Cell Physiology

Ca2+ spark sites in smooth muscle cells are numerous and differ in number of ryanodine receptors, large-conductance K+ channels, and coupling ratio between them

Ronghua ZhuGe, Kevin E. Fogarty, Stephen P. Baker, John G. McCarron, Richard A. Tuft, Lawrence M. Lifshitz, John V. Walsh Jr.


Ca2+ sparks are highly localized Ca2+ transients caused by Ca2+ release from sarcoplasmic reticulum through ryanodine receptors (RyR). In smooth muscle, Ca2+ sparks activate nearby large-conductance, Ca2+-sensitive K+ (BK) channels to generate spontaneous transient outward currents (STOC). The properties of individual sites that give rise to Ca2+ sparks have not been examined systematically. We have characterized individual sites in amphibian gastric smooth muscle cells with simultaneous high-speed imaging of Ca2+ sparks using wide-field digital microscopy and patch-clamp recording of STOC in whole cell mode. We used a signal mass approach to measure the total Ca2+ released at a site and to estimate the Ca2+ current flowing through RyR [ICa(spark)]. The variance between spark sites was significantly greater than the intrasite variance for the following parameters: Ca2+ signal mass, ICa(spark), STOC amplitude, and 5-ms isochronic STOC amplitude. Sites that failed to generate STOC did so consistently, while those at the remaining sites generated STOC without failure, allowing the sites to be divided into STOC-generating and STOC-less sites. We also determined the average number of spark sites, which was 42/cell at a minimum and more likely on the order of at least 400/cell. We conclude that 1) spark sites differ in the number of RyR, BK channels, and coupling ratio of RyR-BK channels, and 2) there are numerous Ca2+ spark-generating sites in smooth muscle cells. The implications of these findings for the organization of the spark microdomain are explored.

  • wide-field imaging
  • spontaneous transient outward current
  • microdomain

many cells display brief and highly localized cytosolic Ca2+ transients because of Ca2+ release from the endoplasmic reticulum (ER) or the sarcoplasmic reticulum (SR). These phenomena were first discovered by Parker and Yao (18), who found that photorelease of inositol 1,4,5-trisphosphate (IP3) in oocytes caused asynchronous, focal Ca2+ release events, termed “Ca2+ puffs,” that occurred spontaneously on occasion. Spontaneous miniature transients were then found in myocytes, where they are mediated by ryanodine receptors (RyR), and were called “Ca2+ sparks” (6). Ca2+ sparks in cardiac and skeletal muscle constitute elementary events underlying global elevation of Ca2+ caused by release from the SR upon activation by Ca2+ influx and depolarization, respectively (5, 14, 16, 25). In contrast, in smooth muscle, Ca2+ sparks do not appear to serve as elementary events for global elevations in [Ca2+] that elicit contraction. Instead, the Ca2+ that triggers contraction appears to enter the cytosol from outside the cell and not from the SR. In this case, the Ca2+ sparks activate large-conductance, Ca2+-sensitive K+ (BK) channels within their microdomain and, in certain smooth muscle types, Ca2+-sensitive Cl (ClCa) channels (30). The currents elicited by Ca2+ sparks in smooth muscle have been designated as spontaneous transient outward currents (STOC) in the case of the K+ channels and spontaneous transient inward currents (STIC) in the case of the Cl channels. In smooth muscle, sparks and STOC have attracted attention because they appear to mediate relaxation in cerebral arteries (17) and because their dysfunction plays a role in hypertension (4, 20). In airway smooth muscle (30), the same sparks elicit both STIC and STOC, and they may play a role in stabilizing the cell's membrane potential and contractile state. By tipping the balance of activation in favor of the BK or ClCa channels, the relaxed or contracted state, respectively, might be favored. In the present study, we used one of the smooth muscle types that manifests only STOC.

In recent studies of Ca2+ sparks and STOC mediated by BK channels, investigators at our laboratory (28) developed a new imaging-based methodology to measure the mean intracellular Ca2+ current flowing from the SR into the cytosol through the RyR [ICa(spark)] in groups of Ca2+ sparks of differing amplitude. Two of the major findings of that study were as follows. First, the mean ICa(spark) varied by more than fivefold for different groups of sparks. This finding indicated that large-amplitude sparks were due to the opening of a greater number of RyR rather than simply to a single RyR opening for a longer duration. Second, there was variability in the ratio of active RyR to activated BK channels, i.e., the coupling ratio, when sparks elicited STOC. In that study, signal averaging techniques were used that merged data from multiple spark sites. Hence the characteristics of individual spark sites could not be examined systematically, and a number of questions about individual sites remained unanswered. For example, did the magnitude of a spark and a STOC at a given site tend to be constant? Did those spark sites that failed to elicit STOC always fail to do so, and did the coupling ratio differ from site to site? In the present study, we have quantitatively examined the properties of individual sparks sites using high-speed florescence Ca2+ imaging in combination with the patch-clamp technique. The main findings of the present study were reported previously in preliminary form (27).


Preparation of Cells and Patch-Clamp Recording

Bufo marinus were killed by rapid decapitation in accordance with Institutional Animal Care and Use Committee guidelines. Single smooth muscle cells from the stomach muscularis were enzymatically dispersed as described previously (8). Membrane currents were recorded using the tight-seal, whole cell patch-clamp recording configuration. The extracellular solution contained (in mM) 130 NaCl, 3 KCl, 1.8 CaCl2, 1 MgCl2, and 10 HEPES, with pH adjusted to 7.4 with NaOH. The pipette solution contained (in mM) 137 KCl, 3 MgCl2, 10 HEPES, 3 Na2ATP, and 0.05 fluo-3, with pH adjusted to 7.2 with KOH. All experiments were performed at room temperature. Whole cell currents were recorded at a holding potential of 0 mV and low-pass filtered using the Axopatch 1D amplifier (200 Hz cutoff), and then they were digitally sampled at 1 kHz and stored for analysis. Events were counted as STOC if they exceeded a threshold of 10 pA as detected using a custom automated algorithm. The events were then checked by visual inspection to eliminate anomalies such as multiple events overlapping in time or excessively noisy traces. The data for this study were drawn from 326 sparks sites in 25 cells from 13 animals.

