INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Three ryanodine receptor (RyR) isoforms have been described in various mammalian tissues (Takeshima et al., 1989; Nakai et al., 1990; Hakamata et al., 1992) and related variants in avian and amphibian tissues (Sutko and Airey, 1996). The three mammalian RyRs share a high sequence identity (>65%), although key differences have been noticed and these may be related to tissue-specific functions. In muscle, RyR type 1 (RyR-1) is predominantly expressed in adult skeletal muscles, RyR type 2 (RyR-2) is the major adult cardiac isoform, and RyR type 3 (RyR-3) is present in smooth muscle and in embryonic striated muscle (Sorrentino and Reggiani, 1999). RyR types 1 and 2 play a central role in voltage-dependent activation of Ca²⁺ release in their respective tissues during excitation-contraction (EC) coupling (Coronado et al., 1994). Both isoforms interact closely with the dihydropyridine receptor (DHPR) and colocalize with DHPRs in junctions formed by the sarcoplasmic reticulum (SR) and t-system membranes (Block et al., 1988; Carl et al., 1995).

The contribution of RyR types 1 and 3 to EC coupling in skeletal muscle was addressed using RyR-1 or RyR-3 knockout mice (Takeshima et al., 1994, 1996; Bertocchini et al., 1997; Barone et al., 1998). Absence of RyR-1 leads to loss of EC coupling function, paralysis of skeletal muscles, and absence of postnatal survival (Takeshima et al., 1994; Nakai et al., 1996). The EC coupling function is unique to RyR-1 because neither RyR-2 nor RyR-3 supports skeletal-type EC coupling when expressed in RyR-1-deficient skeletal muscle cells (Nakai et al., 1997; Ward et al., 1999). Absence of RyR-3 affects some dynamic aspects of embryonic muscle tension but is not essential for mouse survival (Takeshima et al., 1996; Bertocchini et al., 1997). There are additional differences in sensitivity to cytosolic ligands that set RyR-1 and RyR-3 apart. Avian skeletal muscle is endowed with  and  RyR isoforms, of which  was shown to be homologous to mammalian RyR-3 (Ottini et al., 1996). Under the same ligand conditions, 60% of the openings of chick RyR type  channels are ~50-fold longer than the bulk of the openings of chick RyR type  channels (Percival et al., 1994). Also, type  channels remain open over a wider range of cytoplasmic Ca²⁺ and, in the presence of ATP, are not inactivated by Ca²⁺ (Percival et al., 1994). A long mean open time has also been reported in the mammalian RyR-3 channel (Chen et al., 1997). This result and the low sensitivity of mammalian RyR-3 to inactivation by Ca²⁺ (Sonnleitner et al., 1998) and by Mg²⁺ (Murayama and Ogawa, 1997) suggest that RyR-3 channels in situ could remain open for considerably longer periods than RyR-1 channels.

In heart and skeletal muscle the spontaneous activity of RyR channels in cells at rest results in Ca²⁺ sparks with highly stereotypic properties (Cheng et al., 1993; Klein et al., 1996; Conklin et al., 1999a). The RyR isoforms directly responsible for these spark activities are unknown. RyR types  and  in amphibian skeletal muscle and RyR-2 in mammalian ventricle are highly likely to engage in sparks because 1) sparks were originally reported in these adult tissues (Cheng et al., 1993; Tsugorka et al., 1995) and 2) so far, these are the only isoforms reported in these adult tissues. However, sparks were not observed in the adult rat EDL muscle (Shirokova et al., 1998) although they were clearly seen, albeit at a low frequency, in the adult mouse FDB muscle (Conklin et al., 1999b). Thus, whether mammalian RyR-1 or RyR-3 engages in stereotypic Ca²⁺ spark activity remains to be thoroughly elucidated.

