A mechano- and heat-gated two-pore domain K⁺ channel controls excitability in adult zebrafish skeletal muscle

Significance

Activity of mechano-gated TRAAK K⁺ channels has been reported in neurons but not in skeletal muscle, yet an archetype of tissue challenged by mechanical stress. We show here that K⁺ channels sharing properties of TRAAK channels are present in adult zebrafish skeletal muscle fibers. The kcnk4b transcript encoding TRAAK channels was cloned and found to be present in adult zebrafish muscles but absent in larvae. Activation of TRAAK channels by heat and stretch in zebrafish muscle led to acceleration of action potentials repolarization suggesting that heat production and membrane deformation associated with muscle activity can control muscle excitability through TRAAK channel activation. TRAAK channels may represent a teleost-specific evolutionary product contributing to improve swimming performance at critical stage of development.

Abstract

TRAAK channels are mechano-gated two-pore-domain K⁺ channels. Up to now, activity of these channels has been reported in neurons but not in skeletal muscle, yet an archetype of tissue challenged by mechanical stress. Using patch clamp methods on isolated skeletal muscle fibers from adult zebrafish, we show here that single channels sharing properties of TRAAK channels, i.e., selective to K⁺ ions, of 56 pS unitary conductance in the presence of 5 mM external K⁺, activated by membrane stretch, heat, arachidonic acid, and internal alkaline pH, are present in enzymatically isolated fast skeletal muscle fibers from adult zebrafish. The kcnk4b transcript encoding for TRAAK channels was cloned and found, concomitantly with activity of mechano-gated K⁺ channels, to be absent in zebrafish fast skeletal muscles at the larval stage but arising around 1 mo of age. The transfer of the kcnk4b gene in HEK cells and in the adult mouse muscle, that do not express functional TRAAK channels, led to expression and activity of mechano-gated K⁺ channels displaying properties comparable to native zebrafish TRAAK channels. In whole-cell voltage-clamp and current-clamp conditions, membrane stretch and heat led to activation of macroscopic K⁺ currents and to acceleration of the repolarization phase of action potentials respectively, suggesting that heat production and membrane deformation associated with skeletal muscle activity can control muscle excitability through TRAAK channel activation. TRAAK channels may represent a teleost-specific evolutionary product contributing to improve swimming performance for escaping predators and capturing prey at a critical stage of development.

The resting muscle K⁺ conductance plays the pivotal role of setting the resting skeletal muscle membrane potential at a value close to the K⁺ equilibrium potential (1). Two major types of K⁺ channels are involved in the establishment and maintenance of the resting membrane potential, the inward rectifier K⁺ channel (Kir) and the two-pore domain K⁺ channel (K_2P) (2). Whereas there is compelling evidence for the functionality of Kir channels in fully differentiated vertebrate skeletal muscle fibers, K_2P channel activity has never been described in this cell type. By contrast, K_2P channels have been extensively studied in heterologous expression systems. The K_2P channels family is composed of six subfamilies, TWIK, TREK/TRAAK, TASK, TALK, THIK, and TRESK, encoded by KCNK genes, that all produce instantaneous and time-independent background K⁺ currents (3, 4). However, a number of physicochemical changes have been shown to modulate K_2P channel opening, so that K_2P channels are now considered to be implicated in various physiological processes (5). Stretch and/or deformation of the membrane specifically activate the subfamily of TREK/TRAAK channels. Obviously, mechano-sensitivity should confer to TREK/TRAAK channels a potentially important role in contractile cells, and stretch-activated K⁺ currents sharing comparable properties to K⁺ currents flowing through TREK1 channels in heterologous expression systems have indeed been recorded in cardiac and in smooth muscle cells (6, 7). But, in vertebrate skeletal muscle fibers, specifically in mouse skeletal muscle fibers, mechano-sensitivity of ion channels has only been revealed at the single-channel level for Ca²⁺ channels at resting membrane potential and for high conductance Ca²⁺-activated K⁺ channels at depolarized membrane potentials (8–10), whereas no functional K_2P channels have been described.

Zebrafish has become a leading model for gaining insight into the genetic, histological, and molecular aspects of muscle development and pathophysiology, but electrophysiological approaches at the cellular level are very sparse or even lacking. Moreover, studies have been almost exclusively performed at the first stage of the larval period (11–13). Yet, it is known that zebrafish undergoes metamorphosis and changes in the musculature at 3 to 4 wk of age during transition from larval to juvenile stage (14), suggesting that important changes in the functional properties of muscle fibers may occur during this transition. Furthermore, a number of muscle pathologies worsen with age or come up at ages far beyond embryonic stages, so that the use of zebrafish larvae to model those kinds of muscle disorders is limited. With the objective to improve and optimize the use of the zebrafish model, we have recently been able to implement the voltage- and current-clamp techniques on isolated zebrafish skeletal muscle fiber to initiate an exhaustive characterization of the electrophysiological properties and excitation–contraction coupling of fast skeletal muscle fibers from zebrafish at fully developed stages (15). In the present study, we reveal the presence of a functional TRAAK K_2P channel in fast skeletal muscle fibers from adult zebrafish that contributes to increase the resting muscle conductance and to speed the repolarization phase of action potentials (APs) when activated by membrane stretch and heat. We also cloned the kcnk4b gene and show that the transfer of zebrafish TRAAK channel cDNA in adult mouse skeletal muscle fibers, that do not express functional TRAAK channels, gave rise to activity of mechano-gated channels displaying main TRAAK channel properties. Finally, we demonstrate that this channel is absent in larvae but arises at 1 mo of age of fish. Our work thus provides experimental evidence of the functionality of a K_2P channel in vertebrate skeletal muscle with regulation properties that confer to this channel a pivotal role in skeletal muscle excitability and contractility and contributes to improve swimming performance of zebrafish at a critical stage of its development.

Results

Single-Channel Recordings in Cell-Attached Membrane Patches of Adult Zebrafish Skeletal Muscle Fibers Reveal the Presence of a K⁺-Selective Mechano- and Heat-Gated Channel.

