Introduction
Nemaline myopathy (NM) is a genetically heterogeneous disease of skeletal muscle caused by mutations in genes that are generally involved in muscle contraction, in particular those related to the structure and/or regulation of the thin filament. Mutations in
ACTA1 (skeletal muscle actin) or
NEB (nebulin) together make up the majority of cases, whilst other causative genes (to date,
TPM3,
TPM2,
TNNT1,
CFL2,
KBTBD13,
KLHL40,
KLHL41,
LMOD3,
MYPN, RYR3 or
MYO18B) together with unidentified genes, are implicated in the remainder [
20,
38]. These mutations result in weakness at the contractile level, while other cellular pathological hallmarks include dense accumulations of proteins known as nemaline rods, arrested muscle fibre growth, impaired fibre type differentiation, and disarray of contractile filaments [
20]. However, the underlying mechanisms behind many of these features remain uncertain, even though the mutations affecting thin filament structure and function are likely to be involved [
7,
17,
28,
29,
40,
42,
60,
63]. In the present study, we aimed to acquire a clearer understanding of muscle fibre dysfunction in NM by specifically studying nuclei and the related cortical cytoskeleton.
Skeletal muscle fibres are large syncytial cells containing many, often hundreds, of nuclei (termed myonuclei). Sufficient numbers and regular spacing of myonuclei throughout the muscle fibre are a prerequisite for its function, allowing the efficient delivery of gene products to all parts of the cell, with minimal transport distances. Therefore it is thought that each nucleus is responsible for maintaining a certain volume of the muscle fibre, termed the myonuclear domain [
5,
32]. The nuclei of skeletal muscle fibres are linked with various cytoskeletal components including non-sarcomeric/cytoplasmic actins, microtubules and intermediate filaments such as desmin. All three of these cytoskeletal networks have been implicated in the spacing and positioning of nuclei in models of skeletal muscle development: microtubules in the initial translocation/spacing of nuclei along the fibre [
12,
14,
15,
35], and actin and desmin in their movement to the fibre periphery [
45].
At the organelle level, nuclear function and transcription are regulated by a host of external factors; the nuclear envelope acts as a signalling hub that is capable of transducing a range of chemical and mechanical signals to regulate gene expression [
18,
34,
58]. The cytoskeleton is known to regulate nuclear shape and morphology via interactions with the nuclear envelope [
10,
23], a process which can itself impact on gene transcription, with different morphologies being linked to cell type, function, differentiation, and disease states [
10,
23,
62]. Given that the force-generating properties of NM muscle fibres are severely limited, we hypothesised that cytoskeletal components as well as nuclear function, positioning and integrity might be affected in this disease, and possibly contribute to pathology.
Using single muscle fibres from mouse models and NM patients with mutations in
ACTA1 or
NEB, we found that myonuclei display a range of defects, including irregular spacing and morphology, abnormal nuclear envelope and altered chromatin distribution. We also observed severe disruption within the microtubule, desmin and cytoplasmic (β- and γ-) actin networks, as well as alterations in their anchorage at the nuclear surface. We next sought to investigate the underlying pathological mechanisms, and found that impaired contractile force production is responsible for the nuclear spacing and morphological defects. We further demonstrated the role of a properly organised microtubule network in regulating nuclear shape. Our findings suggest that these alterations are likely to contribute to some of the features observed in NM, which include: broad transcriptional alterations and hindered muscle fibre growth [
30,
32] (perhaps due to the nuclear disruption which is likely to affect gene expression programmes); myofibril disarray (since desmin and the nuclear envelope are known to contribute to their organisation [
1,
6]); and altered mechanical properties of muscle fibres (known to be related in part to the cortical cytoskeleton) [
21]. In addition, our results suggest that nuclear and cytoskeletal defects might be a secondary feature and/or source of pathology in other (muscle) diseases, even in those where these structures are not primarily affected.
