contraction in cardiac muscle is initiated by a release of calcium into the sarcoplasm that subsequently binds to the troponin complex, causing a series of interprotein interactions and resulting in a cyclic interaction between actin and myosin (see review, Ref. 33). The steric blocking model for regulation of contraction (see review, Ref. 29) involves troponin and tropomyosin that lie along the thin filament. When calcium binds to troponin, tropomyosin moves to a position that does not block the actomyosin interaction. On calcium removal, tropomyosin moves back to a position that blocks the actomyosin interaction. Whereas this model explains physiological contraction, it does not explain the properties of the system at low MgATP concentrations ([MgATP]). Early work by Bremel and Weber (3) showed that activation occurred without calcium if [MgATP] was low in reconstituted systems, and Godt (10) showed that skinned fibers would develop tension at low [MgATP] without calcium. Recent biochemical studies by Geeves and colleagues (16) led to a different model for thin filament regulation. In this model, there are three states: a blocked state that cannot bind myosin, a closed state that binds myosin weakly, and an open state that binds myosin strongly and allows force development. The transition from the blocked to the closed state is favored by calcium binding to troponin, whereas the transition from the closed to the open state is favored by strong cross-bridge binding. The important caveat is that activation requires cross-bridge binding. Light microscopic studies with myofibrils (32) and electron micrographic reconstruction studies on regulated thin filaments (35) support this model. These more current studies ascribe a central role to myosin-head binding in the activation process.

In addition to the proposed direct role of cross bridges in activation, there is an additional effect of cross bridges on calcium binding of troponin C (TnC) to the thin filament. Bremel and Weber (3) showed that rigor cross bridges increased calcium binding to reconstituted thin filaments. Studies with skinned cardiac preparations showed similar results, with increases in cross-bridge attachment being associated with increased calcium binding (18). Additional studies (14), in which conformationally sensitive probes on cardiac TnC in trabeculae were used, showed that cycling cross bridges as well as rigor cross bridges influence TnC structure. These studies suggest that cross bridges play an additional role in activation in the form of enchancing calcium binding and enhance the cooperativity of the thin filament system (28).

Cross bridge-dependent activation of the thin filament may have pathological implications to the energy-deprived (anoxic or ischemic) heart. During this condition, there is a series of events occurring that can result in irreversible damage (20). In the early phase of ischemia, arrhythmias develop, followed by complete loss of cyclical contractions (see review, Ref. 1). Subsequently, there is a quiescent period followed by a gradual increase in diastolic pressure or contracture of the heart associated with a decline in ATP (17, 23) that ultimately leads to the “stone heart” condition (5). Two possible mechanisms responsible for the ischemic contracture have been proposed (1): 1) a slow release of calcium, resulting in activation of the thin filament, and2) cross bridge-dependent activation, which can occur independent of the calcium level but that requires low levels of MgATP. The latter mechanism has been termed rigor activation (3) because it requires a low level of rigor cross bridges. Studies on the calcium mechanism showed that a rise in cytosolic calcium followed rather than preceded ischemic contracture, suggesting that rigor activation may be responsible for ischemic contracture (21). Studies on skinned myocyte preparations showed an increase in tension with decreases in MgATP levels at low calcium, supporting the hypothesis that cross bridge-dependent activation occurs in cardiac muscle (26).

To further investigate cross bridge-dependent activation of the thin filament and its potential role in anoxic/ischemic contracture, we studied the ATPase activity of cardiac myofibrils over a range of MgATP levels and at different pH levels. We also investigated the extent of shortening of the myofibrils using fluorescence microscopy to determine whether changes in ATPase activity on cross bridge-dependent activation were associated with mechanical changes in the sarcomeres. These studies with myofibrils afforded a shorter diffusion distance for MgATP than with skinned fibers and allowed for both biochemical and structural studies. Our data show that cardiac myofibrils are susceptible to cross bridge-dependent activation of ATPase activity at low calcium and that the highest ATPase activities were associated with the greatest extent of sarcomere shortening. These studies support the hypothesis that contracture in the energy-deprived heart may be the result of cross bridge-dependent activation of the thin filaments and point to the important role of strong cross-bridge binding in thin filament activation in both physiological and pathological conditions.

MATERIALS AND METHODS

Myofibril isolation.

