DEPENDENCE OF INACTIVATION SUBUNIT IN THE KINETICS AND Ca2+ 1C α ENCODED BY EXONS 40-42 IN THE CARBOXYL-TERMINAL REGION Calcium

,

DHP 1 -sensitive Ca 2ϩ channels of class C (1) are voltagegated channels, which start to open at membrane voltages more positive than Ϫ40 mV and slowly inactivate if Ba 2ϩ is the charge carrier. Inactivation is usually greatly accelerated if Ba 2ϩ is replaced by Ca 2ϩ (2). The channels are also designated as L-type and are multisubunit proteins composed of the poreforming ␣ 1C subunit, which contains high affinity binding sites for DHPs (3)(4)(5)(6)(7), and of the auxiliary ␤ and ␣ 2 ␦ subunits (8,9). Analysis of the hydrophobicity profile of ␣ 1C indicates four repetitive motifs of similarity (I-IV), each composed of six transmembrane segments (S1-S6) (10). Both, the short aminoterminal tail encoded by exons 1 and 2, and the long carboxylterminal tail encoded by exons 38 -50 of the human ␣ 1C gene (11) are located in the cytoplasm.
Expression of ␣ 1C is regulated through alternative splicing (12), which has primarily been detected in the membranespanning regions of the molecule. However, there is evidence that the carboxyl-terminal tail is also affected by alternative splicing. Two partial transcripts have been identified in a cDNA library of human hippocampus (11,13). They show that exons 40 -43 encoding the second quarter of the putative cytoplasmic tail of the ␣ 1C molecule are subject to alternative splicing and may give rise to new ␣ 1C splice variants in the brain.
The functional role of the carboxyl-terminal tail attracts much attention because of its potential involvement in channel gating. Removal of approximately 70% of the tail causes an increase in the opening probability of the rabbit cardiac ␣ 1C channel (14). A similar deletion mutant of the human cardiac ␣ 1C showed faster inactivation of the channel as compared to the wild-type channel (15). It has been concluded that this tail part of ␣ 1C may serve as a critical component of the gating structure that influences inactivation properties of the channel (15).
In this report we describe two recombinant plasmids, pHLCC72 and pHLCC86, which contain alternative exons encoding parts of the carboxyl-terminal tails that are found in human hippocampus transcripts. After expression in Xenopus oocytes, we have analyzed electrophysiological properties of ␣ 1C,72 and ␣ 1C,86 channels and compared them with the reference ␣ 1C,77 channel (16). The results of our study show that amino acids encoded by exons 40 -42 are important for the voltage dependence of activation and inactivation of the current through these channels, as well as for the kinetics and Ca 2ϩ dependence of inactivation.

MATERIALS AND METHODS
Preparation of cDNAs Encoding ␣ 1C Subunit Splice Variants-All splice variants were constructed within the frame of pHLCC77 (16) composed of exons 1-20, 22-30, 32-44, and 46 -50 using the pBluescript SK(Ϫ) vector (Stratagene) flanked at the 5Ј-end with HindIII/ BglII and at the 3Ј-end with BglII/BamHI fragments of the Xenopus ␤-globin gene untranslated region sequences, respectively (17,18). The recombinant plasmid pHLCC86 was prepared by replacing nucleotides 5104 -5482 of pHLCC77, encoding exons 41 and 42, with the BsaI/BglII fragment of the coding frame of h54 cDNA, containing exons 40A and 40B and an upstream region of exon 43A (11). The recombinant plasmid pHLCC72 was constructed by substituting the SfuI (3726)/ScaI (5348) fragment of pHLCC77 with the respective fragment of h2.05 cDNA (11,13), containing exon 41A with its 57-nt extension in the upstream direction. Nucleotide sequences of all obtained cDNAs were verified by the modified dideoxy termination method (19).
Template DNAs were linearized by digestion with BamHI (pHLCC77 and pHLCC86) or NotI (pHLCC72), and capped transcripts were synthesized in vitro with T7 RNA polymerase using the mRNA cap kit (Stratagene).
In some experiments Ca 2ϩ was used as charge carrier through the channels. One to 5 h prior to the recording of Ca 2ϩ currents (I Ca ), oocytes were injected with 50 nl of a BAPTA solution containing 40 mM Na 4 -BAPTA and 10 mM HEPES (pH 7 with KOH). The bathing solution contained (in mM): Ca(NO 3 ) 2 40, NaOH 50, KOH 1, HEPES 10 (pH 7.4 with methanesulfonic acid).