We previously showed that RyR-3 augments the dimensions of spontaneously occurring Ca²⁺ sparks in embryonic mouse skeletal muscle cells (Conklin et al., 1999b). However, this study could not resolve whether RyR-3 by itself in the absence of RyR-1 could produce Ca²⁺ sparks. In the present study we used RyR-1 null myotubes, which are known to express RyR-3 (Takeshima et al., 1995), and RyR-3 null myotubes, which are functionally normal and are known to express RyR-1 (Takeshima et al., 1996) to test the separate contributions of RyR-3 and RyR-1 to Ca²⁺ sparks in embryonic skeletal muscle. As controls, we used double RyR knockout types 1 and 3 null myotubes. Spontaneous Ca²⁺ sparks were not detectable when both RyRs were knocked out. The Ca²⁺ spark parameters suggested that RyR-3 has the same ability as RyR-1 to generate these miniature Ca²⁺ release events. Part of these results were previously published in abstract form (Conklin et al., 1999c).

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

RyR-1 null and RyR-3 null mice

Screening of the RyR-1 and RyR-3 wild-type (wt) and null alleles was done respectively by PCR and Southern blot analysis using established protocols (Takeshima et al., 1994; Bertocchini et al., 1997). Double-null embryos were obtained by breeding mice heterozygous for the RyR-1 null allele (RyR-1^+/) and homozygous for the RyR-3 null allele (RyR-3^/).

Single cell preparations

Embryonic myotubes were enzymatically dissociated from intercostal muscles of embryonic day 18 (E18) RyR-1 null, RyR-3 null, and RyR types 1 and 3 double-null mice as described (Conklin et al., 1999b). Isolated cells were allowed to settle in a culture dish with the bottom replaced by a thin glass coverslip for at least 1 h before imaging. Embryonic cells were viable for several hours, as demonstrated by their ability to maintain a relatively constant low resting fluo-3 fluorescence and absence of Di-8-ANEPPS penetration into the cytosol. The total number of cells from which sparks were collected was 67 cells from 12 RyR-1 null embryos and 66 cells from 30 RyR-3 null embryos. No sparks were seen in >150 cells imaged from 10 RyR types 1 and 3 double-null embryos.

Ca²⁺ spark measurements

T-system and Ca²⁺ imaging was performed as described (Conklin et al., 1999b). Cells were allowed to settle at room temperature in Krebs buffer plus 2 mM CaCl₂ and stained with 17 µM Di-8-ANEPPS (Molecular Probes, Eugene, OR) for up to 10 min, followed by dye washout. Alternatively, cells were loaded with 4 µM fluo-3 acetoxymethyl (AM) ester (Molecular Probes) for up to 20 min. Cells were viewed with an inverted microscope with a 40× oil immersion objective (N.A. = 1.3) and a Fluoview (Olympus, Melville, NY) confocal attachment. The 488-nm spectrum line provided by a 5 mW argon laser attenuated to 6% was used for excitation of fluo-3 and Di-8-ANEPPS. The pixel size was 0.1-0.3 µm and the line-scan rate was 2.05 ms per 512-pixel line. Two-dimensional (2-D) images of Di-8-ANEPPS fluorescence were Kalman-averaged three times. Stock solutions of fluo-3 AM and Di-8-ANEPPS were prepared in DMSO.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

The top images of Fig. 1 show intercostal muscle cells from RyR-1 null, RyR-3 null, and double type 1/type 3 null embryos loaded with fluo-3 AM in Krebs solution containing 2 mM Ca²⁺ at room temperature. All cells were capable of maintaining a low average cytosolic fluorescence for several hours, although sites of persistent elevated Ca²⁺ were clearly present in some cases. Ca²⁺ sparks were identified during repetitive scans performed at a fast speed and low laser intensity to avoid photobleaching. On average, ~25% of the RyR-1 null or RyR-3 null cells produced Ca²⁺ sparks and typically, these cells had several simultaneously active sites. Single sparks can be seen in the images of RyR-1 null and RyR-3 null cells, and the small insets show the same location in the cell without the spark. In double-null cells we failed to identify sparks in >150 cells repetitively scanned for several minutes in each case. The inset in the latter image shows that exposure of a double-null cell to 1 µM thapsigargin resulted in a large elevation in cytosolic Ca²⁺. This result demonstrated that the inability of the double-null cell to generate sparks was not due to a generalized incapacity of these cells to store Ca²⁺ in the SR. We also used the cell-impermeant dye Di-8-ANEPPS to verify the integrity of the cell membrane and transverse tubular system (Shacklock et al., 1995) in the three cell types investigated. The bottom images of Fig. 1 shows that Di-8-ANEPPS fluorescence accumulated on the cell surface and in areas immediately adjacent to the cell surface. Some features of the nascent t-system, such as faint threads, were seen in all cells and are consistent with the longitudinal disposition of transverse tubules in mouse embryonic muscle seen in EM images (Franzini-Armstrong, 1991).

