Dynamin-2 mutations associated with centronuclear myopathy are hypermorphic and lead to T-tubule fragmentation

Abstract

Skeletal muscle requires adequate membrane trafficking and remodeling to maintain its normal structure and functions. Consequently, many human myopathies are caused by mutations in membrane trafficking machinery. The large GTPase dynamin-2 (Dyn2) is best known for catalyzing membrane fission during clathrin-mediated endocytosis (CME), which is critical for cell signaling and survival. Despite its ubiquitous expression, mutations of Dyn2 are associated with two tissue-specific congenital disorders: centronuclear myopathy (CNM) and Charcot–Marie–Tooth (CMT) neuropathy. Several disease models for CNM-Dyn2 have been established to study its pathogenic mechanism; yet the cellular and biochemical effects of these mutations are still not fully understood. Here we comprehensively compared the biochemical activities of disease-associated Dyn2 mutations and found that CNM-Dyn2 mutants are hypermorphic with enhanced membrane fission activity, whereas CMT-Dyn2 is hypomorphic. More importantly, we found that the expression of CNM-Dyn2 mutants does not impair CME in myoblast, but leads to T-tubule fragmentation in both C2C12-derived myotubes and Drosophila body wall muscle. Our results demonstrate that CNM-Dyn2 mutants are gain-of-function mutations, and their primary effect in muscle is T-tubule disorganization, which explains the susceptibility of muscle to Dyn2 hyperactivity.

Introduction

Muscle is a unique tissue with elaborated membrane and cytoskeleton organization designed for excitation–contraction coupling and force generation (1). Many human myopathies are caused by mutations of proteins responsible for membrane trafficking, and one of them is the critical regulator of clathrin-mediated endocytosis (CME), dynamin-2 (Dyn2) (2,3).

Dynamin is best known for catalyzing membrane fission to release clathrin-coated vesicles from the plasma membrane and is essential for synaptic vesicle recycling, cell signaling and growth (4–6). Dynamin is composed of five domains: an N-terminal GTPase domain, followed by the middle and pleckstrin homology (PH) domains, the GTPase effector domain (GED), and the C-terminal proline/arginine-rich domain (PRD) (Fig. 1A). The GTPase domain binds and hydrolyzes GTP, and the middle and GED interact with each other to form the stalk region for dynamin self-assembly (7,8). The PH domain binds phosphatidylinositol-4,5-bisphosphate (PI(4,5)P₂), and the PRD binds to SH3 domain-containing partners and targets dynamin to clathrin-coated pits (CCPs) (9,10). During CME, dynamin is recruited to nascent CCPs through interactions with the BAR and SH3 domain-containing proteins endophilin and/or amphiphysin (6,11,12). Dynamin then binds to PI(4,5)P₂ through its PH domain and self-assembles into a spiral structure at the highly curved neck of mature CCPs, where it directly catalyzes membrane fission through GTP-hydrolysis-dependent powerstroke mechanism (13,14).

Figure 1.

Biochemical properties of disease-associated Dyn2 mutations. (A) Domain structure of Dyn2 and disease-associated mutations analyzed in this study. (B and C) Membrane fission activities of purified Dyn2 proteins. The indicated concentrations of Dyn2 proteins were incubated with SUPER templates for 30 min at room temperature in the presence of GTP. Membrane fission was measured by the release of fluorescent vesicles from SUPER template. Fission activity of Dyn2 mutants alone (B) or Dyn2 with purified N-BAR domain (C) was compared with wild-type at the same concentration by Student's t-test. *P < 0.05. (D) Curvature sensitivity of Dyn2. The GTPase activity of Dyn2 was measured in the presence of LN or LUV of different sizes. (E) Tubulation of SUPER templates. An aliquot of 0.5 µm Dyn2 or N-BAR domain was incubated with SUPER templates for 10 min in the absence of GTP. Arrowheads indicate the tubules. Scale bar, 10 µm. (F) Membrane fission activities of CNM-associated Dyn2 mutant proteins. Additional CNM-Dyn2 mutations were purified and analyzed for their fission activities as described in (B). All values reported in this study represent the mean ± SD of at least three independent experiments.

