INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Calcium ion influx through the voltage-dependent calcium channel plays a major role in the excitation-contraction coupling of cardiac myocytes. To date, molecular cloning has identified the primary structures for seven to nine distinct calcium channel ₁ subunits (_1A, _1B, _1C, _1D, _1E, _1F, _1G, _1H, _1S) (Hofmann et al., 1994; Perez-Reyes et al., 1998; Tsien, 1998). The ₁ subunit of the cardiac L-type calcium channel has been cloned (Perez-Reyes, 1990; Wei et al., 1991). Reminiscent of the native channel, the _1C subunit, whether expressed in Xenopus oocytes or in mammalian HEK-293 cells, typically inactivates faster in the presence of Ca²⁺ than in the presence of Ba²⁺ ions (Neely et al., 1994; Parent et al., 1995; DeLeon et al., 1995). In contrast, _1E channels showed similar rates of inactivation in the presence of Ba²⁺ and Ca²⁺ (Parent et al., 1997). Several mechanisms have been proposed to explain Ca²⁺-dependent inactivation. Single-channel data strongly suggest that calcium-dependent inactivation is triggered by Ca²⁺ ions binding directly to the intracellular face of the channel (Imredy and Yue, 1994). Inactivation is rapid and calcium sensitive in patches with a single L-type calcium channel (Imredy and Yue, 1992) and in planar lipid bilayers (Haack and Rosenberg, 1994). Calcium influx through one L-type calcium channel can therefore selectively facilitate the inactivation of another adjacent channel without a generalized elevation of bulk intracellular calcium concentration (Imredy and Yue, 1992; Risso and DeFelice, 1993). For instance, in cardiac cells, local internal Ca²⁺ is elevated from baseline values of 0.05 µM to peak values of 7 µM after opening of voltage-dependent calcium channels (Risso and DeFelice, 1993; Cannell et al., 1995; Lòpez-Lòpez et al., 1995). A sharp decrease in the free intracellular concentration of Ca²⁺ could also slow and/or reduce calcium-dependent inactivation, prompting the suggestion that Ca²⁺ acts from the internal face of the channel (Kramer et al., 1991). Furthermore, intracellular perfusion with trypsin was shown to decrease Ca²⁺-dependent inactivation in guinea pig myocytes (You et al., 1995), suggesting the presence of a intracellular "Ca²⁺-dependent inactivation particle" similar to the "ball-and-chain" inactivation particle of Shaker K⁺ channels (Hoshi et al., 1990). Thus Ca²⁺ binding could occur at a site with a high affinity (Johnson and Byerly, 1993) or a site with moderate affinity close enough to the calcium channel pore, such as the channel inner mouth. Direct binding of Ca²⁺ to the channel protein would thus be consistent with most experimental results and has emerged as the most likely chemical initiation event for inactivation.

It has therefore been suggested that calcium binding sites in cytoplasmic domains are critical in calcium-dependent inactivation, with a possible role for the C-terminus of _1C. Ca²⁺ ions can interact with neutral oxygen donors such as carbonyl and alcohol groups with a coordination number varying between 6 and 8 (DaSilva and Williams, 1993). The major ligands in calcium-binding proteins are thus oxygen-containing residues with either carboxylate groups (aspartate, glutamate) or carboxyl groups (asparagine, glutamine, serine, threonine), as in calmodulin and parvalbumin (McPhalen et al., 1991; Nakayama et al., 1992). The calcium binding motif recurrent in these proteins is called an EF-hand binding motif. Functional EF-hands occur in pairs, with the two hands (-helices) related by an approximate twofold axis of symmetry around a calcium-loop binding site. EF-hand proteins bind Ca²⁺ with dissociation constants in the micromolar (µM) range (Kretsinger, 1976; Kohama, 1979), which is compatible with the reported observation that intracellular Ca²⁺ causes inactivation with a K_d  4 µM (Haack and Rosenberg, 1994). The C-terminus of all calcium channel ₁ subunits has retained the sequence of a EF-hand motif in a section located close to IVS6 (Babitch, 1990). It shows high sequence similarity to Ca²⁺ binding sites in its central region, where Ca²⁺ binding is contributed by a hydrophilic residue (aspartate, asparagine, glutamate, glutamine, serine, threonine). The proposition that a EF-hand binding motif in the C-terminus may play a active role in calcium-dependent inactivation is thus quite attractive, as it meets the basic criteria of intracellular calcium binding site requirements. Indeed, a role for the EF-hand binding motif in calcium-dependent inactivation was first suggested by chimeric studies conducted by DeLeon and colleagues (1995). Amino acid sequences in the C-terminus are shown in Fig. 1 for _{1C, 1D}, _1A, _1B, and _1E calcium channel subunits. The C-terminus of _1G is not shown, as it does not bear any homology to the other ₁ subunits (Perez-Reyes et al., 1998). The inferred calcium ligands are assigned to the vertices of an octahedron as X, Y, Z, Y, X, Z and are provided by oxygen-containing side chains. Residues that comply with the consensus sequence are underlined. As seen, residues K1539 and K1543 in _1C fail to comply with the classical model of an EF-hand motif. However, the ligand found at the Y position, which corresponds to amino acid E1537 in _1C, is replaced by a hydrophobic alanine residue in _1A, _1B, and _1E channels that all lack faster calcium-dependent inactivation (DeWaard and Campbell, 1995; Parent et al., 1997). Herein the role of the glutamate residue E1537 was tested with a series of point mutations to document the nature of Ca²⁺ binding, if any, at this site. Inactivation data reported herein most unexpectedly suggest that the residues within the EF-hand binding motif may rather contribute to voltage-dependent inactivation in _1C channels.

