Molecular Determinants of L-type Ca2+ Channel Inactivation

Recently we have described a splice variant of the L-type Ca2+ channel (α1C,86) in which 80 amino acids (1572–1651) of the conventional α1C,77 were substituted by another 81 amino acids due to alternative splicing of exons 40–42. Ba2+ current (IBa ) through α1C,86 exhibited faster inactivation kinetics, was strongly voltage-dependent, and had no Ca2+-dependent inactivation. An oligonucleotide-directed segment substitution and expression of the mutated channels in Xenopus oocytes were used to study the molecular determinants for gating of the channel within the 80-amino acid domain. Replacement of segments 1572–1598 or 1595–1652 of the “slow” α1C,77 channel with the respective segments of the “fast” α1C,86 gave rise to rapidly inactivating α1C,86-like channel isoforms. We found that replacement of either motifs 1572IKTEG1576 or1600LLDQV1604 of α1C,77 with the respective sequences of α1C,86 caused strong but partial acceleration of I Ba inactivation. Replacement of both sequences produced an α1C,86-like fast channel which had no Ca2+-dependent inactivation. These results support the hypothesis that motifs 1572–1576 and 1600–1604 of α1C,77 contribute cooperatively to inactivation kinetics of α1C and are critical for Ca2+-dependent inactivation of the channel.

The voltage-gated class C ("cardiac") L-type Ca 2ϩ channel is composed of the pore-forming ␣ 1C subunit, containing the high affinity binding sites for dihydropyridines and other organic Ca 2ϩ channel blockers, and the auxiliary ␤and ␣ 2 ␦-subunits (1,2). Unlike dihydropyridine-insensitive Ca 2ϩ channels, which exhibit voltage-dependent inactivation of Ca 2ϩ current (I Ca ), L-type Ca 2ϩ -channels are in addition inactivated by Ca 2ϩ , but not Ba 2ϩ ions permeating through the channel (3). The Ca 2ϩ -binding site responsible for the Ca 2ϩ -induced inactivation of the class C Ca 2ϩ channel is thought to be located near the inner mouth, but outside the electric field of the pore (4). Substituting a small 142-amino acid segment of the Ca 2ϩinsensitive ␣ 1E channel with a homologous segment of the cytoplasmic tail of ␣ 1C , confers Ca 2ϩ sensitivity on the ␣ 1E channel (5).
Alternative splicing of the human ␣ 1C subunit generates multiple isoforms of the channel (6), including those with struc-turally altered carboxyl-terminal tail (7). Recently two splice variants of the principal 2138-amino acids pore-forming ␣ 1C subunit, ␣ 1C,86 and ␣ 1C,77 , exhibiting strong differences in their gating properties were described (8). The ␣ 1C,86 channel has 80 amino acid residues (1572-1651) in the second quarter of the 662-amino acid cytoplasmic tail of the conventional channel (␣ 1C,77 ), replaced with 81 non-identical amino acids ( Fig. 1) due to alternative splicing of exons 40 -42 (7). Both splice variants retained high sensitivity toward dihydropyridine blockers, but I Ba through ␣ 1C,86 inactivated 10 times faster than ␣ 1C,77 at ϩ20 mV. The inactivation rate of ␣ 1C,86 was strongly voltagedependent but essentially Ca 2ϩ -independent suggesting that the segment 1572-1651 of the carboxyl-terminal tail of ␣ 1C is critical for the kinetics as well as for voltage and Ca 2ϩ dependence of inactivation of ␣ 1C channel. Extended segment-substitution studies, reported here, show that amino acid residues 1572-1576 and 1600 -1604 of ␣ 1C,77 contribute in a cooperative manner to the Ca 2ϩ -binding motif(s) responsible for the feedback inhibition of ␣ 1C channel by Ca 2ϩ and the kinetics of decay of I Ba in the absence of Ca 2ϩ .
