INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

MECHANICAL STRESS changes the electrophysiological properties of the heart, a phenomenon known as mechanoelectric feedback (25). Stretching intact hearts or excised muscle can raise the beat rate (2, 3), cause diastolic depolarization (30), change the action potential configuration (7, 24), and induce arrhythmias (17). Stretch-activated channels (SACs), considered to be the origin of mechanoelectric transduction (21, 37, 53), are found in many cardiomyocytes, including those from molluscan ventricle (40), chick embryo ventricle (20, 33), rat atrium (22, 50) and ventricle (8), rabbit sinoatrium and atrium (12), guinea pig ventricle (6, 39), and pig atrium (19). Most of the above recordings were made with the cell-attached single-channel patch-clamp technique, in which the open probability (P_o) of the channels increased with negative pressure applied to the patch pipette. None of the above data were obtained from adult ventricular cells, and there are almost no data on whole cell responses of cardiocytes to stretch under voltage clamp.

To record whole cell mechanosensitive currents under stretch, the cells must not only be voltage clamped but also stretched without damage. Stretching intact, isolated cells without damage has proven to be extremely difficult (5, 11), and only one paper has shown whole cell currents (39). In some experiments hydrostatic or osmotic pressure was used as a mechanical stimulus to inflate the cells, but it is unlikely that these stimuli are equivalent to axial stretch (20). Hagiwara et al. (12) evoked mechanosensitive Cl currents by inflating the cells through the clamping pipette, whereas Sorota (44) and Tseng (47) found an increase of Cl permeability in response to hypotonic swelling. Attempting to evoke axial strain, Sasaki et al. (39) attached cells to a coverslip or to a fire-polished glass tool and pulled on the other end with another glass tool or suction pipette. Currents were recorded with a separate pipette and had a reversal potential of 15 mV in physiological saline. Hu and Sachs (20) recorded currents in chick heart cells through a perforated patch and stimulated them by compressing the rounded cells with a second pipette. In agreement with the results of Sasaki et al. (39), Hu and Sachs found a mechanosensitive cation current reversing at 16 mV. In none of these papers was strain under reliable control.

In this paper, we present evidence of a gadolinium-sensitive whole cell stretch-activated current (I_stretch) in adult rat ventricular cells under controlled strain. The cells were pulled with a pair of concentric pipettes as described by Palmer et al. (29). The axial strain was measured from calibrated displacements of the ends of the cell and from Fourier transforms of the sarcomere spacing. These measurements formed the basis for quantifying the relationship between strain and current.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Cell preparation. Ventricular myocytes were enzymatically isolated by retrograde perfusion of the heart (28, 49). Briefly, Sprague-Dawley (2-3 mo old) rats were injected with heparin (2,000 U/kg) and then Nembutal (60 mg/kg). When the rat was anesthetized, the heart was quickly excised, cannulated through the aorta in cold Tyrode buffer, and then mounted on a dual-channel tube Langendorff perfusion apparatus. Perfusion of the heart proceeded at 37°C for 10 min in Tyrode solution, 4.5 min in Ca²⁺-free Tyrode solution, 30 min in the enzyme solution, and 5 min in low-Ca²⁺ Tyrode solution. All perfusion solutions were equilibrated with 100% oxygen. The enzyme solution was limited to 60 ml and allowed to recirculate. After perfusion, the ventricle was cut off and minced. Cells were dispersed from the tissue by agitation, filtered into Tyrode solution through a 200-µm-mesh net, and stored at 4°C until use. All experiments were done within 1-20 h after isolation.

Dye loading. Isolated cells were loaded for 30 min at room temperature in normal Tyrode solution containing 2 µM fluo 3-AM (Molecular Probes) and 0.2% Pluronic-127 (Molecular Probes) (41). Cells were then rinsed three times in saline and left for 20 min to further hydrolyze the ester form of the dye.

