Abstract
Dynamin 2 (DNM2) is a ubiquitously expressed GTPase implicated in many cellular functions such as membrane trafficking and cytoskeleton regulation. Dominant mutations in DNM2 result in tissue-specific diseases affecting peripheral nerves (Charcot–Marie–Tooth neuropathy, CMT) or skeletal muscles (centronuclear myopathy, CNM). However, the reason for this tissue specificity is unknown, and it remains unclear if these diseases share a common pathomechanism. To compare the disease pathophysiological mechanisms in skeletal muscle, we exogenously expressed wild-type DNM2 (WT-DNM2), the DNM2-CMT mutation K562E or DNM2-CNM mutations R465W and S619L causing adult and neonatal forms, respectively, by intramuscular adeno-associated virus (AAV) injections. All muscles expressing exogenous WT-DNM2 and CNM or CMT mutations exhibited reduced muscle force. However, only expression of CNM mutations and WT-DNM2 correlated with CNM-like histopathological hallmarks of nuclei centralization and reduced fiber size. The extent of alterations correlated with clinical severity in patients. Ultrastructural and immunofluorescence analyses highlighted defects of the triads, mitochondria and costameres as major causes of the CNM phenotype. Despite the reduction in force upon expression of the DNM2-CMT mutation, muscle histology and ultrastructure were almost normal. However, the neuromuscular junction was affected in all DNM2-injected muscles, with the DNM2-CMT mutation inducing the most severe alterations, potentially explaining the reduction in force observed with this mutant. In conclusion, expression of WT and CNM mutants recreate a CNM-like phenotype, suggesting CNM mutations are gain-of-function. Histological, ultrastructural and molecular analyses pointed to key pathways uncovering the different pathomechanisms involved in centronuclear myopathy or Charcot–Marie–Tooth neuropathy linked to DNM2 mutations.
Introduction
Dynamins are mechanoenzymes able to fission membrane and interact with cytoskeletons. These large GTPase enzymes are thus involved in many cellular processes ranging from cytoskeleton regulation to membrane remodeling and endocytosis (1–3). Classical dynamins are composed of several functional domains: a GTPase domain at the N-terminal which is able to hydrolyze GTP, a middle domain and a GTPase effector domain (GED) forming the stalk and involved in protein oligomerization, a pleckstrin homology (PH) domain binding phosphoinositides and a C-terminal proline-rich domain (PRD) which binds SH3-containing proteins (2,4). Dynamin 2 (DNM2), encoded by the DNM2 gene is the ubiquitously expressed classical dynamin. Despite its ubiquitous expression, dominant mutations in DNM2 have been associated to two different diseases affecting different tissues: centronuclear myopathy (CNM), and Charcot–Marie–Tooth peripheral neuropathy (CMT). It is unclear how different mutations in the ubiquitous exons of DNM2 lead to very different diseases and if the pathomechanisms of these two diseases partially overlap.
CMT neuropathies, or hereditary motor and sensory neuropathies (HMSNs), are a group of inherited peripheral neuropathies. The main symptoms are loss of the sensation of touch, pain, muscle weakness and atrophy (5,6). The classification of CMT into two forms relies on the histological and electrophysiological characteristics: demyelinating forms (CMT1) causing a defect in nerve conduction velocity, and axonal forms (CMT2) in which nerve conduction velocity is near normal (7). Additionally, forms of CMT with intermediate electrophysiology have been described (dominant intermediate, DI-CMT). In 2005, heterozygous mutations in DNM2 were reported to cause a DI-CMTB intermediate form (OMIM 606482) (8,9). Patients with DI-CMTB caused by the K562E mutation in DNM2 protein display mild reductions in nerve conduction velocity, loss of myelinated axons, rare segmental demyelination and remyelination with onion bulb formation and focal hypermyelination (9).
Heterozygous mutations in DNM2 have also been associated with autosomal dominant CNM (CNM1; OMIM 160150) (10,11). CNM is a congenital muscle disease characterized by muscle weakness and myofiber atrophy (12,13). Individuals affected by CNM present with facial weakness and delayed motor milestones. At the histological level, muscle fibers present abnormally localized nuclei at the center of the fiber, fiber type I predominance and atypical sarcoplasmic reticulum organization as radial strands (14). The severity of the disease is variable, and cases of DNM2 related-CNM have been reported with neonatal or adulthood onset (10,15). The most common mutation present in patients with the severe neonatal onset form is the S169L mutation. Severe hypotonia is often present, and patients may require mechanical ventilation (16). In comparison, the R465W mutation is the most common mutation found in patients with adolescent or adult-onset of the disease and is associated with milder disease symptoms (11). To date, a genotype–phenotype correlation has not been identified in vivo.