View larger version (30K): [in this window] [in a new window]   FIGURE 1   Predicted secondary structure for the cardiac 1C channel with the four homologous repeats I, II, III, IV. The N- and C-termini are predicted to be facing the cytoplasm. The E1537 residue is located within 100 nucleotides (or 30 AA) of the IVS6 transmembrane segment in a short section of the C-terminus that bears high homology to a Ca2+-binding motif. The amino acid alignment for 1C, 1D, 1A, 1B, and 1E calcium channels is shown enlarged for this section of the C-terminus. The EF-hand binding motif is absent in the newly cloned 1G channel. There is a high degree of homology between the calcium channel 1 subunits, with a few notable exceptions, such as E1537.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Polymerase chain reaction mutagenesis for E1537 mutants

Standard methods of plasmid DNA preparation and DNA sequencing were used (Sambrook et al., 1989). A wild-type, full-length cardiac _1C subunit cDNA (Genebank X15539) was cloned from rabbit (Wei et al., 1991). Point mutations were prepared by overlap extension at the junctions of the relevant domains, using sequential polymerase chain reaction (PCR) (Hoffman-La Roche) (Ho et al., 1989) as described earlier (Parent et al., 1995; Parent and Gopalakrishnan, 1995). Briefly, 1.1-kbp DNA fragments were amplified between nucleotides 3752 and 5236 from the full-length _1C template, using Gold Ampli-Taq polymerase (Perkin-Elmer Cetus) in reactions performed with a DNA Thermal Cycler 2400 (Perkin-Elmer Cetus) or a DNA engine PTC-200 (MJ Research). The final PCR product containing the overlapping region was ligated back into the host _1C subunit between the EcoRV (4350) and BstEII (4646) sites. Restriction enzymes were obtained from New England Biolabs (Beverly, MA). Constructs were verified by restriction mapping, and recombinant clones were screened by double-stranded sequence analysis of the entire ligated cassette. DNA constructs were linearized at the 3' end by HindIII digestion, and run-off transcripts were prepared using methylated cap analog m⁷G(5')ppp(5')G and T7 RNA polymerase included in the mMessage mMachine transcription kit (Ambion, Austin, TX). The final cRNA products were resuspended in 0.1 M KCl at a concentration of 2 µg/µl and stored at 80°C. The integrity of the final product and the absence of degraded RNA were determined by denaturing agarose gel stained with ethidium bromide.

Functional expression of wild-type and E1537 mutant channels

Oocytes were obtained from female Xenopus laevis clawed frog (Nasco, Fort Atkinson, WI) as described previously (Parent et al., 1995, 1997; Parent and Gopalakrishnan, 1995). Individual oocytes free of follicular cells were obtained after 30-40 min of incubation in a calcium-free saline solution (in mM: 82.5 NaCl; 2.5 KCl; 1 MgCl₂; 5 HEPES; pH 7.6) containing 2 mg/ml collagenase (Sigma-Aldrich, St. Louis, MO). Forty-seven nanoliters of cRNA coding for the wild-type or mutated _1C subunit was injected 16 h later at a concentration of 100 ng/µl in 0.1 M KCl (4.7 ng total cRNA) into stage V and VI oocytes. cRNA coding for rat brain _2b (Williams et al., 1992) and rat brain/cardiac _2a (Perez-Reyes et al., 1992) was typically coinjected with _1C at a 1:1:1 molar ratio. Oocytes were incubated at 18°C in a Barth solution (in mM): 88 NaCl; 3 KCl; 0.82 MgCl₂; 0.41 CaCl₂; 0.33 Ca(NO₃)₂; 5 HEPES; pH 7.6, supplemented with 5% horse serum, 2.5 mM Na pyruvate, 100 units/ml penicillin; 0.1 mg/ml streptomycin.

Electrophysiological recordings and result analysis

Wild-type and mutant _1C channels were screened for macroscopic barium current 4-7 days after RNA injection, with the Warner OC-725C amplifier oocyte clamp as desribed earlier (Parent et al., 1995, 1997). Voltage and current electrodes, filled with 3 M KCl, were broken slightly under the microscope to decrease the electrode resistance to 1.5 M tip resistance. Whole-cell currents were measured at room temperature in a 10 BaMeS solution (in mM: 10 Ba(OH)₂; 110 NaOH; 1 KOH; 0.5 niflumic acid; 10 HEPES titrated to pH 7.2 with methanesulfonic acid) or a 10 CaMeS solution in which Ca(OH)₂ replaced Ba(OH)₂ equimolarly. Niflumic acid was added to block endogenous Ca²⁺-dependent Cl currents (White and Aylwin, 1990). To further palliate contamination by these Cl currents, a volume of 100 nl of a 10 mM 1,2-bis(2-aminophenoxy)ethane-N,N,N',N'-tetraacetic acid titrated with HEPES and KOH to pH 7.4 (BAPTA-HEPES) stock solution was injected directly into oocytes 1-2 h before experiments for a final concentration of 1 mM. Alternatively, oocytes were also preincubated in a saline solution containing 100 µM 1,2-bis(2-aminophenoxy)ethane-N,N,N',N'-tetraacetic acid tetra lacetoxymethyl) ester (BAPTA-AM) (Sigma-Aldrich, St. Louis, MO). Whole-cell current traces were filtered at 1 kHz, using the built-in filter of the oocyte clamp, and were acquired at 5 kHz through a Digidata 1200 analogue-to-digital board (Axon Instruments, CA). The pClamp programs Clampex and Clampfit, version 6.04, were used to generate voltage protocols and to digitally acquire and analyze data. Capacitive transients and leak currents were digitally subtracted. Isochronic inactivation curves were generated by measuring tail currents obtained after 5-s voltage pulses applied from 100 to +30 mV. Fractional currents were fitted to Eq. 1 (Parent et al., 1995, 1997):

(1)

where i is the peak current obtained after a 5-s pulse to voltage V_m; i_max is the peak current measured after a 5-s voltage pulse to 100 mV; and i/i_max is their ratio when normalized to 1; Y₀ is the fraction of noninactivating current; E_0.5 is the midpotential of inactivation; z is the slope factor; and R, T, F have their usual meanings. Pooled data points i/i_max (mean ± SEM) were fitted to this modified Boltzmann equation using user-defined functions and the fitting algorithms provided in Origin 5.0 (Microcal Software) analysis software. Fit parameters were estimated with their corresponding errors. For the activation and inactivation time constants, leak subtracted current traces recorded at 10 kHz were fitted with exponential functions of the first order for Ba²⁺ currents (Eq. 2) or the second order for Ca²⁺ currents (Eq. 3), using the Chebyshev algorithm in Clampfit 6.04. Whole-cell Ba²⁺ current time constants were systematically measured between time t = 0 s and t = 2 s with the following equation:

