Cav1.1 controls frequency-dependent events regulating adult skeletal muscle plasticity

Excitable cells can evoke a large number of physiological events in response to plasma membrane depolarization. For instance, a single action potential may induce massive intracellular Ca²⁺ release in muscle cells that produces muscle fiber contraction and may trigger neurotransmitter release in neurons. Additionally, most excitable cells, notably muscle cells and neurons, respond not only to single depolarization events, but also to complex patterns of depolarizing stimuli of varying durations and frequencies. Consequently, a train of action potentials at a defined frequency is no longer an all or none message, but a code that excitable cells decipher to activate different process that determine their behavior. For example, spike-bursting patterns appear important for information processing and memory acquisition in neurons (Harris et al., 2001; Xu et al., 2012). In skeletal muscle fibers, different stimulation patterns activate diverse signaling pathways that ultimately define their phenotype (Schiaffino et al., 2007). Adult skeletal muscle contains different types of muscle fibers, which vary in both the speed of contraction and force generation and in their resistance to fatigue. Accordingly, muscle fibers can be classified generally as fatigue resistant slow-twitch fibers or as fast-twitch fibers that are either fatigue resistant or fatigable. Fatigue resistant slow-twitch fibers rely on their ATP supply mainly via oxidative metabolism that confers them their fatigue resistance, whereas fast-twitch fibers rely mainly on glycolytic metabolism as a source of ATP. These different fiber phenotypes can change in response to changes in activity demands, leading to a gradual switch from one fiber type to another. This process is known as muscle plasticity.

Nerve activity plays a major role in the specification and maintenance of skeletal muscle fibers phenotype, which depends on both myoblast lineage and motoneuron innervation (Gunning and Hardeman, 1991; DiMario and Stockdale, 1997; Kalhovde et al., 2005). Importantly, external electrostimulation with different firing patterns corresponding to different motoneuron subclasses allows the establishment of specific transcriptional programs that control fiber-type identity and growth and can provoke fiber phenotype transition by inducing the expression of specific myosin heavy chains and other contractile proteins as well as metabolic enzymes (Pette and Vrbová, 1985; Murgia et al., 2000; Serrano et al., 2001; Kubis et al., 2002). In particular, Troponin I is a protein of the contractile machinery which possesses two isoforms that are differentially expressed in distinct muscle fibers types (Banerjee-Basu and Buonanno, 1993) and whose regulation is under nerve activity control (Calvo et al., 1996; Rana et al., 2009). Expression of the slow Troponin I (TnIs) isoform is restricted to slow-twitch fibers while the fast isoform (TnIf) is restricted to fast-twitch fibers; both isoforms are downregulated in denervated muscles. External electrical stimulation of the denervated muscle with a slow pattern (10 Hz) induces a specific upregulation of the TnIs gene, while stimulation with a fast pattern (100 Hz) upregulates TnIf (Calvo et al., 1996). This feature makes TnI isoforms good markers of the transcriptional changes that occur in fibers during the muscle plasticity process.

In spite of the physiological importance of the ability of muscle cells to decipher different electrical stimulation patterns, the molecular mechanisms involved in muscle plasticity remain largely unknown. Changes in intracellular Ca²⁺ concentration have been proposed extensively to mediate cellular responses to electrical stimulation. It was proposed that a train of action potentials with a defined frequency induces a characteristic train of Ca²⁺ release events that, in turn, differentially activate particular Ca²⁺-dependent signaling pathways that determine the expression of genes responsible for the slow or fast muscle phenotype (Chin, 2004). These signaling pathways include the calcineurin–NFAT, Ca²⁺/calmodulin-dependent kinases II and IV (CaMKII and CaMKIV) and protein kinase C (PKC) pathways (Chin et al., 1998; Liu et al., 2001; Serrano et al., 2001; Wu et al., 2002).

Muscle cells present a complex pattern of Ca²⁺ transients under depolarization that are related to both excitation-contraction (E–C) coupling and excitation-transcription (E–T) signaling. In adult muscle cells, depolarization of the transverse-tubule (T-tubule) membrane induces a conformational change in Cav1.1 when transmitted to the ryanodine receptor (RyR1), causes channel opening and Ca²⁺ release from the sarcoplasmic reticulum (SR) to induce fiber contraction. In addition to this canonical signal, sustained depolarization stimulates inositol 1,4,5-trisphosphate receptor [Ins(1,4,5)P₃R]-mediated Ca²⁺ release that generates slow, long-lasting Ca²⁺ transients (Jaimovich et al., 2000). These slow transients, not related to muscle contraction, regulate several transcription-related events following membrane depolarization (Carrasco et al., 2003; Juretić et al., 2007). The signaling pathway activated by depolarization of muscle cells and triggered by Cav1.1 includes the sequential activation of G protein, phosphatidylinositide 3-kinase (PI3K) and phospholipase C (PLC) to produce inositol (1,4,5)-trisphosphate [Ins(1,4,5)P₃] that causes Ca²⁺ release via Ins(1,4,5)P₃R present in the SR membrane as well as in the nucleus (Cárdenas et al., 2005; Eltit et al., 2006). Electrical stimulation of cultured myotubes also promotes ATP release, likely through pannexin-1 (Panx1) channels. This ATP released is a key step in the onset of slow Ins(1,4,5)P₃-dependent Ca²⁺ signals, acting through P2X and P2Y receptors to modulate the process of E–T coupling (Buvinic et al., 2009).