A first series of experiments was carried out to investigate single-channel activity in cell-attached patches from isolated adult zebrafish skeletal muscle fibers bathed in a solution containing 140 mM K⁺ to zero the fiber internal potential. Depolarizing voltage pulses applied to membrane patches from a holding potential of −90 mV induced opening of outward unitary currents whose amplitude and opening increased with depolarization and inactivated within a few seconds on maintaining depolarization (SI Appendix, Fig. S1A). Channels responsible for this activity were unambiguously identified as delayed rectifier K⁺ channels, similar to those recorded in mammalian skeletal muscle fibers in the same experimental conditions, although displaying apparent lower unitary conductance (16, 17). When maintained at 0 mV, cell-attached patches displayed extremely low single-channel activity, but applying negative pressure in the patch pipette instantaneously and reversibly induced an opening burst of channels carrying outward currents (Fig. 1A). This stretch-induced opening burst was observed in all the 214 cell-attached patches tested in these experimental conditions throughout this study. Fig. 1B shows that application of calibrated negative pressure of increasing values in the patch pipette augmented channel activity in a dose-dependent manner until loosing patch sealing before reaching saturation. A 4 kPa was then applied in cell-attached patches at different holding membrane potentials and in the presence of either 5 mM or 140 mM K⁺ at the external face of the membrane patch (in the pipette) to measure the single-channel conductance and the ion selectivity of the stretch-activated channel. In the presence of 5 mM external K⁺, amplitude of stretch-induced unitary currents increased with depolarization and reversed at an extrapolated voltage around −80 mV, as expected for K⁺ selective channels considering a theoretical K⁺ equilibrium potential of −85 mV (Fig. 1C). The current–voltage relationship exhibited an outward rectification and the mean unitary current at 0 mV was 4.8 ± 0.2 pA, yielding a mean unitary conductance of 56 ± 2 pS. In the presence of 140 mM external K⁺, stretch-induced unitary currents reversed at a mean value of 0.6 ± 1.8 mV, close to the expected value of the K⁺ equilibrium potential in these experimental conditions (Fig. 1D). The current–voltage relationship was found to be linear, and the mean chord unitary conductance was 223 ± 11 pS (n = 7). Taken together, these data indicate that the stretch-activated channels are K⁺ selective and are predicted to follow the Goldman–Hodgkin–Katz current equation for a K⁺-selective leak current. Fig. 1 C and D show that open probability of the stretch-activated K⁺ channel increased with depolarization for a fixed negative pressure of 4 kPa. Measurement of channel activity at different holding potentials in seven cell-attached membrane patches indeed indicated that mean channel activity was gradually augmented up to saturation by depolarizations above +30 mV (Fig. 1E).

Fig. 1.

Unitary conductance, voltage, and heat sensitivity of mechano-gated K⁺ channels in cell-attached patches of adult zebrafish skeletal muscle fibers. (A) Activation of single-channel outward currents at 0 mV in response to negative pressure applied in the patch pipette. (B) Increase in channel opening in response to increasing negative pressures in a same patch held at 0 mV (Left). Relationship between the mean relative channel activity and negative pressure applied in the pipette in patches held at 0 mV (n = 7). (C) Single-channel currents through mechano-gated channels at the indicated membrane potentials (Upper) and relationship between mean unitary current amplitudes and membrane potential (Lower) in the presence of 5 mM K⁺ in the pipette. (D) Same as in C but in the presence of 140 mM K⁺ in the pipette. (E) Relationship between the mean relative channel activity and membrane potential with continuous application of −4 kPa in the pipette. (F) Increase in channel activity in response to exchange of a 20 °C to a 37 °C external solution.

Taken together, these data suggest that the stretch-activated K⁺ channel recorded in cell-attached membrane patches of zebrafish muscle fibers belongs to the class of the mechano-gated TRAAK or TREK channels of the K_2P channel family. To confirm this, we investigated the effects of heat and arachidonic acid (AA) exposure, two known activators of TRAAK and TREK channels. Fig. 1F shows that heating to 37 °C a cell-attached membrane patch held at 0 mV elicited an increase in opening of channels previously demonstrated to be stretch-activated in the same membrane patch by applying negative pressure. Measuring the unitary current at 0 mV (4.8 ± 0.3 pA) in the 5 patches tested confirmed that the heat-activated channel corresponded to the stretch-activated K⁺ channel. In contrast, we never succeeded in activating stretch-gated channels in cell-attached patches exposed to an extracellular solution containing AA at concentrations ranging from 20 to 100 µM. However, we suspected that the absence of the effect of AA may be due to the fact that, in this patch clamp configuration, the fatty acid did not reach the close environment of the channel or because channel opening was repressed by an intracellular component. A next series of experiments were then performed using the inside-out configuration to further characterize the channel properties by gaining direct access to the cytoplasmic face of the channel.

Single Mechano-Gated K⁺ Channels Display Functional Properties of TRAAK Channels in Inside-Out Patches.

Membrane patches held at 0 mV displayed a high activity of channels carrying outward currents when they were excised and their cytoplasmic face was exposed to an ATP- and Ca-free 140 mM K⁺-containing solution. SI Appendix, Fig. S1B shows that this channel activity was completely and reversibly abolished upon addition of 1 mM ATP to the cytoplasmic face of these inside-out patches. Another activity of single channels carrying outward currents of larger unitary amplitude was observed in inside-out membrane patches upon exposition of the cytoplasmic face to a solution containing 100 µM Ca²⁺ and was reversibly suppressed when Ca²⁺ was omitted (SI Appendix, Fig. S1C). Based on their unitary conductance and regulation of their activity, these two channels were unequivocally identified as ATP-dependent K⁺ and high conductance Ca²⁺-activated K⁺ channels respectively, again comparable to those described in mammalian skeletal muscle fibers (9, 18, 19).

