Introduction

Mechanosensitive (MS) ion channels are found in a variety of organisms from bacteria to human. These channels transduce mechanical stimuli into intracellular signals, such as elevation in intracellular Ca2+ concentration and electrical activities 1, 2, 3, 4, 5, 6, which are important in many physiological functions and pathological conditions, for example, sound detection 4, cell volume regulation 7, touch sensation 8, hypo-osmotic shock protection 9, cell locomotion 10, muscular dystrophy and cardiac arrhythmias 11. The BK channel is one type of MS channels which is widely expressed in both excitable and non-excitable cells and plays an important role in, for example, vascular smooth muscle tone regulation, neuronal firing and endocrine cell secretion 12, 13, 14, 15, 16. The MS properties of the BK channels expressed in skeletal muscle 17, kidney 18, smooth muscle 19, neuroepithelial cells and osteoblasts 20, 21 have been characterized. We previously identified a stretch-activated BK channel (referred as SAKCaC) in cultured chick embryo ventricular myocytes 22. Like other BK channels, SAKCaC can be activated by membrane voltage and intracellular Ca2+. However, the unique property of this channel is that it can also be activated by membrane stretch. Molecular cloning and subsequent functional characterization of the SAKCaC have shown that the channel subunit contains a 59-amino-acid insert (called stress-axis regulated exon (STREX)) in the C-terminus, and that STREX domain is the stretch sensor conferring the membrane stretch sensitivity to the channel 23, 24. However, it is not well understood how the membrane stretch is transferred to the gate to open the channel. In the model based on the MthK channel structure, the pore-forming a subunit contains two regulator of K+ conductance (RCK) domains in the intracellular C-terminus, and eight RCK domains from the tetrameric channel form a gating ring 25, 26. The linker that connects the S6 gate to the RCK domain and the gating ring form the linker-gating complex, which functions as a passive spring and transduces the force generated by intracellular Ca2+ to the channel gate 27. The STREX domain is located between RCK1 and RCK2 domains in the long intracellular C-terminus of SAKCaC (Figure 1) 19. The present study showed that shortening the linker rendered the channel more sensitive to membrane stretch, whereas lengthening the linker resulted in the opposite effect, suggesting that, similar to Ca2+ activation, the membrane stretch sensed by the STREX domain seems to be transmitted to the channel gate by the linker-gating complex.

Figure 1
figure 1

Schematic presentation of the linker, RCK1, RCK2 and STREX domains of BK channel α-subunit.

Full size image

Results

Activation of SAKCaC by membrane stretch

We first characterized the activation of wild-type (WT) SAKCaC by membrane stretch in excised inside-out patches. Figure 2A shows typical single channel current traces in response to stepwise negative pressures with 10 mmHg increments. Figure 2B plots the channel open probability (Po) against the level of pressure applied. The Po increased with increased pressure. In contrast, there were no detectable channel activities induced by membrane stretch in mock transfected cells (data not shown). The SAKCaC activation occurred in the absence of intracellular Ca2+, suggesting that membrane stretch directly activated the channel. All these results are in agreement with those reported previously 9, 23, 28.

Figure 2
figure 2

Pressure dependency of Po of SAKCaC. Pipette solution contained 145 mM K+, 0 mM Ca2+ (with 1 mM EGTA), bath solution contained 145 mM K+ and 0 mM Ca2+ (n = 18). (A) Representative single-channel current traces of SAKCaC in response to membrane stretch (C denotes the closed current level). Membrane potential was held at +60 mV. (B) Pressure-Po curve was fit by the Boltzmann equation.

Full size image

Effect of linker length on Po of SAKCaC

To address the role of the linker in SAKCaC activation, we produced mutant channels, in which the S6-RCK1 linker was shortened by deleting one residue or lengthened by inserting three residues. Figure 3A shows representative current traces of the mutant BK channels in response to membrane depolarization (+60 mV), and Figure 3B summarizes the Po. It is obvious that the Po was greatly increased by shortening the linker (−1), but decreased by lengthening the linker (+3). Similar results were reported previously in the Slo BK channels and the interpretation is that the linker-gating ring complex forms a passive spring to exert the force on the gate 27. We also generated mutant subunits in which more than one residue was deleted or more than three residues were inserted in the linker. However, no detectable channel activities were observed in cells expressing these mutants, even at voltages of greater than +100 mV and in the presence of 100 μM intracellular Ca2+ (data not shown), possibly due to defective channel assembly and/or membrane trafficking.

