The skeletal muscle L-type Ca channel serves a dual role as a calcium-conducting pore and as the voltage sensor coupling t-tubule depolarization to calcium release from the sarcoplasmic reticulum. Mutations in this channel cause hypokalemic periodic paralysis (HypoPP), a human autosomal dominant disorder characterized by episodic failure of muscle excitability that occurs in association with a decrease in serum potassium. The voltage-dependent gating of L-type Ca channels was characterized by recording whole-cell Ca currents in myotubes cultured from three normal individuals and from a patient carrying the HypoPP mutation R528H. We found two effects of the R528H mutation on the L-type Ca current in HypoPP myotubes: (1) a mild reduction in current density and (2) a significant slowing of the rate of activation. We also measured the voltage dependence of steady-state L-type Ca current inactivation and characterized, for the first time in a mammalian preparation, the kinetics of both entry into and recovery from inactivation over a wide range of voltages. The R528H mutation had no effect on the kinetics or voltage dependence of inactivation.
- dihydropyridine receptor
- L-type calcium channel
- human skeletal muscle
- cultured cells
- familial periodic paralysis
- patch clamp
The skeletal muscle L-type calcium channel occupies an important position in muscle physiology, serving both as a slowly activating voltage-dependent calcium channel and as a voltage-dependent activator of calcium release from the sarcoplasmic reticulum (Rios and Brum, 1987; Tanabe et al., 1988). The gating properties of the L-type Ca channel have been well studied in frog fibers (Cota et al., 1983; Sanchez and Stefani, 1983; Cota and Stefani, 1989; Feldmeyer et al., 1990; Francini et al., 1992, 1996), native and cultured rodent muscle cells (Beam and Knudson, 1988; Delbono, 1992;Delbono and Stefani, 1993; Dirksen and Beam, 1995), and artificial bilayers fused with t-tubule membrane vesicles (Ma et al., 1991, 1996;Mejia-Alvarez et al., 1991). However, the human isoform of the channel has only recently begun to be characterized (Garcia et al., 1992; Sipos et al., 1995, 1997; Jurkat-Rott et al., 1998). Studies to date have revealed important differences between the human isoform and its relatives in other species (Garcia et al., 1992; Sipos et al., 1997), underscoring the need for a comprehensive description of the gating properties of the human channel.
Hypokalemic periodic paralysis (HypoPP) is an autosomal dominant disorder of muscle in which patients experience episodes of weakness in association with a reduction in the serum potassium concentration. The episodes begin in adolescence and often wane by the fifth or sixth decade of life, although patients can experience a late-onset chronic proximal weakness accompanied by myopathic changes (Lehmann-Horn et al., 1994). In vitro studies of biopsied muscle fibers revealed that caffeine-induced contracture was normal in HypoPP fibers (Ruff, 1991) but that reducing [K+]o caused an unexpected paralytic depolarization of the fibers, whereas normal fibers became hyperpolarized under these conditions (Rüdel et al., 1984). Although HypoPP is primarily a disorder of muscle membrane excitability, microelectrode studies failed to identify an abnormal conductance or pump (Rüdel et al., 1984).
Linkage studies placed the genetic locus of HypoPP at chromosome 1q31–32, in the vicinity of the gene encoding the α1subunit of the skeletal muscle L-type calcium channel (α1S) (Fontaine et al., 1994). Further analysis demonstrated the existence of specific α1S mutations that cosegregated with the disease in families. Three mutations were found: R528H (∼50% of cases), a substitution of histidine for the outermost arginine residue in the S4 region of domain II of the channel (Jurkat-Rott et al., 1994); R1239H (∼50% of cases), an arginine to histidine substitution at the outer end of domain IV S4; and R1239G (only one family), a substitution of glycine for arginine at the same position (Ptacek et al., 1994; Fouad et al., 1997).
The location of the missense mutations within putative voltage sensors of the L-type Ca channel predicted an alteration of the voltage dependence of channel gating. In an effort to define the effect of the HypoPP mutations, Sipos et al. (1995) recorded Ca currents in myotubes cultured from HypoPP patients heterozygous for the R528H and R1239H mutations. R1239H myotubes had a 62% reduction in L-type Ca current density but exhibited no gating defect. In contrast, R528H myotubes had no reduction in L-type Ca current density but showed a remarkable 40 mV hyperpolarizing shift in the voltage dependence of steady-state inactivation. Lerche et al. (1996) failed to confirm this finding in human embryonic kidney cells transiently expressing the cardiac isoform of the α1 subunit (α1C) carrying the homolog of the R528H mutation (the β, α2δ, and γ subunits were coexpressed). The voltage dependence of both activation and inactivation was shifted 5 mV in the hyperpolarizing direction, and the current density was reduced by 38%.Lapie et al. (1996) recorded L-type Ca currents in mouse L cells expressing the rabbit α1S subunit carrying R528H and found that the voltage dependence of activation, the voltage dependence and rate of inactivation, and sensitivity to the dihydropyridine agonist Bay K 8644 were the same as in cells transfected with the normal construct. Most recently, additional recordings from human myotubes carrying the R528H mutation and recordings from dysgenic myotubes expressing R528H in the rabbit α1S subunit both failed to reveal a large shift in the voltage dependence of inactivation. Instead, a small parallel shift of 6–8 mV was found in the voltage dependence of both activation and inactivation (Jurkat-Rott et al., 1998). Thus, the gating defect initially found in human myotubes has not been reproduced, either in several heterologous expression systems or in the native human muscle cells in which it was originally described.
