Nuclear alignment in myotubes requires centrosome proteins recruited by nesprin-1

Muscle dysfunctions in several diseases have been correlated with mutations in nuclear envelope genes, such as emerin, lamin A, and nesprins 1 and 2 (nesprins 1–4 are also known as SYNE1–SYNE4, respectively). In particular, mutations in these proteins have been reported in human patients and in mouse models with Emery–Dreifuss muscular dystrophy, as well as ataxias and cardiomyopathies (Ellis, 2006; Gros-Louis et al., 2007; Wheeler et al., 2007; Zhang et al., 2007a; Attali et al., 2009; Puckelwartz et al., 2009, 2010). It has been speculated that these mutations might affect signalling pathways in muscle cells, but the exact role of nuclear envelope components in myopathies remains unclear. Whereas emerin and lamins are proteins that associate with the inner nuclear membrane, the nesprins are transmembrane proteins that span a wide distance from the perinuclear space to the cytoplasm. In the perinuclear space, nesprins are anchored to proteins of the SUN family by their conserved C-terminal KASH domain, whereas in the cytoplasm, they are believed to interact with the cytoskeleton through N-terminal calponin homology or plectin-binding domains and spectrin repeats. At least four nesprin proteins exist in humans and mice (nesprins 1, 2, 3 and 4), of which nesprin-4 is specifically expressed only in secretory epithelia (Zhang et al., 2001; Wilhelmsen et al., 2005; Roux et al., 2009).

In muscle cells, nesprin-1 has been implicated in nuclear positioning in the syncytial cytoplasm and in clustering a subset of synaptic nuclei at the neuromuscular junction (Grady et al., 2005; Zhang et al., 2007b, 2010; Lei et al., 2009). Consistent with this, nesprin-1 is particularly enriched on nuclei beneath the neuromuscular junction (Apel et al., 2000), as well as localizing less strongly to non-synaptic nuclei. Interference with the function or the localization of nesprin-1 abolishes correct myonuclear anchorage and leads to respiratory failure and death in mice (Zhang et al., 2007b, 2010; Puckelwartz et al., 2009). However, it remains unclear by what mechanism nesprin-1 mediates nuclear anchorage. In various studies on non-muscular cells, nesprins have been implied to interact with the actin cytoskeleton (Starr and Han, 2002; Zhen et al., 2002; Padmakumar et al., 2004). In addition, nesprins and related proteins have been shown to interact with components of the microtubule network, such as kinesin, dynein and dynactin, and with the centrosome (Malone et al., 2003; Roux et al., 2009; Zhang et al., 2009; Zhou et al., 2009; Fridolfsson et al., 2010; Yu et al., 2011; Wilson and Holzbaur, 2012, 2015). Several groups have reported that proteins of the centrosome, such as PCM-1, pericentrin and γ-tubulin, are relocated from the pericentriolar material to the nuclear envelope upon onset of myoblast differentiation, and that a substantial amount of microtubules grow from the nuclear surface following this reorganization (Tassin et al., 1985a; Bugnard et al., 2005; Srsen et al., 2009; Fant et al., 2009). In this study, we show that nesprin-1 is essential for the relocalization of centrosome proteins and components of microtubule motor complexes to the nuclear envelope in differentiating mouse myoblasts, and that nesprin-1 and the centrosome protein PCM-1 are needed for regular positioning of nuclei in these cells.

To identify factors that recruit centrosome proteins to the surface of the nuclear envelope during myoblast differentiation, we investigated the potential role of proteins of the outer nuclear membrane. To date, the only known proteins specifically localized at the outer nuclear membrane are members of the KASH family; in mouse muscle cells, these include nesprins 1, 2 and 3 (Hetzer, 2010).