Figure 3
figure 3

Change in the linker length alters the channel activity in the absence of Ca2+ (n = 11). (A) Representative recording of currents at a membrane voltage of +60 mV. (B) Plot of Po versus change in the linker length.

Full size image

Effect of changing the linker length on the mechanosensitivity of SAKCaC

We next examined the effect of changing the linker length on SAKCaC activation by membrane stretch. As shown in Figure 4, membrane stretch greatly increased the Po of both WT and mutant channels in a dose-dependent manner. However, the mechanosensitivity was strongly enhanced by shortening the linker (−1), and decreased by increasing the linker (+3). The relationship of Po to the pressure, resulting from membrane stretch, could be fitted with the Boltzmann equation: Po=1/[1+exp(p-p1/2)k], where p is the pressure, p1/2 is the pressure required to induce half-maximal channel activation, and k is the reciprocal of the curve steepness or slope. The p1/2 values were 23 mm Hg, 18 mm Hg*, and 27 mm Hg*, and the k values were 22 mm Hg, 23 mm Hg, and 20 mm Hg for WT, −1, and +3 mutant channels, respectively. The p1/2 values for both mutant channels were significantly different from that for WT (* indicate significant difference p < 0.05). The unitary channel conductance (260 ± 3.1 pS), however, remained unaltered by either membrane stretch or change in the linker length (data not shown).

Figure 4
figure 4

Mechanosensitivity of SAKCaC with linkers of different length (n = 10). The pipette and bath solution contained 145 mM K+ with 0 mM Ca2+. Membrane potential was held at +60 mV. The data points were fitted with the Boltzmann equation.

Full size image

Effect of changing the linker length on the mechanosensitivity of SAKCaC at different membrane voltages and intracellular Ca2+ concentrations

One key feature of the BK channels is Ca2+ and voltage dependence of channel activation 12. Here we therefore examined the role of the linker in Ca2+-dependent SAKCaC activation in the presence of membrane stretch. As shown in Figure 5, increase in intracellular Ca2+ concentration from 0 to 1.2 μM greatly enhanced the channel activity of both WT and mutant channels at all the levels of membrane stretch tested. Compared to WT, the Po-Ca2+ curve for the mutant channel with a shorter linker (−1) was left-shifted, and the slope was not significantly changed (Figure 5B). In contrast, the Po-Ca2+ curve was shifted to the right for the mutant channel with a lengthened linker (+3), and the slope was also decreased (Figure 5C). In the presence of intracellular Ca2+concentrations, application of negative pressure induced an increase in the Po. Furthermore, for both WT and mutant channels, increase in negative pressure resulted in a parallel left-shift in the Po-Ca2+ curves without obvious change in the slope (Figure 5; Table 1). Thus, the enhanced SAKCaC activation in the presence of membrane stretch seems not due to an increase in Ca2+ sensitivity, suggesting that change in membrane stretch does not alter the Ca2+ sensitivity.

Figure 5
figure 5

Effect of changing the linker length on Ca2+ dependence of SAKCaC in the presence of membrane stretch (n = 11). Membrane potential was held at +60 mV. The data were fitted with the Hill equation: Po=[Ca2+]iηH/(KdηH+[Ca2+]iηH), where ηH is the Hill coefficient and Kd is the dissociation constant.

Full size image
Table 1 The effect of membrane stretch on the Hill coefficient of SAKCaC with a linker of different length
Full size table

We also examined whether change in the linker length and membrane stretch affects the sensitivity of SAKCaC to membrane voltage. Figure 6 plots the Po against the membrane voltage (V) for the WT and mutant channels in the absence of intracellular Ca2+ and in the presence of membrane stretch ranging from 0 to 40 mmHg. These data were fitted with the Boltzmann equation to derive the voltage for half-channel activation, V0.5. The results show that lengthening the linker increased, whereas shortening the linker decreased, the V0.5 values (Table 2). In addition, compared to WT at all the membrane stretches tested, the Po-V curve was left-shifted for the mutant channel with a shortened linker (Figure 6B), but right-shifted with decreased slopes for the mutant channel with a lengthened linker (Figure 6C). The Po at fixed pressure was voltage-dependent, being higher when the membrane was depolarized. When the pressure was increased, the Po-V curves were left-shifted. However, for both WT and mutant channels, there was almost no further increase in the slope at the relatively high membrane stretch. Our results also indicate that the channels were less sensitive to membrane stretch at relatively high intracellular Ca2+ concentrations (Figure 5) or membrane depolarization (Figure 6) (for original data see Supplementary information, Data S1).