Mammals express three dynamins in a tissue-specific manner. Dyn2 is ubiquitously expressed yet mutations in Dyn2 lead to two autosomal dominant, tissue-specific diseases: centronuclear myopathy (CNM) and Charcot–Marie–Tooth (CMT) disease (3,15,16). CNM is a group of congenital muscular disorders characterized by progressive muscle weakness, myofiber atrophy, abnormal nuclear centralization and disorganization of triads, which are the membrane structures composed of two sarcoplasmic reticulum cisternae located on both sides of a sarcolemma invagination called T-tubule (15,17). In contrast, CMT is a group of inherited neuronal disorders affecting both sensory and motor neurons (18). It has been reported that the CNM-associated mutations cause Dyn1 or Dyn2 to form higher oligomers in solution (19,20), and expression of one of these CNM-Dyn2 mutants resulted in disorganized mitochondria, triad, neuromuscular junction (NMJ) and sarcomere structure in muscle of mouse or zebrafish (21–24). However, the molecular pathomechanism of these mutations remains unclear: whether CNM-related Dyn2 mutants have abnormal membrane remodeling activities and what would the primary effect of these mutant proteins be in the muscle.

Given that different mutants are associated with different diseases, we hypothesize that CNM- and CMT-related mutations must have distinct effects on the biochemical and cellular activities of Dyn2. To test this, we have conducted a comprehensive comparison of Dyn2 mutants from CNM or CMT regarding their biochemical activities, including membrane binding, self-assembly, GTP hydrolysis and membrane fission abilities. In parallel, we also examined the cellular effects of these Dyn2 mutants within C2C12 murine myoblasts as well as muscles in Drosophila melanogaster. Our biochemical results demonstrated that CNM-Dyn2 mutants are hypermorphic with enhanced membrane fission activity, and our cell-/animal-based experiments demonstrated that the hyperactivity of Dyn2 would lead to T-tubule disorganization in muscle.

Results

CNM-associated mutations result in Dyn2 hyperactivity

To examine the biochemical properties of disease-associated Dyn2 mutants, we purified these recombinant mutant proteins from Sf9 insect cells. We studied three CNM-associated mutants, p.Arg465Trp (R465W), p.Ala618Thr (A618T) and p.Ser619Leu (S619L), and two CMT-associated mutants, p.Gly537Cys (G537C) and p.Lys562Glu (K562E) (Fig. 1A and Supplementary Material, Fig. S1A). These were chosen because they are either the most frequent mutant alleles (R465W and K562E) or localized within the same domain (25).

We first examined the ability of these proteins in catalyzing membrane fission from supported bilayer with excess membrane reservoir (SUPER) templates (26,27). Previous studies reported that Dyn2 is less effective than Dyn1 in catalyzing fission, and Dyn2 could act in concert with the curvature-generating N-BAR domain to drive vesicle release more efficiently from SUPER templates (26,28). We found that the CNM-Dyn2 mutants exhibit higher fission activity than WT-Dyn2, whereas CMT-Dyn2 exhibits reduced fission activity compared with WT (Fig. 1B). To further distinguish them, we measured their fission activity in the presence of endophilin N-BAR domains. Consistent with the previous studies, the N-BAR enhanced the fission efficiency of the WT-Dyn2 and CNM-Dyn2 mutants, whereas the CMT-Dyn2 mutants remained significantly impaired (Fig. 1C).

Dyn2 has been reported to be highly dependent on membrane curvature for binding and assembling onto a membrane template to stimulate its GTPase activity (28). When WT-Dyn2 is incubated with different sized liposomes to provide templates with different membrane curvatures, we observed the expected elevation in stimulated GTPase activity upon membrane curvature increase (Fig. 1D). In contrast, CNM-Dyn2 has lost its curvature sensitivity. We found that the CNM-Dyn2 mutants exhibit abnormally high GTPase activity on 1000 nm large unilamellar vesicle (LUV), which is indistinguishable from highly curved lipid nanotube (LN). Furthermore, CNM-Dyn2 could also tubulate SUPER template (planar membrane) in the absence of GTP, whereas WT-Dyn2 could not (Fig. 1E). On the contrary, the CMT-Dyn2 mutants showed impaired assembly-stimulated GTPase activity regardless of the liposome template, but especially with larger liposomes (Fig. 1D and E). To draw a general conclusion, we also examined the membrane fission activity of other CNM-Dyn2 mutations, including E368K, E560K, Δ625 and E650K, and found that these mutants have higher fission ability than WT-Dyn2 (Fig. 1A and F). Together, these results demonstrate that CNM-Dyn2 mutants are hyperactive with regard to their GTPase, membrane tubulation and fission activities.