We reinvestigated the localization of nesprins in C2C12 cells, given that conflicting reports on the localization of nesprin-1 and -2 in skeletal muscle cells have been published (Apel et al., 2000; Zhang et al., 2001, 2005; Mislow et al., 2002). Because nesprins exist as alternative splice variants, we performed immunofluorescence and western blot experiments with antibodies against the C-terminal KASH domain that is present in nuclear-envelope-bound isoforms. In undifferentiated C2C12 myoblasts, nesprin-2 and nesprin-3 were expressed at substantial levels and localized to the nuclear envelope, whereas the expression of nesprin-1 was very weak (Fig. 1A,B). Upon differentiation into myotubes, the nesprin-1 signal increased and became prominent at the nuclear envelope, while at the same time the centrosome protein pericentrin relocalized to the nuclear surface. The accumulation of nesprin-1 at the nuclear envelope correlated with the appearance of a 115-kDa band on western blots, corresponding to the nesprin-1α isoform (Fig. 1B), which is the prominent isoform expressed in human muscle cells (Randles et al., 2010). Concomitantly, the nesprin-2 signal disappeared from the nuclear envelope, whereas nesprin-3 remained (Fig. 1A). This pattern of nesprin localization at the nuclear envelope was confirmed using two different sets of antibodies against nesprins 1 and 2 (Fig. S1A,B). Depending on the antibody used, nesprin-2 immunofluorescence in C2C12 myotubes was seen in intranuclear structures as previously described (Zhang et al., 2005) or was diffuse in the cytoplasm (Fig. 1A, Fig. S1B). The nature of the intranuclear structures is unknown, but RNA interference experiments suggest that these might represent unspecific targets of the antibody (Fig. S1B). Given that immunoblotting against nesprin-2 failed to provide specific bands, we examined expression levels of nesprin-2 in comparison to those of nesprin-1 by performing reverse transcription from C2C12 RNA obtained prior to and after differentiation, followed by quantitative PCR (qPCR). Using PCR primers complementary to the spectrin-repeat-containing sequences found in a large variety of possible splice variants of nesprin-1 or nesprin-2, we obtained strong signals for nesprin-1 only upon differentiation of C2C12 into myotubes, in contrast to nesprin-2, for which levels dropped in differentiating cells (Fig. 1C).

Because nesprin-2 was found to be absent from the nuclear envelope in myotubes, we tested whether nesprins 1 or 3 are involved in the recruitment of centrosome proteins to the nuclear envelope in differentiating myoblasts. We depleted up to 80% of nesprin-1α or nesprin-3 protein by siRNA treatment of C2C12 cells (Fig. 2A,B). As a consequence, the recruitment of pericentrin to the nuclear envelope was inhibited in nesprin-1-depleted cells that had fused into myotubes (Fig. 2C). Interestingly, numerous myotubes had nesprin-1-positive as well as nesprin-1-negative nuclei in the same cytoplasm. This was likely due to the fact that nesprin-1 depletion had already started prior to fusion of C2C12 cells, enabling non-depleted cells to fuse with depleted cells. Thus, in heterogeneous myotubes, only nuclear envelopes with a high level of nesprin-1 also displayed enriched amounts of pericentrin (Fig. 2C, asterisk). Given that the pericentrin levels at the nuclear envelope varied between myotubes, we quantified the immunofluorescence intensity within a perinuclear rim of ∼0.5 µm thickness. For these measurements, we analysed myotubes that contained both nesprin-1-positive and -negative nuclei. Nuclei that were depleted of nesprin-1 showed only 37±15% (mean±s.d., n=20) of the pericentrin intensity of control nuclei. The effects of nesprin-1 siRNA could be reproduced with double-stranded RNAs (dsRNAs) against two different targeting sequences (see Materials and Methods), thus reducing the risk of unspecific off-target effects. In contrast to nesprin-1 depletion, the removal of nesprin-3 did not substantially alter the amount of nuclear-envelope-bound pericentrin (Fig. 2D). Quantification of the pericentrin signal was performed in myotubes containing nesprin-3-depleted and undepleted nuclei (Fig. 2D, asterisk) in the same cytoplasm. Pericentrin levels on nesprin-3-depleted nuclear envelopes were 89±20% (n=21) of the level of non-depleted nuclear envelopes. As a control, silencing of nesprin-2 was attempted (see Materials and Methods), but was difficult to monitor due to the absence of nesprin-2 immunofluorescence in C2C12 myotubes and the absence of specific nesprin-2 signal on immunoblots. Treatment with siRNA against nesprin-2 was without effect on the localization of centrosome proteins (Fig. S1B). When investigating the fate of centrosome proteins other than pericentrin, we found that depletion of nesprin-1 also led to the reduction of PCM-1 at the nuclear envelope (Fig. 2E), γ-tubulin (see Fig. 4A) and CDK5RAP2 (data not shown). None of these reductions were seen after depletion of nesprin-3 (Fig. 2F). We verified by immunoblotting that the depletion of nesprin-1 or nesprin-3 did not affect the overall protein levels of centrosome proteins, such as PCM-1 (Fig. 2A,B). To test whether the anchoring of centrosome proteins to the nuclear envelope involved proteins of the Golgi or of the endoplasmic reticulum (ER), we tested the effects of the Golgi-disrupting drug Brefeldin A and siRNA against the ER-resident chaperone calnexin (Fig. 2G,H). Although it is known that Golgi membranes accumulate around the nuclear surface upon differentiation of muscle cells (Tassin et al., 1985b), and that ER membranes are continuous with the outer nuclear membrane, neither of these seemed to be essential for the anchorage of centrosome proteins, because neither Brefeldin A nor calnexin siRNA reduced the accumulation of pericentrin at the nuclear envelope (Fig. 2G,H).

