Excitation–contraction coupling (ECC) is the physiological mechanism whereby an electrical signal detected by the dihydropyridine receptor, is converted into an increase in [Ca2+], via activation of ryanodine receptors (RyRs). Mutations in RYR1, the gene encoding RyR1, are the underlying cause of various congenital myopathies including central core disease, multiminicore disease (MmD), some forms of centronuclear myopathy (CNM) and congenital fibre-type disproportion. Interestingly, patients with recessive, but not dominant, RYR1 mutations show a significant reduction in RyR protein in muscle biopsies as well as ophthalmoplegia. This specific involvement of the extraocular muscles (EOMs) indicates that this group of muscles may express different amounts of proteins involved in ECC compared with limb muscles. In the present paper, we report that this is indeed the case; in particular the transcripts encoding RyR3, cardiac calsequestrin (CSQ2) and the α1 subunit of the cardiac dihydropyridine receptor are up-regulated by at least 100-fold, whereas excitation-coupled Ca2+ entry is 3-fold higher. These findings support the hypothesis that EOMs have a unique mode of calcium handling.
Extraocular muscles (EOMs) are among the fastest and most fatigue-resistant skeletal muscles . More than 20 years ago, they were categorized as a separate group of muscles or ‘allotype’ since they represent a unique group of highly specialized muscles, anatomically and physiologically different from other skeletal muscles . Indeed, their embryonic origin is distinct from that of limb muscles: the former, together with the striated muscles of the face, jaw and throat, develop from the first seven somitomeres whereas the latter derive from somites . In humans, there are six types of EOMs and they are characterized by the presence of two layers: the inner global layer and the outer orbital layer . The fibre type classification of limb muscles does not fit the six fibre types described in EOMs  and the innervation of EOMs is also different, as they are innervated by cranial nerves and contain both singly innervated muscle fibres (SIFs) and multiply innervated fibres (MIFs), whereas mammalian limb muscle fibres are singly innervated by motoneurons originating from the spinal cord . The distinct origin and innervation of EOMs are probably responsible for the differences in gene and protein expression that have been described in mammalian EOMs .
The ryanodine receptor 1 (RyR1) is the calcium channel of striated skeletal muscle responsible for releasing Ca2+ from the sarcoplasmic reticulum leading to muscle contraction [7,8]. Over the last two decades, more than 200 mutations in RYR1 (the gene encoding RyR1) have been identified in the human population and linked to neuromuscular disorders and/or the pharmacogenetic disorder malignant hyperthermia [9,10]. Whereas the dominant RYR1 mutations are predominantly associated with central core disease and/or a susceptibility to malignant hyperthermia, recessive mutations are found in patients with multiminicore disease (MmD), centronuclear myopathy (CNM) and congenital fibre-type disproportion [10–15]. Interestingly, patients with recessive, but not dominant, RYR1 mutations show a decreased amount of RyR1 protein in biopsied muscles, as well as specific involvement of EOM (ophthalmoplegia) [11–14,16]. Selective involvement of EOM or lack thereof, also characterizes particular neuromuscular diseases: in myasthenia gravis and mitochondrial myopathies they are the first and most affected muscle group  whereas they are characteristically spared from pathology in aging, Duchenne muscular dystrophy and congenital muscular dystrophy [18–20]. Such sparing of EOM in muscular dystrophies has been attributed to constitutive differences between EOM and other skeletal muscles  but the factors controlling these differences remain largely unknown.
In order to help clarify the cause(s) for the specific involvement of EOM in patients with recessive RYR1 mutations, we investigated their excitation–contraction coupling (ECC) machinery and calcium homoeostasis. Our results show that there are significant differences between leg muscles (LMs) and EOM; in particular the latter have developed a chimaeric configuration in that they express significantly lower levels of RyR1, the α1 subunit of the dihydropyridine receptor (Cav1.1) and calsequestrin-1 (CSQ1), whereas the cardiac isoforms of the Cav1.2 and CSQ2 are highly expressed as is RyR3. Such changes result in different characteristics of calcium homoeostasis, as myotubes explanted from EOM exhibit a large component of excitation coupled Ca2+ entry (ECCE). The results of the present study shed light on the underlying causes leading to EOM involvement in congenital muscle disorders due to RYR1 mutations causing a decrease in RyR1 protein.
