Voltage dependence of I_SAC

To understand the voltage-dependence of the stretch activated current, we plotted the currents flowing at the end of the 140 ms pulses (late currents I_L) as iv curves versus the potential (Fig. 4A). Before the mechanical stimulus (filled circles), the resulting iv curve was flat small because Cs⁺ had blocked the K⁺ current components, the slope at negative potentials indicated a high input resistance of 500 MOhm. A 10 μm local stretch displaced the iv curve to the open circles (input resistance 62.5 MOhm). During stretch, the iv curve was almost linear and intersected the voltage axis at -5 mV. Panel B of Fig. 4 shows the difference current, i.e. the current induced by stretch. In this definition, I_SAC showed an almost linear voltage-dependence, i.e. its conductance G_SAC was constant (no dependency on membrane potential). The data of Fig. 4B could be fitted by linear regression where the conductance was 16.5 nS and the reversal potential was E_rev = -0.5 mV (r=0.996).

Figure 4

Voltage dependence of the stretch induced current I_SAC through non-selective cation channels. Currents at the end of 70 ms pulses were plotted versus the potential of the clamp pulse. Superimposed K⁺ currents were blocked by replacing K⁺ by Cs⁺ ions in (more...)

Linear iv curves with reversal potentials between 0 and -15 mV were reported for both axial stretch [5, 81] and for local stretch [36, 37, 82]. Since at potentials negative to -10 mV I_SAC is inward, its activation can depolarize the diastolic membrane potential, and it can retard the repolarization phase 2, from the plateau back to the resting potential. At positive potentials I_SAC is outward. This stretch activated outward current can shorten the initial repolarization phase 1. Thus, the properties of the mechanically activated current I_SAC can explain the stretch induced modification of the resting and action potentials in non-clamped cardiomyocytes.

I_SAC is a non-selective cation current

I_SAC was characterized in terms of the ion selectivity. Substitution of chloride by aspartate or by glutamate did not change amplitude or reversal potential suggesting that movement of Cl^--ions does not contribute to I_SAC, or that the channel is a cation selective channel [7, 36, 81]. To quantify the relative permeability of the stretch activated channel for different cations, the 150 mM extracellular Na⁺ ions were replaced by the same concentration of an other cation species, resulting in a change of both amplitude and reversal potential of I_SAC. From the shifts we estimated that the SACs are best permeable for Cs⁺ ions [36, 82]. Usually, the reversal potential of I_SAC is approx. -10 mV, this would indicate that K⁺ efflux and Na⁺ influx pass through the channel with the same permeability. For mouse ventricular myocytes superfused with a Ca²⁺ free medium, the permeability ratio was P_Cs: P_K :P_Na :P_Li :P_TMA :P_NMG = 1.3 :1.1 :1.0 :0.8 :0.1 :0.03 [37]. That is, even the large organic cations tetramethyl ammonium (TMA⁺) or N-methyl glucosamine (NMG⁺) could carry a small inward current when local stretch activated SACs. The numbers can't be easily compared with others because of different temperatures (22 °C in [81, 82]) and extracellular Ca²⁺ concentrations in the extracellular superfusate (0 mM in [37], 0.2 mM in [81] and 1.8 mM in [82]) and the patch electrode (low or high concentrations of EGTA or BAPTA, see below). In rat atrial myocytes [82] the permeability ratio was P_Cs :P_Na :P_Li =1.05 : 1 : 0.98 for the stretch activated current and P_Cs :P_Na :P_Li= 1.49 : 1.0 : 0.79 for the non-selective background current.

In guinea pig ventricular cells, stretch induced a negative I_SAC also when the 150 mM Na⁺ ions were replaced by 90 mM Ca²⁺. At -100 mV, there was a Ca²⁺ carried inward current that reversed polarity at +15 mV as one would expect for Ca²⁺ ions passing inwardly at the same ease as K⁺-ions flowing outwardly. However, the iv curve was no longer linear with Ca²⁺ as the charge carrier, negative to -20 mV it bended to become nearly independent of membrane potential [36]. In murine ventricular myocytes isotonic substitution of Na⁺ by Ca²⁺ induced outward current components that hampered the quantification of I_SAC and its Ca²⁺ permeability. In summary, we conclude that SACs are Ca²⁺ permeable, albeit with low permeability (compare [73], for review see [21, 59]).

