Store-Operated Calcium Entry in Skeletal Muscle: What Makes It Different?

Current knowledge on store-operated Ca2+ entry (SOCE) regarding its localization, kinetics, and regulation is mostly derived from studies performed in non-excitable cells. After a long time of relative disinterest in skeletal muscle SOCE, this mechanism is now recognized as an essential contributor to muscle physiology, as highlighted by the muscle pathologies that are associated with mutations in the SOCE molecules STIM1 and Orai1. This review mainly focuses on the peculiar aspects of skeletal muscle SOCE that differentiate it from its counterpart found in non-excitable cells. This includes questions about SOCE localization and the movement of respective proteins in the highly organized skeletal muscle fibers, as well as the diversity of expressed STIM isoforms and their differential expression between muscle fiber types. The emerging evidence of a phasic SOCE, which is activated during EC coupling, and its physiological implication is described as well. The specific issues related to the use of SOCE modulators in skeletal muscles are discussed. This review highlights the complexity of SOCE activation and its regulation in skeletal muscle, with an emphasis on the most recent findings and the aim to reach a current picture of this mesmerizing phenomenon.


Overview of the Muscle Structure and Ca 2+ Handling
Skeletal muscle fibers are very large multinucleated cells that are formed by the fusion of precursor cells, namely, myoblasts. They have a highly organized internal architecture, with most of the cell volume being occupied by the contractile elements, mainly actin and myosin, grouped as myofibrils. Sarcomeres are the functional units of the contractile apparatus, delineated by the Z-lines, which are oriented perpendicular to the long axis of the fibers. Within the sarcomeres, the strict arrangement of actin and myosin gives rise to the typical striated pattern of skeletal muscles, with alternating A and I bands, where the Z-line is found in the middle of the I-band (Figure 1). Each myofibril is surrounded by the sarcoplasmic reticulum (SR), which is a specialized region of the endoplasmic reticulum (ER) forming an interconnected network with a high Ca 2+ buffering capacity (rev in [1]). Twice per sarcomere, the plasma membrane has deep invaginations called the t-tubules. Remarkably, it is estimated that the t-tubules encompass around 80% of the plasma membrane [2]. Facing each side of the t-tubules are enlargements of the SR forming the junctional SR (jSR; also called the terminal cisternae, TC) that, together with the t-tubules, compose the triad. The jSR is enriched in the acidic Ca 2+ buffer calsequestrin (CASQ) and is in continuity with the longitudinal SR (lSR), which harbors a high density of sarco-endoplasmic reticulum Ca 2+ ATPase (SERCA) pumps (rev in [3]). Fast and slow muscle fibers, so named according to their kinetics of contraction, compose a muscle. These fiber types differ in many aspects, among them, the size and buffering capacity of the SR, which are larger in fast fibers, together with a greater SERCA activity (rev in [4]). Excitation-contraction (EC) coupling depicts the succession of events leading to cytosolic Ca 2+ elevation and, eventually, skeletal muscle contraction. It is initiated by the release of acetylcholine from the motor neuron nerve, which binds to its cognate ionotropic receptors at the neuromuscular junction (NMJ). The resulting cation influx depolarizes the plasma membrane (PM), activates Na v 1.4 voltage-gated sodium channels, and triggers action potentials (APs) that propagate along the PM and into the t-tubules of skeletal muscle fibers. Within the t-tubules, the AP is sensed by L-type Ca v 1.1 channels, which are also called dihydropyridine receptors (DHPRs). Through physical interaction, the change in DHPR conformation is transmitted to the RyR1 Ca 2+ channel, which is located in the membrane of the jSR. The opening of RyR1 leads to an explosive and simultaneous release of Ca 2+ from the SR into the cytoplasm across sarcomeres, initiating muscle contraction. After stimulation, Ca 2+ clearance from the cytosol is accomplished by the SERCA pumps, which mediate the re-uptake of Ca 2+ into the SR store. Due to the very brief muscle AP and the intrinsic biophysical properties of Ca v 1.1, there is virtually no Ca 2+ flux through these channels during membrane depolarization (rev in [5]). Hence, EC coupling per se is a process that does not require extracellular Ca 2+ entry, in contrast to cardiac muscle, where Ca 2+ influx is mandatory for the RyR (RyR2 in that case) to open and permit muscle contraction. The non-requirement of Ca 2+ entry for EC coupling in part explains the delayed interest for SOCE in skeletal muscle.
SOCE is a ubiquitous mechanism that allows Ca 2+ to enter the cells in response to a decrease in the ER/SR Ca 2+ content. This peculiar mechanism was originally described in 1986 in the pioneering work of J. Putney on salivary glands, where this mechanism was called capacitive Ca 2+ entry at that time [6]. After 20 years of intense research to elucidate the molecular components of this Ca 2+ entry, a handful of studies uncovered the proteins supporting SOCE, namely, the stromal interaction molecule 1 (STIM1; [7][8][9]) and Orai1 [10][11][12]. STIM1 is a single-pass transmembrane protein that is localized on the ER that binds Ca 2+ via its luminal EF-hand motifs. Upon store depletion, the unbinding of Ca 2+ leads to STIM1 oligomerization and translocation toward the plasma membrane (PM). The unfolding of the protein exposes a polybasic region at the C-terminal end of the molecule, which promotes its recruitment/stabilization at the PM via the binding of phosphoinositides [13][14][15]. Other exposed key residues of STIM1 within the CAD (CRAC activation domain), allow the gating of Orai1 and eventually Ca 2+ entry. The whole process from ER Ca 2+ depletion to the activation of Ca 2+ entry takes tens of seconds, mainly as a result of STIM1 translocation to the PM (rev in [16]). The translocation toward the PM is associated with a profound remodeling of the ER, forming thin and elongated ER cisternae (cortical ER (cER)) visible in electron micrographs [17,18]. STIM2, the other member of the STIM family, has a higher affinity for phosphatidylinositol 4,5-bisphosphate (PiP 2 ) due to a modified polybasic domain [19,20]. In addition, the Ca 2+ affinity of STIM2 (K d for Ca 2+ binding around 400 µM) is lower than that of STIM1 (K d : 200 µM), making STIM2 a regulator of the basal cytosolic Ca 2+ concentration and the ER Ca 2+ level (rev in [21]). The PM Ca 2+ -selective Orai channel comprises three members, namely, Orai1-3. The current flowing through Orai, called Ca 2+ release activated Ca 2+ current (I CRAC ), has been known since the nineties (well before the molecular identification of Orai channels), and is nowadays very well characterized. Its peculiar electrophysiological signature comprises a tiny unitary conductance in the fS range, which precludes single-channel recording, a strong inward rectification with a very positive reversal potential highlighting its Ca 2+ selectivity, and a complex regulation both by intra-and extracellular Ca 2+ concentration [22]. In particular, I CRAC undergoes fast and slow Ca 2+ -dependent inactivation (CDI) that are proposed to prevent excessive Ca 2+ entry and, thus, a potential overload that would be detrimental for cell function [23,24].
