Ca2+-activated Cl− channel TMEM16A/ANO1 identified in zebrafish skeletal muscle is crucial for action potential acceleration

The Ca2+-activated Cl− channel (CaCC) TMEM16A/Anoctamin 1 (ANO1) is expressed in gastrointestinal epithelia and smooth muscle cells where it mediates secretion and intestinal motility. However, ANO1 Cl− conductance has never been reported to play a role in skeletal muscle. Here we show that ANO1 is robustly expressed in the highly evolved skeletal musculature of the euteleost species zebrafish. We characterised ANO1 as bonafide CaCC which is activated close to maximum by Ca2+ ions released from the SR during excitation-contraction (EC) coupling. Consequently, our study addressed the question about the physiological advantage of implementation of ANO1 into the euteleost skeletal-muscle EC coupling machinery. Our results reveal that Cl− influx through ANO1 plays an essential role in restricting the width of skeletal-muscle action potentials (APs) by accelerating the repolarisation phase. Resulting slimmer APs enable higher AP-frequencies and apparently tighter controlled, faster and stronger muscle contractions, crucial for high speed movements.

xcitation-contraction (EC) coupling in vertebrate skeletal muscle is initiated at the neuromuscular junction by an action potential (AP) from a single somatic efferent motor neuron. This neuronal AP causes the release of acetylcholine which binds to nicotinic acetylcholine receptors in the motor endplate of the muscle fibre, inducing influx of Na + ions. This in turn, causes a depolarising excitatory postsynaptic potential which as soon exceeds a certain threshold level, triggers a sarcolemmal AP. Subsequently, these depolarisations travel downwards into specific sarcolemmal invaginations, the transvers (t)-tubules, where they are detected by the voltage-sensing α 1S subunit of the dihydropyridine receptor (DHPR). This induces a conformational change in the DHPR which, via allosteric coupling, is transduced to the sarcoplasmic Ca 2+ release channel or ryanodine receptor type-1 (RyR1) 1 . Opening of RyR1 leads to a massive release of Ca 2+ ions from sarcoplasmic reticulum (SR) stores into the cytoplasmic gap 2,3 of this triadic junction, which bind to troponin C of the thin filaments, finally inducing muscle contraction via actin-myosin cross-bridge interactions.
Previous studies on EC coupling in zebrafish (Danio rerio) skeletal muscle revealed that the skeletal musculature of euteleost fishes is evolutionary highly advanced compared to mammals 4,5 . Similar to mammals, teleost skeletal muscles are composed of two major types of muscle fibres, classified as slow-(type I) and fasttwitch (type II) fibres. However, in contrast to mammals, where slow-and fast-twitch fibres are intermingled, teleost axial musculature displays a clear separation, with slow (oxidative/red) muscles found on the lateral surface and fast (glycolytic/white) muscles forming the deeper layers. Additionally, the selective expression of each of the two isoforms of DHPRα 1S subunit 4 and RyR1 6 in red and white zebrafish skeletal muscles resulted in the formation of two different muscle-type specific DHPR-RyR1 couplons. Again in contrast, mammals express only one type of couplon, regardless if it is in type I or type II fibres. This couplonspecification in zebrafish is just one example of how the 3rd round of teleost-specific whole-genome duplication (Ts3R) at the basis of the teleost clade 7,8 resulted in isoform formation during phylogenetic organ differentiation and hence in physiological specialisation 4,5 . Ts3R was also the initiation point of the loss of Ca 2+ conductance through both the DHPRα 1S isoforms 5 , interestingly, by involving distinct point mutations 4 . Notably, our investigations on the zebrafish skeletal muscle EC coupling machinery enabled us to detect another intriguing physiological and biophysical difference in EC coupling between euteleost and mammalian skeletal muscle-the unexpected participation of a Ca 2+ -activated Cl − channel (CaCC).
CaCC currents are involved in multiple physiological processes, ranging from sensory transduction 9 , epithelial secretion 10 , to smooth muscle contraction 11 . CaCC opening in smooth muscle cells results in membrane depolarisation due to Cl − efflux, since in contrast to skeletal muscle cells the intracellular Cl − concentration in smooth muscle cells is high due to active accumulation by Cl − /HCO 3 − exchange and Na + , K + , Cl − cotransportation 11 . Smooth muscle CaCCs are activated by different sources of intracellular Ca 2+ increase, viz. via Ca 2+ entry through voltage-gated Ca 2+ channels, Ca 2+ release from intracellular stores as a result of G-protein-coupled pathways, or Ca 2+induced Ca 2+ release through ryanodine receptors. Despite their crucial role in a wide range of biological processes, the molecular identity of CaCCs remained elusive until the year 2008, when three independent research groups [12][13][14] identified 'Transmembrane protein with unknown function 16' (TMEM16A), which is part of the 10-member mammalian TMEM16 family 15 , to be accountable for CaCC currents. Hydropathy analysis indicated that the anion-selective channel embraces 8 (octa) transmembrane helices with cytosolic C-and N-terminal ends, and hence was referred to as anoctamin 1 (ANO1) 14 . In general, ANO proteins are homo-dimers 16,17 , and in the case of ANO1 each subunit is activated independently 18,19 , suggesting a doublebarrelled channel similar to the ClC Cl − channel 20 . Newer reports, based on crystal structure analyses and cryo-electron microscopy, proposed that the membrane-spanning domain of ANO1 consists of 10 transmembrane α-helices 21,22 . Hallmark biophysical features of ANO1 currents are voltage-and Ca 2+dependent activation 23,24 with an outward rectification at lower micromolar Ca 2+ concentrations and a linear current-voltage (I-V) relationship at higher Ca 2+ concentrations 25,26 .