Because silencing of nesprin-1 expression prevented the accumulation of centrosome proteins at the nuclear envelope, we tested whether displacement of nesprins from the nuclear membrane caused a similar effect. We found that a high overexpression of the nesprin-1 KASH domain displaced endogenous nesprin-1 from the nuclear envelope in myotubes and prevented the relocalization of centrosome proteins to the nuclear envelope (Fig. 3A,B). This was also achieved by depletion of the SUN proteins 1 and 2, which localize to the inner nuclear membrane (Fig. 3C,D), and which are known to anchor nesprins to the nuclear envelope (Padmakumar et al., 2005; Crisp et al., 2006). To test the association between nesprin-1 and centrosome proteins biochemically, we performed immunoprecipitation of nesprin-1 from extracts of nuclear envelopes prepared from C2C12 myotube nuclei (see Materials and Methods and Fig. S1C for details). The centrosome protein PCM-1 was found to co-immunoprecipitate under these conditions, whereas γ-tubulin was absent from the precipitate (Fig. 3E). Consistent with this, immunoprecipitation of PCM-1 co-precipitated a portion of the nesprin-1α pool in the reverse experiment (Fig. 3E).

To ensure that the depletion of nesprin-1 did not cause unspecific interference with centrosome protein relocalization because of a global disorganization of the nuclear envelope, we stained for nuclear pore complexes using monoclonal antibody 414, which recognizes a subset of nucleoporins (Davis and Blobel, 1986), and for nesprin-3. Both were still correctly localized in nesprin-1-depleted myotubes, indicating that the structure of the nuclear envelope was not globally affected (Fig. 3F; Fig. S1D).

Taken together, our data suggest that nesprin-1 is specifically involved in anchoring centrosome proteins to the nuclear envelope during myoblast differentiation and myotube formation. To test whether nesprin-1 is sufficient for the recruitment of centrosome proteins to the nuclear envelope, we expressed exogenous mCherry-tagged nesprin-1α in undifferentiated C2C12 myoblasts, that is, under conditions that show only minor amounts of endogenous nesprin-1 on the nuclear envelope. We observed that the tagged exogenous nesprin-1α localized to the nuclear envelope even in undifferentiated C2C12 cells (Fig. 3G). Immunofluorescence of the centrosome protein PCM-1 revealed that 47±1% (mean±s.d.) of C2C12 cells that incorporated mCherry–nesprin-1α at the nuclear envelope also showed relocalization of PCM-1 (data from four different experiments counting a total of 7163 cells). In these cells, PCM-1 was either entirely recruited to the nuclear surface (top row of Fig. 3G, left), or a portion of the PCM-1 remained in the pool of cytoplasmic pericentriolar satellites (second and third row of Fig. 3G, left). PCM-1 accumulation at the nuclear envelope was not seen in non-transfected neighbouring cells. In control transfections, we expressed mCherry-tagged tubulin-α or mEmerald-tagged nesprin-3 (Fig. 3H,I). Tagged α-tubulin incorporated into the microtubule network, but failed to relocalize PCM-1 to the nuclear envelope (only 2±1% of a total of 1224 cells with nuclear-envelope-bound PCM-1, in three different experiments). Likewise, tagged nesprin-3 localized to the nuclear envelope in addition to structures resembling the Golgi, but induced relocalization of PCM-1 only in 9±3% of the transfected cells (n=124; Fig. 3H). Moreover, comparison of the ratio of fluorescence intensity of PCM-1 between the nuclear envelope (within a rim of 0.5 µm thickness) and the neighbouring cytoplasm revealed that nesprin-1 expression increased PCM-1 immunofluorescence by a factor of 1.7±0.3 (n=5), whereas nesprin-3 expression led to an insignificant increase of 1.1±0.1. Taken together, this suggests that PCM-1 relocalization to the nuclear envelope upon nesprin-1α expression is specific. To exclude that those C2C12 cells displaying PCM-1 at the nuclear envelope had undergone premature differentiation in culture, we performed immunofluorescence for the differentiation marker myosin II heavy chain (MYH1E) (Fig. 3J). We quantified that only 1% of mCherry–nesprin-1α-expressing cells stained positively for myosin II heavy chain (in a total of 970 cells), demonstrating that relocalization of PCM-1 to the nuclear envelope could be induced by nesprin-1α, independently of a full differentiation program.