View larger version (44K): [in this window] [in a new window]   FIGURE 1   Top panels show confocal pseudocolor images of cells loaded with fluo-3 during a Ca2+ spark. Full-size images have a size of 512 × 512 pixels (0.2 × 0.2 µm per pixel) and were acquired at a rate of 1.2 s per image. Insets show the same area of the cell without the spark. An 8-pixel smoothing and a 32-color table were applied to highlight the location of the Ca2+ spark. Blue and red colors depict lower and higher fluorescence intensities, respectively. F/Fo = 0, 1, and 2 are indicated. The 5-µm scale bar is the same for the three top images. Bottom panels show confocal gray-scale images of embryonic cells stained with 17 µM Di-8-ANEPPS.

Fig. 2 shows line-scan images of Ca²⁺ sparks in each of the two RyR null cells. Time increases from left to right and only the region covered by the spark is shown in the image. Ca²⁺ sparks occurred repetitively at the same location, although the overall density of sparks was low in both cases. The overall rate of spark occurrence at sites engaged in the formation of sparks was ~1.9 sparks/scan (417 events captured in 221 scans) for RyR-3 null cells and ~1.6 sparks/scan (341 events captured in 208 scans) for RyR-1 null cells. In both cases, scans had a duration of 2.05 s. Considering the duration of the line-scan and the size of the cell, the spark frequencies were ~0.03 events/s/µm for both cell types. This frequency of spark occurrence represents a small fraction of that reported for depolarized cells, but it is close to that reported for frog skeletal muscle at rest (Klein et al., 1996). Inspection of the traces of the time course of fluorescence through the center of sites of sparks (Fig. 2, bottom) revealed a remarkable similarity between events in RyR-1 null and RyR-3 null cells. The spark onset was fast in both cell types and the peak amplitude and decay phases were also similar. In RyR-3 null cells there was a tendency for sparks to occur in rapid succession at the same site. These tandems were also noticed previously in normal cells (Conklin et al., 1999a) but were not observed in RyR-1 null cells.

View larger version (57K): [in this window] [in a new window]   FIGURE 2   Ca2+ sparks in embryonic myotubes. The top shows confocal line-scan images of fluo-3 fluorescence from sparks each in a separate cell. Images have a size of 20 µm (vertical) × 2.05 s (horizontal) and a pixel size of 0.2 µm × 2.05 ms. The bottom shows the time course of the change in fluorescence intensity during a spark obtained by integration of the line-scan fluorescence of the images above.