The cellular functions of DNM2 rely on its ability to hydrolyze GTP, oligomerize and bind lipids (1). Lipid binding promotes GTP activity that correlates with oligomerization. Several years ago, the tetrameric structure of the dynamin superfamily was solved which helped in predicting the effect of mutations related to CNM and CMT diseases (17). It was suggested that mutations may not strongly affect protein misfolding, but may rather impair its regulation. In vitro studies have shown that the K562E-CMT mutant was not able to bind lipids, in contrast to CNM mutants that were able to bind lipids in an equivalent manner to the WT protein (18). Furthermore, the GTPase basal activity was also differentially altered in R465W- and S619L-CNM versus K562E-CMT mutants: the CNM-related mutants had high GTPase activity irrespective of lipid binding while the CMT mutant displayed low GTPase activity even in the presence of lipids (18,19). In addition, CNM mutants formed abnormally stable oligomers (19). These combined studies suggest that various disease-causing DNM2 mutations affect different molecular properties of DNM2 function in vitro.
The reason for this tissue specificity and the heterogeneous severity of the different DNM2 mutations is currently unknown, and it remains unclear if these diseases share a common pathomechanism. To investigate these questions, we selected the most common DNM2-CNM mutations, R465W and S619L causing adult and neonatal forms respectively, and the DNM2-CMT mutation K562E, and compare the cellular and physiological effects of their exogenous expression in skeletal muscle in vivo in mice.
Results
Exogenous expression of CNM and CMT patient mutations in WT mouse muscle results in reduced muscle force
To investigate the effect of specific DNM2 mutations in skeletal muscle, we selected the DNM2-CNM mutations R465W and S619L causing adult and neonatal forms of CNM, respectively (10,15), and the DNM2-CMT mutation K562E causing CMT neuropathy (9) (Fig. 1A). The R465W-CNM mutation is located in the middle/stalk domain at the interface between dynamin dimers, whereas the S619L-CNM mutation is located in the PH domain, at the interface between the PH and stalk domains (Fig. 1B). While the K562E-CMT mutation is also in the PH-domain, 3D modeling suggests that this mutation is in a loop potentially contacting lipid membranes. Expression of patient mutations from human cDNA was achieved by performing intramuscular injections using AAV1 into the tibialis anterior (TA) muscles of 3-week-old wild-type mice. At this age, muscles have completed the early stages of postnatal muscle restructuring and growth and from 3–5 weeks undergo muscle hypertrophy to reach near-adult muscle mass, which will reach its maximum at 8 weeks of age (20). Mice were analyzed 2 weeks post injection at 5 weeks of age, and patient mutations were analyzed in comparison to wild-type DNM2 (WT-DNM2) or empty vector (empty) injection. No difference in TA muscle mass relative to body weight was detected between groups (Fig. 1C). Expression of WT-DNM2 resulted in a trend (P = 0.10) for reduced muscle force (Fig. 1D), as described previously (21). Expression of the K562E-CMT mutation through AAV1 injection into the muscle resulted in a significant drop in absolute muscle force in contrast to empty control muscles (P = 0.002). Expression of the R465W-CNM and S619L-CNM mutations also resulted in a strong reduction in force (P = 0.0003 and P < 0.0001, respectively). These results were confirmed on specific muscle force, relative to muscle mass (Fig. 1E). Exogenous expression of DNM2 proteins was confirmed for all DNM2-injected muscles by immunoblot analysis and densitometry, relative to empty-injected control (Fig. 1F and G). While WT-DNM2 and CNM-DNM2 mutations result in a 3–10-fold increase in DNM2 expression relative to endogenous levels, expression of the K562E-CMT mutation resulted in a stronger increase (approximately 25-fold) despite administration of equivalent AAV1 titers. The DNM2 RNA level was then investigated by RT-qPCR, and similarly to the protein level, DNM2 RNA in muscle was higher for the K562E-CMT mutation (Fig. 1H). Of note, there was an inverse correlation between the levels of DNM2 proteins and the decrease in force, with the K562E-CMT mutant being the most expressed with a mild reduction in muscle force in contrast to the S619L-CNM mutant being the least expressed with the strongest reduction in force. The R465W-CNM mutant was intermediate in expression and impact on the force. These data are consistent with the fact that the S619L-CNM mutation is observed in patients with the strong myopathic phenotype while the K562E-CMT mutant is observed in patients with a primary neuropathy (11).
AAV-mediated expression of CNM and CMT patient mutations in muscle results in reduced muscle force. (A) Domain organization of DNM2 and localization of the mutations studied in this work. The middle and GED domains form the structural stalk domain. (B) Mutations were visualized on equivalent amino acid of DNM3 published structure (4), showing CNM mutations (in red) and CMT mutation (pink) used in this study. (C) Ratio of TA muscle weight and total body weight (n = 4–6 muscles). (D) Absolute maximal muscle force measured 2 weeks after injection (n = 4–6 muscles). (E) Specific force corresponding to the absolute maximal muscle force divided by muscle weight (n = 4–6 muscles). (F) Skeletal muscle lysates immunoblotted for DNM2 and stained with Ponceau solution to determine total protein. (G) Relative expression of total DNM2 signals revealed by an anti-DNM2 antibody and standardized to Ponceau staining. Expression is represented as fold change in contrast to empty-AAV injected muscles (n = 4–6 muscles). (H) Transcript expression of the different DNM2 constructs quantified by RT-qPCR after DNAse treatment. Total DNM2 level was detected using primers recognizing DNM2 and Dnm2 and normalized to Rpl27 expression. Statistical analysis: Shapiro–Wilk normality test and Brown–Forsythe test for equal variances, followed by one-way ANOVA or Kruskal–Wallis ANOVA, followed by multiple comparisons test with Tukey–Kramer or Dunn’s correction; ***P < 0.0001; *P < 0.05; ns not significant. P values > 0.40 are not displayed in the graph.