When Ca²⁺ was removed from an extracellular Cs⁺ solution during continuous stretch, the amplitude of I_SAC increased by a factor of 2.5 as if Ca²⁺ ions would hinder the permeation of the Cs⁺ ions [37]. An even stronger hindrance was achieved by addition of the 3-valent cations La³⁺ or Gd³⁺, both blocked I_SAC. The concentration for this block was low, 8 μM were saturating provided one waited for 10 min. This contrasts with the 100 μM concentrations used in other reports; we suggest that these high concentrations are the result from the fact that the on-rate of the block increases with the concentration (low of mass action). The Gd³⁺ block of I_SAC was not completely reversible, i.e. I_SAC did not recover within 30 min, as can be expected from the high affinity of Gd³⁺. The block of SACs by divalent and trivalent cations has been described in the literature [79]. It has been interpreted as a fast flickering block, i.e. the permeating cation interacts with the inner side of the channel protein; the single channel current is interrupted for the time of interaction (binding). If the channel is blocked for several femto seconds, this block appears as a reduced current amplitude if the recording system has a time resolution in the order of milliseconds.

In rat atrial myocytes, two types of non-selective channels have been described, SACS and non-selective background channels. Gadolinium inhibited the background channels in a concentration dependent manner with an IC₅₀ value of 46 μM. In presence of 100 μM Gd³⁺, stretch still induced I_SAC and diastolic depolarization [82]. The result can't be explained in the context of the other publications, at the moment. The question whether or not SACs are open without the application of exogenous stretch, would have consequences for the cellular Na⁺ homoeostasis (see below), for ventricular myocytes a clear answer is missing. For the experiments applying axial stretch to guinea-pig ventricular myocytes, the answer would be no channel openings without sarcomere lengthening [5], and Gd³⁺ was reported not to hyperpolarize the resting membrane [25]. This conflicts with the Gd³⁺ induced hyperpolarization reported by those groups that applied local stretch [36, 37, 82]. Also Fig. 4A indicates that Gd³⁺ shifted the iv-curve in the continuous presence not back to the control before stretch but to less inward current (triangles more positive than filled circles in Fig. 4A). At present, one can not rule out that part of the difference is due to the mechanical effects of attaching the glass stylus, even if the stylus had not been used for stretch.

Unfortunately, Gd³⁺ is not a selective blocker of SACs, it reduces also the potassium currents I_K1 and I_Ko [25]. Hence, 30 μM streptomycin has been introduced as tool to separate I_SAC from the stretch induced net currents [5, 17]. I_SAC strongly resembled when these currents were defined as difference current by 8 μm Gd³⁺ or by 30 μM streptomycin [44]. In both definitions, I_SAC had a voltage-independent conductance and a reversal potential of approx. -10 mV. Some groups have used the drug amiloride to block SACs in oocytes [20], also the selectivity to SACs is low, it binds to a variety of other channels. For purification of the channel protein a highly specific toxin would be desirable. A peptide isolated from the spider Grammtula spalata (GTx) has been identified and shown to be useful [67]. Unfortunately, the effort to synthesize this protein has not yet been successful. For a recent review on the pharmacology of mechanogated channels see [23].

Single channel analysis

The first evidence for the existence of SACs came from single channel recordings performed on oocytes [45, 80], myoblasts [19], smooth muscle cells [73] and embryonic chicken heart cells [57, 63]. In most experiments, the single channel current was activated by negative pressure (suction) applied to the open end of the electrode forming the cell attached patch (Fig. 5). The open probability could be enhanced equally well by suction or by pressure. There was no obvious time delay between suction and appearance of channel activity. The SACs in oocytes were cation-selective with P_K >P_Cs >P_Na >P_Li >P_Ca [80]. Tetramethylammonium was impermeable. In smooth muscle cells, single channel conductance followed the rank P_K >P_Na >P_Cs >P_Ba >P_Ca [24]. The single channel conductance was between 20 pS and 40 pS, depending on the species of the permeable cation and on the temperature of the experiment.

Figure 5

Currents flowing through single SACs. Left part, AA-AH: recordings from tissue-cultured chick ventricular myocytes, membrane hyperpolarized by 40 mV. Aa no suction, Ab 20 mm Hg suction. Ac, Ad amplitude histograms demonstrating the stretch activation (more...)

Results from single channel analysis in oocytes [21, 80] and embryonic chicken skeletal muscle fibres [19] described a time dependence of SACs. Upon step like changes in the negative pressure, SACs activate within several milliseconds [22] and than inactivate with a time constant in the order of 200 ms [52]. This inactivation could or could not lead to a sustained channel activity. Kinetic analysis indicated that there were three closed states and one open state. The open time was independent of both pressure and membrane potential, however, dependent on the ionic species [80]. The single channel kinetics should be reflected in the time course of whole cell currents, however, this link to the physiological significance of adaptation of single SACs has not yet been achieved [59].