Histograms of the half-width (FWHM), the maximum fluorescence intensity (F/F_o) and the half-duration (FTHM) of individual sparks collected in RyR-1 null (341 events from 67 cells) and RyR-3 null (412 events from 66 cells) cells are shown in Fig. 3. Sparks occurring in rapid succession for which the fluorescence did not decay to the baseline between events were excluded from these distributions. The frequency distributions of the three spark parameters were similar, consistent with the visual inspection of the line-scan data. The distributions were roughly symmetrical in the case of the half-width and asymmetric in the case of the peak intensity and half-duration. However, in neither case were these distributions bimodal. The distributions of half-durations were skewed, with a mean longer than the mode. Thus, brief events were overrepresented in both cases. A better appreciation of the kinetics of the spark was obtained by plotting the three parameters of a single spark, namely FWHM, FTHM, and F/F_o, in three dimensions. These are shown at the bottom of Fig. 3. Three-dimensional (3-D) plots showed that there was a broad tendency for sparks to vary in intensity more than in dimension and duration. Accordingly, the shape of the 3-D distributions was roughly that of a vertical ellipsoid extending upward in the intensity (z) axis. In RyR-1 null cells there was a significant number of long-lasting sparks (FTHM > 100 ms, n = 36 events) clearly not present in RyR-3 null cells. This group of sparks also had a higher half-width and a higher peak fluorescence, and produced a scattering of the 3-D distribution toward the top and the back of the 3-D plot. This observation is consistent with recordings of the RyR-3 avian homolog channel, RyR type , in planar bilayers. Under the same ligand conditions, 60% of the openings of chick -RyR channels are ~50-fold longer than the bulk of the openings of chick -RyR channels (Percival et al., 1994). Notwithstanding the group of large sparks in RyR-1 null cells, the overriding conclusion when the data is taken as a whole is that the specific absence of the RyR-1 or RyR-3 isoform in each of the null cell types did not affect the spatial and temporal properties of the bulk of the Ca²⁺ sparks produced by the remaining RyR isoform.

View larger version (48K): [in this window] [in a new window]   FIGURE 3   Histograms of spark parameters and relationship of Ca2+ spark width to intensity and duration. The peak fluorescence intensity in F/Fo units, full-width at half-maximal fluorescence intensity (FWHM), and full-duration at half-maximal fluorescence intensity (FTHM), of sparks identified in RyR-1 null cells (341 events) and RyR-3 null cells (412 events). Plotted in three dimensions are the FTHM (x axis), FWHM (y axis), and F/Fo (z axis) of each spark.

We further established that sparks were mediated by RyRs by treating cells with caffeine. Panel A of Fig. 4 shows line-scans in cells during a control period and ~5 min after the addition of 0.1 mM caffeine to the bath solution. In RyR-1 null cells Ca²⁺ sparks were prolonged and, in some cases, the frequency of events was also increased. In RyR-3 null cells this caffeine concentration produced no significant effect. This result is consistent with previous observations (Bertocchini et al., 1997) showing that RyR-3 null cells are remarkably insensitive to caffeine. Several sparks from the same site were averaged before and after exposure to caffeine and are shown in panel A as cropped line-scans. The time course of fluorescence change through the center of the spark average is shown in panel B. In RyR-1 null cells there was a significant prolongation of the spark duration, but the peak fluorescence remained unchanged. Averaged for three sites, the fold-change of the Ca²⁺ spark parameters produced by 0.1 mM caffeine in RyR-1 null cells vs. RyR-3 cells were +1.09 ± 0.07 vs. 1.03 ± 0.03 for the change in F/F_o, +1.91 ± 0.24 vs. 1.01 ± 0.03 for the change in FTHM, and +1.17 ± 0.11 vs. 1.09 ± 0.01 for the change in FWHM (mean ± SE; + or  denote increase or decrease, respectively; 1.0-fold denotes no change). We also attempted to measure the response of cells to a higher caffeine concentration. Panel C of Fig. 4 shows compressed line-scan images in cells exposed to 1 mM caffeine. In the millimolar range, caffeine produced a large increase in background fluorescence in RyR-1 null cells or a moderate increase in RyR-3 null cells that in both cases was incompatible with spark measurements. However, this high caffeine concentration had no effect on the resting fluorescence of double-null cells. The main conclusion from these experiments was that RyR-3 channels were significantly more sensitive to caffeine than RyR-1 channels and, furthermore, all the Ca²⁺-mobilizing activity appeared to be mediated by these two RyR isoforms.