In vivo expression of CNM and CMT patient mutations induce distinct histopathological phenotypes
We next determined whether the reduction in muscle force correlated with specific histopathological phenotypes. Transverse TA sections were stained with hematoxylin and eosin (HE) to determine the fiber size and nuclei position within muscle fibers. Indeed, decreased fiber size and centralization of nuclei are main hallmarks of CNM. In contrast to empty injected control muscles, R465W-CNM and S619L-CNM mutations induced strong nuclei mislocalization within the muscle fibers, with the latter mutant exhibiting the most severe alteration (Fig. 2A and B), consistent with this mutation inducing the most severe phenotype in patients (15). Similarly, WT-DNM2 also induced mislocalization of nuclei to a lesser degree, as shown previously (21). Furthermore, exogenous expression of CNM mutations and WT-DNM2 induced reduced fiber size (Fig. 2C), consistent with a CNM histopathological phenotype. These results support a gain-of-function mechanism in CNM. In contrast, nuclei positioning was not significantly altered in K562E-CMT-injected muscles (Fig. 2A and B), despite the strong increase in DNM2 expression (Fig. 1F and G). This correlates with normal muscle histology reported for a DNM2-CMT patient (23). In addition, fiber size was not affected (Fig. 2C). Overall, this suggests a potential loss-of-function mechanism in DNM2-CMT.
To further investigate the histological phenotype, muscles were stained with succinate dehydrogenase (SDH) and nicotinamide adenine dinucleotide (NADH), indicating oxidative activity from the mitochondria and the mitochondrial/reticulum, respectively. WT-DNM2 and DNM2-CNM mutations induced an abnormal staining pattern with accumulation towards the center and periphery of fibers, in both SDH and NADH staining (Fig. 2A). In addition, SDH staining was visibly reduced in the severe S619L-CNM mutation, suggesting a strong defect in mitochondrial function. Interestingly, in K562E-CMT muscles NADH staining was severely affected in contrast to SDH staining which was largely unaffected. Of note, slight sub-sarcolemmal accumulation of NADH was previously seen in the muscle of a DNM2-CMT patient (23). Overall, these results suggest that different pathomechanisms occur in CNM and CMT resulting in distinct histopathological features.
Ultrastructural defects in muscles overexpressing CNM but not CMT DNM2 mutations
As histological analysis suggested defects in myofiber organization as a potential cause of decreased force for the DNM2-CNM mutants, we next analyzed AAV1-transduced muscles at the ultrastructural level by transmission electron microscopy. Muscles transduced with R465W-CNM exhibited misaligned Z-lines, enlarged abnormally shaped mitochondria with associated membrane accumulations and altered triad structures (Fig. 3A). Triads are comprised of one t-tubule surrounded by two junctional sarcoplasmic reticuli. We investigated if this structure was abnormal in the different muscles (Fig. 3B). Quantitative analysis revealed that the number of identifiable triads per sarcomere was significantly reduced in R465W-DNM2-transduced muscles (Fig. 3C). This correlated with an increased circularity of t-tubules (Fig. 3D). The more severe S619L-CNM mutation exhibited the strongest increase in t-tubule circularity; however, one striking difference was the enlarged mitochondrial structures with the internal cristae network clearly perturbed (Fig. 3A), correlating with the reduced SDH staining observed (Fig. 2A). Overexpression of WT-DNM2 also induced some minor mitochondrial abnormalities, with misaligned Z-lines, altered triad number per sarcomere and increased circularity of t-tubules (Fig. 3A–D). These combined results suggest DNM2-CNM mutations induce CNM-like clinical, histological and ultrastructural phenotypes when overexpressed in vivo in WT muscles, and these phenotypes correlate with the CNM severity observed in patients.
In contrast, K562E-CMT expression did not induce any obvious ultrastructural defects apart from an increase in t-tubule circularity (Fig. 3A, B and D), consistent with the normal histological phenotype (Fig. 2A), and patient muscle biopsies (23).
Overexpression of CNM and CMT patient mutations induce distinct histopathological phenotypes. (A) Transverse 8 μm TA sections from AAV-transduced muscles stained with hematoxylin and eosin (HE), SDH or reduced NADH. Scale bar = 50 μm. (B) Percentage of myofibers with central, internal or peripheral nuclei (n = 400–500 fibers × 4–6 muscles). (C) Transverse 8 μm TA sections were analyzed for fiber diameter. Fiber diameter (minimum Feret’s diameter) is grouped into 5 μm intervals and represented as the percentage of the total fibers in each group (n = 400–500 fibers × 4–6 mice).
