Cardiac-type EC-Coupling in Dysgenic Myotubes Restored with Ca²⁺ Channel Subunit Isoforms α_1C and α_1D Does not Correlate with Current Density

Abstract

Ca²⁺-induced Ca²⁺-release (CICR)—the mechanism of cardiac excitation-contraction (EC) coupling—also contributes to skeletal muscle contraction; however, its properties are still poorly understood. CICR in skeletal muscle can be induced independently of direct, calcium-independent activation of sarcoplasmic reticulum Ca²⁺ release, by reconstituting dysgenic myotubes with the cardiac Ca²⁺ channel α_1C (Ca_V1.2) subunit. Ca²⁺ influx through α_1C provides the trigger for opening the sarcoplasmic reticulum Ca²⁺ release channels. Here we show that also the Ca²⁺ channel α_1D isoform (Ca_V1.3) can restore cardiac-type EC-coupling. GFP-α_1D expressed in dysgenic myotubes is correctly targeted into the triad junctions and generates action potential-induced Ca²⁺ transients with the same efficiency as GFP-α_1C despite threefold smaller Ca²⁺ currents. In contrast, GFP-α_1A, which generates large currents but is not targeted into triads, rarely restores action potential-induced Ca²⁺ transients. Thus, cardiac-type EC-coupling in skeletal myotubes depends primarily on the correct targeting of the voltage-gated Ca²⁺ channels and less on their current size. Combined patch-clamp/fluo-4 Ca²⁺ recordings revealed that the induction of Ca²⁺ transients and their maximal amplitudes are independent of the different current densities of GFP-α_1C and GFP-α_1D. These properties of cardiac-type EC-coupling in dysgenic myotubes are consistent with a CICR mechanism under the control of local Ca²⁺ gradients in the triad junctions.

INTRODUCTION

Excitation-contraction (EC) coupling in muscle depends on the close interaction of a voltage-gated Ca²⁺ channel in the t-tubule or the plasma membrane with a Ca²⁺ release channel (ryanodine receptor, RyR) in the sarcoplasmic reticulum (SR). In cardiac EC-coupling this interaction is mediated by Ca²⁺ entering the cell through a voltage-gated Ca²⁺ channel that in turn activates Ca²⁺ release from the SR. In skeletal muscle, SR Ca²⁺ release is activated in the absence of Ca²⁺ influx, presumably by a physical interaction of the two Ca²⁺ channels. Under experimental conditions, the skeletal muscle Ca²⁺ release channel (RyR1) can also be activated by μM concentrations of cytoplasmic free Ca²⁺ (Nagasaki and Kasai, 1983; Smith et al., 1986). Whether Ca²⁺-induced Ca²⁺-release (CICR) in skeletal muscle occurs under physiological conditions, and if so, to what extent it contributes to skeletal muscle EC-coupling is still not resolved.

Reconstitution of Ca²⁺ channel null-mutant skeletal myotubes with wild-type and mutant heterologous Ca²⁺ channels has been a potent tool for analyzing the mechanism of EC-coupling in an intact muscle system. Normal function can be restored in dysgenic myotubes, which lack the skeletal muscle Ca²⁺ channel α_1S subunit, by heterologous expression of α_1S (Tanabe et al., 1988). When dysgenic myotubes are reconstituted with the cardiac Ca²⁺ channel α_1C subunit, “cardiac-type” EC-coupling is restored, which is characterized by its dependence on Ca²⁺ influx (Tanabe et al., 1990). Presumably the skeletal muscle Ca²⁺ release channel is activated by trigger Ca²⁺ entering the myotube through α_1C. However, not all examined Ca²⁺ channel isoforms that produced Ca²⁺ currents when expressed in dysgenic myotubes also triggered CICR from the SR. The neuronal isoform α_1A, which is not targeted into triads of dysgenic myotubes, only very rarely displayed evoked contractions (Adams et al., 1994) or Ca²⁺ transients in response to electrically induced action potentials (action potential-induced Ca²⁺ transients) (Flucher et al., 2000). Apparently the correct subcellular distribution of the channel is important for the activation of cardiac-type EC-coupling in skeletal myotubes. Skeletal-type EC-coupling properties could be conferred onto the cardiac α_1C subunit by replacing a short sequence of the cytoplasmic loop connecting the homologous repeats II and III with that of α_1S (Nakai et al., 1998). Thus, the chimera CSk53 was created, which combines skeletal-type activation of EC-coupling with cardiac current properties.

If CICR participates in normal skeletal muscle EC-coupling, it is unlikely that the trigger Ca²⁺ comes from the Ca²⁺ currents across the sarcolemma, because the slow activation of the skeletal Ca²⁺ current lags behind activation of SR calcium release (Garcia et al., 1989; Feldmeyer et al., 1990). Instead, trigger Ca²⁺ must be provided by Ca²⁺ released from the SR by the population of the skeletal muscle RyR1 that is under the direct control of the voltage-sensor α_1S subunit (reviewed in Rios and Stern, 1997). In other words, Ca²⁺ released through the RyR1 provides a positive feedback resulting in release of more Ca²⁺ from the same or adjacent RyRs. A recent report by O'Brien et al. (2002) indicates that, whereas CICR appears not to be necessary for activation of EC-coupling in skeletal muscle, SR Ca²⁺ release is reduced by fivefold when Ca²⁺ activation of RyR1 is impeded. This indicates a significant contribution of CICR to skeletal muscle EC-coupling.

Here we studied the properties of CICR in skeletal myotubes separate from skeletal-type activation of Ca²⁺ release by expressing nonskeletal α₁ subunits with distinct current characteristics and with distinct targeting characteristics in dysgenic myotubes. Analyzing the capability of different types of Ca²⁺ channels to generate action potential-induced and voltage-clamp-induced Ca²⁺ transients indicates that in the absence of direct skeletal coupling, Ca²⁺ currents of similar size as those of the skeletal α_1S are sufficient to trigger CICR. Moreover, the results indicate that activation of SR Ca²⁺ release seems to depend less on the magnitude of the whole-cell current than on the correct localization of the channels in the triads.

MATERIALS AND METHODS

Cell culture and transfections

Myotubes of the homozygous dysgenic (mdg/mdg) cell line GLT were cultured as described in Powell et al. (1996). At the onset of myoblast fusion (2 days after addition of differentiation medium), GLT cultures were transfected using FuGene transfection reagent (Roche, Basel, Switzerland). Cultures were analyzed 3–5 days after transfection.

α₁ subunit constructs

The following GFP-tagged Ca²⁺ channel constructs have been used: α_1S, α_1A, α_1C (Grabner et al., 1998), α_1D (Koschak et al., 2001), and α_1G (Monteil et al., 2000). GFP-CSk53: The BamHI-EcoRV fragment (nt 1265–4351; α_1C numbering) of clone CSk53 (Nakai et al., 1998), where part of the II-III loop coding α_1C cDNA sequence (nt 2549–2690) was replaced by α_1S sequence (nt 2156–2297), was ligated into the corresponding restriction enzyme sites of clone GFP-α_1C (Grabner et al., 1998) resulting in an N-terminally GFP-tagged CSk53.

GFP and immunofluorescence labeling

Differentiated GLT cultures were fixed and immunostained as previously described (Flucher et al., 1994; Flucher et al., 2000). For double-immunofluorescence labeling we used an affinity-purified anti-GFP antibody (Molecular Probes, Eugene, OR) at a final dilution of 1:4000 and the affinity-purified antibody 162 (Giannini et al., 1995) against RyR1 at a dilution of 1:5000. Alexa-488- and Alexa-594-conjugated secondary antibodies (Molecular Probes) at dilutions of 1:4000 were used to achieve a wide separation of the fluorescence signals. Alexa-488 was usually used with the anti-GFP antibody so that the antibody label and the intrinsic GFP signal were both recorded in the green channel. Controls, for example multiple combinations of primary and secondary antibodies and the omission of primary antibodies, were routinely performed.