CNM-associated mutations result in abnormal self-assembly of Dyn2

To investigate how these CNM-Dyn2 mutants become hyperactive, we next analyzed their binding and self-assembly onto membranes by sedimentation assay (29). WT-Dyn2 bound more efficiently to highly curved LN than to 100 nm LUV, whereas CNM-Dyn2 bound to 100 nm LUV as efficiently as LN (Fig. 2A). In contrast, the CMT mutant G537C could only bind to highly curved LN, whereas K562E completely lost its membrane binding ability. Electron microscopy (EM) analysis of assembled proteins showed that the spirals formed by CNM-Dyn2 on LN were less ordered with irregular periodicity when compared with WT (Fig. 2B and C). Remarkably, we often observed ring structures formed by Dyn2S619L lying on the grid without associating with the lipid template (bottom right panel, Fig. 2B). The diameter of Dyn2S619L rings ranged from 80 to 100 nm, and this structure was reminiscent of the dynamin ring assembled in low-salt condition (30). This result indicates that abnormal self-assembly may underpin the hyperactivity of CNM-Dyn2. Indeed, the Dyn1 proteins bearing A618T and S619L mutations have shown higher basal GTPase activity and formed higher oligomers in solution (19). To test this hypothesis, we first introduced the assembly-deficient mutant, R399A (29), into CNM-Dyn2 and found that R399A brings the higher basal GTPase activity back to WT levels (Fig. 2D). The much higher basal GTP hydrolysis of Dyn2S619L than those of R465W and A618T implies that S619L has higher propensity to self-assemble into oligomers. Indeed, this speculation is demonstrated by EM images in which S619L could form ring-like oligomers without binding to membrane (Fig. 2B).

Figure 2.

Membrane binding and self-assembly activity of disease-associated Dyn2 mutations. (A) Membrane binding of Dyn2 mutations onto LN or LUV. An aliquot of 0.5 µm purified Dyn2 was incubated with 150 µm LN or 100 nm LUV for 5 min at 37°C. After centrifugation, the Dyn2 fraction in the pellet was considered as ‘membrane-bound’ fraction. (B) Electron micrographs of Dyn2 mutants assembled onto LN. Bar, 100 nm. (C) Quantification of the periodicity of WT- or CNM-Dyn2 assembled on LN. n = 15. (D) Basal GTPase activity of Dyn2 with or without R399A mutations. (E) Liposome binding of different Dyn2 proteins in the presence or absence of GTP; 0.5 µM Dyn2 was incubated with 100 nm LUV with (gray bars) or without (black bars) 1 mm GTP at 37°C for 5 min, and the membrane-bound Dyn2 fraction was quantified by the sedimentation assay. (F) Fraction of disassembled Dyn2 from (E). n.s., not significant.

Secondly, we reasoned that hyper-assembly may prevent Dyn2 disassembly; therefore, we examined disassembly of the mutants upon GTP incubation (Fig. 2E). Similar to Dyn1 (31), WT-Dyn2 dissociated from membranes upon GTP addition; however, the dissociation efficiency of CNM-Dyn2 was lower (Fig. 2F). Together, we demonstrate that CNM-Dyn2 mutants are hypermorphic and exhibit enhanced, but altered, self-assembly properties. Notably, the CMT-Dyn2 mutants G537C and K562E show different activities upon highly curved membrane template. K562E completely lost its membrane binding and membrane-stimulated GTPase activity, whereas G537C could still bind to and be activated by LN (Figs 1D and 2A). The high dependence on the membrane curvature of G537C, which locates in the variable loop 1 (VL1) of PH domain, is similar to the membrane insertion-defective mutant Dyn1I533C, which also locates at VL1 (32). These results indicate that K562E is a membrane-binding-defective mutant but G537C is defected in curvature generation.

Effects of CNM-dynamin-2 in C2C12-derived myotubes

To explore the cellular effects of these hyperactive Dyn2 mutants in disease-relevant cells, we took advantage of the mouse myoblast C2C12 cell line that can undergo differentiation, fusion and myotube formation in culture. We first examined the expression of endogenous Dyn2 in C2C12 and found that its protein level increased upon myoblast differentiation (Fig. 3A). Similarly, many muscle-specific proteins, including BIN1 (amphiphysin II), caveolin-3 and myosin heavy chain (MyHC), were also induced upon differentiation, but not the endocytic accessory dynamin-interacting protein, SNX9 (Fig. 3A).

Figure 3.

