Molecular determinants of voltage-dependent slow inactivation of the Ca2+ channel.

Ba(2+) current through the L-type Ca(2+) channel inactivates essentially by voltage-dependent mechanisms with fast and slow kinetics. Here we found that slow inactivation is mediated by an annular determinant composed of hydrophobic amino acids located near the cytoplasmic ends of transmembrane segments S6 of each repeat of the alpha(1C) subunit. We have determined the molecular requirements that completely obstruct slow inactivation. Critical interventions include simultaneous substitution of A752T in IIS6, V1165T in IIIS6, and I1475T in IVS6, each preventing in additive manner a considerable fraction of Ba(2+) current from inactivation. In addition, it requires the S405I mutation in segment IS6. The fractional inhibition of slow inactivation in tested mutants caused an acceleration of fast inactivation, suggesting that fast and slow inactivation mechanisms are linked. The channel lacking slow inactivation showed approximately 45% of the sustained Ba(2+) or Ca(2+) current with no indication of decay. The remaining fraction of the current was inactivated with a single-exponential decay (pi(f) approximately 10 ms), completely recovered from inactivation within 100 ms and did not exhibit Ca(2+)-dependent inactivation properties. No voltage-dependent characteristics were significantly changed, consistent with the C-type inactivation model suggesting constriction of the pore as the main mechanism possibly targeted by Ca(2+) sensors of inactivation.

Ba 2؉ current through the L-type Ca 2؉ channel inactivates essentially by voltage-dependent mechanisms with fast and slow kinetics. Here we found that slow inactivation is mediated by an annular determinant composed of hydrophobic amino acids located near the cytoplasmic ends of transmembrane segments S6 of each repeat of the ␣ 1C subunit. We have determined the molecular requirements that completely obstruct slow inactivation. Critical interventions include simultaneous substitution of A752T in IIS6, V1165T in IIIS6, and I1475T in IVS6, each preventing in additive manner a considerable fraction of Ba 2؉ current from inactivation. In addition, it requires the S405I mutation in segment IS6. The fractional inhibition of slow inactivation in tested mutants caused an acceleration of fast inactivation, suggesting that fast and slow inactivation mechanisms are linked. The channel lacking slow inactivation showed ϳ45% of the sustained Ba 2؉ or Ca 2؉ current with no indication of decay. The remaining fraction of the current was inactivated with a single-exponential decay ( f ϳ 10 ms), completely recovered from inactivation within 100 ms and did not exhibit Ca 2؉ -dependent inactivation properties. No voltage-dependent characteristics were significantly changed, consistent with the Ctype inactivation model suggesting constriction of the pore as the main mechanism possibly targeted by Ca 2؉ sensors of inactivation.
The voltage-gated inward current of Ca 2ϩ ions is a common mechanism of transient increase in the cytoplasmic free Ca 2ϩ concentration that stimulates a great variety of cellular responses. The rapid and complete inactivation of Ca 2ϩ current is the critical step terminating Ca 2ϩ influx and preventing Ca 2ϩ overloading of the cell. In the case of L-type (␣ 1C ) Ca 2ϩ channels, two different mechanisms are in control of the Ca 2ϩ current inactivation (1). One mechanism is driven by Ca 2ϩ ions on the cytoplasmic side of the membrane (2), whereas the other depends on transmembrane voltage. Replacement of Ca 2ϩ ions by Ba 2ϩ ions in the extracellular medium eliminates Ca 2ϩ -dependent inactivation so that Ba 2ϩ -conducting Ca 2ϩ channels are inactivated in a voltage-dependent manner.
Two major mechanisms have been previously implicated in voltage-dependent inactivation (3). The ball and chain mechanism of an ion pore occlusion by a positively charged segment of the N-terminal tails was first described in the tetrameric Shaker K ϩ channel where it supports fast N-type inactivation (4). Somewhat similarly, the hinged-lid mechanism in the Na ϩ channel is mediated by the IFM motif of the cytoplasmic linker between repeats III and IV (5). In both Na ϩ and K ϩ channels, receptor sites for the different inactivation gates are located in S4 -S5 intracellular loops (6,7). The second, C-type mechanism of slower K ϩ channel inactivation (8) was found to involve a constriction of the pore by the S6 segments lining the intracellular part of the pore and arranged as an inverted teepee structure (9,10).
The voltage-dependent inactivation of ␣ 1C Ca 2ϩ channels appears to be more complex (for review, see Refs. 11,12). Experimental trials of chimeras between ␣ 1C and the faster inactivating Ca 2ϩ channels showed that multiple regions are involved in inactivation, including transmembrane segments IS6 (13), IIIS6 (14,15), IVS5 (16), IVS6 (17)(18)(19), repeats I-II linker (20 -22), as well as the C-terminal determinants, E1537 of EF-hand motif (23) and the Ca 2ϩ -sensing 80-amino acid domain 1572-1651 (24). In these regions, multiple amino acids were shown to be critical for the rate of inactivation of the Ba 2ϩ current (12). A systematic analysis of these multiple determinants has not been performed.