Voltage-clamp commands, current recordings and leak current subtraction were performed by means of the EPC software (Cambridge Electronic Design, Cambridge, UK). The EPC software analysis module, the KaleidaGraph software (Abelbeck, Reading, CA), and the FigP software (Biosoft, Ferguson, MO) were used for the data analysis. Statistical values are given as means Ϯ S.E. Membrane currents, filtered at 0.5-1 KHz and sampled at 2 KHz, were triggered by 0.25-or 1-s step depolarizations applied from V h of Ϫ90 mV at a frequency of 0.033 Hz. All experiments were performed at room temperature (20 -22°C).
Inactivation characteristics of I Ba through the three ␣ 1C splice variants were measured with 2-s conditioning pre-pulses. An increase of the duration of conditioning pre-pulses from2st o2 0sproduced, in all tested ␣ 1C splice variants, an additional shift of the inactivation curves by 6 Ϯ 2 mV toward more negative potentials without changes in their steepness, indicating that a steady state had not been reached with 2-s pre-pulses. However, the long pre-pulses were poorly tolerated by many oocytes; therefore, all inactivation curves reported in this paper have been obtained with a 2-s pre-pulse protocol and consequently are called "isochronic" inactivation curves.
To compare the sensitivities of ␣ 1C splice variants toward DHPs, we have measured the fractional inhibition of I Ba at V h ϭϪ 90 mV by different concentrations of (ϩ)-isradipine ranging from 10 nM to 1 M. After application of isradipine, I Ba was monitored at 30-s intervals until an equilibrium of the inhibition was reached. In these experiments endogenous, DHP-insensitive Ca 2ϩ or Ba 2ϩ currents (20) were not subtracted from recorded peak I Ba amplitudes.

Structural Features of the Studied Splice Variants-
We have compared electrophysiological properties of human ␣ 1C splice variants: ␣ 1C,77 and two of its homologues, ␣ 1C,72 and ␣ 1C,86 . Both homologues contain substitutions in the region of exons 40 -42 encoding the second quarter of the putative cytoplasmic tail of ␣ 1C (Fig. 1, upper panel, see diagram). The recombinant plasmids for ␣ 1C,72 (pHLCC72) and for ␣ 1C,86 (pHLCC86) were prepared by incorporation of partial cDNA clones (h2.05 and h54) into the nucleotide sequence of pHLCC77. These partial clones have been isolated earlier from the human hippocampus cDNA library (11,13) and proved to be products of alternative splicing of one and the same ␣ 1C gene (11). The nucleotide sequence of exon 41 in pHLCC72 is extended by 57 nt in the upstream direction and thus produces an insertion of 19 residues into the amino acid sequence of the ␣ 1C,72 channel at position 1575. In pHLCC86, 17 nt are deleted from the 3Ј-end of exon 40, the 102-nt exon 41 is replaced by the 118-nt exon 40B, and the 128-nt exon 42 is replaced by a 132-nt extension of exon 43 in the upward direction. Thus, 247 nt of the original pHLCC77 cDNA are replaced in pHLCC86 by 250 nt of a new coding sequence. At the amino acid level, this results in the replacement of 80 amino acid residues (1572-1651) of ␣ 1C,77 with 81 essentially non-identical amino acids in ␣ 1C,86 (Fig. 1). Amino acid alignments of the variable parts of the constructs have already been published (11). Because of this long stretch of non-identical amino acids, no mutagenesis has been attempted so far. All other parts of the recombinant channels studied in this work were unchanged.
None of the studied splice variants has yet been shown to be expressed in the brain. The ␣ 1C,86 channel is an "artificial" splice variant of the human ␣ 1C . The partial cDNA clone h54 used to construct pHLCC86 shows further variability due to alternative splicing downstream of exon 43, which was not incorporated into the recombinant plasmid. A full-size transcript has not yet been cloned. Moreover, a fragment of an intron upstream of exon 40A indicates that h54 is not a part of a functional transcript but rather a product of post-transcriptional processing of the ␣ 1C mRNA. However, both ␣ 1C,72 and ␣ 1C,86 showed a number of new characteristics pointing to an  Table II. G-I, averaged current-voltage relationships (10 -18 experiments) of I Ba through ␣ 1C,86 , ␣ 1C,72 , and ␣ 1C,77 . Test pulses were applied at 30-s intervals. The equation for fitting the data is given in Table III. V h ϭ Ϫ90 mV. Pore-forming ␣ 1C subunits were co-expressed with auxiliary ␤ 1 and ␣ 2 ␦ subunits at a 1:1:1 molar ratio. involvement of sequences encoded by exons 40 -42 in important gating properties of the channel.