There were only very few reports that reported successful single channel analysis in atrial myocytes [39, 82]. In ventricular myocytes the reports remain "episodic" [57] although several groups have spend long periods of unsuccessful experiments to find them. Nowadays one interprets that in ventricular myocytes SACs may be "hidden" in parts of the surface membrane where the patch pipette has only limited access, such as the transversal tubules [28] or the caveolae. Alternatively, it has been suggested that cytoskeletal structures that may be important for the mechanosensitivity, are destroyed when the membrane seals to the glass pipette [59].

The amplitude of I_SAC and "mechanosensitivity"

Usually, the amplitude of the measured membrane current is normalized by the cell size, in the literature. For this, one measures the membrane capacity (proportional to the surface membrane) and divides the current by the given number of pF. In our own experiments, the membrane capacity varied between 76 and 135 pF, in dependence on species and age. Division of I_SAC by 100 pF yields current densities between 5 and 20 pA/pF [36, 37], in dependence on membrane potential, extent of local stretch, and species (see below). These values are larger than those reported for end-to-end stretch (0.5–1.5 pA/pF) [5, 81]. We hesitate to apply the normalization process in case of the local stretch because this method modifies only part of the surface membrane. Essentially, this fraction is not known, it ranges between 5 and 20% of the outer cell surface. That is, the actual divisor should be smaller and the current density higher than given by the above numbers. An other estimate tries to find how many open channels (N_o) are necessary to generate a current of -1 nA at -100 mV, when the single channel conductance is 25 pS (see above). Using N_o = 1 nA/(25 pS*0.1 V) we obtain an estimate that 400 channels should open simultaneously.

To define the mechanosensitivity of whole cell I_SAC, the amplitude of I_SAC has been plotted versus the mechanical input. In case of axial stretch, the current at -100 mV increased linearly according to I_SAC = -0.3*SL% (Fig. 6D, SL%: relative increase in the sarcomere length [81]). In case of local stretch [36], the relation was curvilinear. The slope of the curve defines the mechanosensitivity. Fig. 6C indicates that the mechanosensitivity was not constant but increased with the extent of local stretch. Fig. 6 also indicates that mechanosensitivity differs between ventricular myocytes isolated from hearts of different animals. Low mechanosensitivity was found for cardiomyocytes from mice and young rats (filled circles). Cardiomyocytes from young guinea pigs (empty circles) and rats in the age of 15 months (filled squares) had a higher mechanosensitivity. Cells from rats with spontaneous hypertension (SHR, 15 months, filled triangles) had a mechanosensitivity that was several times higher than the one of the age-matched controls (compare the third with the second trace in Fig. 6A). Also the human ventricular myocyte displayed a high mechanosensitivity (Fig. 6B, Fig. 6C open squares). The literature suggests that cardiac hypertrophy goes along with a proliferation of the subsarcolemmal cytoskeleton in the individual hypertrophied cardiomyocytes [72]. We speculate that these results may suggest that both cytochalasin D and colchicine soften cytoskeletal network; in the consequence the coupling between the mechanical stimulus and the activation of SACs would be less efficient.

Figure 6

Mechanosensitivity of I_SAC measured at -45 mV holding potential (currents due to 140 ms depolarizations to 0 mV superimposed). A,B: pen recordings, period and extent of local stretch indicated. A: Increase in mechanosensitivity in rat ventricular myocytes (more...)

How does mechanical deformation modulate channel activity?

Activation and inactivation of ion channels could be caused by mechanically induced changes in the surface tension in the lipid bilayer that would induce conformational changes in the channel protein. Such a mechanism has been made plausible for channels incorporated in the membranes of bacteria (MscL channel of E.coli, for details see the review [21]). For eukaryotes this explanation is unlikely because the membrane lipids are fluid, and there redistribution would rapidly abolish possible tension. In ventricular myocytes, stretch is suggested to unfold the 'slack' surface membrane primarily from subsarcolemmal invaginations near the z-line [47] and to induce incorporation of sub-membrane caveolae [40]. Hence, lengthening of the sarcomeres does not necessarily increase the stress in the lipid bilayer membrane. It is more likely that the structures of the extracellular matrix (ecm) and/or the cytoskeletal proteins bear and transmit the tension [58, 59]. In case of isolated ventricular myocytes the ecm is cleaved away, including the glycocalix [33], hence, the channel activation seems to be independent of ecm proteins. Activation of Cl^--channels by pulling microbeads succeeded only when these beads were coated with antibodies binding to integrins. Further more, experiments studying the influence of mechanical stress on cellular hypertrophy plated the cells on matrix proteins interacting with integrins. Thus, mechanical stimuli may activate different targets via different signalling pathways.