View larger version (47K): [in this window] [in a new window]   FIGURE 4   Stimulation of Ca2+ sparks by caffeine. Panel A shows line-scan images (15 µm vertical and 2.05 s horizontal) of sparks before and 5 min after the addition of caffeine at a final bath concentration of 0.1 mM. The cropped line-scans (7 µm × 205 ms) shown as insets are images of the same site obtained by averaging 5 sparks before and 8 sparks after in the RyR-1 null cell, and 8 sparks before and 6 sparks after in the RyR-3 null cell. Panel B shows the time course of the integrated fluorescence intensity of the average spark images above. Panel C shows line-scan images of the increase in cytosolic Ca2+ produced by a high caffeine concentration (1 mM). The two line-scan images for each cell type are before (left) and after (right) 1 mM caffeine. Image size is 60 µm (vertical) and 2.05 s (horizontal) in all cases.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

A first step toward understanding Ca²⁺ sparks in molecular terms is to establish which RyR isoform is capable of generating Ca²⁺ sparks. We previously showed that RyR-1 in the absence of RyR-3 engages in Ca²⁺ spark activity in embryonic intercostal muscle cells (Conklin et al., 1999b). The present data further demonstrated that RyR-3 in the absence of RyR-1 also engages in Ca²⁺ spark activity. Hence both isoforms have an inherent ability to engage in Ca²⁺ spark activity. Moreover, inspection of double knockout cells led us to conclude that the two RyRs in question were the only intracellular channels that produced the sparks. Intracellular Ca²⁺ release channels formed by the inositol trisphosphate receptor could have contributed to the events measured here, given their ability to generate a multitude of miniature local Ca²⁺ transients in amphibian oocytes (Yao et al., 1995). Moreover, it could be argued that spark-producing inositol trisphosphate receptors were down-regulated in the double-null muscle cell due to an unforeseen pleiotropic effect. The contributions of inositol trisphosphate receptors to the present results was deemed highly unlikely, because "spark"-like activities produced by inositol trisphosphate receptors are much more heterogeneous than those encountered here. Also, sparks in each of the two RyR null cells were found to be inhibited by ryanodine (not shown) and inositol trisphosphate receptors have not been convincingly shown to be expressed in skeletal muscle.

Not only did each RyR subtype produce sparks, but a detailed comparison of the kinetics of the sparks showed events produced by RyR-1 or RyR-3 were indistinguishable. These observations are significant for two reasons. First, RyR-1 and RyR-3 have fundamentally different roles, as RyR-3 channels are not activated by voltage and do not support EC coupling. Our results suggest that resting Ca²⁺ sparks result from an inherent activity of RyRs unrelated to EC coupling. Specific roles of Ca²⁺ sparks in embryonic muscle unrelated to the EC coupling function were previously discussed (Conklin et al., 1999b). Second, the density of RyR-3 in adult skeletal muscle is low (Jeyakumar et al., 1998), yet RyR-3 colocalizes with RyR-1 at junctional triads forming clusters in which both isoforms are present (Flucher et al., 1999). Our results suggest that in these clusters of RyR-1 channels essentially "doped" with RyR-3 channels, the impact of RyR-3 could be significant. An initial opening of an RyR-3 channel could activate RyR-1 channels present at a much higher density in the immediate surroundings of the activated RyR-3 channel. Hence, a single RyR-3 channel need only act as a trigger of a spark, while additional RyR-1 channels could quickly amplify the trigger Ca²⁺ to produce a full spark.

Additional inferences concerning clusters of RyRs can be brought about by comparing Ca²⁺ sparks in wt and RyR null cells. Table 1 summarizes the parameters of Ca²⁺ sparks reported previously in wt cells of the same age using the same microscope and imaging techniques (Conklin et al., 1999b) and those reported here for RyR-1 null and RyR-3 null cells. Sparks in wt cells lasted longer and were wider than in any of the two null cells. Also, sparks in wt muscle were brighter than those produced by RyR-3 in RyR-1 null cells. The single channel conductance is similar for RyR-1 and RyR-3 channels (Percival et al., 1994; Chen et al., 1997; Sonnleitner et al., 1998). Therefore, there are two explanations to consider for the difference in spark size and duration between the wt and RyR null muscle. Either the opening probability of RyR channels is higher in the wt cell or, alternatively, more RyRs were open during a spark in a wt cell than in each of the two mutant cells. The former possibility was considered unlikely because the overall frequency of sparks was considerably lower in wt cells, given the number of cells and the number of sparks collected from these cells (Table 1). In the context of the latter possibility, we suggest Ca²⁺ sparks could arise from densely packed clusters of RyRs such as those depicted in Fig. 5. Large clusters of RyRs, of the size depicted for E18 wt cells or larger, have been reported in primary myotubes cultured from embryonic muscle (Takekura et al., 1994). As indicated in Fig. 5, the specific absence of RyR-1 would lead to clusters composed exclusively of RyR-3. Briefer sparks in RyR-1 null cells could result from the fact that the density of RyR-3 is sparse, hence the probability of activating near-neighbor RyRs is reduced. It is also possible that these dilute clusters of RyR-3 channels could condense into small clusters of highly packed RyRs. Again, the expectation would be that sparks should be briefer and smaller because the size of the cluster is reduced. However, the specific absence of RyR-3 would lead to clusters of RyR-1 channels with defects in the packing lattice. These defects could be severe to the point that the activation of near-neighbor RyRs is compromised. It may also be possible that RyRs within this defective lattice could repack into smaller clusters. In either case, the parameters of sparks would be reduced.