Differences between ␣ 1C,77 , ␣ 1C,72 , and ␣ 1C,86 in Inactivation, Current-Voltage Relations, and Sensitivity to DHP Blockers-When cRNAs for ␣ 1C,77 , ␣ 1C,72 ,o r␣ 1C,86 were co-injected into Xenopus oocytes with cRNAs for auxiliary ␣ 2 ␦ (21) and ␤ 1 subunits (22,23), they gave rise to functional Ca 2ϩ channels with significantly different electrophysiological properties. Fig.  1( A-C) show traces of I Ba through splice variants of the poreforming ␣ 1C subunit recorded in response to depolarizing voltage clamp steps to ϩ20 mV (1 s) from V h ϭϪ 90 mV. The inactivation kinetics of I Ba through ␣ 1C,86 was much faster than that through ␣ 1C,77 and ␣ 1C,72 channels. A direct comparison of time constants () of inactivation obtained from exponential fits of the current traces is shown in Table I. In the case of ␣ 1C,77 and ␣ 1C,72 , the kinetics of the I Ba decay was fitted best by a single-exponential function. For ␣ 1C,86 an exponential fit indicated two time constants, where the slow time constant, s , was approximately 4 times that of the fast time constant, f ( Table  I). Subtraction of the DHP-insensitive, endogenous I Ba did not change significantly the absolute values. With ␤ 1 co-expressed, the fast component of the inactivation phase of I Ba through ␣ 1C,86 comprised 84.6 Ϯ 1.3% (n ϭ 12) of the total current recorded with a 1-s pulse, while the slow component was 15.4 Ϯ 1.3% (n ϭ 12) ( Table I). The slow component of I Ba through ␣ 1C,86 was still significantly faster than that through ␣ 1C,77 or ␣ 1C,72 channels ( Table I).
To characterize further the inactivation properties of the three splice variants of ␣ 1C , we examined the rate of recovery of I Ba from inactivation. Fig. 3 shows the ratio of maximum amplitudes of I Ba elicited by two consecutive test pulses with different intervals. The duration of the first pulse was 0.4, 2, or 3 s for ␣ 1C,86 , ␣ 1C,72 , and ␣ 1C,77 , respectively, a time required to reach 80 -90% of inactivation of the currents through these channels. The second pulse lasted 400 ms. In Fig. 3 the ratios of I Ba at pulse 2 divided by I Ba at pulse 1 are plotted as function of the time intervals between the two pulses. This represents the fractional recovery of I Ba from inactivation. Only the initial phase of recovery of I Ba from inactivation could be fitted with a single-exponential function. This phase had approximately the same time constant for all three ␣ 1C splice variants (Fig. 3). However, I Ba through the ␣ 1C,86 channel reached full recovery much faster than I Ba through ␣ 1C,72 and ␣ 1C,77 . At the 0.15-s interval between pulses, when 93 Ϯ 1% (n ϭ 4) of I Ba through ␣ 1C,86 had recovered, only 57 Ϯ 2% of I Ba through ␣ 1C,72 and 56 Ϯ 2% for ␣ 1C,77 were available. With 16-s intervals between pulses, all measured I Ba had almost completely recovered from inactivation.
As reported previously (21,24), auxiliary ␤-subunits affect, among other properties, the kinetics of the Ca 2ϩ channel current. We have found that ␤ 1 , ␤ 2A or ␤ 3 subunits, when coexpressed with ␣ 2 ␦ and the splice variants of ␣ 1C subunits, caused modulatory effects on the inactivation kinetics of I Ba , which, however, were smaller than the differences between ␣ 1C,86 and ␣ 1C,77 or ␣ 1C,72 (Table I).
