for Clinical Research BIOMAT within the Faculty of Medicine; RWTH Aachen University (to J.S. and B.L.); the Swiss National Science Foundation (to R.C. and U.S.); the National Centre of Competence in Research; Neural Plasticity and Repair (to U.S.); Deutsche Forschungsgemeinschaft (to J.S. and B.L.).
Supplementary material is available at Brain online.
Charcot–Marie–Tooth neuropathy type 4C
Dulbecco’s modified Eagle’s medium
green fluorescent protein
phosphate buffered saline
peripheral nervous system
sodium dodecyl sulphate–polyacrylamide gel electrophoresis
Src homology 3 domain and tetratricopeptide repeats 2