In adult muscle, Cav1.1 also mediates the activation of slow Ins(1,4,5)P₃-dependent Ca²⁺ signals; their amplitude depends on both the duration and frequency of stimulation, with a maximum at 10–20 Hz. At this stimulation frequency, an increase of TnIs and a decrease in TnIf mRNA levels take place (Casas et al., 2010), a characteristic feature of fast-to-slow fiber phenotype transition.

In the present work, we describe that both Ins(1,4,5)P₃ production and ATP release are frequency-dependent events in adult skeletal muscle fibers, controlled by Cav1.1 acting as both a voltage sensor and a frequency decoder. Furthermore, Cav1.1 colocalizes with Panx1 channels in the T-tubule, suggesting the possible interaction between these two molecules to trigger ATP release. This is a major advance in the comprehension of the molecular events underlying activation of different signaling pathways by different patterns of stimulation that are the basis of skeletal muscle plasticity.

We investigated if molecular events upstream of Ins(1,4,5)P₃-dependent Ca²⁺ signals were sensitive to stimulation frequency. Adult flexor digitorum brevis (FDB) muscle fibers were stimulated with 270 pulses, 0.3 ms each, at different frequencies, measuring the generation of Ins(1,4,5)P₃ and ATP release at different times after stimulation. Fibers stimulated at 20 Hz displayed two peaks of Ins(1,4,5)P₃ production, at 15 s and 5 min after stimulation, whereas fibers stimulated at 90 Hz did not increase their Ins(1,4,5)P₃ levels (Fig. 1A). After 20 Hz stimulation Ins(1,4,5)P₃ increased from 0.021±0.001 pg/mg protein to 0.048±0.014 pg/mg protein in the first peak and to 0.054±0.013 pg/mg protein in the second peak. ATP release has a bell shape curve dependency on frequency of stimulation, showing a maximum at 20 Hz and almost no ATP release at frequencies higher than 60 Hz (Fig. 1B). Similarly to Ins(1,4,5)P₃ production, fibers stimulated at 20 Hz exhibited an increase in extracellular ATP levels with two peaks, reaching levels of almost 15-fold over non-stimulated conditions (typical trace in Fig. 1C, mean of maximum release in Fig. 1D). For most cells, the first peak occurs between 15 and 30 s after stimulation and the second between 3 and 5 min later (supplementary material Fig. S1). Average basal ATP levels were 11.1±3.5 pmol ATP/µg RNA, increasing to 166.1±67.5 and 165.2±75.5 pmol ATP/µg RNA for the first and second peak respectively, after stimulation. Interestingly, extracellular ATP levels did not increase when fibers were stimulated at 90 Hz (Fig. 1B–D). The increase in extracellular ATP levels observed at 20 Hz decreased significantly in cells treated with 5 µM carbenoxolone (CBX). At this concentration, carbenoxolone selectively blocks pannexin channels (Huang et al., 2007), 5–20 fold higher concentration is required for connexin hemichannel blockade (Bruzzone et al., 2005), suggesting that pannexin channels mediate ATP release (Fig. 1E,F).