In order to focus our inside-out recordings on activity of the mechano-gated channels precedingly recorded in cell-attached patches, inside-out membrane patches held at 0 mV were excised from isolated zebrafish muscle fibers in the presence of 5 mM K⁺ in the pipette and a Ca-free solution containing 140 mM K⁺, and 1 mM ATP was superfused to the cytoplasmic face of the channel in order to block ATP-dependent K⁺ channels and high conductance Ca²⁺-activated K⁺ channels. As observed in cell-attached membrane patches, channel activity was low in these inside-out patches, but applying negative pressure in the pipette induced openings of channels carrying outward currents of 5.0 ± 0.2 pA (n = 20) (Fig. 2A). Equal concentrations of Cl⁻ being present on both sides of the membrane patch, stretch-activated unitary currents recorded in this configuration could be only carried by K⁺ ions. Given the extracellular (5 mM) and intracellular (140 mM) K⁺ concentrations, such an amplitude of single current yielded a unitary conductance of 60 ± 2.3 pS at 0 mV, thus indicating that the stretch-gated K⁺ channel recorded in these inside-out membrane patches was the same as the one recorded in cell-attached membrane patches. More importantly, in all the inside-out membrane patches tested, exposition of the cytoplasmic face to 20 µM AA induced a progressive and marked opening of these 60 pS stretch-gated K⁺ channels (Fig. 2B), confirming that excision of the membrane patch allows to reveal AA activation likely by disrupting cytoskeletal elements exerting a repressive effect on channel activity (see below AA effect on macroscopic current). Heating of the membrane patch to 37 °C also gave rise to activation of the same 60 pS stretch-gated K⁺ channel (Fig. 2C). TRAAK and TREK channels are two mechano-gated, AA- and heat-sensitive K_2P channels, but TREK is also activated by intracellular acidic pH, whereas TRAAK is activated by alkaline pH (20, 21). Fig. 2D shows that acidification to pH 5 of the solution facing the cytoplasmic side of inside-out membrane patches did not induce opening of the mechano-gated channel (n = 5), while alkalization gave rise to a reversible increase in channel activity (n = 6) (Fig. 2E). These data led us to conclude that the channel activated by membrane stretch in adult zebrafish skeletal muscle fibers shares all the properties of the TRAAK K_2P channel.

Fig. 2.

Effects of stretch, AA, heat, and pH changes on mechano-gated K⁺ channels in inside-out patches of adult zebrafish skeletal muscle fibers. (A) Activation of single-channel outward currents in response to negative pressure applied in the patch pipette. (B) Increase in channel activity in response to exposition of the cytoplasmic face to AA. (C) Increase in channel activity in response to exchange of a 20 °C to a 37 °C superfused internal solution. (D) Absence of effect of acidification of the superfused internal solution on channel activated by applying negative pressure in the pipette (three large deflections at the beginning of the current trace). (E) Increase in channel activity in response to alkalization of the superfused internal solution. The horizontal bars indicate the period during which negative pressure, AA, 37 °C solution, and pH changes are applied. The patch membrane was held at 0 mV. Insets in B–D show the current trace at the level of *on an expanded scale.

Functional TRAAK Channels Encoded by the kcnk4b Gene Are Present in Adult Muscle but Absent from Larval Muscle.

Surprisingly, applying negative pressure in patch pipettes sealed on 13, 5, and 9 isolated muscle fibers from 5-d, 1-wk, and 2-wk-old zebrafish respectively failed to induce activation of K⁺ channels. At 1 mo of age, membrane stretch did elicit opening of K⁺ channels (in all the 12 cell-attached patches tested) exhibiting the same characteristics as those recorded in adult zebrafish fibers, suggesting that a transition exists around 3 wk of age, mechano-gated channels being absent or not functional in sarcolemma of animals younger than 3 wk but present and functional in older animals. In silico search for zebrafish (Danio rerio) sequences homologous to the human gene encoding the TRAAK channel in genome databases revealed that, whereas human TRAAK protein is encoded by a single gene, KCNK4, two paralogs exist in zebrafish, kcnk4a and kcnk4b. Alignment of human and zebrafish kcnk4 deduced amino acid sequences indicated that the overall primary structure of the TRAAK proteins is conserved in the zebrafish proteins such as the two pore channel domains (SI Appendix, Fig. S2). Relative quantification of kcnk4a and kcnk4b expression levels in trunk muscle using qRT-PCR showed that kcnk4a was not expressed in trunk muscle. We thus assumed that the TRAAK channel recorded in juvenile and adult zebrafish muscle fibers was encoded by kcnk4b. In agreement with our electrophysiological data, we further showed that kcnk4b expression level was very low in 1-wk-old fish muscle but substantially increased after metamorphosis in 1-mo and even more in 1-y-old fish (Fig. 3A).

Fig. 3.

Expression of the kcnk4b gene in trunk muscle of developing zebrafish and channel activity of kcnk4b heterologously expressed in HEK-293 cells and in mouse skeletal muscle fibers. (A) Relative quantification of kcnk4b gene by real-time RT-PCR analysis using RNA extracts of 1-wk postfertilization trunks (n = 30 for each experiment), skeletal muscle isolated from 1-mo postfertilization trunks (n = 4 for each experiment), and 1-y old adult fish trunks (n = 1 for each experiment). Three independent experiments were performed. Data were normalized to polr2d. Expression levels were compared to kcnk4b gene expression at 1 wpf, which was arbitrarily set to 1. (B) Confocal image of HEK cells expressing TRAAK-GFP proteins. Arrows indicate the TRAAK-GFP-positive membrane regions on which patch pipettes were sealed. (C) Burst of single-channel outward currents evoked at 0 mV in response to negative pressure applied in a patch pipette sealed on a TRAAK-GFP-positive membrane region of a HEK cell membrane. (D) Recording of single-channel activity elicited by negative pressure in HEK cell-attached patches held at different membrane potentials. (E) Relationship between mean stretch-activated unitary current amplitudes in HEK cell–attached patches and membrane potential (n = 11). (F) Confocal image of a mouse muscle fiber expressing TRAAK-GFP protein after electroporation of kcnk4b-EGFP plasmid. (G) Burst of single-channel outward currents evoked at 0 mV in response to a brief negative pressure applied in a patch pipette sealed on the GFP-positive region of a mouse muscle cell. (H) Recording of single-channel activity elicited by negative pressure in mouse muscle cell–attached patches held at different membrane potentials. (I) Relationship between mean stretch-activated unitary current amplitudes in mouse muscle cell-attached patches and membrane potential (n = 14). For all cell-attached experiments, Tyrode and K⁺-rich solution were present in the pipette and in the bath, respectively.

kcnk4b Expression in HEK Cells and in Adult Mouse Muscle Fibers Gives Rise to Activity of Mechano-Gated K⁺ Channels Displaying TRAAK Channel Characteristics.