Alternative splicing of exons 40 -42 affects gating properties of the channel. Table II and Fig. 1 (D-F) show isochronic (2-s TABLE I Dependence of the kinetics of the I Ba decay on the type of co-expressed ␣ 1C and ␤ subunits Inactivation time constants, ,ofI Ba were determined by test pulses to ϩ20 mV from V h ϭϪ90 mV. For ␣ 1C,72 and ␣ 1C,77 channels, was obtained from a single-exponential equation: I(t) ϭ I ϱ ϩ I⅐exp(Ϫt/), where I ϱ is the steady state amplitude of the current and I is the amplitude of the initial current. The best fit for the current through the ␣ 1C,86 channel was obtained by a bi-exponential equation: I(t) ϭ I ϱ ϩ I f ⅐exp(Ϫt/ f ) ϩ I s ⅐exp(Ϫt/ s ), where f and s stand for fast and slow components, respectively. I Ba fractions refer to the respective contributions of fast and slow components to the total current through ␣ 1C,86 . In all cases the subunit composition of the analyzed channels was ␣ 1C :␤:␣ 2 ␦ (1:1:1, moles). n ϭ number of tested oocytes. *, p Ͻ 0.05 compared to the respective ␣ 1C,77 channel (one-way ANOVA and Tukey test).  2.Dependence of I Ba inactivation kinetics on membrane potential. A, traces of I Ba recorded at 0, ϩ20, and ϩ40 mV. The voltage dependence of the inactivation time constant () was determined by fitting current traces of I Ba in the range of 0 to ϩ40 mV with exponential functions (B). Values of for ␣ 1C,72 (•, n ϭ 10) and ␣ 1C,77 (E, n ϭ 15) were determined by mono-exponential fitting. A bi-exponential approximation was used to obtain values for ␣ 1C,86 ; only the fast component has been plotted in B and C (Ⅺ, n ϭ 10 -14). To illustrate differences in the voltage dependence of for the three ␣ 1C splice variants, the values of at each potential were normalized with respect to at ϩ40 mV (C). The subunit composition of the analyzed channels was ␣ 1C :␤ 1 :␣ 2 ␦ (1:1:1, mol). pre-pulses) inactivation characteristics of I Ba through the three ␣ 1C splice variants. Isochronic inactivation curves were shifted toward negative potentials by 5 mV (␣ 1C,72 ) and 11 mV (␣ 1C,86 ) with respect to that of ␣ 1C,77 (Fig. 1, D-F ; Table II, see V 0.5 values). The slopes of isochronic inactivation curves were less steep for ␣ 1C,86 and ␣ 1C,72 channels than for ␣ 1C,77 (Table II). Thus, cooperativity in the mechanism leading to inactivation of I Ba may be different for ␣ 1C,86 and ␣ 1C,72 than for ␣ 1C,77 .
Current-voltage relationships also point to differences in the voltage dependence of I Ba through ␣ 1C,77 as compared to the other two splice variants (Table III, Fig. 1, G-I). In contrast to the negative shift of the inactivation curves of I Ba through ␣ 1C,72 and ␣ 1C,86 , their values for half-maximal activation were shifted toward more positive potentials by 6 mV and 11 mV, respectively (Fig. 1, G-I, and Table III). These data suggest that structural changes produced by alternative splicing of exons 40 -42 in ␣ 1C influence the voltage sensors of the channels for activation and inactivation in different ways. Since the reversal potentials of the current flowing through ␣ 1C,77 , ␣ 1C,72 , and ␣ 1C,86 channels are not significantly different (Table III), the pore region determining the selectivity of the channel is probably the same in the studied splice variants (25).
All three splice variants retain a high affinity for DHP block-ers. When measured at V h ϭϪ 90 mV, the IC 50 value for (ϩ)-isradipine inhibition of I Ba through ␣ 1C,77 is about 3.5 times higher than those for the other splice variants (Table III).

Differences between Splice Variants in Ca 2ϩ -dependent Inactivation-Besides voltage-dependent inactivation, many L-type
Ca 2ϩ channels exhibit Ca 2ϩ -dependent inactivation (2). This latter mode of inactivation has also been shown for heterologously expressed L-type Ca 2ϩ channels (26 -28). In view of the marked kinetic differences in inactivation between ␣ 1C,86 , ␣ 1C,77 , and ␣ 1C,72 , we studied their respective Ca 2ϩ -dependent inactivation properties.