Hearts were isolated from male Sprague-Dawley rats (300–400 g) and then transferred to ice-cold relaxation buffer [75 mM KCl, 50 mM 2,3-butanedione monoxime, 10 mM imidazole (pH 7.2), 2 mM MgCl₂, 2 mM EGTA, 2 mM ATP, 1 mM dithiothreitol (DTT), 1 mM benzamidine-HCl, 1 mM NaN₃, 0.1 mM phenylmethylsulfonyl fluoride, 1 μg/ml leupeptin, 1 μg/ml pepstatin, and 1% Triton X-100] and rinsed of blood. The atria and aorta were removed, and the ventricular tissue was minced with scissors in relaxation buffer. The minced tissue was homogenized in 5 vol of relaxation buffer per heart by using two 10-s bursts with a Polytron homogenizer at a speed setting of 4. The homogenate was filtered through one layer of cheesecloth into a Dounce homogenizer (to remove large connective tissue pieces) and homogenized by 100 strokes with a loose pestle. The myofibrils and fiber pieces were collected by centrifugation at 1,500 g for 5 min at 4°C. Myofibrils were resuspended in 10 vol of relaxation buffer, homogenized briefly with a loose pestle, and then filtered through four layers of cheesecloth. The suspension was homogenized by another 100 strokes with a tight pestle, and myofibrils were collected by centrifugation. The homogenization was repeated once more, giving a white pellet of myofibrils and fiber pieces. The resulting pellet was resuspended (5 ml/heart) in rigor buffer [75 mM KCl, 10 mM imidazole (pH 7.2), 2 mM MgCl₂, 2 mM EGTA, 1 mM NaN₃, and 1 mM DTT] with 0.5 mM ATP and then collected by centrifugation. The supernatant fraction was removed, and the pellets were rapidly resuspended in ice-cold rigor buffer plus 30 mM EDTA at 10 ml/heart, with one heart per centrifuge tube. This procedure rapidly induced the rigor state by chelation of MgCl₂with minimal sarcomere shortening. The myofibrils were collected by centrifugation and washed three times with rigor buffer. The final myofibril pellet was resuspended in glycerol rigor buffer with a final concentration of 50% glycerol, 75 mM KCl, 10 mM imidazole (pH 7.2), 2 mM MgCl₂, 2 mM EGTA, 1 mM NaN₃, and 4 mM DTT and was stored at −20°C. Before use, the myofibrils were washed with pCa 9.0 buffer (see Solutions) containing 1 mM DTT and 1 mg/ml BSA to remove glycerol. Myofibrillar protein concentration was measured with the biuret assay (13) using BSA as the standard.

Solutions.

The solutions used for the ATPase assays were made from base solutions of pCa 9.0 and 3.0 using the program of Fabiato (7) and the stability constants of Godt and Lindley (11). The base solution was 20 mM buffer [PIPES for pH 7.0 and MES for pH values of 6.4, 6.0, and 5.4], KCl to 200 mM ionic strength, 4 mM EGTA, 2 mM MgCl₂ (free), 2 mM creatine phosphate, 1 mM NaN₃, and either 0 (pCa 9.0) or 5 mM (pCa 3.0) CaCl₂. Different pCa levels were made by mixing pCa 9.0 and 3.0 solutions, whereas variations in MgATP levels were made by changing the total ATP concentration. The calcium-buffering ability of EGTA is less at lower pH levels, so the pCa 9.0 solutions were nominal rather than absolute, especially at pH levels <6.4.

ATPase assay.