Functional Expression of Ca 2ϩ Channels in Xenopus Oocytes-Xenopus laevis oocytes were defolliculated 6 -12 h prior to injection with 50 -100 nl of a mixture containing mRNAs for ␣ 1C -, ␤ 1 -(11), and ␣ 2 ␦subunits (12) in equimolar ratio. Injected oocytes were incubated at 18°C in sterile Barth's medium supplemented with 10,000 units/liter penicillin, 6 mg/liter streptomycin, 50 mg/liter gentamicin, and 90.1 mg/liter theophilline. Membrane currents were recorded at room temperature (20 -22°C) by a two-electrode voltage clamp method using a GeneClamp 500 amplifier (Axon Instruments). Electrodes were filled with 3 M CsCl and had resistance between 0.2 and 1 megohms. The Ba 2ϩ and Ca 2ϩ extracellular (bath) solutions contained 50 mM NaOH, 1 mM KOH, 10 mM HEPES, and 40 mM Ba(OH) 2 or 40 mM Ca(NO 3 ) 2 , respectively (pH adjusted to 7.4 with methanesulfonic acid). In some experiments, about 1 h prior to recording, Ca 2ϩ current (I Ca )-expressing, oocytes were injected with 50 nl of 94 mM Cs 4 -BAPTA (pH 7.4). Currents were filtered at 1 kHz. Data were acquired using pClamp 5.5 software (Axon Instruments), corrected for leakage using an on-line P/4 subtraction paradigm and analyzed with KaleidaGraph software. Results are shown as mean Ϯ S.E. "Endogenous" current, determined in the presence of 5 M (Ϯ)-PN200 -110 to block the L-type current, did not exceed 3% of the total current.

RESULTS
Strategy of Segment Exchange Analysis-Since it was unclear whether the altered properties of the ␣ 1C,86 channel were due to the lost determinants normally present in ␣ 1C,77 and/or were due to those imposed by the new 81-amino acid motif, we chose a well defined slow L-type Ca 2ϩ channel isoform (␣ 1C,77 channel (9)) as a primary target for the mutation studies. A series of segment exchange experiments were performed on pHLCC77 plasmid encoding ␣ 1C,77 to map the molecular determinants for the ␣ 1C inactivation kinetics as well as its voltage and Ca 2ϩ dependence within the experimentally targeted 80amino acid (1572-1651) sequence of the carboxyl-terminal tail. To narrow the search, we substituted initially two large fragments of this sequence in ␣ 1C,77 with respective sequences from ␣ 1C,86 (Fig. 1, mutants 77L and 77K) by introducing an HLCC86-specific restriction site into the ␣ 1C,77 -encoding DNA sequence (mutant 77M2, Fig. 1). The mutated channels were then expressed in Xenopus oocytes by coinjecting an equimolar mixture of cRNAs encoding the mutated ␣ 1C -, ␤ 1 -(11), and ␣ 2 ␦-subunits (12). Because both mutants showed accelerated inactivation kinetics of I Ba and loss of Ca 2ϩ -dependent inactivation, shorter segments of 5-7 amino acids in this region of slowly inactivating ␣ 1C,77 were replaced by residues from the equivalent positions of rapidly inactivating ␣ 1C,86 using oligonucleotide-directed segment exchange technique (mutants 77M1, 77M3, and 77M5 shown in Fig. 1). As we shall show below, it was the double mutant 77M1,3 ( Fig. 1) that best mimicked the properties of the ␣ 1C,86 channel.

Introduction of HLCC86-specific SacI Restriction Site within the Segment (1595-1598)-coding Sequence Does Not Change
Properties of Mutated ␣ 1C,77M2 -The nt sequence of the ␣ 1C,86encoding plasmid contains a convenient SacI restriction site which is absent from pHLCC77. This site, when introduced into the ␣ 1C,77 -coding DNA sequence, allows the transfer of large segments of the 81-amino acid motif of the "fast" ␣ 1C,86 channel (e.g. 77L and 77K, Fig. 1 We introduced this SacI restriction site into pHLCC77 by replacing the segment (nt 4783-4794), which encodes the amino acid residues 1595-1598 of the "slow" ␣ 1C,77 channel, with the respective segment of pHLCC86 coding for SSHP (mutant 77M2, Fig. 1). The resulting ␣ 1C,77M2 channel showed no significant electrophysiological differences compared with ␣ 1C,77 (Table II, 3A) show that about 90% of I Ba decayed rapidly in all these channels (Table II), and their voltage dependence were almost identical (Table III, Fig. 3B). Comparison of the steady-state inactivation curves for ␣ 1C,77L , ␣ 1C,77K , and ␣ 1C,86 using 2-s conditioning prepulses, however, showed that the steady-state inactivation curve for ␣ 1C,77L was shifted by about Ϫ20 mV with respect to that of ␣ 1C,86 and by Ϫ30 mV relative to ␣ 1C,77 (Table IV). Thus, structural changes produced by the substitution of the 27-amino acid segment alone altered the inactivation properties of the ␣ 1C,77L mutant more effectively than the exchange of the 81-amino acid segment of ␣ 1C,86 , containing the 27-amino acid motif.