Expression and localization of Dyn2 in C2C12. (A) Expression of Dyn2 and muscle-specific proteins in dividing and differentiated C2C12 myoblasts. Cell lysates from undifferentiated myoblast (day 0) or from myoblasts differentiated for the indicated days were harvested and analyzed with western blotting. (B) Localization of Dyn2 in myotubes. Endogenous Dyn2 and caveolin-3 in day 5-differentiated C2C12 were stained and imaged with confocal microscopy. Images of z-stack maximum intensity projections, 3D reconstruction or single focal plane were shown as indicated. Boxed areas were magnified in the lower panels. Arrows highlight the T-tubule, and arrowheads indicate Dyn2 puncta adjacent to caveolin-3-decorated T-tubule. Scale bars, 10 µm.

We next examined the subcellular localization of Dyn2 in C2C12. In undifferentiated myoblast, both endogenous Dyn2 and exogenous Dyn2-GFP were found in the cytoplasm and at the plasma membrane CCPs, similar to its distribution in epithelia or fibroblast cells (33,34) (Supplementary Material, Fig. S1B and C). The specificity of anti-Dyn2 antibody was validated with Dyn2 knockdown or antibody depletion assay (Supplementary Material, Fig. S1C and D). After 5 days of differentiation, endogenous Dyn2 showed plasma membrane as well as intracellular punctate distribution (Fig. 3B and Supplementary Material, Fig. S1D). The peripheral Dyn2 co-localizes with clathrin adaptor AP-2 (Supplementary Material, Fig. S1E), whereas the intracellular Dyn2 puncta often located adjacent to T-tubules, labeled by caveolin-3 (Fig. 3B, arrowheads). Reconstructed three-dimensional image rotation further illustrates the close association of intracellular Dyn2 puncta with T-tubule (Supplementary Material, Movies S1–S3). We also double-stained Dyn2 with its interacting protein BIN1 (35), which is critical for T-tubule biogenesis and aligns with caveolin-3 (36,37). Similar to caveolin-3, Dyn2 partially co-localizes with BIN1-labeled tubule structures in C2C12-derived myotube (Supplementary Material, Fig. S1F). Given its subcellular distribution, we next examined the effect of CNM-Dyn2 on CME or T-tubule morphology.

We utilized a tet-regulatable adenoviral vector to express mutant Dyn2 in C2C12 myoblast and analyzed transferrin uptake to monitor CME. Consistent with previous reports (38,39), we could not detect significant impairment in CME upon CNM-Dyn2 expression. However, CME was inhibited in cells expressing CMT mutant G537C and, as a positive control, the dominant-negative K44A (Supplementary Material, Fig. S2). Expression of CNM-Dyn2 mutants did not impair the differentiation of C2C12 myoblast, as muscle-specific proteins, MyHC and BIN1, were induced as expected (Supplementary Material, Fig. S3A). However, short-term expression (12 h) of CNM-Dyn2 in mature C2C12-derived myotubes resulted in pronounced fragmentation of caveolin-3-labeled T-tubules (Fig. 4A, blue arrowheads). We noted that overexpression of WT-Dyn2 at high, but not low, levels also leads to T-tubule fragmentation (Fig. 4A, high and low). This result further supports the hypothesis that hyperactivity of Dyn2 would lead to T-tubule fragmentation. On the contrary, more clustered, but not vesiculated, T-tubules were observed in myotubes expressing either G537C or K44A mutant, indicating that vesiculation of the T-tubule network depends on the GTPase (K44A) and membrane fission (G537C) activity of Dyn2 (Fig. 4A). The effect of Dyn2 on the T-tubule morphology was further quantified by the percentage of myotubes containing caveolin-3 tubule longer than 10 µm (Fig. 4B). Consistently, similar effects of CNM-Dyn2 on T-tubule morphology could also be observed using another T-tubule marker dihydropyridine receptor (Supplementary Material, Fig. S3B). Notably, the protein level of caveolin-3 was not significantly affected upon acute Dyn2 expression (Supplementary Material, Fig. S3C). In addition, the detection of caveolin-3 at the periphery indicated that the transport of caveolin-3 to plasma membrane is not perturbed (Fig. 4A).

Figure 4.