Imaging and Measurement of Ca2+ Sparks

Fluorescent images were obtained using fluo-3 as the Ca2+ indicator and a custom-built wide-field, high-speed digital imaging system, which was described in detail elsewhere (31). Rapid imaging was made possible by using a cooled, high-sensitivity, charge-coupled device camera (128 × 128 pixels) developed in conjunction with the Massachusetts Institute of Technology Lincoln Laboratory (Lexington, MA). Imaging was performed at 100 Hz with 10-ms acquisition times, and there were 200 images/sequence (2 s). Multiple sequences of 2 s were taken, and the interval between sequences was ∼30 s. The camera was interfaced with a custom-made inverted microscope equipped with a ×40 oil-immersion lens (NA 1.3), and each pixel covered a 333 × 333-nm area of the cell. The 488-nm line of a multiline argon laser provided fluorescence excitation of the indicator fluo-3, and a laser shutter controlled the exposure duration. Emission of the Ca2+ indicator was monitored at wavelengths >500 nm. To obtain a constant [Ca2+] indicator, fluo-3 (50 μM) was delivered through the patch pipette, and measurements were not commenced until 10–15 min after disruption of the patch. After that time, no significant change in background fluorescence was detected. Subsequent image processing and analysis were performed offline using a custom-designed software package running on a Silicon Graphics workstation.

Because the localization of sparks within a cell was a critical feature of the present study, we examined the lateral resolution of the imaging system by measuring the point spread function of the system with the same objective used during the study using 190-nm-diameter fluorescent beads. The point spread function shows that we approached the theoretical limit of resolution for this system, ∼0.3- to 0.4-μm full width at half-maximum intensity (FWHM), while achieving axial resolution of ∼1-μm FWHM (see, e.g., Ref. 11, Table 2.1). Because in the present study we were interested in determining the location of the epicenter pixels of sparks that were both spatially and temporally distinct, the resolution of the imaging system was more than sufficient for our purposes.

Two measures of Ca2+ sparks were used: 1) the conventional fluorescence ratio, ΔF/F0, within a restricted volume; and 2) the change in total fluorescence, F − F0, over a larger volume, also designated as the Ca2+ signal mass, which is proportional to the total quantity of Ca2+ released into the cytosol. These measurements were described in detail previously (28, 29), so only a brief description follows. For the fluorescence ratio measure, the fluo-3 images, with pixel size 333 × 333 nm, were first smoothed by convolution with a 3 × 3-pixel approximation to a two-dimensional (2-D) gaussian curve (σ = 1). Fluorescence ratios were then calculated and expressed as a percentage on a pixel-to-pixel basis from the following equation: Math(1) where F is the fluorescence at each pixel in the time series and F0 is the “resting” level derived from the fluorescence time series by computing the median pixel value during quiescent times at each F(x,y). During a spark, the single pixel that had the highest fluorescence ratio was designated as the epicenter pixel, and the ΔF/F0 traces followed the time course of the epicenter pixel. To determine the epicenter pixel with precision, images of each spark had to be scaled properly. The value of a second measurement, the Ca2+ signal mass for each spark was computed from the 2-D, wide-field fluorescence images of fluo-3 using the following equations: Math(2) Math(3) The signal mass (SMt) is the product of the detector gain (G) times the change in total fluorescence (FT) summed over a 13.7-μm2 region (41 pixels on a side in the x-y plane), surrounding the spark epicenter pixel (x,y), determined from the ΔF/F0 images calculated in Eq. 1. The [Ca2+] in moles bound to fluo-3 was calculated using the following equation: Math where 2.44 is a calibration factor determined previously (28). This calculation assumes the presence of no buffers other than fluo-3 in the patched and dialyzed cell and hence provides a minimum value for the total Ca2+ released. Investigators at our laboratory (28) previously showed that the signal mass remained proportional to the total Ca2+ released into the cytosol over a wide range of buffering conditions, including a maximum estimate of fixed buffer in smooth muscle cells (3). Moreover, the signal mass was proportional to the total Ca2+ released for Ca2+ currents ranging from 0.1 to 10 pA, which easily encompassed the expected range of ICa(spark). Next, we demonstrated that the duration of ICa(spark) corresponded to the point in time when the STOC began to decay, that is, the time to onset of decay (TTD). Thus we calculated ICa(spark) by dividing total signal mass by TTD to take advantage of the higher temporal resolution of the STOC measurement. These results were described in detail in ZhuGe et al. (28). Finally, to determine the “coupling ratio,” we measured the STOC amplitude 5 ms after its onset. For this purpose, the onset of the STOC was taken as the point in time when the current trace was 2 SD above baseline noise. For further treatment of the “coupling ratio,” see ZhuGe et al. (28) and results below.

Statistical Analysis

Variability in the parameters of Ca2+ sparks and STOC within and between sites was evaluated by performing analysis of variance after log transformation (see Table 1 and Figs. 3 and 4).

View this table:
Table 1.

Heterogeneity in sparks and their STOC


All chemicals except fluo-3 (Molecular Probes, Eugene, OR) were purchased from Sigma Chemical (St. Louis, MO).