Because the localization of the microtubule nucleator γ-tubulin was reduced at nesprin-1-depleted nuclear envelopes in myotubes (Fig. 4A, second column, right nucleus), we tested whether the microtubule network in these cells was altered. Unlike control myotubes, nesprin-1-depleted myotubes contained a large number of nuclei (∼70%) that failed to re-grow microtubules efficiently after recovery from nocodazole-induced depolymerization (Fig. 4B), suggesting that the absence of centrosome proteins from the nuclear surface prevented microtubule nucleation. Surprisingly, in cell cultures that had not been subjected to microtubule depolymerization, both control myotubes and nesprin-1-depleted myotubes displayed a similar microtubule network organization (Fig. 4C). To test whether the number of microtubules in the vicinity of nuclei were reduced upon nesprin-1 depletion, we tracked individual microtubules within a distance of ∼1 µm of the nuclear surface, in image stacks of myotubes. Optical sections of complete myotubes were grouped by performing sum projections of partial stacks, allowing the tracking of individual microtubules over a large volume (Fig. 4D). We counted ∼600 microtubules close to each nucleus, for controls as well as for nesprin-1-depleted nuclei (Fig. 4E). Long microtubules, oriented parallel to the axis of the myotube, were usually seen curving around the surface of the nuclei in both conditions, with only 12±2% (mean±s.d.; control) compared to 14±1% (nesprin-1 depletion) of microtubules oriented approximately perpendicular to this axis (Fig. 4F). Analysis of deconvolved 3D data sets of microtubules suggested that end-on contacts of microtubules with the nuclear envelope were rare, although the exact origins and ends of microtubules were difficult to trace. To find out whether there was a substantial number of microtubules emanating from the nuclear surface in myotubes at a steady state (with an established equilibrium of polymerized versus non-polymerized tubulin), we quantified the immunofluorescence signal intensity of microtubules in myotube segments containing nuclei (segment 1 in scheme Fig. 4G), in tangential segments at the periphery of nuclear envelopes (segment 2) and in areas free of nuclei (segment 3). Both segments 1 and 2, at or near nuclei, showed a microtubule intensity that was 10–20% higher compared to nuclei-free cytoplasmic segments, raising the possibility that there was a small contribution of myotube nuclei to microtubule nucleation or microtubule anchoring. However, this percentage was not significantly altered in myotubes with nesprin-1-depleted nuclei (Fig. 4G), indicating that the small number of microtubules at the nuclear envelope did not require substantial amounts of nuclear-envelope-bound centrosome proteins under equilibrium conditions, unlike microtubule re-growth after depolymerization, which is favoured only if many nucleation sites are instantly available (Fig. 4A,B).

The most notable change in myotubes depleted of nesprin-1 was an abnormal positioning of nuclei within the syncytial cytoplasm (Fig. 5A): whereas 76% of control myotubes showed nuclei aligned evenly along the long axis of the cell, only 21% of nesprin-1-depleted myotubes contained aligned nuclei (Fig. 5A, graph). Moreover, the spacing between nuclei was reduced and multiple nuclei often concentrated into small packages (Fig. 5A). To determine whether centrosome proteins mediated any aspect of nesprin-1-dependent nuclear positioning, we depleted PCM-1, given that this has previously been characterized as a protein necessary for the recruitment of many other centrosome components (Dammermann and Merdes, 2002), including binding of pericentrin to the nuclear envelope (Fig. S1E; in the reverse experiment, that is, depletion of pericentrin, PCM-1 was not altered). As compared to nesprin-1 depletion, a similar although weaker effect on nuclear positioning was seen after depletion of PCM-1, yielding 55% of cells with aligned nuclei (Fig. 5A, graph). The depletion led to removal of 74% of PCM-1 in cultures overall (Fig. 5B, quantification of immunoblot), but for the quantification of nuclear positioning, only myotubes lacking PCM-1 at the nuclear envelope were counted. We verified that depletion of PCM-1 did not affect the overall protein levels of nesprin-1 (Fig. 5B). Moreover, we verified by immunofluorescence that nuclear envelopes depleted of PCM-1 still retained nesprin-1 (Fig. 5B, right).