We recorded Ca currents in human myotubes cultured from three normal individuals and from a patient with the R528H mutation to define further the voltage dependence of gating for both the wild-type and the R528H L-type Ca channel. We found that the density of L-type Ca current was mildly reduced in HypoPP cells compared with normals. The mutation slowed the kinetics of activation but did not affect the voltage dependence of activation, the kinetics of deactivation, or the voltage dependence or kinetics of inactivation. Our findings suggest that the R528H mutation modestly reduces current density and affects a transition state along the activation pathway but leaves the channel otherwise unaffected.
MATERIALS AND METHODS
Cell culture. Human muscle satellite cells (myoblasts) were obtained from muscle biopsies taken from a HypoPP patient carrying the R528H mutation. The biopsies were performed for clinical diagnostic purposes in accordance with a protocol approved by the Subcommittee on Human Studies at the Massachusetts General Hospital. Normal satellite cells were obtained from muscle tissue discarded from surgery on three patients with no neuromuscular disease. Muscle samples were dissociated at 37°C in PBS (Life Technologies, Gaithersburg, MD) containing 1 mg/ml trypsin (Sigma, St. Louis, MO), 1.5 mg/ml collagenase (Sigma), and 1 mg/ml bovine serum albumin (Sigma). The digested muscle cells were centrifuged for 10 min at 1000 rpm, and the pellet was resuspended in high-glucose DMEM (Life Technologies) containing 4.5 gm/dl glucose, 25 mm HEPES (Mediatech, Washington, DC), 20% fetal bovine serum (FBS; Intergen, Purchase, NY), 2 mm l-glutamine (Sigma), 100 U/ml penicillin (Sigma), and 100 μg/ml streptomycin (Sigma) and was plated in a 100 mm Petri dish for growth in a 5% CO2 atmosphere at 37°C. When the freshly dissociated cells were 75–80% confluent, they were treated lightly with a solution of trypsin and EDTA (Sigma) in PBS, spun down, resuspended in growth medium containing 5% DMSO (Sigma), and frozen in aliquots at −80°C for later use.
Aliquots of myoblasts were periodically thawed and propagated (in a 5% CO2, 37°C incubator) in a growth medium containing MCDB 120 medium (JRH Biologicals, Lenexa, KS), 20% FBS, 2 mm l-glutamine, 100 U/ml penicillin, and 100 μg/ml streptomycin. After they had grown to confluency (3–4 d), myoblasts were induced to fuse and differentiate into myotubes by replacing the growth medium with a differentiation medium containing DMEM, 1.5% FBS, 2 mm l-glutamine, 100 U/ml penicillin, 100 μg/ml streptomycin, and 25 mm HEPES. The differentiation process took 4–6 d, after which myotubes could be maintained in the differentiation medium and used for recording for 2–3 weeks.
We verified that both the R528H and wild-type alleles were transcriptionally active in the primary cultures of HypoPP myotubes. RNA was isolated from a 75 cm2 confluent culture of myotubes using the Micro-FastTrack Kit (Invitrogen, San Diego, CA). Reverse transcription was primed with random hexamers, and a 140 base pair (bp) cDNA was amplified using PCR. Because the forward primer (5′-GGAGATCCTGCTGGTGGAGTCGGG) and reverse primer (5′-GGAAAGACTCAGGACTGATGTG) span an intron (∼800 bp), the 140 bp PCR product was amplified from complementary, and not genomic, DNA. The R528H mutation causes the loss of a BbvI restriction site within the amplified segment. The PCR product amplified from normal myotubes was completely cut by BbvI. The product obtained from HypoPP cells was only partially cut, consistent with expression of both wild-type and mutant alleles.
Calcium current recordings. Myotubes were detached from the bottom of the culture dish with a brief (3–5 min) treatment with trypsin and EDTA solution and were triturated gently using a 1 ml plastic pipette to create rounded myotube fragments. The myotube fragments were plated on 12 mm coverslips and allowed to settle for 30–60 min. Whole-cell patch-clamp recordings were obtained from the fragments using an Axopatch 200A patch-clamp amplifier (Axon Instruments, Foster City, CA) under control of a custom stimulation/recording program written in AxoBASIC and running on an IBM-compatible 486-based computer. Patch-clamp electrodes were fabricated from borosilicate glass capillary tubes (1.65 mm outer diameter; VWR Scientific, West Chester, PA) using a multistage puller (Sutter Instruments, Novato, CA), tip-coated with Sylgard, and fire-polished to a final tip resistance (in bath solution) of 2–4 MΩ. Whole-cell capacitance compensation was not used in the recordings because the large size of the myotube fragments [the average membrane capacitance was 123.9 ± 3.3 pF (n = 149) and 170.9 ± 7.6 pF (n = 101) for normal and HypoPP fragments, respectively] exceeded the maximal analog compensation for the CV201A head stage (100 pF). Series resistance was partially compensated by the analog circuitry of the amplifier. On average, the series resistance after compensation was 6.1 ± 0.2 and 5.6 ± 0.2 MΩ for normal and HypoPP fragments, respectively. For all of the cells used in this study, the voltage-clamp time constant was <1.4 msec (on average, τclamp was 0.70 ± 0.02 and 0.88 ± 0.03 msec for the normal and HypoPP fragments, respectively). The voltage error attributable to residual series resistance was always <5 mV. The amplifier output was filtered at 1 kHz and sampled at 2 kHz using an LM900 interface (Dagan Corporation, Minneapolis, MN). All experiments were performed at room temperature (22°C) except those studying the rate of deactivation; for those experiments, the recording chamber was maintained at 12–13°C using a Peltier device connected to a temperature controller (Medical Systems, Greenvale, NY).