(2)

I(t) is the current at time t; _act and _inact are the time constants of the activation and inactivation processes, and I_act and I_inact are the amplitudes of these processes. For simplicity's sake, the single inactivation time constant measured in Ba²⁺ is often referred to as ^inact_Ba in the text. Whole-cell Ca²⁺ current time constants were measured between time t = 0 and t = 2 s with the following equation:

(3)

I(t) is the current at time t; _act, _inact¹, and _inact² are the time constants of the activation, the fast inactivation, and the slow inactivation processes, respectively; I_act, I_inact¹, and I_inact² are the amplitudes of the same processes. For simplicity's sake, the two inactivation time constants measured in Ca²⁺ are referred to, respectively, as _fast^inact and _slow^inact in the text. Experiments were performed at room temperature. Figures were drawn using Designer 4.1 (Micrografx Software).

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Fig. 1 compares the primary sequences of the C-termini of _1C, _1D, _1A, _1B, and _1E subunits. The C-terminus of the newly cloned T-type calcium channel _1G (Perez-Reyes et al., 1998) does not exhibit a putative EF-hand binding motif. The rationale for targeting the putative EF-hand motif in calcium-dependent inactivation pertains to its potential ability to form an intracellular Ca²⁺ binding site. In particular, we were interested in the E  A mutation at position 1537. Alanine is a small, neutral, relatively hydrophobic residue, whereas glutamate is a hydrophilic (acidic) residue with a pK_a value of 4.3 that would thus be negatively charged under physiological conditions. Acidic residues are also known to be effective chelators of metal ions such as Ca²⁺ and Cd²⁺ (Creighton, 1993). Thus the respective side chains of alanine and glutamate are expected to display large differences in their Ca²⁺ affinity, regardless of their role in the putative EF-hand. If position 1537 were critical in calcium-dependent inactivation, substitution at this position by a less polar residue could significantly slow Ca²⁺-dependent inactivation while leaving Ba²⁺-dependent inactivation unaltered.

E1537 mutant channels displayed slower inactivation kinetics in Ba²⁺ and Ca²⁺ solutions

To investigate the possible role of the negatively charged E1537 residue in calcium-dependent inactivation, mutant channels E1537D, E1537Q, E1537S, E1537G, and E1537A were expressed in Xenopus oocytes. Fig. 2 shows typical whole-cell current recordings, for the wild-type and mutant E1537Q, E1537S, E1537A cardiac _1C calcium channels, obtained in the presence of 10 mM Ba²⁺ (top) or 10 mM Ca²⁺ (bottom) solutions, using 450 ms voltage steps. In these and the following experiments, the _1C subunits (wild-type and mutants) were systematically coinjected with the _2b and the _2a auxiliary subunits. Except for the E1537D mutant, for which we never got expression, all E1537 mutant yielded measurable Ba²⁺ and Ca²⁺ currents with bona fide Ca²⁺ channel characteristics (see also later in Fig. 4). Peak Ba²⁺ currents ranged from 400 to 900 nA (see Table 1). Although such current amplitudes are well above background, E1537 mutant peak current amplitude was on average twofold smaller than the wild-type _1C peak current recorded under the same experimental conditions. In addition, macroscopic Ba²⁺ currents were at least twice as large as Ca²⁺ currents in all mutants. This ratio likely represents a minimum value, as Ca²⁺ currents were generally recorded a few days later than Ba²⁺ currents to improve the signal-to-noise ratio. Macroscopic Ba²⁺ currents activated within 3 ms. As expected for L-type cardiac calcium channels, the macroscopic wild-type _1C currents typically inactivated faster in the presence of 10 mM Ca²⁺ than in the presence of Ba²⁺ (left). In fact, all channels displayed a faster Ca²⁺-dependent than Ba²⁺-dependent inactivation, as Ba²⁺ currents invariably inactivated more slowly than the corresponding Ca²⁺ traces, from the wild-type to the E1537A channel. This observation also applied to E1537G whole-cell currents that are not shown in this figure. Six independent series of channel expression in oocytes confirmed these observations. The slower Ba²⁺- and Ca²⁺-dependent inactivation in E1537A channels was also observed in the presence of the ₃ subunit (Castellano et al., 1993) (results not shown), indicating that the reduced inactivation was independent of the nature of the  subunit. Based on the observation that inactivation remained faster in the presence of Ca²⁺, one may reasonably conclude that calcium-dependent inactivation was not abolished by mutations at position E1537, a conclusion also reached by Zhou and colleagues (1997). A careful examination of the whole-cell recordings nonetheless indicates that both Ba²⁺ and Ca²⁺ current traces inactivated faster for the wild-type than for the mutant channels in the following order: wt > E1537Q > E1537S  E1537A, thus suggesting that overall macroscopic inactivation was reduced for E1537 mutant channels. Indeed, all mutant channels displayed significantly slower kinetics than the wild-type _1C calcium channel. To investigate and quantify the mutant channel kinetics, whole-cell Ba²⁺ and Ca²⁺ currents were fitted to a sum of exponential functions (see Table 2 and Fig. 3). Inactivation time constants at V_m = +10 mV were estimated at time t = 2 s for wild-type _1C/_2b/_2a; E1537Q/_2b/_2a, E1537G/_2b/_2a, E1537S/_2b/_2a, and E1537A/_2b/_2a channels in the presence of 10 mM Ba²⁺ (right panel) or 10 mM Ca²⁺ (left panel). As shown in the right panel of Fig. 3, the inactivation time course of whole-cell Ba²⁺ traces could be fit well by a single exponential function. Ba²⁺ inactivation time constants increased from 672 ± 43 ms (n = 7) for the wild-type _1C channel to 988 ± 40 ms (n = 5) for E1537S and to 2501 ± 740 ms (n = 5) for E1537A. As seen, all E1537 mutant channels displayed significantly slower Ba²⁺ inactivation than the wild-type channel. Indeed, _Ba^inact for the wild-type channel was different from the E1537Q channel time constant at the level p < 0.1 (*), whereas they were significantly different at the level p < 0.05 (**) for E1537G, E1537S, and E1537A channels. The inactivation time course of whole-cell Ca²⁺ traces required, in contrast, at least a sum of two exponential functions _fast^inact and _slow^inact. The faster inactivation time constant _fast^inact was not significantly affected by mutations at E1537 as _fast^inact ranged from 49 ± 2 ms (n = 15) for the wild-type channel to 56 ± 3 ms (n = 12) for E1537A. On the other hand, the slower Ca²⁺ inactivation time constant _slow^inact increased from 496 ± 33 ms (n = 15) for the wild-type _1C channel to 712 ± 41 ms (n = 6) for E1537S and to 2378 ± 660 ms (n = 12) for E1537A. The increase in _slow^inact in the E1537A mutant was accompanied by a parallel increase in its relative importance as the relative amplitude of the slower Ca²⁺ inactivation time constant increased from 41 ± 4% for the wild-type channel to 65 ± 7% for E1537A. The fit values, including the relative amplitudes of _fast^inact and _slow^inact, are given in detail in Table 2. From our whole-cell recordings, it thus appears that the increased fraction of noninactivating current in E1537 mutant channels was caused by a combination of factors such as the relative decrease in the amplitude of the fast and calcium-dependent inactivation time constant, and the increase in the slower and Ba²⁺-dependent inactivation time constant. The mutant slow inactivation time constants measured in Ca²⁺ were all significantly (p < 0.01) larger than the wild-type _slow^inact, whereas _fast^inact was not significantly different at the level p < 0.1. Furthermore, _slow^inact in Ca²⁺ and _Ba^inact in Ba²⁺ appeared remarkably similar for all calcium channel combinations and were found to increase in parallel in E1537Q, E1537S, E1537G, and E1537A channels. These observations suggest that mutations at position 1537 specifically affected the slow inactivation time constant in Ba²⁺ and Ca²⁺ solutions. As the inactivation time constant _slow^inact appeared to be independent of the charge carrier, there is a strong possibility that mutations at position 1537 modified the rate of the voltage-dependent transition to the inactivated state in _1C channels. Assuming that voltage-dependent inactivation could proceed more rapidly from the open than from the closed state in _1C/_2b/_2a and _1C/_2b/₃ channels, in sharp contrast to neuronal channels (Patil et al., 1998), _slow^inact in our experiments is likely to reflect changes in the rate of transition from the open to the inactivated state. It can always be argued that our whole-cell data could be explained as well by a faster open-to-closed transition or by a slower closed-to-open transition rate in E1537 mutant channels. However, any decrease in the channel open probability would also influence calcium-dependent inactivation unless Ca²⁺-dependent and voltage-dependent inactivation proceeds from two distinct open states, as could happen in a modal kinetic model. Such a model has already been proposed to explain Ca²⁺-dependent inactivation in L-type calcium channels (Imredy and Yue, 1994). At this stage, it can in any event be safely concluded that the apparent or macroscopic rate of Ba²⁺ inactivation has been modified in E1537 mutant channels.