Materials and methods
Human subjects
All tissue was consented, stored, and used in accordance with the Human Tissue Act, UK, under local ethical approval (REC 13/NE/0373). Details of patients providing samples for light microscopy are given in Table
1, and of patients for electron microscopy in Tables
2 and
3. No electron microscopy samples were available from the patients used for light microscopy studies, hence, the different cohorts of patients for the two techniques.
Table 1
Nemaline myopathy patient muscle biopsy samples used for light microscopy
ACTA1
| 20 | M | c.16G>A | E6K | Copenhagen, Denmark |
ACTA1
| 30 | F | c.841T>C | Y281H | Genoa, Italy |
NEB
| 36 | F | c.2836-2A>G and c.5763+5G>A | Mutation in splice site | Copenhagen, Denmark |
NEB
| 56 | M | c.17234C>T and c.2271_22713del | R5745X and K7571del | Copenhagen, Denmark |
NEB
| 71 | F | c.508-7T>A and c.19097G>T | Mutation in splice site; and S6366I | Helsinki, Finland |
Table 2
Nemaline myopathy patient muscle biopsy samples used for electron microscopy
NEB
| 2 (a) | M | c.17737-2A>T and c.21315delA | Mutation in splice site and R7105 frameshift | Milan, Italy |
NEB
| 2 (b) | M | c.4082+5G>T (splice site) and g.112388C>T | Mutation in splice site and Q>X | Milan, Italy |
NEB
| 17 | M | c.22249A>C and c.8392-8395 duplication | T7417P and R2799L frameshift | Helsinki, Finland |
NEB
| 23 | M | c.11164C>T and c.19097G>T | R3722* (nonsense) and S6366I | Helsinki, Finland |
NEB
| 30 | F | c.508-7T>A and c.19097G>T | Mutation in splice site and S6366I | Helsinki, Finland |
ACTA1
| 11 weeks | F | c.796T>C | F226L | London, UK |
ACTA1
| 3 | M | c.235A>G | T79A | Milan, Italy |
ACTA1
| 10 | M | c.841T>C | Y281H | Milan, Italy |
Table 3
Patients with sporadic late-onset nemaline myopathy (SLONM), used for electron microscopy
SLONM 1 | 65 | 64 | M | IgG lambda | Rome, Italy |
SLONM 2 | 69 | 68 | F | Not available | Rome, Italy |
SLONM 3 | 72 | 67 | F | No | Leuven, Belgium |
SLONM 4 | 79 | 54 | F | Not available | Leuven, Belgium |
SLONM 5 | 34 | 32 | F | No | Leuven, Belgium |
Mouse models
Acta1H40Y mouse tibialis anterior muscle samples, with and without transfer of the
Myl4 gene remained frozen from a previous study [
30]. To summarise, data were collected from four wild type and four mutant male mice. At 4 weeks of age, the compartment containing the tibialis anterior muscle was injected with rAAV6 virus containing the
Myl4 transgene, and contralateral legs served as controls, being injected with virus lacking the functional gene. Mice were sacrificed 4 weeks later at 8 weeks of age. Colony maintenance and experiments were approved by the Uppsala Local Ethical Committee on Animal Research.
Neb cKO mice [
28] were maintained at the University of Arizona in accordance with the US National Institutes of Health guidelines “Using Animals in Intramural Research”. Mutants and WT/heterozygous littermates were sacrificed at 3 months of age by cardiac perfusion with 4% paraformaldehyde (PFA)/PBS, to properly preserve microtubule structure. Extensor digitorum longus muscles were dissected for whole-mount immunolabelling of cytoplasmic (β- and γ-) actins, desmin or microtubules (see below).