Human muscle cell cultures
Primary muscle cell cultures were established from fragments of quadriceps muscles obtained from biopsies of healthy donors (five donors) and EOM samples obtained from patients undergoing squint corrective surgery (four donors), as described previously [21,22]. Cells were cultured on laminin-coated 0.17-mm-thick glass coverslips in growth medium (Skeletal Muscle Cell Growth medium; Promo Cell) and induced to differentiate into myotubes by culturing them in differentiation medium (Skeletal Muscle Cell Differentiation medium; Promo Cell) for 7–14 days. This research was carried out in accordance with the Declaration of Helsinki (2013) of the World Medical Association and was approved by the Ethikkommission beider Basel (permit number EK64/12); all subjects gave written informed consent to carry out this work.
Myotubes were loaded with fura-2 (Calbiochem) or fluo-4 (Life Technologies) (final concentration 5 μM) in differentiation medium for 30 min at 37°C, after which the coverslips were mounted on to a 37°C thermostatically-controlled chamber which was continuously perfused with Krebs–Ringer medium; individual cells were stimulated by means of a 12- or 8-way 100 mm diameter quartz micro-manifold computer-controlled microperfuser (ALA Scientific Instruments), as described previously . For global changes in the intracellular Ca2+ concentration, the fluorescent ratiometric Ca2+ indicator fura-2 was used. Online measurements were recorded using a fluorescent Axiovert S100 TV inverted microscope (Carl Zeiss) equipped with a 20× water-immersion FLUAR objective [0.17 numerical aperture (NA)] and filters (BP 340/380, FT 425, BP 500/530) and attached to a Cascade 128+ CCD (charge-coupled device) camera. Changes in fluorescence were analysed using Metamorph imaging software (Molecular Devices) and the average pixel value for each cell was measured at excitation wavelengths of 340 nm and 380 nm as described previously [21,22]. Fura-2 fluorescent ratio signals were converted into [Ca2+] using the fura-2 Ca2+ imaging calibration kit from Molecular Probes/Invitrogen following the manufacturer's instructions, as previously described . The dynamics of [Ca2+] influx were investigated by TIRF (total internal reflection fluorescence) microscopy using the fast Ca2+ indicator fluo-4, as described previously . Briefly, differentiated human myotubes were mounted on a thermostatically controlled perfusion chamber, bathed continuously in Krebs–Ringer containing 2 mM Ca2+. ECCE was measured after the application of 60 mM KCl to myotubes pre-treated with 50 μM ryanodine (Calbiochem) to block RyR1-mediated Ca2+ release. Nifedipine (50 μM) (Calbiochem) was used to block the dihydropyridine receptor or 0.5 mM EGTA to chelate Ca2+. Online fluorescence images were acquired using an inverted Nikon TE2000 TIRF microscope equipped with an oil immersion CFI Plan Apochromat 60× TIRF objective (1.49 NA) and an electron multiplier Hamamatsu CCD camera C9100–13. Data were analysed using Metamorph imaging software (Molecular Devices).
Quantitative real-time PCR
Total RNA was extracted from muscle biopsies using TRIzol® (Invitrogen) and following the manufacturer's protocol. The RNA was first treated with DNase I (Invitrogen) and then 1000 ng were reverse-transcribed using the high-capacity cDNA reverse transcription kit (Applied Biosystems); cDNA was amplified by quantitative real-time PCR (qPCR) using SYBR Green technology (Fast SYBR Green Master Mix, Applied Biosystems) as described previously . The sequence of the primers used to amplify and quantify the different genes is given in Supplementary Table S1. qPCR was performed on a 7500 Fast Real-Time PCR machine from Applied Biosystems using the 7500 software v2.3. The standard protocol was selected and the running method consisted of a holding stage at 50°C for 20 s and a denaturing step at 95°C for 10 min, followed by 40 cycles of denaturing at 95°C for 15 s and an extension step at 60°C for 1 min. Gene expression was normalized to expression ACTN2, which is present in all muscle fibre types. Results are expressed as fold change compared with expression of the gene in LMs.
