INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

In skeletal and cardiac muscle, depolarization of the transverse tubular (T-tube) membrane by the action potential (AP) triggers Ca²⁺ release from the sarcoplasmic reticulum (SR), an intracellular Ca²⁺ storage organelle (Sandow, 1965; Beeler and Reuter, 1970). The transduction mechanisms that link T-tube depolarization and SR Ca²⁺ release in skeletal and cardiac muscle are quite different. In cardiac muscle, T-tube depolarization activates the L-type Ca²⁺ channel, resulting in a small Ca²⁺ influx that activates the SR Ca²⁺ release channel (Fabiato, 1983; Beuckelmann and Wier, 1988; Näbauer et al., 1989). This transduction mechanism in cardiac muscle is commonly referred to as Ca²⁺-induced Ca²⁺ release (CICR). In skeletal muscle, the transduction mechanism is not dependent on Ca²⁺ influx across the T-tube membrane (Caputo, 1968; Armstrong et al., 1972; Miledi et al., 1977). Instead, T-tube depolarization is thought to be sensed by voltage-sensing proteins, and then this information is relayed through a physical protein-protein link to the SR Ca²⁺ release channel (Schneider and Chandler, 1973; Shirokova et al., 1996). Thus the T-SR transduction mechanism in cardiac muscle involves a diffusing second messenger (i.e., Ca²⁺), while the T-SR transduction mechanism in skeletal muscle involves a direct physical link.

The study of intracellular Ca²⁺ signaling in isolated skeletal and cardiac muscle cells has a long, distinguished, and productive history. Traditionally, these studies used a wide array of clever but invasive experimental approaches. For example, intracellular Ca²⁺ signaling in skeletal muscle has been well defined in highly stretched, cut fiber segments (Hille and Campbell, 1976; Vergara et al., 1978; Kovacs et al., 1979). Intracellular Ca²⁺ signaling in cardiac muscle has been elegantly defined in mechanically skinned myocytes (Fabiato, 1983). Traditionally, the SR Ca²⁺ release process is triggered by a wide array of nonphysiological perturbations (i.e., ionic exchange, voltage-clamp steps, and/or pharmacological agents like caffeine). The use of nonphysiological invasive experimental manipulations is well justified because it is difficult to accurately quantitate local intracellular Ca²⁺ signals in moving intact cells. The consequence is that the spatiotemporal distribution of local intracellular SR Ca²⁺ release in contracting cells at normal resting sarcomere lengths in the absence of potentially invasive experimental manipulations has not been well defined. This is unfortunate because this information is fundamental to delineating the constraints that govern intracellular Ca²⁺ release and distribution under normal physiological conditions.

The blurring produced by the mechanical movement (i.e., contraction) and/or diffusion represents a persistent experimental obstacle to fluorescence imaging of local intracellular Ca²⁺ in muscle cells. To overcome this, a series of high-resolution snapshot images of Rhod-2 fluorescence were taken during very brief (7 ns) flashes of excitation light from a frequency-doubled Nd:YAG laser. Each snapshot image is essentially an instantaneous freeze-frame picture of the Ca²⁺ profile in the cell during the flash. Blurring artifacts are minimized because there is no significant cell shortening or molecular diffusion during the brief 7-ns flash. Precisely timed flashes during successive APs revealed the spatiotemporal distribution of local intracellular SR Ca²⁺ release in contracting muscle cells in two spatial dimensions at high temporal resolution.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Cardiac muscle preparation

Rat ventricular myocytes were enzymatically dissociated with the Langendorff retroperfusion technique (Mitra and Morad, 1985). Briefly, the heart was removed and placed on a petri dish containing a Ca²⁺-free Tyrode (Tyrode_noCa) solution containing 140 mM NaCl, 5.4 mM KCl, 1 mM MgCl₂, 0.33 mM NaH₂PO₄, 10 mM HEPES, and 10 mM glucose, pH 7.4, at 37°C. Then the aorta was cannulated and the coronary arteries were washed with Tyrode_noCa solution for 6 min. The heart was then perfused with Tyrode_noCa solution containing 2 mg/ml collagenase (Worthington Biochemical Corporation, Lakewood, NJ; 257 U/mg) and 0.1 mg/ml protease (Pronase E; Sigma, St. Louis, MO; 4.4 U/mg). The dissociation procedure was concluded by perfusing the heart with Tyrode solutions containing 0.2 and then 2 mM CaCl₂. The dissociated cells were used the same day.