Recording solutions. Solutions were designed to maximize calcium currents and to block sodium and potassium currents. T-type Ca currents were distinguished from L-type Ca currents using depolarizing conditioning pulses or, in a few cases, by subtraction of dihydropyridine (nitrendipine or nimodipine)-sensitive current. The bath solution contained (in mm): 120 TEA-Cl, 10 CaCl2, 1 MgCl2, 10 HEPES, 5 glucose, 0.002 TTX, and 0.1 EGTA, pH adjusted to 7.4 with TEA-OH. The pipette solution contained (in mm): 130 CsCl, 0.5 MgCl2, 10 HEPES, and 1 EGTA, pH adjusted to 7.2 with CsOH. Because rundown of the L-type Ca current did not occur even during long experiments, the internal solution was not supplemented with 5 mm Mg-ATP and 5 mm phosphocreatine as in the experiments of Sipos et al. (1995) and Jurkat-Rott et al. (1998). Nitrendipine was diluted from a 1 mm stock in DMSO to a final concentration of 0.5, 5, or 10 μm in the bath solution. Nimodipine was diluted from a 3 mm DMSO stock to a final concentration of 5, 10, or 50 μm.
Data analysis. Leak subtraction was performed on-line by scaling and subtracting the average response (n = 5–15) to 30 mV steps from the holding potential, which was usually −100 mV. Because the cell capacitance was too large to be fully compensated by the amplifier circuitry, the capacitance transients at the onset and offset of depolarized command pulses often saturated the patch-clamp amplifier. For this reason, leak-subtracted currents usually had subtraction artifacts at the onset and offset of the voltage pulse, which were routinely blanked from traces presented in the figures. Curve fitting was performed off-line using AxoBASIC or SigmaPlot (Jandel Scientific, San Rafael, CA). L-type Ca current density was calculated as the maximum current amplitude divided by the cell capacitance. Error bars in the figures indicate the SEM, and results presented in the text are means ± SEM. Student’st test was used to determine the statistical significance of differences between two sets of averaged data, whereas differences among multiple data sets were analyzed using one-way ANOVA and the “protected t test” [a version of Student’st test weighted for multiple comparisons (Welkowitz et al., 1976)].
L-type Ca current density in normal and R528H HypoPP myotubes
Figure 1 A shows the response of a normal myotube and a HypoPP myotube to a slow ramp protocol, in which the voltage was increased from −100 to +50 mV over a period of 1.25 sec (giving a rate of change of 0.12 mV/msec). In agreement with previous studies on cultured mammalian muscle cells (Beam and Knudson, 1988; Dirksen and Beam, 1995), the ramp protocol demonstrated two major components of Ca current. T-type current activated at −50 to −40 mV and showed prominent inactivation during the ramp, giving rise to a sharp peak at −20 to −30 mV. L-type Ca current activated slowly at much more depolarized potentials (0 to +20 mV) and showed little inactivation during the ramp (the reduction in inward current toward the end of the ramp occurred both because of the reduction in the driving force on calcium ions and because of a small amount of background outward current). As expected, the L-type component, but not the T-type component, was selectively blocked by the dihydropyridine drug nitrendipine (500 nm). The T-type current density was similar in normal and HypoPP cells (as in Fig.1 A). The density of L-type Ca current (measured as the maximum L-type Ca current during a 500 msec voltage pulse from −100 to +20 or +30 mV divided by the cell capacitance) was modestly reduced in HypoPP cells compared with all three sets of normal cells tested (Fig. 1 B). The mean current density values in the three sets of normal myotubes were 1.78 ± 0.09 pA/pF (normal #1; n = 149), 1.39 ± 0.15 pA/pF (normal #2;n = 30), and 1.50 ± 0.19 pA/pF (normal #3;n = 15), whereas the current density in HypoPP myotubes was 1.14 ± 0.06 pA/pF (n = 101). One-way ANOVA performed on the four data sets demonstrated that there was a significant difference among the means (F = 10.7;p < 10−5). When the differences in means of the four data sets were compared in all possible pairwise combinations using the protected t test (Welkowitz et al., 1976), the only statistically significant difference found was between normal #1 and the set of HypoPP cells (p < 0.001). The three normals were not significantly different from each other, and normal #2 and normal #3 were not significantly different from the HypoPP set. The current density in all sets of myotubes was quite variable, perhaps because of day-to-day and batch-to-batch changes in the degree of myotube differentiation.
In contrast to other studies of human myotubes (Rivet et al., 1992;Sipos et al., 1995, 1997; Jurkat-Rott et al., 1998), we could not detect a distinct “third-type” of transient calcium current in our myotubes. In both normal and HypoPP cells studied using voltage steps (n = 11), the current–voltage relationship of the transient current appeared to have a single maximum between −30 and −20 mV, with no additional increase above −10 mV (data not shown). Moreover, only one distinct peak, having the low threshold of the T-type current, was observed in slow ramp traces recorded in the presence of saturating nitrendipine or nimodipine (n = 5 normal and 5 HypoPP cells; see Fig. 1 for examples).
The range and steepness of activation are the same in normal and R528H HypoPP myotubes
Despite the location of the R528H mutation in the S4 region of domain II of the skeletal muscle L-type calcium channel, to date no study of this mutation has revealed a substantial effect on the steepness or position of the G(V) curve. We examined the voltage dependence of activation by measuring the peak L-type Ca current during a series of 1 sec test pulses (Fig.2). A 2 sec conditioning pulse to −30 mV was applied before each test pulse to inactivate the T-current, facilitating the detection of the peak L-type Ca current. Typically, as shown in the examples in Figure 2 A, the L-type Ca current was detectable at 0 mV, reached a maximum value at +20 mV, and then decreased with further depolarization because of the reduction in the driving force for calcium. The peak L-type Ca current was measured as a function of test potential and normalized to the maximum value (observed at +20 or +30 mV).