As previously shown, adult FDB muscle fibers stimulated at 20 Hz exhibit an increase in TnIs and a decrease in TnIf mRNA levels 4 h after the stimulus, a signature of fast-to-slow muscle fiber phenotype transition (Casas et al., 2010). These transcriptional changes required functional Ins(1,4,5)P₃R because they did not occur in fibers pre-incubated during 30 min with 5 µM XestosponginB (XB), an Ins(1,4,5)P₃R inhibitor (Fig. 2A). Stimulation at 90 Hz produced the opposite effects: a decrease in TnIs and an increase in TnIf mRNA levels. These changes, however, were independent of Ins(1,4,5)P₃R because pre-incubation with XB had no effect (Fig. 2A). Changes observed in TnI mRNA levels were controlled at the transcriptional level because they were suppressed by pre-incubation of muscle fibers with 1.5 µM Actinomycin D, a transcription blocker (supplementary material Fig. S2A). We examined if ATP released after 20 Hz stimulation plays a role in the transcriptional changes in TnI genes described above. ATP release blockage with Panx1 inhibitor CBX suppressed the changes in TnIs and TnIf mRNA levels produced by stimulation at 20 Hz (Fig. 2A). Similarly, we found no increment in TnIs mRNA levels after stimulation at 20 Hz in fibers pre-incubated for 30 min with 2 U/ml apyrase, a nucleotidase that metabolizes extracellular ATP to AMP (Fig. 2B). Moreover, we found that addition of 30 µM ATP induced an increase in intracellular Ins(1,4,5)P₃ levels, which peaked 60 s after ATP addition (Fig. 2C), going from 0.015±0.002 pg/mg protein to 0.026±0.003 pg/mg protein. In agreement, addition of 30 µM ATP induced the same transcriptional changes in TnI genes produced by stimulation at 20 Hz (Fig. 2D). We tested other ATP concentrations, ranging from 10 to 100 µM (supplementary material Fig. S2B); we chose 30 µM because at this concentration the transcriptional changes observed were more prominent.

We measured extracellular ATP levels in dysgenic (mdg) myotubes that do not express the α₁s subunit of Cav1.1 (the subunit bearing the voltage sensor), after stimulation with 100 pulses at 20 Hz. In these cells, electrical stimulation did not evoke any Ca²⁺ transient due to the absence of the Cav1.1 α₁s subunit (supplementary material Fig. S3). The typical traces of ATP release in Fig. 3A show practically no response after electrical stimulation of dysgenic myotubes as compared to myotubes from wild-type mice. In wild-type cells ATP increased from 6.05±0.71 to 63.11±13.76 and 60.23±5.44 pmol/µg RNA for the first and second peak respectively, while in mdg myotubes ATP release went from 28±3.79 pmol/µg RNA to 55.82±8.61 and 62.83±7.07 pmol/µg RNA for the first and second peak (see Fig. 3B for fold changes).

To evaluate the corresponding role of Cav1.1 in adult muscle fibers, FDB fibers were stimulated at 20 Hz in the presence of nifedipine to inhibit Cav1.1 function. We found that 25 µM nifedipine produced a marked decrease in ATP release relative to the controls measured at different times after stimulation, as shown in Fig. 4A (quantification of peaks is shown in Fig. 4B). Moreover, we found that the increase in TnIs and decrease in TnIf mRNA levels observed in controls did not occur in fibers treated with nifedipine (Fig. 4C), suggesting a blockade of the whole signaling process.

We have previously shown that post-tetanic Ins(1,4,5)P₃-dependent Ca²⁺ signals in adult skeletal muscle fibers are independent from Ca²⁺ entry (Casas et al., 2010). Consequently, we hypothesized that the involvement of Cav1.1 in E–T coupling corresponds to a different and Ca²⁺ channel-independent function of this protein. To test this hypothesis, we stimulated at 20 Hz muscle fibers previously incubated for 20 min with 10 µM (-)S-BayK 8644 (BayK), another dihydropyridine that behaves as a Cav1.1 agonist, and measured ATP release and mRNA levels of both TnI isoforms. We found that this agonist inhibited ATP release (Fig. 4A,B) and prevented the changes in TnIs and TnIf mRNA levels produced by 20 Hz electrical stimulation (Fig. 4C). Interestingly, these two drugs had opposite effects on Ca²⁺ currents through Cav1.1 in adult fibers. Nifedipine produced an almost complete inhibition of the Ca²⁺ current whereas BayK enhanced the peak current amplitude and shifted the voltage-dependence towards more negative values (Fig. 5).