Sparks Can Be Measured in Two Ways

The evolution of a Ca2+ spark is shown in Fig. 1, along with the simultaneously recorded STOC that it elicited. Figure 1A shows one spark produced using the fluorescent Ca2+ indicator dye fluo-3. Figure 1Ba shows the STOC, and Fig. 1Bb shows the conventional method of measuring a spark as a fraction of the background fluorescence (ΔF/F0) in a relatively small region. In the present study, the region was a single 333 × 333-nm pixel that lies at the spark's center and displayed the brightest fluorescence, i.e., the epicenter pixel. This measure is often considered proportional to the average [Ca2+] in the region, but there are difficulties with this view. Because the indicator dye and Ca2+ do not come into equilibrium in the microdomain of the Ca2+ release site where the BK channels reside (29), calculations of [Ca2+] there are problematic. Consequently, another approach is needed.

Fig. 1.

Ca2+ spark, Ca2+ signal mass (SM), and elicited spontaneous transient outward currents (STOC). A: portion of smooth muscle in the field of view; Ca2+ spark is shown at top left. Changes in cytosolic Ca2+ were measured using fluo-3 (50 μM), which was introduced into the cell in salt form through the patch pipette. Images were acquired at 100 Hz, with exposure time of 10 ms/image. The change in Ca2+ concentration is expressed as a conventional fluorescence ratio, ΔF/F0 (%), and displayed on a pseudocolor scale calibrated at right. Numbers above images correspond to the arrowheads in B, bottom, and indicate time when image was obtained. The cell was bathed in medium containing 1.8 mM Ca2+ and held at a membrane potential of 0 mV. Ba: time course of STOC elicited by the spark shown in A, as monitored by whole cell patch-clamp recording. Bb: for the same spark, time course of change in fluorescence in the pixel (333 × 333 nm) in which peak fluorescence was reached, i.e., the epicenter pixel. Bc: time course of Ca2+ SM for the same spark, computed as the total fluorescence for a volume subtended by an area 41 pixels/side in the x-y plane and centered on the epicenter pixel. SM is proportional to the total amount of Ca2+ released by the spark (see materials and methods). The error bars in the trace represent SD due to the calculated photon noise. Bd: time course of Ca2+ current flowing through ryanodine receptors (RyR) [ICa(spark)], which was obtained by differentiation of SM (Bc) and calibrated without consideration of endogenous buffers.

A different approach is the signal mass method pioneered by Sun et al. (24) and developed in earlier work at our laboratory for wide-field microscopy (28), for which it is ideally suited. This is illustrated by the trace in Fig. 1Bc showing the Ca2+ signal mass, which differs from the ratio measure in the middle image in two ways. First, it tracks the total increase in fluorescence [FT(t) − FT(t0)], not the ratio to the background; second, the increase is gathered over a much larger region, in this case 41 pixels on a side. Moreover, because the imaging and fluorescence detection system uses wide-field microscopy, fluorescence is gathered from the entire thickness of the cell. The result, as shown elsewhere (28), is that the signal mass recorded in this way is directly proportional to the total amount of Ca2+, or the total charge due to Ca2+ released, during a spark. Hence the derivative of this trace as shown in Fig. 1Bd is a function of the underlying Ca2+ current passing through the RyR [ICa(spark)] from the SR to the cytosol.

Two features of the signal mass trace merit attention. First, as noted in an earlier study (28), the point in time at which the signal mass reaches its peak, i.e., when ICa(spark) terminates, corresponds to the time that the STOC begins to decay. Hence, the time from STOC to TTD provides an approximate measure of the duration of ICa(spark). Second, the signal mass trace remains in a plateau phase for a considerable time after reaching its peak, i.e., its decay is very much slower than its rise. This indicates that diffusion accounts for almost all the removal of Ca2+ from the vicinity of the release site once the peak is reached and suggests that sequestration or extrusion processes do not substantially diminish the magnitude of the measured signal mass. These features of the spark and STOC shown in Fig. 1 are representative of the spark population as a whole in these smooth muscle cells. Similarly, in cardiac cells, diffusion away from the release site appears to account for 80% of the fall in [Ca2+] during a spark (9).

Localizing Ca2+ Sparks and STOC in the Cell

It is a straightforward task to construct a map of spark locations in a cell from images such as those shown in Fig. 2A. Such a map is shown in two ways, in Fig. 2, B and C, which show about one-third of the cell. In Fig. 2B, each circle again indicates the occurrence of a single spark, with the center of the circle lying at the spark's epicenter pixel and the diameter of the circle proportional to the magnitude of ΔF/F0 at that pixel. In Fig. 2C, each occurrence of a spark is shown simply by a small, closed circle at the epicenter pixel, i.e., the brightest pixel during the spark. Figure 2B suggests that the amplitude of a spark at a given site is relatively constant, described earlier in materials and methods. Visual inspection of such maps also reveals that there are many spark sites per cell, considerably more than the 1–4 sites cited in earlier studies (12, 19). In the results below, we make a quantitative estimate of the number of sites and describe how occasionally a brief burst of 2–4 sparks (Fig. 2A, bottom) is generated at one location.

Fig. 2.

Ca2+ spark maps of cells revealing numerous spark-generating sites. A: continuous recordings of change in fluorescence from 2 different spark sites. Traces were acquired during 2 different recording periods. Images shown at right of traces were acquired when the designated sparks (a and b) reached their peaks. Insets show sparks on an expanded spatial scale. B: map of Ca2+ spark location, amplitude, and frequency for the cell shown in A. Ca2+ sparks were recorded during ten 2-s periods for a total of 20 s. Each circle denotes the occurrence of 1 Ca2+ spark, and the area of the circle is proportional to the peak intensity of the spark (not the spatial extent of the spark). Arrows labeled a and b indicate the locations of sparks a and b in A. Note that the cluster of sparks occurring in the region of b are not counted as occurring at the same site, because they are not all centered on the same pixel. C: a second map of Ca2+ spark location and frequency using the same data set used in B. Amplitude information is omitted to present a clearer indication of resolution of site location. Each solid circle denotes the location of a spark-generating site. The frequency of each site is color coded as follows: black, 1 spark; green, 2 sparks; red, 3 sparks; and blue, 4 sparks. Inset shows 1 region on an expanded scale.