Families of L-type Ca currents were recorded in cells from one HypoPP patient and three normal controls. In Figure 2 B the average of the amplitude-normalized peak L-type Ca current measured in HypoPP cells and in one set of control cells is plotted as a function of test potential. Normal and HypoPP cells showed a similar threshold for activation of the L-type Ca current (−5 to −10 mV) and a similar position of the peak current (+20 to +30 mV). In Figure 2 Cthese data have been transformed to a conductance–voltage [G(V)] curve using the equation:G(V) =I(V)/(V −E rev), whereE rev, the reversal potential for the L-type Ca current, was estimated by linear extrapolation of theI–V curve. TheG(V) curves measured for normal and HypoPP cells overlap throughout the range of activation, from the threshold near −5 mV to the point at which conductance is maximal, around +40 to +50 mV. The solid and dashed lines are fits to the normal and HypoPP data, respectively, of the Boltzmann function:G(V)/G max = 1/(1 + exp[−(V −V 1/2)/k]), whereV 1/2 is the voltage at which the conductance is half maximal and k describes the steepness of activation. For the normal cells, V 1/2 = 10.2 ± 1.3 mV, and k = 7.4 ± 0.7 mV; whereas for the HypoPP cells, V 1/2 = 13.0 ± 1.9 mV, andk = 7.3 ± 0.7 mV (not significantly different).G(V) curves recorded in the two additional sets of normal cells were nearly identical (data not shown).
The rate of activation is slower in R528H HypoPP myotubes than in normal myotubes
To characterize further the voltage dependence of activation, we measured the rate of activation of L-type Ca current. A 2 sec conditioning pulse to −30 mV to inactivate the T-type current was followed by a 500 msec or 1 sec pulse to a series of test potentials. Traces were digitally smoothed with a 100 Hz Gaussian filter, normalized to peak amplitude, and then averaged over all cells to construct a mean activation time course for test potentials ranging from +10 to +50 mV. The mean time courses obtained in cells from the three control patients and the HypoPP patient are compared in Figure3. At all test potentials except +10 mV, the mean activation time course of the HypoPP current was markedly slower than the mean time course recorded in the three sets of normal cells.
Activation time courses were fit to the equation:I(t) = m(t) * h(t), wherem and h are single exponentials describing the activation and inactivation, respectively, of the channel, as has been done by a number of others (Mejia-Alvarez et al., 1991; Tanabe et al., 1991; Delbono, 1992; Garcia et al., 1992; Dirksen and Beam, 1995; Ma et al., 1996; Sipos et al., 1997; Jurkat-Rott et al., 1998). Figure4 A shows activation time courses recorded from one normal and one HypoPP cell using the protocol shown in the inset. To avoid interference by a residual capacitive transient (blanked in Fig. 4 A), we began the fits 8 msec after the start of the test pulse. The dashed lines shown with the traces are fits derived from the m *h model, and the estimated value of τm, the time constant of activation, is given with each trace. For most of the Ca currents we recorded, this model fit the activation time course well, implying that the activation process, most likely involving several transitions, was dominated by one rate-limiting eigenvalue. [The mean time courses that were calculated from the raw traces (Fig. 3) are more complicated because of the contribution of multiple single-exponential components.]
Figure 4 B, left, shows the voltage dependence of the activation time constants measured in the three sets of normal cells and in the HyopPP cells. As has been reported in other mammalian preparations, including neonatal mouse myotubes (Dirksen and Beam, 1995) and human myotubes (Sipos et al., 1997; Jurkat-Rott et al., 1998), the activation time constant in the normal cells was smallest at +10 mV and was only weakly voltage dependent between +20 and +50 mV, with a slight trend toward slower activation with increasing voltage. The activation time constant in HypoPP cells was slower at all test voltages and became progressively more sluggish compared with the control value at more depolarized test voltages. At test potentials of +40 and +50 mV, the channels from mutant myotubes activated almost three times as slowly as did the channels at +10 mV.
Activation kinetics were also quantified by a model-independent parameter, the time to half peak (t 1/2) of the activation time course. This parameter was measured from the same activation time courses that were used in the single-exponential fits, and the results are shown in Figure 4 B,right. The behavior of t 1/2 was qualitatively similar to the behavior of the activation time constant seen in Figure 4 B; in normal cells the time to half peak was only weakly dependent on voltage, whereas in HypoPP cells the time to half peak increased sharply with depolarization, showing a nearly fourfold difference in magnitude between +10 and +50 mV. Thus, a comparison of three different measures of activation kinetics (mean time courses, single-exponential fits, and measurements oft 1/2) suggested that the R528H mutation slowed the activation rate of the channel.
A recent study of the activation kinetics of the L-type Ca current in mouse muscular dysgenesis (mdg) myotubes expressing the cloned rabbit α1S subunit concluded that slow activation was correlated with low channel expression (Adams et al., 1996). This finding raises the possibility that the slower activation we observed in HypoPP myotubes was a consequence of the observed reduction in current density rather than a direct effect of the R528H mutation. In agreement with Adams et al. (1996), we found that lower current density was correlated with slower activation in both normal and HypoPP myotubes. In fact, τmappeared to be even more steeply dependent on current density in our cells than in the previous study; the slope was −5 to −12.2 msec per pA/pF in our cells, compared with −3.3 to −5.5 msec per pA/pF in mouse myotubes. Nevertheless, we found that the relationship between τm and current density was different for the normal and HypoPP myotubes at all test potentials. When τm (measured at +40 mV) is plotted against current density, the data from the HypoPP cells clearly fall along a different line than the data from the normal cells (Fig.5 A). Figure 5 Bshows the dependence of τm on test potential for only those cells having a current density between 0.5 and 1.0 pA/pF. The difference in τm between normal and HypoPP cells having similar current densities is just as striking as when all cells are pooled (as in Fig. 4 B,left). The slow activation seen in the HypoPP cells is therefore not solely caused by low expression and is likely to be an independent effect of the mutation, because it clearly exceeds what would be predicted from the normal variation of τm with current density.