Images of the double labeling experiments were recorded on an Axiophot microscope (Carl Zeiss, Jena, Germany) using a cooled CCD camera and Meta View image processing software (Universal Imaging Corporation, West Chester, PA). Quantitative analysis of the labeling patterns was performed by systematically screening the cover glasses for transfected myotubes using a 63× objective. The labeling pattern in transfected myotubes with more than two nuclei was classified as “clustered” when punctuate α₁ subunit fluorescence was co-localized with a similar RyR1 label (Fig. 1). The counts were obtained from several samples of at least three different experiments.

FIGURE 1.

The L-type Ca²⁺ channel subunit GFP-α_1D is targeted into skeletal muscle triads of dysgenic myotubes. Dysgenic myotubes transfected with GFP-α_1D were double immunofluorescence labeled with antibodies against GFP (top) and against the skeletal muscle RyR1 (middle). GFP-α_1D is distributed in clusters co-localized with RyR1 (examples indicated by arrows), which indicates a localization of GFP-α_1D in junctions of the SR with the plasma membrane or the t-tubules (triads). The color overlay (bottom) shows the co-localization of GFP-α_1D clusters (green) and RyR1 clusters (red) as yellow foci. The inset shows a fourfold enlarged area. Bar, 10 μm.

Patch-clamp and intracellular Ca²⁺ recording

Whole-cell patch-clamp recordings were performed with an Axopatch 200A amplifier controlled by pClamp 8.0 software (Axon Instruments Inc., Union City, CA, USA). The bath solution contained (mM): 10 CaCl₂, 145 tetraethylammonium chloride, and 10 HEPES (pH 7.4 with TEA-OH). Patch pipettes, pulled from borosilicate glass (Harvard Apparatus, Kent, UK) had resistances of 1.8 to 2.5 MΩ when filled with 145 Cs-aspartate, 2 MgCl₂, 10 HEPES, 0.1 Cs-EGTA, 2 Mg-ATP (pH 7.4 with Cs-OH). For simultaneous recording of whole-cell currents and Ca²⁺ transients the pipette solution additionally contained 0.2 mM fluo-4-K₅ (Molecular Probes). The Ca²⁺ fluorescence signal was recorded by a photometer system (PTI, S. Brunswick, NJ, USA) adjusted to the Zeiss Axiovert epifluorescence microscope. Ca²⁺ transients were normalized by the resting fluorescence (ΔF/F). To attempt block of SR Ca²⁺ release, ryanodine and ruthenium red (both Sigma-Aldrich, Vienna, Austria) were dissolved in absolute ethanol or water, respectively, and added to the pipette solution at final concentrations of 10–40 μM for ryanodine and 10–160 μM for ruthenium red.

Leak currents were digitally subtracted by a P/4 prepulse protocol. Recordings were low-pass Bessel filtered at 1 kHz and sampled at 5 kHz. Currents were determined with 200 ms depolarizing steps from a holding potential of −80 mV to test potentials between −40 and +80 mV in 10 mV increments. Test pulses were preceded by a 1-s prepulse to −30 mV to inactivate endogenous T-type Ca²⁺ currents (Adams et al., 1990). Current recordings of cells transfected with the T-type Ca²⁺ channel α_1G subunit were measured with 200 ms depolarizing steps from a holding potential of −90 mV to test potentials between −80 mV and +80 mV in 10 mV increments. Ca²⁺ current densities were normalized by linear cell capacitance and expressed in pA/pF.

Action potential–induced Ca²⁺ transients were recorded in cultures loaded for 45–60 min at room temperature with 5 μM fluo-4-AM plus 0.1% Pluronic F-127 (Molecular Probes) in HEPES and bicarbonate-buffered DME, as previously described (Flucher et al., 1993; Powell et al., 1996). Action potentials were elicited by passing 1 ms pulses of 30 V across the 19-mm incubation chamber. 0.5 mM Cd²⁺ and 0.1 mM La³⁺ were added to block Ca²⁺ influx and therefore allow discrimination between CICR and skeletal-type EC coupling. Application of 6 mM caffeine to the bath solution resulted in rapid Ca²⁺ release and thus proved the capacity of SR Ca²⁺ release.

Statistics

Data were expressed as mean ± SE where n is the number of cells examined. Data analysis, unpaired Student's t-test, and one-way-ANOVA analysis followed by the Tukey post test were performed with Clampfit 8.0 (Axon instruments, Union City, CA, USA) and GraphPad Prism software (GraphPad Inc., San Diego, CA, USA). P < 0.05 was considered statistically significant.

RESULTS

It has previously been demonstrated that heterologous expression of different muscle α₁ subunit isoforms in dysgenic myotubes restores EC-coupling by two distinct mechanisms. Whereas the skeletal muscle α_1S triggers SR Ca²⁺ release independently of Ca²⁺ influx through the L-type Ca²⁺ channel, the cardiac α_1C depends on Ca²⁺ influx for activation of EC-coupling, presumably by CICR (Garcia et al., 1994). In contrast to the two muscle isoforms α_1S and α_1C, the neuronal non-L-type α_1A subunit was not capable of restoring EC-coupling in dysgenic myotubes, although α_1A generated sizable Ca²⁺ currents (Adams et al., 1994; Flucher et al., 2000). Here we expressed α_1S, α_1C, α_1A, and for the first time the L-type Ca²⁺ channel α₁ subunit isoform α_1D in dysgenic myotubes to compare their current properties and targeting characteristics with their ability to restore EC-coupling.

GFP-α_1D is targeted into triads of dysgenic myotubes

Immunofluorescence analysis (Fig. 1) shows that a fusion protein of GFP attached to the N-terminus of α_1D (GFP-α_1D) is efficiently expressed in myotubes of the dysgenic cell line GLT and is distributed in a clustered pattern. Double labeling of GFP-α_1D and RyR1 demonstrated that GFP-α_1D clusters were co-localized with clusters of RyR1, thus identifying the clusters as t-tubule/SR or plasma membrane/SR junctions, all of which will from here on be called triads. GFP-α_1D/RyR1 coclustering was found in 62% of 454 analyzed GFP-α_1D-expressing myotubes, which was similar to the degree of clustering obtained with the native α_1S or with α_1C (see also Fig. 4 C). Myotubes in which GFP-α_1D was expressed in a nonclustered pattern typically showed ER/SR labeling (not shown), presumably because these myotubes were still very immature or due to unbalanced expression of the heterologous channel and endogenous muscle proteins (Flucher et al., 1994). Localizing GFP-α_1D (Fig. 1, green) in clusters coincident with the RyR (Fig. 1, red), makes α_1D the third member of the L-type Ca²⁺ channel family that, when expressed in skeletal muscle cells, is targeted into triads.

FIGURE 4.