To understand the potential mechanisms that caused misalignment of nuclei in myotubes depleted of nesprin-1 or PCM-1, we investigated proteins that are known to mediate microtubule-dependent transport. Treatment of differentiated C2C12 cultures with erythro-9-[3-(2-hydroxynonyl)]adenine (EHNA), an ATPase inhibitor that has been shown to interfere with dynein function (Bouchard et al., 1981; Penningroth et al., 1982), led to inhibition of nuclear alignment (Fig. 5A, graph). The effect of the dynein inhibitor ciliobrevin A was also tested, but ciliobrevin A was found to significantly prevent the fusion of myoblasts into myotubes (data not shown). Moreover, the protein p150glued (also known as DCTN1, hereafter referred to as p150), a component of the dynactin complex that has been implicated in dynein activation, as well the kinesin light chain 1 (KLC1) were found to localize to the nuclear envelope in control myotubes (Fig. 5C,D). Depletion of nesprin-1, and also of PCM-1, prevented the localization of both p150 and KLC1 to the nuclear envelope (Fig. 5C,D). By contrast, depolymerization of the microtubule network in myotubes did not remove p150 from the nuclear envelope (Cadot et al., 2012; Fig. 5E), suggesting that motor-complex-associated proteins are bound to the nuclear surface in a manner that is dependent on nesprin-1 and PCM-1, but independent of microtubules.

In the present study, we employed a cell culture model of muscle differentiation to show that centrosome proteins are anchored to the nuclear envelope by nesprin-1, and that this interaction is necessary for the recruitment of motor proteins that position nuclei in syncytial myotubes. Our findings are consistent with earlier studies, in which nesprin-related KASH domain proteins have been implicated in nuclear migration and positioning (Starr, 2007), such as UNC-83 in C. elegans (Starr et al., 2001), and MSP-300 and Klarsicht in Drosophila photoreceptor cells and in striated muscle (Mosley-Bishop et al., 1999; Patterson et al., 2004; Elhanany-Tamir et al., 2012). In mammals, the KASH domain of nesprins is required for nuclear positioning in neurons and in muscle cells, as elimination of KASH-encoding exons or displacement of endogenous nesprins by overexpressed KASH results in defective nuclear migration, localization and anchorage (Grady et al., 2005; Zhang et al., 2007b, 2009, 2010; Wilson and Holzbaur, 2012, 2015). Nuclear movement has previously been suggested to be mediated by a physical interaction between nesprins or related proteins such as ZYG-12 and the centrosome, which is at the centre of a radial microtubule network (Malone et al., 2003; Zhang et al., 2009). In early differentiated myotubes, the centrosome no longer exists as a microtubule-organizing centre, but its constituent proteins are relocalized all around the surface of the nucleus (Tassin et al., 1985a; Bugnard et al., 2005; Srsen et al., 2009). We show that this depends on the differentiation-specific expression of the nesprin-1α isoform, and that other nesprins are not involved in this process. We did not find nesprin-2 at nuclear membranes of newly differentiated C2C12 cells, although splice variants of nesprin-2 have previously been identified in skeletal muscle (Duong et al., 2014; Zhang et al., 2005). It is possible that nuclear and non-nuclear forms of nesprin-2 are expressed at more advanced stages of muscular differentiation, for example, upon sarcomere formation, that are not mimicked under our experimental conditions (Zhang et al., 2005). The only other nesprin present at the nuclear envelope during the early differentiation stages, nesprin-3, has a cytoplasmic domain that is largely dissimilar from nesprin-1α (only 8% of sequence identity outside its KASH domain), and does not recruit centrosome proteins (Fig. 2D,F). The detection of premature PCM-1 relocalization to the nuclear envelope prior to differentiation upon mCherry–nesprin-1α expression led us to conclude that nesprin-1α is a receptor for centrosome proteins at the nuclear envelope. It is possible that additional proteins are involved for efficient anchoring, because we only obtained partial recruitment of PCM-1. Another possibility would be that differentiation-specific post-translational modifications of nesprin-1α are needed for a strong interaction with centrosome proteins, and that these modifications are missing in our transfected cells at the myoblast stage.