In vivo expression of CNM mutations results in ultrastructural defects. (A) AAV-transduced TA muscles imaged by transmission electron microscopy. Images show an overview of muscle sarcomeric ultrastructure, mitochondrial shape and triad structure. Scale bar = 2 μm. (B) Zoomed image in triad structure. (C) Graph representing the number of triads per sarcomere (mean ± SEM) (n = 10–30 sarcomeres × 2 mice per groups). (D) Circularity of t-tubule using 1 = perfect circle and 0 = straight line (mean ± SEM) (n = 10–30 t-tubules per group). Statistical analysis: Shapiro–Wilk normality test and Brown–Forsythe test for equal variances, followed by Kruskal–Wallis ANOVA, and multiple comparisons test with Dunn’s correction; ***P < 0.0001; *P < 0.05; ns not significant. P values > 0.40 are not displayed in the graph.
DNM2-CNM mutants are linked to defects in t-tubule and costamere markers
As ultrastructural analysis suggested t-tubule and triad defects in muscles expressing DNM2-CNM mutants, we next performed immunofluorescence analysis of dihydropyridine receptor alpha (DHPRalpha), a marker of t-tubules, on longitudinal sections. Expression of WT-DNM2 and K562E-CMT led to a DHPRalpha staining comparable to empty control (Fig. 4A). In contrast, both R465W-CNM and S619L-CNM displayed disorganized DHPRalpha staining patterns. This defect confirmed the observation of altered triad structures by electron microscopy and aligned with the severity of the CNM mutations, with muscle expressing the S619L-CNM mutant most severely affected.
Overexpression of DNM2-WT and DNM2-CNM mutants disrupt muscle organization. (A) Longitudinal 8 μm or TA sections were stained by immunofluorescence using antibodies against DHPR and DNM2. Scale bar = 20 μm. (B) Longitudinal 8 μm TA sections were stained by immunofluorescence using antibodies to detect caveolin-3 and CHC. Images displayed are projections of confocal stacks. Scale bar = 20 μm.
To further investigate the molecular mechanism that could explain the defects in muscle force observed with DNM2-CNM mutants, we focused on DNM2 and known functional partners. DNM2 localization in the sarcomere was previously reported to be associated to Z-lines (21). We investigated the localization of DNM2 on longitudinal muscle sections using an antibody recognizing both exogenous and endogenous DNM2 proteins. AAV1-empty-injected muscles displayed striated transverse staining on longitudinal sections (Fig. 4A) consistent with Z-line localization as seen previously (21). DNM2 appeared to be mislocalized in muscles expressing both CNM mutants. In particular, the more severe S619L mutation induced areas of longitudinal staining (Fig. 4A, arrow). In contrast, K562E-CMT exhibited a similar DNM2 localization to empty control muscles (Fig. 4A).
DNM2 plays an important role in clathrin-mediated endocytosis (2,24) and together with clathrin participates in the correct attachment of the muscle fibers to the extracellular matrix (25). Immunolabeling of clathrin heavy chain (CHC) was performed on longitudinal muscle sections from TA muscles. In AAV1-empty injected muscles, CHC exhibited the expected transverse staining pattern reminiscent of costameres (Fig. 4B). Muscles transduced with K562E-CMT exhibited a transverse staining pattern indistinguishable from control sections. However, muscles transduced with R465W-CNM and S619L-CNM showed a more disorganized staining pattern for CHC. Additional peri-nuclear staining in WT-DNM2 and S619L-CNM transduced muscles was also observed. Caveolin-3 is an important regulator of clathrin-independent endocytosis and participates in t-tubule biogenesis. In adult muscle, caveolin-3 is localized at the sarcolemma. We did not detect obvious changes in caveolin-3 distribution in any groups (Fig. 4B).
Dysferlin is another muscle protein implicated in t-tubule biogenesis, membrane trafficking and membrane repair. While dysferlin localization appeared normal upon expression of the K562E-CMT mutant, expression of WT-DNM2 or DNM2-CNM mutants induced an abnormal accumulation of dysferlin around the centralized nuclei (Supplementary Material, Fig. S1A).
Overall, the structural defects induced by DNM2-CNM mutants parallel disorganization of t-tubule and costamere markers and are probably the cause of decreased muscle force. Conversely, expression of the K562E-CMT mutant did not have a strong impact on muscle fiber organization, raising the question of the molecular mechanism for the decrease in muscle force observed.