Glass coverslip grown and differentiated myotubes were fixed with 4% paraformaldehyde (made in Phosphate buffered saline (PBS)), permeabilized with 1% Triton in PBS for 20 min and processed as previously described . The following antibodies were used: mouse anti-RyR1 (Thermo Scientific; MA3–925), goat anti-Cav1.1 (Santa Cruz Biotechnology, sc-8160), rabbit anti-Cav1.2 (Santa Cruz Biotechnology, sc-25686), Alexa Fluor® 488-conjugated chicken anti-rabbit, Alexa Fluor®-conjugated donkey anti-goat IgG (Life Technologies) and Alexa Fluor®-conjugated goat anti-mouse IgG (Life Technologies). Cells were stained with DAPI (Life Technologies) to visualize nuclei and observed using a Nikon A1R confocal microscope with a CFI Apo TIRF 100× (1.49 NA) objective.
Electrophoresis and immunoblotting
The total sarcoplasmic reticulum fraction was isolated from flash-frozen muscle samples (human EOM, quadriceps muscles and mouse heart) stored in liquid nitrogen as previously described . Protein concentration was determined using Protein Assay Kit II (Bio-Rad Laboratories) using BSA as a standard. SDS/PAGE, protein transfer on to nitrocellulose membranes and immunostaining were performed as described previously . The following primary antibodies were used: mouse anti-RyR1 (Thermo Scientific, MA3–925), goat anti-Cav1.1 (Santa Cruz Biotechnology, sc-8160), rabbit anti-Cav1.2 (Santa Cruz Biotechnology, sc-25686), rabbit anti-CSQ-1 (Sigma, C-0743) and CSQ2 (Abcam, ab-3516), goat anti-SERCA1 (sarcoplasmic/endoplasmic reticulum Ca2+-ATPase 1) (Santa Cruz Biotechnology, sc-8093), goat anti-SERCA2 (Santa Cruz Biotechnology, sc-8095), mouse anti-sarcalumenin (Thermo Scientific, MA3-932), goat anti-junctophilin1 (Santa Cruz Biotechnology, sc-51308). Secondary horseradish peroxidase conjugates were Protein G–horseradish peroxidase (Sigma, P8170) and horseradish peroxidase-conjugated goat anti-mouse IgG (Sigma, A2304). The immunopositive bands were visualized by chemiluminescence using the Super Signal West Dura kit (Thermo Scientific) or the chemiluminescence kit from Roche.
Statistical analysis was performed using Student's t test for two populations. Values were considered significant when P<0.05. When more than two groups were compared, analysis was performed by the ANOVA test followed by the Bonferroni post-hoc test, using the GraphPad Prism 4.0 software. The Origin Pro 8.6 software was used for generating dose–response curves and calculating EC50 values.
Excitation–contraction coupling gene and protein expression levels in human EOM
Figure 1(A) shows the expression level as assessed by qPCR, of the main genes involved in skeletal muscle ECC in EOM biopsies, compared with quadriceps biopsies from healthy donors; the latter were taken as the reference tissue and the expression of the different set of muscle-specific genes in human quadriceps was set as 1. The results represent the mean expression levels of samples pooled from four to five biopsies and are expressed on a logarithmic scale. The most interesting finding is that the expression levels of major genes involved in skeletal muscle ECC, namely RYR1, CACNA1S, SERCA1 and CASQ1 were ~10-fold lower (P<0.0001) than in LM biopsies. Intriguingly, the transcript level of RYR3 was ~800-fold higher in EOM (P<0.0001). A second interesting result was that the expression level of CASQ2 was found to be ~100-fold higher in EOM (P<0.0001) as was that of the structural proteins TRDN and ASPH1 (~10-fold and ~1000-fold respectively; P<0.0001) and of the calcium buffering and binding proteins SRL and PVALB (10-fold; P<0.0001) As expected, the MYH1 transcript level was significantly lower in EOM (100-fold; P<0.0001) than in LMs, since this isoform is mainly expressed in slow-twitch muscles. In contrast, the transcript of MYH13, which is the isoform characteristically expressed in eye muscles, was ~5000-fold higher in EOM (P<0.0001), corroborating the validity of our assay. Taken together, these results indicate that the gene expression pattern of the main components involved in ECC of human EOM is specific and distinct from that of other striated skeletal muscles.