Cell stretch and patch clamp. Isolated rod-shaped ventricular cells with clear sarcomeres were held by two concentric glass pipettes (29), with the inner pipette serving as a stop to prevent the cell from being sucked up the outer pipette. The outer pipette was pulled from a glass capillary (ID 1.0 mm, OD 1.5 mm; Drummond Scientific) with an inner tip diameter of ~15 µm. The inner pipette was made from a glass capillary (ID 0.5 mm, OD 1.0 mm; Dagan) with an outer tip diameter of ~12 µm. The inner pipette was inserted into the outer pipette by a manipulator, leaving a gap of ~8 µm to the tip of the outer pipette. The tip of the outside pipette was then cut by fusion of the tip to the filament of a microforge as described by Hilgemann (18). The cut end was lightly fire-polished so that the tips of the inside and outside pipettes were forged together and formed a cup to hold the cell.

Data recording and analysis. Standard whole cell recording methods (14) were used to clamp the membrane voltage and record currents. The fire-polished patch pipette had a tip ID of 1-2 µm, with a resistance of 0.5-2.0 M when the pipette was filled with high-K⁺ pipette solution. The seal resistance was usually >2.0 G. Cell capacitance was derived by fitting the transient currents generated by a voltage pulse after the whole cell configuration was formed (26).

View larger version (141K): [in this window] [in a new window]   Fig. 1.   Cell stretching and measurement of sarcomere length (SL). A: cell was attached to 2 concentric pipettes (left and right) and voltage-clamped by a third (middle). Right concentric pipette moved right to stretch cell, and patch pipette moved with local strain to reduce stress around tip. Images were sampled by a frame grabber at specific times. White dot indicates a fiducial fixed point. B: diffraction patterns of cell in A. Spatial frequency (fcell) corresponds to SL. Actual SLs in a-d are 1.78, 1.84, 1.84, and 1.78 µm, respectively, obtained by comparison with a calibrated stage micrometer.

Solutions. The Tyrode solution contained (mM) 137 NaCl, 5.4 KCl, 0.5 MgCl₂, 1.8 CaCl₂, 10 HEPES, and 5.0 glucose, pH 7.4 with NaOH. The Ca²⁺-free solution was Tyrode solution without added Ca²⁺. For the low-Ca²⁺ Tyrode solution we reduced Ca²⁺ in Tyrode solution from 1.8 to 0.2 mM. Tris-Tyrode solution was Tyrode solution with Na⁺ replaced by Tris⁺. The enzyme solution was 60 ml of Ca²⁺-free Tyrode plus 30 mg of collagenase A (Boehringer Mannheim) and 6 mg Protease XIV (Sigma). The pipette solution was (mM) 130 KCl, 10 NaCl, 5 MgCl₂, 11 EGTA, 1 CaCl₂, and 10 HEPES, pH 7.4 with KOH. In some experiments, Cl was replaced by F as indicated.

Mathematical modeling. I_stretch was modeled using a simple linear nonspecific current with two components, the Na⁺-carrying element and the K⁺-carrying element

where E_Na and E_K are the reversal potentials for sodium and potassium respectively, E_m is the membrane potential, and _stretch,Na and _stretch,K are the whole cell conductances of the stretch current to potassium and sodium ions, respectively. The components were kept separate to ease calculation of ion concentration changes within the cell as well as the electrogenic effect of the current.

1

1

1

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Stretch increases action potential duration and depolarizes resting potential. In normal-Ca²⁺ solution, heart cells are apt to contract when touched by a glass pipette, particularly when subjected to suction. We presume that this represents action of SACs. To check the effect of stretch on cell excitability, we recorded the membrane potential while stretching the cell. In current-clamp mode, pulses of 1 ms and 2 nA repeated every 5 s could elicit action potentials. After the action potential was stable, we pulled on one of the concentric pipettes to stretch the cell. The displacement was ~5 µm and lasted 20 s. The pipette was then returned to its original position. Figure 2 shows action potentials recorded during stretching.