The ATPase activity of myofibrils was measured using the phosphate assay of Carter and Karl (4) with modifications. The base conditions of the reaction buffer were the pCa solutions with DTT and BSA added to 1 mM and 1 mg/ml, respectively, and 10 μg/ml (3.5 U) of creatine phosphokinase. The absence of creatine kinase was found to result in a lack of a sharp peak of ATPase activity even though studies by others (34) suggest that the intrinsic creatine kinase is sufficient. This may have resulted from some oxidation of the intrinsic creatine kinase (24) during the myofibril isolation. Titration of these preparations with different levels of creatine kinase showed that no further enhancement of the peak activity was observed when creatine kinase was elevated above 3.5 U/ml. Variations in the buffer included pCa, pH, and MgATP levels. The reaction (at 23°C) was started by the addition of myofibrils to the reaction solution (0.2 mg/ml in a final volume of 100 μl). At specified times, the reaction was quenched by the addition of 20 μl of ice-cold 25% TCA and was held on ice before centrifugation. The protein was removed by centrifugation at 13,500 g for 15 s, and an aliquot was removed for analysis of phosphate. To minimize the hydrolysis of creatine phosphate during the phosphate assay, the sample was added to ice-cold molybdate-HCl solution, followed immediately by the addition of malachite green and mixing. This was immediately followed by addition of H₂SO₄and mixing, followed by incubation for a minimum of 1 h for color to develop. In myofibril-free samples, these modifications resulted in 0.25% hydrolysis of creatine phosphate compared with 16% when the assay is used at room temperature and with 2 min of incubation before the addition of H₂SO₄. Assays were done in a microplate with “in plate” standards prepared just before analysis of ATPase samples. The phosphate level was corrected for contamination in the buffers by measuring the phosphate level in samples with TCA added first, followed by the addition of myofibrils. Incubation time was varied when different pCa levels were used (5–30 min) and was 30 min for pCa 9.0 and the different pH levels. Myofibrils that were assayed at different pH levels were first washed with pH 7.0 buffer, the protein concentration was determined, and then samples were diluted at least 10-fold into the appropriate pH buffer before analysis.

Sarcomere length measurement.

To monitor the extent of sarcomere shortening, myofibrils were pretreated with fluorescent α-actinin to decorate the Z lines, allowing for measurement of sarcomere lengths using fluorescence microscopy (31). Rabbit skeletal muscle α-actinin was purified and labeled as described in Swartz (30) by using Texas Red 2 maleimide (Molecular Probes, Eugene, OR). The cardiac myofibrils were pretreated for 1–4 h with 0.5 μM Texas Red α-actinin and then treated with MgATP as was done for the ATPase assay. After 30 min of incubation with MgATP (0, 0.1, 1, 8, or 3,000 μM), the myofibrils were spread on a coverslip and fixed by the addition of 3% formaldehyde in reaction buffer. The myofibrils were fixed for 10 min, rinsed with rigor buffer, drained, mounted in 75% glycerol, 20 mM Tris (pH 9.0), 2 mM EGTA, 2 mM MgCl₂, and 1 mg/mlp-phenylenediamine, and sealed to the slide with nail polish. Images of the α-actinin-treated myofibrils were obtained with a Zeiss axiovert TV equipped with a narrow-band pass Texas Red filter, ×100 (NA 1.3) phase-contrast lens, and a Photometrics charge-coupled device camera (32). The camera was controlled via a Macintosh 840AV with a Matrox board using IPLab software (version 3.0; Scanalytics, Fairfax, VA). Both phase and fluorescent images were obtained, and the fluorescent images were used to determine sarcomere length. The distance between the Z lines was measured in three contiguous sarcomeres using IPLab software to give the myofibril average, and, for each treatment, 80 myofibrils were measured. This approach was used because of the poor resolution of the phase-contrast image of sarcomeres at <1.8 μm (31).

RESULTS

The rate of hydrolysis of ATP by cardiac myofibrils is dependent on both calcium and substrate level. Calcium influences the rate by increasing the rate of the calcium-activated actomyosin ATPase, whereas the substrate influences the activity via Michaelis-Menten kinetics. To more clearly show the influence of substrate on the activity at different calcium levels, the ATPase activity at different pCa levels was plotted as a function of [MgATP] in Fig.1. The first feature to note is that at low levels of MgATP (0.5–5 μM), calcium had little influence on the rate, whereas at higher levels, calcium did influence the rate. At the low levels of MgATP, the rate was likely substrate limited. The second feature is that there was a biphasic response of ATPase activity to substrate level for pCa levels of ≥5.5. At ∼10 μM MgATP, there was a peak in activity at pCa levels of 9.0, 7.0, and 6.0 in that higher and lower levels of MgATP resulted in lower ATPase activity. At pCa 5.5, the peak in activity was at ∼50 μM, and the rate also declined at higher levels of MgATP. The biphasic response was not observed at pCa 5.0 and 4.0. The highest activity was at pCa 4.0, reaching a value of 7.5 ×10⁻² nmol P_i ⋅ min⁻¹ ⋅ μg⁻¹at 1 mM MgATP. At pCa 9.0 and 1 mM MgATP, the activity was 1.5 × 10⁻² nmol P_i ⋅ min⁻¹ ⋅ μg⁻¹. Comparison of the values at high and low calcium showed that calcium increased the activity fivefold, demonstrating calcium regulation of the ATPase activity. These data demonstrate that, at pCa levels of ≥5.5, there were interactive effects of calcium and substrate level on ATPase activity.