The rate of deactivation is the same in normal and R528H HypoPP myotubes
Because the R528H mutation slowed the rate of activation, we also measured the rate of deactivation, which provides a measure of the energy barrier traversed by the channel during closing (between the open state and the nearest closed state). Because the settling time of the whole-cell clamp was not fast enough to record deactivation at room temperature (22°C), we used a Peltier device to cool the recording chamber to between 12 and 13°C, thus slowing the rate of deactivation and allowing tail currents to be distinguished clearly from the “off” capacitive transient.
Figure 6 A shows tail currents recorded after repolarization to various voltages after a 250 msec step to +50 mV. The sequence was preceded by a conditioning step to −30 mV to avoid T-type current contamination. Tail currents were fit with single exponentials (dashed lines superimposed on the traces). At −10 mV, the fit required a nonzero asymptote, suggesting that there were some reopenings occurring during the tail current. At −30 and −50 mV, reopenings are not expected [see theG(V) curve, Fig. 2 C], and the single-exponential relaxation reflects deactivation. Figure6 B shows the voltage dependence of τtail, measured at tail potentials between −50 and −10 mV in HypoPP cells and in one set of control cells. There were no significant differences in the rates of deactivation in the normal and HypoPP cells at all tail potentials tested. In both types of cells, the deactivation rate was only mildly voltage dependent, changinge-fold per 86 mV between −50 and −20 mV.
The voltage dependence of steady-state inactivation is the same in normal and R528H HypoPP myotubes
Sipos et al. (1995) reported that the voltage dependence of inactivation induced by a 15 sec conditioning pulse was shifted 40 mV in the hyperpolarized direction for L-type Ca currents in R528H HypoPP myotubes compared with normals, although this finding was not successfully reproduced in recordings made by the same investigators in myotubes from additional HypoPP patients (Jurkat-Rott et al., 1998). We measured the steady-state voltage dependence of inactivation using 60 sec conditioning pulses that preceded test pulses to +30 mV (Fig.7 A). The peak L-type Ca current measured during the test pulse is plotted as a function of conditioning voltage in Figure 7 B for the HypoPP cells and for one set of normal cells. In contrast to the findings of Sipos et al. (1995) but in agreement with those of Jurkat-Rott et al. (1998), the voltage dependence of inactivation was not markedly shifted in HypoPP myotubes. The solid and dashed lines are Boltzmann fits to the data from normal and HypoPP cells, respectively, withV 1/2 = −16.9 ± 2.3 mV andk = 8.7 ± 1.4 mV for the normal cells andV 1/2 = −14.0 ± 1.3 mV andk = 8.8 ± 0.94 mV for the HypoPP cells (not significantly different). Steady-state inactivation curves measured in another set of normal cells (data not shown) gaveV 1/2 = −18.0 ± 3.4 mV andk = 8.5 ± 2.1 mV (n = 4).
Kinetics of inactivation of the skeletal muscle L-type Ca current
Although R528H did not affect the voltage dependence of steady-state inactivation, it was still possible that the mutation affected the kinetics of inactivation. Moreover, although the rate of inactivation from the open state of the skeletal muscle L-type Ca current at depolarized voltages has been well studied in frog fibers (Cota et al., 1983; Sanchez and Stefani, 1983; Cota and Stefani, 1989;Francini et al., 1992) and mammalian preparations (Ma et al., 1991;Mejia-Alvarez et al., 1991; Delbono and Stefani, 1993; Caffrey, 1994;Jurkat-Rott et al., 1998), the rates of entry into and recovery from inactivation at more hyperpolarized voltages have not been studied comprehensively in any single species. For these reasons, we measured the rate of inactivation and recovery from inactivation of the L-type Ca current in HypoPP myotubes and myotubes from normal #1.
The rate of inactivation of L-type Ca channels from the open state was measured by applying 20 sec pulses to a range of strongly depolarized test potentials from a holding potential of −100 mV. Figure8 A shows the current measured in normal and HypoPP cells at +10, +30, and +50 mV. The solid lines are single exponential fits to the slowly decaying phase of the currents, whose time constants (τh values) are plotted as a function of voltage in Figure 8 B. τh was similar in normal and HypoPP myotubes, and on average, the rate of inactivation was approximately twice as fast at +50 mV (τh = ∼1 sec) as it was at +10 mV (τh = ∼2 sec).
The rate of entry into the inactivated state at more hyperpolarized voltages was measured using the three-pulse protocol shown in theinset of Figure 9 A. In each trial, the control response was measured during a 200 msec pulse to +30 mV. After a 5 sec gap at −100 mV, a variable-length conditioning pulse was issued, followed by a 100 msec gap at −100 mV and a 200 msec test pulse to +30 mV to measure the degree of inactivation. The traces in Figure 9 A show test currents recorded in one normal cell and one HypoPP cell after 1–90 sec conditioning pulses to −10 mV. To determine the rate of entry into the inactivated state, we plotted the test current, relative to the control current, as a function of T cond. Figure9 B shows that the kinetics of entry at −30, −10, and +10 mV was the same in normal and HypoPP myotubes. The time constant of entry τentry was determined by fitting a single exponential with a nonzero asymptote to each set of data; the dotted lines are fits to the mean data recorded in normal cells. The entry rate was strongly voltage dependent, occurring quickly and most completely at +10 mV, more slowly at intermediate voltages, and quickly but incompletely at more hyperpolarized voltages. This pattern was confirmed by measurements of the rate of entry at −20 and −40 mV (data not shown).