In order to understand the mechanism allowing Cav1.1 to control ATP release via Panx1 channels, we looked for a possible protein–protein interaction between these two molecules. We performed an in situ proximity ligation assay (PLA) with probes for Panx1 and Cav1.1 α1s subunit in non-stimulated fibers (Fig. 6) and 1 min after 20 Hz or 90 Hz stimulation (supplementary material Fig. S4A). We observed that immunofluorescence against Cav1.1 α1s subunit shows a clear double-striated pattern as expected for a protein located in the transverse tubules (Fig. 6A, bottom panels). Panx1 immunostaining shows a preponderant fluorescence near the surface and also a striated signal that runs at both sides of the z-line (labeled with anti-α-actinin antibody), suggesting a T-tubule localization for this protein as well (Fig. 6A, upper panels). This signal appears weaker towards the center of the fiber (Fig. 6C, right). In situ PLA probes for Panx1 and Cav1.1 α1s show a well-defined protein–protein interaction in non-stimulated (Fig. 6B,C) as well as in 20 Hz stimulated muscle fibers (supplementary material Fig. S4A), with no apparent differences. PLA probes in fibers after 90 Hz stimulation gave variable results, the mean being smaller than those after 20 Hz or basal conditions (supplementary material Fig. S4A). Indirect immunofluorescence against RyR1 was used to define fiber structure in this assay (Fig. 6B). PLA labeling appears to be concentrated near the surface (up to a depth of 4–5 µm), as can be seen in the z-projection for the whole fiber and in the correspondent fluoresce profiles of the center of the fiber (dashed line; Fig. 6B,C). As control we explored the PLA probes using antibodies against α-actinin, a structural protein that marks Z-disk and SERCA1, sarco-endoplasmic reticulum Ca²⁺ ATPase, that is expressed in non junctional sarcoplasmic reticulum. Even when the immunofluorescence for these two proteins gave a clear and strong labeling (α-actinin in Fig. 6A, upper right panel; supplementary material Fig. S5, SERCA1), there was a poor signal for PLA when using α-actinin/SERCA1 probes (Fig. 6D), as expected for two proteins that do not interact. After α-actinin/SERCA1 PLA was performed, indirect immunofluorescence against RyR1 was used to define fibers structure (Fig. 6D, lower panel).

We also tested the Panx1/Cav1.1 interaction by co-immunoprecipitation of these proteins from triad's enriched fractions of skeletal muscle. Characterization of this fraction is shown in Fig. 6E. We observed that Cav1.1 co-immunoprecipitated with Panx1 (Fig. 6F), but not with dysferlin, another protein present in triads extracts and used here as a negative control for co-immunoprecipitation essay (supplementary material Fig. S4B). These results suggest a direct protein–protein interaction between Cav1.1 and Panx1 channels, allowing us to hypothesize that control exerted by Cav1.1 over ATP release via Panx1 channels may be through a conformational change of this molecule somehow transmitted to Panx1.

In adult skeletal muscle, stimulation frequency regulates activation of specific transcriptional programs that help to define and adapt the muscle fiber phenotype. Ever since the role of nerve stimulation in muscle plasticity process was demonstrated (Buller et al., 1960; Pette and Vrbová, 1985; Schiaffino et al., 1999), a series of hypotheses have been proposed to explain how nerve stimulation modifies the transcription of specific genes. As in response to each action potential there is a rapid increase in intracellular Ca²⁺ via RyR1, stimulation at different frequencies could be expected to induce different patterns of intracellular Ca²⁺ signals to differentially activate Ca²⁺-dependent signaling pathways leading to activation of specific transcription programs proper of one fiber type or another (Hughes, 1998; Spangenburg and Booth, 2003). In this regard, it was described that Ca²⁺ signal amplitudes associated to each stimulation train was larger at higher than at lower frequencies (Chin, 2004). Fiber type differences in expression of Ca²⁺-related proteins such as parvalbumin, SERCA, Cav1.1 and RyR1 also contributed to sustain this hypothesis (Lamb, 1992; Delbono and Meissner, 1996; Chin et al., 2003). Nevertheless, the mechanism underlying differential activation of signaling pathways at the plasma membrane remains elusive. In the present work, we evidence a molecular sensing system in adult skeletal muscle fibers that decodes stimulation frequency at the plasma (T-tubule) membrane, linked to Cav1.1 activation, pannexin-mediated ATP release and Ins(1,4,5)P₃ production.

We have previously described that membrane depolarization of skeletal myotubes evokes Ins(1,4,5)P₃-dependent Ca²⁺ signals triggered by Cav1.1 and related to E–T coupling. In these cells, ATP released through Panx1 channels after electrical stimulation plays an important role in this process (Buvinic et al., 2009). In electrically stimulated adult muscle fibers, Ins(1,4,5)P₃-evoked Ca²⁺ signals mediate the frequency-dependent activation of slow-phenotype muscle fiber genes (TnIs) and repression of fast-phenotype fibers genes (TnIf) (Casas et al., 2010). In a recent publication (Blaauw et al., 2012), lack of calcium signals in response to Ins(1,4,5)P₃ uncaging in adult muscle fibers was reported; there are several reasons why such signals may be hard to record, including some dependence of Ins(1,4,5)P₃-induced Ca²⁺ release on membrane potential and/or intracellular Ca²⁺ concentration (Rojas and Jaimovich, 1990) or even mitochondria dumping of the calcium transient (Eisner et al., 2010). The fact that such transients can be evidenced after tetanic stimulation (Casas et al., 2010) and that Ins(1,4,5)P₃R inhibition blocks expression of particular genes, suggests that the precise location, magnitude and regulation of this signaling system should be a matter of further studies. Still, evidence from other groups also shows the presence of functional Ins(1,4,5)P₃R in adult muscle fibers (Volpe et al., 1985) and Ins(1,4,5)P₃ production after muscle stimulation (Vergara et al., 1985; Mayr and Thieleczek, 1991).