The biphasic response of the ATPase activity to MgATP at low calcium has been observed in skeletal myofibrils and was described as substrate inhibition of activity (12). Another explanation is that at subsaturating levels of MgATP, the existing rigor cross bridges activated the thin filament, at least at very low levels of MgATP, to the same level as high calcium. Increases in MgATP levels would decrease the number of rigor cross bridges and subsequently cause the decline in ATPase activity at submaximal calcium levels. To further investigate the biphasic response of cardiac myofibrillar ATPase to MgATP levels, the activity was measured using a greater number of MgATP concentrations at low calcium (pCa 9.0). This allowed for a more detailed determination of the MgATP level that gave the highest activity. Figure 2 shows the ATPase activity as a function of MgATP at pCa 9.0 and pH 7.0. The peak activity was at 10 μM MgATP and precipitously dropped at higher levels of MgATP. This peak activity was 229% of that at the highest level of MgATP tested (3 mM). These results demonstrate that rigor activation occurs in cardiac myofibrils similar to that observed in skeletal muscle (36) and that maximal ATPase activity occurs at a similar level of MgATP to that of maximal tension observed in skinned cardiac cell preparations (26).

For this rigor-dependent mechanism of thin filament activation to function in ischemic contracture, it must occur at pH levels below the physiological levels likely to occur in the ischemic heart. Figure3 shows the ATPase activity that was measured at pH levels of 6.4, 6.0, and 5.4 at pCa 9.0. There was little change in activity at 3 mM MgATP when comparing the different pH levels. Others (37) have observed a decline in activity at pCa 9.0 with lower pH, and this difference may be explained by contamination of the myofibril preparation with mitochondria and sarcoplasmic reticulum because of a lack of detergent washing of the myofibrils. Our data show that, at all of these levels, there was a biphasic response of ATPase activity to MgATP. At pH 6.4 and 6.0, the peak activity was at 6–8 μM, whereas at pH 5.4, the peak was not as sharp and was at ∼50 μM. The peak activity was less with lower pH and was 155, 127, and 130% of that at 3 mM MgATP for pH 6.4, 6.0, and 5.4, respectively.

To determine whether the increase in ATPase activity is mechanically associated with sarcomere shortening, we did experiments to monitor the extent of sarcomere shortening at different levels of MgATP after 30 min of incubation with substrate (i.e., the same conditions as the ATPase assays). For this, we pretreated the myofibrils with fluorescent α-actinin. This protein binds specifically to the Z line and affords a fluorescent signal to use for measurement of sarcomere length. The MgATP levels used were 0, 0.1, 1, 8, and 3,000 μM to cover the range of substrate levels including the level associated with peak activity at most of the pH levels. Figure 4 shows phase-contrast and fluorescence images of myofibrils at pH 7.0 and 5.4 after incubation in 3,000, 8, and 1 μM MgATP. The phase-contrast images are difficult to interpret, but there is a distinct Z-line signal in the fluorescence images. The double-arrowhead lines within the fluorescent images delimit the internal two sarcomeres and show that the shortest sarcomeres occurred at 8 μM MgATP. Images such as these were used to measure the sarcomere length at the different pH and MgATP levels at pCa 9.0. The results are shown in Fig.5, which contains a histogram of the sarcomere lengths determined after the different treatments. At the highest MgATP level (3,000 μM), there was a decrease in sarcomere length with decrease in pH. This may have been the result of a partial loss of regulation or low-level activation because of the weak calcium buffering of calcium by EGTA at the low pH levels. The greatest decrease in sarcomere length was observed at 8 μM MgATP for all the different pH levels, and there was little difference between the different pH levels. The percentage of shortening relative to the relaxed condition (3,000 μM MgATP) ranged from 27 to 32%, with the greatest percentage of shortening occurring at pH 7.0. The smallest change in sarcomere length was observed at 0.1 μM MgATP, where the contractile machinery was likely substrate limited. Intermediate levels of shortening were observed at 1 μM MgATP. We did not measure the sarcomere length of myofibrils treated with 10–100 μM MgATP at pH 5.4, where the highest ATPase activity was observed, so there may have been a greater extent of shortening at these levels of MgATP for this particular pH. These data show that, at least for pH levels of 7, 6.4, and 6.0, the greatest extent of sarcomere shortening was associated with the highest ATPase activity, suggesting that the ATPase activity was coupled to mechanical events.