DNM2 overexpression leads to neuromuscular junction defects
Muscle force generation depends on muscle innervation through the neuromuscular junction (NMJ) formed by contact between a motor neuron and a muscle fiber. Neuromuscular junctions were visualized with fluorescently labeled alpha-bungarotoxin which stains nicotinic acetylcholine receptors located in NMJs. In contrast to AAV1-empty control muscles, all muscles injected with the different DNM2 constructs exhibited a strong alteration in NMJ shape that appeared more elongated (Fig. 5A). We calculated the circularity for different NMJs present in transduced muscles. In particular, there was a tendency for reduced circularity, in contrast to control muscles, in all NMJ from the different constructs, except for CNM-R465W (Fig. 5B). In addition, NMJ fragmentation was highly increased in all the analyzed constructs in contrast to NMJ from empty-injected muscles (Fig. 5C). However, the surface of NMJs was similar between DNM2 constructs and AAV1-empty control (Fig. 5D). Moreover, the number of NMJs in the analyzed muscles was strongly reduced in K562E-CMT-expressing muscles (Fig. 5E). The largest reduction was observed in K562E-CMT-injected muscles. Overall, NMJ defects correlated with decreased muscle force with all DNM2 constructs including WT-DNM2. The strongest defects observed with the CMT mutant could explain in part the reduction in force observed in injected muscles, despite the lack of significant muscle histopathological features.
Neuromuscular junction defects due to the overexpression of DNM2-WT, CMT and CNM mutants. (A) Images of representative neuromuscular junctions labeled with using fluorescent alpha-bungarotoxin. Scale bar = 20 μm. (B) NMJ shape; circularity was calculated (value of 1 indicates a perfect circle, as the value approaches 0 it indicates an increasingly elongated shape) (n = 1–3 NMJ from 2 mice). (C) Number of fragments per NMJ (n = 1–3 NMJ × 2 mice). (D) NMJ Area (n = 1–3 NMJ × 2 mice). (E) Graph representing the number of NMJs per muscle area calculated using different images of longitudinal 8 μm TA muscle sections. Kruskal–Wallis ANOVA, followed by multiple comparisons test with Dunn’s correction; ***P < 0.0001; *P < 0.05; ns not significant. P values > 0.40 are not displayed in the graph.
Discussion
While DNM2 is ubiquitously expressed, different dominant point mutations in DNM2 result in two tissue-specific disorders. It was unclear however if these diseases arise from a common pathomechanism. In this study, we investigated the molecular, cellular and physiological alterations induced by DNM2 mutations in two distinct human diseases: CNM (neonatal and adult onset forms) and Charcot–Marie–Tooth neuropathy, with a focus on skeletal muscle. Moreover, the basis of the difference in severity between the mutations leading to CNM was not previously defined.
Differential functional impact of CNM and CMT mutations
CNM patients present with a myopathic phenotype varying from neonatal to adult onset, with a wide range of severity. Most often symptoms are limited to skeletal muscles. Conversely, CMT neuropathy patients exhibit predominantly neuronal symptoms. Bitoun et al. reported no obvious muscle defects in a biopsy of a CMT patient with the K559del mutation in the PH domain (23). However, another patient with a DNM2 mutation exhibiting both CNM and CMT-like features has been reported (26). This patient was carrying a mutation in the middle/stalk domain (G359D), where typically CNM mutations are located and close to another mutation already reported in CNM. Histological analysis of the patient’s muscle revealed variation in fiber size and fiber atrophy, and a sural nerve biopsy showed a severe loss of large myelinating fibers (26). Nevertheless, the effect of CMT-related mutations on skeletal muscles has not been extensively explored.
Muscle weakness is common to both CNM and CMT patient presentation (11). Here, we showed that the exogenous expression of CNM and CMT DNM2 mutations led to muscle force decrease in mice. However, expression of both CNM mutations led to muscle structural alterations mimicking the histopathological hallmarks seen in CNM patients with hypotrophic myofibers with mislocalized nuclei, central accumulation of oxidative activity, decreased triads number and abnormal t-tubule shape, while expressing a CMT mutant left the muscle structure almost intact.
Importantly, overexpression of WT-DNM2 also caused the CNM phenotype with the same histological and structural findings, as shown previously (21), suggesting increasing DNM2 may be pathogenic for skeletal muscles. This is supported by data suggesting overexpression of DNM2 may be in part responsible for the CNM phenotype observed in the X-linked form of the disease due to mutations in the lipid phosphatase MTM1. Indeed, DNM2 protein level was found increased in muscle from mice and patients lacking MTM1, and normalization or decrease of DNM2 through transgenesis, shRNA or antisense oligonucleotides prevented and reverted the X-linked CNM in mice (27–30). DNM2 reduction with similar approaches also ameliorated the phenotypes of the transgenic mouse mice Dnm2R465W/+ (31), supporting that reduction of DNM2 overall activity counteracts DNM2-CNM mutation effects. Of note, the heterozygous KO Dnm2+/− mice do not develop CNM suggesting that CNM mutations are not loss-of-function (27).