The ability of Ca²⁺ channel isoforms to restore EC-coupling compared to their current densities and targeting properties. (A) Myotubes responding to field stimulation with Ca²⁺ transients (see Fig. 3) were counted in cultures transfected with GFP-α_1D, GFP-α_1C, GFP-α_1A, and GFP-α_1S. Counts from seven experiments are expressed in percent (mean ± SE) with the values of GFP-α_1D normalized to 100%. All three L-type channels (GFP-α_1D, GFP-α_1C, GFP-α_1S) restored EC-coupling with similar efficiency, whereas with GFP-α_1A only a single responding cell was observed. (B) Average peak current densities shown for the same channel isoforms as above (see also Table 1). Although the nonskeletal α₁ isoforms activate action potential-induced Ca²⁺ transients by CICR, no correlation of current density and restoration of EC-coupling exists (cf. A and B). (C) Fraction of myotubes in which the channel isoforms were targeted into the triads (see Fig. 1). The triad targeting properties of the Ca²⁺ channel isoforms exactly correspond to their ability to restore EC-coupling (cf. A and C).

GFP-α_1D restores Ca²⁺ currents in dysgenic myotubes

Whole-cell patch-clamp measurements of dysgenic myotubes expressing GFP-α_1D revealed Ca²⁺ currents with properties distinct from those of GFP-tagged skeletal and cardiac α₁ subunits (Fig. 2). Myotubes were selected for electrophysiological analysis based on the GFP fluorescence. Ca²⁺ currents in response to depolarizing steps from a holding potential of −80 mV to voltages between −40 and +80 mV in 10 mM extracellular Ca²⁺ were recorded and used to calculate the current-to-voltage (I/V) relationship of GFP-α_1D. Representative current traces for the three α₁ subunit isoforms are shown in Fig. 2 C. Comparing the average I/V curves of GFP-α_1D currents to those of GFP-α_1S and GFP-α_1C expressed in dysgenic myotubes (Fig. 2 A) shows that the average maximal current density of GFP-α_1D is 3.7 ± 0.4 pA/pF, which is close to the value obtained with GFP-α_1S but significantly lower (P < 0.01) than the average current density of GFP-α_1C (Table 1). Furthermore, GFP-α_1D activates at more negative potentials than the two other α₁ subunits. The point of half-maximal activation for GFP-α_1D is ∼13 mV more negative than that of GFP-α_1C and 37 mV more negative than that of GFP-α_1S (Table 1). The distinct activation characteristics can be best appreciated when the I/V curves are normalized for maximal amplitudes (Fig. 2 B). This analysis demonstrates that GFP-α_1D is functionally expressed in dysgenic myotubes and that its activation characteristics resemble those reported for α_1D in native cells and for GFP-α_1D in a nonmuscle expression system (Zidanic and Fuchs, 1995; Koschak et al., 2001).

FIGURE 2.

Current properties of GFP-α_1D expressed in dysgenic myotubes compared to those of GFP-α_1S and GFP-α_1C. (A) Voltage dependence of peak Ca²⁺ current densities (pA/pF; mean ± SE) of GFP-α_1D (▪; n = 42), GFP-α_1S (□; n = 33), and GFP-α_1C (•; n = 28). Maximal current density of GFP-α_1D is similar to that of GFP-α_1S and about four times lower than that of GFP-α_1C. (B) Normalized I/V curves show that GFP-α_1D activates at 13 mV more negative potentials than GFP-α_1C and at 37 mV more negative potentials than GFP-α_1S (for values of half maximal activation, see Table 1). (C) Representative whole-cell current recordings of GFP-α_1D, GFP-α_1S, and GFP-α_1C expressed in dysgenic myotubes. Cells were held at −80 mV and after a prepulse to inactivate low-voltage activated currents test pulses to increasing voltages from −40 to +80 mV were applied; currents were recorded in 10 mM Ca²⁺. GFP-α_1D activates faster than GFP-α_1S.

TABLE 1.

Properties of Ca²⁺ currents and Ca²⁺ transients recorded in reconstituted dysgenic myotubes with combined patch-clamp and fluo-4 Ca²⁺ measurements

α1-subunit	Current density (pA/pF)	Gmax (nS/nF)	V1/2 (mV)	k (mV)	(ΔF/F)max
α1D	3.7 ± 0.4 (42)	51.3 ± 5.3	−0.6 ± 1.2	6.7 ± 0.3	0.27 ± 0.03 *** (9)
α1S	3.1 ± 0.3 (33)	112.9 ± 6.8	36.0 ± 1.4	7.5 ± 0.3	0.96 ± 0.09 (34)
α1C	14.1 ± 1.1** (28)	236.2 ± 20.2	12.2 ± 1.6	6.2 ± 0.4	0.30 ± 0.03 *** (10)
CSk53	22.3 ± 2.2*** (35)	177.7 ± 18.0	5.9 ± 1.7	5.8 ± 0.5	0.41 ± 0.03 *** (18)
α1A	24.1 ± 4.5*** (18)	264.7 ± 55.9	6.0 ± 1.2	1.0 ± 0.2	0.50 ± 0.12 ** (12)

Entries correspond to mean ± SE; numbers of cells are given in parentheses.

Significance (^*** P < 0.001; ^** P < 0.01) is given in relation to α_1S.

(ΔF/F)_max values other than that of α_1S are not significantly different from one another.

GFP-α_1D restores cardiac-type EC-coupling in dysgenic myotubes

Next it was of interest to examine whether the small Ca²⁺ currents generated by GFP-α_1D would be sufficient to stimulate EC-coupling in dysgenic myotubes. This was first analyzed by recording intracellular Ca²⁺ signals in response to 1-ms current pulses passed through the recording chamber, in myotubes loaded with the fluorescent Ca²⁺ indicator fluo-4-AM. In normal myotubes and in dysgenic myotubes transfected with GFP-α_1S or GFP-α_1C, this field stimulation protocol has previously been shown to elicit Ca²⁺ transients with an all-or-none response to increasing voltages. This characteristic identifies the transients as action potential-induced Ca²⁺ transients (Flucher et al., 1993), which represent the most physiological parameter for assessing the restoration of EC-coupling. However, even in cultures transfected with the native GFP-α_1S, action potential-induced Ca²⁺ transients are observed only in a subset of expressing myotubes, presumably because excitability is achieved only in well differentiated cultured myotubes.

Cultures transfected with GFP-α_1D responded to field stimulation with brief Ca²⁺ transients (Fig. 3). The transients were characterized by a rapid upstroke that terminated abruptly, indicative of a strong voltage-dependence of both, the activation and the deactivation of the EC-coupling mechanism. The average amplitudes of the transients reached a (ΔF/F)_max of 0.46 ± 0.04, which was not significantly different from the amplitudes recorded with GFP-α_1C or GFP-α_1S (Table 2). The speed of the upstroke and the decay of the transients were similar between GFP-α_1D and GFP-α_1C; however, the average values obtained with these channel isoforms were significantly slower than those of GFP-α_1S. Unless the expression of nonskeletal channels adversely affects the overall differentiation of the myotubes, for which there is no independent evidence based on the morphology of the myotubes and on the expression of proteins seen with immunocytochemistry, this difference in the time course of the transients probably reflects the different modes of EC-coupling (see below). But most important, despite the differences observed in the whole-cell current properties (Fig. 2), time-to-peak, amplitudes, and time constant of the decline of the Ca²⁺ transients were not significantly different (P>0.05) between GFP-α_1D and GFP-α_1C (Table 2).

FIGURE 3.