Figure 6
figure 6

Effect of changing the linker length on voltage dependence of SAKCaC in the presence of membrane stretch in the absence of Ca2+ (n = 16). The data were fitted with the Boltzmann equation: Po=1/{1+exp[-ZF/RT(V-V1/2)]}, where V1/2 is the membrane potential for half-maximal channel activation.

Full size image
Table 2 The effect of membrane stretch on V1/2 (mV) of SAKCaC with a linker of different length
Full size table

Discussion

This study investigated the role of the linker in activation and particularly mechanosensitivity of SAKCaC. Membrane stretch induced an increase in the channel activity in a dose-dependent manner. The sensitivity to membrane stretch increased greatly by shortening the linker, and reduced by lengthening the linker, both in the presence and in the absence of intracellular Ca2+ (Figures 4 and 5). Furthermore, the SAKCaC was less sensitive to membrane stretch in the presence of relatively high intracellular Ca2+ concentrations or membrane depolarization. These results provide evidence consistent with the mechanical model recently proposed for the BK channel gating by Niu et al. 27. In this model, eight RCK domains form the gating ring, and the linker that connects the S6 gate with RCK1 domain acts as a passive spring, transducing the force to the gate and pulling it open when the membrane is stretched. This model is able to account for the mechanosensitivity of SAKCaC. The membrane tension stretches the channel protein, in turn pulling the linker and increasing the tension on the spring. The tension on the spring opens the gate and thereby increases the SAKCaC activity. Thus we suggest that the membrane stretch is also transferred to the gate by the linker ring complex. More recently, we have demonstrated that the sensitivity of SAKCaC to membrane stretch is detected by the STREX domain locating between RCK1 and RCK2 in the intracellular C-terminus. Deletion of STREX or substitution of serine (S24) within the STREX domain diminishes the stretch sensitivity 28. Thus, it is obvious that the mechanical forces are detected by the STREX domain and then transferred to the gate by the linker connecting the S6 gate with RCK1. Then a question arises as to how the STREX senses and transmits the force in the membrane to the channel. A simple explanation may be that the STREX domain interacts with a membrane-associated protein, which senses or transmits the force in the membrane, as illustrated in Figure 7. One notable feature of the SAKCaC channels with linkers of different lengths is the high Ca2+ sensitivity (Figure 5). The present study suggests that it may arise from the STREX domain, as previously described for the BK channels in human glioma cells 29. It has been proposed that binding of Ca2+ to the gating ring may increase the Po by expanding the gating ring to pull the channel open 26, 27, 30. Therefore, we conclude that the force generated by Ca2+ or membrane stretch is transferred to the gate via a common pathway.

Figure 7
figure 7

Regulation of BK channel gating by membrane stretch based on the spring model proposed for BK channel gating by Niu et al 27. The membrane stretch (Fm) was conveyed by a membrane-associated protein and STREX, and then transferred to the gate by the linker connecting the transmembrane segment 6 (S6) gate. The charged transmembrane segment 4 (S4) voltage sensor is indirectly connected to the S6 gate through a spring. Addition of amino acids into the linker leads to changes of linker length (ΔL).