The voltage dependence of f for I Ba through ␣ 1C,77L and ␣ 1C,77K channels was similar to that of ␣ 1C,86 (Fig. 3C). There was approximately a 2.5-3-fold decrease of f at ϩ40 mV compared with f at 0 mV. In both channels the fraction of I Ba following the fast decay accounted for approximately 95% of the total current at ϩ40 mV. However, a larger fraction of the current through 77L and 77K mutants continued to inactivate rapidly at 0 mV. Thus, voltage dependence of inactivation of I Ba through ␣ 1C,77L was stronger than in ␣ 1C,86 .
Similar to ␣ 1C,86 (8), ␣ 1C,77L and ␣ 1C,77K channels did not exhibit Ca 2ϩ -dependent inactivation of I Ca (Fig. 3D) as indicated by the absence of characteristic U-shaped voltage dependence of f (e.g. see for example, ␣ 1C,77M2 , Fig. 2C). Therefore, ␣ 1C,77L and ␣ 1C,77K channels appear to have electrophysiological properties quite similar to those of ␣ 1C,86 channel. This data suggests that there may be at least two molecular determinants for the gating kinetics and Ca 2ϩ dependence of inactivation of the channel, one located possibly within the 22-amino acid segment 1572-1593 (see Fig. 1), and the other within the 54-amino acid motif (1599 -1652).
Motif M1 Determines the Fractional Ratio of Fast to Slow Inactivation-To further narrow the 22-amino acid motif (1572-1593) that confers the inactivation properties of ␣ 1C,86 to ␣ 1C,77L channel, a number of ␣ 1C,77 mutants containing shorter segments of this motif were prepared. The carboxyl-terminal part (1588 -1592; mutant M5 in Fig. 1) of the 77L motif did not cause appreciable changes in the properties of the mutated ␣ 1C,77M5 channel as compared with ␣ 1C,77 (Fig. 4A, Tables II  and III). Within the remaining 16-amino acid sequence (1572-1587), we identified a 5-amino acid segment (1572-1576, mutant 77M1 in Fig. 1) that was critical for the faster inactivation. Electrophysiological properties of the ␣ 1C,77M1 channel are presented in Fig. 4. The inactivation kinetics of I Ba through  ␣ 1C,77M1 was much faster than through the ␣ 1C,77 channel (Fig.  4A). This may have resulted from about 2-fold increase in the fraction of I Ba that inactivates rapidly (Table II). Interestingly, the time constant of inactivation of the fast current component remained virtually unchanged at ϩ20 mV as compared with that of the ␣ 1C,77 channel. Furthermore, the voltage dependence of I Ba through ␣ 1C,77M1 channel ( Fig. 4B) was not significantly different from that of the ␣ 1C,77 channel (Table III). However, the steady-state inactivation of I Ba through ␣ 1C,77M1 shifted by approximately Ϫ20 mV with respect to the ␣ 1C,77 channel (Table IV). At the same time, the voltage dependence of time constants of inactivation of I Ba through ␣ 1C,77M1 (Fig.  4B) was steeper than in ␣ 1C,77 but did not reach values seen with the ␣ 1C,77L channel (Fig. 3C). In contrast to the other rapidly inactivating ␣ 1C,77 mutants, f for I Ca through ␣ 1C,77M1 channel exhibited weak U-shaped voltage dependence (Fig. 4C) consistent with the idea that ␣ 1C,77M1 remains somewhat Ca 2ϩ sensitive. Thus, amino acids in positions 1572-1576 may be important for the fractional ratio of fast to slow inactivation of I Ba .

A Second 5-Amino Acid Motif Critical for the Faster Kinetics of Inactivation of I Ba and Its Ca 2ϩ and Voltage Dependence-
Segment exchange analysis also revealed a second 5-amino acid motif 1600 -1604 (mutant 77M3 in Fig. 1) involved in the gating of the channel. Inactivation of I Ba through ␣ 1C,77M3 occurred with time constants similar to those of the fast ␣ 1C,86 (Fig. 5A, Table II). However, the fraction of the slow component of I Ba was approximately 3 times greater at ϩ20 mV in the ␣ 1C,77M3 channel, causing considerable retardation of the inactivation compared with ␣ 1C,86 . Both channels exhibited approximately the same strong voltage dependence of f for I Ba (compare Figs. 3C and 5B). When f for the inactivation of I Ca through the ␣ 1C,77M3 channel was plotted versus membrane potential, neither strong nor U-shape voltage dependence were observed (Fig. 5D). Thus, ␣ 1C,77M3 channel lacks the Ca 2ϩ -dependent inactivation.