How do centrosome proteins affect nuclear positioning in myotubes? Whereas in non-muscular cells, forces can be transmitted to the nucleus by the radial centrosomally anchored microtubule network owing to the physical link between the centrosome and the nucleus, a different mechanism is likely active in myotubes: we suggest that the relocalized centrosome proteins all around the nuclear envelope anchor microtubule-dependent motors of the kinesin and dynein family, allowing nuclei to move laterally along microtubules and to position along the length of the myotube. The dependence of myonuclear distribution on microtubules was originally inferred from pharmacological experiments and from correlative studies of nuclear position and the microtubule network (Englander and Rubin, 1987; Bruusgaard et al., 2006), and has more recently been shown by Cadot et al. (2012) and Metzger et al. (2012). Initial formation of microtubule filaments in myotubes might involve microtubule-nucleating activity on the nuclear surface (Tassin et al., 1985a; Bugnard et al., 2005; Fant et al., 2009). However, microtubules in our early differentiated C2C12 cultures were mostly seen in a parallel orientation along the length of the cell, with long microtubules running past nuclei instead of attaching end-on to their nuclear surface (Fig. 4C,E). Later in the differentiation process, the microtubule network is further remodelled into an orthogonal-grid-like organization, with a large fraction of microtubules growing off Golgi elements (Oddoux et al., 2013). It is possible that upon formation of a microtubule network in myotubes, release of microtubule ends from their initial nucleation sites or generation of new distant nucleation sites occurs, as seen in many non-muscular cell systems undergoing differentiation (Mogensen, 1999; Musa et al., 2003; Stiess et al., 2010). This would explain loss of binding of microtubule ends at the nuclear surface, as seen in our study. Moreover, previously reported microtubule nucleation from nuclear envelopes in myotubes might have been experimentally enhanced by depolymerization of the microtubule network: under these conditions, the concentration of free tubulin is high, and re-nucleation can occur from sites that might not specifically nucleate microtubules under physiological conditions (Mitchison and Kirschner, 1985).

The concept of nuclear positioning by gliding along microtubules is supported by previous studies demonstrating an interaction of kinesin and dynein–dynactin motor complexes with mammalian nesprins 1, 2 and 4, and with the related KASH-domain proteins ZYG-12 and UNC-83 in C. elegans (Malone et al., 2003; Roux et al., 2009; Zhang et al., 2009; Fridolfsson et al., 2010; Yu et al., 2011; Wilson and Holzbaur, 2012, 2015). In myotubes, we propose that this interaction is indirect, mediated by nuclear-envelope-bound centrosome proteins, because the depletion of the centrosome protein PCM-1 induces loss of dynactin and kinesin subunits from the nuclear surface. Given that PCM-1 is necessary for the targeting and assembly of a variety of proteins of the pericentriolar material (Dammermann and Merdes, 2002), anchoring of motor complexes and nuclear alignment might occur through other centrosome proteins such as pericentrin and Cep135, both known to bind directly to dynein and dynactin polypeptides (Purohit et al., 1999; Uetake et al., 2004). Consistent with this, the centrosome protein pericentrin was lost from the nuclear envelope when PCM-1 was depleted (Fig. S1C). Besides contributing to nuclear positioning, the accumulation of centrosome proteins at the nuclear envelope might serve additional roles in muscular cell types, given that a similar accumulation has been found in mono-nucleate mouse cardiomyocytes (Zebrowski et al., 2015). It has been speculated that in mammalian cardiomyocytes, that relocalization of centrosome proteins helps to maintain the post-mitotic cell cycle state (Zebrowski et al., 2015).

In skeletal muscle, the ordered positioning of nuclei would be expected to be important for correct muscle function, given that abnormal positioning has been found in mouse models in which the KASH domain of nesprin-1 was deleted, and which mimicked aspects of Emery–Dreifuss muscular dystrophy (Zhang et al., 2007b, 2010; Puckelwartz et al., 2009). Interestingly, point mutations in the nesprin-1 gene have been found in a subgroup of patients suffering from Emery–Dreifuss muscular dystrophy (Zhang et al., 2007a). Given that defective nuclear positioning is reproduced in our cell culture model, and shown to involve PCM-1 in addition to nesprin-1, it will be important to test in the future whether the localization of centrosome proteins and microtubule motors at the nuclear envelope is affected in any way in Emery–Dreifuss muscular dystrophy, and whether this is involved in pathogenetic mechanisms.

C2C12 mouse myoblasts (American Type Culture Collection) were cultured in Dulbecco's modified Eagle's Medium (DMEM)+Glutamax-I (Invitrogen) supplemented with 20% fetal bovine serum (FBS) and 0.5% chicken embryonic extract. Cells were induced to differentiate by switching to differentiation medium (DMEM+Glutamax-I supplemented with 0.5% FBS, 5 µg/ml insulin and 5 µg/ml transferrin).