In order to confirm that kcnk4b actually encodes the mechano-gated K⁺ channel that we recorded in adult zebrafish muscle, we next decided to express the zebrafish kcnk4b cDNA in HEK cells. With this aim, we cloned and sequenced the full-length cDNA of the zebrafish kcnk4b gene and a kcnk4b-GFP construct was prepared to transfect the cells. Confocal fluorescence images of transfected cells indicated that the TRAAK-GFP protein is expressed at the plasma membrane (Fig. 3B). Patch pipettes were then sealed on the fluorescent region of the plasma membrane and, in the presence of Tyrode in the pipette and a 140 mM K-containing solution in the bath, application of negative pressure in the pipette gave rise to reversible activation of channels carrying outward currents at 0 mV (Fig. 3C), whereas such an activity was never detected in nonfluorescent cells. The mean unitary conductance of these channels at 0 mV was 50 ± 2 pS (n = 11) (Fig. 3 D and E), not significantly different from the mean unitary conductance of mechano-gated channels measured in cell-attached membrane patches from adult zebrafish muscle fibers (56 ± 2 pS) using the same experimental conditions (P = 0.09, unpaired t test).

Mouse skeletal muscles were also electroporated with the kcnk4b-GFP construct. Electroporated muscle fibers exhibited a peripherical and striated expression pattern compatible with expression of the TRAAK-GFP protein at the plasma membrane and in the transverse tubules (Fig. 3F). Application of negative pressure in patch pipettes sealed on fluorescent regions of the fibers gave also rise to activation of single K⁺ channels currents displaying a mean unitary conductance of 57 ± 2 pS (n = 13) at 0 mV, not significantly different from the mean unitary conductance of mechano-gated channels measured in cell-attached membrane patches from adult zebrafish muscle fibers using the same experimental conditions (P = 0.1, unpaired t test) (Fig. 3G), whereas channel opening was never detected in TRAAK-GFP negative zones just as in nonelectroporated cells (Fig. 3 G–I). Additionally, we found that the mechano-gated K⁺ channel recorded in inside-out patches from TRAAK-GFP positive regions of mouse muscle fibers was activated by alkaline pH and AA (SI Appendix, Fig. S3).

Taken together, these data further confirm that kcnk4b, showing a specific temporal expression profile in zebrafish muscle, does encode the mechano-gated TRAAK channel recorded in adult zebrafish muscle and establish its potency to generate mechano-sensitive K⁺ channel activity in mouse muscle.

AA, Heat, and Membrane Stretch Activate K⁺ Currents at the Macroscopic Level.

In order to investigate the potential role of the TRAAK channel in zebrafish skeletal muscle excitability and contractility, we implemented the whole-cell voltage- and current-clamp techniques on isolated adult zebrafish muscle fibers (15). We first tested the effects of AA exposure, heat, and stretch on membrane currents recorded at different voltages under whole-cell voltage-clamp conditions. In the 49 fibers tested, addition of AA either in the extracellular solution or in the intracellular milieu at concentrations ranging from 20 to 100 µM failed to induce any change in outward currents evoked by depolarization pulses from −40 to +40 mV. As put forward to explain the absence of effect of AA in cell -attached patches, we hypothesized that the presence of an intracellular component prevented AA added in the extracellular medium and even in the intracellular solution to exert its activating effect on TRAAK channels. Maingret et al. (22) reported that TRAAK is tonically repressed by the cytoskeleton since disruption of the cytoskeleton by colchicine was shown to potentiate TRAAK opening by membrane stretch. Accordingly, in order to reveal TRAAK activation by AA in whole-cell conditions, prior to recordings, muscle fibers were bathed in a Tyrode solution containing 500 µM colchicine during 20 min and subsequently fibers were dialyzed with an internal solution also containing 500 µM colchicine. Additionally, to focus recordings on background K⁺ currents, fibers were bathed in a Na⁺-free, low-Cl⁻ solution containing the Cl⁻ channel inhibitor 9-AC. Fig. 4A shows that in these experimental conditions, AA added in the extracellular solution at 20 µM induced a marked potentiation of outward currents elicited by depolarizing pulses of increasing amplitudes from a holding potential of −80 mV. The mean current recorded at +40 mV in the presence of AA was significantly larger than in its absence (P = 0.0002). Measurement of the change in current at −80 mV induced by a short −10 mV voltage pulse also indicated that the membrane conductance was significantly larger at −80 mV in the presence of AA (264 ± 41 S/F) than in its absence (216 ± 31 S/F) (P = 0.019). In all cells tested, subtracting the control current to the current recorded in the presence of AA at each potential yielded an AA-induced current that displayed an outward rectification and that nullified close to the theoretical value of the K⁺ equilibrium potential (−87 mV). A significant AA-induced potentiation of K⁺ currents was also observed in the same cells when pulses were given from a holding potential of −40 mV that produced inactivation of voltage-gated K⁺ channels (P = 0.013 for comparison of current amplitudes at +40 mV) (Fig. 4B). These data indicate that AA specifically activates noninactivating outwardly rectifying K⁺ channels in zebrafish muscle fibers of which cytoskeleton is disrupted by colchicine.

Fig. 4.

Effects of AA on whole-cell K⁺ currents in adult zebrafish skeletal muscle fibers. (A) Increase in outward K⁺ currents (Upper traces) elicited by depolarizing pulses of increasing amplitudes (Lower traces) from a holding potential of −80 mV in response to AA exposure. (B) Same as in A in the same fiber from a holding potential of −40 mV. (C) Relationship between the mean current difference (end pulse current in the presence of AA minus end pulse control current) and voltage in eight fibers depolarized from a holding potential of −80 mV. (D) Same as in C for the same fibers depolarized from a holding potential of −40 mV.

Fig. 5 shows that increasing the temperature of the extracellular solution from 20 to 37 °C also led to an increase of the outward currents elicited by depolarizing pulses of increasing amplitudes from a holding potential of −80 mV as well as from a holding potential of −40 mV. For both holding potentials, the current recorded at +40 mV was significantly larger at 37 °C than at 20 °C (P = 0.03), and the current difference corresponding to the heat-activated current displayed a slight outward rectification and reversed at a voltage close to the K⁺ equilibrium potential. Measurement of the change in current at −80 mV induced by a short 10 mV-hyperpolarizing pulse also indicated that the membrane conductance was significantly larger at −80 mV at 37 °C (243 ± 26 S/F) than at 20 °C (182 ± 19 S/F) (P = 0.0002).