In contrast to the great interest in SOCE in non-excitable cells, studies on skeletal muscle SOCE failed to appear for several years following the initial description by J. Putney. Several reasons could account for this relative disinterest: as mentioned, EC coupling per se does not require an external Ca 2+ influx and skeletal muscle contraction can occur, at least for a while, in a medium devoid of Ca 2+ [25]. Furthermore, once Ca 2+ is released from the SR into the cytosol, at least during moderate muscle activity, almost all of it is pumped back into the SR with basically no loss across the PM (Ca 2+ flux across the PM is orders of magnitude smaller than Ca 2+ flux across the SR membrane; [26]). Hence, it was assumed that there was no need for external Ca 2+ entry to replenish the Ca 2+ stores in skeletal muscle. More technically and linked to muscle size, the classical protocol that is used to activate SOCE, i.e., blocking the SERCA pumps in the absence of external Ca 2+ to passively deplete the stores, does not result in a massive store depletion in skeletal muscle. Indeed, the SR Ca 2+ store is a large compartment with a very high Ca 2+ -buffering capacity. In addition, the SR membrane in skeletal muscle is less leaky compared to non-excitable cells, impeding SERCA blockage-induced store depletion [27,28], and thus SOCE. Lastly, electrophysiological recordings of I CRAC in skeletal muscle are scarce, with only two papers reporting such a current in myotubes [29,30]. Actually, the group of B. Allard claimed that in skeletal muscle, I CRAC is below the limit of detection, even using the silicon clamp approach, which allows for accurate voltage clamping of the membrane [31,32]. Hence, this strongly limits the knowledge we have on the biophysical properties of the skeletal muscle SOCE current. For all these reasons, the interest in SOCE in the muscle system really emerged only after the first work clearly revealed this pathway in 2001 [33]. In this study performed on isolated fibers from extensor digitorum longus (EDL), the SR was depleted by successive applications of high K + solution in a Ca 2+ -free medium, followed by a treatment to block the SERCA pumps. The subsequent Ca 2+ re-addition replenished the stores, which was indicative of SOCE having taken place. Mn 2+ quenching experiments further confirmed the activation of SOCE. With the identification a few years later of the proteins supporting SOCE and the finding that mutations in STIM1 or Orai1 are associated with muscle pathologies, the field gained much attention. Indeed, loss-of-function mutations of STIM1 and Orai1 lead, besides a severe immunodeficiency, to congenital myopathy, which is characterized by hypotonia and reduced muscle endurance. Gain-of-function mutations in SOCE molecules also result in progressive muscle weakness, known as tubular aggregate myopathy (TAM) syndrome (rev in [34]). at the triad decreases; the mechanism ensuring this localization remains uncertain (rev in [37]). In a recent elegant paper, Sébastien et al. [44] revisited triad protein mobility and showed, using a photoactivatable form of the protein, that a small fraction of triadin (also called Trisk 95), which is a single-pass transmembrane protein of the triad, is mobile and can move in and out of the triad. Once at the triad, and in line with the report of Cusimano [43], the mobility of triadin is greatly reduced. The authors also identified the transmembrane (TM) domain of Trisk 95 as an essential motif that retains the protein at the jSR [44]. It would be very informative to know whether STIM1 diffuses in the SR membrane just as it does in non-muscle cells, whether its diffusion is limited like for Trisk 95, or whether it is permanently retained at the triad. Two main arguments are in favor of STIM1 being retained at the triad (at least for a large part of the pool): First, the very fast kinetics of SOCE activation in muscle cells is incompatible with the translocation of STIM1 upon activation (see Section 3 about phasic SOCE). The second aspect is related to the site of store depletion during EC coupling. Indeed, SR Ca 2+ depletion occurs mainly at the jSR [45,46] and thus for STIM1 to sense this local Ca 2+ decrease, the protein must be located there as well. One should also consider that STIM1 can be activated following other types of stimulation, for instance, after IP 3 receptor (IP 3 R)-induced Ca 2+ release. However, while this pathway seems to be relevant in developing muscle, there are controversies about its existence in adult muscle. Blaauw et al. [47] claimed that no IP 3 R-induced Ca 2+ release could be recorded in adult skeletal muscle after using different approaches, such as direct IP 3 injection, IP 3 uncaging, or stimulation by agonist-induced IP 3 formation. In contrast, IP 3 R1 was reported to be localized at the lSR, colocalizing with the Z-line in differentiating myotubes [43], and the group of Jaimovich described a small amount of IP 3 R-induced Ca 2+ release that takes place preferentially around the nuclei and induces transcriptional activity [48,49]. More recently the same group showed that mitochondria Ca 2+ uptake, resulting from Ca 2+ release, is partially sensitive to IP 3 R blockers [50]. Hence, it remains to be defined whether STIM1 could potentially sense SR Ca 2+ decrease after IP 3 R-induced Ca 2+ release and thus induce SOCE in muscle fibers. Related to the IP 3 R, the ER compartment found around the nuclei and in continuity with the SR localized around the myofibrils and implicated in EC coupling [1] should be considered. Indeed, a high density of IP 3 R was found around nuclei, in particular, those at the NMJ [51,52], and one can speculate that a subset of STIM molecules is also present under the NMJ ( Figure 2D) and could be implicated in the specific Ca 2+ signals that are required for the proper formation of the NMJ [52]. Thus, several open questions related to the localization of STIM1 molecules at different places in adult muscle fibers and the putative mobility of STIM1 remain to be addressed. (B) SOCE at the CEU is well described as a newly formed structure that is induced by strenuous exercise. STIM and Orai are expressed at the CEUs but the respective SOCE Ca 2+ -flux has not yet been demonstrated to date. (C) STIM at the lSR moves to the triad upon activation or activates Orai at the PM. It was also reported to modulate the activity of the SERCA pump. (D) Localization at the NMJ remains speculative for the time being.

Regulation of SOCE in Skeletal Muscle
The triad was the first ER/SR-PM contact site described [53]. Nowadays, many membrane contact sites (MCS), which are defined as a tight apposition between the ER membranes and other organelle's membrane or PM, are identified within cells, with the ER-PM region being the best characterized [54]. The skeletal muscle triad is intensively studied in terms of the formation, molecular composition, and regulation of Ca 2+ release. The DHPR and the RyR1, which are abundantly expressed at the triad, are the two key Ca 2+ channels supporting EC coupling, but several other proteins are located there, providing a structural function or acting as regulators of Ca 2+ handling. We will here only briefly mention the most relevant of these and highlight their putative function as direct or indirect SOCE modulators ( Figure 1). Junctophilin 1 and 2 (JPH1 and JPH2) are anchored in the SR membrane and possess a large N-terminal hydrophobic region that binds phospholipids in the PM, making JPHs key proteins to couple the SR to the t-tubule membrane. Indeed, knockdown of both JPHs leads to a malformation of the triad that is associated, among other defects, with reduced SOCE [55,56]. Interestingly, neither RyR1 nor DHPR fulfills a structural function, as shown by animal models lacking one or the other protein, that maintain a proper triad formation [57,58]. With RyR1 being the most important Ca 2+ release channel of skeletal muscle, it is indirectly coupled to SOCE activation and, in addition, SOCE (induced by SERCA inhibition) is reduced in myotubes from mice lacking RyR1 [59]. Triadin is a family of SR-associated proteins that originate from the alternative splicing of the TRDN gene [60]. In skeletal muscle, three members of the family are expressed, namely, Trisk 95, Trisk 51, and Trisk 32, with the first two being expressed the highest. Trisk 95 is associated with RyR1 and CASQ1, and by promoting membrane deformation, it favors the interaction between RyR1 and DHPR and thus EC coupling (rev in [61]). Junctin and junctate are two ER/SR proteins that are generated by a complex splicing of the same gene, which also encodes humbug and aspartyl β-hydroxylase (rev in [62]). Junctin expression is restricted to cardiac and skeletal muscle, where it associates with RyR1, triadin, and CASQ1. It was shown to modulate the open probability of RyR1 but also to play a role in the SR Ca 2+ storage size [63]. In contrast, junctate is more broadly expressed and is a high-capacity Ca 2+ binding protein [64]. Overexpressing junctate specifically in skeletal muscle resulted in an increased SR Ca 2+ content and an enhanced caffeine-induced Ca 2+ release, but also an increased Ca 2+ entry after store depletion [65]. A few years later, it was reported that junctate indeed facilitates the recruitment of STIM1 to the ER-PM junction. This mechanism was thus postulated as the third one favoring the localization of STIM1 at the PM upon store depletion, in addition to phospholipids and Orai binding [66]. These experiments were conducted in T cells, but one can reasonably hypothesize that such an association between STIM1 and junctate also takes place in skeletal muscle to promote SOCE. CASQ1 is highly expressed at the jSR of skeletal muscle and was reported to have an inhibitory effect on SOCE [67], which was confirmed in CASQ1 knockdown experiments [68]. Later, CASQ1 was indeed shown to interact with STIM1, preventing its binding to Orai1 and thus limiting SOCE [69,70]. Mitsugumin 29 (mg29) is a synaptophysin-like protein that is localized at the t-tubules [71]. Mg29 −/− animals presented structural defects, such as swollen t-tubules, vacuolization of the SR, and a misalignment of the triad, even if the triad is present [59,72]. These structural alterations were associated with a reduced SR Ca 2+ content, together with a reduced/slower SOCE [59,73]. Functionally, these animals were more prone to fatigue compared with wild-type controls [59,74]. Interestingly, the decreased expression of mg29 with age was postulated to account for the reduced SOCE observed in old animals, despite normal levels of STIM and Orai [73]. However, the role of mg29 as a potential regulator of muscle SOCE was not confirmed by another study that was performed on adult tissue [75], and thus the involvement of mg29 as a SOCE modulator remains an open question. Furthermore, the observed reduction of SOCE in aged mice was not confirmed by others [76].