To measure the rate of recovery from inactivation, we used a similar three-pulse sequence, shown in the inset to Figure10 A. A control current was elicited by a 200 msec pulse to +30 mV to assay the amount of current in the absence of inactivation. After a 5 sec gap, a 20 sec conditioning pulse to +30 mV was applied to inactivate all of the channels, followed by a recovery step of variable length to allow recovery from inactivation and a test pulse to +30 mV to measure how much current had recovered. The traces in Figure 10 Ashow the test currents recorded in one normal cell and one HypoPP cell after recovery intervals ranging from 1 to 15 sec at −50 mV. The time course of recovery was measured by plotting the test current, relative to the control current, against T recovery. The rate of recovery, quantified by fitting a single exponential to the time course (dotted lines), was similar in normal and HypoPP cells at all voltages tested (Fig. 10 B). In all of the fits, the recovery curve relaxed to an asymptote that was lower than would be expected based on the steady-state inactivation curve (Fig. 7), suggesting that the fitted single-exponential phase was just the first phase of a slower recovery process.
Figure 11 shows τrecovery, τentry, and τh plotted together on a semilogarithmic scale. As shown also in Figures 8-10, the kinetics of inactivation is similar in normal and HypoPP myotubes. The solid line is a fit to the normal data of a two-state model of inactivation, in which the rates governing transitions of the L-type Ca channel to and from the inactivated state vary exponentially with voltage. The model predicts single-exponential relaxations having a τ equal to 1/[α(V) + β(V)], where α(V) and β(V) are the rates leading to and from the inactivated state, respectively. To model the apparent saturation of the time constants at extreme voltages (more obvious for τrecovery than for τh), α(V) and β(V) are made saturating functions of voltage (see Fig. 11 legend).
The primary motivation of this study was to test for a gating defect in the skeletal muscle L-type calcium channel caused by the hypokalemic periodic paralysis mutation R528H. A prerequisite of this goal was further characterization of the gating of the wild-type channel. We found two effects of the R528H mutation on the L-type Ca current in human myotubes. First, the current density was mildly reduced in heterozygous HypoPP myotubes. This observation agrees with the findings of Lerche et al. (1996) and Lapie et al. (1996), which showed reductions in current density of 38 and 75%, respectively, when the R528H mutation was expressed in heterologous systems without contamination from the wild-type allele. On the other hand, it contrasts with results from native muscle cells or muscular dysgenesis myotubes expressing the rabbit α1S subunit (Sipos et al., 1995; Jurkat-Rott et al., 1998). Second, the rate of activation of the L-type Ca current was slowed in HypoPP myotubes, which has not been observed previously. The steepness and midpoint of theG(V) curve and the deactivation rate were unaffected.
Activation of the skeletal muscle L-type calcium channel in human myotubes
In both normal and HypoPP cells, activation of the L-type Ca current followed a Boltzmann distribution, withV 1/2 of approximately +10 mV and k of ∼7 mV. This range and steepness of activation agrees fairly well withI(V) curves measured with 10 mm Ca2+ as the charge carrier in other mammalian systems, including neonatal mouse myotubes (Beam and Knudson, 1988; Dirksen and Beam, 1995), mouse mdg myotubes expressing the rabbit α1S subunit or a chimeric skeletal-cardiac L-type Ca channel (Garcia et al., 1997; Jurkat-Rott et al., 1998), human myotubes (Sipos et al., 1995, 1997; Jurkat-Rott et al., 1998), and adult human muscle fibers (Garcia et al., 1992). Under similar conditions, the range of activation in frog muscle is 30–40 mV more negative (Cota et al., 1983; Sanchez and Stefani, 1983; Cota and Stefani, 1989).
A single exponential was adequate in most cases to describe the kinetics of activation, in agreement with previous studies in mammalian systems including muscle fibers from human (Garcia et al., 1992) and rat (Mejia-Alvarez et al., 1991; Delbono, 1992; Garcia et al., 1992), mouse myotubes (Tanabe et al., 1991; Dirksen and Beam, 1995), human myotubes (Sipos et al., 1997; Jurkat-Rott et al., 1998), and bilayers containing the rabbit skeletal muscle L-type Ca channel (Ma et al., 1996). In frog fibers, activation of the L-type Ca channel is clearly sigmoidal and better fit by multiple-exponential schemes, such as the Hodgkin–Huxley m 3 * h model (Sanchez and Stefani, 1983) or more complicated multistate models (Francini et al., 1996). HypoPP is an autosomal dominant disorder, and therefore the HypoPP myotubes used in this and previous studies have all been heterozygous for R528H. Surprisingly, a single exponential described activation in the mutant cells as well as in the normal cells, suggesting that the time constant of activation of the mutant channels was never far enough from that of the normal channels to allow the resolution of two distinct components. The “pure” time constant of activation of the R528H channel is expected to be somewhat slower than the τm values we measured in R528H myotubes.
Although studies in adult frog and human fibers have shown that the rate of activation becomes faster with depolarization (Sanchez and Stefani, 1983; Feldmeyer et al., 1990; Garcia et al., 1992; Francini et al., 1996), we found that the rate of activation was essentially voltage independent in normal myotubes and became slower with depolarization for HypoPP myotubes. Sipos et al. (1997) and Jurkat-Rott et al. (1998) observed a similar effect in normal human myotubes; the single-exponential time constant of activation increased from ∼40–55 msec at +10 mV to 60–75 msec at +20 to +50 mV, with very weak voltage dependence at +20 mV and above. In neonatal mouse myotubes, Dirksen and Beam (1995) also saw an upward trend of τmwith voltage from ∼60 msec at 0 mV to 75 msec at +60 mV.