Importantly for this study, the temporal correspondence between a spark and STOC, illustrated in Fig. 1, makes possible the localization of the BK channels giving rise to a STOC and hence the construction of a STOC map. This, of course, is not possible simply by using whole cell patch-clamp recording, which monitors STOC everywhere in the cell.

An Individual Spark Site Can Be Characterized by the Amount of Ca2+ Released Into the Cytosol During a Spark

We define a single spark site in the following explicit way to determine properties of individual sites in an objective fashion. Two or more sparks occur at the same site if they are centered on the same 333 × 333-nm pixel, that is, if the epicenter pixels for each spark are identical in location. To determine the consistency of spark properties at a given site, we examined those sites that produced two or more sparks. Figure 3Aa shows the total Ca2+ released (i.e., the signal mass) for individual sparks arising from this group of sites. The data are arranged by individual sites and ranked in ascending order along the abscissa according to the mean amount of total Ca2+ released per spark at each site. Figure 3, Ab and Ac, shows the ΔF/F0 and FWHM as determined from the ΔF/F0 image for the same group of sites, in each case following the same order shown in Fig. 3Aa, that is, according to the ascending order of the signal mass. Hence, data from the same spark site appear vertically above and below one another in all the plots shown in Fig. 3A.

Fig. 3.

Spark and STOC parameters at individual spark sites, ranked according to mean spark SM at a site. A: scatterplots of SM (a), fluorescence ratio (b), and full width at half-maximum intensity (FWHM) (c). Each point represents a single spark. Each position along the abscissa represents 1 of the 32 spark sites used for this analysis, so that points representing sparks at the same site appear above each other in vertical register. In a, b, and c, spark sites are ranked in ascending order according to the mean SM of sparks at each site. B: scatterplots of STOC amplitude (a) and time to onset of decay (TTD) (b). As in A, the sites are ranked in ascending order according to mean Ca2+ SM at a site. Note that the rank order by STOC amplitude is not the same as that according to spark SM. Omitted from the plots in B are data from sites where Ca2+ sparks failed to cause STOC or where the STOC parameters could not be measured adequately because of noise or the simultaneous occurrence of several STOC.

Examination of Fig. 3Aa conveys the impression that a given spark site releases a characteristic amount of Ca2+ into the cytosol. To address this question quantitatively, we asked whether the variation in signal mass from spark to spark within a site was significantly different from the variation between sites, using an analysis of variance measure. If all sites released the same amount of Ca2+ on average, the within-site variance likely would not differ from the between-site variance. This was not the case; that is, the between-site variance was significantly greater than the within-site variance (Table 1). Table 1 shows the data for all sites with multiple sparks and the results for those sites that generated STOC, which were 76% of the total in this data set. Sparks from 11 sites were excluded from the analysis because of uncertainty in measuring the corresponding STOC. In a larger data set (28), 21% of sparks were STOC-less. As shown in Table 1, the between-site variance was significantly greater than the intrasite variance for two other measures of the spark: ΔF/F0 and FWHM.

A Given Spark Site Elicits a STOC of Characteristic Amplitude

A second way to assess possible heterogeneity among spark sites is to examine the properties of the STOC that are generated at a given site. This approach has the advantage of being independent of the fluorescence measure of the spark amplitude, because the spark serves only to locate the STOC in the cell. The amplitude of STOC and TTD are shown in Fig. 3B for those sites that generated two or more STOC. In each case, the sites are ranked according to increasing signal mass, that is, in the same order as shown in Fig. 3A; hence data from the same site are in vertical register. Figure 4A shows the same data as Fig. 3Aa but ranked by site according to increasing STOC amplitude. As was the case with signal mass data (Fig. 3Aa), this plot suggests heterogeneity in the STOC sites. Again, using analysis of variance, we tested whether the intrasite variance in STOC parameters was different from intersite variance.

Fig. 4.

Scatterplots of STOC amplitude (A), ICa(spark) (B), and coupling ratio (C) at individual spark sites. As in Fig. 3, each point represents a single spark, and each position along the abscissa represents a single spark site. In each case, data are ranked in ascending order of the mean value of the parameter plotted. Hence, in A, sites are ranked from left to right in ascending order by mean STOC amplitude; in B, by mean ICa(spark); and in C, by mean coupling ratio. The numbers above the points lying on the x-axis in C indicate the number of observations at the site. Note that the values for ICa(spark) at those sites that are STOC-less are not shown in B, because a STOC TTD is needed to calculate ICa(spark).

As is evident in Table 1, intrasite variance was significantly less than the intersite variance for STOC amplitude, but not for STOC TTD. The fact that the TTD of the STOC is not significantly different from site to site is of interest because the TTD should reflect the duration of ICa(spark) (29). This finding indicates that there is no variability in the mean duration of ICa(spark) from site to site, and it suggests that there are not major differences in the open times and/or burst durations of RyR at different sites.

Spark Sites Differ in Number of Active RyR

The variation in signal mass among spark sites evident in Fig. 3Aa could result from either a difference in the number of RyR underlying a spark or a difference in spark duration. Because the number of RyR is proportional to the magnitude of ICa(spark), we examined the variation in ICa(spark) within and between sites. ICa(spark) was calculated by dividing the signal mass of the spark by the TTD of the elicited STOC and converting the quotient into units of current. These values for current cannot be corrected for nonmobile buffers, because the buffer kinetics are unknown. Thus, in their uncorrected sate, the values provide a minimum value for the peak ICa(spark) (29). The results are shown in Fig. 4B, where the sites are ranked in ascending order of the mean ICa(spark) at a site and in Table 1. These data indicate that ICa(spark), and hence the number of active RyR, varies from site to site. Note that Fig. 4A shows the data in ascending order of STOC amplitude and Fig. 4B shows the data in ascending order of ICa(spark) amplitude.