View larger version (49K): [in this window] [in a new window]   FIGURE 2   The whole-cell current 1C kinetics are faster in the presence of Ca2+ as the charge carrier. Wild-type 1C and mutant E1537Q, E1537S, E1537A were expressed in Xenopus oocytes with auxiliary 2b and 2a subunits. The current traces were recorded with the two-electrode voltage-clamp technique, 2 h after injection of 10 mM BAPTA, in the presence of 10 mM Ba2+ (top traces) and 10 mM Ca2+ (bottom traces). The holding potential was 80 mV throughout. Voltage pulses (450 ms) were applied from 40 to +50 mV in 10-mV steps at 0.2 Hz. In all channels, wild-type and mutants, calcium traces were always faster than barium traces. Leak currents were digitally subtracted. Capacitive transients were erased for the first millisecond after the voltage step. The current scale varies between 0.05 and 0.5 µA. Time scales are 100 ms throughout.

View this table: [in this window] [in a new window]   TABLE 1   Whole-cell peak currents for wild-type 1C/2b/2a, E1537A/2b/2a, E1537G/2b/2a, E1537Q/2b/2a, and E1537S/2b/2a channels in 10 mM Ba2+ and 10 mM Ca2+ solutions as the mean ± SEM of n independent experiments

View this table: [in this window] [in a new window]   TABLE 2   Inactivation time constants for wild-type 1C/2b/2a, E1537Q/2b/2a, E1537G/2b/2a, E1537S/2b/2a, and E1537A/2b/2a channels recorded in 10 mM Ba2+ or 10 mM Ca2+ solutions

View larger version (11K): [in this window] [in a new window]   FIGURE 3   Mutations at E1537 slowed 1C inactivation in Ba2+ and Ca2+. In the presence of Ba2+, two exponential functions can satisfactorily account for the current time course, whereas the current time course in the presence of Ca2+ could be best described by a sum of three exponential functions (act, fastinact, slowinact) for all channels herein investigated. Only the inactivation time constants are reported here. Inactivation time constants for Vm = +10 mV were estimated at time t = 2 s for wild-type 1C/2b/2a; E1537Q/2b/2a, E1537G/2b/2a, E1537S/2b/2a, and E1537A/2b/2a channels in the presence of 10 mM Ba2+ (right) or 10 mM Ca2+ (left). (Right) Inactivation time course of whole-cell Ba2+ traces could be well fitted by a single exponential time constant. Ba2+ inactivation time constants increased from 672 ± 43 ms (n = 7) for the wild-type 1C channel to 988 ± 40 ms (n = 5) for E1537S and to 2501 ± 740 ms (n = 5) for E1537A. All mutant channel inactivated significantly more slowly in Ba2+ than did the wild-type channel. The fit values are given in detail in Table 2. (Left) Inactivation time course of whole-cell Ca2+ traces could be best described by a sum of two exponential functions. The faster inactivation time constant fastinact was not significantly affected by mutations at E1537 as fastinact ranged from 49 ± 2 ms (n = 15) for the wild-type channel to 56 ± 3 ms (n = 12) for E1537A. On the other hand, the slower Ca2+ inactivation time constant slowinact increased from 496 ± 33 ms (n = 15) for the wild-type 1C channel to 712 ± 41 ms (n = 6) for E1537S and to 2378 ± 660 ms (n = 12) for E1537A. Moreover, the relative amplitude of the slower Ca2+ inactivation time constant increased from 41 ± 4% for the wild-type channel to 65 ± 7% for E1537A. slowinact was different from the wild-type channel at the significance level p < 0.01 (***) for all E1537 mutant channels, whereas fastinact was not significantly different at the level p < 0.1. Bainact for E1537Q was different at the level p < 0.1 (*), and Bainact for E1537G, E1537S, and E1537A was different at the level p < 0.05 (**). The fit values, including the relative amplitude of fastinact and slowinact, are given in detail in Table 2.