View larger version (17K): [in this window] [in a new window]   Fig. 2.   Longitudinal stretch increased action potential duration (APD) and depolarized the cell. Action potentials were recorded in current-clamp mode with 1 ms, 2 nA stimulation at 0.2 Hz. A: action potentials (nos. refer to ordinal location of stimulus: 3, before stretch; 6 and 8, during stretch; 10, after stretch). B: APD at both 90% and 50% repolarization (APD90 and APD50) increased during stretching (5-8) and returned close to control duration after strain was released (9-14). C: resting potential was depolarized during stretching. Cell was in low-calcium (0.2 mM) Tyrode solution.

Longitudinal stretch-activated inward currents. When the membrane potential was held at 100 mV in Tyrode solution, stretching the cell elicited an inward current (Fig. 3). This current activated without visible delay after initiation of stretch and was maintained during stretch. After release, the current returned to its prestretch level. While the cell remained attached, repeated stretching gave nearly the same response. For example, in one of our more extensive experiments (n = 4), the mean current was 1.03 ± 0.18 pA/pF at 100 mV. Currents elicited in Ca²⁺-free or low-Ca²⁺ (0.2 mM) solutions were similar to those obtained in normal Tyrode solution. Because the cells were much more likely to contract after being attached to the pulling pipettes in normal-Ca²⁺ solution than in low-Ca²⁺ solutions, we usually conducted experiments in Ca²⁺-free or low-Ca²⁺ Tyrode solution.

View larger version (18K): [in this window] [in a new window]   Fig. 3.   Stretch-induced inward current of a rat ventricular cell. A: trapezoidal stimulus pulse that drove piezomanipulator. Stretching paradigm was trapezoidal to reduce parasitic mechanical oscillation. a-d, time at which images were taken corresponding to Fig. 1. B: whole cell inward current that started without obvious delay and was sustained during stretch; cell membrane potential was held at 100 mV. Bath contained normal Tyrode solution; cell capacitance was 87 pF.

View larger version (41K): [in this window] [in a new window]   Fig. 4.   Fluo 3 Ca2+ images did not show local leakage during stretch in normal Tyrode solution. A: SIT camera image of a fluo 3-labeled cell being held by pipettes. A small amount of transmitted light was added to epi-illumination to provide contrast that makes pipettes visible. Small squares indicate areas of interest used to measure fluo 3 signals over time. B: fluorescence images of cell before stretch (transmitted illumination removed). C: fluorescence image during stretch (as in B). D: fluo 3 brightness signals (arbitrary units) from small regions of interest marked in A. White is value 0, and black is value 255. Stretch was applied at 3 min. Cell was loaded with 2 µM fluo 3 for 30 min in normal Tyrode solution. Difference in absolute intensities between ends and middle may reflect compression of cell by pipettes forcing more of the cell to be in the effective focal volume.

View larger version (15K): [in this window] [in a new window]   Fig. 5.   Stretch-activated inward current blocked by 100 µM Gd3+. A: stretching protocol. B: control. C: with Gd3+. D: partial recovery after 5 min of washing. Cell was in normal Tyrode solution at a holding potential of 100 mV.

View larger version (22K): [in this window] [in a new window]   Fig. 6.   Octanol does not significantly affect currents. A: stretching protocol. B: control. C: with 100 µM octanol in bath. Cell was in normal Tyrode solution at a holding potential of 100 mV.

I_stretch has a reversal potential of 6 mV. We examined the voltage dependence of the currents in Tyrode solution by applying test pulses between 120 and +20 mV from a holding potential of 100 mV. Because stretching the same cell more than four times without losing the attachment was very difficult, we collected the data from four different cells and normalized the currents by the cell capacitance. The cells were stretched 5 µm, equivalent to ~3% strain. All currents were calculated as the mean current between the rising and falling phases. The mean current-voltage (I-V) curve (Fig. 7) showed an essentially linear relationship with a reversal potential of 6 mV.

View larger version (12K): [in this window] [in a new window]   Fig. 7.   Current-voltage (I-V) relationship, normalized to cell capacitance, is nearly linear, with a reversal potential of 6 mV in Tyrode solution. Cells were stretched by 5 µm, equivalent to ~3% strain; mean resting SL was 1.79 ± 0.02 µm (n = 4). Line, linear fit of experimental data by I (pA/pF) = 0.012V (mV) + 0.07.