Expression of CNM-related Dyn2 mutants disrupts the T-tubule in C2C12-derived myotube. (A) Day 5-differentiated C2C12 myotubes were infected with tetracycline-regulatable, Dyn2-expressing adenoviruses and induced with 10 ng/ml tetracycline. The cells were fixed and stained after 12 h induction. Representative images of confocal z-stack projections are shown. Red and blue arrowheads indicate tubular and fragmented T-tubule, respectively. The control cell is empty viral vector-infected C2C12 myotube. (B) Quantification of myotubes containing at least one caveolin-3 tubule longer than 10 µm. The percentage of each condition was compared with control cells by Student's t-test. *P < 0.05. Scale bars, 10 µm.

To directly observe the membrane fission in vivo, we generated artificial membrane tubules by overexpressing BIN1-GFP in C2C12 (37) and monitored its dynamics in the presence of different Dyn2-mCherry mutants (Fig. 5A). BIN1-GFP was found on shorter tubules in the presence of Dyn2S619L-mCherry. Furthermore, the time-lapse microscopy showed that BIN1-GFP is highly dynamic in WT and Dyn2S619L-mCherry expressing cells, but it is very stable in Dyn2G537C-mCherry cell (Supplementary Material, Movies S4–S7). The different effects of Dyn2 mutations were further quantified by the percentage of cells containing BIN1-GFP tubule longer than 10 µm (Fig. 5B). Together, these real-time images support the higher membrane fission activity of CNM-Dyn2 mutants in cells.

Figure 5.

Membrane fission activity of Dyn2-mCherry regulates the morphology of BIN1-GFP-decorated membrane tubules in C2C12 myoblasts. (A) C2C12 myoblasts were co-transfected with BIN1-GFP and different Dyn2-mCherry constructs: WT, S619L (CNM) or G537C (CMT), respectively. After 24 h, cells were imaged with an inverted epi-fluorescent microscope at 37°C for the dynamics of BIN1-GFP tubule. (B) Cells expressing comparable amount of BIN1-GFP were quantified for the length of BIN1-GFP tubule >10 µm. n = 3 experiments, and 12–15 cells for each condition were quantified. Scale bars, 10 µm.

Dynamin has been reported to directly regulate actin organization in podocytes (40). However, we did not detect significant alteration of F-actin staining in CNM-Dyn2 expressing myotubes (Supplementary Material, Fig. S4). Together, these results suggest that the primary effect of Dyn2 hyperactivity in muscle is to disrupt T-tubule organization.

Effects of CNM-dynamin-2 in Drosophila body wall muscle

To further confirm the critical role of Dyn2 activity on T-tubule organization, we generated transgenic Drosophila and analyzed the effect of human Dyn2 expression in muscle. We used the GAL4-UAS system to specifically express HA-tagged human Dyn2 in fly muscle and found that larvae with WT- or CNM-Dyn2 but not CMT-Dyn2G537C expression had impaired locomotion activity (Fig. 6A and Supplementary Material, Movie S8). Moreover, both WT- and CNM-Dyn2 flies had defects in the subsequent eclosion. Therefore, we first examined the larval body wall muscle with anti-HA and phalloidin staining for its sarcomere structure in each mutant fly. We found that Dyn2 could partially localize to I-band, especially Dyn2G537C, and the sarcomere structure of WT- and CNM-Dyn2 expressing muscles aligned well and had similar organization as compared to control or Dyn2G537C flies (Fig. 6B). However, the WT- and CNM-Dyn2 flies developed muscle fiber atrophy as quantified by measurement of muscle fiber area (Fig. 6C and D).

Figure 6.

Effects of Dyn2 mutations on muscle structure and function in fly. (A) The locomotion of indicated Dyn2-expressing larvae was analyzed at room temperature. n = 15. (B) The sarcomere structure of DNM2 third-instar larvae was examined by phalloidin staining. Single confocal images are shown. (C) Muscle fiber atrophy in WT- and CNM-Dyn2 expressing Drosophila. Third-instar larvae were dissected and stained with phalloidin (green) and DAPI. The abdominal hemisegment A3 was imaged, and z-stack projection images are shown. (D) Quantification of muscle fiber area. The area of muscle fiber 6 and 13 in the A3 hemisegment was determined using ImageJ (National Institutes of Health; n = 5). Scale bars, 10 µm. ^#P < 0.10 and *P < 0.05.

Interestingly, we observed that WT-Dyn2 localized not only to sarcolemma and I-band (green and red arrowheads in Fig. 7A, respectively), but also to T-tubule and NMJ (arrow and white arrowhead in Fig. 7A). Nonetheless, the expression of CNM-Dyn2 did not cause dramatic alteration on the morphology of NMJ in larval muscle (Supplementary Material, Fig. S5A).