Now, we show that ATP release in adult muscle fibers after electrical stimulation is frequency dependent, occurring in fibers stimulated at 20 Hz but not at 90 Hz (Fig. 7). We also show that ATP release plays a key role in transcription of the slow-type TnI gene induced by 20 Hz stimulation, because fibers treated with apyrase to hydrolyze ATP, or CBX to inhibit ATP release by inhibition of Panx1 channels, did not exhibit the changes in mRNA levels produced at this stimulation frequency. Moreover, extracellular addition of 30 µM ATP induced the same transcriptional changes in TnI genes as electrical stimulation at 20 Hz. The fact that external ATP increased Ins(1,4,5)P₃ levels suggests that these transcriptional events occur in response to the same signaling pathways elicited by 20 Hz electrical stimulation or by external ATP addition.

In the last few years, ATP has been increasingly considered as an extracellular messenger for autocrine and paracrine signaling (Corriden and Insel, 2010). Several studies have recently demonstrated that ATP can be released by pannexin channels in a variety of cell types that include myotubes (D'Hondt et al., 2011). The widespread distribution of Panx1 has been confirmed in several tissues, with the highest levels in skeletal muscle (Baranova et al., 2004). In this work we show that Panx1 channels are located at the sarcolemma and at the T-tubules in skeletal muscle fibers.

We have shown that Cav1.1 forms part of the mechanism regulating ATP release after electrical stimulation of muscle cells. Thus, ATP release is practically absent after electrical stimulation of myotubes lacking the α1 subunit of this channel (impairing the assembly of the whole channel at the plasma membrane), while in adult muscle fibers the Cav1.1 blocker nifedipine abolished both ATP release and the transcriptional changes produced by 20 Hz stimulation. Interestingly, (-)S-BayK 8644 (BayK) also inhibited these two processes, albeit it had the opposite effect on Ca²⁺ current through the Cav1.1 channel than nifedipine. We have previously shown that Ins(1,4,5)P₃-dependent Ca²⁺ signals are independent from extracellular Ca²⁺ entry (Casas et al., 2010). The fact that nifedipine and BayK have the same inhibitory effects on ATP release and gene expression indicates that the Ca²⁺ current through Cav1.1 is not relevant to activate the E–T coupling process. More importantly, it suggests the existence of a Ca²⁺-current-independent function of Cav1.1, related to activation of E–T coupling that is blocked by these two drugs. In fact, the voltage gated Ca²⁺ channel (driving an L-type current) function of Cav1.1 is also irrelevant for the voltage sensor function in E–C coupling activation. Hence, the roles of Cav1.1 as voltage sensor in E–C coupling and E–T signaling, both activated by membrane depolarization, are independent of the function of Cav1.1 as Ca²⁺-current carrier. The molecular mechanisms behind these different functions are not still completely established, but several mutations in Cav1.1 differentially affect Ca²⁺ currents and E–C coupling, suggesting that different parts of the Cav1.1 molecule would be responsible for each function (Beam and Bannister, 2010). This could also be the case for the activation of the signaling cascade presented here. Another and not excluding possibility is that different states of the protein during activation, possibly reached with different kinetics, could be involved in E–T signaling. Several lines of evidence suggest complex transitions of Cav1.1 and multiple intermediate states of the protein during activation. In frog muscle fibers, there are different components of charge movement, with different voltage activation dependence and different kinetic parameters (Ríos and Pizarro, 1991). Even when charge movement is amenable to modeling by only one component, different fractions of this charge movement could be associated to E–C coupling and to Ca²⁺ current through the pore. For instance, nifedipine inhibits only a fraction of charge movement (nifedipine sensitive Q; Q_ns) (Lamb and Walsh, 1987). In this condition, fibers maintain E–C coupling, showing that Q_ns is not necessary for the onset of E–C coupling, while the Q_in (insensitive to nifedipine) is. In that work, the authors suggest that the difference in time course between the charge movement Q_ns and the appearance of the Ca²⁺ current through Cav1.1 is possibly due to the existence of an intermediary state (occurring after Q_ns), with an associated charge movement too slow to be recorded. Interestingly, in muscle fibers treated with BayK, charge movements appear unaffected (Lamb and Walsh, 1987). It is also possible then, that both drugs inhibit a charge movement with slow kinetics that might be responsible for the activation of the signaling pathway described here after stimulation at 20 Hz. Along this line, there is clear indication from the tail currents observed in BayK treated fibers that Cav1.1 deactivation is slowed-down in the presence of this drug (Johnson et al., 1997). Therefore, one possible explanation for our BayK results is that a slowed Cav1.1 configuration precludes activation of the ATP release and Ins(1,4,5)P₃ production machinery induced by 20 Hz stimulation. Following this view, in control adult fibers, the channel will be unable to activate ATP release at high frequencies (as 90 Hz) because the Cav1.1 activation kinetics for this process would not be fast enough. The bell-shaped curve of ATP release at different frequencies also supports this hypothesis, showing that beyond 20 Hz, ATP release decrease with increasing the frequency of stimulation to be lost at a frequency between 60 and 75 Hz. Further experiments with Cav1.1 mutants and drugs affecting the channel behavior to evaluate their effect on the frequency-decoding process described here would also shed light on the possible mechanism of E–T coupling activation by Cav1.1.