Inactivation kinetics and whole-cell current density in E1537 mutant channels

Our first series of experiments clearly demonstrated that Ba²⁺ inactivation was significantly reduced in E1537 mutants as compared to inactivation in the wild-type channel. Whether these slower inactivation kinetics were primarily caused by an intrinsic change in the channel inactivation properties or rather were linked to a lower current density was further investigated in the following series of experiments. It is well established that calcium channel inactivation tends to process faster for larger currents. As whole-cell currents for the E1537 mutants were generally smaller than the wild-type currents recorded under the same conditions, the role of current density (or under our particular recording conditions, current amplitude) plays in conferring slower inactivation kinetics to E1537 mutants had to be carefully reviewed. The influence of peak current amplitude on E1537 channel inactivation was thus examined in Fig. 4. Whole-cell Ba²⁺ traces were recorded at the peak voltage for the wild-type, E1537Q, and E1537A channels, normalized and superimposed to magnify the differences in inactivation kinetics. As shown in Fig. 4 A, E1537A Ba²⁺ currents did not appreciably decay over the 450-ms voltage pulse to +10 mV, in contrast to wild-type Ba²⁺ currents. Whole-cell traces shown in the inset suggests that the rate of current activation may also be reduced in E1537A channels with, on average, _act = 3.8 ± 0.7 ms (n = 12) as compared to _act = 2.1 ± 0.8 ms (n = 9) for the wild-type channel. The permeation parameters were otherwise quite similar, as seen in the normalized I-V curve (Fig. 4 B) between the wild-type and the E1537A channel, because both activated around 25 mV and peaked at 0 mV. The E1537Q peaked at +10 mV in the presence of Ba²⁺. Macroscopic I-V curves were also typically shifted to the right in the presence of Ca²⁺ for all channels (Fig. 4 E). For instance, the peak current shifted from 6 ± 2 mV (n = 11) in the presence of Ba²⁺ to 14 ± 1 mV (n = 14) in the presence of Ca²⁺ for the wild-type channel. This +8 mV shift was comparable to the shifts experimentally recorded for E1537Q (10 ± 0.5 mV to 19 ± 2 mV, n = 11) and E1537A (0 ± 0.5 mV to 6 ± 2 mV, n = 14) channels. The macroscopic current expression was generally higher for the wild-type channel as whole-cell Ba²⁺ currents averaged 1.3 ± 0.2 µA (n = 11), whereas E1537Q and E1537A generated smaller Ba²⁺ currents (see also Table 1). As seen in Fig. 4, A and D, Ba²⁺ and Ca²⁺ inactivation were not necessarily faster for larger currents. Two lines of evidence indeed suggest that whole-cell current amplitude was not the sole determinant in E1537 mutant slower inactivation kinetics. The first argument is circumstantial and pertains to the significant kinetic differences between E1537Q and E1537A, despite similar expression levels. Indeed, for all V_m  10 mV, whole-cell Ba²⁺ and Ca²⁺ currents recorded for E1537Q were not significantly different in size than E1537A currents (Fig. 4, C and F), despite its slightly faster rate of inactivation. To circumvent the current density problem, we further designed a series of experiments whereby the relative ratio of the wild-type _1C to the auxiliary subunits was progressively decreased such that wild-type _1C currents would match the expression level of E1537 mutants. Alternatively, in a separate series of experiments, Ca²⁺ currents for the E1537 mutants were measured 4 days later than wild-type currents, hence increasing the probability of finding a mutant channel with Ca²⁺ currents comparable in size to those of the wild-type channel. The two approaches yielded similar results. As shown in Fig. 4 D, the superimposed yet not normalized whole-cell Ca²⁺ current traces measured at V_m = +10 mV for the wild-type, the E1537Q, and the E1537A channels yielded similar current amplitudes when recorded under the same experimental conditions. Despite generating a identical peak current of 0.66 µA, wild-type Ca²⁺ traces nonetheless inactivated significantly faster than E1537A traces, as only 15% of the wild-type currents remained at the end of the 450-ms pulse compared to 70% of the E1537A currents. Yet again, E1537Q currents displayed an intermediate rate of inactivation, with 38% noninactivating currents. Results obtained in Fig. 4 D thus minimized the possible role of current-dependent inactivation in the slower inactivation kinetics of E1537 mutants. From Fig. 4 it also appears that the current time course measured in Ca²⁺ solutions was qualitatively similar for all channels. As the curve fitting analysis demonstrated, the major kinetic effect brought about by Ca²⁺ is seen in an additional inactivation time constant that is relatively fast, _fast^inact, and was absent in Ba²⁺ traces. This faster inactivation time constant _fast^inact remained present in all mutant channels, furthering the notion that calcium-dependent inactivation was not specifically modified in E1537 mutant channels.