Fig. 5.

Effects of heat on whole-cell K⁺ currents in voltage-clamped adult zebrafish skeletal muscle fibers. (A) Increase in outward K⁺ currents (Upper traces) elicited by depolarizing pulses of increasing amplitudes (Lower traces) from a holding potential of −80 mV in response to heat. (B) Same as in A in the same fiber from a holding potential of −40 mV. (C) Relationship between the mean current difference (end pulse current at 37 °C minus end pulse current at 20 °C) and voltage in eight fibers depolarized from a holding potential of −80 mV. (D) Same as in C for the same fibers depolarized from a holding potential of −40 mV.

Finally, in order to explore the activating effect of membrane stretching on TRAAK channels at the macroscopic level, a fire-polished glass rod displaced with a micromanipulator was pressed against the voltage-clamped portion of the cell emerging from the silicone grease (23). The experiments were performed in the presence of the same external and internal solutions as the ones used for AA experiments. Fibers were maintained at −80 mV and, every 2 s, changes in background membrane conductance were measured by applying, not pulses, but voltage ramps of 0.5-s duration bringing the internal potential from −80 to +40 mV since the effects of cell compression were found to be transient. Fig. 6 A and B show that cell compression induced a positive increase of the current generated by the voltage ramp. The mean outward current recorded at +40 mV in response to cell compression was significantly larger than the one recorded without compression (P = 0.0025). Subtracting the current induced by the voltage ramp before cell compression to the current recorded during cell compression in seven fibers yielded a mean outward current which nullified at a value close to the K⁺ equilibrium potential (Fig. 6C).

Fig. 6.

Effects of cell compression on whole-cell K⁺ currents in voltage-clamped adult zebrafish skeletal muscle fibers. (A) Increase in outward K⁺ currents (Upper traces) elicited by successive depolarizing ramps (Lower traces) bringing the membrane potential from −80 mV to +40 mV in 0.5 s every 2 s, preceded by a 50-ms pulse to −90 mV. The horizontal bar indicates the period during which cell was compressed. (B) Currents (Upper traces) before and during cell compression evoked by voltage ramps (Lower trace). Currents correspond to currents recorded at the level of * and ** in A on an expanded scale. (C) Relationship between the mean current difference (current during minus current before cell compression) and voltage (n = 6).

Taken together, these results strongly suggest that AA, heat, and stretch activate K⁺ channels at the whole-cell level sharing the properties of TRAAK channels.

Heat and Membrane Stretch Increase Resting Membrane Conductance and Accelerate AP Repolarization under Physiological Conditions of Excitation.

A last series of experiments were performed to assess the consequences of TRAAK channel activation on muscle fiber excitability under physiological conditions of excitation. To this end, fibers were current clamped, and trains of APs were elicited at 50 Hz during 500 ms from a resting membrane voltage of −100 mV by current injection in order to maximize voltage-gated Na⁺ channel opening and AP overshoot (15). Fig. 7 presents the changes induced by warming fibers from 20 to 37 °C. Warming to 37 °C led to depolarization of the resting membrane potential which on average got closer to the K⁺ equilibrium potential and produced an increase in membrane conductance revealed by a decrease in the voltage change induced by a 2 nA negative current injection (n = 7, P = 0.0062) (Fig. 7B). Warming also induced a reduced AP amplitude (P = 0.0026) and a significant reduced elapsed time between AP peak and 50 % repolarization (P = 0.0002) as well as 75% repolarization (P < 0.0001) (Fig. 7C). All these changes induced by warming are compatible with the opening of K⁺ channels at rest and during AP.

Fig. 7.

Effects of heat on AP and resting membrane potential in current-clamped adult zebrafish skeletal muscle fibers. (A) Changes of AP induced by warming of the external solution. The two first and the two last AP in a 0.5-s 50-Hz train of APs are presented at 20 and 37 °C in the same fiber after re-establishing the resting potential at −100 mV at 37 °C. (B) Depolarization of the resting membrane potential and reduction in the voltage change induced by a 50-ms current step of −2 nA at 20° C and 37 °C in the same fiber before re-establishing the resting potential at −100 mV at 37 °C (Left). Histogram showing mean voltage change induced by negative current step (Left axis) and mean resting membrane potential (Right axis) at 20 °C and 37 °C in seven fibers (Right). (C) Superimposition of the first AP of the train shown in A at 20 °C and 37 °C on an expanded scale (Left). Histogram showing mean change in AP amplitude and mean elapsed time between AP peak and 50% or 75% repolarization induced by heat in the same fibers (n = 7) (Right).

Fig. 8 shows the effects of cell compression on resting membrane potential, conductance, and AP in the same current-clamp conditions as the one used for testing the effects of warming. As observed with a change in temperature, cell compression evoked a depolarization of the resting membrane potential which on average got closer to the K⁺ equilibrium potential and an increase in membrane conductance (P = 0.0009) (Fig. 8B), a reduction of AP amplitude (P = 0.0083) and reduced elapsed time between AP peak and 50% repolarization (P = 0.016) as well as 75 % repolarization (P = 0.0002) (Fig. 8C). All these observed changes in resting voltage and AP likely suggest that the stretch of the sarcolemma induced by cell compression provokes opening of K⁺ channels.

Fig. 8.

Effects of cell compression on AP and resting potential in current-clamped adult zebrafish skeletal muscle fibers. (A) Changes of AP induced by cell compression. The two first and the two last AP in a 0.5-s 50-Hz train of AP are presented before and during cell compression in the same fiber after re-establishing the resting potential at −100 mV during cell compression. (B) Depolarization of the resting membrane potential and reduction in the voltage change induced by a 50-ms current step of −2 nA before and during cell compression in the same fiber before re-establishing the resting potential at −100 mV during cell compression (Left). Histogram showing mean voltage change induced by the negative current step (Left axis) and mean resting membrane potential (Right axis) before and during cell compression in six fibers (Right). (C) Superimposition of the first AP of the train shown in A before and during cell compression on an expanded scale (Left). Histogram showing mean change in AP amplitude and mean elapsed time between AP peak and 50% or 75% repolarization induced by cell compression in the same fibers (n = 6) (Right).