In this work we focused on determinants situated in all four transmembrane segments S6. The approach was based on our earlier observation (25) that Ca 2ϩ channel inactivation was impaired by the A752T mutation at a position Ϫ2 from the cytoplasmic end of IIS6 identified in the human fibroblast ␣ 1C Ca 2ϩ channel transcript (26). Similarly, Val-1504 in IVS6 of the rabbit cardiac ␣ 1C was shown to be critical for the channel inactivation (18). The goal of our work was to determine, by single and combined amino acid substitutions, the role and molecular requirements for the involvement of S6 segments in voltage-dependent inactivation. The results are consistent with the C-type inactivation model (4,8,12,13) and suggest that the slow inactivation of Ca 2ϩ channels is mediated by an annular determinant composed of amino acid residues situated in the cytoplasmic ends of transmembrane segments S6 in repeats I-IV. Complete removal of slow inactivation by a specific set of mutations in all four repeats gave us a unique opportunity to investigate the properties of Ca 2ϩ channels that retain only the fast component of inactivation.
Expression in Xenopus Oocytes-Wild-type (␣ 1C,WT ) and mutated ␣ 1C subunits were co-expressed with ␤ 1 (30) and ␣ 2 ␦ subunits (31) as previously described (29). Membrane currents were recorded at 20 -22°C by a two-electrode voltage clamp method using a TEV-200A amplifier (Dagan). The bath solution contained, in millimolar, 40 Ba 2ϩ , 50 Na ϩ , 1 K ϩ , 5 HEPES, and 0.3 niflumic acid (pH 7.4 with methanesulfonic acid). Electrodes were filled with 3 M KCl and had resistances between 0.2 and 1 M⍀. One hour before recordings, oocytes were injected with 50 nl of 100 mM BAPTA (1,2-bis(O-aminophenoxy)ethane-N,N,NЈ,NЈ-tetraacetic acid) (pH 7.4 with CsOH). The duration of test pulses used for the calculation of inactivation kinetics was 1 s to avoid deterioration of the stability of oocyte preparation. Currents were filtered at 1 kHz and sampled at 2.5 kHz. pClamp 8 software (Axon Instruments) was used for data acquisition, and KaleidaGraph software was used for the analysis. The values of the double-exponential fitting are presented only for the purpose of comparison of different mutants; they characterize only the 1-s current kinetics and do not represent the absolute values. Data are given as means Ϯ S.E. Endogenous current, determined in the presence of 5 M (Ϯ)-PN200-110 to block the ␣ 1C current, did not exceed 1-2% of the total current.

RESULTS
The Role of Hydrophobic Amino Acids in the Ϫ2 Position of the Cytoplasmic Ends of Segments S6 of Repeats II, III, and IV in Ca 2ϩ Channel Inactivation-The mutation A752T (26) that impairs Ca 2ϩ channel inactivation (25) is located at the position Ϫ2 from the cytoplasmic end of the putative transmembrane segment IIS6 (Table I). The analogous S6 positions in repeats III and IV of ␣ 1C are also occupied by hydrophobic amino acids that are essentially conserved among the other Ca 2ϩ channels except in the cases of ␣ 1A and ␣ 1B , that have an equivalent replacement V1165I (12). To examine the role of these hydrophobic amino acids in Ca 2ϩ channel inactivation, the homologous mutations V1165T in IIIS6 (␣ 1C,III ) and I1475T in IVS6 (␣ 1C,IV ) have been introduced alone and in combinations with each other or with A752T (␣ 1C,II ).
The test pulses to ϩ10 mV eliciting maximum currents were selected to compare the kinetics (Table II and Fig. 1A) and voltagedependence of activation and inactivation of the mutated channels (Fig. 1, B and C). For every trace of Ba 2ϩ current, the steady-state current was determined from the fit (see Table II) and normalized to the peak amplitude. This sustained component of the current (I o ) is a measure of the non-inactivating fraction of the Ba 2ϩ current. The inactivating component of the current was further analyzed by doubleexponential fitting as a sum of the fast and slow components. Increase of the pulse duration to 2 s did not greatly change the calculated parameters (see Table II, asterisk).