Restoration of action potential-induced Ca²⁺ transients by GFP-α_1D in dysgenic myotubes. Field stimulation at low frequency (30 V, 1 ms, 0.3 Hz/0.5 Hz) evoked rapid Ca²⁺ transients in myotubes expressing GFP-α_1D, GFP-α_1C, or GFP-α_1S. Bath application of 0.5 mM Cd²⁺/0.1 mM La³⁺ immediately stopped the activation of transients in myotubes expressing GFP-α_1D (A) and GFP-α_1C (B), indicating that the transients were dependent on Ca²⁺ influx through the L-type channels. Addition of 6 mM caffeine (A, right trace) triggered a strong Ca²⁺ transient, proving that SR Ca²⁺ release was still functional during the Cd²⁺/La³⁺ block. (C) Persistent Ca²⁺ transients after addition of Cd²⁺/La³⁺ in myotubes expressing GFP-α_1S characterizes skeletal-type EC-coupling, which is independent of Ca²⁺ influx. In contrast, GFP-α_1D restored action potential-induced Ca²⁺ transients with cardiac characteristics. Marks underneath the traces indicate the approximate times of stimulation; bars indicate periods of Cd²⁺/La³⁺ and caffeine application.

TABLE 2.

Properties of action potential-induced Ca²⁺ transients recorded in reconstituted dysgenic myotubes with fluo-4 Ca²⁺ measurements

α1-subunit	Time to peak (s)	(ΔF/F)max	Decay time constant τ (s)	N
α1S	0.068 ± 0.010	0.65 ± 0.09	0.255 ± 0.019	26
α1D	0.370 ± 0.057 ***	0.46 ± 0.04	1.408 ± 0.170 ***	9
α1C	0.257 ± 0.030 ***	0.55 ± 0.10	1.113 ± 0.103 ***	18

Entries correspond to mean ± SE. Significance (^*** P < 0.001) is given in relation to α_1S; values of α_1D and α_1C are not significantly different from each other (P>0.05).

When Ca²⁺ currents were blocked by the addition of 0.5 mM Cd²⁺/0.1 mM La³⁺ to the bath solution while continuously stimulating, the action potential-induced Ca²⁺ transients in GFP-α_1D-transfected myotubes ceased immediately (Fig. 3 A). This dependence of the Ca²⁺ transients on Ca²⁺ influx was similarly observed with GFP-α_1C (Fig. 3 B) and is the hallmark of cardiac-type EC-coupling. In contrast, because skeletal EC-coupling is independent of Ca²⁺ influx, blocking Ca²⁺ currents with Cd²⁺/La³⁺ in myotubes transfected with GFP-α_1S did not stop action potential-induced Ca²⁺ transients (Fig. 3 C). To make sure that the inhibition of Ca²⁺ transients in myotubes expressing GFP-α_1D was indeed due to blocking the activation mechanism, SR Ca²⁺ release was tested by the application of 6 mM caffeine immediately after a blocking experiment. The strong Ca²⁺ release induced by caffeine (Fig. 3 A, right trace) showed that the failing response to field stimulation in Cd²⁺/La³⁺ was not due to depletion of the SR or other effects on the release apparatus. Thus, GFP-α_1D restored cardiac-type EC-coupling in dysgenic myotubes.

Restoration of cardiac-type EC-coupling depends on triad targeting but not on the magnitude of the current density

In the field stimulation experiments GFP-α_1D was as efficient as GFP-α_1C in restoring action potential-induced Ca²⁺ transients in dysgenic myotubes. In Fig. 4 A we compare the frequencies at which action potential-induced Ca²⁺ transients were observed in cultures transfected with GFP-α_1D, GFP-α_1C, GFP-α_1A, and GFP-α_1S. The number of responding cells in cultures transfected with GFP-α_1D was not significantly different from those of GFP-α_1C and GFP-α_1S. As previously shown (Flucher et al., 2000), GFP-α_1A very rarely produced action potential-induced Ca²⁺ transients, probably due to the fact that this channel is not targeted into the triads. Out of six culture dishes transfected with GFP-α_1A, only a single myotube in one dish responded with an action potential-induced Ca²⁺ transient; therefore, the bar is not visible at this scale (Fig. 4 A). Comparing the capability of each of these channel isoforms to restore EC-coupling with the average densities of their Ca²⁺ currents shows no correlation of the two parameters (cf. Fig. 4, A and B). For GFP-α_1S, which directly activates SR Ca²⁺ release, such a correlation would not have been expected. However, for the constructs which elicit cardiac-type EC-coupling—presumably by CICR—a correlation of current density and their ability to activate action potential-induced Ca²⁺ transients was to be expected. Most interestingly, GFP-α_1D, with a current density of less than a third of that of GFP-α_1C, activated EC-coupling just as efficiently as the cardiac channel. In contrast to current density, restoration of EC-coupling correlated very well with the triad targeting properties of the individual channel isoforms (cf. Fig. 4, A and C). Those constructs which are targeted into the triads (GFP-α_1S, GFP-α_1C, GFP-α_1D) all activated EC-coupling with similar efficiency. However GFP-α_1A, which generated the largest current density of all examined channel isoforms (see also Table 1) but is not targeted into triads, hardly ever activated action potential-induced Ca²⁺ transients in dysgenic myotubes. Thus, for the activation of cardiac-type EC-coupling, the spatial arrangement of the Ca²⁺ channels opposite the RyR appears to be more important than the pure size of the Ca²⁺ currents.

Ca²⁺ transients activated by GFP-α_1C and GFP-α_1D are qualitatively and quantitatively different from those activated by GFP-α_1S

To further characterize the dependence of cardiac-type EC-coupling on Ca²⁺ currents we performed combined patch-clamp and fluo-4 Ca²⁺ recording experiments in dysgenic myotubes expressing different α₁ subunit isoforms. This approach enables us to simultaneously monitor the whole-cell Ca²⁺ influx and intracellular Ca²⁺ transients in the same myotubes, and it allows us to analyze the voltage-dependence of Ca²⁺ transients. But these experiments differ from the analysis of action potential-induced Ca²⁺ transients in several important aspects: 1), Sampling—whereas with field stimulation we see and record only those myotubes that have achieved excitability and a good degree of EC-coupling (probably the peak of the population of expressing cells), with the patch-clamp approach we get a random sample of differently well developed and expressing myotubes; 2), Stimulation—instead of a 1-ms extracellular current pulse, which stimulates an action potential lasting only a few milliseconds, in the patch-clamp experiments we depolarize the cells for 200 ms giving rise to unnaturally long periods of Ca²⁺ influx; and 3), Intracellular milieu—while the myotube is being loaded with the Ca²⁺ indicator through the patch pipette, the cells are dialyzed and Ca²⁺ buffers and the entire intracellular milieu are altered. Thus, in this set of experiments we trade in better control and quantitative analysis for less physiological conditions. This is reflected in the shape of the Ca²⁺ transients in response to the 200-ms depolarization (Fig. 5 C), which does not resemble that of action potential-induced Ca²⁺ transients (Fig. 3).

FIGURE 5.