Caffrey (1994) fit the kinetics of activation with a sum of two exponential components (τ1 = 2–20 msec and τ2 = 10–180 msec) for L-type Ca currents recorded in mouse BC3H cells. Although the two components, which Caffrey (1994)described as “modes” of activation, each became faster with voltage, the relative contribution of the slower mode was greater at more depolarized voltages, leading to a modest slowing of activation overall. A different type of voltage-dependent switch between activation modes has been described for L-type channels in rat heart cells, which shift from a low-Po , slowly opening mode to a high-Po , rapidly opening mode when the cell is depolarized 25–50 mV above the half-maximal voltage of the G(V) curve (Pietrobon and Hess, 1990). By analogy with these results, the apparent slowing of activation with depolarization seen by us and others in human and mouse myotubes might result from a modal shift in activation gating, with the two modes having time constants that are too similar to be resolved well in myotube recordings.
Another possible explanation for the apparent slowing is contamination by a high-threshold component of fast calcium current such as the third-type current described by Rivet et al. (1992), which would be expected to contribute most significantly to the time course at 0 to +10 mV. However, this latter possibility seems unlikely given that (1) we were never able to isolate third-type current when we looked for it in normal and HypoPP myotubes and (2) third-type current, like T-type current, would be expected to be completely inactivated by a 2 sec prepulse to −30 mV (Rivet et al., 1992).
The rate of activation of the L-type Ca current in R528H myotubes was slower than in normal myotubes at all voltages, which is consistent with the location of R528H, a partial neutralization of a charged arginine residue, at the outer end of the S4 voltage sensor of domain II. Substitutions for charged S4 residues in heterologously expressed potassium, sodium, and calcium channels also cause changes primarily in the voltage dependence or kinetics of activation (Stühmer et al., 1989; Liman et al., 1991; Papazian et al., 1991; Chen et al., 1996;Garcia et al., 1997). Neutralization of the analogous (outermost) arginine in the S4 segment of domain II of the skeletal muscle Na channel to cysteine causes a slight rightward shift and modest reduction in slope of the G(V) curve, and modification of the mutant cysteine with the cationic sulfhydryl reagent MTSET has further effects on activation, both biasing the channel toward the open state and dramatically slowing displacement of domain II S4 back and forth across the membrane (Mitrovic et al., 1998). Replacement of the outermost arginine of domain IV S4 with histidine in the skeletal muscle Na channel has negligible effects on activation, instead disrupting the coupling of activation to inactivation (Chahine et al., 1994). Histidine has a pKa of 6–8 in proteins (Fersht, 1985; Lu and MacKinnon, 1995) and affects gating in a pH-dependent manner when engineered into the S4 region of the Shaker K+ channel (Starace et al., 1997) or the S4 region of domain IV of the sodium channel (Chahine et al., 1994). However, varying the extracellular pH from 6 to 9 did not alter activation rates in R528H myotubes (data not shown).
Surprisingly, the R528H mutation did not affect the position or steepness of the G(V) curve, suggesting that the apparent gating charge of the channel, as well as the equilibrium energy difference between states along the activation pathway, was unaffected. The effect of R528H on the activation rate, however, implies that this mutation affected a transition state along the activation pathway, thereby either slowing the rate-limiting step in activation or slowing some other step enough to make it rate-limiting. Because deactivation was unaffected by the mutation, it is unlikely that R528H affected the final step in a sequential gating scheme. However, if the channel is governed by a cyclic gating scheme, as proposed by several others (Feldmeyer et al., 1990; Ma et al., 1996;Sipos et al., 1997), the mutation could have affected the final step in the process of activation without affecting the rate of deactivation. Our results, which implicate the S4 segment of domain II in helping to set the activation kinetics of the skeletal muscle L-type Ca channel, contrast with chimeric studies and scanning mutagenesis studies in the mouse mdg myotube system that identify parts in or near the S4 regions of domains I and III, as opposed to domains II and IV, as the crucial determinants of the activation rate of the channel (Tanabe et al., 1991; Nakai et al., 1994; Garcia et al., 1997). It is possible that the activation rate is sensitive to selective alterations in the outermost part of domain II S4, a region that was not specifically targeted in these previous studies.
Although an early paper comparing normal and R528H human myotubes found that the mutation caused a striking 40 mV hyperpolarizing shift in the voltage dependence of steady-state inactivation (Sipos et al., 1995), a recent report from the same group failed to confirm this finding in myotubes from additional HypoPP patients and revealed only a modest (8 mV) leftward shift in the steady-state inactivation curve in R528H myotubes (Jurkat-Rott et al., 1998). Using 60 sec conditioning pulses, we found no difference in the voltage dependence of steady-state inactivation between normal and HypoPP myotubes. TheV 1/2 in normal and HypoPP cells was approximately −17 mV, which is more hyperpolarized than the value measured by Sipos et al. (1995) and Jurkat-Rott et al. (1998) in normal human myotubes (V 1/2 = −1 to −5 mV). This difference is most likely explained by the shorter conditioning pulse lengths of 15 and 20 sec, respectively, used in the previous studies, which did not allow inactivation to proceed all the way to steady state at hyperpolarized potentials (see below). OurV 1/2 of −17 mV was significantly more depolarized than the values measured in frog fibers [V 1/2 = −30 to −40 mV in 10 mmCa2+ (Sanchez and Stefani, 1983; Cota and Stefani, 1989)], cut rat fibers [V 1/2 = −48 mV in 2 mm Ca2+ (Mejia-Alvarez et al., 1991)], or lipid bilayers containing the rabbit channel [V 1/2 = −32 mV in 100 mmBa2+ (Mejia-Alvarez et al., 1991)]. The steepness factor was 8.7, in good agreement with values previously found in human myotubes [k = 8–9 mV (Jurkat-Rott et al., 1998)].