Antibodies
Primary antibodies were as follows (species, isotype, manufacturer, catalogue number, and dilution are given): lamin A (mouse monoclonal IgG3, Abcam, ab8980, 1:200); nesprin-1 (rabbit monoclonal IgG, Abcam, ab192234, 1:400); pericentrin (rabbit polyclonal IgG, Abcam, ab4448, 1:200); β-tubulin (clone TUB2.1, mouse monoclonal IgG1, Santa Cruz, sc-58886, 1:500); desmin (clone D33, mouse monoclonal IgG1, Dako, M076001-2, 1:400); β-actin (rabbit polyclonal IgG, Abcam, ab8227, 1:300); γ-actin (clone 2A3, mouse monoclonal IgG2b, Bio-Rad, MCA5776GA, 1:300); acetyl lys9/lys14 histone H3 (rabbit polyclonal IgG, cell signaling, #9677, 1:200).
Enzymatic isolation and culture of intact single muscle fibres
Intact single muscle fibres were prepared as described previously, using enzymatic dissociation with collagenase I (Sigma Aldrich) and gentle trituration (Sigma Aldrich) [
47]. After isolation, fibres were plated into 6-well plates (~ 30 fibres per well) in DMSO/high glucose/GlutaMAX supplement/pyruvate (Thermo Fisher Scientific, Cat# 31966021), containing 10% horse serum and 1% penicillin/streptomycin solution. Freshly isolated fibres were treated overnight with nocodazole (20 μM), taxol (10 μM) or epothilone D (10 μM). Final DMSO (vehicle) concentration was 0.5% in all cases. To assess myonuclear spacing, fibres were cultured for 72 h in the presence of the aforementioned drugs.
Immunohistochemistry (single muscle fibres)
Myofibres were fixed in 4% PFA/PBS for 15 min, and washed 3× in PBS. Fibres were permeabilised in 0.1% triton-X/PBS for 10 min, washed 3× and blocked in 10% normal goat serum/PBS for 1 h. Fibres were then treated with primary antibodies in blocking solution overnight (β-tubulin) or for 3 h (lamin A, nesprin-1, pericentrin) at 4 °C. Fibres were washed 3× in PBS for a total of 30 min, and then treated with Alexa 594 or 488-conjugated secondary antibodies and DAPI (all at 1:1000 in PBS) for 3 h. Fibres were washed 3× in PBS for a total of 30 min and mounted in Fluoromount mounting medium (Southern Biotech) with coverslip (thickness #1.5).
Immunohistochemistry (whole-mount muscles)
Neb cKO mice and WT/heterozygous littermates (all female) were sacrificed at 3 months of age by cardiac perfusion with 4% PFA/PBS, to properly preserve microtubule structure. Extensor digitorum longus muscles were dissected for whole-mount immunolabelling of cytoplasmic (β- and γ-) actins, desmin or microtubules. Muscles were then permeabilised in 0.5% triton-X/PBS (20 min) and 0.1% Triton-X/PBS (20 min), with each solution being replaced at least once during the incubation. Samples were then blocked in mouse-on-mouse block/PBS for 3 h, and then blocked in 8% bovine serum albumin overnight. Primary antibodies (tubulin, cytoplasmic actins) in blocking buffer were applied for 5 h, followed by 2 h washing in 0.1% triton-X/PBS.
Fluorescence imaging
Fibres were imaged on a Zeiss Axiovert 200 spinning disc confocal microscope equipped with BD CARV II and a motorised Z drive at 20× magnification (for imaging of nuclear morphology and nuclear envelope). For nuclear number and distribution, Z-stacks with 1 μm Z increments were taken through the entire depths of fibres, as described previously [
48,
49]. For imaging of cytoskeleton and nuclear volume, a Nikon A1 laser scanning confocal microscope with a 100× oil immersion objective (1.4 NA) was used, with Z-stacks taken with 0.3 μm Z increments (Nikon Imaging Centre, King’s College London).
Image analysis
Analysis of nuclear number and spacing: Coordinates of myonuclei were identified in 3D within Z-stacks of muscle fibres. A custom-made Matlab programme was used to a measure fibre CSA, nuclear number, nearest neighbour distances and order score (‘g’) of nuclei within fibres, as described previously [
4,
48].