Simultaneous recording of whole-cell Ca²⁺ currents and Ca²⁺ transients in dysgenic myotubes transfected with skeletal and nonskeletal α₁ subunit isoforms. (A, B) Voltage-dependence of Ca²⁺ transients (A) and of Ca²⁺ currents (B). Ca²⁺ transients of GFP-α_1S (▪; n = 34) have a sigmoidal voltage relationship that does not follow the I/V curve. Transient-to-voltage curves for GFP-α_1C (•; n = 10), GFP-α_1D (□; n = 9), and GFP-α_1A (○; n = 11) are bell-shaped and follow their I/V curves with respect to activation properties but not with respect to amplitudes. Whereas current densities for GFP-α_1D are significantly smaller than those of GFP-α_1C (P = 0.0011) or GFP-α_1A (P = 0.014), their maximal amplitudes of Ca²⁺ transients are not significantly different from each other (P > 0.05). The dotted line shows the shift in the GFP-α_1A I/V curve after removing one exceptionally high recording from the analysis. But transients of all nonskeletal isoforms are significantly smaller (P < 0.001) than that of GFP-α_1S. (C) Representative examples of Ca²⁺ transients recorded with fluo-4 in parallel to whole-cell currents show the difference between skeletal and nonskeletal α₁ isoforms. (D) A scatter plot of the maximal peak current densities versus peak ΔF/F values for GFP-α_1S (▪), GFP-α_1C (•), and GFP-α_1D (□) shows that skeletal and nonskeletal Ca²⁺ signals clearly separate into two distinct groups. Superthreshold Ca²⁺ transients of GFP-α_1C and GFP-α_1D are confined to a narrow window of amplitudes over a wide range of current densities. (E) I/V curves of GFP-α_1C (•) and GFP-α_1D (□) (both the same as in B) compared to the average I/V curves of those currents for each group, which did not evoke Ca²⁺ transients (▴, GFP-α_1C, n = 5; ▵ GFP-α_1D, n = 16). For both channels, currents without transients were relatively smaller than those that evoked transients, but the absolute values differed significantly.

Nevertheless, important functional characteristics of the transients, like the skeletal- and cardiac-type activation of EC-coupling, can be distinguished in the voltage-dependence curves of Ca²⁺ transients (Garcia et al., 1994). The amplitudes of Ca²⁺ transients stimulated by GFP-α_1C mirror those of the Ca²⁺ currents over a wide voltage range (Figs. 5, A and B). In contrast, with GFP-α_1S Ca²⁺ transients activate at more negative potentials as the Ca²⁺ currents and stay fully activated at voltages near the reversal potential where currents decline. As expected from the dependence of action potential-induced Ca²⁺ transients on Ca²⁺ influx shown in Fig. 3, the voltage-dependence curves of Ca²⁺ transients in myotubes expressing GFP-α_1D showed the cardiac characteristics (Fig. 5 A). With this construct the Ca²⁺ transients activated in parallel with the currents at more negative potentials than those of GFP-α_1C and declined at voltages above +20 mV. Interestingly, the maximal amplitudes of the transients for GFP-α_1D and GFP-α_1C were not significantly different (P > 0.05) despite the more than threefold difference in current densities (Fig. 5; Table 1). This suggests that currents of both channels reach the activation threshold of CICR and trigger equal amounts of Ca²⁺ release. However, the maximal amplitudes of the Ca²⁺ transients activated by GFP-α_1D or GFP-α_1C were much lower than that activated by GFP-α_1S. This indicates that the native channel either activates a larger Ca²⁺ pool or activates SR Ca²⁺ release more efficiently than the two nonskeletal channel isoforms.

Unexpectedly, also GFP-α_1A, which in the field stimulation experiments restored action potential-induced Ca²⁺ transients very rarely (Fig.4 A), frequently produced cardiac-type Ca²⁺ transients in the combined patch-clamp/fluo-4 Ca²⁺ recording experiments. Apparently the mechanism that limited activation of cardiac-type EC-coupling in response to action potential-induced Ca²⁺ transients to the correctly targeted channel isoforms broke down under the conditions of the patch-clamp experiments. But even in this case, the average maximal amplitudes of Ca²⁺ transients were similar to those of GFP-α_1C and GFP-α_1D (Fig. 5). Whereas the transient-to-voltage relationship for GFP-α_1A is somewhat higher than that of GFP-α_1C and GFP-α_1D (Table 1), this difference is not statistically significant (P > 0.05) and mainly reflects a single recording with an exceptionally high current and transient. If this recording is excluded from the analysis, the transients of GFP-α_1A are also in the exact same range as GFP-α_1D and GFP-α_1C (dotted curve in Fig. 5 A). At this point we have no explanation for the occurrence of such exceptionally strong responders. However, one needs to remember that also in the field stimulation experiments occasional responders were observed with GFP-α_1A (see above; Adams et al., 1994; Flucher et al., 2000).

Analysis of the combined patch-clamp/fluo-4 Ca²⁺ recording experiments revealed striking differences between the skeletal and nonskeletal α₁ subunits expressed in dysgenic myotubes. Plotting the amplitudes of the Ca²⁺ transients versus the peak current densities for all individual experiments (Fig. 5 D) shows that dysgenic myotubes reconstituted with GFP-α_1S produce Ca²⁺ transients of greatly variable size (ΔF/F values up to 2.60) with no correlation to current densities. Ca²⁺ transients are observed even in myotubes with no detectable currents, and some of the myotubes with the highest current densities showed very small transients. In contrast, GFP-α_1C and GFP-α_1D display a completely different picture. Data points of both channels spread out over a wide range of current densities but within a relatively narrow window of Ca²⁺ transient amplitudes (ΔF/F values between 0.15 and 0.42). The maximal transient amplitudes of both nonskeletal channel isoforms do not reach those of the skeletal GFP-α_1S.

Furthermore, a fraction of GFP-α_1C- and GFP-α_1D-expressing myotubes showed Ca²⁺ currents but no detectable Ca²⁺ transient (33% and 64% of myotubes expressing GFP-α_1C and GFP-α_1D, respectively). This was never the case with GFP-α_1S. In Fig. 5 E a second set of I/V curves is shown for GFP-α_1C and GFP-α_1D. In addition to the I/V curve of those myotubes responding with both currents and transients (same as in Fig. 5 B), I/V curves were calculated from those recordings, in which Ca²⁺ currents were not accompanied with Ca²⁺ transients. For both channel isoforms the currents not associated with Ca²⁺ transients were significantly smaller (α_1D: P = 0.0001; α_1C: P = 0.0180) than those associated with transients. Thus, in both cases the larger currents activated transients whereas the smaller currents did not. This could be interpreted as the existence of a threshold for triggering the Ca²⁺ signal. However, the whole-cell current densities necessary to activate Ca²⁺ transients differed greatly between the two channel isoforms. The amplitudes of Ca²⁺ currents of GFP-α_1D that were associated with Ca²⁺ transients were almost identical to those of GFP-α_1C that were not associated with transients (Fig. 5 E). Thus, whereas current amplitudes matter for the activation of Ca²⁺ release in skeletal myotubes, the absolute values of whole-cell current densities do not reliably describe the activation threshold in the triad.

The two distinct groups of Ca²⁺ transient amplitudes do not depend on the skeletal versus cardiac EC-coupling mechanisms

The data presented above show that GFP-α_1C and GFP-α_1D expressed in dysgenic myotubes give rise to Ca²⁺ transients with an average maximal amplitude at about one third of that activated by GFP-α_1S (Table 1). Therefore it was reasonable to assume that the difference in skeletal versus cardiac EC-coupling mechanisms (direct or Ca²⁺-dependent) is responsible for this striking difference. This hypothesis was tested with a channel chimera, GFP-CSk53, which consists of GFP-α_1C with a 46-amino-acid sequence in the II-III cytoplasmic loop replaced by the corresponding sequence of α_1S (Nakai et al., 1998). This sequence is sufficient to confer skeletal-type EC-coupling properties onto the cardiac channel. Action potential-induced Ca²⁺ transients persist after application of Cd²⁺/La³⁺ to block Ca²⁺ currents (Fig. 6 E). This observation is in agreement with earlier reports showing that GFP-CSk53 activated EC-coupling by the skeletal Ca²⁺-independent mechanism (Nakai et al., 1998). Fig. 6, A–D shows the voltage-dependence of Ca²⁺ transients and currents of GFP-CSk53 compared to those of GFP-α_1S and GFP-α_1C. Peak current densities similar to those of GFP-α_1C (Fig. 6 B; Table 1) demonstrate the normal functional expression of the channel chimera. But surprisingly, on first sight the voltage relationship of the transients of GFP-CSk53 resembled that of the cardiac and not of the skeletal isoforms in both amplitude and shape (Fig. 6 A). However, the analysis of individual recordings from the combined patch-clamp/fluo-4 Ca²⁺ experiments showed that the cardiac-like decline of the transients at voltages above +20 mV was preferentially seen in those recordings where the baseline shifted upward during the duration of the experiment. Dividing the recordings into those with baseline shifts below and those above 10% (Fig. 6 C) revealed that the records with little or no shift displayed the typical skeletal characteristics of the voltage-dependence of the Ca²⁺ transients, i.e., no decline at higher test potentials. The same analysis performed with the recordings of GFP-α_1C showed typical cardiac characteristics in both groups (Fig. 6 D). Therefore, the apparent decline of the amplitudes of Ca²⁺ transients at positive potentials was an experimental artifact that could be avoided by excluding the data of recordings with a baseline shift greater than 10%.