By analogy with proteins at the triad, the MCS associated with STIM and Orai contains proteins that are involved in the establishment of the contact and several SOCE regulators. Extended synaptotagmin (E-syt1) was shown to be recruited by SOCE and then served to stabilize the MCS [77] (rev in [54]). In contrast to the triad, where RyR1 and DHPR do not have a structural function, the MCS containing SOCE are formed by the recruitment of STIM1 to the PM and its binding to Orai1. Several SOCE-interacting proteins were reported, such as SOCE-associated regulatory factor (SARAF), STIM-activating enhancer (STIMATE), partner of STIM (POST), or CRAC regulator 2A (CRACR2A), which serve as fine regulators of Ca 2+ entry (rev in [78]). To the best of our knowledge, however, none of these regulatory proteins were reported to modulate SOCE in skeletal muscles. Nevertheless, the lipid composition of the PM was proposed to negatively regulate Orai1, with two residues in the N-terminus of the channel being implicated directly or indirectly in this regulation [79]. Interestingly, the t-tubule membrane is about four times more enriched in cholesterol than the sarcolemma [80,81], potentially influencing Orai1 function. Furthermore, PiP 2 content modulates SOCE. Indeed, among the regulatory SOCE mechanisms, one is the well-known prevention of Ca 2+ overload via the slow Ca 2+ -dependent inactivation (CDI) of SOCE/I CRAC , which is in part linked to SARAF binding. It was reported that the accessibility of SARAF to Orai1 is associated with the movement of the SOCE complex from PiP 2 -poor to PiP 2 -rich regions upon activation [82]. Whether such regulation of SOCE takes place in skeletal muscle is not known and will be difficult to address due to the quasi impossibility of I CRAC recordings in these cells [83]. Hence, compared to what is known about SOCE/I CRAC regulation in non-excitable cells, little is known about skeletal muscle SOCE regulation. In particular, how Ca 2+ concentration within the narrow space between the t-tubule and the jSR influences both the RyR1 channels and the SOCE process (including CDI) is so far not addressed but would be fundamental to understand. Some years ago, a different mechanism of muscle SOCE activation was proposed by the group of F. Protasi, arguing that the high density of RyR1 molecules at the triad would hinder the diffusion of STIM1 to the same place [84]. Along this line, they reported that STIM1 is localized at the lSR, more specifically at the I-band [41]. Orai1, on the other hand, was found at the triad, as previously reported [38]. Unexpectedly, electron microscopy revealed that after high-intensity (HIT) exercise, the t-tubules were massively reorganized, and to a lesser extent, the SR too. Essentially, the t-tubules elongated and moved parallel to the lSR, resulting in the formation of numerous contacts between both membranes and the appearance of characteristic stacks. Hence, this finding would imply that in skeletal muscle, the plasma membrane (in that case, the t-tubule) moves toward STIM1 and not vice versa, as is observed in other cell types. These peculiar t-tubule structures, called Ca 2+ entry units (CEUs), were also occasionally seen in muscles from non-exercised mice. Immunostaining confirmed that STIM1 and Orai1 co-localization was minimal at rest but significantly enhanced upon HIT exercise [41]. In a follow-up paper, the same group reported that these were reversible structures, even if it took several hours (>6 h) to revert [85]. Functionally the CEUs were associated with better resistance to fatigue. Indeed, when EDL muscles (ex vivo experiments) from animals that previously ran intensively for one hour (and thus had newly formed CEUs) were undergoing high-frequency stimulations, the force decline was less compared to muscles from animals that did not run beforehand. This "beneficial" effect was gone in the presence of the SOCE blockers BTP2 or 2-APB [41], and also in muscles from Orai1 −/− animals, highlighting a role of Orai1-dependent SOCE in this effect [85]. In addition, the formation of the CEUs upon intense exercise was associated with an increased basal SOCE, an enhanced resting cytosolic Ca 2+ concentration, and a decrease in total SR Ca 2+ [85]. The authors hypothesized that the increased basal Ca 2+ entry might be due to STIM2-induced SOCE rather than STIM1, but this remains to be determined. These observations raised an additional question about the beneficial effect of the formation of the CEUs, as increased basal SOCE is known to be detrimental for cells and, in particular, for skeletal muscle, as highlighted by the pathologies linked to gain-of-function mutations of STIM1 or Orai1 (rev in [86]). So far, nothing is known about the mechanism leading to such t-tubule rearrangement, nor whether this is a general way SOCE presents upon physiological muscle use. What was, however, recently shown was an enhanced CEUs in mice lacking CASQ1, which is the main SR Ca 2+ buffer [87]. In these animals, the Ca 2+ storage capacity was greatly reduced and correlated with higher SOCE capacity. In addition, the expression levels of STIM1, Orai1, and SERCA were elevated, further promoting an increased SOCE that would permit sustained Ca 2+ release upon repetitive stimulations in a context of reduced SR Ca 2+ stores [87]. Overall, the massive rearrangement of the t-tubules and the formation of CEUs is an interesting new way of considering SOCE in skeletal muscle; a potential regulatory mechanism that is entirely different from what takes place in non-excitable cells. It should, however, be stressed that there is so far no direct evidence that Ca 2+ entry takes place at the CEUs. Importantly, the lSR has a greatly reduced RyR1 density [88] and the RyR1 density at the CEUs was not increased upon exercise [41,85]. Given the pivotal role of RyR1 in the activation of SOCE [30,39,59,89,90], this raises the question of how SOCE could be activated at these sites. While it is appealing to speculate that the observed increase in SOCE is causally linked to and not only correlated with the newly formed CEUs, it might also be that the observed increase in SOCE upon exercise [85] is caused by an altered SR Ca 2+ -buffering capacity and/or structural/functional changes within the triads. Thus, strikingly, knock out of key SR and triad proteins, namely, triadin/junctin [91], CASQ [87], and microtubuleassociated protein 6 (MAP6 [92]), induce CEU structures that fully resemble those found after strenuous exercise [41,84]. Interestingly, alterations in t-tubular structure and SOCE were also reported in human muscle after heavy-load resistance exercise [93]. Thus, Cully et al. [93] described the formation of vacuoles within the longitudinal t-system upon highforce eccentric exercise. These vacuoles, which formed within several hours after exercise and seem to be reversible after a few days, could store large amounts of Ca 2+ but were devoid of SOCE. The authors argued that the absence of SOCE in the longitudinal t-system compartment reflected the fact that the SR does not form junctions and that RyRs are absent in these regions. This would prevent the activation of SOCE, even if a respective SOCE protein machinery is present at these sites. It will require further studies to better define the physiological mechanisms that lead to the observed changes in the t-system architecture upon different forms of exercise, as well as the physiological consequences that are related to such alterations in muscle structure.
Based on the different studies on skeletal muscle SOCE, we can conclude that it is more than likely that SOCE is taking place at various locations within an adult muscle fiber. Obviously, it occurs at the triad (see also the chapter below) and potentially at the CEUs, possibly depending on the fiber type and the intensity of exercise. Other putative locations, which are, however, so far not supported by literature, are the PM in connection with the SR, but also with the ER around the NMJ, with the latter likely being involved in gene regulations ( Figure 2).

Using "Skinned" Fibers to Measure SOCE
Probably the largest fraction of SOCE is conducted across the transverse tubules of the tubular (t-) system membrane specifically and not the PM [32,83,94] nor the longitudinal tubules of the t-system [93]. The fine-and complex-branched structure of the t-system and its location entirely within the fiber's body made it inaccessible for conventional electrophysiological approaches [83,95,96]. It needed the development of fluorescent measurements from within the t-system of skinned fibers to obtain most of our current knowledge on the Ca 2+ -handling properties of the t-system, including SOCE [97]. These fluorescent techniques proved superior to the classical approaches because they allowed for a better temporal and spatial resolution and greater sensitivity as, e.g., compared to SOCE measurements using Mn 2+ -quenching experiments. Importantly, they also allowed for studying SOCE simultaneously with SR Ca 2+ release in physiological salt solutions, normal activation patterns, and functional SR proteins, which cannot be achieved in standard electrophysiological measurements [96]. Therefore, we briefly sketch the development here.
Mechanically skinned fibers were originally described by Natori in 1954 [98]. Using this technique, access to the intracellular space of the fiber is gained by physically rolling back the sarcolemma with fine forceps [99,100]. The procedure leaves the integrity of the SR and t-system intact. In addition, the t-system entry mouths, which are connected to the sarcolemma beforehand, seal off to form a closed compartment after skinning. Fluorescent dyes diffuse into the open t-system when applied in an extracellular buffer prior to skinning [101] and are trapped therein upon skinning [102] because of the induced sealing-off of the t-tubule mouths. The necessary protein machinery that is used to conduct EC coupling resides within the t-system and is not affected by the removal of the sarcolemma during the skinning procedure. Hence, EC coupling is preserved in such a preparation, with normal [Ca 2+ ] cyto transients [45,96,103,104] and force responses [103,105], as seen in intact fibers. Recently, low-affinity Ca 2+ -sensitive dyes (Rhod-5N, Fluo-5N, and Mag-indo-1) trapped in the sealed t-system were used to investigate t-system Ca 2+ handling [94,95,102] and were used to measure SOCE in a quantitative manner [93,97].