Skeletal muscle preparation

Single skeletal muscle fibers were obtained from the lumbricalis muscles of the tropical frog Leptodactylus insularis. Whole muscles were incubated in Ringer's solution containing 115 mM NaCl, 2.5 KCl, 1.8 CaCl₂, 5 HEPES (pH 7.3), and 8 mg/ml collagenase (Worthington Biochemical Corporation; 257 U/mg) at 37°C for 35 min. Removing the collagenase and adding 5 mg/ml bovine serum albumin stopped the fiber dissociation process. Fibers were mechanically separated with a Pasteur pipette tip and placed on a coverslip at slack length for experimentation.

Optical measurements

Isolated cardiac and skeletal muscle cells were loaded with Rhod-2 (Molecular Probes, Eugene, OR) by incubation in a 10 µM Rhod-2/AM solution for 30-60 min at room temperature. After Rhod-2 loading, cells were rinsed continuously for 15 min. The loaded washed cells were then placed on the stage of an inverted fluorescence microscope (Nikon Diaphot, Tokyo, Japan) modified for flash laser imaging (Escobar et al., 1994). All experiments were carried out at room temperature. Action potentials were triggered by field stimulation using micropositioned platinum wires. A Zeiss planapo 63× (NA 1.4) oil immersion objective was used to image the cells. Snapshot images were obtained by epiilluminating a relatively large field on the cell with a single 7-ns-duration, 532-nm light flash from a frequency-doubled Nd-YAG laser (Spectra Physics, Mountain View, CA). The size of the illuminated field was defined by the magnification of the objective and the size of the aperture through which the excitation light was directed.

Images of the illuminated area were acquired with a cooled CCD camera (MCD 600; Spectra Source, Ventura, CA). Image collection was synchronized to electrical stimulation. Individual images were collected after precisely timed delays from the electrical stimulus. The delay between the electrical stimulus and image acquisition was digitally mastered with the computer hardware. The synchronization of laser, the CCD camera, and the A/D conversion system was controlled with a LabVIEW (National Instruments, Austin, TX)-based computer program.

The inherent blurring of fluorescent images (confocal or not) makes it difficult to precisely define local (submicron) Ca²⁺ gradients. Thus no formal attempt was made to define the submicron Ca²⁺ gradients here. The intention of our study was not to precisely define the amplitude of the subsarcomeric Ca²⁺ gradients detected. Instead, our intent was to show that clear differences exist in contracting cardiac and skeletal muscle cells at similar sarcomere lengths. To thoroughly and formally define subsarcomeric Ca²⁺ gradients would require confocal spatial resolution combined with the temporal resolution of laser flash microscopy.

Image analysis

Line scan measurements were made on digital images with the aid of the Scion Image analysis program (Scion Corp., Frederick, MD). One-dimensional fast Fourier transforms (FFTs) were implemented on 128 line scan measurements of 128 pixels, using the FFT algorithm provided in the Origin 5.0 software package (Microcal, Northampton, MA). The polar forms of the FFT of each line scan were averaged and plotted for different snapshot images at different times during the fluorescent transients.

Mathematical modeling of Ca²⁺ distribution

To theoretically simulate the myoplasmic free Ca²⁺ concentration changes that may occur during a twitch, a multicompartment unidimensional diffusion model was evaluated (Cannell and Allen, 1984; Pizarro et al., 1991). The model consists of a segmented hemisarcomere cut into n slices. In our modeling, n was equal to 12. The first slice (compartment 0) was presumed to be the site of the Ca²⁺ input flux (i.e., the junctional space). Free Ca²⁺ concentration changes in all compartments were calculated as the influx into minus the efflux out of the compartment. Formally,
where t is time, J is the directional flux, and i is the segment number. The Ca²⁺ influx component was assumed to arise from three different sources: 1) SR Ca²⁺ release, 2) unbinding of Ca²⁺ from known buffers, and 3) Ca²⁺ flux from neighboring compartments. The Ca²⁺ influx due to SR Ca²⁺ release was governed by the Ca²⁺ gradient across the SR membrane and SR Ca²⁺ release channel function. The Ca²⁺ influx due to Ca²⁺ unbinding from known buffers included known endogenous Ca²⁺ binding proteins (e.g., parvalbumin) and the exogenously added Ca²⁺ indicator Rhod-2. The Ca²⁺ efflux component was also assumed to arise from different sources: 1) SR Ca²⁺ reuptake and Ca²⁺ binding to known buffers and 2) Ca²⁺ flux into neighboring compartments. It was assumed that the SR Ca²⁺ reuptake process did not take place in the junctional segments. The SR Ca²⁺ reuptake was assumed to be active Ca²⁺ transport, the maximum rate and kinetics of which were defined by a second-order kinetic scheme. The only entity allowed to diffuse is the Ca²⁺ ion. All equations were numerically integrated using a finite-difference approximation (Euler method), with the aid of the Scope 3.5 simulation package.