When SACs were first described in embryonic chicken myotubes, the cells were treated with tubulin and actin reagents to test the involvement of the cytoskeleton. It was shown that colchicine depolymerising tubulin had no significant effect, but cytochalasins depolymerising F-actin increased the channel's stretch sensitivity [19]. The results suggested that neither tubulin nor f-actin would be necessary for mechanosensitive gating. It was postulated that the SACs were normally linked into a component of the membrane skeleton that runs parallel to the actin network, so that when actin was depolymerised by cytochalasin D, more stress could be transferred to the channel-linked component. The identity of that component has not yet been resolved [59]. Our lab tested colchicine and cytochalasin D on whole cell I_SAC of murine ventricular myocytes [37]. Both toxins reduced steady state I_SAC activated by stretch. Further, preincubation with one of the toxins prevented the mechanical activation of I_SAC. The result that the basic effects of colchicine or cytochalasin D on I_SAC did not differ, was somewhat unexpected. If depolymerization of tubulin and F-actin has very similar effects, mechanosensitivity is unlikely to be caused by a specific interaction between the SAC with one of these cytoskeletal elements [34]. That is, SAC opening should not be attributed to a direct interaction with tubulin or F-actin (protein-protein interaction).

At present, we speculate with the tensegrity model that both actin and tubulin are necessary for the mechanical tension within the cell. The tensegrity model [71] compares the surface membrane of a cell, e.g. of a fibroblast, with the tent supposing that tubulins were the stiff rods bearing the external mechanical load and the F-actin strands were the cables that are necessary to provide the tension. The intermediate filaments were supposed to be tension-free, until the cell would be deformed or would swell. The tensegrity model postulates that the exogenous mechanical energy would be distributed to all intracellular components including the nucleus. Whilst the model has been nicely elaborated for fibroblasts or endothelial cells, details applicable to cardiomyocytes are missing. That is, at present we do not know how much of the exogenous force is dissipated by the cytoskeleton and how much by the sarcomeric proteins (titin, proteins of the z-band etc.). If we apply the tensegrity model to cardiomyocytes, the above similarity of the effects of colchicines and cytochalasin would suggest that activation of SACs needs tension in the cell surface, or, that removal of this surface tension by breaking down one of the essential cytoskeletal constituents will hamper the spread out of energy from the stylus to distant structures. When the single channel activity was facilitated this would mean that local interactions within an isolated membrane patch would still function, and that cytochalasin has impaired the dissipation of mechanical energy from the patch to its neighbourhood.

The cellular elements involved in mechanical channel activation, i.e. mechanical deformation of the cell surface, stress propagation and modulation of protein conformation, may localize in different parts of the cell. In the example of endothelial cells, the streaming blood deforms the luminal surface, cellular cables transport this mechanical stimulus to the focal adhesions on the basolateral side of the cell, where integrins interact with the proteins of the extracellular matrix. The mechanical stimulus on these ligated integrins induces phosphorylation cascades that finally activate the NO-synthase that is localized in the caveolae of these cells. In cardiomyocytes, most have activated SACs with "naked tools" (fire polished glass, naked carbon fibres), i.e. channel activation seems to be independent of interaction between matrix proteins and integrins. An analogy of this signalling pathway has been suggested for rat ventricular myocytes: stretch can induce fluorescence signals indicative for the production of NO [56]. Using specific inhibitors, the likely signalling pathway is that axial stretch (sarcomere length increased from 1.81 to 1.99 μm) activates PI(3)kinase that phorphorylates Akt and eNOS in turn. Part of the stretch-induced signals would be caused by NO. NO is considered as a second messenger that can increase the open probability of the Ca²⁺ release channel RyR2 by nitrosylation [56]. Whether this NO dependent pathway would apply to activation SACs has to be clarified by future experiments.

Mechanosensitivity of the inward rectifying current I_K1

To study the possible mechanosensitivity of I_K1 one has to superfuse the cells with "normal" extracellular solutions, e.g., K⁺ ions have to present in both patch electrode and bath. Under those conditions, I_K1 generates the positive current hump at -60 mV to the N-shaped iv curve of the cell (Fig. 7, blue curve). Local axial stretch reduced this hump of outward current and reduced the slope of the iv curve, indicative for reduction of I_K1. Separation of I_K1 from other current components such as I_SAC is possible because of their peculiar voltage dependence: The conductance G_K1 deactivates with positive potentials (Fig. 7B, blue curve, scaled on the left ordinate), in contrast to G_SAC that is voltage-independent.