siRNAs were designed to target the following sequences: 5′-CCAGGGTGAAGAAGCTAAA-3′ and 5′-GCTCCTGCTGCTGCTTATT-3′ for nesprin-1, 5′-AGCCACAGAACTCCAAAAT-3′ for nesprin-2, 5′-GCTACGTAGAATCATCACA-3′ for nesprin-3, 5′-TGAGCTTCGTGATTCTCAG-3′ for PCM-1 (Dammermann and Merdes, 2002), 5′-CGTCGGATGCTCTGGATTT-3′ and 5′-CAGGTGCCTTCGAAATATT-3′ for SUN-1, 5′-CCGCTGCTCTGAGACTTAT-3′ and 5′-GCCCTTGCTGCAGACTTTT-3′ for SUN-2, and 5′-AAGCATCATGCCATCTCTGCT-3′ for calnexin. Control experiments were done using the scrambled siRNA sequence 5′-UUCUCCGAACGUGUCACGU-3′ (Qiagen). C2C12 cells were seeded at 80,000 cells/cm², and induced to differentiate 4 h later. Transfection was performed after 24 h in differentiation medium, with Lipofectamine RNAiMAX (Invitrogen). The final concentrations of most siRNAs were 25–30 nM, except for PCM-1, for which a final concentration of 500 nM was necessary to achieve efficient depletion. For nesprin-1 depletion, cells differentiated for 2–3 days were transfected twice with HiPerfect reagent (Qiagen) and 150 nM siRNA at 48 h intervals. For western blot analysis, cells were recovered by scraping, washed in PBS, and subsequently washed in ‘wash buffer’ (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, and 2 mM EDTA). Cells were resuspended in wash buffer containing 0.5% Triton X-100, 1 mM DTT and a cocktail of protease inhibitors (Complete/EDTA free, Roche). Cells were finally lysed by sonication. 70 µg of extracts were subjected to SDS-PAGE.

Differentiated C2C12 cells, scraped from petri dishes, were centrifuged in 10 mM HEPES, pH 8.0, 1.5 mM MgCl₂, 1 mM DTT, 1 mM PMSF, followed by lysis with 20 mM Tris-HCl, pH 7.0, 150 mM NaCl, 1% NP-40, 10% glycerol, 5 mM EDTA, and a protease inhibitor cocktail (Complete/EDTA free, Roche). After sonication and centrifugation at 16,100 g for 30 min at 4°C, the resulting nuclear pellets were extracted with 20 mM Tris-HCl, pH 7.0, 400 mM NaCl, 1% octyl β-D-glucopyranoside, 5 mM MgCl₂, 1 mM DTT and protease inhibitors. Protein extracts were diluted to a final concentration of 1 mg/ml in buffer containing 200 mM NaCl and 0.5% octyl β-D-glucopyranoside, and samples of 1 ml were incubated with rabbit anti-nesprin-1 or rabbit anti-PCM-1 antibody (see later for details) for several hours at 4°C. Immunoprecipitation was performed with 50 µl magnetic protein A beads (Biorad) per sample, followed by rinsing four times in the same buffer, and elution in gel-loading buffer containing 2.3% SDS.

Microtubules were depolymerized by treatment with 20 µM nocodazole for 40 min at 37°C, followed by 40 min on ice. Cells were rinsed once with PBS and twice with growth medium, before re-polymerization was induced in growth medium at 37°C. Dynein ATPase activity was inhibited by using erythro-9-[3-(2-hydroxynonyl)] adenine (EHNA) at 0.1 mM for 24–48 h. The Golgi was disassembled by treatment with 5 µg/ml Brefeldin A for 90 min.

The mouse nesprin-1 cDNA IRAKp961K22116Q was obtained from ‘Deutsches Ressourcenzentrum für Genomforschung’. A plasmid encoding mEmerald–nesprin-3 was obtained from Addgene (#54203), originally provided by Dr Michael Davidson, Florida State University, Tallahassee, FL. Nesprin-1α and the nesprin-1 KASH domain were amplified by PCR using the forward primers 5′-AAAAAAGAATTCATGGTGGTGGCAGAGGACTTG-3′ or 5′-AAAAAAGAATTCTCCCGTTCTGACCCCAGGCC-3′, respectively, in combination with the reverse primer 5′-TTTTTTGGTACCTCAGAGTGGAGGAGGACCGTT-3′. The resulting DNA was digested with EcoRI and KpnI, and cloned into the respective sites of pmCherry-C2 or pEGFP-C2 (Clontech). Transfection of myoblasts was performed using an Amaxa Nucleofector system according to the manufacturer's protocol (KitV, Lonzabio). Cells were induced to differentiate at 4 h to 6 h after transfection. For quantitative PCR (qPCR) experiments, total RNA was isolated from C2C12 myoblasts and myotubes using an RNeasy Mini Plus kit (Qiagen). Equal amounts of RNA (1 µg) were added to a reverse transcriptase reaction mix (Superscript II Reverse Transcriptase, Invitrogen) with random primers. The resulting cDNA was subjected to quantitative PCR using a Biorad C1000 thermal cycler, coupled to the CFX96 real-time system, using the SSoFast Evagreen Supermix (Biorad) for 40 cycles. Primers were used for nesprin-1 (forward, 5′-CTTCCTGTTCCGGATCCTC-3′; reverse, 5′-AGGTGAGTCCAATAAGCAGCA-3′), nesprin-2 (forward, 5′-ATGTCACCAGCCCAGAGG-3′; reverse, 5′-GACGGGCTACCAACTCCTTT-3′), as well as for two housekeeping genes, hypoxanthine guanine phosphoribosyl transferase (HGPRT1, forward, 5′-TCCTCCTCAGACCGCTTTT-3′; reverse, 5′-CCTGGTTCATCATCGCTAATC-3′), and ribosomal protein S16 (Rps16, forward, 5′-AGGAGCGATTTGCTGGTGTGG-3′; reverse, 5′-GCTACCAGGGCCTTTGAGATGG-3′).