Figure 7.

Effects of Dyn2 mutations on T-tubule morphology in Drosophila third- instar larvae. (A) The distribution of HA-Dyn2 WT in larval body wall muscle. Representative images from confocal z-stack projections are shown. Arrow indicates the T-tubule. White, red and green arrowheads indicate the NMJ, I-band and sarcolemma distributions of Dyn2, respectively. (B) The effect of Dyn2 mutations on T-tubule morphology. The expression of Dyn2 was driven by the MHC-GAL4 promoter. Third-instar larval body wall muscle was stained with anti-HA (green) and anti-DLG (red) antibodies and imaged by confocal microscopy.

Next, we examined the morphology of T-tubule with anti-discs large (DLG) antibody, a postsynaptic and T-tubule marker (41). We found that the CNM-Dyn2 expressing third-instar larvae had severely fragmented T-tubules in their body wall muscle (Fig. 7B). Although less striking, the expression of exogenous WT-Dyn2 also altered the T-tubule morphology, and many WT-Dyn2 puncta were detected adjacent to T-tubules (Fig. 7B). No obvious fragmentation was detected in G537C expressing larvae, and this difference is not due to lower expression of Dyn2G537C (Supplementary Material, Fig. S5B). In contrast, we detected much higher protein levels of G537C in the larvae, probably because the muscle with higher CNM-Dyn2 expression would deteriorate. To further examine whether Dyn2 could vesiculate T-tubule after its formation, we employed MHC-GAL4 together with tub-GAL80^ts to block Gal4 activity in the early developmental stage. After 4 days of induction in the larval stage, we detected T-tubule fragmentation in CNM-Dyn2 flies, but not in WT- and G537C-Dyn2 (Fig. 8). Consistently, these CNM-Dyn2 larvae showed defective locomotion ability when compared with WT-Dyn2 and G537C-Dyn2 flies (Supplementary Material, Fig. S5C). Again, this differential effect of WT- and CNM-Dyn2 is not due to their expression level (Supplementary Material, Fig. S5D). Together, these results demonstrate that T-tubule is highly sensitive to Dyn2 activity and is likely the primary target of CNM-Dyn2 mutations in muscle.

Figure 8.

Effect of temporally regulated Dyn2 expression in larval muscle. The expression of Dyn2 was regulated by tub-GAL80ts and was thus restricted to larval stages. After 4 days of induction at 30°C, third-instar larvae were stained with anti-HA (green) and anti-DLG (red) antibodies, as in Figure 7B.

Discussion

In this study, we demonstrate that CNM- or CMT-associated mutations have distinct effects on the membrane fission activity of Dyn2, and the hyperactivity of CNM-Dyn2 is due to its abnormal self-assembly. Although the hyperactivity of Dyn2 does not significantly interfere with CME, it causes T-tubule fragmentation in cultured myotubes and fly muscle. These findings revealed the molecular basis and primary cellular effect of these disease-associated Dyn2 mutations.

Dynamin has been proposed to be regulated by an autoinhibitory mechanism that prohibits its self-assembly through interaction between the PH domain and the stalk region from previous biochemical and structural studies (14,19). Our results here support this autoinhibitory mechanism with the CNM-Dyn2 mutants, which are presumably defective in this autoinhibition (E368K, A618T, S619L and Δ625 in the inhibitory interface) or possess enhanced self-assembly (R465W and E650K in the stalk surface) and thus lead to curvature insensitivity and irregular membrane fission (Supplementary Material, Fig. S1A). One special mutant, E560K, is located within the VL2 of PH domain, which is responsible for PI(4,5)P₂ binding. Thus, we speculate that the negative-to-positive charge conversion of E560K may increase the binding affinity of Dyn2 to PI(4,5)P₂, thus enhancing its assembly on membrane. Taken together, our results suggest that under physiological conditions, autoinhibition is sufficient to prevent endogenous Dyn2 from randomly severing T-tubule, which is highly curved and enriched in PI(4,5)P₂ and BIN1 (37). Furthermore, the T-tubule fragmentation by higher levels of WT-Dyn2 expression suggests that the inhibitory mechanism can be overcome by high Dyn2 concentrations.