I_stretch increase with SL. By programming the voltage sent to the piezomanipulators, we could apply different global strains to the cell. Video images of cells before and after stretching were grabbed into the computer, and mean SL was computed from the power spectrum of the images. Figure 8 shows that in the narrow range of strain available, the change in mechanosensitive current was approximately linear with both SL and strain. Both SL and SL strain were fit to the current by linear regression. At 100 mV, the former follows the relationship I (pA/pF) = 8.3  5.0SL (µm), and the latter follows I (pA/pF) = 0.30SL%. The first equation predicts that SAC currents will persist to 1.66 µm. The 95% confidence limits, however, include zero current at longer SLs, and we do not trust the precision of this prediction. There are probably small-scale nonuniformities in SL that do not show up in our first-order analysis of the diffraction pattern. The sample size and the optical resolution in three dimensions limit the effective resolution of the SL. The net result of these systematic and random errors is scatter in the data. In Fig. 8B, the large fraction of data points outside the 95% confidence intervals suggests some significant nonrandom errors such as a nonuniform SL distribution.

View larger version (12K): [in this window] [in a new window]   Fig. 8.   Relationship between SL and mechanically activated current at a holding potential of 100 mV. A: linear regression of SL (µm) vs. current density (pA/pF) by equation Y = aX + b gives relationship I = 8.3  5.0SL. B: linear regression of strain, SL% (% change of SL) vs. current density (pA/pF) by equation Y = aX gives relationship I = 0.30SL%. Solid line, mean regression; dashed lines, 95% confidence limits calculated in Origin 5.0. Confidence limits are based on Gaussian distributions of errors. Data were collected from 6 different cells labeled by different symbols.

Na⁺ is the main carrier of mechanically sensitive currents at resting potential. Because I_stretch was inward at 100 mV, that current must be carried by an influx of cations, an efflux of anions, or a mixture of the two. Replacing Cl in the pipette solution with F produced similar currents (n = 5, see Fig. 9 for example), suggesting that the current was carried by cations rather than anions (13).

View larger version (14K): [in this window] [in a new window]   Fig. 9.   Replacing Cl in pipette solution with F produced a typical inward current, suggesting that stretch-activated current is not a Cl efflux. A: stretching protocol. B: stretch-activated current. Cell capacitance was 72 pF; holding potential was 100 mV. Cell was in normal Tyrode solution.

View larger version (18K): [in this window] [in a new window]   Fig. 10.   Stretch-activated current was reduced when extracellular Na+ was replaced by Tris+. A: stretching protocol. B: stretch-induced currents in normal Tyrode solution. Bath contained Tyrode, as did perfusate. Beginning of superfusion flow produced a transient upward-going current caused by changes in liquid junction potential. C: currents with cell superfused with Tris-Tyrode solution (Na+ replaced by Tris+). Stretch-induced current is greatly reduced, suggesting that most of it is carried by Na+. Exposure of cell to Tris-saline produced a large outward background current. Cell membrane was held at 100 mV, and Ca2+ concentration was 0.2 mM in bath solution.

Modeling I_stretch. The experimentally obtained I-V curve for I_stretch was well fit by a straight line representing the equation

where _stretch,K = 0.9 nS and _stretch,Na = 1.17 nS, given the model rat's nominal whole capacitance of 200 pF (see Fig. 7). These values were used with the equation above to test the effect of I_stretch on the rat action potential and compare it with the experimentally obtained results.

View larger version (7K): [in this window] [in a new window]   Fig. 11.   Simulated rat ventricular action potentials at 0.2-Hz stimulation frequency with 10 mM EGTA inside cell and 0.2 mM Ca2+ extracellular solution. Solid line, control; dashed line, stretch. Resting potential is elevated and action potential is prolonged with stretch-activated current.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

This is the first study of stretch-activated whole cell currents in mammalian cardiocytes under controlled strain. We were able to show that in rat ventricular cells stretch activated a gadolinium-sensitive, noninactivating, inward current. Our data are similar to those published for other heart cells (20, 39) and smooth muscle cells (9, 52).