In the present study, we identify and characterise ANO1 expression pattern, subcellular distribution, and current properties in zebrafish skeletal muscle by implementing molecular, immunocytochemical, and biophysical approaches. Experiments on myotubes derived from the DHPRβ 1 -null zebrafish mutant relaxed which lack SR Ca 2+ release 27 , reveal that Cl − influx via sarcolemmal ANO1 channels is activated by Ca 2+ ions released through ryanodine receptors. Interestingly, our results demonstrate that this ANO1-mediated Cl − influx accelerates the repolarisation phase and thereby decreases the duration of the skeletal muscle AP, apparently acting synergistically with the canonical K + efflux repolarisation mechanism. Furthermore, we show that evolution of this accelerated AP repolarisation mechanism in the euteleost species Danio rerio allows enhanced muscle stimulation frequencies during AP trains, consequently expected to generate tighter muscle control as well as increased force production 28-32 -an all over improvement in muscle properties especially vital for the aquatic pray-predator environment.

Results
SR Ca 2+ release awakes zebrafish skeletal muscle Cl − current. Patch-clamp recordings from freshly dissociated zebrafish skeletal myotubes (Fig. 1a) displayed a very distinct picture compared to mouse myotubes, even under identical experimental conditions 30,33 . Although the standard depolarisation protocols elicited the expected robust SR Ca 2+ release in zebrafish myotubes, the archetypal slow DHPR Ca 2+ inward current was missing 4 , and surprisingly was 'replaced' by a huge outward current (Fig. 1a). To test if this outward current could be massive Cl − influx, we measured whole-cell currents under Cl − free conditions 34 . And indeed, under Cl − -free conditions the outward current nearly extinguished whereas SR Ca 2+ release remained intact (P > 0.05) (Fig. 1b). To assess if this Cl − current was Ca 2+ -dependent or more precisely dependent on Ca 2+ released from the SR during EC coupling, we repeated the recordings under standard Cl − conditions (165 mM Cl − in external and 4 mM Cl − in internal solution), but on myotubes isolated from the DHPRβ 1 -null zebrafish mutant relaxed, which lacks skeletal muscle EC coupling 27 . As demonstrated in Fig. 1c, d the lack of considerable Cl − current (left panels) is concordant with the lack of SR Ca 2+ release (right panels). Overall, integrating the above results we identified this molecular participant of the zebrafish skeletal muscle EC coupling mechanism as a Ca 2+ -activated Cl − channel (CaCC). The voltage dependence of the zebrafish skeletal muscle CaCC current (Fig. 1d) strongly resembles Anoctamin (ANO) currents described in mammalian smooth muscles 35,36 .
Zebrafish mutant relaxed contains functional CaCCs. As a first step, to validate if zebrafish relaxed myotubes are a suitable system for studying skeletal muscle CaCCs, we tested the physiological availability of CaCCs in relaxed myotubes by SR store depletion experiments in the presence of the RyR1 agonist caffeine, using the pulse protocol depicted in Fig. 2a. As expected, even before application of caffeine we recorded robust CaCC outward currents at +40 mV and smaller inward currents at −120 mV from normal control myotubes, but very marginal currents from relaxed myotubes (Fig. 2b). We observed a pronounced current rundown in normal myotubes (Fig. 2d), which is a characteristic of ANO currents and is most probably produced by phosphorylation of the channel by Ca 2+ /calmodulin-dependent protein kinase II (CamKII) 37,38 .
Notably, application of 8 mM caffeine after the 14th sweep of the pulse protocol (Fig. 2a) induced comparable (P > 0.05) augmentation of CaCC currents, in both normal as well as relaxed myotubes (Fig. 2c, d). These results unambiguously demonstrate that relaxed myotubes (i) contain intact SR Ca 2+ stores and (ii) express functional CaCCs which can be activated by intracellular Ca 2+ ions. Thus, zebrafish relaxed myotubes serve as an ideal experimental system for studying CaCCs under SR Ca 2+ -release free conditions. Two ANO1 isoforms are expressed in zebrafish skeletal muscle. Since strict Ca 2+ -dependence (Fig. 1c, d; Fig. 2), nearly-linear outward voltage dependence (Fig. 1d, left graph), and pronounced current rundown (Fig. 2d) are characteristics for Cl − currents of the ANO channel family, the apparent question arose which ANO isoform(s) is/are expressed in zebrafish skeletal muscle. Out of the 10 isoforms of the ANO protein family, only ANO1 and ANO2 are verified CaCCs ( Supplementary Fig. 1a), while the other 8 ANO proteins, either work as scramblases or have unknown functions 39,40 . As a result of the Ts3R genome duplication 7,8 zebrafish has two ANO1 and two ANO2 genes. The pedigree in Fig. 3a, with the Drosophila ANO protein as outgroup, indicates that the predicted translational products of the two ANO1 and two ANO2 zebrafish isoforms cluster well with their respective mouse counterparts but show a higher degree of phylogenetic advance. ANO1 and ANO2 isoform-specific RT-PCR primers were designed (Supplementary Table 1) to amplify DNA fragments   coding for a region between transmembrane α-helices 4 and 9 containing amino acid residues N650, E654, E702, E705, E734, and D738, supposed to delineate the ANO Ca 2+ -binding site 21,41 . Supplementary Figure 1b depicts the positions of these six residues on the proposed 10-helix membrane folding model of ANO channels 21 . Amino acid sequence alignment of ANO1 isoforms of zebrafish and mouse shows conservation of these six residues in both the ANO1 isoforms of zebrafish ( Supplementary   Fig. 1c). PCR amplification of ANO1 and ANO2 isoforms from first strands of whole adult zebrafish confirmed the accuracy of the four primer pairs and amplified fragments from all targeted ANO isoforms (Fig. 3b) to a similar extent (P > 0.05). On the contrary, RT-PCR amplification of ANO1 and ANO2 isoform fragments with first strands from red and white skeletal musculature showed no signals for both ANO2 isoforms ( Fig. 3c) but signals for both ANO1 isoforms (Fig. 3d). More specifically, Relaxed myotubes express functional CaCCs and contain intact SR Ca 2+ stores. a Test-pulse protocol from a holding potential of −80 mV to +40 mV for 100 ms followed by a 100-ms pulse to −120 mV. The entire protocol was repeated 100 times. Brown bar indicates the application of 8 mM caffeine to induce SR Ca 2+ store depletion between 15th and 45th sweep, which was terminated by perfusion with caffeine-free standard bath solution.