Importantly, the mechanisms of frequency decoding by Cav1.1 could be important for other excitable cells. In neurons, it has been shown that intracellular Ca²⁺ increases depending on Cav1 occurs at 10 Hz stimulation but not at 100 Hz and this Ca²⁺ increase has an important role in CREB activation (Wheeler et al., 2012).

We have presented here evidence indicating that both Cav1.1 and Panx1 are located in a region of the T-tubule membrane and a close association between these two molecules was evidenced by the in situ PLA method, which gives positive labeling for proteins located at a distance of less than 40 nm. This suggests the possibility of a direct protein–protein interaction. Consistent with a stronger Panx1 labeling near the fiber surface, interaction with Cav1.1 appears also to be concentrated in a region of the T-tubules at 4 to 5 µm from the sarcolemma. Considering a precise conformational change of Cav1.1 which could be responsible for ATP release, this conformational change could be directly linked to Panx1 activation via a protein–protein interaction between these two molecules, probably at the T-tubules. The fact that PLA gives the same labeling in control and in fibers stimulated at 20 Hz suggest the presence of a pre-assembled signaling protein complex that may be activated by conformational changes of some of its components after stimulation, as has been already be proposed in other models (Galés et al., 2006; Halls and Cooper, 2010). The difference found after 90 Hz stimulation could imply that at this frequency, additional conformational changes in the proteins involved make PLA less efficient. Some modification of Cav1.1 or Panx1 affecting antibody recognition cannot be excluded at this point. In any case, this should be matter of further studies; the probably dynamic nature of Panx1–Cav1.1 interaction makes the use of real-time studies (as FRET) advisable.

It has been described that pannexin channels have themselves some voltage dependency (Iglesias and Spray, 2012) and at this stage we cannot rule out direct effects of membrane potential over pannexin channels; it appears unlikely, though, that this could be the single mechanism responsible for ATP release. More likely, the possibility of a cooperative effect between the two molecules, Cav1.1 and Panx1, should be considered albeit a detailed knowledge of pannexin channel properties is lacking.

In this work, we have observed that there were, at least, two peaks of ATP release as well as two peaks in Ins(1,4,5)P₃ production. This result is coherent with Ins(1,4,5)P₃ production occurring in response to extracellular ATP release, as reinforced by the fact that external addition of ATP induced an increase in Ins(1,4,5)P₃ levels. The mechanism for the generation of the second ATP peak is unknown; we can only speculate on the possibility that Ca²⁺ release by Ins(1,4,5)P₃R following Ins(1,4,5)P₃ production might induce the opening of pannexin channels (Locovei et al., 2006) to induce the second peak, but further experiments are needed to elucidate this point.

In conclusion, this is the first report to describe molecular events displaying frequency dependency in skeletal muscle fibers. These findings make a critical advance to our current understanding of the mechanisms by which muscle cells decipher different patterns of stimulation that underlie muscle plasticity. A new role for Cav1.1 in this frequency decoding process appears to emerge.

5- to 7-week-old mice were used throughout this work. All protocols were approved by the Bioethics Committee, Faculty of Medicine, Universidad de Chile. Isolated fibers from the flexor digitorum brevis (FDB) muscle were obtained by enzymatic digestion with collagenase type II (90 min with 400 U/ml; Worthington Biochemicals Corp., Lakewood, NJ, USA), and mechanic dissociation with fire-polished Pasteur pipettes, as described previously (Casas et al., 2010). Isolated fibers were seeded in Matrigel-coated dishes and used 20 h after seeding.

Isolated muscle fibers seeded in a dish were stimulated with 270 squared pulses of 0.3 ms duration at different frequencies with a stimulation device that consists of a row of six platinum wires intercalated 0.5 cm apart with alternate polarity across a circular plastic holder that fits in the dish. The number of pulses was kept constant at 270 pulses at every frequency tested.