SACs in heart cells have been reported to be nonselective cation channels (8, 33) or Cl channels (12). In our experiments, I_stretch was not significantly different when the pipette solution contained Cl or F, suggesting that these currents are carried by cations. Further experiments are needed to define the selectivity more precisely. Removing external Ca²⁺ did not significantly affect the current, suggesting that Ca²⁺ does not act as a significant charge carrier or a second messenger, although Ca²⁺ is probably a permeant ion. Replacing Na⁺ with Tris⁺ reduced the current, although the blockage was not complete. I_stretch in rat heart cells appears to be cation selective, with Na⁺ acting as the major permeant ion at resting potential. Wellner and Isenberg (52) reported that in guinea pig smooth muscle cells, stretch increased a voltage-activated potassium current and reduced a calcium current. In heart cells, however, there is no evidence to suggest that stretching has any effect on voltage-sensitive channels (20, 39). Because of the experimental difficulty with our preparation, we did not directly check the effect of stretching on voltage-sensitive currents. However, we did observe that changing the membrane potential to 40 mV and removing Ca²⁺ from the bath solution produced little difference in the time course of the mechanically activated currents. Thus it is unlikely that voltage-dependent K⁺, Na⁺, or Ca²⁺ channels are responsible for the mechanosensitive currents we observed.

Heart cells shorten >10% during contraction, but they are not readily stretched. In our system, 2-5% strain was possible. In some experiments, we moved the stretching pipette 10% of the cell length, but the attached cells would either contract immediately or slip out of the holding micropipette. In the latter case, the cells usually contracted into a ball in a few seconds. With a resting SL of 1.77 µm, our results showed that <5% strain was sufficient to evoke stretch-sensitive currents. Comparing the passive elastic properties from detergent-skinned isolated cells and intact cardiac tissue, Brady (4) suggested that intracellular structures may contribute measurably to total cardiac passive elasticity at SL < 2.2 µm, whereas extracellular elements form the major limitation at more extended lengths. Furthermore, he suggested that these intracellular structures were probably related to the cytoskeleton rather than membrane elements (5). Our data are consistent with a model in which stress in the membrane is coupled through the intracellular cytoskeleton to the channels.

How much current is contributed by SACs at "resting" SLs is difficult to determine and could only be measured with a specific inhibitor, which Gd³⁺ is not. Our definition of a mechanically sensitive current is one that changed with stretch of the cell, so that it is only a differential measurement. In Fig. 8 we plotted current versus SL during stretch and versus SL as strain. Taken literally, Fig. 8A suggests that at SL = 1.75 µm there would be ~1 pA/pF of inward current from SACs. This plot is of necessity made from differential data arising from different cells, and without allowance for a variation in SL throughout the cell (only the 1st-order diffraction line was used to define SL) we cannot place much confidence in the amplitude of the resting current. Furthermore, the resting current in isolated cells is unlikely to be the same as in cells in situ, where the normal extracellular contacts are in place.

We made multiple attempts to record stretch-activated ion channels from tight seal patches on the rat ventricular cells, but we never recorded single channel activity. Reviewing the literature, we found that all the reports of single channel stretch-activated activity in heart cells were obtained from neonatal (8, 20, 22), atrial (19, 50), or cultured (33, 38) cells. There are no data on freshly isolated ventricular cells. There appear to be three possible explanations for this absence of single channel data.

First, the channel density may be low, so the chance of catching one channel in a pipette is small. Is this reasonable? Our maximal currents were ~1 pA/pF at 100 mV. This current corresponds to a product of the unitary current and the probability (P_o) of being open. If the single channel conductance had a typical value of ~25pS, the single channel current would be ~2.5 pA/channel at 100 mV. There is 1 pF of capacitance for every 100 µm² of membrane, so we would expect that the minimal density (if P_o =1) is 1 channel/40 µm². Because cell-attached patches have areas of 10-30 µm² (34, 43), we might expect to see a channel in every other patch. P_o is probably <<1 in the whole cell experiments because the strains were small and there was no hint of saturation in the plot of current versus SL. Consequently, there should have been a higher density of channels, and we should have seen them in most patches. SACs typically occur with a density of ~1-3/patch. It does not seem likely, therefore, that the absence of single channel activity came from a low channel density.