Full size image

The influence of the linker length on the Ca2+ and voltage dependence of SAKCaC activation in the presence of membrane stretch was also examined. The Po versus membrane potential (V) or Ca2+ curves were shifted by altering the linker length; shortening the linker had little effect on the slopes of Ca2+-Po curve and V-Po curve, suggesting that shortening linker length does not significantly affect the voltage sensitivity and Ca2+ sensitivity of SAKCaC. However, lengthening the linker decreased the slopes of Po-Ca2+ curve and voltage-Po curve, revealing that increasing the linker length decreases the effectiveness of Ca2+ and voltage activation of the channel. Furthermore, the Po against membrane potential (V) curve was left-shifted for SAKCaC when membrane stretch was applied. However, the change in membrane stretch from −10 to −40 mm Hg had little effect on the slope over the range of Po values, suggesting that membrane stretch does not significantly affect the voltage sensitivity of SAKCaC. These results are consistent with the gating properties observed from the BK channels in mouse skeletal muscle fibers 30, 31, which were activated by membrane stretch without altering the voltage sensitivity. We also evaluated whether the SAKCaC activation by membrane stretch was due to an increase in Ca2+ sensitivity. We examined the effects of Ca2+ on the SAKCaC gating when membrane stretch was applied. The channel activity was enhanced by membrane stretch, resulting in a parallel left-shift of the Ca2+-Po curve without affecting the slope, suggesting that Ca2+ sensitivity is not altered by membrane stretch. This is in agreement with the results previously described for the BK channels in smooth muscle cells 19 and skeletal muscle cells 32, whose activation in response to membrane stretch is not due to enhancement of Ca2+ sensitivity. Although activation of SAKCaC by membrane stretch is independent of the Ca2+ or voltage sensitivity, the spring-based gating model suggests that the effect of membrane stretch on the SAKCaC gating is similar to that produced by increasing intracellular Ca2+ or membrane depolarization, as illustrated in Figure 7. Binding of Ca2+ expands the gating ring and pulls the gate open, while depolarization opens the SAKCaC via the S4 voltage sensor pulling a spring to open the gate 27, 33. It should be emphasized that tension on the spring is essential for influences of Ca2+, membrane stretch, and voltage on the SAKCaC activation. The tension, caused by either membrane stretch or Ca2+ binding or membrane depolarization, would destabilize the closed conformation and favor the open conformation. The spring model predicts that the major effect of change in Ca2+ concentration or membrane potential occurs effectively only in the closed-state conformation (Figure 7). The effectiveness of SAKCaC activation by membrane stretch was reduced at relatively high intracellular Ca2+ or membrane depolarization, suggesting that when the channel is in open-state conformation, the membrane stretch also becomes less effective.

MS channels are found in many mammalian cell types, in which they are thought to be involved in a number of physiological or pathological functions. In the heart cells, the SAKCaC is thought to be involved in dilation or relaxation of heart. However, it still remains to be determined how SAKCaC functions in the heart under physiological and pathological conditions. The present study has provided evidence shedding light on how the channel is activated in response to membrane stretch.

Materials and Methods

Mutagenesis and expression of SAKCaC

The a subunit of the MS BK channel or SAKCaC (accession number AB 072618) was previously cloned from a chick embryonic heart cDNA library 23. Introduction of mutations in the linker between S6 and RCK1 domain was described previously 27. In brief, the linker was shortened by deletion of one amino acid residue from GGSYSAVSG to GGSYAVSG (referred as −1), or lengthened by insertion of three amino acid residues from GGSYSAVSG to GGSYSAAGAVSG (referred as +3), using the QuickChange site-directed mutagenesis kit (Stratagene). All mutations were confirmed by sequencing. Chinese hamster ovary (CHO-K1) cells were used to express WT and mutant BK channels as described previously 2, 23.

Electrophysiology

All recordings were performed at room temperature (22-25 °C) in the inside-out configuration of patch-clamp technique using an Axon 200-B amplifier. Patch pipettes were fabricated from disposable micro-pipettes (Drummond Scientific) in two stages on a PP-83 puller (Narishige Scientific Instrument Lab, Tokyo) to a tip diameter of 0.5–1.0 μm and fire-polished on a MG-83 microforge (Narishige Scientific Instrument Lab, Tokyo) to give an average resistance of 5-10 MΩ in recording solution. Negative pressure was applied via the patch pipette using a pneumatic transducer tester (DPM-IB, BIO-TEK Instruments INC.VT.). Currents were sampled at 5 kHz and filtered at 1.5–2.9 kHz via a four-pole low pass Bessel filter. The program package Clampfit 8.0 was used for data acquisition and analysis. Continuous recordings of 2 000-4 000 ms were used to estimate the channel open probability (Po). The pipette solution contained 145 mM potassium-gluconate, 1 mM EGTA, 10 mM HEPES, and 5 mM glucose with pH adjusted to 7.4. To minimize potential Ca2+-dependent SAKCaC activation, EGTA was used to eliminate possible cytoplasmic Ca2+ concentration via putative Ca2+-permeable stretch-activated channels and/or stretch-induced Ca2+ leakage. The bath solution contained the same components as the pipette solution, except for inclusion of indicated Ca2+ concentrations. For Ca2+ concentration lower than 1 μM, proper concentrations of CaCl2 and 1 mM EGTA were used, on the basis of calculation using the EQCAL program (Biosoft, MO). All reagents were purchased from Sigma.

(Supplementary Information is linked to the online version of the paper on the Cell Research website.)