View larger version (29K): [in this window] [in a new window]   FIGURE 4   Whole-cell currents for wild-type 1C/2b/2a, E1537A/2b/2a, and E1537Q/2b/2a channels were recorded in the presence of 10 mM Ba2+ (upper panels) and 10 mM Ca2+ (lower panels) by the pulse protocol previously described. (A) Whole-cell current traces recorded in the presence of 10 mM Ba2+ at +10 mV were normalized to 1.0 and superimposed. Ba2+ inactivation was faster for wild-type > E1537Q > E1537A channels. As shown in the insert, the E1537A activation also appeared to be noticeably slower. (B) The corresponding normalized I-V curves obtained in the presence of 10 mM Ba2+ are shown. The wild-type and E1537A current-voltage curves peaked at 0 mV, whereas the E1573Q channel peaked at +10 mV. On average, I-V curves peaked at the following voltages: 6 ± 2 mV (n = 11) for the wild-type 1C/2b/2a, 0 ± 0.4 mV (n = 15) for E1537A/2b/2a, and 10 ± 1 mV (n = 14) for E1537Q/2b/2a channels. (C) Ba2+ whole-cell current amplitude was higher on average for the wild-type channel with a peak current of 1.2 ± 0.1 µA (n = 11) as compared to peak currents of 0.81 ± 0.06 µA (n = 15) for E1537A/2b/2a and 0.63 ± 0.04 µA (n = 14) for E1537Q/2b/2a channels. For clarity, only the positive error bars are shown for the wild-type channel. (D) Whole-cell currents obtained in 10 mM Ca2+ are shown superimposed at Vm = 10 mV. Current traces were not normalized because in that particular case, wild-type 1C/2b/2a, E1537A/2b/2a, and E1537Q/2b/2a channels yielded whole-cell Ca2+ currents in the same range. As seen, wild-type 1C Ca2+ currents were typically faster than E1537A Ca2+ currents, independently of the whole-cell current amplitude. E1537Q channels yielded current traces with an intermediate inactivation time course. (E) Corresponding I-V curves. As expected, whole-cell Ca2+ I-V curves are shifted to the right. Wild-type and E1537Q Ca2+ currents peaked, respectively, at Vm = 14 ± 1 mV (n = 14) and 19 ± 3 mV (n = 11), whereas E1537A Ca2+ currents peaked at Vm = 5 ± 1 mV (n = 14). (F) Whole-cell Ca2+ current amplitude was higher on average for the wild-type channel with a peak current of 0.61 ± 0.08 µA (n = 14) as compared to peak currents of 0.43 ± 0.05 µA (n = 14) for E1537A/2b/2a and 0.27 ± 0.04 µA (n = 11) for E1537Q/2b/2a channels.

Voltage dependence of the E1537A inactivation time constants

In voltage-dependent ion channels, voltage controls kinetic transitions, and more specifically, positive voltage encourages transitions from the closed to the open state, and ultimately to the inactivated state. Thus in a traditional model of voltage-dependent inactivation where inactivation is strongly coupled to the open state, inactivation is expected to speed up with membrane depolarization (Armstrong and Bezanilla, 1977; Bean, 1981; Imredy and Yue, 1994). Mutations at E1537 can lead to a slower inactivation by one of the following mechanisms, either by slowing the macroscopic rate of transition to the inactivated state or by making the channel unresponsive to membrane depolarization. We thus investigated the influence of voltage on the inactivation time constants for the E1537A mutant channel. Inactivation time constants for the wild-type _1C/_2b/_2a and E1537A/_2b/_2a channels were estimated at time t = 2 s and reported as a function of the applied membrane potential between 10 and +20 mV (Fig. 5). As seen in the left panel, _Ba^inact for the wild-type channel decreased from 918 ± 43 ms to 627 ± 46 ms (n = 7) between 10 and +20 mV; hence membrane depolarization appeared to speed up inactivation in the presence of Ba²⁺. This observation has previously been reported for native L-type calcium currents in isolated myocytes from adult rat (Imredy and Yue, 1994) when it was shown that the decay of Ba²⁺ currents accelerates monotonically with depolarization. In contrast, the E1537A inactivation time constants increased at least threefold over the same voltage range, with _Ba^inact increasing from 1221 ± 246 ms at 10 mV to 3609 ± 1329 ms (n = 5) at +20 mV. Note that the inactivation time constants estimated at 10 mV were not significantly different for the wild-type and the E1537A channel. Similar results were obtained for _slow^inact in Ca²⁺ solutions. Moreover, the slow inactivation time constants between Ba²⁺ and Ca²⁺ turn out to be remarkably similar between 10 and +20 mV for wild-type and mutant channels. The apparent similarity between _Ba^inact and _slow^inact in Ca²⁺ argues that the mutations are primarily affecting an inactivation path present in both Ba²⁺ and Ca²⁺ solutions. As seen previously in Fig. 3, _fast^inact was not significantly different between the wild-type and the E1537A channel, with _fast^inact ranging from 41 ± 5 ms to 73 ± 5 ms (n = 7) for the wild-type channel and from 49 ± 3 ms to 69 ± 4 ms (n = 10) for E1537A. In contrast, _slow^inact in Ca²⁺ was consistently higher for E1537A than for wild-type _1C channels at all membrane potentials. For the wild-type channel, _slow^inact decreased from 732 ± 7 ms to 521 ± 51 ms (n = 7), whereas in E1537A channels, _slow^inact actually increased from 990 ± 99 ms to 1823 ± 115 ms (n = 10) between 10 and +20 mV. Thus, not only was the E1537A channel slower at all membrane potentials than the wild-type channel; it also appeared to become slower with membrane depolarization in both Ba²⁺ and Ca²⁺ solutions.

View larger version (29K): [in this window] [in a new window]   FIGURE 5   Inactivation time constants for the wild-type 1C/2b/2a and E1537A/2b/2a channels were estimated between time t = 0 and time t = 2 s and reported as a function of the applied membrane potential between 10 and +20 mV. (Left) For the wild-type Ba2+ traces, the inactivation time constants decreased from 918 ± 43 ms to 627 ± 46 ms (n = 7) between 10 and +20 mV; hence membrane depolarization appeared to speed up inactivation in the presence of Ba2+. In contrast, the E1537A inactivation time constants increased at least threefold over the same voltage range with Bainact = 1221 ± 246 ms at 10 mV to 3609 ± 1329 ms (n = 5) at +20 mV. Note that the inactivation time constants estimated at 10 mV are not significantly different for the wild-type and the E1537A channel. (Right) The Ca2+ inactivation time course can best be fitted with two exponential functions. Again, the fast Ca2+ inactivation time constant is not significantly different between the wild-type and the E1537A channel, with fastinact ranging from 41 ± 5 ms to 73 ± 5 ms (n = 7) for the wild-type channel and 49 ± 3 ms to 69 ± 4 ms (n = 10) for E1537A. In contrast, the slow Ca2+ inactivation time constant slowinact was consistently higher for E1537A than for wild-type 1C channels at all membrane potentials. For the wild-type channel, slowinact decreased from 732 ± 7 ms to 521 ± 51 ms (n = 7), whereas slow actually increased from 990 ± 99 ms to 1823 ± 115 ms (n = 10) between 10 and +20 mV. Thus the E1537A channel appeared to become slower in response to membrane depolarization in both Ba2+ and Ca2+ solutions.