At Some Sites Sparks Consistently Elicit STOC, and at Others They Consistently Fail

RyR-BK coupling ratio varies from site to site.

The coupling ratio is defined for the purposes of this study as the ratio of the magnitude of the STOC, taken at an isochronic point that lies at or before the peak of all the STOC studied, to that of the eliciting ICa(spark) measured at the same isochronic point. For the purposes of this study, the isochronic point was 5 ms after the onset of the STOC. The coupling ratio provides a measure of the number of active BK channels to active RyR during each spark. The usefulness of the 5-ms isochronic point was described in a previous study (28). Briefly, this isochronic point ensures that neither spark nor STOC has gone beyond its peak value. Figure 4C shows these values for each site in ascending order of mean coupling ratio; the points lying on the abscissa represent sparks that failed to elicit STOC. Strikingly, the sites that failed to produce STOC failed without exception, and the sites that produced STOC also did so without exception. Moreover, examination of all spark sites showed that the within-site variance for the coupling ratio is significantly different from the between-site variance (Table 1). Hence, the coupling ratio varies from site to site. The results described in this section and the preceding three subsections indicate that the choice of a single pixel, 333 nm on each side, is valid within limits. That is, if our definition of a site were excessively large, then it would be possible to obtain results that showed all sites are equivalent; i.e., the intrasite variation would be no different from the intersite variation. For the intrasite variation to be less than the intersite variation, as we found, then the definition must have validity, although, of course, it is entirely possible that the definition could be refined further.

There are numerous spark sites in a single smooth muscle cell.

Simple visual inspection of Fig. 2C discloses many more spark sites than the average of 1–4 reported in previous studies (12, 19). This is true even when only those sparks that are far removed from one another are considered to occur at different sites. This result is also strikingly evident in videotaped imaging of sparks, even for rather short periods of time, an example of which is shown in Supplemental Fig. 1. Please refer to the Supplementary Material for this article to view movies.1 To determine the number of spark sites in quantitative fashion, it is necessary to define a spark site for this purpose. A difficulty in doing so is apparent from the inset of Fig. 2C, in which a number of epicenter pixels belonging to different sparks lie close together in a cluster. However, such clusters comprise only a relatively small (∼20%) proportion of the total number of sites observed and make little difference in the overall count. When we count such clusters as a single site, we find an average of 13 different spark sites per observed field (i.e., ∼⅓ cell), based on observations over a 10-s period (n = 9 cells).

Another approach to defining a spark is to rely on the knowledge we have acquired about the BK channels at the sites that give rise to a STOC. The site can be defined on the basis of the region over which BK channels are activated by a spark as follows. We demonstrated previously that BK channels underlying a STOC are exposed to a mean [Ca2+] that is ∼10 μM. Using this finding, we were able to determine, using reaction-diffusion simulations, the extent of the membrane area containing the BK channels activated by a spark, as shown in the plot of Fig. 5A. To produce this plot, the Ca2+ release site was placed 25 nm from the plasma membrane in accordance with the observed distance between the SR and plasma membranes (22). The x-axis in the plot indicates the profile of the [Ca2+] extending laterally along the membrane from the site of release. Depending on the buffer, the mean concentration of 10 μM is found at distances 150–300 nm from the release site. Because the pixel size in our experiments was 333 nm/side, we counted separately those sites that were separated by two intervening pixels (i.e., a site being defined as an area of 3 pixels/side). With this criterion, we estimated that, on average, there were 14 spark sites per field of observation (Fig. 5B).

Fig. 5.

Estimation of number of spark-generating sites based on the area of BK channels activated by Ca2+ sparks. A: Ca2+ concentration ([Ca2+]) spatial profile at the end of 15-ms pulses of 1 pA of ICa(spark) (50 μM fluo-3 + 27 μM fixed buffer; solid line) and 3 pA (50 μM fluo-3 + 230 μM fixed buffer; dashed line). Inset shows the expanded scale of the same plot and indicates the lateral distance from the Ca2+ release site where [Ca2+] reaches 10 μM in the 2 buffer conditions. The plot is adapted from Fig. 3B in ZhuGe et al. (29). B: histogram of number of spark-generating sites in the field of view per cell. Ca2+ sparks were recorded for 10 s, and a spark-generating site was defined as the volume subtended by an area of 1 μm2 in the x-y plane (i.e., 3 pixels in both x and y directions). The mean number of spark-generating sites per field of view was 14. Because ∼⅓ of the cell was in the field, the mean number of spark-generating sites for the entire cell was ∼42.

Because we estimated that, on average, roughly one-third of the cell was in the field of observation, the number of sites would appear to be closer to 42 per cell. To determine whether there were additional sites that remained silent over our observation period (10 s), we applied 0.5 mM caffeine, a concentration that increases STOC frequency from 2.4 to 5.9 per second in these cells (31). As shown in Fig. 6B, a greater number of spark sites were observed, ∼2.0 times as many as in the precaffeine condition (n = 4 cells). Hence the increase in STOC frequency upon application of 0.5 mM caffeine is due, for the most part, to recruitment of new spark sites. This in turn argues for a low probability of discharge at any given site. Adding this to the number of sites observed before caffeine yields a total of 39 spark sites per field of observation, on average, and ∼117 sites for the entire cell. This estimate takes into account a 20% overlap, discussed below. In contrast to the increased frequency, the amplitude of the sparks was not significantly different (Fig. 6B), so that, at least in this respect, the sparks observed in the presence and absence of 0.5 mM caffeine did not differ.

Fig. 6.