The Ca²⁺ release waveforms that were used to drive the model were obtained from the experimental data. Skeletal and cardiac simulations were driven with the measured skeletal and cardiac Ca²⁺ release waveforms, respectively. Thus differences in the rate of SR Ca²⁺ release in skeletal and cardiac muscle are considered in the modeling. Recently it was demonstrated that the rate of SR Ca²⁺ release is not affected by the rate of SR Ca²⁺ uptake (Caputo et al., 1999). Thus differences in uptake rate were not considered here. It is clear that intracellular Ca²⁺ buffer capacity is very important and will have an impact on whether subsarcomeric Ca²⁺ gradients can be detected (Cleeman et al., 1998). Higher Ca²⁺ buffer capacity would make it easier to detect gradients. Skeletal muscle has higher Ca²⁺ buffer capacity than cardiac muscle because of the presence of parvalbumin. Our skeletal muscle simulations take this into consideration, as Ca²⁺ binding and Mg²⁺ binding to parvalbumin are included. Simulations using skeletal parameters (including parvalbumin) and either the cardiac or skeletal Ca²⁺ release waveforms were performed (data not shown). These simulations revealed that the presence of higher buffer capacity was not sufficient to generate detectable gradients when the slower Ca²⁺ release waveforms were used. The skeletal and cardiac simulations shown in this study were carried out using skeletal or cardiac parameters, respectively. The parameters used are listed in Table 1.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Intracellular Ca²⁺ distribution in intact cardiac myocytes

Intact rat ventricular cardiac myocytes were isolated and loaded with the fluorescent Ca²⁺ indicator Rhod-2. An image of transilluminated rat cardiac ventricular myocyte is shown in Fig. 1 A (left). The striations (~2 µm apart) arising from the ordered arrangement of the contractile proteins of the sarcomere were clearly evident. A snapshot image of the same cell taken 70 ms after an AP was triggered by field stimulation is also shown in Fig. 1 A (right). The pair of images in Fig. 1 A illustrates the relationship between cell morphology and image area. Note that the image area, the area illuminated by excitation light, represents only a portion of the entire cell.

View larger version (61K): [in this window] [in a new window]   FIGURE 1   (A) A transillumination image of a rat ventricular cardiac myocyte (left) and a single 7-ns exposure fluorescence image (i.e., snapshot image) of the same cell taken 70 ms after field stimulation (right). The field stimulation induced an action potential (AP) on the surface membrane of the cell. These data are representative of 27 similar experiments. (B) Sequence of snapshot images of a rat ventricular cardiac myocyte taken at different times along the AP-induced Ca2+ transient. One image was taken at the specified delay after each field stimulus. The series of images represent data collected on successive APs. The pseudo-color scale corresponds to F/F measurements. (C) The fluorescent signal obtained from the same point in the cell at different delays from the field stimulation are plotted as a function of time.

A series of snapshot images taken at different times after an AP was triggered is shown in Fig. 1 B. Before the AP, the relatively low resting fluorescence indicates that the indicator was homogeneously distributed and that there was a relatively low resting Ca²⁺ level (~100 nM). After the AP, an elevation in intracellular Ca²⁺ concentration was first evident at the 10-ms mark. The AP-triggered Ca²⁺ release nearly reached peak intensity at the 70-ms mark. Some heterogeneity in the Ca²⁺ distribution during the release process was evident. This heterogeneity, however, could not be clearly correlated with any morphological determinant like sarcomere spacing. Thus our snapshot imaging approach was unable to reliably detect any subsarcomere Ca²⁺ gradients during the AP-triggered Ca²⁺ release process. This was disappointing because subsarcomere Ca²⁺ gradients must arise if Ca²⁺ is released rapidly at the T-SR junction and then slowly diffuses across the sarcomere. Nevertheless, the overall AP-triggered Ca²⁺ transient (F/F) in the cardiac myocyte was reconstructed from several snapshot images (Fig. 1 C). The reconstructed Ca²⁺ transient contains points that are not represented in Fig. 1 B. The temporal characteristics of the AP-triggered Ca²⁺ transient were nearly identical at all points in the snapshot image. The AP triggered a rapid Ca²⁺ rise that spontaneously decays over several hundred milliseconds. The rise time of the AP-triggered Ca²⁺ transient in the cardiac myocyte had a time constant of 16.4 ± 0.71 ms (mean ± SEM).