View larger version (29K): [in this window] [in a new window]   FIGURE 5   Models of RyR clustering in embryonic muscle. The center diagram shows a cluster of RyR-1 (red) and RyR-3 (green) arranged in a conventional side-to-side disposition described in myotubes in cell culture (Takekura et al., 1994). Dotted RyRs represent EC-coupled RyRs. According to the tetrad size (26 nm on its side), the cluster is ~0.3 µm wide. RyR-3 in RyR-1 null muscle (top) and RyR-1 in RyR-3 null muscle (bottom) may be present as dispersed RyRs or in high-density clusters smaller than those present in wt embryonic muscle.

The ability of caffeine to activate RyR-3 channels has been difficult to determine in situ. Caffeine-induced SR Ca²⁺ release assayed by a Ca²⁺ indicator in permeabilized limb muscle bundles was found to be depressed in RyR-1 null compared with wt embryonic muscle. (Takeshima et al., 1995). This study suggested RyR-3 channels were less sensitive to caffeine than RyR-1 channels. We now find through confocal Ca²⁺ imaging that caffeine is at least two orders of magnitude more efficacious in RyR-1 null muscle than in RyR-3 null muscle and, therefore, RyR-3 channels must be considerably more responsive to caffeine than RyR-1 channels. In amphibian skeletal muscle the main effect of submillimolar caffeine (0.5 mM) is to increase the peak spark fluorescence ~1.3-fold (Gonzalez et al., 1999). In RyR-1 null embryonic cells, similar to previous determinations in wt cells of the same age (Conklin et al., 1999a), we found that the main effect of caffeine (0.1 mM) was to increase the event duration ~1.9-fold. A mechanism involving the recruitment of additional channels by caffeine was proposed to explain the increase in peak fluorescence (Gonzalez et al., 1999). In our case, we suggest that caffeine increases the ability of channels already committed to the spark to reopen a second time following the initial opening and closing event that produces the spark. Reopening of channels would be expected to lengthen the decay phase of the spark without necessarily increasing the Ca²⁺ released at the peak. Reopenings could be driven by an increase in the Ca²⁺ sensitivity of RyR-3 channels, which we suggest might be due to enhanced channel phosphorylation rather than to a direct "nucleotide effect" of caffeine on the RyR (Coronado et al., 1994). Direct stimulatory effects of caffeine on the RyR occur at concentrations 10- to 50-fold larger than those reported here. Moreover, the phosphodiesterase inhibitor IBMX (3-isobutyl-1-methylxanthine) prolonged the time course of sparks in RyR-1 null cells much like submillimolar caffeine (not shown). Sequence comparisons have shown 21 potential phosphorylation sites, 3 of which are common to all RyRs, and 7 others are only found in the RyR-3 (Marziali et al., 1996). In addition, one potential cAMP-dependent phosphorylation site KRXXS/T is conserved in all RyR-3 splice variants (Marziali et al., 1996). Mutagenesis and expression analyses followed by Ca²⁺ spark determinations should provide additional information on the molecular basis of the high caffeine sensitivity of RyR-3.