FIGURE 6.

Comparison of the cardiac-skeletal chimera CSk53 with α_1C and α_1S. The voltage-dependences of Ca²⁺ transients (A) and of peak current densities (B) of CSk53 (□, n = 18) resemble in size and shape those of GFP-α_1C (•) but not those of GFP-α_1S (▪). The bell-shaped (cardiac-type) Ca²⁺ transient-to-voltage curve of CSk53 is in disagreement with skeletal-type EC-coupling properties as indicated by the insensitivity of action potential-induced Ca²⁺ transients to Cd²⁺/La³⁺ block (E). (C) Dividing the CSk53 records into two groups based on the occurrence and magnitude of a fluorescence baseline shift (above and below 10%) shows that the stable recordings (baseline shift < 10%; □, n =7; > 10%, ♦, n = 11) display a sigmoidal transient-to-voltage relationship. (D) In contrast, the transient-to-voltage curves of α_1C are bell-shaped regardless of the size of a baseline shift (baseline shift < 10%, C, n = 5; >10%, •, n = 5). Therefore, the bell-shaped appearance of the curve for CSk53 in A is due to this artifact, and the true character of the CSk53 transients is skeletal. (F) A scatter plot of maximal peak current densities versus ΔF/F values emphasizes the difference between Ca²⁺ transients of CSk53 and GFP-α_1S. Whereas current densities for CSk53 vary between 1.5 and 40 pA/pF, the amplitudes of the transients do not exceed ΔF/F values of 0.6.

GFP-CSk53 activates Ca²⁺ transients by the skeletal mechanism

Nevertheless, its maximal Ca²⁺ transient amplitudes were similar to those of the nonskeletal α₁ subunits (Fig. 6, A and F). The average maximal amplitude of Ca²⁺ transients for GFP-CSk53 was 0.41 ± 0.03, which is not significantly different from those of GFP-α_1C and GFP-α_1D but significantly lower than that of GFP-α_1S (P < 0.001; Table 1). Thus, the difference in amplitudes of Ca²⁺ transients observed in myotubes transfected with GFP-α_1C and GFP-α_1D on one hand, and with GFP-α_1S on the other, cannot be solely explained by the difference between the cardiac and skeletal activation mechanisms.

Cardiac-type and skeletal-type Ca²⁺ transients can be individually activated in the same myotube

Finally, we wanted to exclude the possibility that the relatively low Ca²⁺ transients in myotubes transfected with nonskeletal muscle α₁ subunit isoforms and the chimera CSk53 were simply due to a reduced Ca²⁺ release capacity of those myotubes. Therefore we examined whether the same myotube responded with different size Ca²⁺ transients when activated by the native GFP-α_1S or by a nonskeletal channel. The low-voltage-activated Ca²⁺ channel GFP-α_1G (Monteil et al., 2000) also gave rise to Ca²⁺ transients in dysgenic myotubes. These activated at potentials of −60 mV and peaked at about −40 mV, both at ∼40 mV more negative potentials than transients in GFP-α_1S-expressing myotubes (Fig. 7 A). Similar to the Ca²⁺ transients of all other examined nonskeletal α₁ subunits, GFP-α_1G transients reached maximal amplitudes that were substantially smaller than those of GFP-α_1S. When coexpressed with GFP-α_1S, Ca²⁺ transients activated by GFP-α_1G could be well separated from transients activated by the high-voltage-activated GFP-α_1S (Fig. 7 C). The transient-to-voltage curve of cotransfected myotubes displayed a shoulder at −30 mV and continued to rise at −10 mV (Fig. 7 A). Subtracting the transient-to-voltage curve of GFP-α_1G alone from that of the cotransfection experiments eliminated the shoulder, resulting in a curve with typical skeletal characteristics (Fig. 7 B). Thus, we conclude that the shoulder represents the contribution of GFP-α_1G. The height of this shoulder is near ΔF/F values of 0.3, which corresponds to transients observed with GFP-α_1C, GFP-α_1D, GFP-α_1A, and GFP-CSk53, and it is only about one third of the maximal Ca²⁺ signal activated by GFP-α_1S in the same cells at more positive voltages. Thus, in our experiments Ca²⁺ transients activated by channels other than the native GFP-α_1S peak at amplitudes significantly below that activated by GFP-α_1S, even in the same myotube.

FIGURE 7.

Cardiac-type and skeletal-type EC-coupling activated independently in the same myotube. (A) Voltage-dependence of Ca²⁺ transients in dysgenic myotubes transfected with GFP-α_1G alone (♦; n = 9) and in combination with GFP-α_1S (▪; n = 9). The low-voltage activated Ca²⁺ channel GFP-α_1G generates cardiac-type Ca²⁺ transients activated at negative voltages with maximal amplitudes (ΔF/F) not significantly different from those of GFP-α_1C and GFP-α_1D (cf. Fig. 5). The I/V curve of cotransfected myotubes displays a shoulder at negative potentials and continues to rise at positive potentials to levels similar to those achieved with GFP-α_1S alone (□; data from Fig. 5 A). Maintained transients at +80 V near the reversal potential characterize this component as skeletal. (B) Subtraction of the I/V curve of GFP-α_1G from that of the GFP-α_1G/GFP-α_1S-cotransfected myotubes results in a curve (•) resembling that of GFP-α_1S (□). (C) Examples of Ca²⁺ transients and currents from a cotransfected cell depolarized to −10 mV (left traces) and to +80 mV (right traces). Ca²⁺-dependent (cardiac-type) and Ca²⁺-independent (skeletal-type) transients can be activated independently from each other in the same cell.

DISCUSSION

In this study we investigated the properties of “cardiac-type” EC-coupling activated in skeletal muscle cells by reconstituting dysgenic myotubes with nonskeletal α₁ subunit isoforms. GFP-α_1D—an L-type Ca²⁺ channel isoform found in pancreatic β cells, hair cells of the cochlea, and in the sinus node of the heart (Hell et al., 1993; Iwashima et al., 1993; Kollmar et al., 1997; Takimoto et al., 1997)—was targeted into the triads and restored action potential-induced Ca²⁺ transients as efficiently as GFP-α_1C, even though its whole-cell current density was much smaller. In combined patch-clamp/fluo-4 measurements, Ca²⁺ transients activated by GFP-α_1C and GFP-α_1D peaked at amplitudes significantly below those of the skeletal GFP-α_1S. However, this difference in the size of Ca²⁺ transients was not due to the different EC-coupling activation mechanisms, because CSk53, a cardiac channel with skeletal EC-coupling characteristics, also gave rise to transients of reduced magnitude.