DISCUSSION

The slow contraction associated with a prolonged ischemic episode involves a complex series of events. These include changes in metabolite levels (decreases in glycogen, pH, phosphocreatine, and ATP; increases in lactate, P_i, ADP, and calcium). The mechanism responsible for the sarcomeric shortening very likely involves a cyclic actomyosin interaction. In vivo experiments suggest that the ischemic contracture precedes changes in intracellular calcium (21). Therefore, it appears that the contracture is caused not by the classic calcium-based mechanism but by the cross bridge-dependent mechanism. Other mechanical studies on isolated cardiac cells (6, 8, 34) suggest that increases in phosphate and decreases in pH depress calcium sensitivity, suggesting that even if there is an increase in cytosolic calcium, the thin filament will be less responsive under ischemic conditions. Interestingly, increases in phosphate do not inhibit rigor tension development (i.e., cross bridge-dependent activation) in cardiac preparations, suggesting that this mechanism of activation could operate in the ischemic heart (25). In the current study, we investigated the cross bridge-dependent mechanism of activation in isolated cardiac myofibrils with special emphasis on whether it occurs at low pH as observed in the ischemic heart.

Cross bridge-dependent activation of skeletal muscle myofibrils was first demonstrated by Weber (36), who observed a biphasic relationship between ATPase activity and MgATP with maximal activity at ∼10 μM MgATP. A schematic model of this process is shown in Fig.6 in which the tropomyosin is diagrammed to show its change in position caused by the rigor cross bridge. Recent models of thin filament activation incorporate a three-state model of the thin filament in which high calcium favors a shift from the blocked state to the closed state, whereas strong cross-bridge binding favors a shift from the closed state to the open state (16). It is the open state that allows for cross-bridge cycling and thus contraction. This models emphasizes the importance of strong cross bridges in activation (22). The important feature of the model is that it reconciles biochemical and physiological data showing that the thin filament could be activated by strong cross bridges in the absence of calcium. The classic two-state steric blocking model did not per se incorporate strong cross bridges in the activation process. The three-state model allows activation of the thin filament in the absence of calcium in that there is a dynamic equilibrium between the three states, and, although in the absence of calcium the blocked state predominates, there is a small population of closed states to which cross bridges can bind weakly and induce the transition into the open state on strong binding. This binding then shifts adjacent sites into the open state, allowing for cross-bridge cycling at these sites. This sequence of events can be used to explain rigor activation of myofibrillar ATPase and tension development when MgATP is low.

Studies on the interaction between calcium and MgATP levels (28) in skeletal myofibrils showed that the steepness of the relationship between calcium and ATPase activity was decreased with lower [MgATP] and that the calcium level for half-maximal activation was lowered with decreased [MgATP]. Hofmann and Fuchs (18) demonstrated increased calcium binding under rigor conditions compared with relaxed conditions. Similar studies have led to the suggestion that strong cross-bridge binding influences the properties of TnC via the thin filament, especially in cardiac muscle (14, 18). Our studies investigated this phenomenon in cardiac myofibrils from the perspective of ATPase activity. The data in Fig. 1demonstrate that low levels of MgATP stimulated the ATPase activity at pCa values of ≥5.5 in cardiac myofibrils in that there was a biphasic response of ATPase activity to MgATP level. The peak activity occurred at higher levels of MgATP with lower pCa, suggesting an interplay between strong cross-bridge binding and calcium binding. Studies with isolated cardiac cells suggest a similar phenomenon in that calcium sensitivity is increased when MgATP levels are lowered to the 10–100 μM range (2). Although we did not investigate this particular phenomenon at lowered pH, it has relevance to the anoxic heart in that MgATP levels will decline, and this could result in increased calcium sensitivity of the contractile apparatus.