Acceleration of inactivation kinetics and disruption of Ca 2ϩ sensitivity of inactivation produced by M3 mutation in ␣ 1C,77 channel were not accompanied by changes in the properties of the voltage sensor observed in ␣ 1C,86 (8). The voltage dependence of I Ba through ␣ 1C,77M3 (Fig. 5B, Table III) and its steadystate inactivation (Table IV) remained similar to those of ␣ 1C,77 (Tables III and IV). Therefore, the 5-amino acid motif (1600 -1604) responsible for the faster kinetics, and Ca 2ϩ and voltage dependence of inactivation is distinct from the molecular structures that cause voltage shifts of activation and inactivation in ␣ 1C,86 (8).
We extended the M3 mutation of the ␣ 1C,77 channel and found that amino acids immediately preceding the M3 motif may cause changes in voltage dependence of inactivation. The "extended" mutant ␣ 1C,77M3m contained uncharged hydrophobic Val replacing the positively charged Lys-1599, common for both the slow (␣ 1C,77 ) and fast (␣ 1C,86 ) channels (Fig. 1), and preceded by bulky Phe residue. I Ba through the ␣ 1C,77M3m channel exhibited a strong shift of steady-state inactivation curve toward negative potentials (Table IV) with opposite but much smaller shift of the current-voltage relation (Table III). Similar strong shift of the steady-state inactivation was displayed by the ␣ 1C,77L mutant, which had no overlapping mutated structure (Fig. 1). Therefore, within the cytoplasmic motif (1572-1651) of ␣ 1C,77 , there are at least two non-overlapping regions that may strongly affect intramembrane voltage sensors of the channel responsible for inactivation.
Double Mutant M1 ϩ M3 Shows the Best Fit to the ␣ 1C,86 Channel-When M1 mutation was combined with the M3 mutation, the resulting ␣ 1C,77M1,3 channel exhibited all the prop- erties of the ␣ 1C,86 channel. Fig. 5A shows representative trace of I Ba through ␣ 1C,77M1,3 elicited by step depolarization to ϩ20 mV from V h ϭ Ϫ90 mV. Approximately 90% of the current inactivated with f of 40 to 45 ms, a distinct characteristic of ␣ 1C,86 channel (Table II). Furthermore, the time constant of inactivation of I Ba through ␣ 1C,77M1,3 was strongly voltage-dependent (Fig. 5C). As compared with ␣ 1C,77M1 and ␣ 1C,77M3 , the voltage dependence of I Ba through the double mutated channel was shifted by about ϩ15 mV (Fig. 5C) but all other parameters were essentially identical to those of ␣ 1C,86 channel (Table III).
Double-mutated ␣ 1C,77M1,3 channel did not show Ca 2ϩ -dependent inactivation as evidenced by the absence of characteristic U-shaped voltage dependence of f for I Ca (Fig. 5E). Unlike ␣ 1C,77M3 channel (Fig. 5D), inactivation kinetics of I Ca through ␣ 1C,77M1,3 was strongly voltage dependent (Fig. 5E). Taken together, these data suggest that motifs M1 and M3 complement each other in disrupting the functional site of Ca 2ϩ -dependent inactivation. Thus, two 5-amino acid motifs located at positions 1572-1576 and 1600 -1604 of ␣ 1C,77 cooperatively participate in the gating function of the channel. DISCUSSION Unlike Ba 2ϩ current, the magnitude of inactivation of I Ca depends on the size of I Ca so that the time constant of inactivation exhibits a U-shaped voltage dependence. Ca 2ϩ accelerates inactivation of the L-type Ca 2ϩ channel by reducing its open probability by interacting with the ␣ 1C subunit (4,14) in equimolar ratio (15). The Ca 2ϩ -binding site for Ca 2ϩ -dependent inactivation has been postulated to be very close to the internal opening of the pore (4) but outside of the conduction pathway (16). The Ca 2ϩ -dependent inactivation was thought to be linked (17) to a 29-amino acid domain (1506 -1534) with homology to Ca 2ϩ -binding EF-hand motif (18). Recent experiments using mutation analysis (5), however, have excluded amino acids  Ba 2ϩ currents were measured in response to 1-s test pulses to ϩ20 mV from V h ϭ Ϫ90 mV applied at 30-s intervals in the range of Ϫ40 to ϩ60 mV with 10-mV increments. The fit was obtained by equation: where E rev is reversal potential, V 0.5 Ϫ voltage at 50% of I Ba activation, and k I-V Ϫ slope factor. G max , maximum conductance (not shown); instead are presented maximum amplitudes of I Ba (I Ba(max) ) measured at voltage for the peak current.  Steady-state inactivation curves were measured using a 2-step voltage clamp protocol. A 2-s conditioning pulses were applied at 30-s intervals with 10-mV increments up to ϩ20 mV from V h ϭ Ϫ90 m V followed by a 250-ms test pulse to ϩ20 m V. Peak current amplitudes were normalized to maximum value. The curves were fitted by a Boltzmann function: where V is the conditioning pulse voltage; V 0.5 is the voltage at half-maximum of inactivation, and k is a slope factor.