GFP-α_1D, together with the native α_1S and the cardiac α_1C, is now the third member of the voltage-gated Ca²⁺ channel family that has been demonstrated to become normally targeted into the skeletal muscle triads. Apparently all the members of the L-type subclass so far examined possess the triad targeting signals, whereas the neuronal, non-L-type α_1A does not (Flucher et al., 2000). We have previously shown that an important triad targeting signal resides in the C-terminus of α_1S and can be conferred onto GFP-α_1A by replacing a short C-terminal sequence with the corresponding sequence of α_1S (Flucher et al., 2000). Similarly the C-terminus of GFP-α_1D was able to confer triad targeting properties onto GFP-α_1A (Flucher and Grabner, unpublished results). Thus, triad targeting signals could be a conserved property of the L-type Ca²⁺ channel family. Alternatively, since α_1S, α_1C, and α_1D isoforms are all expressed in various muscle tissues whereas α_1A is not, the triad targeting property could be a muscle-specific feature.

Ca²⁺ currents generated by GFP-α_1D expressed in dysgenic myotubes showed the typical characteristics previously reported for α_1D in native cells (Zidanic and Fuchs, 1995) and of GFP-α_1D expressed in mammalian heterologous expression systems (Koschak et al., 2001). GFP-α_1D currents activated at 13 mV more negative potentials compared to GFP-α_1C. The current density of GFP-α_1D in dysgenic myotubes was less than one third of that of GFP-α_1C. Nevertheless, both channels restored action potential-induced Ca²⁺ transients in dysgenic myotubes equally well and as efficiently as the native GFP-α_1S. The fact that these Ca²⁺ transients were blocked by Cd²⁺/La³⁺ demonstrated that the mechanism by which GFP-α_1D activated action potential-induced Ca²⁺ transients was CICR, with the trigger Ca²⁺ provided by the L-type Ca²⁺ current. Therefore it was surprising that the considerable difference in the size of the currents of GFP-α_1D and GFP-α_1C was not reflected in the size of the transients or in the rate at which action potential-induced Ca²⁺ transients were observed. Thus, activation of CICR is not strongly dependent on the total influx of Ca²⁺ manifested in the whole-cell current densities. This conclusion is further supported by data on α_1A collected by us and others (Adams et al., 1994; Flucher et al., 2000; and present study). While α_1A expressed in dysgenic myotubes produces high current densities the size of GFP-α_1C or larger, it restores contractions or action potential-induced Ca²⁺ transients very rarely.

Whereas the ability of the examined α₁ subunits to trigger Ca²⁺ transients does not correlate with the magnitude of their Ca²⁺ currents, it correlates well with their triad targeting properties. All channel isoforms that are efficiently targeted into the triads (GFP-α_1S, GFP-α_1C, GFP-α_1D) also restore action potential-induced Ca²⁺ transients. On the other hand, GFP-α_1A fails to do either. This suggests that activation of EC-coupling by CICR is dependent on local Ca²⁺ concentrations achieved in the restricted space of the triad in the moment of depolarization by an action potential. In the triad, even the small GFP-α_1D currents—which are of similar size as those of α_1S—attain the concentration necessary for activating SR Ca²⁺ release. However, the Ca²⁺ activation threshold of Ca²⁺ release is not reached during fivefold larger Ca²⁺ currents generated by the nontargeted GFP-α_1A channel isoform. A local activation mechanism for CICR is also consistent with the observation that the termination of the Ca²⁺ release during action potential-induced Ca²⁺ transients is under the tight control of the membrane potential. At least in the dysgenic system, CICR is not a regenerative process that is self-sustained by Ca²⁺ released through the RyRs alone. In that case transients would be expected to continue to rise after repolarization and the deactivation of voltage-gated Ca²⁺ channels. On the contrary, the abrupt termination of the transients upon repolarization argues that the sustenance of the CICR requires the continued influx of Ca²⁺ through a voltage-gated channel. This tight control of Ca²⁺ activation of the Ca²⁺ release channels by voltage-gated Ca²⁺ currents in dysgenic myotubes reconstituted with GFP-α_1C or GFP-α_1D is reminiscent of EC coupling in cardiac myocytes (Cannell et al., 1987; Stern et al., 1999; Bers, 2000) and can only be envisioned in the restricted space of the triad junction, where α_1C or α_1D come within nanometers of the Ca²⁺ release channel (Rios and Stern, 1997).

A role of local Ca²⁺ gradients in the activation of CICR in skeletal myotubes is further corroborated by the observation that the lower limits of current densities required for triggering Ca²⁺ transients differ between GFP-α_1C and GFP-α_1D. In both cases failure of activation of Ca²⁺ transients was preferentially observed in myotubes with low current densities, whereas the myotubes with higher current densities usually produced Ca²⁺ transients. However, the current density required for the activation of Ca²⁺ transients was not constant. It varied between individual myotubes, and it was clearly different between GFP-α_1C and GFP-α_1D. The existence of a threshold of current density below which no Ca²⁺ release would be stimulated is expected for CICR. But this threshold should be the same in myotubes expressing GFP-α_1C and GFP-α_1D. Thus, we have to conclude that the whole-cell current densities only incompletely reflected the actual Ca²⁺ concentration at the activation site. Either the geometry of the channels in the triad is such that the spatiotemporal Ca²⁺ gradients required for Ca²⁺ activation of the RyR are similar for GFP-α_1C and GFP-α_1D even though their channel properties differ or, alternatively, the fractions of channels inside and outside the triad might differ for the two α₁ isoforms, so that there is no linear relationship between whole-cell current densities and current densities in the triads.

GFP-α_1A, which is not targeted into triads and usually fails to trigger action potential-induced Ca²⁺ transients, generated Ca²⁺ transients in the combined patch-clamp and fluo-4 Ca²⁺ recording experiments. On first sight, this result appears to contradict the hypothesis that triggering cardiac-type EC-coupling by CICR depends on the co-localization of the t-tubular and SR Ca²⁺ channels in the triads. However, if activation of CICR in the triads depends on local Ca²⁺ gradients, these could easily break down during the 200-ms depolarization of the patch-clamp experiments. During a brief action potential locally high Ca²⁺ concentrations near the mouth of a channel may activate SR Ca²⁺ release in the case of channels concentrated in the triad junction, whereas similarly high Ca²⁺ concentrations generated by α_1A channels diffusely located outside the triads may dissipate in the highly buffered cytoplasm before reaching the RyRs in the triad. However, during continued activation over the period of 200 ms, Ca²⁺ entering the myotube through α_1A may flood the cell to a degree that spatial gradients can no longer be maintained and CICR is triggered even by a channel positioned outside the triad. Interfering with the Ca²⁺ buffering milieu in the patched myotubes may further contribute to this inconsistent behavior under the different experimental conditions. The different shapes of Ca²⁺ transients produced by the field stimulation and in the combined patch-clamp recordings support this explanation (see also Garcia et al., 1994). The fact that in these experiments Ca²⁺ transients do not decline immediately after repolarization suggests that in contrast to action potential-induced Ca²⁺ transients, CICR stimulated by long depolarization contains a regenerative component, which sustains Ca²⁺ release even in the absence of influx. Also the slow decline of the Ca²⁺ transients indicates that the cytoplasmic Ca²⁺ buffering capacity has been exceeded under these experimental conditions. Applying shorter test pulses (10 ms) partially reversed this problem, and Ca²⁺ transients looked more like those of the action potential-induced Ca²⁺ transients (not shown). However, analyzing the voltage-dependence of current and transient activation and a reliable comparison of current densities of channels with different activation kinetics necessitated long depolarization. Therefore, even though analysis of Ca²⁺ signals under voltage-clamp control is common practice and has contributed significantly to our current understanding of EC-coupling (e.g., Baylor et al., 1983; Brum et al., 1987; Melzer et al., 1984; Beurg et al., 1997; Dietze et al., 1998; Jurkat-Rott et al., 1998; Grabner et al., 1999; Wilkens et al., 2001), it has to be kept in mind that spatiotemporal Ca²⁺ gradients, which appear to be critical for activation of CICR under physiological conditions, probably break down during prolonged patch-clamp recordings.