Analysis of nuclear shape parameters: For 2- and 3-dimensional measurements (area, aspect ratio, circularity, and volume), nuclei in the DAPI channel were thresholded by pixel intensity until fully highlighted. Inbuilt ImageJ functions were used to measure 2D parameters and Voxel Counter plugin for volume. For accurate shape analysis, nuclei positioned on the sides or the backs of fibres (relative to the microscope objective) were excluded, as were those in clusters where nuclei were touching.
Microtubule quantifications: density (% area) was calculated on binary converted images in ImageJ. Microtubule directionality was calculated using the TeDT tool [
31].
Statistics
Graphs were prepared and analysed in Graphpad Prism. Linear regression lines and statistical comparisons were performed using inbuilt algorithms (ANCOVA test was used to compare elevations/intercepts and slopes of different regression lines). For statistical comparisons of nuclear organisation and nuclear shape in human subjects (column graphs), control data points were pooled, since no significant age-related differences were observed amongst healthy control subjects. Owing to the different origins of disease (mutation and gene affected), patient data points were not pooled. One-way ANOVA with Tukey post-correction was used to compare each patient against the pooled controls; in addition, a random effect algorithm was incorporated into the model, to account for any potential inter-individual differences that might exist amongst the control cohort. For studies with animals, no significant differences were observed between animals of the same genotype, and as such, individual data points for both control and mutant/treated animals were pooled. For column comparisons in the animal studies, a two-tailed t test was used to compare two groups, and a one-way ANOVA with Tukey post-correction was used to compare more than two groups. For drug treatments with nocodazole, taxol, and epothilone D in ex vivo culture experiments, a cumulative probability test (Kolmogorov–Smirnov test) was used to compare groups against control/vehicle only. Asterisks denote the following statistical significance levels: *P < 0.05), **P < 0.01, ***P < 0.001.
Discussion
In this study, we identify a range of nuclear and cytoskeletal defects in skeletal muscle fibres that occur as part of ACTA1 or NEB-related NM pathology. We show that such defects in nuclear morphology and spacing are caused by the impairment in force production that is a characteristic of this disease. We also highlight the role of the microtubule cytoskeleton in the regulation of nuclear shape.
Myonuclear spacing defects (Fig.
1) have also been observed in other mutant mouse models, including those in genes encoding nuclear envelope proteins (SUN1/SUN2 double KO, Nesprin-1 KO [
26,
57,
64]), and also in ageing [
5]. Interestingly, this effect appears to be specific, since knockout of other nuclear envelope proteins such as lamin A/C does not significantly alter myonuclear organisation [
16,
27]. Currently, there is little insight into whether defects in nuclear spacing contribute to myofibre dysfunction, or whether they are merely a secondary phenomenon. However, given that regular spacing of myonuclei is a highly conserved feature across invertebrates and vertebrates, it is assumed that it is important for proper muscle fibre function, such as inter-nuclear cooperation and the efficient distribution of gene products throughout the cell [
32]. Therefore, one might envisage that any deviation from a “normal” nuclear arrangement would result in sub-optimal muscle fibre function.
Studies to ascertain the mechanisms of nuclear spacing in skeletal muscle have largely taken place in myotubes and embryos of
Drosophila and mouse, and have identified a number of mediators including nesprins, microtubules, MTOC proteins and the motor proteins kinesin and dynein [
14,
15,
57]. Myotubes, as an in vitro system, are analogous to the events during embryonic development, whereby myonuclei are located in the centre/core of the fibre, where their spacing takes place along the axis of the fibre. As such, myotubes are not anatomically or developmentally equivalent to mature muscle fibres, in which nuclei are anchored at the fibre periphery. In these examples of NM, it is unclear whether the nuclear organisation defects are a result of aberrant spacing occurring during development, or in events that occur in maturity. However, our results show that nuclear positioning can be remodelled in mature fibres, since delivery of the
Myl4 isoform into adult
Acta1H40Y mice rescued nuclear spacing defects (Fig.