The same experiment was performed on single frog skeletal muscle fibers. The sarcomere length of the skeletal fibers was nearly identical to that of the cardiac myocytes tested above. The sarcomere lengths of a typical cardiac myocyte and skeletal muscle fiber are compared in the transillumination images shown in Fig. 2. Sarcomere length in cardiac and skeletal muscle cells corresponds to the distance between transverse tubules. This distance is important because SR Ca²⁺ release occurs at the junction (i.e., the T-SR junction). All cells tested (cardiac or skeletal) had a sarcomere spacing of ~2 µm (± 0.2 µm). Thus the SR Ca²⁺ release sites were equally spaced along the length of the cells.

View larger version (110K): [in this window] [in a new window]   FIGURE 2   Structural comparison between a rat ventricular cardiac myocyte (A) and a single lumbricalis frog skeletal muscle fiber (B). Transillumination images show that the two cell types have a similar sarcomere length (~2 µm).

Like the cardiac myocytes, the intact skeletal muscle fibers were loaded with the fluorescent Ca²⁺ indicator Rhod-2. A segment of a typical skeletal muscle fiber is shown in Fig. 3 A (left). A snapshot image of the same fiber taken 2.5 ms after an AP was triggered by field stimulation is shown in Fig. 3 A (right). Like the cardiac data, the snapshot image area represents only a portion of the skeletal muscle fiber. The image area is the region of the cell illuminated by the 7-ns-long flash of excitation light. Heterogeneity in the free Ca²⁺ distribution at the 2.5-ms mark is evident. The heterogeneity is in the form of transverse high- and low-intensity fluorescence bands that repeat regularly along the length of the cell. This banding (Fig. 3 A, right) corresponds well to the sarcomere spacing of the cell observed in the transillumination image (Fig. 3 A, left).

View larger version (65K): [in this window] [in a new window]   FIGURE 3   (A) Transillumination image of a single lumbricalis fiber (left) and the corresponding snapshot image of the same cell (right) taken 2.5 ms after field stimulation. These data are representative of 22 similar experiments. Note that a larger area of the skeletal fiber is illuminated than in the cardiac case. (B) Sequence of snapshot images of a single lumbricalis fiber taken at different delays after field stimulation. One image was taken at the specified delay after each field stimulus. The series of images represent data collected on successive APs. Strong subsarcomeric Ca2+ gradients were observed during the first 4 ms of the AP-induced Ca2+ transient. (C) The fluorescent signal obtained from the same point in the cell is plotted as a function of time.

A series of snapshot images taken at selected times after an AP was triggered is shown in Fig. 3 B. Before the AP, the uniform resting fluorescence of the Rhod-2-loaded cell suggests relatively even indicator distribution and a low resting Ca²⁺ level (~100 nM). Intracellular Ca²⁺ release was evident 0.5 ms after the AP was triggered. The AP-triggered Ca²⁺ release reached peak intensity at the 6-ms mark. Periodic distinct regions of high and low Ca²⁺ concentration (i.e., high- and low-intensity fluorescence bands) were evident for the first 4 ms after the AP. The fluorescent banding becomes more and more marked as time progresses. This in part reflects the inherent signal-to-noise characteristics of the detection system. At early times, the amount of Ca²⁺ release is small and near the limit of resolution. Maximum subsarcomeric Ca²⁺ gradients occur at the peak of the Ca²⁺ release and later on (i.e., at the 2.5-ms mark). The appearance of Ca²⁺ gradients within a few milliseconds indicates a high synchrony of release. Escobar et al. (1994) demonstrated a similar pattern of intracellular Ca²⁺ release in highly stretched skeletal muscle fibers (sarcomere length ~4 µm). These authors argued that the bands of high Ca²⁺ concentration (i.e., the regions of high fluorescence intensity) correspond to the position of the T-tube SR junction and that the bands of low Ca²⁺ concentration correspond to the interjunctional space. If this interpretation is equally valid here, then our data illustrate that significant subsarcomeric Ca²⁺ gradients occur in response to an AP in contracting skeletal muscle fibers at normal resting sarcomere lengths (~2 µm).