1506 -1534 from this function. Other observations (5,8) suggest that the location of the site for Ca 2ϩ -dependent inactivation may be at least 40 amino acids toward the carboxyl terminus of the polypeptide chain. None of the mutations described in this paper have caused appreciable changes in the kinetics of I Ca inactivation. However, the kinetics of I Ba decay was changed in a gradual manner (Table II). Since f decreases in the following order among the mutants: 86 , it is likely that retardation of inactivation can be eased through specific structural changes not requiring Ca 2ϩ . The values of f for the decay of I Ba through the fast mutants are almost identical to those for I Ca through the slow channel, suggesting that both modes of inactivation may be thermodynamically similar.
It would be reasonable to assume that mutations leading to faster decay rates of I Ba disrupt intraprotein interactions responsible for retardation of inactivation, which otherwise may be achieved through interaction with Ca 2ϩ . First, inactivation time constants of I Ba through fast channels (␣ 1C,77K , ␣ 1C,77L , ␣ 1C,77M1,3 , and ␣ 1C,86 ) do not exhibit a U-shaped voltage dependence, and therefore Ba 2ϩ does not influence inactivation in a current-dependent manner. Second, all fast channels are deprived of Ca 2ϩ -dependent inactivation.
It is the 77M1 motif (1572-1576, Fig. 1) that is common to both fast mutants. However, when each motif is introduced alone, it causes only partial disruption of Ca 2ϩ sensitivity and acceleration of I Ba . Therefore, for the full effect to occur, M1 must be supplemented with another determinant. The latter may be mobilized either from the adjacent sequence (1577-1587) of the 77L motif or from a distant 77M3 motif. These segments in ␣ 1C,77 channel are responsible for Ca 2ϩ -dependent inactivation and, in the absence of Ca 2ϩ , for the accelerated kinetics of I Ba inactivation. Thus, these motifs are involved in the gating function of the channel possibly through a direct interaction with the pore.
It is possible that 8 out of 11 amino acids (1574 -1584, Fig. 1) of ␣ 1C,77 represent residues that may form coordination bonds with Ca 2ϩ . However, disruption of this motif by insertion of 19 additional amino acids at position 1576 neither changed the kinetics of I Ba decay, nor the Ca 2ϩ -dependent inhibition of otherwise invariant channel isoform, ␣ 1C,72 (8). Moreover, ␣ 1C,77M1 mutant retained some Ca 2ϩ -induced inactivation property. Since segment 1577-1584 in both the slow ␣ 1C,72 and the fast ␣ 1C,77M1,3 channels is intact (Fig. 1), the remaining segment (1572-1575) is apparently critical for gating. Coordination of Ca 2ϩ in Ca 2ϩ -binding sites involves seven bonds that are spatially distributed as an octahedron (19) and can be separated into two vertices. Our data suggest that either of the two motifs, 1577-1587 or 1600 -1604, may serve as the second critical element for the channel gating. Additional point mutation analysis is needed to make a final judgment about critical amino acids in the proposed Ca 2ϩ -coordination motif responsible for the channel gating.
Our data point to the complexity of molecular determinants for Ca 2ϩ dependence of inactivation. In addition to motifs discussed above, there may be other sequences which determine fast inactivation kinetics and loss of Ca 2ϩ -induced inactivation of ␣ 1C,77K channel. Investigation of these motifs is in progress.