Discussion

Our study demonstrates the presence of a functional mechano- and heat-gated TRAAK channel in vertebrate skeletal muscle. By exploring in detail its biophysical and regulation properties at the unitary as well as at the macroscopic level, we show that this mechano-gated K⁺ channel shares all the properties exhibited by the TRAAK isoform of the K₂P channel family studied in heterologous expression systems, in terms of unitary conductance, activation by membrane stretch, heat, AA, and internal alkalization, and slight voltage dependence (21, 22).

At the molecular level, the search for TRAAK transcripts in adult zebrafish skeletal muscle led us to reveal the presence of the kcnk4b isoform in adult fast muscle fibers, the muscle fiber type on which we recorded TRAAK channel activity. Cloning and transfer of the zebrafish gene encoding TRAAK channels led to expression of a fully functional mechano-gated K⁺ channel sharing comparable properties of channels recorded in adult zebrafish muscle, in HEK cells, but overall in mouse muscle fibers which do not express native mechano-gated K⁺ channels. These data firmly attest that kcnk4b encodes the mechano-gated K⁺ channel that we recorded in zebrafish muscle and that TRAAK channels can be fully functional in a native muscle environment. In agreement with our electrophysiological data showing that membrane stretch failed to induce activation of the K⁺ channel in zebrafish larvae muscle, this kcnk4b isoform was found to be absent in whole larvae muscle but arose between 3 wk and 1 mo of age, which is coincidental with a major zebrafish developmental transition (14). In that context, emergence of TRAAK channels during development should represent an important step contributing to maturation of skeletal muscle excitation properties.

Considering the regulating parameters of TRAAK channel activity, specifically stretch and heating, we foresaw that the TRAAK channel could play an important role in skeletal muscle excitability and contractility. We first showed in current–clamp conditions that warming zebrafish muscle fibers induced a significant acceleration of the repolarization phase of APs, a reduced AP amplitude, an increase in the resting membrane conductance, and change in resting membrane potential compatible with activation of K⁺ channels corresponding to TRAAK channels. A rise in temperature is of course known to affect other ion channels and in particular the voltage-gated channels. However, the fact that in voltage-clamp conditions, in the presence of K⁺ as the unique permeant ion, warming increased the muscle resting membrane conductance and activated an outward rectifying noninactivating K⁺ current strongly suggests that TRAAK channel activation contributes at least in part to the observed changes of APs and resting membrane conductance in response to warming. We also showed that membrane stretch induced by cell compression activated a K⁺-selective current exhibiting a slight outward rectification, likely flowing through TRAAK channels. More importantly, we were able to compress fibers during trains of AP in current-clamp conditions and to show that changes in resting voltage and conductance and AP induced by membrane stretch are also compatible with activation of K⁺ channels corresponding to TRAAK channels.

This contribution of the mechano- and heat-activated TRAAK channel to muscle fiber excitability has important physiological relevance. First, whatever the animal species, muscle activity is indeed associated with heat production by muscle fibers (24). Up to now, it was recognized that elevated muscle temperature improves muscle performance by increasing the rate of myofibrils cross-bridge cycling and muscle fiber conduction velocity essentially through acceleration of opening and closing of voltage-gated Na⁺ channels (25, 26). Our study states that TRAAK channels now stand as muscle temperature sensors that contribute to further accelerate the repolarization phase of APs in response to an elevation of muscle temperature.

In addition, during muscle activity, the skeletal muscle sarcolemma sustains vigorous deformations associated with cycles of contraction–relaxation of the muscle fiber. At rest, and even during activity, muscle can be also subject to stretch. Although the magnitude of tensions the membrane undergoes in an active or in a resting fiber is poorly documented, the membrane may be stretched to the point of activation of mechano-gated TRAAK channels. To our knowledge, a change in skeletal muscle excitability provoked by increasing muscle length has never been demonstrated. Yet, stretching is known for long to enhance force production in active muscle (27). However, this so-called residual force enhancement process is thought to represent purely sarcomeric properties. Our study demonstrates that TRAAK channels’ activation adds upstream to this potentiating effect by accelerating the repolarization phase of AP. Moreover, specifically in fish, alternate contraction of the myotomes on each side of the body during fast swimming may increase the open probability of TRAAK channels by stretching muscles of the relaxed side, hyperpolarizing the cells, and thereby reducing Na⁺ channel inactivation to ensure a large Na⁺ inward current upon depolarization.

Whether it be heat or stretch, acceleration of the repolarization phase of AP in fast muscle should contribute not only to increase conduction velocity of excitation of muscle fibers but also muscle force. Indeed, especially during escape responses of zebrafish which are powered by fast muscle fibers, acceleration of the repolarization phase of AP allows reaching higher frequency of AP and fusion of contraction, and higher muscle force (15, 28). Despite the voltage-dependence of TRAAK channels revealed in our single-channel recording experiments, our current-clamp experiments clearly show that compression and heat open K⁺ channels at resting membrane potential, suggesting a role of TRAAK channels not only during muscle activity but also at resting membrane potential. It indeed can be postulated that TRAAK channels overactivated by extreme hot temperature or in response to deleterious muscle stretching play a protective role by hyperpolarizing the cell and increasing the resting membrane conductance to the point of reducing excitability and forcing the cell to rest.

From a phylogenetical point of view, the euteleost zebrafish is more advanced than mammals (29). It can thus be postulated that this animal species developed this mechano- and heat-gated channel as an originative bigger gear to improve muscle performance as proposed for the Ca²⁺-activated Cl^- channel ANO1 (30). Our study also highlights the necessity to functionally phenotype zebrafish in fully developed animals and to not restrict investigations to larvae stages as done in the vast majority of studies.

Materials and Methods

Experimental Animals and Ethics Statement.

AB/TU wild-type zebrafish were raised and bred according to standard procedures (31) and housed at the fish facility (PRECI, SFR Biosciences UAR3444/CNRS, US8/Inserm, ENS de Lyon, UCBL). All animal manipulations were performed in agreement with EU Directive 2010/63/EU. The developmental stages are given in weeks (wpf) and months (mpf) postfertilization at 28 °C, according to morphological criteria (14).