b Representative sweep of CaCC currents before application of caffeine showing robust CaCC outward current at +40 mV (I max = 77.12 ± 7.69 pA pF −1 , n = 5) and smaller inward current at −120 mV (I max = −27.85 ± 3.99 pA pF −1 , n = 5) from normal control myotubes (left trace), in contrast to only marginal outward (P < 0.001) and inward (P < 0.05) currents (I max = 9.32 ± 0.97 and −12.27 ± 3.20 pA pF −1 , n = 5, respectively) from relaxed myotubes (right trace). c Representative sweep of CaCC currents after application of 8 mM caffeine, showing indistinguishable (P > 0.05) CaCC currents at +40 and −120 mV between normal (I max = 79.50 ± 7.29 and −31.12 ± 4.22 pA pF −1 , respectively, n = 5) (left trace) and relaxed (I max = 81.42 ± 10.14 and −28.18 ± 3.52 pA pF −1 , respectively, n = 5) (right trace) myotubes. Scale bars, 50 ms (horizontal), 30 pA pF −1 (vertical). d Plots of CaCC currents vs. sweeps at +40 and −120 mV with corresponding representative CaCC current traces (insets) from normal (left graph) and relaxed (right graph) myotubes. After caffeine application, the maximal CaCC outward current at +40 mV is indicated by dark blue or dark red arrows and the maximal inward current at −120 mV by light blue or light red arrows, from normal or relaxed myotubes, respectively. Data are presented as mean ± s.e.m.; P determined by unpaired Student's t-test ANO1-a isoform-specific primers predominantly amplified the DNA fragment from red muscle and ANO1-b-specific primers from white muscle. Furthermore, RT-PCR amplification by using two additional ANO2 pan-primer pairs confirmed the nonexistence of ANO2 isoforms in skeletal muscle ( Supplementary   Fig. 2). In order to test the purity of the tissue sample preparations, positive control PCRs with DHPRα 1S -a and DHPRα 1S -b isoform-specific primers and identical first strands were performed (Fig. 3e). Since α 1S -a and α 1S -b isoforms are exclusively expressed in red and white muscle, respectively 4 , the Signal intensity (%)  Table 1) showed signals with similar intensity (P > 0.05, n = 5) for RNAs that were isolated from adult normal zebrafish. c RT-PCR amplification with ANO2-specific primers under identical PCR conditions did not show any signal, neither from red muscle (r.m.) nor white muscle (w.m.). d In contrast, ANO1-specific primers amplified PCR fragments from both muscle types but to different extents. e Similarly, control PCR amplifications with DHPα 1S -specific primers detected both, α 1S -a and α 1S -b in both tissues but again to different extents. f Left graph, quantification of band intensities of ANO1 isoforms revealed that ANO1-a is predominantly expressed in the red muscle compared (P < 0.001) to white muscle (43.75 ± 1.89%, n = 5), and ANO1-b is mainly expressed in the white muscle compared (P < 0.001) to red muscle (45.56 ± 2.08%, n = 5). Right graph, DHPRα 1S -a and α 1S -b, known to be exclusively expressed in red and white muscles, respectively 4 , displayed a similar cross-contamination pattern for DHPRα 1S -a in white (24.21 ± 3.12%; n = 5) and DHPRα 1S -b in red muscle (49.90 ± 8.09%; n = 5). Bars represent mean ± s.e.m. and overlaying blue stars indicate individual data points; *** P < 0.001 determined by unpaired Student's t-test respective weaker amplification products in Fig. 3d, e are merely due to cross-contaminated red and white muscle preparations from the relatively small-sized zebrafish. Hence, from the very similar amplification intensity profile of ANO1 and DHPRα 1S isoforms ( Fig. 3f) we can conclude that ANO1-a is predominantly expressed in the red and ANO1-b in the white skeletal musculature of zebrafish.