Our results also provide direct evidence to support the hypothesis that triads disorganization is the foundation of CNM (1,24,42), and the Dyn2-related CNMs are caused by hyperactive membrane fission of T-tubules. T-tubule is the muscle-specific subdomain of the plasma membrane; therefore, its sensitivity to Dyn2 activity may account for the susceptibility of muscle to Dyn2 hyperactivity relative to other tissues. The disorganization of T-tubules must disturb the structure and function of triads, which underlies the excitation–contraction coupling of muscle. In contrast, we reason that Dyn2-related CMT may be caused by the defective fission ability of Dyn2 and impaired CME. Recently, the requirement of CME and dynamin for axon myelination has been reported (39). Our biochemical results are consistent with this hypothesis that impaired CME is likely the basis of Dyn2-related CMT.

The lethality of Dyn2 transgenic flies is due to eclosion defect. However, the flies lacking amphiphysin display defective T-tubules, but can eclose and grow into adult properly (43).Therefore, it suggests that T-tubule disorganization is not the major reason for the lethality in CNM-Dyn2 flies. One alternative explanation of the lethality may be the jammed endocytic pathways caused by excess membrane structure/vesicles resulted from Dyn2 hyperactivity. The accumulated vesicles may cause imbalance in lysosomal function or autophagy, thus leading to severe muscle dysfunction even more than amphiphysin null mutant. Further studies are needed to elucidate the following consequences of T-tubule fragmentation mediated by Dyn2 hyperactivity.

Recently, the importance of clathrin heavy chain, AP-2 and Dyn2 in skeletal muscle sarcomere organization has been addressed, and these endocytic proteins are proposed to play unconventional roles in the formation and maintenance of the contractile apparatus (44). Although we have demonstrated that the major target of CNM-Dyn2 is the T-tubule, we do not exclude the importance of Dyn2 on F-actin organization or sarcomere structure in muscle. Indeed, we detected WT-Dyn2 localizing to the actin filament-distributed I-band in fly muscle, thus supporting its role for actin organization in this tissue.

Given that CNM-related Dyn2 is a hypermorphic mutation, it suggests that downregulation of Dyn2 activity in muscle would be a potential therapy of CNM. Interestingly, the Laporte group has reported that reducing the expression of Dyn2 could rescue X-linked myotubularin (MTM1)-related CNM (45). Therefore, a tissue-specific attenuation of Dyn2 activity will be a potential therapy for both Dyn2- and MTM1-related CNM. Together, our findings reveal the molecular mechanism of disease-related Dyn2 mutations and pave the way for future studies of this essential GTPase of its unique roles in different tissues.

Materials and Methods

Plasmids, adenoviruses and antibodies

For protein expression and purification, human DNM2 or mutants were cloned into pIEX6 (Novagen, Madison, WI, USA) for expression in Sf9 cells. Adenoviruses encoding N-terminal HA-tagged DNM2 mutants under tet-regulatable promoter were generated, as described previously (46). Plasmids used in this study were listed in Supplementary Material, Table S1, and antibodies used here were described in Supplementary Material, Table S2.

Protein purification

Dynamin proteins were expressed in Sf9 cells transiently transfected with various constructs and purified, as described previously (28). The N-BAR domain from endophilin was expressed and purified from Escherichiacoli, as described previously (47).

Preparation of lipid templates

For liposomes, lipid mixtures (DOPC:DOPS:PIP2 at 80:15:5) were dried, rehydrated in buffer [20 mm HEPES (pH 7.5) and 150 mm KCl] and subjected to a series of freeze–thaw cycles before extrusion through polycarbonate membranes (Whatman) with a pore size ranging from 50 to 1000 nm using an Avanti Mini-Extruder. LNs (DOPC:DOPS:PIP₂:GalCer at 40:15:5:40) were generated as described previously (28). SUPER templates were generated as reported previously (4). Briefly, silica beads with 2.5 µm diameter (Corpuscular) were incubated with 100 µm, 100 nm liposomes (DOPC:DOPS:PIP₂:Rhodamin-PE at 79:15:5:1) in 20 mm HEPES (pH 7.5) and 600 mm NaCl. The mixture was incubated for 30 min at room temperature, washed with water and immediately used for fission assay.

Fission, membrane tubulation and GTPase activity assays

Fission activity of dynamin was measured as reported previously (28). Briefly, SUPER templates were incubated with 1 mm MgCl₂ and 1 mm GTP and indicated dynamin for 30 min at room temperature. The released vesicles were separated from SUPER template by 260g swing-out centrifugation. The fission activity is expressed as the percentage of total fluorescence on SUPER templates. To monitor the curvature generation ability, SUPER templates were incubated with 0.5 µm Dyn2 without GTP on a cover slip, and pictures were taken after an incubation of 10 min by an inverted Leica DM IRB microscope with a 100×, 1.35-NA oil-immersion objective equipped with CCD monochrome Photometrics CoolSNAP HQ.