Most mutations causing CNM are localized at the interface between the middle/stalk domain and the PH domain (4,32). This interface was proposed to have an autoinhibitory effect on the PH domain insertion into the membrane and consequently dynamin oligomerization and GTPase activity. As CNM mutations located there disrupt potential hydrogen bounds between the middle/stalk and PH domains, it is expected that CNM mutations alleviate the autoinhibition, leading to DNM2 ‘hyperactivation’. In agreement, in vitro studies showed that CNM mutations induce a higher GTPase hydrolysis activity of DNM2 regardless of their binding to lipids, combined with increased oligomer stability (18,19). Several studies in cells also suggested CNM mutations increase the oligomeric state and stability of DNM2 (33,34), and endocytosis of transferrin was found increased in murine muscle cells with the R465W mutation and in CNM patient myoblasts with R465W or R369Q mutations (35).
Altogether, our present in vivo data combined to the literature support that the DNM2-CNM mutations are gain-of-function.
Conversely, expressing the K562E-CMT mutant in WT muscles did not cause any of the CNM histopathological hallmarks with the exception of an increase in t-tubule circularity. In particular, myofiber size and nuclei position were normal (Fig. 2A–C), despite the high level of protein expression (Fig. 1F and G). Our data suggest that DNM2-CMT mutations are not gain-of-function. Most mutations causing CMT are localized in the loops of the PH domain implicated in membrane recognition, and in vitro studies showed CMT mutants exhibit a strong reduction in lipid binding (18). The pathogenesis of CMT and CNM mutations has previously been investigated in cellular systems modeling peripheral nerves. Decreased myelination was observed in cultured motor neurons from dorsal root ganglia explants only upon expression of CMT mutants but not CNM mutants (36). In the same cellular system, DNM2-CMT mutants impaired clathrin-mediated endocytosis in motor neurons and Schwann cells, whereas CNM mutants had no effect. Furthermore, overexpression of CNM but not CMT mutations could rescue endocytosis defects observed in Schwann cells from Dnm2-deficient mice (36). In vivo, Dnm2 deletion in Schwann cells impaired myelination and deletion in adult mice caused demyelination (37), reminiscent of histopathology seen in CMT patients and suggesting that the disease is due to decrease of DNM2 in peripheral nerves. In addition, our finding that AAV1 transduction of the K562E-CMT mutant leads to a significant increase in its RNA expression can be explained if this mutant decreases the overall activity of DNM2, in light of recent data showing that decreasing DNM2 promotes adenovirus replication probably by increasing the release of virus from endosomes (22). Taken together, DNM2-CMT mutants are most probably loss-of-function.
Overall, our in vivo data support that DNM2-CNM mutations are gain-of-function while DNM2-CMT mutants are loss-of-function, suggesting different pathomechanisms are involved in the two diseases.
Different pathomechanisms for DNM2 mutations related to CNM and CMT
The exogenous expression of different DNM2 mutants in muscle revealed marked differences in the structure and organization of the neuromuscular system that may underlay the common decrease in muscle force. In particular, expression of the K562E-CMT mutant is not associated to strong structural defects of the myofibers but correlated with pronounced structural defects of the neuromuscular junctions. For DNM2-CNM mutants, alterations of both NMJ and myofibers potentially cause muscle weakness; obviously, the strong abnormalities in triads, mitochondria and the sarcomere structure should at least partially decrease muscle force. Overall, we propose that muscle weakness in CNM is induced largely from alteration of the myofiber structure while muscle weakness in CMT mainly comes from NMJ defects.
While our AAV serotype 1 injections were intramuscular, we cannot exclude the possibility that AAV1 underwent retrograde transport resulting in expression of DNM2 constructs in motorneurons, as previously found with AAV serotype 2 (38). Primary muscle defects have been shown to induce NMJ defects. For example, muscle-specific deletion of Dnm2 induced structural defects of myofibers with metabolic alterations, together with irregular NMJs and peripheral nerve damage (39). Also, overexpression of uncoupling protein 1 (UCP1) in skeletal muscle led to defects in mitochondrial function resulting in secondary NMJ defects and degeneration of motor neurons (40). Bragato et al. overexpressed human DNM2 constructs in zebrafish, harboring different mutations than the one tested here (the R522H CNM mutant and the G537C CMT mutant) and also found that they both impacted on the structure of NMJ (41). Conversely, the CMT mutant also induced a high number of central nuclei in the myofibers. This discrepancy may be due to the different animal models used or the fact that there are two orthologs of DNM2 in zebrafish.
Both CNM mutants tested in this study induced mitochondrial shape defects (Fig. 3A). As mice with muscle-specific deletion of Dnm2 display abnormal mitochondria as well as increased NMJ area (39), the balance in DNM2 activity is potentially important for mitochondria homeostasis. Contradictory reports were recently published on a direct role of DNM2 in mitochondria fission (42, 43); whether mitochondria structural defects we observed, especially with the S619L-CNM mutant, are caused by a direct role of DNM2 on mitochondria dynamics thus remains unclear. When muscles were injected with K562E-CMT mutant, no obvious disruption in mitochondria was observed, however NMJs were clearly affected (Figs. 3 and 5). DNM2 may affect mitochondria or be directly involved in the maintenance of the NMJ.