To buffer intracellular Ca 2ϩ ions and to minimize contaminating Ca 2ϩ -dependent Cl Ϫ currents, 50 nl of 40 mM BAPTA solution were injected into the oocytes prior to the recordings. The BAPTA injection did not affect properties of I Ba . However, it could have reduced the response to Ca 2ϩ of Ca 2ϩ -dependent inactivation, although Neely et al. (26) have shown that the time course of Ca 2ϩ -dependent inactivation remains virtually unchanged over a 20-fold range of buffering capacity. In Fig. 4, representative current traces recorded from oocytes during superfusion with 40 mM Ba 2ϩ solution and after switching to 40 mM Ca 2ϩ solution were superimposed. Ca 2ϩ current (I Ca ) amplitudes were much smaller in all three channels than I Ba amplitudes. This is consistent with a lower conductance for Ca 2ϩ than for Ba 2ϩ ions of L-type calcium channels (29). The reduction was less pronounced in ␣ 1C,86 compared to ␣ 1C,77 and ␣ 1C,72 . The accelerated inactivation rate seen in ␣ 1C,77 and ␣ 1C,72 , when Ca 2ϩ was the charge carrier, was absent in ␣ 1C,86 . This is illustrated in Fig. 4B, where peak I Ca has been scaled up to the level of peak I Ba . The scaling factors for ␣ 1C,77 , ␣ 1C,72 , and ␣ 1C,86 were 3.3, 2.9, and 1.8, respectively. In contrast to I Ba inactivation kinetics of ␣ 1C,77 and ␣ 1C,72 , I Ca kinetics could not be fitted by a single exponential. With a bi-exponential fitting procedure, at ϩ20 mV the fast time constants, f ,o fI Ca inactivation were 27.7 Ϯ 1.9 ms (n ϭ 7) and 34.4 Ϯ 5.5 ms (n ϭ 4) for ␣ 1C,77 and ␣ 1C,72 , respectively. The I Ba inactivation time constants were 398.7 Ϯ 39.6 ms (n ϭ 7) and 348.1 Ϯ 22.1 ms (n ϭ 7) in these experiments. Thus, an acceleration of the inactivation kinetics by a factor of 13 and 10 was observed if Ca 2ϩ ions were the charge carriers through ␣ 1C,77 and ␣ 1C,72 . By contrast, inactivation kinetics of ␣ 1C,86 were only slightly influenced by Ca 2ϩ ions. The time constant, f , observed at ϩ20 mV in Ca 2ϩ containing solution was 78.2 Ϯ 8.0 ms (n ϭ 13) compared to 59.0 Ϯ 3.0 ms (n ϭ 10) in Ba 2ϩ . The apparent slowing of the inactivation kinetics of ␣ 1C,86 by Ca 2ϩ ions could be explained by a different surface potential with Ca 2ϩ ions in the solution (30). This is also indicated by a slight shift toward more positive potentials of the current-voltage curve of all three calcium channel constructs when switching from Ba 2ϩ to Ca 2ϩ solution . The pore-forming ␣ 1C subunit was co-expressed with auxiliary ␤ 1 and ␣ 2 ␦ subunits at equimolar ratio.

TABLE II
Dependence of isochronic inactivation curves for I Ba on ␣ 1C and ␤ subunits Isochronic inactivation curves were measured using a two-step voltage clamp protocol. A 2-s conditioning pre-pulse was applied from V h ϭϪ90 mV (10-mV increments up to ϩ40 mV) followed by a 1-s test pulse to ϩ20 mV. The intervals between each cycle were 30 s. Recorded peak current amplitudes were normalized to the maximum value determined in the range Ϫ60 to ϩ20 mV. Isochronic inactivation curves were fitted by a Boltzmann function: where V is the conditioning pre-pulse voltage, V 0.5 is the voltage at half-maximum of inactivation, and k is a slope factor. n ϭ number of tested oocytes. *, p Ͻ 0.05 compared to the respective ␣ 1C,77 channel (one-way ANOVA and Tukey test).