Application of low (0.5 mM) caffeine concentration reveals a large number of silent spark sites. A: spark maps based on 10-s observation in the absence (a) and presence (b) of 0.5 mM caffeine for the same cell. The map in c was produced by superimposing a and b. Each solid circle represents a Ca2+ spark; blue circles indicate sparks observed before caffeine, red, after caffeine, and green, those observed in both conditions. Inset shows an expanded representation of a region of c; each box corresponds to a single pixel, providing an indication of the resolution obtained. B, left: the sum of the number of Ca2+ spark sites from 4 cells in the absence of 0.5 mM caffeine and its presence (P < 0.05, paired t-test); right, comparison of Ca2+ spark amplitude under these 2 conditions. Amplitudes were not significantly different.

A consideration of the overlap of sites observed in the presence and absence of 0.5 mM caffeine indicates that the number of spark sites per cell is even greater. Approximately 20% of the sites that generated sparks in the absence of caffeine also generated sparks in the presence of caffeine. Hence, we can estimate that caffeine activates approximately one in five sites, and therefore the total number of sites is approximately five times the number of sites observed in the presence of caffeine. This estimate rests on the assumptions that the sites are independent and identical in their sensitivity to caffeine and that the sites observed before addition of caffeine are a random sample of the whole population. Thus the average number of total spark sites per cell is calculated as the number of sites observed in the absence of caffeine multiplied by 2.0, multiplied by 5, which equals 140 in the field of observation and likely ∼420 sites per cell. Our estimates are several orders of magnitude greater than observed previously using different technology in smooth muscle cells, but they correspond to morphological observations (Ref. 15; see discussion).

The rate of spark generation at a single site.

Because the spark sites are heterogeneous as outlined above, it was of interest to know whether they were heterogeneous in their rate of spontaneous discharge. If the sites were independent and all discharged at a uniform rate constant at a low probability, then the observed frequency of discharge likely would conform to a Poisson distribution. In constructing such a distribution of observations, it is critical to estimate the number of sites that fail to discharge during the period of observation. Each of the nine cells included in the data described in There are numerous spark sites in a single smooth muscle cell was observed for a total of 10 s. The total record for a cell comprised five 2-s periods separated by 30-s intervals during which the data buffers holding the images were read out. We used these 2-s intervals to measure the number of failures at an observed site. By “failures,” we mean the number of 2-s intervals during which no spark was observed at that site, given that we observed at least one spark at some time during the course of the five 2-s intervals. (For example, if a site generated 1 spark in one 2-s period and failed to generate sparks in the other four 2-s periods, we then credited this site with 1 spark and 4 failures.) The results are shown in Fig. 7, and the values plotted on the ordinate axis are the totals for the portion of all nine cells in the field of observation. The number of failures obtained simply by counting is 734 (hatched bar). However, this number leaves out the silent sites (i.e., sites that failed to spark even once in 10 s). On the basis of the results of the experiments with a low dose of caffeine (see There are numerous spark sites in a single smooth muscle cell), we estimate, for the nine cells, a total of 1,719 additional silent sites [additional silent sites = 5 × (no. of observed sites × 1.8)]. This number must be added to 734, giving a total of 2,453 failures (Fig. 7, open bar) for all cells over the entire 10 s. (Note that this is a minimum count, because we do not know the fractions of silent sites revealed by caffeine application.) A χ2 test shows that this is not consistent with a Poisson distribution (P < 8.6 × 10−7 that observed and expected distributions are the same). However, it is only a small number of observations (those sites yielding 2 or more sparks) that cause a deviation from a Poisson distribution.

Fig. 7.

Variation in Ca2+ spark frequency among sites. The number of Ca2+ sparks (ordinate) is the total from 9 cells (portion of the cell within the field of observation), each observed for a total of five 2-s periods. Hatched bar represents the number of failures observed, and open bar above hatched area represents the number estimated from the application of 0.5 mM caffeine. Closed circles and connecting line represent best fit of data by Poisson distribution. Inset shows expansion of the bins with frequencies >2 sparks per site every 2 s and the corresponding portion of the fitted line. The probability that the data actually correspond to such a distribution is 8.6 × 10−72 test).

For the reasons that follow, it appears that the deviation from a Poisson distribution may arise from occasional rapid bursts of sparks from a single site imposed on a low rate of spontaneous discharge. A simple calculation indicates that the overall rate of discharge was low. The mean rate of discharge at a given site was 1.25 times the number of STOC per second per cell (because ∼20% of spark sites are STOC-less), divided by the number of spark sites per cell. Performing this calculation results in 3/420 = 0.007 sparks/site/s, corresponding to an intrasite, interspark interval of 143 s. If we neglect the overlap with the caffeine experiments, the number of sites per cell is 126 [i.e., 42 + (2 × 42)], then the spark rate is 0.024 per site per second, and the intrasite, interspark interval is 42 s. For the set of observations recorded in Fig. 7, there were only 26 sites that generated more than one spark over the entire 10-s period. Of these, 19 sites generated multiple sparks occurring in short bursts in only one of the five 2-s recording intervals. Hence only seven sites discharged multiple sparks not in a burst, and these seven sites were found in a total of five of nine cells. In summary, it appears that the small but significant deviation from a Poisson distribution shown in Fig. 7 is most readily accounted for by bursting behavior that may well characterize all sites, discussed below.


We have examined Ca2+ sparks and the STOC that they generate at single spark sites in amphibian smooth muscle cells. We have found that there is greater variance among sites than within sites for the following spark properties: Ca2+ signal mass, which is proportional to the total quantity of Ca2+ released per spark; ΔF/F0, a second measure of spark intensity; STOC amplitude; ICa(spark), which provides a measure of the number of active RyR at a spark site; and the coupling ratio, which is a measure of the ratio of BK channels/RyR at a site. We conclude that individual spark sites differ in the number of RyR, in the number of BK channels, and in the ratio of the RyR to BK channels. In the present study, we have found no clear evidence for differences in the duration of ICa(spark) from site to site and thus no evidence for variation in duration that RyR remain open at different sites. In this limited respect, we have found no difference in the kinetics of RyR at different sites.