Soon after the first description of SOCE in skeletal muscle by Kurebayashi and Ogawa [33], it became apparent that SOCE in muscle was markedly different from that observed in non-muscle cells, in particular regarding its fast activation kinetics. Using a fluorescent dye trapped in the t-system of a skinned fiber preparation allowed Launikonis et al. to monitor [Ca 2+ ] t-sys and derive a measure of SOCE during SR Ca 2+ release [95]. SR Ca 2+ release was induced by lowering the free cytosolic magnesium concentration ([Mg 2+ ] cyto ), which removed the Mg 2+ -dependent inhibition of the RyR1 and caused a cell-wide Ca 2+ -release [97,[106][107][108]. Note that if not otherwise stated, [Ca 2+ ] or [Mg 2+ ] will always refer to the respective free ion concentrations. Under these conditions, SOCE was activated within one second upon exposure to low [Mg 2+ ] cyto . This was the first demonstration of fundamental differences in skeletal muscle SOCE because it was at odds with the classical view of SOCE activation as observed in, e.g., immune cells: STIM oligomerization, recruitment of PM and ER contact sites, puncta formation, and activation of Orai channels. SOCE, as observed in these early experiments, activated way too fast to allow for such a complicated mechanism of activation. A refinement of the technique led to further insights. Thus, Edwards et al. demonstrated that SOCE in mouse EDL muscle even occurs on a millisecond timescale [109]. To isolate SOCE and avoid the activation of voltage-dependent currents, SR Ca 2+ release was induced via the direct activation of the RyR1, again using conditions of low [Mg 2+ ] cyto . Besides a cell-wide Ca 2+ release, low [Mg 2+ ] cyto commonly induces the emergence of propagating Ca 2+ waves across the preparation, a condition under which SR Ca 2+ buffers are constantly depleted and refilled, and SOCE can be studied. A low-affinity Ca 2+ -sensitive dye loaded into the t-system of a skinned fiber allowed for the continuous monitoring of SOCE, which reported the respective activation of SOCE only 27 ms after SR Ca 2+ release had occurred. This fast kinetics indicated a steep relationship between changes in [Ca 2+ ] SR and activation of SOCE and suggested a physical pre-coupling between STIM1 and Orai1 within the triad region to account for the observed fast kinetics [26,95]. This was in agreement with the localization of STIM1 and Orai1 within the triad regions [29,38] and that disruption of the triad structure by knockout of junctophilin [55,110] or exposure of the fiber to sustained high [Ca 2+ ] cyto [94,95,102] reduced/abolished SOCE. Moreover, these experiments showed that SOCE activated as the RyR1 began to release Ca 2+ and that this occurred well before the Ca 2+ inside the SR was significantly depleted. This was an important step forward compared to experiments that relied on completely depleted SR Ca 2+ stores, which also showed a RyR dependence of SOCE (the SOCE amplitude was greatly reduced in the myotubes of RyR1/RyR3-deficient mice [59] and in dyspedic (lacking RyR1) myotubes [89]. Furthermore, the skeletal muscle I CRAC current, as assessed using a whole-cell patch clamp, was reduced three-fold in RyR1-null myotubes and inhibited by 100 µM ryanodine [30]), but could not resolve the full dynamics linking the activation of SOCE to the opening of the RyR1.
While these early experiments defined key hallmarks of SOCE in muscle, by tracking the t-system Ca 2+ simultaneously with SR Ca 2+ release triggered via direct stimulation of the RyR1 [95,109], it was of interest to determine whether SOCE could also be activated using voltage stimulation and thereby operate during single muscle twitches. A first hint that this was possible was given in 2009 when Launikonis et al. [96] reported on an APdependent Ca 2+ -influx that was independent of L-type Ca 2+ channels and that they named AP-activated Ca 2+ current (APACC). While the authors argued for an SOCE-independent mechanism at that time, in retrospect, their results surprisingly reflect several key features of phasic SOCE observed at low cytosolic buffering conditions (see below). Another finding was that the rate of fura-2 quenching by Mn 2+ changed upon electrical burst stimulation in interosseous muscle fibers [32]. The observed increase in the Mn 2+ quenching rate was attributed to an electrically silent pathway suggested to be again independent of L-type Ca 2+ channel function [32].

How to Measure SOCE during EC Coupling?
A combination and advancement of different experimental approaches involving skinned muscle fibers [97,105], as described above, then led to the recent demonstration of SOCE during physiological activation patterns when SOCE was observed during single muscle twitches in skinned rat EDL fibers [39]. It was named phasic SOCE (pSOCE) to reflect the transient nature of the observed Ca 2+ flux and to discriminate it from slower and longer-lasting forms of SOCE, referred to as chronic SOCE (cSOCE; [39]). The recording of pSOCE during AP-induced SR Ca 2+ release was made possible by the simultaneous tracking of [Ca 2+ ] cyto and [Ca 2+ ] t-sys at high temporal resolution and fidelity using highspeed confocal microscopy and concomitant electric field stimulation in skinned muscle fibers. An overview of two important, fluorescence-based experimental paradigms for measuring SOCE using skinned skeletal muscle fibers is given in Figure 3.
One key point of the technique involves the use of high levels of cytosolic Ca 2+ buffering with 10 mM EGTA [39,111]. This allowed us to isolate SOCE by reducing the [Ca 2+ ] cyto reached during an AP (see below). Moreover, it enabled calculating the amount of Ca 2+ released during every muscle twitch and thereby allowed for determining the dependence of pSOCE on SR Ca 2+ release [39]. Another key point was that the technique employed fluorescence averaging across the entire muscle fiber, which is possible because electrical field stimulation (EFS) triggers synchronous APs in every sarcomere. This significantly improved the signal-to-noise ratio, in particular regarding the weak Rhod-5N fluorescence emanating from within the t-system, which constitutes only a small fraction of the total fiber volume. The necessary fast imaging became possible only with the advancement of fast microscopy techniques, e.g., resonant scanners or spinning disc devices. In contrast to measurements of cSOCE, where SR Ca 2+ levels are strongly decreased, pSOCE was activated under conditions when SR Ca 2+ stores were full, i.e., loaded to endogenous levels prior to activation. While these bulk SR Ca 2+ levels are largely maintained upon low-frequency EFS [45,46,112], SOCE was found to be activated with individual APs nevertheless [39].
What were the main arguments for actually defining this Ca 2+ influx as being storedependent? First of all, the amount of Ca 2+ lost in the t-system due to the activation of pSOCE showed a clear dependence on the amount of Ca 2+ released from the SR upon an AP [39], which is the defining property of SOCE. If not SOCE, this dependence could also be explained by a Ca 2+ -dependent mechanism in which Ca 2+ released from the SR per se would activate Ca 2+ -dependent ion channels at the sarcolemma, as it is known, e.g., from the family of transient receptor potential (TRP) channels. This, however, was ruled out by showing that pSOCE was unaffected when the fiber was bathed in a solution buffered with the fast Ca 2+ -chelating agent BAPTA instead of EGTA [39]. Second, pSOCE was blocked when Ca 2+ release from the SR was inhibited by blocking the RyR1. Thus, micromolar concentrations of both tetracaine and ryanodine abolished SR Ca 2+ release and pSOCE [39]. Moreover, increasing [Mg 2+ ] cyto from 1 to 3 mM, a well-known condition that inhibits RyR activity, largely suppressed SR Ca 2+ release, in agreement with previous findings [107,113,114], and abolished pSOCE [39]. Of note, while tetracaine and ryanodine can affect Ca v 1.1 and Na v 1.4 function, despite using higher concentrations than reported within this study [39], increased [Mg 2+ ] cyto did not affect t-system excitability [113,114]. Third, pSOCE was reduced by increasing the SR Ca 2+ load. Thus, exposing the fiber to [Ca 2+ ] cyto that was increased from an approximately physiological level of 67 nM to 200 nM and 1.3 µM led to an increase in [Ca 2+ ] SR and a subsequent reduction in pSOCE [111], proposedly by increasing the distance (concentration-wise) to the threshold of SOCE activation. The observed inhibition was not due to the increased [Ca 2+ ] cyto per se, as preloading the fiber at an increased [Ca 2+ ] cyto of 200 nM and then returning it to the original [Ca 2+ ] cyto of 67 nM immediately before the pSOCE assessment (establishing identical recording conditions but different SR Ca 2+ load) resulted in marked suppression of pSOCE at the beginning of the recording, again in agreement with an increased distance to the threshold of SOCE activation.