Expression of ECC transcripts and proteins in human EOM biopsies
In order to confirm the changes in gene expression on the protein level, we prepared the total sarcoplasmic reticulum fractions and probed them with a panel of antibodies against skeletal and cardiac ECC protein components. The protein expression levels were then normalized for SERCA2 whose content was similar in LM and EOM as assessed by real time PCR (Figure 1A) and immunoblotting. Figure 1(B) shows the representative Western blots and quantitative histograms, as can be seen, for RyR1, Cav1.1, SERCA1 and CSQ1 the protein levels are decreased by more than 50% in EOM compared with LM samples. The cardiac isoform of the RyR2 was not detectable in EOM (Supplementary Figure S1) whereas Cav1.2, the cardiac isoform of the Cav CSQ2, as well as sarcalumenin, were significantly up-regulated in EOM (2-fold for Cav1.2 and by 50% for CSQ2 and sarcalumenin); qPCR could not be used reliably to compare the expression levels of the Cav1.2 transcript because of its extremely low level of expression in leg skeletal muscles (the reference tissue in the qPCR); its level of expression is comparable with that found in mouse heart microsomes (Supplementary Figure S1).
Ca2+ homoeostasis in EOM cultured myotubes
Muscle biopsies have been used as a starting material to obtain primary muscle cultures from patients with different neuromuscular disorders and the myotubes that have been obtained have been a useful model to study ECC under normal and pathological conditions [21,22,24,25]. In the present study, we successfully used a protocol similar to that used for LM biopsies, in order to obtain EOM-derived myotubes (Figure 3A). Once a sufficient number of myoblasts grew out of the biopsy, they were further expanded and then induced to differentiate into myotubes. Figure 2 shows photomicrographs of myotubes stained with anti-RyR1 (red, Figure 2A), anti-Cav1.1 (green, Figure 2B) merged plus DAPI (Figure 2C) and anti-RyR1 (red, Figure 2D), anti-Cav1.2 (green, Figure 2E) and merged plus DAPI (Figure 2F) and observed by confocal microscopy. As can be seen, the myotubes are positive for RyR1 and Cav1.1 and the two proteins co-localize within an intracellular membrane compartment, but as previously established  their distribution is punctuated, not highly regular and lacks the distinct structure observed in mature fibres. The localization of Cav1.2, however, is clearly different from that of Cav1.1 (compare Figures 2B and 2E); indeed the Cav1.2 appears to be almost exclusively localized on the plasma membrane. These results point to a potential alternative form of ECC in EOM-derived myotubes and we next studied calcium homoeostasis in these cells. Results from LM-derived myotubes from five donors and EOM-derived myotubes from four donors were pooled and averaged. Figure 3(B) shows that the KCl-induced Ca2+ release from intracellular stores (in the presence of 100 μM La3+ and therefore reflecting skeletal ECC) was similar in LM and EOM-derived myotubes. However, in EOM-derived myotubes the EC50 for 4-chloro-m-cresol-induced Ca2+ release was increased (332±26 and 234±58 μM for EOM and LM respectively). Additionally, Figure 3(C) shows that the resting [Ca2+] in EOM-derived myotubes is significantly lower than that of LM-derived myotubes (P<0.0001; Student's t test), whereas the size of the rapidly releasable intracellular Ca2+ stores is more than 2-fold larger in EOM-derived myotubes compared with LM-derived myotubes (Figure 3D).