Figure 7

Mechanosensitivity of K⁺ currents. A: Iv curves of net membrane currents measured with ramp commands repolarizing the cell from +60 to -100 mV at a rate of 100 mV/s. Blue trace before, red trace during 10 μm stretch. Arrows mark the shift of resting potential (more...)

During control, the iv curve is determined by the current I_K1 (Fig. 7C). During axial stretch, the current is shifted to the negative direction (Fig. 7A, red curve). In part, this effect is due to the induction of I_SAC (Fig. 7D, blue curve). Stretch activation of I_SAC, however, should have increased the slope conductance at negative potentials whilst stretch had reduced the conductance. Also, the reduced hump of the N-shaped curve can not be explained by activation of I_SAC with a voltage-independent conductance G_SAC. With both arguments, we modelled the stretch-induced changes in the membrane currents by superimposition of a mechanically deactivated I_K1 on I_SAC and could fit the results (Fig. 7D). The results indicate that local stretch reduces G_K1 without changing its voltage-dependence. Such an effect could mean that the availability of the K1-channel is the parameter that is mechanosensitive. Unfortunately, G_K1 is reduced also by low concentrations of Gd³⁺, hence, a pharmacological separation between the two mechanosensitive currents I_SAC and I_K1 has not yet been performed. The observed mechanosensitivity of I_K1 contributes to the membrane depolarization of the diastolic potential and to the retardation of the repolarization phase 2. In line with the missing effect of axial stretch on this part of repolarization (Fig. 2B), the current recorded during axial stretch did not indicate a component due to reduced I_K1. In the embryonic cell cultures from rat or chicken, stretching the cell did not affect I_K1 [27, 61], however, in these non-adult cells I_K1 amounts only to a fraction of its counterpart in adult mammalian cardiomyocytes.

Mechanosensitivity of the outwardly rectifying currents I_Ko

The literature on reconstituted K⁺-channels describes that mechanical deformation can activate a K⁺ current through TREK-1 channels [50, 55]. Single channel analysis has demonstrated the activation of this channel by positive pressure (and by arachidonic acid) in rat ventricular myocytes [38]. We speculate that the stretch-induced increase in outward current seen at positive potentials (olive curve between Fig. 7C, D) could be caused by a channel like this. In addition, mechanical deformation has been described to activate the ATP-depletion K⁺ current I_K,ATP [69]. Also this K⁺ current could contribute to the increase in outward current at positive potentials, illustrated in Fig. 7A. In adult rat atrial myocytes, a stretch-activated single K⁺ channel (probably of the 2P domain K⁺ channel) could be analyzed in cell attached patches [54]. The mean conductance was 65 pS (-60 mV, 150 mM [K⁺]_o). The open probability of this channel increased with the amplitude of suction, saturating at -50 mm Hg. The channel did not activate immediately but after a latency of 100 ms. Time to peak was 500 ms, then, the current inactivated to a plateau of 20% within 1 s.

Results from fluorescence indicators

Enhanced Na⁺ influx through SACs is expected to increase the cytosolic concentration [Na⁺]_c. We have measured stretch-induced increments in [Na⁺]_c by means of fluorescence indicators and by means of electronprobe microanalysis (EPMA). Fig. 9 shows pseudoratiometric images of [Na⁺]_c, using Sodium Green® as the indicator. Before local axial stretch, the image was flat, i.e. the Na⁺ concentration within the cell was homogenous. A 5 μm stretch increased the fluorescence suggestive for increments in [Na⁺]_c. These increments grew with the period of stretch. During a 3 min stretch, fluorescence was on average 2-fold higher than before stretch. Most importantly, stretch induced spatial heterogeneities in [Na⁺]_c, i.e. there were patches with low and elevated [Na⁺]_c close together. Fig. 9C shows these patches with more than doubled fluorescence as white spots. Unexpectedly, these white spots occurred not only close to the stylus or close to the patch pipette but also far away from the glass tools. Calibration of the fluorescence signal is problematic. Our own ratiometric SBFI measurements yield 11 mM for the isolated murine cardiomyocytes (16 mM in the literature [6]), and during 3 min stretch the local concentration could increase up to 30 mM (white-colored patches in Fig. 9C).

Figure 9

Stretch-induced increments in free cytoplasmic Na⁺ concentration [Na⁺]_c analyzed by Sodium Green fluorescence in a camera imaging system. Guinea pig ventricular myocyte stimulated at 2 Hz. A: Image before stretch, position of the glass stylus (S) and (more...)