The reconstructed fluorescence transient (F/F) from several snapshot images is illustrated in 3 C. This transient represents the fluorescence intensity at the T-SR junction (i.e., site of SR Ca²⁺ release). The AP triggered a very rapid Ca²⁺ rise (time constant 0.85 ± 0.23 ms) that spontaneously decayed with a time constant of 10.8 ± 3.7 ms (mean ± SEM). These values are ~20 times faster than those measured in cardiac cells. The data in Fig. 3 illustrate the relatively high spatial and temporal resolution of the snapshot imaging method applied. For example, the high- and low-fluorescence banding (Fig. 3 A, right) demonstrates submicron (x-y) spatial resolution. The high number of points on the rising phase of the Ca²⁺ transient (Fig. 3 C) demonstrates submillisecond temporal resolution. The point is that the applied snapshot imaging method has the capacity to detect subsarcomeric Ca²⁺ gradients at a sarcomere spacing of 2 µm. Thus the absence of subsarcomeric Ca²⁺ gradients during AP-triggered Ca²⁺ release in cardiac myocytes (Fig. 1) is not due to the resolving power of the imaging system.

To quantitate the spatial fluorescence banding, the fluctuations in intensity along lines perpendicular to the Z-lines were analyzed using the fast Fourier transform (Fig. 4). The profile of fluorescence intensity along a line perpendicular to the sarcomeres (i.e., z-lines) in a skeletal and cardiac muscle is illustrated in Fig. 4 A. The skeletal and cardiac data were collected 2.5 ms and 100 ms after electrical stimulation, respectively. These plots of line intensity have two fundamental features. First, there is lower intensity at the ends of the plots because of the finite size of the illuminated circular spot. Second, the variance in the signal is substantially greater in the skeletal muscle case. This greater variance is generated by the distinct fluorescence gradients that occur along the line in skeletal muscle. The variance in the cardiac case is smaller because of the absence of big fluorescence gradients. To provide a more quantitative description of the size and periodicity of the fluorescence gradients, a one-dimensional fast Fourier analysis of 128 of these line scan measurements was made in both the skeletal and cardiac cases. The results of this Fourier analysis are shown in Fig. 4 B. Each of the four plots (skeletal at left, cardiac at right) shows the modulus of the average polar fast Fourier transform. In all plots, there are high values at the zero frequency point, and this corresponds to the average fluorescence signal along the scan line. Interestingly, a small but distinct second peak (arrows) becomes evident only in the skeletal muscle case. This second peak occurs at the 1.8-ms and 2.5-ms time points at the 0.6-µm¹ mark. This corresponds to a mean periodicity of fluorescence intensity occurring every 1.7 µm (i.e., 1/0.6 µm¹). Note that this second peak was absent at the 0.5-ms and 6-ms time points because there are no clear subsarcomeric gradients at those times. This analysis also shows that the biggest global fluorescence signal (i.e., the "DC" level of the FFT) occurred at the 6-ms point, but the most distinct gradients occurred at the 2.5-ms point. The absence of a distinct second peak in the cardiac muscle case is consistent with the absence of a detectable subsarcomeric Ca²⁺ gradient in those cells.

View larger version (28K): [in this window] [in a new window]   FIGURE 4   (A) Line scan plots of fluorescence intensities of a skeletal fiber (left; 2.5 ms after stimulus) and a cardiac ventricular myocyte (right; 100 ms after stimulus). Lines were drawn perpendicular to the z lines. Note the large variance of the signals. (B) Average one-dimensional fast fourier transform (FFT) of 128 line scan recordings of 128 pixels obtained from a skeletal fiber (left plots) or cardiac cell (right plots) at four different times during a fluorescent transient. The arrow denotes the intensity peak corresponding to >1.7 µm (1/0.6 µm1).