In a previous study (28), investigators at our laboratory found evidence of differences in the number of RyR in groups of sparks of different amplitude, in the number of BK channels in groups of STOC of different amplitude, and in RyR-BK coupling ratios in different groups of sparks. However, those results were derived from averaged data derived from groupings of sparks and STOC from different sites. Hence the differences found might indicate either that different sites varied differently in their properties or that all sites were uniform but had different functional characteristics at different points in time. For example, all sites might contain a constant number of RyR, and activation of a different fraction at different times might account for the apparent variation observed. The present study indicates that at least a part of the variation is due to differences in the properties of different sites.

Variation in Coupling Ratio and STOC-less Spark Sites

One of the most striking findings in the present study is that all spark sites that failed to generate STOC did so consistently and that spark sites that generated STOC also did so without exception. In a previous study, investigators at our laboratory (28) found that sparks with STOC differed from those without STOC, respectively, in the following ways: signal mass, 6.2 ± 0.3 × 10−20 M Ca2+ vs. 4.0 ± 0.2 × 10−20 M Ca2+ (P = 0.001); ΔF/F0 (%), 12.6 ± 0.5 vs. 10.5 ± 0.5 (P = 0.004); FWHM, 2.7 ± 0.1 μm vs. 2.2 ± 0.1 μm (P = 0.001); the t½ of decay, however, was not different: 16.0 ± 0.5 vs. 15.4 ± 0.91 (P = 0.269). Without STOC, values for ICa(spark) could not be calculated. On the basis of these previous results together with the present demonstration that STOC-less sparks arise from unique sites, we conclude that STOC-less sites differ from those that generate STOC in signal mass, ΔF/F0 and FWHM.

There are several possible explanations for the existence of STOC-less spark sites. First, the RyR might be located so far from the plasma membrane that BK channels are not activated by the Ca2+ discharged from the SR. However, this does not appear to be the case, for two reasons. First, 3-D imaging of Ca2+ sparks in feline esophageal smooth muscle preparation showed that STOC-less sparks are not located farther from the cell membrane than sparks that generate STOC, and they may even be somewhat closer, on average (13). Second, ultrastructural studies indicate that the SR and plasma membrane are quite closely positioned in smooth muscle, so that the RyR and BK channels are likely to lie within 12–20 nm of one another (22). A second possible explanation for STOC-less sites is that the BK channels there may lack the β1-subunit, resulting in a decreased sensitivity to Ca2+. Although this possibility cannot be excluded definitively, it seems to be unlikely, for three reasons. First, in an earlier study, the BK channels from the cell type used in the present study were examined in excised patches and found to have a Ca2+ sensitivity that indicated the presence of the β1-subunit (23). Second, with the exception of cells from transgenic mice with the β1-subunit knocked out, smooth muscle cells seem to be characterized by the presence of BK channels that incorporate the β1-subunit (4, 20). Third, sparks in β1-knockout smooth muscle cells generate STOC, albeit at smaller amplitude; the sparks are not completely STOC-less at 0 mV (4, 20). Hence, although this explanation seems unlikely, it would be of considerable interest if there were sites characterized by the absence of β1-subunits.

At the moment, it seems most likely that the STOC-less spark sites lie at one extreme of the variation in coupling ratio, with BK channels absent or not activatable at these sites. If the RyR and BK channels were in physical contact, either directly or via an anchoring protein, then a constant coupling ratio might be expected. The variability in coupling ratio makes this sort of structural arrangement less likely.

The possible existence of spark sites without BK channels is of interest, because it leads to speculation that there may be other targets for Ca2+ released during sparks in the microdomains of these sites. These targets might require a [Ca2+] in excess of 1 μM, and thus they would have to lie close to the RyR in the spark microdomain. The usefulness of this arrangement for specificity of Ca2+ signaling becomes apparent when one considers that the global [Ca2+] in the cells used here cannot be elevated to much more than 1 μM (1). Hence, targets with low Ca2+ sensitivity lying in the microdomain would be activated only by Ca2+ arising from sparks and would be protected from activation by global rises in Ca2+. In a different smooth muscle type, investigators at our laboratory previously showed that the same sparks activated both BK channels and Ca2+-activated Cl channels, with the latter in the form of STIC (30). No evidence for STIC was found in this preparation, hence cytosolic Ca2+ arising from sparks may have targets other than ion channels in smooth muscle. For example, investigators at our laboratory recently observed sparklike events in freshly dissociated single-nerve terminals from the posterior pituitary. We call these events syntillas (from scintilla, Latin for spark, from a nerve terminal, a synaptic structure), and our laboratory found that syntillas do not activate ion channels, although the precise role of syntillas is not yet clear (7). Moreover, because the SR network has contact with virtually every intracellular organelle, Ca2+ sparks may be capable of modulating a variety of cellular processes.

Number of spark sites.

We have found that there were many more spark sites than previously estimated in smooth muscle cells with the use of confocal imaging technology. For example, in rat cerebral arterial cells, the mean number of spark sites per cell was 2.1 (19), and in canine pulmonary arterial cells the mean number was 1–4/cell (2 times more in the presence of 100 μM caffeine) (12). In the present study, using a high-sensitivity, wide-field digital imaging technology, we directly observed 14 sites on average for the approximately one-third of the cell in view, leading to an estimate of ∼42 per cell. (Even if we defined a site as an area of 5 pixels/side, i.e., 1.67 μm2, we would have observed ∼13 sites per region of the cell in the field, hence ∼39 sites per cell.) Furthermore, our experiments with low levels of caffeine led to an even higher estimate and indicate that there are hundreds of spark sites per cell. Qualitative observations that we have made in other cell types suggest that these results are not due to a difference in the smooth muscle cell type or the animal species used. We have observed a large number of spark sites in smooth muscle cells from the guinea pig airway (30), the cat esophagus (13), and the rat cerebral artery (Supplemental Fig. 2), although we did not perform systematic quantitation of the number of sites in those cells. (Note that in Supplemental Fig. 2, 4 of the 6 images show 2 widely separated spark sites occurring at the same time.) It is possible that the difference could result from the expression of different types or different densities of RyR in different smooth muscle cells. Our findings, however, are consistent with immunocytochemical data (15) that show an extensive distribution of RyR in smooth muscle, with a meshlike pattern resembling the interconnected SR network.