The experimental protocol for in vivo DNA electrotransfer of mouse hind-limb muscles was performed using 8- to 16-wk-old Swiss OF1 male mice (Charles-River) weighing 30 to 45 g. The procedure was approved by the Animal Experimentation Committee # C2EA-55 of the Rhône-Alpes Region and by the French Ministry for Higher Education, Research and Innovation.

Sequence Alignments.

Danio rerio sequences of kcnk4a and kcnk4b were retrieved from Ensembl database: ENSDARP00000136028.2 and NCBI Reference Sequence XP_017208343.1, respectively. The KCNK4/TRAAK human sequence was retrieved from Genbank (accession number: AAF64062.1). Alignments of the deduced amino-acid sequences of kcnk4a and kcnk4b with the human sequence were performed using ClustalX 2.1 (32) using a linux system (Debian “Bullseye” using the package clustalx_2.1+lgpl-9_amd64.deb). The software was used with the default configuration for protein in multiple alignment mode (Residue-specific Penalties: On; Hydrophilic Penalties: On; Hydrophilic Residues: GPSNDQEKR; Gap Separation Distance: 4; End Gap Separation: Off). The multiple alignments parameters were: Gap Opening: 10; Gap extension: 0.2; The Protein Matrix Weight was Gonnet.

RNA Extraction.

Fish at 1 wpf (n = 30 per experiment), 1 mpf (n = 4 per experiment), and adults (1-y old fish, n = 1 per experiment) were terminally anesthetized by incubation in MS-222 (0.2% tricaine buffered at pH 7.0) until total arrest of opercular movements. One-mpf fish were staged according to Parichy et al. (14). Total RNA was extracted with TRIzol reagent (Invitrogen) from the entire trunks for the 1 wpf larvae and the dorsal skeletal muscles of 1 mpf and 1-y old fish that were obtained after skin removal.

Quantitative Real-Time PCR (qRT-PCR).

Reverse transcription was carried out at 37 °C for 1 h in the presence of random hexamers (Promega) using M-MLV reverse transcriptase (Promega). qRT-PCR was performed using a SYBR Green mix and fast amplicon protocol according to manufacturer instructions (Roche, Penzberg, Germany). The primers used were specific for kcnk4a (forward 5′-GGCCGACATGCAGATACAAAC-3′; reverse 5′-CCGTGATCCTGGTCTCATCTC-3′); kcnk4b (forward 5′-CAGTCATCGTGCGTGTCATC-3′; reverse 5′-AGTCTCCAAAGCCCACAGTAG-3′) and the housekeeping gene polr2d (forward 5′-CCAGATTCAGCCGCTTCAAG-3′; reverse 5′-CAAACTGGGAATGAGGGCTT-3′). All primers were designed to produce amplicons of approximately 200-bp length. Data were analyzed using the Ct method and normalized with polr2d.

kcnk4b Cloning and Sequencing.

To isolate kcnk4b, cDNA was prepared from total RNA of muscle tissue of 1-y-old fish. Reverse transcription was carried out at 37 °C for 1 h in the presence of random hexamers (Promega) using Superscript™ II reverse Transcriptase (Invitrogen). A 1550-bp fragment corresponding to the full-length coding sequence (CDS) of kcnk4b was amplified by PCR using the following kcnk4b primers (forward 5′-CAGTCATCGTGCGTGTCATC-3′; reverse 5′-AGTCTCCAAAGCCCACAGTAG-3′). The PCR amplicon was subcloned using the PCR Zero Blunt™ TOPO™ kit (#450245, Invitrogen). Sequencing of the full-length kcnk4b cDNA confirmed the sequence found in databases (NCBI Reference Sequence XP_017208343.1 sequence) (SI Appendix, Fig. S2).

GFP-Tagged kcnk4b Plasmid Construct.

kcnk4b cDNA was amplified using the Phusion HF DNA polymerase (M0530S, Biolabs, France) and a pair of primers (forward: 5′-GGGCGAATTGGGCCCTCTAGATGCATGC-5′; reverse: ATACGCGGATCCACTTTTGGTGTAGAAGGATC-3′) in order to introduce the XhoI and BamHI restriction sites respectively at the 5′ and 3′ ends of the amplicon and to subsequently allow cloning into the pEGFP-N1 expression vector (Clontech).

Cell Culture and Transfection.

Human epithelial kidney (HEK) 293 T cells were grown in standard conditions using Dulbecco modified Eagle’s medium (4.5 g L⁻¹ glucose, Gibco) supplemented with 10% fetal calf serum (Gibco), 1% penicillin-streptomycin, and 0.5 mM stabilized L-glutamine and maintained within a humidified 5% CO₂ atmosphere at 37 °C. Cells plated on plastic culture-dishes were grown for 2 d and subsequently transfected with plasmid constructs using the calcium-phosphate precipitation technique. Subsequent experiments were performed 48 h to 76 h posttransfection.

DNA Electrotransfer in Adult Mouse Muscle.

Expression in mouse flexor digitorum brevis (fdb) and interosseus (io) muscles was achieved by plasmid injection and subsequent electroporation following previously described procedures (33, 34). Mice were given analgesic pretreatment (Metacam, 20 mg/kg) and anesthetized 1 h later by isoflurane inhalation (induction: 4%, 2 L/min; stabilization: 3%, 0.6 L/min). Footpads of each hind-limb paws were injected subcutaneously with 20 μL of a solution containing 2 mg/mL hyaluronidase in sterile saline. Mice were allowed to wake up for 1 h and were then reanesthetized and 20 μL of solution containing 30 to 50 μg plasmid DNA diluted in NaCl 0.9% was injected into each hind-limb paw. Control and test plasmid were injected in the contralateral and homolateral hind-limb, respectively. Two gold-plated stainless-steel acupuncture needles connected to the electroporator were then inserted subcutaneously near the proximal and distal portions of the foot, respectively. The electroporation protocol consisted of 20 pulses of 130 V/cm amplitude and 20-ms duration delivered at a frequency of 2 Hz by a BTX ECM 830 square wave pulse generator (Harvard Apparatus). Isolation of muscle fibers, experimental observations, and measurements were performed 7 to 8 d later.

Confocal Microscopy.