Our inability to detect subsarcomeric Ca²⁺ gradients in cardiac muscle does not mean that they do not exist. In fact, our interpretation is that subsarcomeric Ca²⁺ gradients do indeed exist in the cardiac cells, but those subsarcomeric Ca²⁺ gradients are simply more difficult to resolve. Several factors may contribute to the difficulty of detecting subsarcomeric Ca²⁺ gradients in the cardiac cells. For example, the distribution of the fluorescent Ca²⁺ indicator (Rhod-2) could be different in the cardiac and skeletal muscle cells. The spacing of SR Ca²⁺ release sites along the length of the cell could be different. The axial (z axis) alignment of sarcomeres in cardiac cells may not be as tight as that in skeletal muscle fibers. These possibilities are addressed by the experimental data. The relatively uniform resting fluorescence of the Rhod-2-loaded cells suggests that the indicator was evenly distributed in both the cardiac and skeletal cells. Similar sarcomere lengths suggest similar spacing of SR Ca²⁺ release sites along the length of the cells. Although periodic sarcomere registration shifts can been seen in the cardiac cells (Fig. 2), these shifts do not occur frequently enough to explain the absence of detectable subsarcomeric Ca²⁺ gradients. However, other possible factors that may contribute to the difficulty of detecting subsarcomeric Ca²⁺ gradients in the cardiac cells cannot be easily addressed, for example, endogenous cytosolic Ca²⁺ buffers that may be differentially and nonuniformally distributed along the cardiac or skeletal muscle sarcomere. Differences in ultrastructure could explain the different results in cardiac and skeletal muscle (Soeller and Cannell, 1999). Mitochondria and nuclei are sandwiched in and among the contractile apparatus in cardiac muscle, generating ultrastructural irregularities that could have an impact on local Ca²⁺ imaging. The point is that potential alternative explanations should be acknowledged. Perhaps the simplest explanation for our results is described below.

The absence/presence of detectable subsarcomeric Ca²⁺ gradients in cardiac/skeletal muscle cells may simply be the result of the different mechanisms that link T-tube depolarization to the SR Ca²⁺ release process. In skeletal muscle, the brief AP (~2 ms) results in almost synchronous activation of SR Ca²⁺ release sites across the muscle fiber. In contrast, the CICR process in cardiac cells involves a relatively slow Ca²⁺ diffusion step and the recruitment of multiple SR Ca²⁺ release sites. To evaluate this possibility, a very simple diffusional model (see Methods) was used to predict subsarcomeric Ca²⁺ distributions that may arise in response to different Ca²⁺ release functions (Fig. 5). The model was driven by a measured cardiac or a skeletal Ca²⁺ release waveform. The time courses of the driving Ca²⁺ release functions and simulated snapshot images are illustrated. Both the cardiac (Fig. 5 A) and skeletal (Fig. 5 B) Ca²⁺ release functions generate subsarcomeric Ca²⁺ gradients. In the cardiac case (Fig. 5 A), the peak Ca²⁺ levels are relatively small compared to the mean Ca²⁺ level. In contrast, the same simple distributed spatial model driven by the measured skeletal Ca²⁺ release waveform predicts large distinct subsarcomeric Ca²⁺ gradients. Thus the absence/presence of detectable subsarcomeric Ca²⁺ gradients in cardiac/skeletal muscle cells may simply be due to the different kinetics of the free Ca²⁺ input signals in each case.

View larger version (35K): [in this window] [in a new window]   FIGURE 5   Simulated temporal and spatial fluorescence distribution according to our diffusion model. (A) Snapshot image simulations during a Ca2+ transient when the model was driven by an SR Ca2+ flux derived from our measured cardiac Ca2+ release data. (B) Snapshot image simulations when the model was driven by SR Ca2+ flux derived from our measured skeletal Ca2+ release data. Note the large difference in time course between the cardiac and skeletal cases. The parameters used in the model are listed in Table 1. Note that maximum rates of release and uptake do not necessarily correspond to flux because the flux depends on many factors (i.e., gradient direction, dissociation constant, association rate). The concentration of intracellular Ca2+ buffer (700 µM) and its kinetics was obtained from the literature (Cannell and Allen, 1984).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The snapshot laser imaging strategy

In this paper we demonstrate that single snapshot fluorescent images can be taken at the time scale of the decaying lifetime of a fluorescent indicator (Rhod-2) in a living cell. The lifetime of most commercially fluorescent Ca²⁺ indicators is on the order of 2-5 ns. In this study, fluorescent images were acquired during a 7-ns flash of excitation light. Thus the fluorescent signals detected by our CCD camera essentially represent the integration of the lifetime relaxation of the fluorophor. Another feature of the applied snapshot imaging strategy is that it minimizes diffusion and movement artifacts because 7 ns is simply not enough time for a moving particle to blur the image. Thus the snapshot method provides the means to generate "freeze-frame" fluorescence images of local Ca distributions in moving muscle cells. Traditional imaging strategies would require invasive experimental manipulation to restrict movement or be limited to measurements before movement occurs.