To confirm that ANO1 isoforms are indeed the channels responsible for Ca 2+ -activated Cl − conductance in zebrafish skeletal muscle, we implemented a short interfering (siRNA) knock-down strategy previously described to work well in zebrafish 42,43 , since zebrafish ANO1 KO model strains are not available to date. Using the BLOCK-iT™ RNAi Designer algorithm (ThermoFisher Scientific) we designed a series of siRNA expressing DNA constructs targeting the 5′-untranslated region as well as the open reading frame of both ANO1 isoforms. Out of the ten siRNAs expressed in normal zebrafish myotubes seven showed significant (P < 0.05) knock-down of Ca 2+ -activated Cl − currents from 20.6 to 53.1% (n = 8-34) compared to a control scrambled siRNA (n = 28) (see Method section). The strongest current knock-down (53.1 ± 3.6%, n = 34) was obtained with siRNA targeted against a region, showing 91% homology between ANO1-a and ANO1-b isoforms (nucleotide positions 2199-2219 and 2073-2093, respectively). In contrast to the highly significant (P < 0.001) ANO1 current reduction (Fig. 4a), SR Ca 2+ release was indistinguishable (P > 0.05) from control scrambled siRNA expressing myotubes (Fig. 4b), indicating that the observed ANO1 current reduction cannot be attributed to putative siRNA off-target effects hampering SR Ca 2+ release. Altogether, our results explicitly ascertain that the channel responsible for zebrafish skeletal muscle Ca 2+ -activated Cl − conductance is ANO1.
ANO1 is localised in the sarcolemma. Since ANO1 is activated by SR Ca 2+ release, which is initiated by the allosteric interaction between DHPR and RyR1, we next tested if ANO1 channels are located closely to these Ca 2+ release sites in the triadic junctions or reside in the sarcolemma, or are rather distributed in both the subcellular domains. Immunolocalisation assays on zebrafish skeletal myotubes with anti-ANO1 and anti-DHPRα 1S antibodies suggest that ANO1 does not reside in the triads like DHPRα 1S , but is exclusively expressed in the sarcolemma (Fig. 5a). To test for the accuracy of our fixation procedure and resolution of our immunolocalisation approach for ANO protein detection, we immunostained zebrafish myotubes for another member of the ANO family expected to be present in the skeletal muscles of all vertebrate species, namely ANO5 ( Supplementary  Fig. 1a). Several lines of evidence suggest that ANO5, expressed in skeletal muscles of human and mouse 44 is essential for the development and maintenance of skeletal muscle 45 . Recessive mutations in ANO5 cause severe myopathies like gnathodiaphyseal dysplasia, GDD 45,46 , limb-girdle muscular dystrophy type-2L, LGMD2L 47 , and Miyoshi muscular dystrophy-3, MMD3 47,48 . Immunolabeling of zebrafish myotubes with anti-ANO5 antibody showed a similar expression pattern as observed previously for some SR targeted proteins like calsequestrin and the Ca 2+ sensor D1ER [49][50][51] . Consequently, we can conclude that ANO5 is expressed in the triadic SR membrane juxtaposing the t-tubules which were immunostained by anti-DHPRα 1S antibody (Fig. 5b). The strong and clear signal of ANO5 in zebrafish myotubes suggests that with identical cell preparation, fixation, and fluorescent staining procedures ANO1 would have certainly been detected in the triads if it would not solely reside in the sarcolemma.
To further confirm the specificity of ANO1 surface expression, we performed immunolocalisation assays on zebrafish skeletal myotubes expressing siRNA 2073-2093 (ANO1-b numbering) which was most effective in knocking-down the ANO1 current (Fig. 4a). Quantification of the sarcolemmal ANO1 fluorescence signal showed a significant (P < 0.001) reduction of ANO1 in myotubes expressing the ANO1-specific shRNA construct (61.4 ± 1.8%, n = 128) compared to control myotubes (100.0 ± 3.7%, n = 161) expressing the scrambled shRNA construct ( Supplementary Fig. 3). Altogether, results from the above experiments indicate the exclusive sarcolemmal localisation of ANO1 channels in zebrafish skeletal myotubes.