Ϫ16.8 Ϯ 0.9 8.7 Ϯ 0.1* 2 ( Fig. 5, B-D, filled triangles). Ca 2ϩ -induced inactivation is dependent upon the size of Ca 2ϩ influx through the channel pore (2). This can be studied by applying a double-pulse protocol as shown in Fig. 5A (upper  panel). The duration of the pre-pulse was 400 ms. A pulse interval of 50 ms was chosen, which was long enough to minimize incomplete recovery from partial inactivation during the pulse intervals. The test pulse was always to ϩ20 mV and also lasted 400 ms. The interval between cycles was 30 s. Representative current traces for ␣ 1C,77 at the pre-pulse potentials Ϫ40 mV, ϩ20 mV, and ϩ80 mV for I Ba (middle panel) and I Ca (bottom panel) are shown in Fig. 5A.InBa 2ϩ solution, increasing pre-pulse potentials led to a persistent reduction (23.1%) of I Ba through ␣ 1C,77 at the test pulse (Fig. 5, A and B). This is due to incomplete recovery from partial inactivation under these experimental conditions (data not shown). By contrast, with Ca 2ϩ as charge carrier, the test pulse current (I TP ) exhibited a bell-shaped relation as a function of the pre-pulse potential (Fig. 5B). It was inversely related to the current amplitudes at the pre-pulse potentials (I PP ). The maximal current reduction of I TP in Ca 2ϩ solution was 53.5%, and became less with further depolarization during pre-pulses (Fig. 5, A and B). This is a strong indication for Ca 2ϩ -dependent inactivation triggered by Ca 2ϩ influx through ␣ 1C,77 channels. Application of the same protocol to ␣ 1C,72 (Fig. 5C) resulted in a similar relationship between I PP and I TP as with ␣ 1C,77 . The maximal current reductions were 15.9% in Ba 2ϩ solution and 48.9% in Ca 2ϩ solution. However, for ␣ 1C,86 (Fig. 5D) the relationships between I PP and I TP in Ba 2ϩ and Ca 2ϩ solutions (maximal current reductions were 22.8% and 24.3%, respectively) were almost identical and comparable to those obtained with ␣ 1C,77 and ␣ 1C,72 in Ba 2ϩ solution. This provides further evidence that ␣ 1C,86 lacks Ca 2ϩ -dependent inactivation. DISCUSSION Alternative splicing of the ␣ 1 subunit of voltage-dependent Ca 2ϩ channels contributes to the structural diversity of these ion channels, but only little is known about its functional importance. It has been shown that alternative splicing of the gene encoding the ␣ 1C subunit of L-type Ca 2ϩ channels contributes to differences in the voltage dependence of the sensitivity toward DHPs (16) and to the DHP tissue selectivity (31). In this study we have investigated electrophysiologically three putative splice variants of the human class C L-type Ca 2ϩ channel.
We show that a segment of 80 amino acids replaced in ␣ 1C,77 by a nonidentical sequence of 81 amino acids of ␣ 1C,86 in the second quarter of the 662-amino acid carboxyl-terminal tail (1572-1651) caused a 10-fold increase in the rate of inactivation, an 11-mV hyperpolarizing shift in the voltage dependence of inactivation, and elimination of Ca 2ϩ -dependent inactivation, as well as an increase in the affinity of the channel to the DHP blocker (ϩ)-isradipine. Some but not all of these effects were also partially visible in the ␣ 1C,72 channel. It is structurally identical to the reference 2138-amino acid ␣ 1C,77 channel, except for an insertion of 19 amino acids at position 1575 between sequences encoded by exons 40 and 41. There was a 5-mV hyperpolarizing shift of the voltage dependence of inactivation, the kinetics of inactivation of I Ba through ␣ 1C,72 was only 20% faster than that through ␣ 1C,77 (Table I), and the DHP sensitivity of ␣ 1C,72 was the same as that of ␣ 1C,86 but about 4 times higher than that of ␣ 1C,77 (Table III).