Electronprobe microanalysis

Electronprobe microanalysis (EPMA) can quantify the elemental concentrations with the high lateral resolution of the electron microscope. EPMA measures the sum of free and bound sodium concentration, we abbreviate this concentration with σ[Na]. σ[Na] was nearly two-fold higher than [Na⁺]_c estimated from fluorescence indicators. During control, σ[Na] was 23 ±1.8 mM in the peripheral and 17 ±0.8 mM in the central cytosol (see Fig. 11). In the 20 nm narrow subsarcolemmal space σ[Na] could be as high as 100 mM [64, 75]. A 3 min stretch of 5 μm increased σ[Na] in the peripheral cytosol to 48 ±5 mM and in the central cytosol to 29 ±2.4 mM (factors 2.1 and 1.7, respectively). When SACs were blocked by 30 μM streptomycin, stretch did no longer increase σ[Na] in the central cytosol, in the peripheral cytosol the stretch effect was strongly attenuated. The blocker of the Na⁺, H⁺-exchanger, Cariporide®, had no significant effect on the stretch effects on σ[Na] [43].

Figure 11

Analysis of stretch induced changes in the total concentration (σ) of sodium (σNa) and calcium (σCa) with electronprobe microanalysis (EPMA). A: longitudinal cryosection of murine ventricular myocytes frozen after stimulation without stretch (more...)

A 3 min stretch of 5 μm increased σ[Na] in all cell compartments. It increased σ[Na] in both nuclear matrix (factor 1.97) and nuclear envelope (factor 1.65). These changes were insensitive to the treatment of the cell with Cariporide® (factor 2.14 and 1.94, respectively). Stretch doubled σ[Na] in central mitochondria. σ[Na] of peripheral mitochondria was already high before stretch (21.6 ± 4.3 mM), this value was not further increased by stretch (21.2 ± 3.3 mM).

Currents linked to accumulation of [Na⁺]_c

Since Na⁺ ions are the predominant charge carriers of I_SAC, mechanical activation of I_SAC is expected to increase the cytosolic sodium concentration [Na⁺]_c (see below). The increase would become apparent only if the Na⁺, K⁺-ATPase would not completely compensate the extra Na⁺ ion inflow. The Na⁺, K⁺-ATPase generates a pump current I_P because the efflux of 3Na⁺ ions is coupled to the infflux of 2K⁺ ions. The stretch induced pump current could be measured. After a 3 min stretch period (Cs/Cs conditions to reduce superimposed current components), the net current did not recover directly to its respective value before stretch but with an outward current above control that persisted for approx. 2 min. This current was sensitive to the removal of Cs⁺ from or to the addition of strophanthidin to the bath. We interpret that stretch induced I_P belongs to the currents with secondary stretch sensitivity, because it is caused by the stretch induced Na⁺ accumulation.

Another process linked to [Na⁺]_c is the Na⁺, Ca²⁺-exchange. NCX produces an inward current I_x when Ca²⁺ ions are extruded at the expense of Na⁺ influx (coupling 1Ca²⁺ to 3Na⁺ ions). The stretch induced increase in [Na⁺]_c is expected to shift the reversal potential of I_x from -20 mV to more negative potentials. At the normal diastolic potential of e.g. -80 mV, stretch-induced accumulation of [Na⁺]_c should go along with an I_x that is reduced in amplitude but lasting for a longer period of time (see the tail current I_T in Fig. 3B). At the positive plateau potentials, the Ca²⁺ influx coupled to Na⁺ efflux should be more pronounced. With both mechanisms, accumulation of [Na⁺]_c and shift of the reversal potential of Na⁺, Ca²⁺-exchange contribute to the increased Ca²⁺ loading that happens when ventricular myocytes are stretched (see below).

The H⁺, Na⁺-exchanger NHE removes cytosolic protons from the cytosol on the expense of Na⁺ influx. Stretch induced intracellular Na⁺ accumulation should attenuate the driving force and the rate of H⁺ efflux, similar as described for NCX above. Since the coupling is electroneutral, no direct change of net membrane current can be measured. However, one expects a possible intracellular acidification that could modulate other conductances in turn.