SR Ca 2+ release awakes ANO1 outward current close to maximum. To elucidate if ANO1 currents are activated only   (Fig. 6a). While the ANO1 outward current, recorded at +80 mV shows a bell shaped Ca 2+ dependence (Fig. 6a, upper graph), the ANO1 inward current at −140 mV displays a linear Ca 2+ concentration-current relationship (Fig. 6a, lower graph). These specific current properties sufficiently explain the ANO1 current-voltage (I-V) relationship during depolarisation-induced SR Ca 2+ release under physiological conditions (Fig. 6b, left graph). The cytosolic Ca 2+ concentration at negative potentials, below or at the initiation of SR Ca 2+ release around −40 mV (Supplementary Fig. 4a) is apparently too low to induce considerable ANO1 currents. Only intracellular addition of micromolar Ca 2+ can elicit ANO1 inward currents at negative potentials (Fig. 6a, lower graph), which finally yields a linear I-V curve at +13 µM of Ca 2+ addition (Fig. 6b, right graph).   Zebrafish skeletal muscle ANO1 channel is a bona fide CaCC. Similar to mammalian ANO1 channels 24,41 , zebrafish skeletal muscle ANO1 is gated synergistically by intracellular Ca 2+ ions and membrane potential, showing strong outward rectification at low intracellular Ca 2+ concentrations-a biophysical hallmark of CaCCs. Under physiological conditions (+0 cytosolic Ca 2+ addition), the skeletal muscle ANO1 I-V curve shows a zone of combined Ca 2+ -and voltage dependence between −40 and +10 mV and a pure voltage dependence zone between +10 and +80 mV, defined by the raising phase or plateau phase of SR Ca 2+ release, respectively ( Supplementary Fig. 4a). Outward rectification was still observed after cytosolic addition of low (+5 µM) Ca 2+ ions via the patch pipette ( Supplementary Fig. 4b). Regardless of whether Ca 2+ ions were added to the cytosol of normal or relaxed myotubes, inward currents were very small compared to the respective outward currents and consequently lead to a pronounced nick in the I-V curves (Supplementary Fig. 4b). As expected, outward currents recorded from normal myotubes were more pronounced compared to relaxed myotubes, due to additional current amplification by SR Ca 2+ release. Moreover, a linear I-V curve at high [Ca 2+ ] (+13 µM Ca 2+ addition) (Fig. 6b, right graph) depicts the competence of the skeletal muscle ANO1 channel for voltage-independent pore opening i.e., entry into a constitutively active mode.
Altogether, outwardly rectifying as well as ligand/voltagegating properties of zebrafish skeletal muscle ANO1 channels classify them as bona fide voltage-sensitive Ca 2+ -activated Cl − channels, analogous to those expressed in mammalian gastrointestinal epithelia or smooth muscles 40 .
Zebrafish skeletal muscle ANO1 coexists with ClC. Did the evolutionary innovative concept of the Ca 2+ -activated Cl − -channel ANO1 expression in zebrafish skeletal muscle lead to extinction of the solely voltage-gated, canonical skeletal muscle Cl − channel ClC-1 found in mammals 53-55 ? ClC-1 and ClC-2 mRNAs were previously identified in zebrafish skeletal muscle 56 but putative ClC currents remained uncharacterised. To investigate this, we attempted to record ClC currents in zebrafish myotubes using a voltage-clamp protocol previously described for recording ClC-1 currents in mouse FDB fibres 57 . To fully activate ClC currents 57 , normal zebrafish myotubes were stimulated with a 250-ms prepulse from a holding potential of −40 to +60 mV (Fig. 7a) and as expected, we observed robust ANO1 currents (Fig. 7b, left trace). During the 500-ms test pulses to positive test potentials, the expected 'contamination' with ANO1 currents could be minimised by starting the recordings with negative test potentials and hence taking advantage of the pronounced ANO1 current rundown as depicted in Fig. 2d. These recording conditions allowed us to measure a current even at +60 mV in normal myotubes (red trace) which mainly reflects the ClC current (Fig. 7b). However, unquestionably non-contaminated ClC recordings could be harvested only from relaxed myotubes (Fig. 7c), where ANO1 was not activated due to the lack of SR Ca 2+ release (Fig. 1c). Consistent with these findings, we observed a dramatic reduction in peak ClC current density (77% at −140 mV and 66% at +60 mV) in the presence of 1 mM 9AC, a blocker of ClC-1 channels 57 (Supplementary Fig. 5).
As shown in Fig. 7d, the zebrafish ClC current displays a linear I-V relationship with an inward component at potentials below −40 mV and an outward current above −40 mV. Notably, in contrast to ANO1, the ClC current does not show outward rectification. ClC currents at potentials above −10 mV are significantly (P < 0.001) smaller than the corresponding ANO1 currents and hence might have only a minor contribution to the total skeletal muscle Cl − influx during membrane depolarisation.
On the contrary, the ClC inward current component at negative potentials, accounting for 80% of the total membrane conductance in resting human muscle and thus ensuring electrical stability, is a characteristic of skeletal muscle ClC 55 . Lastly, this ClC inward current explains slight Cl − currents that we found in some of our experiments, like in ANO1-current free relaxed myotubes (Fig. 2b, d, right panel) or in normal myotubes at −140 mV without cytosolic Ca 2+ addition (Fig. 6a, lower graph).
Skeletal muscle ANO1 current accelerates AP repolarisation. We were interested why the euteleost species zebrafish evolved such pronounced CaCC currents in skeletal muscle-a phenomenon that has not been reported from mammals. Our prime working hypothesis was that this CaCC current, that immediately follows EC coupling-induced SR Ca 2+ release (Fig. 1a, Supplementary Fig. 4a), plays a role in acceleration of the skeletal muscle action potential (AP) termination. Conventionally, voltage-gated Na + influx and delayed K + efflux contribute to the depolarisation and repolarisation phase of an AP, respectively 58 . However, a number of studies in cardiac and neuronal tissues showed that also CaCC currents are involved in shaping of the AP repolarisation phase [59][60][61][62] . Application of the CaCC blocker 4,4′-diisothiocyanatostilbene-2,2′-disulphonic acid (DIDS) or reduction of the Cl − concentration induced significant broadening of cardiac muscle APs in pig ventricular myocytes 59 as well as in rabbit ventricular and atrial myocytes 60,61 . Broadening of AP width was also observed in hippocampal pyramidal neuronal slices upon ANO2 current reduction, either by applying CaCC blockers niflumic acid (NFA) and 5-nitro-2-(3-phenylpropylamino) benzoic acid (NPPB) or by silencing ANO2 expression with short hairpin RNAs 62 .