A Potential Model of Phasic SOCE Activation
pSOCE was activated with every AP under conditions where the bulk SR Ca 2+ levels were barely depleted [45,46,112]. How is this possible? A potential model to explain pSOCE activation despite full SR Ca 2+ stores was recently proposed by Koenig et al. [39,111]; it is shown in Figure 4 and discussed briefly. Under resting conditions, free [Ca 2+ ] SR was reported at about 0.5 mM in intact FDB fibers [115] and 0.31 mM in mouse tibialis anterior (TA) muscle [112]. In rat skinned EDL fibers with [Ca 2+ ] cyto set to 67 nM, free [Ca 2+ ] SR was determined to be 0.7 mM [111]. The luminal Ca 2+ -affinity of STIM1 was estimated to be around 0.2 mM, with reported Kd Ca values of STIM1 between 200 and 600 µM [116],~200 µM [117],~150 µM [26], and an estimated upper limit of 300 µM [111]. Thus, under resting conditions when [Ca 2+ ] SR is between 0.5-0.7 mM, STIM1 luminal Ca 2+ -binding sites must be mostly occupied; bulk [Ca 2+ ] SR does not drop significantly during an AP [45,46,112] and, hence, cannot be critical for pSOCE activation. Steady [Ca 2+ ] SR is maintained by the strong buffering of CASQ, which is the main Ca 2+ -buffer in the SR [118][119][120]. CASQ is essentially absent in the lSR [121] but is concentrated at the jSR, where it provides large amounts of Ca 2+ to be released despite maintaining [Ca 2+ ] SR at a steady level. This is accomplished by the high buffering capacity of CASQ, which is further increased upon polymerization [122]. When [Ca 2+ ] SR is high, CASQ forms long polymeric tendrils that are anchored to the RyR1 via triadin and junctin [91,123]. These tree-like structures, also known as Ca 2+ wires, are thought to facilitate the diffusion of Ca 2+ toward the RyR1 channel mouth due to a reduction in dimensionality [124,125]. The structural inhomogeneity of Ca 2+ -buffer across the SR and the local release of Ca 2+ through the RyR1 at the jSR must result in spatial and temporal differences in [Ca 2+ ] SR [45]. These SR Ca 2+ gradients may be expected to be small when considering basal RyR1 leakage [90,126,127] but must become significant during AP-induced SR Ca 2+ release. In particular, within the jSR, Ca 2+ nanodomains will form in which Ca 2+ concentrations will differ substantially from the Ca 2+ concentrations present in the bulk SR; however, direct assessment of these concentrations was not possible in the past due to a lack of respective experimental means.  [45,46,112]. Recent advances have, however, overcome these limitations owing to the development of genetically encoded probes. Evidence for local Ca 2+ microdomains was provided recently using a G-CatchER + sensor fused to JP-45 [128] in mouse fast-twitch FDB fibers. Because JP-45 is an integral component of the triad that interacts with Ca v 1.1 at its N-terminus and with CASQ at its C-terminal, it is localized to the jSR within the triad region. The G-CatchER + Ca 2+ probe fused to the C-terminal end can thus report the Ca 2+ concentration within the lumen of the jSR. While this Ca 2+ probe was too slow to catch changes associated with single muscle twitches, it nevertheless proved an important principle: Ca 2+ levels dropped much faster and deeper at the jSR when compared to the bulk SR. This was shown for fibers that were activated with tetanic stimulation, but these experiments certainly imply that changes occur upon single muscle twitches also. It is therefore comprehensible from a physiological point of view that respective STIM1 Ca 2+ -sensing proteins are located at these sites [29,38,41] in close proximity to the RyR1, where they would react most sensitive to changes in free Ca 2+ . Other pools of STIM1 distributed along the lSR would sense bulk [Ca 2+ ] SR and hence react with much less sensitivity. Their role is likely not an immediate one, but may involve recruitment of additional Ca 2+ entry pathways upon heavy demand or functional impairment.
While cSOCE was attributed to STIM/Orai [38,89,129], the molecular correlates of pSOCE have not been elucidated to date. A simple model would assume that pSOCE is carried by the same protein machinery as chronic SOCE [39,111], but this has not been shown to date. Besides a coupling of STIM1 with Orai1, two other main alternatives come to mind considering current knowledge: (i) Ca 2+ -channels other than Orai1, e.g., from the TRP family of ion channels, could mediate pSOCE, although some TRP channels have been neglected as mediators of cSOCE. For example, cSOCE was shown to be unaffected by the knockdown of TRPC3 [130] or by transient expression of dominant-negative TRPC3 and TRPC6 [131]. While apparently not mediating cSOCE, they could potentially carry pSOCE, either alone or together with Orai1 within a heteromeric channel structure [132]. Thus, the coupling of TRPC1 and TRPC4 to STIM1L was recently suggested to underlie fast SOCE in muscle [133], and a ternary complex of TRPC1, Orai1, and STIM1 was proposed to underlie SOCE in human salivary glands and platelets [132,[134][135][136]. (ii) It is also possible that Orai is not gated by STIM proteins but by direct conformational coupling to RyR1 [131,137]. Thus, Orai1 could couple to RyR1 in a similar way, as it was shown for TRPC channels and RyR1 [138,139]; coupling to the RyR1 would confer the necessary "store-dependence," not by sensing SR luminal Ca 2+ , but by coupling Orai1 activation to the conformational changes in RyR1. Apart from RyR1, it was also suggested that cSOCE involves coupling to IP 3 R [94,140], although IP 3 R expression in skeletal muscle is relatively low and preferentially confined to the nuclear envelope and NMJ [52,141].
Elucidating the molecular correlates of pSOCE was limited by non-selective pharmacology (see below) and the fact that pSOCE measurements originally described in rat EDL muscle fibers [39,111] were not easily translatable to mouse muscle fibers. Thus, EFS worked well in rat EDL fibers but, for reasons still unknown, mouse EDL fibers were found to be much less susceptible to EFS. Only recently, Lilliu et al. managed to overcome this limitation by training mice before SOCE measurements were performed [142]. Thus, providing running wheels to the cages of otherwise physically inactive mice enabled successful EFS in mouse skinned EDL fibers. Access to the running wheels for 5-6 days proved sufficient for EFS to work in mouse fibers in a manner that was completely comparable to rat fibers. pSOCE assessed in EDL fibers from these trained mice was indistinguishable from that in rat, suggesting the presence of pSOCE across mammalian species. Importantly, this new experimental paradigm opened the door to use genetically modified mice to test for the molecular nature of pSOCE.

What Is the Physiological Role of Phasic SOCE?
The demonstration of pSOCE upon individual APs [39,111,142] shed new light on the physiological role of SOCE in skeletal muscle function in general. Given that pSOCE was activated already with the first AP under conditions of endogenous SR Ca 2+ loads suggested a primary physiological role of pSOCE and potentially of cSOCE, independent of SR Ca 2+ store refilling. Of note, pSOCE was best assessed under conditions of high cytosolic buffering (10 mM EGTA; [39]). The high EGTA strongly buffered the [Ca 2+ ] cyto that was released from the SR to keep [Ca 2+ ] cyto at relatively low levels during EC coupling. This ensured a larger Ca 2+ gradient across the t-system membrane and hence enabled a larger pSOCE flux (increased driving force compared to low EGTA buffering conditions). When EGTA was lowered to 0.2 mM, equivalent to the endogenous Ca 2+ -buffering power in fasttwitch muscle [39,143], pSOCE could not be resolved anymore [39], indicating no net Ca 2+ flux under these conditions. Apparently, [Ca 2+ ] cyto , in particular within the triadic cleft, reaches levels equaling that within the t-system during a twitch such that pSOCE operated near the reversal potential of Orai with a strongly reduced driving force (the depolarized membrane potential of the AP will not greatly affect the driving force for pSOCE as the AP is almost fully depolarized before Ca 2+ reaches its peak in the cytoplasm (e.g., [144,145]), and pSOCE was activated only an additional 0.3 ms after SR Ca 2+ release [111]).