4). A key feature of muscle tissue in NM (as well as some other myopathies) is a shift towards type I fibres. This alone is unlikely to explain the alterations that we observed in NM patients, since no clear difference in the regularity of nuclear spacing is observed across human fibre types [
9].
Various nuclear envelope and shape defects are also a feature of NM patients (Figs.
2,
3). Perhaps associated with this is the frequent observation of chromatin abnormalities in NM patients by EM and light microscopy (Fig.
3, Suppl Fig. S2, online resource), since both nuclear morphology and the nuclear envelope are known to play a role in transcriptional regulation via chemical and mechanical control of chromatin organisation [
10,
18,
23,
34,
58,
62]. Given that NM is caused by mutations in genes related to contraction, it would appear that these alterations are a secondary defect. However, many of these alterations bear striking resemblance to those seen in primary diseases of the nuclear envelope, for instance those caused by mutations in genes encoding, e.g., lamins, nesprins, or emerin, which frequently present with muscular dystrophy [
2,
13,
19,
33,
52,
55]. Given the severity of disease caused by mutations in genes encoding nuclear envelope proteins, it is likely that the array of nuclear defects that we describe in NM patients contributes to muscle dysfunction. Indeed, broad alterations in the transcriptional profile of skeletal muscle are observed in patients with NM (including in genes related to metabolism and calcium homeostasis), and this may partially result from these characteristic nuclear shape and envelope alterations and/or reorganised chromatin [
51]. It should be noted that intranuclear rods are also a feature of some cases of NM, which might also be expected to affect nuclear function; however, we did not observe any instances of these in the patient samples used in this study (either by EM, or by α-actinin staining in light microscopy samples).
Delivery of the
Myl4 transgene to adult
Acta1H40Y mouse muscles augments contractile force [
30] and rescues nuclear spacing and morphology (Fig.
4). This suggests that (i) a lack of force originating at the sarcomere is responsible for these nuclear defects, and that (ii) this can be reversed by an increase in contractile capacity. The mediators of this effect are unclear, but may involve direct force transmission to the nuclei via, e.g., microtubule, actin or desmin cytoskeletons, or interactions through second messengers that may be responsive to mechanical input. Interestingly, nuclear defects at the ultrastructural level were also observed in patients with acquired forms of NM (SLONM; Suppl Fig S3, online resource; Table
5), where a reduction in muscle contractile force also occurs, but in the absence of mutations in known disease-causing genes. This provides some evidence that contractile dysfunction is a cause of nuclear defects in humans, although this data are currently only supportive. In a broader sense, nuclear abnormalities might be common to other neuromuscular diseases, since various other classes of myopathy are also associated with impaired contraction [
20]. One example that has been studied in detail is Marinesco–Sjögren syndrome, a multisystem disorder with myopathy, which is caused by mutations that affect endoplasmic reticulum trafficking and chaperone function [
25,
46]. Ultrastructural abnormalities in myonuclei of this disorder include highly condensed chromatin and areas of nuclear envelope separation, akin to our observations in nemaline myopathy. It should be noted that disease etiology is highly variable across neuromuscular disorders, and that various aspects of muscle pathology might influence nuclear function and integrity, besides impaired contractile function.
Another of our key findings was the cytoskeletal defects in the
Neb cKO model NM, involving microtubules, desmin and non-sarcomeric actins (Figs.