Consequently, to investigate a putative role of the zebrafish skeletal muscle ANO1 in shaping of the AP repolarisation phase, we performed whole-cell current-clamp electrophysiology on relaxed and normal myotubes. Interestingly, congruent to our hypothesis, the AP ½ width recorded from relaxed myotubes, which lack the SR Ca 2+ release-activated CaCC current (Figs 1  and 2), was with 2.84 ± 0.28 ms (n = 13) nearly double as wide (P < 0.001) as recorded from normal myotubes with 1.60 ± 0.14 ms (n = 14) (Fig. 8a). However, the AP time to peak was comparable (P > 0.05) between relaxed and normal myotubes (1.28 ± 0.12 ms, n = 13 and 1.28 ± 0.07 ms, n = 14, respectively) (Fig. 8a). Finally, to confirm the results obtained from relaxed myotubes under standard Cl − conditions, we performed current-clamp recordings on normal myotubes under Cl − -free conditions (see Methods). The AP ½ width (2.91 ± 0.57 ms, n = 10) as well as the AP time to peak (1.17 ± 0.05 ms, n = 10) was comparable (P > 0.05) between normal myotubes under Cl − free conditions and relaxed myotubes (Fig. 8b). Thus, both series of experiments strongly indicate that the ANO1 Cl − influx in zebrafish skeletal muscle plays a substantial role in modulating the AP by accelerating the repolarisation phase.
Only ANO1-current accelerated APs enable proper spike trains. Apparently, single APs have only limited importance in skeletal muscle functioning. Sustained muscle contractions are evoked only under conditions when motor neurons emit higher frequency APs, resulting in fusion of contractions (tetanic contraction). Muscle contractions start to sum up beyond stimulation frequencies of 5-20 Hz until the response forms a smooth ramped increase of tetanic contraction, which-depending on species and muscle type-reaches 90% of its maximum fusion rate (complete tetanus) at 20-60 Hz. Increased stimulation frequency leads to increased force production and a maximum force of 90% is generated around 60-80 Hz [28][29][30][31] . The usual firing rate of vertebrate motor neurons during voluntary muscle contraction is within this tetanic range 32 .
To investigate the physiological impact of the skeletal muscle AP width on trains of APs, we performed whole-cell currentclamp recordings in the physiological frequency range from normal and relaxed myotubes. Myotubes were subjected to increasing stimulation frequencies between 35 and 80 Hz, in 5-Hz increments to elicit 500-ms long trains of APs with 100-ms recovery intervals. As shown in Fig. 9, there are significant differences (P < 0.05) in the kinetics of AP trains at all spike frequencies between normal and relaxed myotubes. This is due to piling up (Fig. 9b) of the broader APs in relaxed myotubes (Fig. 9d), which leads to the development of a high depolarised plateau 63 , evidently because the sarcolemma is unable to adequately repolarise between the broadened APs. The formation of these depolarised plateaus can already be observed at lowest firing rates (Fig. 9a, c). On the contrary, depolarised plateaus-as the manifestation of membrane potential derailment in relaxed myotubes-hardly develop in normal myotubes even at the highest tested AP train frequencies (Fig. 9a, c). These results clearly demonstrate the crucial influence of the ANO1 current on the electrical membrane stability.
Overall, our data show that zebrafish skeletal muscle contains at least two distinct Cl − channels, namely ANO1 and ClC. While ClC is most likely crucial for its canonical function of stabilising the sarcolemmal resting potential, it seems not to play a major role in shaping the AP because above −10 mV ClC currents are significantly (P < 0.001) smaller than ANO1 currents (Fig. 7d) and thus are not sufficient to support ANO1 in accelerating the AP repolarisation phase. Furthermore, this is supported by the striking similarity in AP kinetics (similar AP ½ width) between normal myotubes under Cl − free recording conditions (where evidently neither ANO1 nor ClC can conduct Cl − ions) and relaxed myotubes under regular Cl − conditions (where ClC conductance is intact) (Fig. 8b).  Fig. 7 Zebrafish skeletal muscle expresses ClC current beside ANO1 current. a To elicit ClC currents, a pulse protocol from −140 to +60 mV in 10-mV increments, preceded by a 250-ms prepulse to +60 mV to fully activate ClC currents, as described previously 57 was used. b Representative recording from normal myotubes (n = 8) demonstrating ClC currents contaminated by ANO1 currents at positive test potentials. Minimisation of contamination in order to harvest close-to-accurate ClC current recordings at positive potentials was achieved by starting the recordings with negative test potentials and thus taking advantage of the ANO1-specific CamKII-induced current rundown. Red trace indicates the last current recorded at +60 mV. Scale bars, 200 ms (horizontal), 10 pA pF −1 (vertical). c Representative non-contaminated ClC recording from relaxed myotubes (n = 6), lacking the SR Ca 2+ release-activated ANO1 currents. d Overlay of the clean ClC I-V curve from relaxed myotubes (n = 6) and the ANO1 I-V curve from normal myotubes (n = 5) (from Fig. 1d), displays the significant difference (P < 0.001) between ClC and ANO1 (CaCC) Cl − currents, starting from the test potential (V test ) of 0 mV. At +60 mV, ClC displays I = 12.03 ± 3.38 pA pF −1 while ANO1 reaches I = 83.12 ± 7.62 pA pF −1 . Data are presented as mean ± s.e.m.; P determined by unpaired Student's t-test Discussion ANO1, a Ca 2+ -activated Cl − channel (CaCC) has never been reported to functionally express in skeletal muscle. In this study we identified and characterised the expression pattern, subcellular distribution and physiological role of ANO1 in zebrafish skeletal muscle. Immunocytochemical results provide direct evidence for the sarcolemmal localisation of ANO1 and RT-PCR amplification experiments revealed the expression profile of the two ANO1 isoforms, with ANO1-a in superficial slow/red and ANO1-b in deep fast/white skeletal musculature of zebrafish. Using a broad range of experiments, we characterised ANO1 as a bona fide CaCC which is gated synergistically by membrane potential and intracellular Ca 2+ concentration and is activated close to maximum by the SR Ca 2+ release during excitation-contraction coupling. Furthermore, ANO1 coexists with ClC in zebrafish skeletal muscle but only ANO1-current accelerated APs enable proper spike trains crucial for high speed muscle contractions. ClC seems not to play a role in shaping the AP, rather stabilises the sarcolemmal resting potential.