Another explanation for the decline in pSOCE amplitude with low cytosolic buffering could be provided by the reported inward rectification of the current-voltage relationship of Orai1 [24]. While this seems to be an intrinsic property of Orai1 channels rather than being caused by the asymmetric ionic composition of electrophysiological test solutions [24], it remains to be tested whether this rectification persists under near symmetric Ca 2+ concentrations at a high micromolar [Ca 2+ ] cyto , as is the case during a twitch. Irrespective of the biophysical mechanism behind the phenomenon, the data suggests that pSOCE, in cooperation with the t-system Ca 2+ -uptake proteins Na + /Ca 2+ exchanger (NCX) and/or plasma membrane Ca 2+ ATPase (PMCA), equilibrates Ca 2+ gradients across the t-system membrane in a bidirectional manner, leading to an influx of Ca 2+ if [Ca 2+ ] cyto transients are lower and an efflux of Ca 2+ if [Ca 2+ ] cyto transients are higher than "normal." These considerations suggest that pSOCE regulates fiber Ca 2+ on a twitch-to-twitch basis by correcting changes in SR Ca 2+ release. Thus, rather than refilling depleted SR Ca 2+ stores, the role of pSOCE is probably better described as maintaining SR Ca 2+ levels. One could also see pSOCE as a mechanism that prevents excessive build-up of Ca 2+ within the t-system when [Ca 2+ ] cyto levels peak during a muscle twitch. By "clamping" the t-system Ca 2+ levels to the levels found in the extracellular space, the function of pSOCE would therefore avoid significant diffusional loss of Ca 2+ from the t-system into the external environment of the fiber [39,111].
As mentioned previously, it remains to be confirmed whether pSOCE depends on the very same protein machinery as cSOCE, i.e., STIM and Orai proteins, but it is also possible that their physiological function and localization are different. Based on what is known today, one can postulate (at least) two pools of SOCE to exist in skeletal muscle cells: one endogenous pool that can be activated rapidly and most likely relies on preformed clusters within the triads (see above); the other one, an inducible pool, is recruited only slowly after strong SR Ca 2+ store depletion and might involve additional membrane structures [42,85], potential relocation of STIM proteins [29,146], and recruitment of additional Orai1 channels [42]. It is possible that the inducible pool of SOCE represents a condition of protein "overexpression," which arises only under unphysiological conditions of SERCA block-induced store depletion, where an inducible pool of SOCE will outperform endogenous SOCE. Thus, its physiological relevance in skeletal muscle should be considered with caution.

Pharmacology of SOCE in Skeletal Muscle: Recent Advances
Since even before the molecular mechanisms of SOCE were elucidated, drugs from the group of imidazoles (e.g., SKF-96365; [147,148]), diphenylboronate (e.g., 2-APB; [149]), and pyrazoles (e.g., BTPs; [150][151][152]; further discussed below) were used to target SOCE. With the discovery of STIM and Orai, SOCE pharmacology significantly advanced based on the recombinant expression of respective proteins in non-excitable cells. However, data on the effects of the most commonly used SOCE modulators in muscle cells and, specifically, in skeletal muscle fibers, are sparse. In this section, we therefore give an overview of the current knowledge on SOCE pharmacology in skeletal muscle, with a particular focus on the common practice and recent developments in the field. For an overview of general SOCE pharmacology, the reader is referred to excellent reviews elsewhere (e.g., [24,[153][154][155]).
SKF96365 was first described as a receptor-mediated Ca 2+ influx blocker [147] and only later as a SOCE blocker in Jurkat T cells [167]. Studies performed in non-excitable and skeletal muscle cells highlighted its low selectivity, as it targets a wide variety of ion channels, including voltage-gated Ca 2+ [96,148,166,168], TRPM4 [169], and K ATP channels [170]. Moreover, SKF96365 at concentrations in the micromolar range was shown to affect ER Ca 2+ pumps in human endothelial cells [171] and NCX in glioblastoma cells with an EC 50 of~10 µM [172]. To block SOCE in skeletal muscle, SKF96365 is commonly applied at a concentration of 20-30 µM [38,59]. Within the same concentration range, it is used to block ECCE [166,173]; it also blocked APACC at 25 µM [96]. Taken together, these studies show that SKF96365 inhibits extracellular Ca 2+ entry into skeletal myotubes and muscle fibers mediated by SOCE, ECCE, or APACC with similar potency.
2-Aminoethyldiphenyl borate (2-APB) was widely used to modulate Ca 2+ fluxes and I CRAC in a variety of tissues and cell types, including skeletal muscle [149,157,174,175]. It has a complex pharmacological profile in terms of both selectivity and mechanism of action. Initially, it was described as an IP 3 R antagonist [51,149,[176][177][178], only to be later recognized as an SOCE blocker [179][180][181]. More recent findings in non-excitable cells have also highlighted the ability of 2-APB to modulate STIM1 multimerization and/or STIM1-Orai1 interaction [182,183], as well as the activity of potassium channels, SERCA pumps, TRPV1-2-3 channels, and interestingly, the DHPR [166,168]. Interestingly, 2-APB also shows a bi-phasic effect by which it can potentiate I CRAC at low concentrations (1-10 µM), while higher concentrations (20-50 µM) inhibit I CRAC [180,182,184]. Such a feature also affected the modulation of individual Orai isoforms. In HEK cells, 5 µM of 2-APB potentiated currents through Orai1 and Orai2 with no effect on Orai3, while 50 µM strongly inhibited Orai1 but potentiated Orai3-mediated currents [156,182,184]. Taken together, these observations suggest that 2-APB does not constitute a valuable tool when looking at specific channel activity.
3,5-bis(trifluorometyl)pyrazole (BTP) compounds were characterized for the first time as nuclear factor of activated T-cells (NFAT) activators and cytokine production inhibitors of T cells [150][151][152]. It was only later when BTP2 (also known as YM58483 or Pyr2) was described as a potent I CRAC inhibitor [185][186][187]. During recent years, multiple studies reported the low selectivity of BTP2, as it was shown to potentiate the activity of TRPM4 [188] and to block TRPC3 and TRPC5 [189]. Despite its low selectivity, it is widely used as a SOCE inhibitor in skeletal muscle [38,41,87,110,190,191] at a concentration of 10 µM. Recently, a more detailed characterization of its effect in skeletal muscle fiber Ca 2+ handling was provided [192]. The authors observed that BTP2 affects Ca 2+ handling in rat skinned EDL fibers at multiple levels. It inhibited SOCE at low concentrations (1-5 µM) but, interestingly, also impaired RyR1 activity at higher concentrations. Thus, it blocked AP-induced SR Ca 2+ release and RyR leak at >10 µM. The impairment of RyR1 occurred in an indirect manner, probably by affecting the DHPR [192], which is a well-known target of other SOCE modulators (e.g., 2-APB and SKF96365 [168]).
GSK-7975A and relatives (GSK-5503A, GSK-5498A) belong to the class of pyrazole derivatives developed by GlaxoSmithKline [193][194][195] that are widely used as I CRAC inhibitors. GSK-7975A was shown to inhibit Orai1-mediated currents with an IC 50 of 4 µM in a heterologous expression system, presumably by altering the Orai1 pore geometry [194,196]. A recent work done in skeletal muscle showed the ability of this compound to block Mn 2+ entry upon store depletion in myoblasts expressing TAM-associated Orai1 mutants, and thus suggested it as a potential candidate drug for TAM treatment [196]. Nevertheless, GSK-7975A does not exhibit high selectivity, as it also affects TRPV6 and L-type Ca 2+ channels at 10 µM [194], and other Orai isoforms, i.e., Orai2 and, to a lesser extent, Orai3 [156]. The mechanism of SOCE inhibition by GSK-7975A appeared to be independent of STIM1-Orai1 coupling or STIM1-oligomerization, but was suggested to involve an allosteric modulation of the Orai1 selectivity filter [194]. However, the observed slow onset of inhibition (several minutes) argues against such a direct effect on the channel pore [24]. To the best of our knowledge, GSK compounds have not been applied to myotubes or skeletal muscle fibers such that respective insights are lacking to date.
Synta66 is structurally closely related to BTP2 and inhibits Orai1 activity at micromolar concentrations via a pore-blocking mechanism [197]. It has weaker reported off-target effects on other ion channels including TRPC1/5 [198], voltage-gated Ca 2+ and Na + channels [199], and other Orai isoforms [156]. However, comparable to BTP2 and the GSK compounds, the speed of inhibition by Synta 66 is rather slow such that cells need to be preincubated for more than 1 h for proper block development [24]. Together with the reported poor reversibility of drug action, this severely hinders the application of Synta66 in muscle fiber experiments. A more comprehensive characterization of the effects of Synta66 and its newly developed derivative in skeletal muscle is currently lacking.