A clear disadvantage of the snapshot imaging strategy is that it is not confocal (Monck et al., 1994). A conventional scanning confocal microscope has slightly better spatial resolution. This disadvantage is somewhat offset by higher temporal resolution and the ability to collect spatial information in two dimensions simultaneously. The traditional conventional scanning confocal microscope operates at its optimal temporal resolution (i.e., line scan mode) at the expense of spatial information (i.e., data collected in only one spatial dimension).

Ca²⁺ distributions in ventricular cardiac myocytes and skeletal muscle fibers

The SR Ca²⁺ release machinery in cardiac and that in skeletal muscle cells share many common features. For example, T-SR communication is mediated by dihydropyridine and ryanodine receptors in both tissues, albeit by different mechanisms. The architecture of the T-SR junction is similar. The spacing of SR Ca²⁺ release sites along the length of the cell is similar. Such striking morphological similarities could imply a common function. However, our data show that cellular architecture is not the only determinant of the local intracellular Ca²⁺ signaling. Subsarcomeric Ca²⁺ gradients were detected in skeletal muscle fibers (Fig. 3) but not in cardiac myocytes (Fig. 1). This is significant because the sarcomere length was constant (i.e., similar SR Ca²⁺ release machinery spacing) between the two cell types and the measurements were made under nearly ideal physiological conditions (i.e., APs instead of voltage clamp steps, low Ca²⁺ indicator level with no other exogenous Ca²⁺ buffer added), with the cell free to contract.

The absence/presence of a subsarcomeric Ca²⁺ gradient in cardiac/skeletal muscle cells (respectively) is most likely due to the marked differences in the synchrony of SR Ca²⁺ release activation. For example, it is thought that AP-triggered SR Ca²⁺ release is highly synchronized across the skeletal muscle cell. A highly synchronized Ca²⁺ release at the T-SR junction could generate the observed subsarcomeric Ca²⁺ gradients if Ca²⁺ diffusion into the nonjunction space is significantly slower than T-SR signal transduction and Ca²⁺ release processes combined. Previously we have shown that the dissipation of subsarcomeric Ca²⁺ gradients in stretched skeletal muscle fibers (i.e., sarcomere ~4 µm) were apparent for nearly 12 ms after triggering by an AP (Escobar et al., 1994; Monck et al., 1994). Here we observed that subsarcomeric Ca²⁺ gradients in skeletal muscle fibers at a sarcomere length of ~2 µm were evident for only ~3-4 ms. The diffusion time in one dimension varies with the square of the distance. Thus doubling distance (2 versus 4 µm) would theoretically increase the diffusion time by about fourfold (3 versus 12 ms). Thus diffusion could explain the temporal redistribution of Ca²⁺ in skeletal muscle fibers reported here and in the previous studies.

In the cardiac myocytes, subsarcomeric Ca²⁺ gradients were not evident, but this does not mean that they did not exist. Two groups have in fact reported subsarcomeric Ca²⁺ gradients in isolated cardiac myocytes (Isenberg et al., 1996; Cleeman et al., 1998). Isenberg et al. (1996) reported the presence of small but detectable subsarcomeric Ca²⁺ gradients in guinea pig ventricular myocytes driven by fast voltage clamp steps after a complex digital manipulation of multiple images. Cleeman et al. (1998) reported subsarcomeric Ca²⁺ gradients in voltage-clamped ventricular myocytes. In the latter study, a very high concentration of indicator and the presence of 5 mM EGTA eliminated contraction and limited Ca²⁺ diffusion out of the junctional region, facilitating the detection of gradients. In our study, cells were triggered to contract by an action potential; thus the rate of SR Ca²⁺ release was not artificially accelerated or synchronized by a fast voltage clamp step. Furthermore, our studies involved single raw images taken in the presence of only 50 µM exogenous Ca²⁺ buffer (i.e., the indicator). Under these conditions, no subsarcomeric Ca²⁺ gradients were detected in cardiac myocytes. The inability to resolve subsarcomeric Ca gradients does not necessarily mean our results are in conflict with those of others. It may simply be that in the absence of certain nonphysiological manipulations the gradients are difficult to detect.