Wild-type (wt) and dysgenic (mdg) primary myoblasts were cultured and differentiated for 3–4 days in DMEM low glucose 4% horse serum as described previously (Casas et al., 2012).

Nifedipine, S(-)BayK 8644 and carbenoxolone were from Sigma. 25 µM N-benzyl-p-toluene sulphonamide (BTS, from Sigma-Aldrich Co., St Louis, MO, USA) was used in all experiments (except in calcium current measurements) as a fiber contraction inhibitor. All other reagents used were of analytical quality.

Fibers were quickly frozen in liquid nitrogen at different times after stimulation and were homogenized in 20 mM Tris-HCl pH 7.5; 2 mM EDTA; 150 mM NaCl; 0.5% Triton X-100. Ins(1,4,5)P₃ production determinations were performed with a Mouse Ins(1,4,5)P₃ ELISA Kit (Cusabio Biotech, Wuhan, P.R. China) following manufacturer's instructions. The data were reported as pg of Ins(1,4,5)P₃/mg of total protein. At difference of ATP measurements, normalization by total RNA was not possible due to lysis buffer used for Ins(1,4,5)P₃ determination which is not compatible with RNA extraction.

50 µl of extracellular media aliquots from electrically stimulated or control fibers were removed at different times post-stimulation. ATP concentrations were measured with the CellTiter-Glo® Luminescent Cell Viability Assay (Promega, Madison, WI, USA), as reported (Buvinic et al., 2009). Data were calculated as pmol extracellular ATP/µg total RNA and the ratios between experimental versus control points were reported. Normalization by total RNA instead of total protein was chosen because adult muscle fibers were seeded on a Matrigel-coated surface (containing a large amount of protein), which may affect the protein determination associated to fibers only.

Experiments were performed on single skeletal fibers isolated from FDB muscles of adult mice, as previously described (Jacquemond, 1997; Collet et al., 2004; Pouvreau et al., 2007). The major part of a single fiber was electrically insulated with silicone grease so that whole-cell voltage-clamp could be achieved on a short portion of the fiber extremity. An RK-400 patch-clamp amplifier (Bio-logic, Claix, France) was used in whole-cell configuration. Command voltage pulse generation and data acquisition were done using WinWCP software (freely provided by Dr Dempster, University of Strathclyde, Glasgow, Scotland) driving an A/D, D/A converter (National Instruments, Austin, TX). Voltage-clamp was performed with a microelectrode of 1-3 MΩ resistance, filled with a solution containing (in mM): 140 potassium glutamate, 5 Na₂-ATP, 5 sodium phosphocreatine, 5.5 MgCl₂, 5 D-glucose, 0.1 EGTA-AM, 5 HEPES, adjusted to pH 7.2 with KOH. The extracellular solution contained (in mM): 140 tetraethylammonium-methanesulphonate, 2.5 CaCl₂, 2 MgCl₂, 0.002 tetrodotoxin,1,4-aminopyridine, 10 HEPES, adjusted to pH 7.2. The tip of the microelectrode was inserted through the silicone, within the insulated part of the fiber. Analog compensation was systematically used to decrease the effective series resistance. The Ca²⁺ current through the DHPR was recorded in response to 0.5 s-long depolarizing steps of increasing amplitude (5 mV increments), applied every 30 s from a holding potential of −80 mV. The linear leak component of the current was removed by subtracting the adequately scaled value of the steady current measured during a 20 mV hyperpolarizing step applied before each test pulse. All experiments were performed at room temperature (20–22°C).

Total RNA was obtained from skeletal muscle fibers employing Trizol reagent (Invitrogen, Corp., Carlsbad, CA, USA) according to manufacturer's protocol. cDNA was prepared from 1 µg of RNA, using SuperScript II enzyme (Invitrogen), according to manufacturer's protocol. Real-time PCR was performed using Stratagene Mx3000P (Stratagene, La Jolla, CA, USA) using the Brilliant III Ultra-Fast QPCR and QRT-PCR Master Mix amplification kit (Agilent Technologies, Santa Clara, CA, USA).

The primers used were: TnIs: 5′-GAGGTTGTGGGCTTGCTGTATGA-3′ (sense), 5′-GGAGCGCATATTAGGGATGT-3′ (antisense); TnIf: 5′-AGGTGAAGGTGCAGAAGAGC-3′ (sense), 5′-TTGCCCCTCAGGTCAAATAG-3′ (antisense); β-actin: 5′-TCTACAATGAGCTGCGTGTG-3′ (sense), 5′-TACATGGCTGGGGTGTTGAA-3′ (antisense). All primers used presented optimal amplification efficiency (between 90% and 110%). PCR amplification of the housekeeping gene β-actin was performed as a control. Thermocycling conditions were as follow: 95°C for 3 min and 40 cycles of 95°C for 10 s, 60°C for 20 s. Expression values were normalized to β-actin and are reported in units of 2^−ΔΔCT± s.d. as described (Pfaffl, 2001). CT value was determined by MXPro software when fluorescence was 25% higher than background. PCR products were verified by melting-curve analysis.