Cellular localization of RyR1, Cav1.1 and Cav1.2 in differentiated EOM-derived myotubes
Global calcium homoeostasis of EOM-derived myotubes
Plasma membrane depolarization of skeletal muscle cells is accompanied by Ca2+ influx through the dihydropyridine receptor and this phenomenon is defined as ECCE [24,27,28]. We next studied ECCE in EOM-derived myotubes using a TIRF microscope . The top panels in Figure 4(A) show a representative myotube pre-treated with 50 μM ryanodine and stimulated with 60 mM KCl; Figure 4(B) shows a representative Ca2+ influx trace initiated by the application of 60 mM KCl in EOM-derived myotubes (continuous line) and LM-derived myotubes (dotted line). The change in fluo-4 fluorescence represents Ca2+ influx via the dihydropyridine receptor and not release from stores as: (i) myotubes were incubated with 50 μM ryanodine in order to block Ca2+ release via RyR; (ii) pre-incubation of myotubes with ryanodine plus 50 μM nifedipine blocked the Ca2+ increase (Figure 4C); and (iii) Ca2+ influx was inhibited when the experiment was performed in Krebs–Ringer solution containing 0.5 mM EGTA (Figure 4C). Thus, compared with LM-derived myotubes, depolarization-induced Ca2+ influx in EOM myotubes was 3-fold higher (Figure 4D).
EOM-derived myotubes exhibit a larger depolarization-induced Ca2+ influx compared with LM-derived myotubes
In the present study, we characterized the ECC machinery of human EOM and report that these muscles are particular in as much as they express important components of the cardiac ECC machinery; in addition, we report for the first time that primary muscle cell cultures can be obtained from biopsies of patients undergoing squint or corrective surgery. Such myotubes maintain their phenotypic characteristics since they exhibit intracellular calcium regulation that is enhanced compared with primary cultures of LM-derived myotubes. The present study is important as it shows the feasibility of performing such experiments not only in biopsies obtained from normal individuals, but also can be extended to patients suffering from neuromuscular disorders causing ophthalmoplegia.
In a previous study, Porter et al.  used a bioinformatic gene profiling approach to identify key differences between murine EOM, LMs and jaw muscles. In their study, they scanned the expression pattern for transcripts encoding proteins involved in transcriptional regulation, signal transduction, intermediary metabolism and sarcomeric ECC. As far as the ECC coupling machinery is concerned, our results do not support their findings as we found significant changes between leg and EOM expression of CSQ1 and CSQ2, RyR1, SERCA1 and Cav1.1, triadin, sarcalumenin and parvalbumin and no change in mitsugumin-29. Such inconsistencies may be due to the fact that (i) they used mouse skeletal muscle and we used human biopsies, (ii) we used qPCR on pooled samples; (iii) the biopsies we analysed were derived from patients with strabismus and therefore not ‘normal’ muscles. Indeed the pattern of expression of transcripts in strabismic and normal EOMs has been reported to be different ; and (iv) we normalized all qPCR results for the content of the muscle-specific gene ACTN2. In support of our findings, it was shown that there are important differences between human EOM and those of other species .