A more careful analysis of cross bridge-dependent activation of ATPase activity at pCa 9.0 and pH 7.0 (Fig. 2) showed that maximal activity was at 10 μM MgATP and was ∼230% of that at high levels of MgATP. Studies on skinned cardiac cells at low calcium and pH 7.0 showed a similar biphasic response of tension to [MgATP] with maximum tension at 4–10 μM MgATP (26). The similarity in [MgATP] for peak tension and ATPase activity is surprising considering that the tension measurements did not employ an ATP-regenerating system and that it was assumed that there was little activation of cross-bridge cycling (26). Our data show that there was an increase in cross-bridge cycling in the form of increased ATPase activity at low [MgATP]. Independent of these differences in approach, the [MgATP] that resulted in peak tension at low calcium corresponds to the same level observed for ATPase activity, suggesting that there is an increase in cross-bridge cycling that develops tension. Indeed, in our studies, we observed a decline in ATPase activity at [MgATP] below the peak value, and a decline in tension was observed in skinned cardiac cells (27), although the decline in ATPase was much more pronounced than the decline in tension. In addition, our structural data demonstrate that the extent of shortening (which required cross-bridge cycling) was greatest at MgATP levels associated with peak ATPase activity (Figs. 4 and 5). These results suggest that rigor cross-bridge activation of the thin filament in the absence of calcium is manifested by increased ATPase activity and sarcomere shortening.

Whereas this mechanism of thin filament activation has been demonstrated at physiological pH, it is important to determine whether it occurs at subphysiological pH. The experiments at low pH showed that there was a peak in ATPase activity at 10–50 μM MgATP at low calcium (Fig. 3). Global ischemia in rat hearts gives pH levels of 6.2 after 12 min and 5.6 after ∼45 min coincident with ischemic contracture (21), justifying the pH range used in the current study. Comparison of the ATPase activity curves at different pH levels showed that decreased pH resulted in decreased peak activity. Studies on skinned cardiac cells showed that tension also increased with decreasing [MgATP] at pH 6.2, reaching a peak at ∼2 μM MgATP, and that lower pH resulted in a peak tension that was 60% of that at pH 7.0 (27). Our studies on ATPase activity at low pH showed a decrement in peak activity of ∼70% relative to that at pH 7.0. As with myofibrils at pH 7.0, the extent of sarcomere shortening was associated with the ATPase activity at the low pH levels: MgATP levels that resulted in the highest ATPase activity were associated with the greatest extent of sarcomere shortening. Thus data from mechanical studies and our studies suggest that rigor cross-bridge activation of the thin filament allows for cross-bridge cycling and that this can occur at subphysiological pH.

Implications for the ischemic heart.

The onset of rigor in dog hearts was observed when total ATP concentration declined to ∼1 mM (23), much higher than the 10 μM level observed for maximal ATPase activation of isolated myofibrils. This difference of about two orders of magnitude may result from several factors. First, the studies by Lowe et al. (23) measured tissue ATP and thus did not give cytosolic levels. Second, compartmentalization of ATP pools has been suggested for the intact cardiac cell (9), which could also yield a different value at the myofilament level than the tissue level. Third, their studies used whole hearts, and thus there is likely cell-to-cell variability in the MgATP level and the rate of rigor development at the individual cell level. Indeed, studies with isolated heart cells showed an asynchronous rigor contracture in the cell population that was not represented in the measured ATP levels (15). Also, studies on striated muscle have shown that the kinetics of rigor contracture were rapid at the isolated fiber level but slow at the fiber bundle level (19). The true level of MgATP at which rigor contracture/tension is maximum is best estimated in skinned fiber preparations or in the isolated myofibril with its shorter diffusion distance. Thus the critical level of MgATP for rigor activation in cardiac myofibrils is ∼10 μM. At the cellular level during an ischemic episode, this MgATP level will be approached from the high substrate side as cellular [MgATP] declines. As pointed out by Hearse et al. (17), there will be an increase in ATPase activity as this critical [MgATP] is approached at a time when glycolysis will be inhibited by the decline in pH and other metabolic products. Thus, from a kinetics standpoint, one would expect a rapid rate of rigor contraction in the isolated cell such that once the level of MgATP is <50 μM, there will be a downward spiral of increasing ATPase activity in the face of a declining ability to generate ATP, which then depletes the cell of high-energy phosphate and results in rigor contraction of the cell.

FOOTNOTES

• The support of the American Heart Association, National Affiliate, (award no. 9630028N to D. R. Swartz) is greatly appreciated.

• The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.