Fluorescence indicators

Increments in [Ca²⁺]_c were among the first effects reported for local mechanical stretch (embryonic chick heart cells: [63]). Without simultaneous analysis of membrane currents for possible involvement of I_SAC, those experiments are relatively easy to perform, and stretch induced increments in [Ca²⁺]_c have been analyzed by both luminescence [2, 8, 12, 60] and fluorescence indicators. In isolated cardiomyocytes, increments in [Ca²⁺]_c due to axial stretch have been measured in guinea-pig ventricular myocytes [26, 78] and discussed in context with the stretch activation of SACs [16]. Fig. 10 compares signals measured by confocal microscopy. Every 2 ms, a new line was scanned along the axis of the cell. Comparison between A and B indicates that local stretch moved the upper cell edge upwardly. The contractions were induced by clamp steps depolarizing 40 ms from -45 to 0 mV. Contraction is seen as edge movement, before stretch the isotonic shortening peaks at 74 ms and falls with a half decay time of 66 ms. During stretch, contraction peaks already at 61 ms, and the half decay time is reduced to 40 ms. The linescans show the change in Fluo4-fluorescence in the pseudocolor code. The global Ca²⁺ transients during control peak within 28 ms to 900 nM and decay with a half times of 163 ms. During local stretch, the peak of 1000 nM is reached within 24 ms, and the decay is slightly faster (144 ms). More importantly, during local stretch sparks occurred during diastole (yellow arrows in Fig. 8B, average from 6 cells: NP_o =12 ±2 before and 20 ±4 during a 10 μm local stretch). A similar result, augmentation of peak amplitude and of number of sparks, has been reported for myocytes axially stretched in the tube [56]. Combining the stretch experiments with fluorescence measurements of NO and protein phosphorylation let the authors conclude that stretch phorphorylates and activates eNOS via activation of PI₃K and Akt, and NO would increase the open probability of ryanodine-receptor by nitrosylation.

Figure 10

Stretch-induced changes in the cytosolic Ca²⁺ concentration [Ca²⁺]_c as analyzed pseudoratiometrically in the confocal microscope. Mouse ventricular myocyte loaded by 5 μM Fluo-4AM, depolarized from -45 to 0 mV for 40 ms at 1 Hz. Upper panels: (more...)

Electronprobe microanalysis

Due to the huge number of cytosolic Ca²⁺ binding proteins the total Ca concentration σ[Ca] in the central cytosol is with a diastolic value of 0.4 ± 0.05 mM 4000-times larger than the free concentration. The ratio of σCa = 400 μM, to [Ca²⁺]_c = 0.1 μm is the buffering power of the cytosol of the ventricular cell. In the peripheral cytosol σ[Ca] was 0.57 ± 0.09 mM (Fig. 11). A 3 min local 5 μm stretch increased σ[Ca] 2-fold to 0.84 ± 0.07 mM in the central and 1.5-fold to 0.84 ± 0.09 in the peripheral cytosol (both changes significant). When stretch was repeated after block of SACs with 30 μM streptomycin, the effect of stretch on σ[Ca] was prevented in the periphery, however only attenuated in the centre of the cell. The latter result may suggest that local mechanical deformation changes σ[Ca] not only via Ca²⁺ influx through streptomycin sensitive SACs (most pronounced in the cell periphery) but also through augmented Ca²⁺ release from the SR (see above).

Ca-compartments

Stretch increased σ[Ca] not only in the cytosol but also in the nuclear matrix from 0.18 ± 0.06 to 0.30 ± 0.05 mM (Fig. 11). It is unlikely that this increase is due to a Ca²⁺ release from the nuclear envelope because in this compartment σ[Ca] increased during local stretch from 0.93 ±0.15 mM to 1.53 ±0.26 mM (significant, factor 1.64). The stretch-induced Ca-increment in these two compartments was not changed when Na⁺- and Ca²⁺-influx through SACs was blocked with 30 μM streptomycin. Hence, stretch may augment the σ[Ca] in nuclear matrix and envelope by modulating Ca²⁺ release from the SR and/or inhibiting Ca²⁺ extrusion via NCX to the extracellular space. There might be a stretch effect on σ[Ca] in the mitochondria, however, these compartments have so low σ[Ca] that significance was not achieved because of the detection limit of EPMA for Ca.

Axial stretch and the passive mechanical properties of the cardiomyocyte

In vivo, the sarcomere length increases from peak systole (1.65 μm) to lack (1.85 μm) and to end of diastole (2.2 μm) by approx. 30%. The passive and active mechanical properties of single cardiomyocytes have been reviewed by Brady [9]. More recently, it has been shown that passive tension is generated largely by the extracellular matrix (collagen) and by the sarcomeric protein titin [18]. In comparison with the sarcomeric proteins, elements of the cytoskeleton, such as desmin intermediate filaments and microtubules, contribute only little to the passive tension [18]. Studies using the carbon fibre technique have revealed many similarities between the mechanical performance of single cell and multicellular preparations. For example, the time course of relaxation was load dependent, stretch prolonged the time course of relaxation in mammalian heart trabeculae and myocytes [62, 77].