Isochronic inactivation measurements in Ba²⁺ and Ca²⁺ for E1537 mutant channels

To provide a more quantitative picture of barium- and calcium-dependent inactivation in E1537 mutant channels, we performed a series of isochronic inactivation experiments at time t = 5 s. In such experiments, the relative amplitude of the tail currents, measured at the test pulse of +10 mV, is proportional to the number of channels present in the open state at that given time or inversely proportional to the number of channels present in the inactivated state. From the relationship between the current amplitude and the conditioning voltage, one can extract "steady-state" information such as the voltage range where channels are experiencing inactivation. Isochronic inactivation data thus provide a snapshot picture of the channel inactivation properties unhampered by time-dependent factors. To achieve inactivation measurements in the absence of kinetic variations, conditioning pulses should be, in theory, many times longer than the slower inactivation time constant. For _1C calcium channels, these considerations, however, must take into account the slow but irreversible time-dependent rundown associated with whole-cell experiments. Hence, our Ba²⁺ isochronic inactivation experiments with 5-s conditioning pulses are not true "steady-state" experiments, but may provide valuable insight into the inactivation process. In the presence of Ba²⁺, the voltage dependence of 5-s isochronic inactivation was investigated for the wild-type _1C/_2b/_2a, E1537Q/_2b/_2a, E1537S/_2b/_2a, E1537G/_2b/_2a, and E1537A/_2b/_2a channels. Fig. 6 shows the whole-cell current traces recorded for wild-type _1C/_2b/_2a and E1537A/_2b/_2a channels by the tripulse protocol shown, in the presence of 10 mM Ba²⁺ (upper panel) and 10 mM Ca²⁺ (lower panel). The fraction of noninactivated whole-cell current remaining at the end of the 5-s pulse was measured at the test pulse of +10 mV (peak voltage), which was then plotted against the prepulse voltage. Pooled fractional currents shown in the extreme right panels were fitted to a Boltzmann equation (Eq. 1). In the presence of 10 mM Ba²⁺, 67 ± 2% (n = 9) of the wild-type channels were completely inactivated by a 5-s pulse to +10 mV. In contrast, only 11 ± 4% (n = 6) of the E1537A channels were completely inactivated under the same conditions. The E1537A inactivation in Ba²⁺ was so shallow that it could not be approximated by Boltzmann functions. The other mutant channels showed intermediate steady-state inactivation properties with a fractional inactivation of 42 ± 3% (n = 11) for E1537Q, 39 ± 3% (n = 3) for E1537S, and 42 ± 2% (n = 4) for E1537S channels. The fitted Boltzmann parameters and the corresponding estimated fit errors are given in the figure legend. In contrast to the Ba²⁺ data, Ca²⁺-induced inactivation appeared to proceed almost to completion for all channel combinations. After a 5-s voltage pulse of +10 mV, whole-cell wild-type currents were inactivated at 83 ± 5% (n = 11), as compared to 75 ± 2% (n = 4) of the E1537Q currents, 80 ± 1% (n = 4) of the E1537G currents, and 81 ± 5% (n = 4) of the E1537S currents. With a inactivation level of 69 ± 3% (n = 7), E1537A yielded inactivation data points in Ca²⁺ that were almost indistinguishable from the wild-type and other E1537 mutants. Isochronic inactivation experiments performed at a shorter time, t = 2 s, proved to be qualitatively similar, with the exception that inactivation was somewhat reduced in E1537A channels with 49 ± 3% (n = 3) (results not shown). Remarkably, inactivation in the presence of Ca²⁺ ensued to a similar extent for all channels, despite the slower Ba²⁺ inactivation in E1537 mutant channels. This suggests again that E1537 mutations altered the slow voltage-dependent transition to the inactivated state without affecting the faster calcium-dependent inactivation. Hence voltage-dependent and calcium-dependent inactivation could involve a series of seemingly independent transitions (Hadley and Lederer, 1991) from kinetically similar closed and open states. Alternatively, voltage-dependent transitions could occur all together in a gating mode parallel to calcium-facilitated ones, as has previously been suggested in the Ca²⁺-induced gating shift model (Imredy and Yue, 1994).