The results, summarized in Table II, show that the mutations to threonine introduced at the Ϫ2 positions in any of the S6 segments in repeats II, III, or IV increased the size of the sustained Ba 2ϩ current at the end of a 1-s depolarization pulse ϳ1.4to 1.7-fold compared with the wild-type channel. None of the mutations have markedly altered the kinetics of inactivation. Indeed, the faster inactivating component representing 60 -70% of Ba 2ϩ current decay has the time constant f ranging from 71 to 86 ms. The remaining slower component of the Ba 2ϩ current inactivated with a s from 395 to 415 ms, i.e. in the range characteristic for the ␣ 1C, WT channel (Table II).
When single mutations were introduced, a 3-to 4-mV shift in both directions in steady-state inactivation curves was observed ( Fig. 1C). At the end of the 2-s conditioning pulse, 6.1 Ϯ 1.4% of the maximum Ba 2ϩ current through the wild-type channel remained noninactivated. This fraction increased to 11.5-21.2% by the threonine substitution in repeats II-IV, suggesting that inactivation of the mutated channels was obstructed. The maximum effect of a single mutation was seen with ␣ 1C, II , thus confirming the results of our earlier study of this mutant expressed in HEK293 cells (25). Thus, the hydrophobic residues of Ala-752 in IIS6, Val-1165 in IIIS6, and Ile-1475 in IVS6 each contribute to the voltage-dependent inactivation of the ␣ 1C channel. Mutation of these amino acids to hydrophilic Thr appears to impair the transition from the open to inactivated state and/or destabilizes the inactivated state of the channel.
Combined Mutations in Repeats II, III, and IV-Given the similarity in the effects of the tested single mutations to Thr, our data may suggest that the hydrophobic amino acids in the Ϫ2 positions of segments IIS6, IIIS6, and IVS6 are concurrently involved in voltage-dependent inactivation. To test this hypothesis, the S6 mutations in repeats II-IV were combined in the double and triple mutants listed in Table II. We observed that each additional homologous mutation inhibited a considerable fraction of the Ba 2ϩ current inactivation ( Fig. 1A and Table II) without significantly altering current-voltage relationships (Fig. 1B). The voltage dependences of the time con- stants of inactivation were not strongly affected (Fig. 1D).
Indeed, the ratio f(Ϫ10) / f (40) of the time constants measured at Ϫ10 mV and ϩ40 mV for Ba 2ϩ currents through the ␣ 1C,II,IV , ␣ 1C,III,IV , and ␣ 1C,II,III,IV channels decreased less than 2-fold compared with ␣ 1C,WT and other mutants. Steady-state inactivation curves showed an increase of the noninactivating component of Ba 2ϩ current from 6.1 Ϯ 1.4% in ␣ 1C,WT to 48.5 Ϯ 3.0% in ␣ 1C,II,III,IV (Fig. 1C). For some of the mutants, these curves were shifted by 3-4 mV toward positive potentials, whereas their slopes in ␣ 1C,II,IV and ␣ 1C,II,III,IV were less steep than in ␣ 1C,WT and other mutants. These data suggest that the combined double and triple mutations to threonine may slightly change the voltage dependence and cooperativity of the voltage sensors for inactivation in these channels.
In the three double mutants tested (␣ 1C,II,III , ␣ 1C,II,IV , and ␣ 1C,III,IV ), the sustained component of Ba 2ϩ current increased 2-fold compared with the single mutants, and accounted for 47-52% of the total current (Table II). An augmentation of the sustained current was due to both the fast and slow components of the current, which decreased about proportionally in the analyzed single and double mutants (I f /I s ϳ 2). In the case of ␣ 1C,II,III,IV , the fraction of the sustained current increased to ϳ70% of the total current ( Fig. 1A and Table II). It is important to note that the ratio of I f /I s ϳ 3 determined for the ␣ 1C,II,III,IV channel suggests a substantial additional decrease of the slower inactivating component.
The triple mutation did not appear to greatly affect the property of Ca 2ϩ -dependent inactivation (Fig. 2). First, the dominating (87.8 Ϯ 0.2%) fast component of the decay of the Ca 2ϩ current at V t ϭ ϩ10 mV ( f ϭ 16.0 Ϯ 1.2 ms, n ϭ 7; Fig.  2A) exhibited a 4.6-fold acceleration as compared with the Ba 2ϩ current (Table II) but had kinetics essentially similar to those through the wild-type channel ( f ϭ 15.3 Ϯ 1.5 ms; I f ϭ 95.1 Ϯ 0.5%; n ϭ 16). Second, we observed a U-shape voltage dependence of f (Fig. 2B) reflecting the relationship between the inactivation rate and the size of Ca 2ϩ current expected for Ca 2ϩ -induced inactivation (32). However, with Ca 2ϩ as a charge carrier, a sustained component of the current was also observed, but its amplitude (11.6 Ϯ 1.8%, n ϭ 5) was smaller than those measured with Ba 2ϩ current (Table II) and appreciably larger than that in the case of I Ca through ␣ 1C,WT (2.8 Ϯ 1.0%; n ϭ 16). One of the possible explanations for the significant residual sustained component of the Ca 2ϩ current in the triple mutant may be partial compensation of the impaired voltage-dependent inactivation in ␣ 1C,II,III,IV by the Ca 2ϩ -induced inactivation. Although the hydrophobic amino acid mutations were introduced in the presumed pore region of the channel (9, 10), the ratio of maximum Ba 2ϩ to Ca 2ϩ currents (ϳ2.7; n ϭ 3) through the ␣ 1C,II,III,IV channel did not change substantially compared with the wild-type channel, indicating that the ion selectivity was not appreciably affected.