Electron microscopy and sedimentation assay

An aliquot of 1.0 µm dynamin was mixed with 25 µm LN and incubated at 37°C for 10 min. The mixture was then adsorbed onto carbon-coated, glow-discharged grids and stained with 1% uranyl acetate. Images were collected using a HITACHI H-7100 EM at 100 kV and a nominal magnification of 80 000×.

To determine membrane binding, dynamin was separated into liposome-bound and soluble fractions by high-speed centrifugation, as described earlier (29). The disassembly fraction is defined as: [(fraction of membrane bound without GTP – fraction of membrane bound with GTP)/fraction of membrane bound without GTP].

Cell culture, transfection and adenoviral infection

Mouse-derived C2C12 myoblasts (American Type Culture Collection, CRL-1772) were cultured in growth medium (GM) and Dulbecco's modified Eagle's medium supplemented with l-glutamine, sodium pyruvate, antibiotics and 10% fetal bovine serum. To induce differentiation, C2C12 were seeded onto gelatin-coated plates or cover slips in GM, grown to 90% confluency and then switched to differentiation medium (DM), same as GM but with 2% horse serum. This time point was considered as day 0 of differentiation. For transfection, proliferating cells in GM at 70% confluency were transfected with interested DNA using Lipofectamine 2000 (Invitrogen), as recommended by the manufacturer.

Immunostaining and imaging

For indirect immunofluorescence staining, cells were fixed with 4% formaldehyde and permeabilized with 0.1% saponin. After blocking with 2% bovine serum albumin and 5% normal donkey serum, cells were stained with the indicated primary and secondary antibodies. Samples were observed under a confocal microscope LSM700 (Carl Zeiss, Jena, Germany). For time-lapse microscopy, cells were cultured in Phenol-red free medium and imaged with an inverted Leica DM IRB microscope with a 63×, 1.35-NA oil-immersion objective at 37°C and 5% CO₂.

Locomotion of larval Drosophila

Larva crawling was analyzed by the locomotor assay described previously with few modifications (48). Larva was allowed to wander under room temperature on agar plate (previously poured and allowed to harden) for 60 s and then the number of grids crossed in 30 s was counted. Each grid is 3 mm × 3 mm.

Fly stocks and genetics

Drosophila stocks were maintained on conventional food medium at 25°C. All general fly stocks and GAL4 lines were obtained from the Bloomington Drosophila Stock Center. The parental strain ZH-51D was used to generate transgenic flies with the pUAST-based constructs injected into Drosophila embryos, thus integrating into the attP (second chromosome) landing site (49).

Dissecting and staining of larval body wall muscle

Third-instar larvae were dissected in Ca²⁺-free buffer (128 mm NaCl, 2 mm KCl, 4 mm MgCl, 5 mm HEPES, 35.5 mm sucrose and 5 mm EGTA), fixed in 4% formaldehyde for 20 min and rinsed in 0.1 m phosphate buffer (pH 7.2), containing 0.2% Triton X-100. After blocking, samples were incubated with primary antibody at 4°C overnight. After wash, samples were incubated with the secondary antibody for 1 h at room temperature and were mounted in glycerol after washes.

Funding

This work was supported by Ministry of Science and Technology (MOST) grant NSC101-2321-B-002-071-MY3 and by National Taiwan University grant NTU-CDP-103R7878 to Y.-W.L. and MOST 103-2320-B-002-018 to S.-T.H.

Acknowledgements

We thank Drs June-Tai Wu [National Taiwan University (NTU)], Henry Sun (Academia Sinica) and the fly core in NTU for help with Drosophila experiments. We thank Drs Sandra Schmid (UT Southwestern), Chun-Liang Pan (NTU), June-Tai Wu (NTU) and Jia-Wei Hsu (NTU) for critical reading and helpful comments on the manuscript. We are grateful to the imaging core at NTU College of Medicine for technical assistance and Drs Nei-Li Chan (NTU) and Te-Sheng Lin (NTU) for PyMol tutorial. We thank Drs Christien Merrifield and Pietro De Camilli for the Dyn2-mCherry and BIN1-GFP constructs.

Conflict of Interest statement. None declared.

Author notes

†

The authors wish it to be known that, in their opinion, the first two authors should be regarded as joint First Authors.