View larger version (34K): [in this window] [in a new window]   FIGURE 6   The voltage dependence of inactivation was investigated for the wild-type 1C/2b/2a, E1537Q/2b/2a, E1537S/2b/2a, E1537G/2b/2a, and E1537A/2b/2a channels at the end of a 5-s prepulse. The inactivation protocol is shown on top of the whole-cell recordings. Holding potential was 80 mV, the test voltage was the peak voltage (usually around 0 mV in Ba2+ and +10 mV in Ca2+), and 14 prepulses were applied from 100 to +30 mV, by 10-mV steps at a frequency of 0.05 Hz. Acquisition frequency was 2 kHz. Inactivation was measured in the presence of 10 mM Ba2+ (top traces) and in the presence of 10 mM Ca2+ (bottom traces) after BAPTA injection. The fraction of the noninactivating current was recorded at the end of the 5-s pulse and reported on the steady-state inactivation curve shown at the right. Only the whole-cell current traces obtained with the wild-type and the E1537A channels are shown. (Upper right) In the presence of 10 mM Ba2+, 67 ± 2% (n = 9) of the wild-type () channels were completely inactivated at the end of a 5-s pulse to +10 mV. In contrast, only 11 ± 4% (n = 6) of the E1537A channels () were completely inactivated under the same conditions. Other mutant channels showed intermediate steady-state inactivation properties with fractional inactivation of 42 ± 3% (n = 11) for E1537Q, 39 ± 3% (n = 3) for E1537S, and 42 ± 2% (n = 4) for E1537S channels. Inactivation data were pooled from independent experiments performed on single oocytes and fitted to Boltzmann functions, using the following fit parameters and the corresponding estimated fit errors (Eq. 1): z = 2.4 ± 0.2, E0.5 = 20 ± 0.7 mV (wild-type); z = 1.4 ± 0.1, E0.5 = 18 ± 2 mV (E1537Q), z = 1.2 ± 0.2, E0.5 = 19 ± 1 mV (E1537S); z = 1.3 ± 0.1, E0.5 = 17 ± 2 mV (E1537G). The E1537A inactivation data point could not be approximated by Boltzmann functions. (Lower right) Inactivation was almost complete at the end of a 5-s prepulse in the presence of 10 mM Ca2+. At +10 mV, whole-cell wild-type currents () were inactivated at 83 ± 5% (n = 11), as compared to the inactivation level of 75 ± 2% (n = 4) for the E1537Q currents (), 80 ± 1% (n = 4) for E1537G (), 81 ± 5% (n = 4) for E1537S (). With its inactivation level of 69 ± 3% (n = 7), E1537A inactivation data points were almost indistinguishable from the wild-type and other E1537 mutants. In the presence of Ca2+, inactivation data points were typically bell shaped for the wild-type and mutant channels alike, but these points were omitted for the sake of clarity. The fit parameters were z = 2.8 ± 0.2 and E0.5 = 22 ± 1 mV (wild-type); z = 1.9 ± 0.2 and E0.5 = 18 ± 2 mV (E1537G); z = 1.9 ± 0.2 and E0.5 = 16 ± 1 mV (E1537S); z = 1.6 ± 0.4 and E0.5 = 22 ± 3 mV (E1537Q); z = 3.0 ± 0.3 and E0.5 = 17.0 ± 0.7 mV (E1537A).

In summary, E1537 mutations significantly reduced Ba²⁺-dependent inactivation. Inactivation proceeded more completely for hydrophilic residues such as glutamate, but was least complete for the hydrophobic residues such as alanine. Five-second pulses to positive membrane potentials in Ba²⁺ failed to inactivate more than 15% of the E1537A channels, in contrast to ~70% of the wild-type channels. These huge differences collapsed in the presence of Ca²⁺, where inactivation proceeded almost to completion for wild-type and mutant channels alike, with 70-85% of calcium currents inactivated after a 5-s pulse. Hence Ca²⁺ preserved its dominant role in the inactivation kinetics of the cardiac _1C wild-type and E1537 mutant channels.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

E1537 mutations slow macroscopic inactivation of Ca²⁺ channels

In cardiac L-type calcium channels, inactivation proceeds through the combined effect of voltage and calcium ions (Kass and Sanguinetti, 1984; Lee et al., 1985; Campbell et al., 1988; Imredy and Yue, 1994), although Ca²⁺-facilitated inactivation arises as the prominent inactivation mechanism under physiological conditions. We have investigated herein the role of E1537 located in the EF-hand binding motif of the cardiac _1C channel in calcium channel inactivation kinetics. Mutant channels E1537Q, E1537S, E1537G, and E1537A were found to display significantly slower inactivation kinetics in Ba²⁺ and Ca²⁺ solutions. Whole-cell Ba²⁺ and Ca²⁺ currents were found to proceed more slowly as the E residue was replaced by hydrophobic residues in the following order: E > Q > G  S > A. Slower kinetics can indeed be readily observed in Ba²⁺ solutions. Hence our results show that point mutations in the proximal part of the C-terminus may additionally impair Ba²⁺-dependent inactivation rather than intrinsically modifying Ca²⁺-dependent inactivation. The result that mutations at E1537 could alter Ba²⁺-dependent inactivation was quite unexpected, as this region is known to be involved in calcium-dependent inactivation. Alanine substitution was most crucial in that regard, with a fourfold increase in _Ba^inact in Ba²⁺ and _slow^inact in Ca²⁺ as compared to the wild-type _1C channel.

Over the years, Ba²⁺ solutions have typically been used to assess voltage-dependent inactivation in L-type calcium channels (Kass and Sanguinetti, 1984; Campbell et al., 1988). More recent observations, however, revealed that ion-dependent inactivation may subsist in the presence of Ba²⁺ in some L-type and non-L-type calcium channels (Ferreira et al., 1997; Parent et al., 1997; Forsythe et al., 1998). By definition, pure voltage-dependent inactivation should proceed unimpaired in the complete absence of divalent cations, with either Na⁺ or Li⁺ as the charge carrier. Thus the Ba²⁺ data reported in this paper cannot be equated a priori with a true measure of voltage-dependent inactivation in _1C channel and mutants. A few lines of evidence suggest, however, that in this series of experiments, Ba²⁺ inactivation could represent a fair approximation of voltage-dependent inactivation. First, as seen in Fig. 5, the slow inactivation time constants between Ba²⁺ and Ca²⁺ turned out to be remarkably similar between 10 and +20 mV for wild-type and mutant channels. The apparent similarity between _Ba^inact and _slow^inact in Ca²⁺ solutions argues that the mutations are probably primarily affecting a inactivation path present in both Ba²⁺ and Ca²⁺ solutions. Such a gating transition could thus be identified as the voltage-dependent transition to the inactivated state, because voltage-dependent inactivation remains in Ca²⁺ solutions. Our kinetic analysis further suggests that the overall Ba²⁺ kinetics probably result mostly from an impaired voltage-dependent inactivation, as the voltage-dependent time constants for E1537 mutants increased rather than decreased, in response to membrane depolarization (see Fig. 5). The impaired voltage-dependent inactivation in E1537 mutant channels could thus be said to be responsible for the overall slower inactivation kinetics in Ba²⁺ as well as in Ca²⁺ solutions. Hence, although E1537 mutant channels displayed overall slower inactivation under the same conditions as the wild-type _1C channel, their fast Ca²⁺-dependent inactivation time constant (_fast^inact) remained relatively constant. Ca²⁺-dependent inactivation remained typically and significantly faster than Ba²⁺ inactivation kinetics for a