Subsuming these findings, we postulate a model (Fig. 10) where depolarisation-induced SR Ca 2+ release, activated via the DHPR-RyR1 interaction, leads to more or less in-parallel binding of Ca 2+ ions to (i) troponin C for initiation of muscle contraction and (ii) ANO1 for induction of a transient Cl − influx. This Ca 2+ -release-induced Cl − influx, together with the delayed K + efflux via the voltage-gated (K V ) channels, causes a quick drop in the membrane potential (V m ) during the decline phase of the AP, thereby repolarising the sarcolemma and returning the electrochemical gradient to the resting state. Due to the rapid decline of cytosolic Ca 2+ by the fast pumping back action of the sarco/ endoplasmic reticulum Ca 2+ -ATPase (SERCA) 64 , ANO1 is prohibited from Cl − efflux at negative membrane potentials and thus is strictly outwardly rectifying. The resulting short-term APs, due to the ANO1-K V synergistic action, are evidently the basis of electrical membrane stability that enables accelerated muscle stimulation rates for high speed (tetanic) muscle contractions with increased force production [28][29][30][31][32] . Altogether, this is crucial for high speed swimming which is vital in the aquatic praypredator context. Apparently, the phylogenetically highest advanced skeletal muscles of euteleost species 4,5 , like zebrafish, developed this innovative 'high-speed gear' by adding ANO1mediated Ca 2+ release-induced Cl − influx to the skeletal muscle EC coupling machinery.
Future investigations on a double transgenic zebrafish strain that selectively expresses fluorescent proteins mCherry and GFP in slow/red and fast/white muscles, respectively, will enable us to gain deeper insights into this mechanism of AP acceleration by studying the putatively distinct input of the muscle-type specific (Fig. 3) ANO1 isoforms, identified in this study. According to our model, AP acceleration is expected to be more pronounced in the fast/white musculature which is dedicated to burst activities and thus dependent on high frequency activation.

Methods
Zebrafish care. Care and maintenance of adult zebrafish, wild-type (wt) and heterozygous for the DHPRβ 1 -null mutation relaxed (red ts25 ) 27 , obtained from the Max Planck Institute (Tübingen, Germany), was according to the established procedures 65,66 and was approved by the Tierethik-Beirat of the Medical University of Innsbruck and Bundesministerium für Wissenschaft, Forschung und Wirtschaft.
RT-PCR detection. Total RNA was extracted from superficial slow (red) and deep fast (white) skeletal muscle of multiple adult zebrafish individually, using the RNeasy Mini Kit (Qiagen) and reverse-transcribed using random primers and M-MLV reverse transcriptase (Promega). PCR primer sequences for amplification of DNA fragments from the two isoforms of ANO1 and ANO2, as well as from DHPRα 1S -a and α 1S -b subunits, used as positive controls to determine the purity of the muscle tissue preparations 4 , were designed according to the sequences deposited in GenBank database (Supplementary Table 1). Quantification of the band intensities was done using ImageJ (open source). Identity and fidelity of the RT-PCR products was confirmed by sequence analysis (Eurofins, Germany).
Primary culture of myotubes. Myoblasts from 2-dpf relaxed zebrafish, homozygous or heterozygous for the DHPRβ 1 -null mutation 27 , or wt zebrafish were isolated and cultured as described 67 . Homozygous relaxed mutants were identified by their inability to move in response to tactile stimulation and motile 'normal' siblings (heterozygous and wt) were used as controls. Myotubes were cultured for 4-6 days in a humidified 28.5°C incubator in L-15 medium supplemented with 3% foetal calf serum, 3% horse serum, 4 mM L-glutamine and 4 U/ml penicillin/ streptomycin.
Images were recorded with a cooled CCD camera (Diagnostic Instruments) and MetaVue image processing software (v 6.2, Universal Imaging, PA). Quantification of ANO1 surface membrane expression after siRNA knock-down was determined by measuring the average fluorescence intensity along the periphery of CFP positive myotubes, obtained from at least two different cultures using the MetaVue software.