We also measured the rates of entry into and exit from inactivation at a range of voltages and found no significant differences between normal and R528H cells. The time course of entry from the open state at depolarized voltages or from closed states at hyperpolarized voltages was monoexponential and occurred on a timescale of seconds, as seen by many investigators (Sanchez and Stefani, 1983; Mejia-Alvarez et al., 1991; Francini et al., 1992; Delbono and Stefani, 1993; Caffrey, 1994;Jurkat-Rott et al., 1998). Where entry was slowest, at −10 to −20 mV, inactivation was complete in 60–90 sec, underscoring the importance of using long (≥60 sec) conditioning pulses to measure steady-state inactivation. Recovery from inactivation had an initial single-exponential phase (over the first 15 sec) that was incomplete, judging from the voltage dependence of steady-state inactivation. This phase became faster and more complete with hyperpolarization, reaching a point of saturation at −90 to −100 mV, but never accounted for >85% of the expected recovery, implying that recovery had a very slow phase that was completed during the 60–90 sec interval between pulse sequences.
Implications for HypoPP
Our study demonstrates that muscle cells cultured from a HypoPP patient carrying the R528H mutation express an abnormal L-type Ca current. In addition to mildly reducing the density of L-type Ca current, the mutation causes a selective slowing of the activation rate of the channel without affecting other gating properties of the ionic current, including inactivation.
Neither of these abnormalities immediately suggests a pathophysiological mechanism for HypoPP. Because the L-type calcium channel plays a dual role in muscle as a calcium channel and as a voltage sensor for excitation–contraction (E–C) coupling (Rios and Brum, 1987; Tanabe et al., 1988), it is possible that activation of the calcium release process is slowed as well as activation of the L-type Ca current. This could give rise to impaired E–C coupling and, if the impairment is severe enough, paralysis. In principle, this possibility can be tested by recording gating currents and by monitoring calcium transients with calcium-sensitive dyes. Although native L-type Ca channel gating currents have been successfully measured in mouse myotubes (Beam and Knudson, 1988), gating current measurements have not been attempted in human myotubes, where the L-type Ca current density is several-fold lower. Calcium transients, on the other hand, have been shown to be robust in human myotubes (Brown et al., 1995) and were recently recorded in HypoPP myotubes carrying the R528H mutation (Jurkat-Rott et al., 1998). In the latter study, calcium release was determined to be identical in normal and R528H myotubes.
Another possible pathophysiological mechanism is that the disruption of L-type Ca channel gating promotes paralysis by impairing electrical excitability. In vitro studies of biopsied fibers showed that the paralysis observed clinically arises from aberrant depolarization of the resting potential of the muscle cell (Rüdel et al., 1984); the cause of this depolarization remains unknown. Although the L-type Ca channel has long been thought to be an effector molecule that senses the voltage across the t-tubular membrane to trigger calcium influx and release, there is no evidence that the channel makes a significant contribution to the muscle resting potential. Moreover, our data give no indication that the mutant channel is open abnormally at potentials near the firing threshold. Thus, a persistent inward current through mutant L-type Ca channels is unlikely to be the cause for the anomalous depolarization. However, it is possible that calcium either conducted through the L-type Ca channel or released from the sarcoplasmic reticulum acts as a messenger that feeds back on the membrane potential, perhaps by opening local calcium-activated potassium channels (Jacquemond and Allard, 1998). In mature muscle, large-conductance K(Ca) channels are found in the t-tubules (Latorre et al., 1989), where the L-type Ca channel is known to reside, and physiologically important colocalization of K(Ca) and Ca channels has been demonstrated in bullfrog saccular hair cells (Roberts et al., 1990) and snail neurons (Gola and Crest, 1993). In such a feedback scheme, the L-type Ca channel is indirectly related to the membrane potential, and its normal function of supplying calcium in a timely manner in response to depolarization serves a protective purpose. A reduction in the L-type Ca current density or a delay in the activation of the channel during activity might disrupt the ability of K(Ca) channels to rescue the muscle cell after a series of action potentials, leading to depolarization and a failure of conduction.
As suggested by Jurkat-Rott et al. (1998), it is also possible that the R528H mutation has no significant effect on E–C coupling or on the electrical behavior of adult muscle cells but rather alters expression of other muscle proteins during development, giving rise to an abnormal electrical environment in the mature muscle fiber that allows potassium-sensitive paralysis. Jurkat-Rott et al. (1998) found modest parallel shifts in the voltage dependence of the L-type, T-type, and third-type calcium currents in R528H myotubes compared with normal myotubes, suggesting that some common regulatory mechanism was different in the mutant cells. Other preliminary work has revealed a reduction in K current density near the resting potential in biopsied mature HypoPP fibers compared with normal fibers (Ruff, 1998) that could render these fibers more susceptible to abnormal depolarization. A resolution of the basis for this diversity of altered behavior in HypoPP fibers and of the pathophysiological basis of fiber weakness will likely require the development of an in vivo model, such as a targeted “knock-in” mutation in mice.
This work was supported by the National Institutes of Health Grant R01-AR42703, by the Esther A. and Joseph Klingenstein Fund, Inc. (S.C.C.), and by a Quan Predoctoral fellowship (J.A.M.). We thank Adriana Pechanova for performing the RT-PCR work. We also thank Dr. Bruce Bean and Dr. Donnella Green for helpful discussions and Vasanth Vedantham for helpful comments on this manuscript.
Correspondence should be addressed to Dr. Stephen Cannon, EDR 413A, Massachusetts General Hospital, Boston, MA 02214.