Recently the elegant work of Soeller and Cannell (1999) clearly showed that T-tubule topology in the rat ventricular myocyte is much more complex than previously thought. This could have an impact on the interpretation of our results. The presence of less ordered T-tubule arrays may imply that SR Ca²⁺ release site topology may not be highly ordered. This could geometrically smear the subsarcomeric Ca gradients and explain why gradients were not detected in cardiac muscle. However, two laboratories have detected subsarcomeric Ca²⁺ gradients in cardiac muscles under certain experimental conditions (Isenberg et al., 1996; Cleeman et al., 1998). The point is that the presence of gradients in those studies implies that topology of SR Ca²⁺ release sites in cardiac muscle must be highly ordered. It might be that not all regions of the T-tubule form T-SR junctions. The T-SR junctions (i.e., SR Ca²⁺ release sites) may occur in highly ordered arrays, even though T-tubes twist and turn in a more complex way. The potential impact of the complex T-tubule topology on our results must be acknowledged, but the absence of detectable subsarcomeric Ca²⁺ gradients here is not likely due to geometric irregularities.

It is important to note that our recording system did resolve Ca²⁺ gradients in skeletal muscle fibers at equal sarcomere length. The absence of detectable Ca²⁺ gradients in the cardiac cells may be the result of a relatively slow AP-triggered SR Ca²⁺ release signal. This assumption is consistent with the longer time until first appearance of detectable Ca²⁺ release and the longer time to peak Ca²⁺ release (both parameters are ~20 times slower in cardiac versus skeletal muscle cells). The apparently slower Ca²⁺ transient in cardiac cells is not likely due to differences in the release rates at the level of individual SR Ca²⁺ release channels. The permeation properties and activation kinetics of single cardiac and skeletal Ca²⁺ release channels are very similar (Györke et al., 1994). Furthermore, the rise time of local spontaneous Ca²⁺ sparks is nearly identical in cardiac and skeletal cells (Cheng et al., 1993; Tsugorka et al., 1995).

One explanation for the slower Ca²⁺ transient in cardiac muscle is the asynchronous recruitment of individual SR Ca²⁺ release units (or channels). Asynchronous recruitment of Ca²⁺ release units after the AP trigger signal would occur over ~100 ms (i.e., measured time to peak Ca²⁺ release). This relatively slow asynchronous recruitment effectively smears the local Ca²⁺ release profile because Ca²⁺ released at some sites would be diffusing away from others. The implication of this interpretation is that the highly synchronized AP-triggered Ca²⁺ release in skeletal muscle would ensure a fast maximum response. In contrast, the more asynchronous recruitment step in cardiac muscle would provide greater control because it represents a potential point for regulating the Ca²⁺ release process. Interestingly, the SR Ca²⁺ release in cardiac muscle is subject to regulation by a number of different pathways (i.e., -adrenergic stimulation).

Numerical simulations of intracellular Ca²⁺ distribution were done to confirm our suspicion that the data could be explained by simple distributed diffusion arguments. The simulations show that subsarcomeric Ca²⁺ gradients are a consequence of the spatial arrangement of the Ca²⁺ release machinery in both skeletal and cardiac muscle. In other words, subsarcomeric Ca²⁺ gradients are a direct logical consequence of striated muscle morphology. The simulation, however, showed that the magnitude of the subsarcomeric Ca²⁺ gradient depends critically on the speed and synchrony of the local Ca²⁺ release function. Fast and highly synchronized Ca²⁺ release generates large subsarcomeric Ca²⁺ gradients. Slow and less synchronous Ca²⁺ release generates small subsarcomeric Ca²⁺ gradients. Our simulation also predicts that increasing Ca²⁺ indicator concentration and/or increasing Ca²⁺ buffer capacity of the cell increase the heterogeneity in Ca²⁺ distribution along the cell. These simulations were in good agreement with our experimental observations and with the modeling predictions of other authors (Cannell and Allen, 1984).

In summary, our data indicate that the spatial distribution of intracellular Ca²⁺ release sites, the speed/synchrony of the local Ca²⁺ release events, and diffusion are all key determinants of the spatiotemporal distribution of the local intracellular Ca²⁺ signals striated muscle cells.