Recent Drug Developments and Future Perspectives
Recent attempts have led to the synthesis of novel SOCE inhibitors, of which CIC-37 seems the most promising candidate [200], but also to the discovery of SOCE enhancers, e.g., AL-2T, NM-3G [200], and "compound 47" [201]. However, a respective evaluation of their effects on SOCE in skeletal muscle is to date pending. 4-((5-phenyl-4-(trifluoromethyl)thiazol-2-yl)amino)benzoic acid (IA65) is currently the only compound amongst the few reported SOCE enhancers whose effects have been characterized to some extent. This was done not only in heterologous expression systems but also in vascular smooth muscle cells and skeletal muscle fibers [202]. Here, IA65 increased the Orai1-mediated current and accelerated CDI in a concentration-dependent manner with an EC 50 of 2 µM. Ca 2+ influx through SOCE measured in rat skinned EDL fibers upon exposure to caffeine and low Mg 2+ showed a comparable increase by about 20-30% at 10 µM IA65. IA65 exhibited some selectivity, enhancing Orai1 activity but only marginal affecting Orai2 and Orai3 [156,202]. It remains to be seen whether these novel drugs will provide helpful tools to study SOCE in skeletal muscle.
What can be learned from the existing data? While drug specificity is always of concern when applying pharmacology, it seems that this particularly applies to SOCE: (i) the different Ca 2+ influx pathways across the t-tubule membrane are difficult to discriminate with current pharmacology; (ii) the importance of several proteins to the function of SOCE, be it direct (STIM, Orai) or indirect (RyR, SERCA, etc.), greatly increase the probability for off-target effects; (iii) the RyR plays a pivotal role in SOCE activation [30,39,59,89,97] but is characterized by substantial drug binding promiscuity toward small molecules and complex drug-receptor interactions, as exemplified by some of the better-studied molecules, namely, ryanodine and tetracaine (rev in [203]). In particular, the situation becomes delicate when considering fully mature muscle fibers, when all of the above applies, and interpretations solely based on drug action are not straightforward, as demonstrated, e.g., by Meizoso-Huesca [192]. Regarding pSOCE, another level of complexity is added when the whole EC-coupling machinery may constitute a potential drug target, including Ca v 1.1 and Na v 1.4, which are known to be affected by current SOCE pharmacology.

Variety of STIM Molecules Expressed in Adult Tissue
Skeletal muscle does not only express STIM1 but also a long splice variant, namely, STIM1L, as well as two STIM2 isoforms. STIM1L, which results from the alternative splicing of the STIM1 gene at exon 11, was discovered 10 years ago to be expressed in human myotubes [146]. In contrast to STIM1, STIM1L is not ubiquitously expressed but is restricted to skeletal and cardiac muscle, as well as to the brain, at least in mice. Another study reported STIM1L mRNA to be exclusively expressed in human skeletal muscle [204]. STIM1L has an extra 106 amino acids in the C-terminal region, which comprises a putative actin-binding domain. STIM1L is absent in myoblasts, and its expression increases during differentiation, eventually reaching a similar level as STIM1 in adult tissue [146]. STIM1L was shown to be pre-localized at the PM before store depletion in clusters co-localizing with Orai1. Knockdown of STIM1L specifically (but not STIM1) slowed down the activation kinetics of SOCE [146], pointing to a specific function of STIM1L in the fast activation of skeletal muscle SOCE. It should be stressed that these results were obtained from experiments that were performed on differentiating myotubes before triad formation had occurred; thus, the localization of STIM1L in adult fibers remains to be determined. Subsequent studies on the intrinsic properties of STIM1L revealed that when expressed in non-muscle cells (mouse embryonic fibroblasts double KO for STIM1/2, MEFDKO), STIM1L did not lead to a faster activation of SOCE, even if the protein was pre-localized at the PM prior to store depletion [205]. This could be explained by the fact that Orai1 did not form clusters co-localizing with STIM1L in MEFDKO before store depletion, as seen in myotubes. These data further imply that additional proteins are necessary for the formation of permanent "preassembled" clusters of STIM1L and Orai1 in myotubes. Additional experiments will tell whether STIM1L is indeed necessary and sufficient for fast SOCE in skeletal muscle. Using MEFDKO cells, it was further shown that SOCE was larger when induced by STIM1L compared to STIM1 [205]. Interestingly, however, the cER did not enlarge after store depletion, as is the case in STIM1 overexpressing cells [17]. Hence, the higher SOCE took place even at reduced contact between the ER and the PM. Several hypotheses can potentially explain the enhanced Ca 2+ entry induced by STIM1L. A reduction in ER-PM contacts could lead to less Ca 2+ accumulation in the proximity of the Orai1 channel to reduce CDI and hence augment SOCE. Another explanation is linked to different channels that are activated by STIM1L. Electrophysiological recordings in HEK cells expressing STIM1 or STIM1L showed that STIM1L is indeed more prone to activating TRPC1 compared to STIM1 [206]. On the contrary, STIM1 is a better Orai1 opener than STIM1L, but in HEK, as well as in MEF cells, STIM1L-induced SOCE was larger than STIM1-induced SOCE [206]. Along this line, Horinouchi et al. [204] reported that STIM1L induced more Ca 2+ entry than STIM1 and interacts more strongly with TRPC (in that case TRPC3 and TRPC6) than STIM1. Interestingly, in differentiated human myotubes, it was shown that STIM1L indeed physically interacts with TRPC1 and TRPC4 [133]. Furthermore, the downregulation of either STIM1L or TRPC1/4 resulted in similar phenotypes, such as a reduced myoblast fusion into myotubes and an impaired ability to sustain repetitive SR Ca 2+ release [133,207]. Along this line, the kinetics of SOCE was slower in myotubes with reduced STIM1L or TRPC1/4 expression [133]. Hence, it appears that besides Orai1, TRPC channels are involved in muscle SOCE, in particular, the ones activated by STIM1L, and that this plays an important role for myotube formation and their ability to release Ca 2+ (rev in [208]).
The other member of the STIM family, namely, STIM2, is expressed in muscle as well and plays a role in differentiation. The downregulation of STIM2 in human skeletal muscle led to an impaired myoblast differentiation, a decreased expression of myogenin and myocyte enhancer factor-2 (MEF2), and a reduction in SOCE [209]. In addition, the basal cytosolic Ca 2+ level was reduced both in myoblasts and myotubes, in line with several reports showing the role of STIM2 in the maintenance of resting Ca 2+ levels in different cell types [210]. Interestingly, the level of STIM2 expression increased during the process of muscle differentiation, which could point to an important role in adult tissue as well. However, STIM2 KO mice did not display alterations in muscle function, which might be explained by an overall low expression level of STIM2 in mice [211]. Considering the strong effect of siSTIM2 on SOCE in human myotubes (about an 80% reduction; [209]), one can speculate that human skeletal muscle expresses STIM2 at a higher level than mouse muscle, or that STIM2 is more important for SOCE in human tissue. Indeed, the downregulation of STIM2 reduced SOCE by only 20% in mouse myotubes [212]. Besides the effect of STIM2 on SOCE, SERCA1a activity was enhanced in STIM2-downregulated cells [212]. Accordingly, the SR Ca 2+ content was higher and, consequently, KCl-and caffeine-induced Ca 2+ release was more pronounced in the case of STIM2 downregulation. The cytosolic part of STIM2 was identified as the binding region to SERCA1a [212]. In contrast, in human myotubes, KCl-induced Ca 2+ release was reduced in STIM2-downregulated cells and repeated KCl stimulations rapidly led to SR Ca 2+ store depletion [209]. The conflicting results obtained after the downregulation of STIM2 might have been due to species differences or the presence of an alternative STIM2 splice form. In 2015, two groups discovered an inhibitory isoform of STIM2, called STIM2β or STIM2.1, which has an insertion of eight amino acids in the CAD domain of STIM2, preventing Orai1 gating; the other STIM2 was called STIM2a or STIM2.2. [213,214]. Interestingly, the expression of this splice form increased in differentiating myotubes [213,215]. The KO of STIM2.1 in C2C12 muscle cells showed a decreased differentiation process, as highlighted by a reduced expression of myosin heavy chain and myogenin [215]. Furthermore, MEF2C and NFAT4 expression were reduced, suggesting that an increased expression of STIM2.1 during myogenesis promotes the formation of myotubes through MEF2C and NFAT4. In line with the inhibitory role of STIM2.1, SOCE was slightly elevated in cells lacking this isoform. Finally, myoblast proliferation was increased in cells lacking STIM2.1, possibly due to the basal cytosolic Ca 2+ level elevation that was observed under this condition [215]. This effect on the basal Ca 2+ level was, however, not seen in CD4 + T cells [214], which might point to different functions of STIM2.1 depending on the specific cell type. What should also be considered is the ratio between both STIM2 isoforms (STIM2.1/STIM2.2), which is likely different between T cells and muscle. In addition, this ratio changes during myogenesis, with an increased expression of the inhibitory STIM2.1 isoform. This might help to prevent excessive Ca 2+ entry during muscle formation, considering that STIM1L expression also increases during muscle differentiation and promotes SOCE [146]. It is well known that overactive SOCE is detrimental for muscle formation/function, as evidenced by the gain-of-function mutation in either STIM1 or Orai1 that leads to TAM [216]. Hence, the expression of an inhibitory isoform, such as STIM2.1, during myogenesis likely participates in the tight regulation of Ca 2+ homeostasis during muscle formation, but also in mature fibers. We are now only at the beginning of understanding the complex regulation of SOCE that is linked to the expression of STIM proteins, and future studies should bring exciting new findings to better understand the interplay between the four STIM isoforms that are expressed in skeletal muscle.