Calibration curves were performed on each myoblast and myotube cDNA, to measure the efficiency of the primer pairs used. Corresponding reaction mix containing RNA without reverse transcriptase was used as a negative control for the qPCR of each sample. Levels of nesprin-1 and nesprin-2 cDNA were expressed relative to those of Rps16 and HGPRT1, using the Biorad CFX Manager software.

C2C12 cells grown on coverslips were fixed in methanol at −20°C and processed for immunofluorescence following standard protocols. For staining of γ-tubulin, microtubules and dynactin complexes, cells were pre-permeabilized in PEM buffer (80 mM PIPES, 5 mM EGTA, 2 mM MgCl₂, pH 6.8) containing 0.5% Triton X-100 (for 30 s to 5 min) or 0.05% saponine (5–15 min). The coverslips were rinsed with PEM containing 4% polyethylene glycol and fixed with 4% paraformaldehyde in PEM for an initial period of 5 min, followed by a second fixation step with paraformaldehyde in carbonate buffer (50 mM, pH 10) for 10 min. Antibody labelling was performed according to standard protocols.

The following antibodies were used: rabbit anti-nesprin-1 (1:500, cat. no ab24742, Abcam), mouse anti-nesprin-1 MANNESA(7A12) (1:100, Randles et al., 2010), rabbit anti-nesprin-2 (1:100, raised against mouse nesprin-2, amino acids 1465 to 1707 of clone NP_001005510), rabbit anti-nesprin-2, #4 (1:100, Zhang et al., 2007b), mouse mAb NSP-3 anti-nesprin-3 (1:100, Ketema et al., 2007), mouse anti-α-tubulin clone B-5-2-1 (1:1000, cat. no T5168, Sigma), rabbit anti-calnexin (1:200, cat. no SPA-860, Stressgen Bioreagents), mouse anti-γ-tubulin TU-30 (1:1000, cat. no 11-465-C025, Exbio), mouse anti-sarcomeric myosin MF20 (1:100, cat. no AB_2147781, DSHB), mouse anti-nuclear pore complex proteins, clone 414 (1:500, cat. no ab50008, Abcam), mouse anti-p150glued (1:100, cat. no 612708, BD Transduction Laboratories), mouse anti-pericentrin (1:100, cat. no 611814, BD Transduction Laboratories), rabbit anti-pericentrin (1:1000, cat. no PRB-432C, Covance), anti-KLC1 H75 (1:100, cat. no SC-25735, Santa Cruz Biotechnology), rabbit mAb anti-GM130 (1:100, cat. no ab52649, Abcam), rabbit and mouse anti-PCM-1 (1:500, Dammermann and Merdes, 2002), rabbit anti-SUN-1 and SUN-2 [raised against peptides NH₂-CHKLEPVFDSPRMSR-CONH₂ and NH₂-CLGRFTYDQEGDSLQ- CONH₂ for SUN-1 (used at 1:50), and NH₂-CVFKDSPLRTLKRKS- CONH₂ and NH2-CGTFAYDQDGEPIQT-CONH₂ for SUN-2 (used at 1:100), respectively; Covalab)]. We used Alexa-Fluor-488- and -568-conjugated secondary antibodies (Molecular Probes) for immunofluorescence, and horseradish-peroxidase-conjugated antibodies (Jackson ImmunoResearch Laboratories) for immunoblotting, followed by ECL.

We thank our colleagues at the Centre de Biologie du Développement (University Toulouse III, France) for their help and for stimulating discussions. We thank Glenn E. Morris (Keele University, Staffordshire, UK), Rener Xu (Fudan University, Shanghai, China) and Arnoud Sonnenberg (Netherlands Cancer Institute) for kindly providing antibody samples against nesprins 1, 2 and 3, respectively.