An interesting observation of the present report is that EOM express low levels of RyR1 but high levels of RyR3, a finding that may explain why patients with recessive ryanodinopathies leading to decreased expression of RyR1 characteristically show extraocular involvement [11,15] whereas those with dominant RYR1 mutations do not. It is plausible that since the levels of expression of RyR1 are low in EOM, mutations leading to a further decrease in its level of expression will severely affect muscle function, leading to ophthalmoplegia. In this context, it should be mentioned that the higher levels of RyR3 expressed in EOM do not compensate for the decrease in RyR1, indicating that the RyR3 isoform must be involved in other aspects of calcium release, cannot be directly activated by the Cav1.1 voltage sensor and cannot functionally replace RyR1. On the other hand, EOM also express high levels of the Cav1.2 and of CSQ ; as far as CSQ2 is concerned, it has half the calcium-binding capacity of CSQ1 , it is phosphorylated to a higher stoichiometry and at least 50-fold more rapidly . Though the role of phosphorylation is unclear, it is thought to increase the Ca2+-binding affinity and to assure that the levels of Ca2+ near the Ca2+ release sites are elevated . The observed up-regulation of Cav1.2 is both novel and intriguing as this isoform is mainly expressed in the heart, where it functions as a voltage sensor and Ca2+ channel, activating RyR2 via a Ca2+-induced Ca2+ release mechanism . In EOM-derived myotubes, this isoform was clearly distributed on the plasma membrane and since RyR2 was not detectable in EOM, either these muscles have evolved a chimaeric cardiac/skeletal ECC, with Cav1.2 activating RyR1, or Cav1.2 may be coupled to RyR3 or it may be involved in Ca2+ influx following plasma membrane depolarization. Such chimaeric expressions of RyR isoforms are not uncommon in smooth muscle cells that are reported to express different combinations of calcium channels and L-type voltage sensors, depending on their tissue of origin and physiological requirements .
Zeiger et al.  showed that rat eye muscle-derived myotubes exhibit superior calcium homoeostasis compared with leg tibialis anterior-derived myotubes and release double the amount of calcium after treatment with ionomycin than LM-derived myotubes. The results of the present study support and extend these findings; indeed human EOMs not only have larger intracellular Ca2+ stores and express different calcium-buffering proteins, but also show lower resting [Ca2+] and enhanced depolarization-induced calcium influx (or ECCE). The enhanced calcium-handling capacity may be due to the higher expression levels of CSQ2 and of the calcium-binding proteins parvalbumin and sarcalumenin. Parvalbumin is a cytosolic high-affinity Ca2+-binding protein preferentially expressed in fast twitch muscles whose function is to promote muscle relaxation [39,40], whereas sarcalumenin is localized on the longitudinal sarcoplasmic reticulum where it is involved in stabilizing the Ca2+ pump and in maintaining rapid contraction and relaxation rates . The physiological function of depolarization-induced calcium influx is far from clear but it is postulated to be important for refilling of intracellular Ca2+ stores, which is essential during repetitive stimulation [42,43]. Interestingly EOM-derived myotubes have significantly larger intracellular calcium stores and enhanced Ca2+ influx and the latter is probably due to the high levels of expression of Cav1.2.
A final observation resulting from the present study and from a previous study  concerns the phenotype of satellite cells; depending on their muscle of origin, be it fast or slow twitch or EOM, the satellite subpopulations are intrinsically different  and they maintain the specific characteristics particular to the muscle from which they originated even after explantation and culture.
In conclusion, the present study provides insights into the ECC characteristics of human EOM. Reduced expression of RyR1, CSQ1 and SERCA1 is a feature of great significance in the context of ophthalmoplegia and neuromuscular disorders. Taking into account the importance of exquisitely efficient calcium management in order to achieve the demanding physiological properties of the EOMs, the increased levels of CSQ2, parvalbumin and sarcalumenin together with reported higher depolarization-induced calcium influx and particular high expression of Cav1.2 indicate that this group of muscles which is in constant use, relies on a chimaeric skeletal/cardiac ECC configuration.
Marijana Sekulic-Jablanovic performed the experiments, analysed the data and drafted the article. Anja Palmowski-Wolfe performed surgeries, provided the biopsies and critically revised the paper for important intellectual content. Francesco Zorzato was responsible for conception and design of the experiments, interpretation of data and critically revised the paper for important intellectual content. Susan Treves was responsible for conception and design of the experiments, collection, analysis and interpretation of data and drafted the article.
We gratefully acknowledge the technical support of Anne-Sylvie Monnet.
This work was supported by the Swiss National Science Foundation (SNF) [grant number 31003A-146198]; and the Department of Anesthesia Basel University Hospital.
α1 subunit of the dihydropyridine receptor
excitation-coupled Ca2+ entry
quantitative real-time PCR
sarcoplasmic/endoplasmic reticulum Ca2+-ATPase
total internal reflection fluorescence