A trivial alternative explanation for the different size, shape, and behavior of Ca²⁺ transients obtained with patch-clamp stimulation would be that the transients recorded from nonskeletal Ca²⁺ channels in the patch-clamp mode are not CICR at all, but simply reflect the Ca²⁺ influx through these channels. This would explain the reduced amplitude compared to that of the skeletal transients as well as the fact that GFP-α_1A, which failed to restore action potential-induced Ca²⁺ transients, generates a Ca²⁺ signal under patch-clamp conditions. Attempts to test this possibility by pharmacologically isolating a possible Ca²⁺ influx signal from that of CICR did not resolve the issue either, because even concentrations between 10 μM and 40 μM ryanodine (Lipp et al., 2002) or 10 μM and 160 μM ruthenium red (Xu et al., 1999) in the patch pipette did not completely block Ca²⁺ release in control myotubes transfected with GFP-α_1S. However, important evidence from this and previous studies (Tanabe et al., 1990; Garcia et al., 1994; Garcia and Beam, 1994) argues against the possibility that these Ca²⁺ transients arise exclusively from Ca²⁺ influx: first, the poor correlation of current densities with the occurrence of associated Ca²⁺ transients in individual myotubes; i.e., myotubes with current densities as small as 3 pA/pF showed transients whereas other myotubes with much larger currents showed no measurable transients (Fig. 5 D) (see also Garcia et al., 1994); second, the poor correlation of current densities with the amplitudes of Ca²⁺ transient for GFP-α_1C and GFP-α_1D, i.e., whereas their current densities differed by almost threefold, GFP-α_1C and GFP-α_1D produced Ca²⁺ transients of similar magnitude (Fig. 5, A and B); and finally, Ca²⁺ transients triggered by the cardiac-skeletal chimera GFP-CSk53 were not significantly larger than those of GFP-α_1C, as would be expected from a construct that triggers skeletal-type Ca²⁺ release in addition to a fluo-4 signal reflecting pure Ca²⁺ influx. If we had merely recorded the influx of Ca²⁺, the magnitude of the Ca²⁺ transients should be highly correlated to the underlying currents. On the contrary, the behavior of the observed Ca²⁺ transients is consistent with a CICR mechanism that depends not only on the size of the Ca²⁺ influx but also on its local density.

Comparing the voltage-dependence of the Ca²⁺ currents with that of the Ca²⁺ transients revealed important properties of CICR in skeletal myotubes. For the nonskeletal channels, the voltage-dependence of the Ca²⁺ transients faithfully followed that of the current density, resulting in the bell-shaped curve typical for cardiac-type EC-coupling (Tanabe et al., 1990). Moreover, differences in the voltage-dependence of the currents, like the left-shifted activation of GFP-α_1D, can similarly be observed for the transients. This strict correlation of current amplitudes and transient amplitudes at different voltages clearly indicates that CICR is not simply an all-or-none mechanism. Smaller currents give rise to smaller transients and larger currents give rise to larger transients. Thus, CICR in individual skeletal myotubes functions as a graded amplification system in which Ca²⁺ entering the myotube through the voltage-gated channels is proportionally amplified by Ca²⁺ released from the SR. Nevertheless, considerable differences in current densities between individual cells and the differences in the average current densities of GFP-α_1C and GFP-α_1D are not reflected in the maximal amplitudes of the transients. These are very similar in all myotubes expressing nonskeletal α₁ subunits, and they peak at a level way below the maximal capacity of SR Ca²⁺ release. Thus, it seems that irrespective of their different current properties, GFP-α_1C, GFP-α_1D, GFP-α_1A, and GFP-α_1G activate the same maximal amount of Ca²⁺ release in the combined patch-clamp/fluo-4 recordings.

This reduced maximum of Ca²⁺ release achieved by the nonskeletal channels is, however, not due to a limited release capacity of the EC-coupling apparatus. Ca²⁺ transients activated by GFP-α_1S were on average three times larger than those of nonskeletal channels. This observation is consistent with similar differences seen in an earlier report comparing α_1S and α_1C expressed in myotubes of dysgenic primary cultures (Garcia et al., 1994). But perhaps myotubes reconstituted with the skeletal GFP-α_1S channel differentiate better than myotubes expressing the heterologous channel isoforms and therefore generate more robust Ca²⁺ transients? However, even on a single myotube level we observed that nonskeletal Ca²⁺ channels released only a fraction of the Ca²⁺ releasable by GFP-α_1S. Myotubes cotransfected with a low-voltage-activated channel GFP-α_1G and the high-voltage-activated GFP-α_1S showed the typical small cardiac-type Ca²⁺ transients at negative voltages followed by up to fivefold larger skeletal-type transients at positive voltages. This indicates that nonskeletal channels only partially activate the Ca²⁺ release capacity of dysgenic myotubes.

Interestingly, the cause of this differential ability to activate Ca²⁺ release is not, as one might think, the difference between skeletal and cardiac-type activation of Ca²⁺ release. GFP-CSk53, a cardiac-based channel chimera with skeletal EC-coupling properties (Nakai et al., 1998), also induced small transients like those of GFP-α_1C and GFP-α_1D. Apparently channel isoforms and chimeras other than α_1S activate SR Ca²⁺ release less efficiently than the native channel, and transferring skeletal EC-coupling properties onto a cardiac-type channel is not sufficient to overcome this deficiency. It may well be that other properties specific to the skeletal α_1S subunit enable α_1S to activate more Ca²⁺ release channels or the same number more efficiently than nonskeletal channels. Thus, the body of the channel—skeletal or nonskeletal—as well as the region in the II-III loop important for skeletal-type activation of EC-coupling seems to be important for the magnitude of Ca²⁺ release achieved in reconstituted dysgenic myotubes.

The combined evidence from this and previous studies allows drafting of the following model of the role of CICR in skeletal muscle contraction. CICR appears to be a significant component of skeletal muscle EC-coupling (O'Brien et al., 2002). To activate CICR in skeletal muscle, a threshold concentration of Ca²⁺ in the triads needs to be exceeded. During the brief depolarization of an action potential, this is achieved only by channels located in close range of the RyRs within the triad junction, but not by Ca²⁺ influx outside the triads. Interestingly, Ca²⁺ currents the size of the skeletal α_1S are capable of activating CICR, whereas SR Ca²⁺ release through the RyRs alone cannot sustain its own further release after repolarization. Thus, in normal EC-coupling CICR is not regenerative but amplifies the trigger Ca²⁺ signal while remaining under the strict control of the voltage-activated channels. Upon long depolarization, and possibly also during tetanic stimulation or certain pathologic conditions like malignant hyperthermia, the strict local control of CICR breaks down and regenerative activation of CICR in skeletal muscle may occur. Thus, the spatial arrangement of voltage-activated channels and SR Ca²⁺ release channels in the triad plays an essential part in the control of Ca²⁺ release in skeletal muscle EC-coupling.