The Starling mechanism

Stretch increases in the twitch force of contraction by a length-dependent change in muscle geometry (overlap and lattice spacing between thick and thin filaments) and by increase the number of formed crossbridges, amplified by a tension-induced increase in the sensitivity of troponin C (TnC) for Ca²⁺ [1, 15]. The expected stretch-induced increments in the [Ca²⁺]_c transient or in the Ca²⁺ influx through L-type Ca²⁺ channels were not detected [26]. A possibility would be that stretch increases the Ca²⁺ binding to TnC during diastole (increased systolic shortening), that during diastole extra Ca²⁺ ions will dissociate from TnC and will be extruded via NCX [35]. Since the Ca²⁺ efflux goes along with a negative current I_x, the plateau of the action potential will be prolonged [10, 41].

The Anrep effect

The immediate stretch-induced increase in force is followed by a slow increase in twitch force over the next 10 –15 min [1]. The Anrep effect is present in single cells, in terms of contraction and [Ca²⁺]_c transients [77].

Results on the contribution of [Na⁺]_c were at controversy. When measured with the fluorescent indicator SBFI, [Na⁺]_c did not increase in cells during the Anrep effect [26]. This result contrasts with other results that are listed in section 8. Also multicellular trabeculae show a stretch-induced increase in [Na⁺]_c during the Anrep effect [70].

The importance of Na⁺ influx through SACs for the stretch-induced [Na⁺]_c accumulation, however, has not yet been widely acknowledged. Instead, an autokrine -parakrine effect was suggested to increase [Na⁺]_c [3]: Stretch should stimulate the release of angiotensin II from the cardiomyocytes, interaction with ATI receptors should stimulate the secretion of endothelin I in turn. ETI binding to its G-protein coupled receptor should stimulate the activity of NHE with the consequence of an increase in [Na⁺]_c, and via NCX of [Ca²⁺]_c. The results are essentially based on the effectiveness of a series of drugs that interact with the individual steps of this cascade [3]. When the effects of local stretch were re-studied in isolated ventricular myocytes, block of NHE with Cariporide® did not interfere with the stretch-induced accumulation of σ[Na], in contrast to streptomycin that blocked SACs and the increase in σ[Na] [43]. Also, the isolated cardiomyocyte is surrounded by a huge build of solution, and the distance to the neighboured cell is so large that parakrine effects are difficult to imagine. Thus, there might be different signalling pathways induced by local stretch of the isolated myocyte and the axial stretch of multicellular specimen.

Stretch-induced arrhythmias

There is no evidence that the rhythmic 30% lengthening of the sarcomeres from peak systole (1.65 μm) to end diastole (2.20 μm) induces arrhythmias. This is in accord with the single cell experiments that applied homogenous stretch. As reported in section 3, homogenous axial stretch increased the duration of the action potential only less than 20%. Since this small effect could be blocked not only by streptomycin but also by loading the cell with the Ca²⁺ chelator BAPTA, it was postulated that this stretch induced APD prolongation is due to electrogenic efflux of Ca²⁺ ions via NCX ([10] see above).

If the large homogenous stretch has so little consequences for electrophysiology in contrast with the big effects of local stretch, most of the channel modulation might be caused by the shear stress component of the local stretch. This conclusion is not in contradiction to the results from single channel analysis because these experiments do apply stretch with large inhomogeneities (for review [28, 59]). It is in line with the clinical observation that ventricular arrhythmias due to a chest wall trauma (commotio cordis) correlate with the peak of diastolic left ventricular pressure that can be increased by the impact to 350 mm Hg [49]. This fast and local event is supposed to locally stretch some myocyte. The present results suggest that these arrhythmias are caused not only by activation of K⁺_ATP-channels [69] but by other mechanosensitive currents such as I_SAC, I_K1 and I_Ko. We have shown that local stretch changes the action potential in a characteristical fashion (Fig. 2C, D), it shortened the APD(0mV) and prolonged the APD(-80 mV). Further, local stretch depolarized the cell and could induce extra systoles. The single cell turned out as a model where activation of I_SAC, activation of I_Ko and deactivation of I_K1 were found to contribute to these changes.

The single cell studies on the mechanosensitivity of electrical activity have produced new findings, in particular the identification of stretch-activated and stretch-deactivated channels. The ability to record mechanical and electrical activity simultaneously with changes in the concentrations [Ca²⁺]_c and [Na⁺]_c is likely to be a powerful tool in the study of the mechanoelectric feedback. The challenge for the future is on one side to characterize better the force components of the mechanical stimuli and the intracellular signalling cascades activated by them. On the other side, forthcoming analysis has to relate the single cell/single channel findings to the observations made in the whole heart.