The S619L-CNM expression induced a more severe muscle phenotype than R465W-CNM, in alignment with the severity and disease age of onset observed in patients with these mutations. Our in vivo models thus recapitulate faithfully the genotype–phenotype correlation seen in patients, supporting that the increased severity and earlier onset linked to the S619L mutation is not due to genetic modifiers or environmental differences with patients harboring other mutations but is an intrinsic property of the mutated residue.
Conclusions
In this study using AAV1-mediated gene expression, we established in vivo models to study DNM2 mutations linked to CNM and CMT in the neuromuscular system. Expression of WT and CNM mutants recreated a CNM-like phenotype, suggesting CNM mutations are gain-of-function. Expression of the CMT mutant in muscle did not lead to severe structural or ultrastructural defects despite similar reduction in muscle force, suggesting that different pathomechanisms are involved in the two diseases.
Materials and Methods
Materials
pAAV plasmids were generated as described before (21) containing full-length human isoform DNM2 cDNA (NCBI Reference Sequence: NM_001005360.2). The different mutations were introduced by primer-directed PCR mutagenesis. All constructs were verified by sequencing.
Primary antibodies used were anti-DNM2 (dilution used 1/100, 2680, described in (21)), anti-DHPR (Santa Cruz Biotechnology; sc-8160; dilution used 1/50), anti-CAV3 (Santa Cruz Biotechnology; sc-5310; dilution used 1/50) and anti-CHC (Abcam; ab21679; dilution used 1/100). Secondary antibodies used were donkey anti-goat Alexa-488, donkey anti-rabbit Alexa-594, goat anti-mouse Alexa-488 and goat anti-rabbit Alexa-594 (Life Technologies). The dilution used for all secondary antibodies was 1/250. To detect neuromuscular junction, we used alpha-Bungarotoxin CF®488A Conjugate (Biotium).
Production and purification of AAV
Recombinant adeno-associated virus serotype 1 were generated by a triple transfection of HEK293T-derived cell line with the expression plasmid pAAV-DNM2 and the auxiliary plasmids pHelper (Agilent) and pXR1 for AAV serotype 1 (UNC Vector Core). DNM2 wild-type and mutated forms were cloned under the control of the CMV promoter in the pAAV-MCS (Agilent). AAV vectors were harvested 48 h after transfection from cell lysate treated with 100 U/ml Benzonase (Merck). AAV1 were purified by iodixanol gradient ultracentrifugation (OptiPrep™, Axis Shield) followed by dialysis and concentration against Dulbecco’s PBS containing 0.5 mm MgCl2 using centrifugal filters (Amicon Ultra-15 Centrifugal Filter Device 100 K). Viral titers were determined by Q-PCR using the LightCycler480 SYBR Green I Master (Roche) and primers targeting the CMV enhancer sequence. Viruses were stored at −80°C until use.
Animals
WT 129/SvPAS mice were handled according to the French and European legislation on animal care and experimentation. Protocols were approved by the institutional Ethics Committee. Protocol No. APAFIS #5640-2016061019332648 v4 was granted to perform animal experiments. Mice were kept on a 12 h light/12 h dark cycle in ventilated cages and given free access to food. All mice analyzed in this study were male. Mice were numbered and after AAV1-injection muscles were analyzed blindly.
Intramuscular injection of AAV1
Three-week-old male WT 129/SvPAS were weighed and anesthetized with a solution of ketamine 20 mg/ml and xylazine 0.4% at 5 μl/g of body weight. The solution was administrated by intraperitoneal injection. Both TA muscles were injected with 20 μl of 5 × 1011 vg/ml AAV1 encoding DNM2 (either wild type-DNM2 or including K562E, R465W or S619L mutations), and the same dose of AAV1 containing an empty AAV construct was injected in the contralateral TA as a control.
Muscle contractile properties
Muscle force measurements were evaluated by measuring in situ muscle contraction in response to nerve stimulation using the Complete 1300A Mouse Test System (Aurora Scientific) as described previously (44). Animals were anesthetized (intraperitoneal injection of pentobarbital sodium, 50 mg/kg) and maintained under deep anesthesia. The distal tendon of the TA was detached and tied to an isometric transducer. The sciatic nerve was stimulated, and response to stimulation (pulse frequency of 1–125 Hz) was recorded to measure absolute maximal force. To determine specific maximal force, TA muscles were dissected and weighed. Muscles were then stored for further analysis.