Rate of spark discharge at a site and existence of bursts of sparks.

We estimated that spontaneous discharge at a spark site occurred, on average, at a rate of 0.007 s−1, or 1:143 s. This estimate was made simply by dividing the observed STOC frequency by the estimated number of spark sites. When we attempted to fit the frequency of spontaneous discharge at a site to a Poisson distribution, there was a small but highly significant deviation from the expected distribution. This deviation can be explained in two ways. First, different sites may discharge with different rate constants. This is similar to the conclusion reached by Gordienko et al. (10) and Bolton and Cannell (2) that there are high-frequency discharge sites in smooth muscle cells. Second, the fact that there are brief bursts of sparks could give rise to a compound Poisson process (21) that departs from a Poisson distribution. (In the latter model, all sites are identical and independent, but whenever a spark discharges, there is a certain probability that a brief burst of 2–4 sparks will occur.) We found no evidence that certain sites tended to burst more than others or that bursting sites had shorter interburst intervals, although we cannot rule this out. Given the present data, it is simplest to postulate that any site may discharge once or else burst each time it discharges. When such bursting behavior was eliminated from consideration, there were very few sites that discharged multiple times (<1 site/2 cells) over our observation period. Thus we favor the second explanation, i.e., the existence of spark bursts at a site, to account for the non-Poisson behavior. This, however, does not negate the existence of the high-frequency discharge sites of Gordienko et al. (10) and Bolton and Cannell (2), which are regions much larger in area than the single-pixel sites examined in the present study. These regions could simply be the result of a high density of identical single-pixel spark sites in one area of the cell. We have found evidence for such a nonuniform spatial distribution of sites in feline esophageal smooth muscle cells (13). Clearly, the kinetics of spark discharge at a site demand further examination in a quantitative fashion.

The existence of short bursts of two to four sparks within a 2-s period, although infrequent, suggests that the frequency of spontaneous spark generation with the usual 1:100 s interburst is not due to local SR Ca2+ depletion and gradual reaccumulation. If reaccumulation were required for another spark, then it certainly could occur within a second, as evidenced by the bursts, and could not explain an interspark duration of 143 s at a site. Moreover, sites where bursting is detected do not seem to be a distinct population; instead, bursting appears to be a relatively infrequent occurrence that may be a property of all sites (see above).

In summary, in the present study, we have found that sparks sites are heterogeneous in the number of RyR and BK channels and in the coupling ratio between the two, and certain sites are consistently “STOC-less.” The differences in coupling ratio allow us to distinguish certain classes of models for coupling RyR and BK channels in the spark microdomain, as discussed above. By inspection alone or using the spatial extent of the region containing the BK channels activated by a spark, we found the number of spark sites per cell was to be substantially greater than previous estimates of one to four per cell, most likely many hundreds of sites per cell, a finding consistent with morphological reports of an extensive SR in smooth muscle.

The results of the present study add to, and are consistent with, the picture of the biophysics of the spark-STOC microdomain found in an earlier study (29). In that previous study, investigators at our laboratory concluded that BK channels at the spark site were exposed to a [Ca2+] of 10 μM or more during a spark. This puts the BK channels and RyR within a radial distance of 150–300 nm from one another in a genuine microdomain, which we termed an “oasis.” This means that the cell membrane can house many such microdomains, and we have found in the present study that indeed it does. Given a BK channel density of 1 per square micrometer in this smooth muscle cell type (23), there must be thousands of BK channels per cell. With an estimate of hundreds of spark sites, a large number of these BK channels must be located at spark sites. In the old model of fewer than five spark sites per cell, most BK channels would be inaccessible to the Ca2+ arising from sparks. In this case, most BK channels would not be activated, except at positive potentials, given 1) the voltage and Ca2+ dependence of BK channels in smooth muscle cells, and 2) that global [Ca2+] rarely reaches a value as high as 1 μM even under intense stimulation and Ca2+ influx (1). Spark sites thus are a feature spread extensively over the SR and not only in one to five privileged regions. In addition, the BK channels residing in the SR are protected from activation by changes in global [Ca2+], responding more selectively to release from the SR at the spark sites. This also suggests that Ca2+-activated proteins with low Ca2+ affinities may be located in microdomains near the ER or SR Ca2+ release sites and not necessarily near the plasma membrane. Thus Ca2+-activated proteins with low Ca2+ affinities, whether located in the cytosol or in the plasma membrane, should be considered for candidacy as residents of Ca2+ microdomains. BK channels may simply be the first examples of such target proteins.


This study was supported by the National Heart, Lung, and Blood Institute Grants HL-21697 (to J. V. Walsh, Jr.) and HL-73875 (to R. ZhuGe) and by British Heart Foundation Grant PG/02/161/14788 and Wellcome Trust Grant 060094/Z/00/Z/JMW/CP/CF (to J. G. McCarron).


We thank Jeffrey Carmichael for excellent technical assistance.


  • 1 Supplemental data for this article may be found at http://ajpcell.physiology.org/cgi/content/full/00153.2004/DC1/.

  • 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. Section 1734 solely to indicate this fact.


View Abstract