STIM and Orai in Different Fiber Types
Based on the different properties regarding the speed of contraction, metabolism, and fatigue resistance, different types of muscle fibers can be distinguished in skeletal muscle. According to their contraction velocity, two main types are defined, namely, type I and type II, which are also called slow-and fast-twitch fibers, respectively [217]. The different expression of myosin heavy chain (MHC) isoforms and ATPase activity (rev in [218]) allows for further subdividing adult fibers into four groups, from the slowest one to the fastest one (IIB < IIX < IIA < I, with IIB not being found in humans). In fast-twitch fibers, the Ca 2+ release upon an AP has a higher amplitude and shorter duration than in slow-twitch fibers [219]. The faster muscle relaxation speed can be explained by a better organization of the SR and t-tubules in the fast-twitch fibers [220] and a more robust expression of the RyR1 and SERCA proteins [219]. Skeletal muscle fiber composition is heterogeneous depending on the muscle type, and muscles containing more slow-twitch fibers, such as in soleus muscles, resist fatigue better and are mainly responsible for the maintenance of posture. In contrast, muscles that are mainly composed of fast-twitch fibers, such as the EDL, are more sensitive to fatigue and are mainly used for intensive and short-time contractions.
The contribution of SOCE toward maintaining contractile force during repetitive muscle activity in the different fiber types is not clearly defined for the time being. Ex vivo experiments showed that the contractile force of soleus muscles, upon stimulation at high frequencies, is more dependent on extracellular Ca 2+ than that of EDL muscles. Hence, in soleus muscles, the force of contraction decreased more strongly in the presence of BTP2 or the absence of extracellular Ca 2+ [190], showing a greater implication of SOCE to avoid fatigue during intensive muscle contractions in slow-twitch fibers. Still, as SOCE inhibition affected both fiber types during tiring stimulations, it is likely that the role of SOCE is the same in the two types of fibers. However, the relative importance of SOCE for preventing muscle fatigue is more pronounced in muscles that are more frequently used, as is the case for slow muscle. Interestingly, upon low-frequency stimulation, SOCE inhibitors did not enhance fatigue, neither in slow-nor in fast-twitch muscle.
A critical role of SOCE in slow muscle was demonstrated recently in muscle-specific Orai1 knock-out mice [129]. The reduction in specific force in soleus muscle from Orai1 KO animals was stronger than in EDL muscle. In addition, Orai1 deletion more markedly affected soleus muscle with a reduction in the percentage and the cross-section area of slow-twitch fibers (type I) in adult mice. As the reduction in muscle force was not observed in inducible Orai1 KO animals (where Orai1 expression was reduced only at adulthood), it was concluded that the decreased specific force specifically in soleus muscle was due to a developmental defect rather than a lack of Ca 2+ entry during EC coupling [129]. It should also be noted that in human patients, Orai1 loss of function, which in this case is not muscle-specific, induces atrophy of fast fibers, with a resulting predominance of slow fibers ( [221] and rev in [86]). Why slow and fast fibers are differently affected by the absence/loss of function of Orai1 is not clear. It could be related to species variants, i.e., human versus mouse, or differences in the tissues lacking Orai1, i.e., all cells versus skeletal muscle. More investigations are required to firmly establish the function of Orai1 (and SOCE) across fiber types.
Another interesting point is related to the formation of tubular aggregates (TAs), which are abnormal structures of the SR that strongly express STIM1 and Orai1 [222]. In humans, TAs are described in patients with a gain-of-function mutation of Orai1 or STIM1 (rev in [216]). It was proposed that these very compact SR structures sequester STIM1 and Orai1 to protect the muscle cell from excess Ca 2+ . In mice, TAs are found in old animals almost exclusively in type IIB fibers [223], which are the fastest, most easily tired ones, but also the least used fibers (rev in [224]). Interestingly, when older mice were allowed to run in a wheel during most of their life span (between 9 and 24 months), the formation of TAs was reduced [222]. In this case, the old muscles from trained mice required extracellular Ca 2+ entry for a stronger contraction and are sensitive to SOCE inhibitors [222], which was not the case in old, untrained animals [190]. Indeed, as mentioned previously (Section 2.2), it was proposed that in old animals, SOCE is not required to maintain force, essentially because SOCE is much less functional with age [73]. The decline in SOCE function with age is, however, in contradiction with another study reporting fully functional SOCE in old animals [76]. From these animal studies, one can hypothesize that the more muscles are used, especially slow muscles, the more SOCE is needed to fight fatigue during intense effort. In very fast fibers, such as IIB fibers, muscle usage overall is less frequent and even less during aging, and the role of SOCE to prevent fatigue is less pronounced. However, the fast fibers develop TAs with age, possibly as a protective mechanism against Ca 2+ overload. Slow fibers, with a higher mitochondrial density and a preferential oxidative metabolism, might be better protected against Ca 2+ overload, thus not requiring TAs formation, but this remains hypothetical. As reported in a recent comprehensive review on CEUs, the structural rearrangement that occurs upon strenuous exercise was so far only studied/observed in fast muscle [84]. It would be of interest to know whether this is specific to fast fibers or whether slow fibers also undergo such t-tubule rearrangement. Considering the significant difference in Ca 2+ handling between the two fiber types, it would not be surprising that SOCE is also divergent in terms of localization within the fiber and regarding respective activation/regulatory mechanisms.
Finally, what do we know about the expression of the SOCE molecules in the different fiber types? Not much is known besides a study that reported a higher expression level of STIM1L in rat soleus muscle compared to rat EDL muscle [97]. The level of STIM1 was, however, similar between slow and fast muscle [97]. Recently, two studies investigated the RNA expression in all nuclei of different mouse muscle cells using single nuclei-RNA seq [225,226]. When we re-analyzed the dataset, we found that in adults, the expression of STIM1 and STIM2 is not homogeneous in the different types of fibers. STIM1 is strongly expressed in type I, IIA, and IIX fibers in the adult soleus, while STIM2 is hardly detected in the soleus muscle. Surprisingly, in adult TA muscle, which is mainly made up of fast fibers, STIM1 is expressed more in IIX fibers than in faster IIB fibers, while conversely, STIM2 is expressed mainly in IIB fibers. Hence, to schematize, STIM1 is more a slow fiber protein, while STIM2 is a fast fiber one. These results highlight potentially distinct functions of STIM1 and STIM2 in different muscle fibers and might confirm a greater role of STIM1-induced SOCE in slow fibers. In contrast, the expression of Orai1 is more uniform in the different fibers of TA muscle [225,226]. The physiological consequences of such distinct expressions of STIM1 and STIM2 (assuming the protein expression follows the same pattern as the mRNA) open new exciting ways for further investigations to understand the function of both proteins and their splicing isoforms in skeletal muscle in depth.

Concluding Remarks
Skeletal muscle is not a homogenous tissue but comprises different types of fibers that have an extremely broad range of activity. The physiological function(s) and regulation(s) of SOCE in this context are so far only poorly understood. Several basic questions remain open, including the localization, mobility, and interaction of the involved SOCE molecules, as well as their differential expression among the different fiber types. Current methods for studying SOCE in muscle, some of which are specific to this tissue, such as the skinned fiber preparation and the force/fatigue measurements, have highlighted the peculiar and fascinating nature of this small but important Ca 2+ flux. The picture of SOCE that emerges is a multifaceted one, envisaging SOCE as both static and dynamic in nature, where SOCE is not confined to the triad alone but acts presumably across multiple sites within a single fiber, where it likely serves different physiological functions.  Data Availability Statement: All data presented in this review are present within the article or associated references.

Conflicts of Interest:
The authors declare no conflict of interest.