Confocal imaging of HEK cells and mouse muscle fibers was conducted with a Zeiss LM5 Exciter microscope equipped with a 63× oil-immersion objective (N.A. 1.4). GFP was excited at 488 nm with an argon laser, and a 505-nm long pass filter was used on the detection channel.

Electrophysiology.

Isolation of zebrafish and mouse muscle fibers.

Fish were terminally anesthetized using 0.2% MS-222 (tricaine methanesulfonate, Sigma-Aldrich), and trunk muscle of 1- and 2-wk fish or the deep layer of muscle in the dorso-caudal region of 1-y-old fish were dissected. Samples were then incubated at 37 °C in a Tyrode solution containing collagenase (1 mg/mL, type 1, Sigma) for 40 min. After enzyme treatment, muscles were rinsed and stored in Tyrode solution at 5 °C until use. Before each electrophysiological experiment, muscles were transferred into disposable 35-mm culture dishes, and intact skeletal muscle fibers were separated from the muscle mass by gently triturating the muscle with a plastic Pasteur pipette. Isolated muscle fibers were exclusively fast type, easily identifiable on the basis of their morphology and their ability to develop voltage-gated Na⁺ currents and APs (15).

Single mouse muscle fibers were isolated from the fdb and io muscles using a previously described procedure (33, 34). In brief, mice were anesthetized by isoflurane inhalation and then killed by cervical dislocation before removal of the muscles. Muscles were incubated in Tyrode solution containing collagenase (2 mg/mL, Sigma, type 1) for 60 min at 37 °C. Single fibers were then obtained by mechanical trituration of a given muscle as done for fish muscles.

Single-channel recordings.

Single-channel currents were recorded from cell-attached or inside-out membrane patches at room temperature using a voltage-clamp amplifier (model RK 400; Bio-Logic, Claix, France). Currents flowing into the pipette were considered to be positive. Channel activity was determined from the average current (I) as NPo = I/i in each patch, where i is the single-channel current, N is the number of open channels, and Po the open-state probability. Currents filtered at 300 Hz were acquired using the pClamp10 software (Molecular Devices) driving an A/D converter (Digidata 1400A, Molecular Devices) at a sampling frequency of 1 kHz.

Whole-cell voltage- and current-clamp recordings.

Zebrafish fast single fibers were voltage-clamped using the silicone clamp technique as previously described (15). In brief, the major part of a single fiber was electrically insulated with silicone grease. A micropipette was inserted into the fiber through the silicone layer to voltage- or current-clamp the portion of the fiber free of grease (50 to 100 μm length) using a patch-clamp amplifier (Bio-Logic RK-400, Claix, France) in whole-cell configuration and to dialyze the cell with the internal solution. Command voltage or current pulse generation and data acquisition were done using the pClamp10 software (Axon Instruments Inc.) driving an A/D converter (Digidata 1400A, Axon Instruments Inc.). Analog compensation was systematically used to decrease the effective series resistance. Currents in voltage-clamp experiments and voltage changes in current-clamp experiments were acquired at a sampling frequency of 10 and 20 kHz, respectively. In voltage-clamp experiments, cell capacitance was determined by integration of a current trace obtained with a 10-mV hyperpolarizing pulse from the holding potential preceding each voltage step and was used to calculate the density of currents (A/F). The holding potential in voltage clamp was fixed at −80 mV. In current-clamp experiments, the resting potential was brought to −100 mV by injecting negative current. At the beginning of current-clamp experiments, positive 0.5-ms current steps of increasing amplitudes were applied to determine the threshold for AP activation. Fibers were subsequently stimulated by a short 50-ms negative current steps to measure cell conductance followed by a 0.5-s or 2-s 50-Hz train of positive 0.5-ms current steps.

Solutions.

For single-channel recordings, the pipette solution was filled with a Tyrode solution containing (in mM): 140 NaCl, 5 KCl, 2.5 CaCl₂, 1 MgCl₂, and 10 HEPES, adjusted to pH 7.2 with NaOH, or with K⁺-rich solution containing (in mM): 140 KCl, 2.5 CaCl₂, 1 MgCl₂, and 10 HEPES, adjusted to pH 7.2 with KOH. For cell-attached experiments, fibers were bathed in the K⁺-rich solution to zero the cell internal potential. For inside-out experiments, the membrane patch was exposed to a solution containing (mM): 140 KCl, 1 MgCl₂, 1 ATP (or 0), 2 EGTA (or 0.1 CaCl₂), and 10 HEPES, adjusted to pH 7.2 with KOH.

For whole-cell voltage-clamp recordings, the external solution contained (in mM): 140 N-methyl-D-glucamine-MeSO₃, 5 KOH, 0.5 CaCl₂, 1 MgCl₂, 0.5 9-anthracenecarboxylic acid, and 10 HEPES, adjusted to pH 7.2 with MeSO₃. For current-clamp experiments, the external solution was Tyrode. In both conditions, the internal solution contained (in mM): 160 K-MeSO₃, 1 MgCl₂, 5 Na₂-ATP, and 5 Na₂-phosphocreatine, 1 MgCl₂, 5 glucose, 5 HEPES adjusted to pH 7.2 with KOH.

Stock solution of arachidonic acid (AA) (Sigma) was prepared in ethanol at 100 mM, flushed with Ar and kept at −80 °C until use. The final solutions containing AA were made daily. Stock solution of ATP (100 mM, potassium salt, Sigma) was buffered at pH 7 (adjusted with KOH). Stock solution of colchicine (Sigma) was prepared in water at 50 mM and diluted daily at final concentration in internal and external solution.

Isolated fibers for cell-attached and whole-cell experiments or membrane patches for inside-out experiments were exposed to different solutions by placing them in the mouth of a perfusion tube from which the rapidly exchanged solutions flowed. For measuring the effect of warming, the external solution at room temperature was rapidly exchanged with the same external solution maintained at 37 °C.

Statistics.

For qRT-PCR experiments, statistical analysis was performed using one-way ANOVA and Tukey’s post hoc tests (n = 3 independent experiments). For electrophysiological experiments, data are given as means ± SEM and were compared using paired, or unpaired when indicated, Student t tests. Differences were considered significant when P < 0.05. Labels *, ** and *** indicate P < 0.05, P < 0.005, and P < 0.0005, respectively.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

All study data are included in the article and/or SI Appendix.