A second possibility is that SACs are not in the surface plasmalemma but are located in the T system and hence invisible to plasmalemmal patching but not to whole cell stretching. An intracellular location for the channels, although an experimental nightmare, has conceptual appeal because strain may be better sensed in the contractile cell interior than in the convoluted plasma membrane. This interior membrane explanation would fit with the sensitivity of the cells to suction applied to the pulling pipettes and with fluorescent imaging showing Gd³⁺-sensitive Ca²⁺ waves induced at the site of mechanical deformation (W. J. Sigurdson and F. Sachs, unpublished observation).

The third possibility is that formation of the patch disrupts channel activity in these cells. Patch formation is a major perturbation of the mechanical properties of the membrane and cortical cytoskeleton (1, 16, 36, 42, 43, 51). It is possible that some important structure in the adult ventricular cortex is disrupted under stress.

Physiologically, the effect of stretch is most likely expressed through changes in the action potential, although it also will affect the filling of Ca²⁺ stores. In many recordings of action potentials from intact tissue under stretch, the duration of the plateau is reduced and the tail of repolarization (phase 3) increases, leading to a crossover, or apparent reversal potential of the mechanosensitive current, at about 20 mV (35, 58). In intact rabbit heart, a long static stretch created by inflating a balloon in the left ventricle extended the APD. The peak amplitude of the stretch-activated depolarization from rest and repolarization from the plateau exhibited a linear relationship to voltage and volume change (58). In single guinea pig cardiac myocytes, a 3% strain did not affect the resting potential but did decrease the APD (10, 54).

The rat ventricular action potential is significantly different from that of the rabbit and guinea pig, because it lacks the prolonged plateau and, during the contraction cycle, has a much greater reliance on calcium released from the SR rather than plasmalemmal calcium entry (46). As a test of whether the observed currents could account for the effects of stretch on the action potential, we simulated the rat action potential using the HEART program from Oxsoft. We introduced a stretch-activated current, I_stretch, that was permeable to potassium and sodium ions and closely fit the experimental I-V curve.

The addition of this simple current could account for the depolarization of the resting potential and the observed changes in APD. At the negative potentials of the rat plateau, the reversal potential of the stretch current near 0 mV caused a net inward current and therefore prolonged and depolarized the action potential throughout its course. In cells with a high plateau and a lower reversal potential of I_stretch (approximately 20 mV), such as those of guinea pig and rabbit, there is a shortening of the plateau but prolongation of APD₉₀ and a crossover potential during phase 3. White et al. (54) observed a stretch-induced reduction in APD in guinea pig, possibly reflecting the differences in the two preparations.

Although stretching single cells would appear to be the most reductionist level for studying the effects of stretch on whole cells, there is a fundamental problem of interpretation that is not readily resolved: What are we pulling on? The response of cells to mechanical deformation depends, in general, on which chemical groups are being distorted. For instance, in a study of cultured vascular smooth muscle cells subjected to periodic strain, Wilson and Kaczmarek (55) found that production and secretion of platelet-derived growth factor and DNA depended on the chemical composition of the substrate. Collagen, fibronectin, and vitronectin were effective, but little response was observed on elastin or laminin. Similarly, if the cells were cultured on pronectin or laminin, cyclic strain caused differential expression of mitogen-activated protein and amino terminal kinase (32). In fibroblasts, mechanically induced increases in cell Ca²⁺ occurred when ₂- or ₁-integrin subunits were stressed but not the transferrin receptor (31). L-type Ca²⁺-channel currents can be activated or inhibited by different ligands for the integrins (56). These kinds of data warn against extrapolating from isolated cells to cells in their normal environment. Aside from the technical difficulties of stretching isolated cells, it is important to get data from cells in their native mechanical environment. Only with that data in hand can we properly extend the studies on isolated cells to discover which ligands produce the responses observed in vivo.