Fibers were rinsed with ice-cold PBS and fixed for 20 min with paraformaldehyde (Electron Microscopy Science, Hatfield, PA, USA) 3% in PBS. Cells were rinsed with ice-cold PBS and incubated with glycine 100 mM in PBS for 10 min. Then, fibers were permeabilized with triton X-100 0.1% in PBS and blocked with 4% (w/v) BSA. Fibers were incubated with a rabbit polyclonal anti-Panx1 antibody (1∶100) (designed by the group of Dr Juan Carlos Saez and directed to the non-conserved region of the C terminus of human, mouse and rat Panx1 (amino acid residues CNLGMIKMD) (Cea et al., 2012), monoclonal anti-α-actinin antibody (1∶200) (Sigma-Aldrich) or monoclonal anti-Cav1.1 α1s antibody (1∶100) (Thermo Scientific, Waltham, MA, USA) over night at 4°C. Finally, the cells were washed three times with PBS for 5 min each, and incubated with anti-mouse and anti-rabbit Alexa Fluor 488/Alexa Fluor 546 as appropriate. The samples were mounted in Dako anti-fading reactive (Dako, Denmark) and store at 4°C until use.

Cav1.1/Panx-1 interaction was detected in situ using the Duolink II red starter kit (Olink Bioscience, Uppsala, Sweden) according to manufacturer instructions. Briefly, primary antibodies against α1s subunit of Cav1.1 and Panx1 were applied over night at 4°C in a humid chamber. Duolink plus and minus secondary antibodies against the primary antibodies were then incubated for 1 h at 37°C. These secondary antibodies were provided as conjugates to oligonucleotides that were tied together in a closed circle by Duolink Ligation Solution, if the antibodies were in close proximity (<40 nm). Finally, polymerase was added, to amplify any existing closed circles, and detection was achieved with complementary fluorescently labeled oligonucleotides. To define fiber structure, after PLA probes an immunofluorescence against RyR1 was performed. As negative control we performed a PLA between two proteins known for do not interact, as α-Actinin and SERCA1.

All images were acquired with a Carl Zeiss Axiovert 135 M Laser Scanning Microscope, with an Apo Plan 63×, NA 1.4 objective. Images deconvolution and processing were performed using ImageJ software (NIH).

Preparation of triad-enriched fractions from back and limbs muscles derived from 6- to 8-week-old BalbC mice were performed as previously standardized in our laboratory for frog and rabbit muscles, using differential centrifugation (Hidalgo et al., 1986; Jaimovich et al., 1986; Hidalgo et al., 2006).

Triad-enriched fractions (100 µg of protein) were solubilized for 1 h in 200 µl of lysis buffer (20 mM Tris-HCl pH 7.4, 0.1% Nonidet P-40, 5 mM EDTA pH 8, 10 mM EGTA pH 7.8, 140 mM NaCl, 10% glycerol and protease inhibitors). A 20 min, 15,000 g supernatant fraction was incubated 30 min with 10 µg A/G agarose as a pre-clearing strategy. The beads were spun down by centrifugation and washed 3 times with 200 µl of washing buffer (25 mM HEPES pH 7.5, 0.2% Nonidet P-40, 140 mM NaCl, 0.1% BSA, 10% glycerol and protease inhibitors). After the pre-clearing step, the whole cell extracts were incubated for 4 h with the correspondent antibody and then incubated 30 min with 50 µg A/G agarose beads. The beads pellet was washed three times with washing buffer. The whole amount of protein obtained after this step was resolved by SDS-PAGE in 7–10% gels, transferred to polyvinylidene difluoride filters and blotted with the corresponding antibody. To evaluate the protein content in the input of triads (prior to immunoprecipitation), 10–15 µg of protein form triads were loaded in each lane.

Results of n experiments are expressed as means ± s.e.m. The significance of differences was evaluated using Student's t-test for paired data and one-way ANOVA followed by Dunnett's post-test for multiple comparisons or Bonferroni's post-test for multiple paired comparisons. A P<0.05 was considered to be statistically significant.

We thank Dr Jordi Molgo for the gift of Xestospongin B, Dr José Miguel Eltit for help in the preparation of this manuscript and Dr Cecilia Hidalgo for valuable comments in reviewing this manuscript.