Over the last few years, a multitude of studies have shed light on the molecular basis of Ca 2ϩ channel inactivation. It has been suggested that voltage-and Ca 2ϩ -dependent inactivation are regulated by distinct sites on the ␣ 1C subunit (32). The structural determinants for voltage-dependent inactivation have been attributed to sequences near or in the S6 segments of domains I, III, and IV of the ␣ 1 subunit (33,34). Substituting as few as 9 amino acids from a rapidly inactivating class A Ca 2ϩ channel near the transmembrane region IS6 for homologous residues in the ␣ 1C subunit was sufficient to transform ␣ 1C into a fast inactivating channel (33). More recently several studies have implicated carboxyl-terminal seg-  measured with 30 s intervals between 1-s test pulses in the range of Ϫ40 to ϩ100 mV (10-mV increments) applied from V h ϭϪ90 mV. In all cases the subunit composition of the analyzed channels was ␣ 1C :␤:␣ 2 ␦ (1:1:1, moles). I-V curves were fitted by where G max ϭ maximum conductance, E rev ϭ reversal potential, V 0.5 ϭ voltage at 50% of I Ba activation, and k I-V ϭ slope factor. n ϭ number of tested oocytes. Inhibition of I Ba by isradipine (IC 50 ) has been measured at a test potential of ϩ20 mV from a V h ϭϪ90 mV without correction for endogenous DHP-insensitive I Ba .* ,pϽ0.05 compared to the respective ␣ 1C,77 channel (one-way ANOVA and Tukey test). ments as being involved in voltage-dependent inactivation (15,35). The membrane-spanning regions, and consequently the voltage sensor in the S4 segments (36), are structurally identical in all three splice variants of ␣ 1C studied in our work. The new amino acid sequences in the cytoplasmic carboxyl-terminal tail, encoded by alternative exons in ␣ 1C,72 and ␣ 1C,86 , do not show hydrophobic stretches that would suggest their insertion into the plasma membrane. The differences in the voltage dependence of gating between ␣ 1C,77 , ␣ 1C,72 , and ␣ 1C,86 may, therefore, be due to an altered interaction of cytoplasmic amino acid sequences with the intramembrane voltage sensor in the S4 segments of the ␣ 1C protein. For example, a direct interaction of the amino acids encoded by exons 40 -42 with the cytoplasmic ends of the charged transmembrane segments S4 may affect the mobility of the charged regions in response to a change in the transmembrane electric field (37). This could influence the transitions between open and closed states of the channel depending on the conformational flexibility of the cytoplasmic polypeptide chains, which may be highest for the ␣ 1C,86 channel. However, we cannot rule out that the fast inactivation observed in ␣ 1C,86 may be due to some modulatory effect on the interaction with auxiliary subunits, which are known to influence Ca 2ϩ channel inactivation properties (38).
Ca 2ϩ -dependent inactivation seems to be mediated directly by binding of Ca 2ϩ ions to the channel (39). Elimination of the Ca 2ϩ selectivity by the E1145Q mutation in the pore region of repeat III of rabbit cardiac ␣ 1C was associated with a loss of Ca 2ϩ -dependent inactivation (27). A Ca 2ϩ binding site has also been implicated for a carboxyl-terminal segment near the transmembrane region IVS6 that includes a putative Ca 2ϩ binding EF-hand motif (40 -42). This motif is essential for Ca 2ϩ -dependent inactivation, although additional residues downstream to the EF-domain are required to exhibit the full effect (42)(43)(44)(45). On the other hand, neither truncation of up to 70% of the carboxyl-terminal tail of cloned ␣ 1C subunits (15,28) nor cytoplasmic modification by trypsin of cloned cardiac Ca 2ϩ channels in HEK 293 cells (15) and endogenous Ca 2ϩ channels in ventricular myocytes (35) had any effect on Ca 2ϩ -dependent inactivation. However, in another study, Ca 2ϩ -dependent inactivation in ventricular myocytes was abolished by trypsin digestion (46), indicating that there is a limit to the extent by which the carboxyl terminus can be shortened before inactiva-tion is impaired.
Our data show that substituting a stretch of 81 amino acids of ␣ 1C,86 for a segment of 80 amino acids in ␣ 1C,77 not only affects voltage-dependent inactivation, it also eliminates Ca 2ϩdependent inactivation. This substitution left the four putative transmembrane domains and the EF-hand motif intact. These structural regions are identical in both ␣ 1C constructs. Thus, our study supports recent observations (43)(44)(45) suggesting that the EF-hand motif is not the only determinant of Ca 2ϩ -dependent inactivation (42). We could narrow down a carboxyl-terminal regulatory domain to a segment of maximally 81 amino acids.
Our studies have shown an alternatively spliced segment in the carboxyl terminus of the human ␣ 1C subunit, which determines inactivation properties of the Ca 2ϩ channel. It remains to be elucidated which parts and residues encoded by exons 40 -42 are involved in voltage-dependent inactivation and whether these same sites are responsible for abolishing Ca 2ϩdependent inactivation of ␣ 1C,86 . Furthermore, it will be of great importance to clarify whether an ␣ 1C,86 -like splice variant is a functional class C Ca 2ϩ channel in the brain.