Protein extraction and western blot
TA muscle cryosections were lysed in RIPA buffer supplemented with PMSF 1 mm and complete mini EDTA-free protease inhibitor cocktail (Roche Diagnostic). Protein concentrations were determined with the Bio-Rad Protein Assay Kit. Samples were denatured at 95°C for 5 min. Then, 15 μg of protein in 5× Lane Marker Reducing Buffer (Thermo Fisher Scientific) were separated in 10% SDS-PAGE gel and transferred on nitrocellulose membrane for 7 min at 2.5 A using a Trans-Blot Turbo Transfer System (Bio-Rad). Total protein was determined by Ponceau S staining. Membranes were blocked for 1 h in TBS containing 5% non-fat dry milk and 0.1% Tween20 before an incubation overnight with primary rabbit polyclonal antibodies against DNM2 2865 (1:500, described in (21)) diluted in blocking buffer containing 5% milk. The secondary antibody coupled to horseradish peroxidase was goat anti-rabbit (Jackson Immunoresearch) (1:10 000) and was incubated for 2 h. Nitrocellulose membranes were visualized in Amersham Imager 600 (GE Healthcare Life Sciences). Images from full western blot membranes are shown in Supplementary Material, Fig. S1B.
RNA extraction and qRT-PCR
Total RNA was isolated from muscle tissue using with TRI Reagent (Molecular Research Center). To eliminate possible detection of DNA, DNAseI treatment was applied to the samples according to the user guide provided (Thermo Fischer Scientific). cDNA synthesis was performed using Superscript IV Reverse Transcriptase (Thermo Fisher Scientific). Quantitative PCR was done using cDNA amplified with SYBR Green Master Mix I (Roche) together with 0.1 μM forward and reverse interexonic primers. Amplicons were analyzed with a LightCycler® 480 (Roche). Primers used were DNM2 and Dnm2: (F) TGATCCTGCAGTTCATCAGC, (R) ATGACACCGATGGTCCGTAG, Rpl27 (45) (F) AAGCCGTCATCGTGAAGAACA, (R) CTTGATCTTGGATCGCTTGGC.
Muscle histology
TA muscles were snap-frozen in liquid nitrogen-cooled isopentane and stored at −80°C for hematoxylin and eosin (H&E), SDH or reduced NADH histology analysis. Transversal cryosections (8 μm) were prepared and stained. Entire muscle sections were imaged with the Hamamatsu 322 NanoZoomer 2HT slide scanner. The percentage of TA muscle fibers with mislocalized (centralized or internalized) nuclei was counted using the cell counter plugin in Fiji image analysis software. The fiber area was measured using the Fiji software. A total of 400–500 fibers from four to six mice were analyzed per group.
Electron microscopy
Transmission electron microscopy (TEM) was carried out on TA muscles fixed in 2.5% paraformaldehyde, 2.5% glutaraldehyde and 50 mm CaCl2 in 0.1 M cacodylate buffer (pH 7.4) as described previously (28). Briefly, sections of 70 nm were obtained from TA muscles and stained with uranyl acetate and lead citrate. They were observed by TEM (Morgagni 268D, FEI). The ratio of triads to sarcomere was calculated by dividing the number of triads identified by the number of sarcomeres present in the field. A total of 10–15 sarcomeres from two mice were analyzed per group.
Immunostaining of muscle transversal sections
For longitudinal immunostaining, TA muscles were fixed in PFA 4% for 24 h, then transferred to 30% sucrose for 12 h and stored at 4°C, as described previously (28). Isopentane-frozen muscles were used to perform transversal stainings. Transversal or longitudinal (8 μm) cryosections of TA were stained with primary antibodies and secondary antibodies listed in Materials and Methods section. Images were taken in the same Leica SP8-UV confocal microscope (Leica Microsystems).
Statistical analysis
For this study, n = 4–5 mice were used. Bar charts show mean ± SEM. All graphs were made with GraphPad Prism software. Normality was tested using the Shapiro–Wilk normality test for each condition to determine if parametric tests were applicable. Differences between groups were analyzed by t test or one-way ANOVA followed by post hoc Tukey’s or Dunn’s multiple-comparison test.
Funding
Institut National de la Santé et de la Recherche Médicale, Centre National de la Recherche Scientifique, Strasbourg University; French State fund managed by the Agence Nationale de la Recherche under the frame program Investissements d’Avenir (ANR-10-IDEX-0002-02, ANR-10-LABX-0030-INRT) and Association Française Contre les Myopathies (AFM 21267-RESCUE). X.M.M. is an Institut de Génétique et de Biologie Moléculaire et Cellulaire International PhD Programme fellow supported by LabEx INRT funds.
Authors’ contribution
B.S.C. and J.L. designed and supervised the research; X.M.M. and S.B. performed the research; X.M.M., B.S.C. and J.L. analyzed the data; X.M.M., B.S.C. and J.L. wrote the manuscript.
Acknowledgements
We would like to thank Raquel Gómez Oca and Nadia Messaddeq for excellent technical assistance and Pascal Kessler for building the macro to analyze histological data. We thank the imaging facility (photonic and electron microscopy), animal house and histology platform of the IGBMC for support.
Conflict of interest statement. B.S.C. and J.L. are cofounders of Dynacure. B.S.C. and S.B. are currently employed by Dynacure, and J.L. is a scientific advisor.
References
Author notes
The authors wish it to be known that, in their opinion, the last two authors should be regarded as equal contributors.