5,
6). Microtubules also show increased density and disorganisation in dystrophic mice with dystrophin or sarcoglycan deficiency, or mice with MAP6 ablation [
3,
54]. In skeletal muscle, microtubules are known to have roles in modulating fibre stiffness and contraction, and in signalling via reactive oxygen species (ROS) [
21]. Indeed, the increased microtubule density in dystrophic mice results in elevated ROS, over-activation of stretch-sensitive Ca
2+ channels, and worsened pathology [
22]. In the mouse model of NM utilised in the present study, defects in non-sarcomeric (β- and γ-) actins and desmin were striking in that the vast majority of fibres showed markedly reduced/virtually absent localisation of all of these components. Ablation of either β- or γ-actin specifically in skeletal muscle causes mild progressive myopathies [
39], and loss of desmin causes disruptions of muscle architecture [
6]. Intriguingly, myofibrils are frequently misaligned and disordered in NM [
60], and this might be due to (i) the loss of desmin, and/or (ii) the defects at the nuclear envelope, since both are involved in the proper arrangement of myofibrils [
1,
6]. One key role of desmin and actins is the lateral transmission of force to the fibre periphery, where actin and microtubules bind to the dystrophin-associated glycoprotein complex at the plasma membrane [
3,
43,
56]. Therefore, the mislocalisation of these components is likely to have implications for the mechanical properties of the fibre.
Cytoskeletal components also have important roles in anchorage at the nucleus, and the normal localisation of microtubules, γ-actin and desmin at the nuclear surface were also largely reduced in
Neb cKO mice. The cytoskeleton is likely to transmit strain from the mechanical forces of contraction to the nuclei, which may be important for fibre integrity and/or regulation of gene expression [
8]. Consistent with this, disruption of microtubules with several agents resulted in alterations to nuclear shape (Fig.
7). Although microtubules are known to regulate nuclear spacing during development, no changes in nuclear distribution were observed when fibres were treated with nocodazole, taxol or EpoD (Fig.
7; Suppl Fig. S4, online resource). Possible explanations for this include: (i) other cytoskeletal systems are instead responsible for nuclear spacing in mature fibres, or are able to compensate when microtubules are disrupted; (ii) nuclei are more mobile in actively contracting fibres, which was not the case in these experiments; or (iii) longer treatments are required to induce significant remodelling of nuclei (which would not be preferable due to the relatively short-term viability of muscle fibres in ex vivo conditions).
The range of disease severity varies greatly in NM, and death in childhood is frequent at the most severe end of the spectrum. In this study, the patients were almost entirely of adult age, representing the milder end of the spectrum (with the exception of some early-onset cases included for electron microscopy studies). This was an experimental design choice, due to the availability of healthy human control tissue at adult, but not childhood ages. Therefore, it is difficult to draw extensive conclusions regarding nuclei in typical congenital cases of NM, although several early-onset cases displayed invaginations, altered chromatin compaction and in one case, separation of inner and outer nuclear membranes (Fig.
3, Table
4). To date, our understanding of human muscle development at the cellular level during infancy and childhood is incomplete, and is likely to be highly dynamic throughout this period of sustained growth [
11]. As such, interpretation of cellular organisation in congenital NM patients would be difficult, without appropriate information on healthy development.
In summary, our results demonstrate that skeletal muscle from NM patients and mouse models display defects in the non-sarcomeric cytoskeleton and in nuclear positioning and integrity. They indicate that abnormal nuclear spacing and morphology are the result of the impaired contractile force production that is a key feature of this disease. In addition, we highlight the role of a properly organised cytoskeleton in the regulation of nuclear morphology. Although nuclear defects are observed in other diseases, including those caused by mutations in nuclear envelope proteins, these findings are somewhat unexpected, given that NM pathology originates at the sarcomere [
7,
41]. They might explain some of the features observed in NM, such as broad transcriptional changes and hindered muscle fibre growth (possibly related to alterations in nuclear envelope and chromatin organisation, which is likely to affect programmes of gene expression [
18,
34,
62]), myofibrillar disarray (due to the roles of desmin and the nuclear envelope in sarcomere organisation [
1,
6]) and altered fibre mechanical properties (due to disrupted cytoskeletal arrangements [
21]). This study raises the possibility that nuclear and cytoskeletal defects may be an overlooked feature and/or source of pathology in other (muscle) diseases.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.