Voltage-clamp electrophysiology. CaCC currents were recorded from cultured myotubes simultaneously with intracellular SR Ca 2+ release by using 0.2 mM Fluo-4 in the patch pipette (internal) solution, and were evoked by a 200-ms pulse protocol from +80 to −50 mV in 10-mV steps from a holding potential of −80 mV, as described 67  To test the integrity of CaCC in the DHPRβ 1 -null relaxed zebrafish, recordings of CaCC currents following caffeine-induced SR Ca 2+ store depletion were performed. From a holding potential of −80 mV, 2-dpf myotubes were voltageclamped at +40 mV for 100 ms followed by a 100-ms step to −120 mV (Fig. 2a). The entire protocol was repeated 100 times and between the 15th and 45th sweep the myotubes were perfused with the standard bath solution supplemented with 8 mM of the RyR agonist caffeine (Merck). Changes in CaCC currents at +40 mV (outward current) and at −120 mV (inward current) before and after application of caffeine were analysed.
To investigate the Ca 2+ -dependence of CaCC currents, intracellular Ca 2+ concentrations were altered via the patch pipette solution. Using the MaxChelator simulation program (http://maxchelator.stanford.edu), pipette solutions for free Ca 2+ concentrations ([Ca 2+ ]) of 0, 2, 5, 7.5 and 13 μM were adjusted (Supplementary Table 2). Due to the unknown kinetics of the CaCC currents under different Ca 2+ concentrations contributed by the patch pipette solution in addition to the normal SR Ca 2+ release, a prolonged 3-s pulse protocol from +80 to −140 mV in 20-mV steps was used.
ClC currents were recorded by applying the voltage-step protocol described by 57 and depicted in Fig. 7a, using standard Cl − -containing external and internal solutions. To confirm that the Cl − current recordings from relaxed myotubes are indeed currents through the skeletal muscle ClC channel, 1 mM 9-anthracene carboxylic acid (9AC), a blocker of ClC-1 channels 57 was added to the external recording solution.
Current-clamp electrophysiology. Action potentials (APs) were recorded from single myotubes in a perforated patch configuration by using 120 µg/ml of amphotericin B in the patch pipette solution under standard Cl − and Cl − -free conditions. The standard Cl − -containing bath solution used was as follows (in mM): 130 NaCl, 4 KCl, 2 MgCl 2 , 2 CaCl 2 , 10 HEPES, and 10 glucose (pH 7.4 with NaOH). For Cl − -free bath solution NaCl, KCl, MgCl 2 and CaCl 2 were replaced by Na-aspartate, K-aspartate, Mg-aspartate and Ca(OH) 2 , respectively. The standard Cl − containing internal solution consisted of (in mM): 135 K-aspartate, 8 NaCl, 2 MgCl 2 , 20 HEPES, and 5 EGTA (pH 7.4 with NaOH). For Cl − free internal solution NaCl and MgCl 2 were replaced by Na-aspartate and Mg-aspartate, respectively. Recordings were initiated after amphotericin B lowered the access resistance below 15 MΩ. A small hyperpolarising current was injected to set the membrane potential to −80 mV and APs were elicited by injecting a current of 3.8 nA for 1 ms. Identical current injections were applied to record 500-ms trains of APs from 35 to 80 Hz, in 5-Hz increments bracketed by 100-ms recovery intervals. Data were acquired with an EPC10 amplifier (HEKA Elektronik, Germany) at a sampling rate of 30 kHz, low-pass-filtered at 3 kHz and analysed using FitMaster (v2x73.2, HEKA Elektronik, Germany) and SigmaPlot 10.0 (Systat Software, Inc.).
General experimental design and statistical analyses. Sample sizes of zebrafish myotubes or tissues are based on previous publications 4,5,27,34,67 , hence, power calculations were not necessary. For siRNA transfection or drug administration assays, myotube cultures according to genotype were arbitrarily allocated to the specific sample groups without the use of an explicit randomisation procedure. Experiments did not require blinding and thus were not performed under blinded conditions. In our study, no data points or samples were excluded from analysis. Based on our previous publications 4,5,27,30,33,34,67 all statistical analyses are considered appropriate. n-values represent the number of independent experiments on zebrafish myotubes or tissues, as specified. Variance is similar between comparison groups. All results are expressed as means ± s.e.m. Statistical significance was determined by using unpaired Student's t-test. P < 0.05 was considered statistically significant and * indicates P < 0.05, **P < 0.01, and ***P < 0.001. Model of ANO1 activation by SR Ca 2+ release as the basis for acceleration of AP repolarisation. Schematic representation of the skeletal muscle triad with sarcolemmal t-tubular invagination (t-tubule) adjacent to the sarcoplasmic reticulum Ca 2+ store (SR) and membrane localisation of some selected channels and pumps involved in cytosolic (Cytosol) Ca 2+ handling. Initially, depolarisation-induced conformational changes in the DHPR are transmitted to the RyR1, which leads to pore opening and release of Ca 2+ ions (blue spheres) from the SR stores. Cytosolic Ca 2+ activates contraction of muscle fibres (m) and also binds to intracellular ANO1 Ca 2+ binding sites 19 , to activate massive Cl − influx (yellow spheres) into the cytosol. Cl − influx together with simultaneous K + efflux (green spheres) via the voltage-gated K + channels (K V ), synergistically and rapidly reduces the membrane potential to accelerate AP repolarisation (inset, green and yellow arrows). Finally, the SERCA pumps back Ca 2+ into the SR and thus resetting the system