Distinct Role of Repeat I Examined by Single and Combined Mutations-Segment IS6 appears to contribute to the voltagedependent inactivation differently as compared with other S6 segments. The inactivation properties of the L404T mutant ␣ 1C,IL are very similar to those of the wild-type channel (Table  II). However, incorporation of the L404T mutation into the ␣ 1C,II,III,IV channel significantly reversed the effect caused by the combined mutations to threonines in repeats II-IV leading to ␣ 1C,II,III,IV (cf. decays of Ba 2ϩ current through the ␣ 1C,II,III,IV channel in Fig. 1A and those through the ␣ 1C,IL-IV channel in Fig. 3A). The sustained component of the Ba 2ϩ current through the ␣ 1C,IL-IV channel was reduced 2.3-fold. Both fast and slow components of the decay increased in size, and the faster inactivating component of the Ba 2ϩ current was accelerated (Table  II). Steady-state inactivation curves for ␣ 1C,IL and ␣ 1C,IL-IV were shifted by 7 and 13 mV toward negative voltages, and their slopes were steeper compared with ␣ 1C,WT and ␣ 1C,II,III,IV , respectively (Fig. 3B). In the case of the ␣ 1C,IL-IV channel, the half-maximal activation was also shifted by 12 mV toward negative potentials (Fig. 3C), but the current-voltage relation Kinetics of Ba 2ϩ current inactivation in ␣ 1C mutants Ba 2ϩ currents were elicited by 1-s test pulses a to ϩ10 mV from a holding potential of Ϫ90 mV. Inactivation time constants () were determined from the double-exponential fitting of the current decay by equation: I(t) ϭ I ϱ ϩ I f ϫ exp(Ϫt/ f ) ϩ I s ϫ exp(Ϫt /( s ), where I ϱ is the steady-state amplitude of the current, I is the amplitude of the initial current, and f and s stand for fast and slow components, respectively. Thus the sum I f ϩ I s represents the apparent inactivating component of the net current. I o is the sustained current component determined as the ratio of steady state to peak current amplitudes. for ␣ 1C,IL was not appreciably different from those for the other isoforms (Fig. 1B). Unlike transmembrane segments IIS6 -IVS6, segment IS6 contains a hydroxyl amino acid Ser-405 at the adjacent position Ϫ1 (Table I). The simultaneous double conversion L404T,S405I was introduced into the ␣ 1C,WT channel to create a microenvironment at the cytoplasmic end of segment IS6 analogous to those in the mutated S6 segments of ␣ 1C,II , ␣ 1C,III , or ␣ 1C,IV . The resulting double mutant ␣ 1C,ILS , however, was inactivated significantly faster than the wild-type channel (Table II). Strong acceleration of the Ba 2ϩ current decay was observed in the composed mutant ␣ 1C,ILS-IV (Fig. 3A, Table II) that exhibited two characteristic features: (a) a robust sustained component of the current that is a hallmark of the mutations to Thr at positions Ϫ2 of S6 in repeats II-IV and (b) an impressive acceleration of the time course of fast inactivation, which is a characteristic result of the S405I mutation. Voltage dependences of activation and inactivation of the ␣ 1C,ILS-IV channel were both shifted to negative potentials (Fig. 3, B and C). Despite acceleration of fast inactivation and impairment of slow inactivation, the voltage dependences of the time constants were not greatly changed (Fig. 3D).
The Combined Mutations Obstructing Slow Inactivation-The kinetic analysis revealed that the slow component of the Ba 2ϩ current inactivation was substantially reduced in the ␣ 1C,ILS-IV channel. It appears that Ser-405 is one of the major determinants of slow inactivation of Ca 2ϩ channel. Indeed, the S405I mutation alone greatly reduced the slow inactivation of Ba 2ϩ current through the ␣ 1C,IS channel and com-pletely inhibited it in the ␣ 1C,IS-IV channel (Table II). In fact, ϳ45% of the maximum Ba 2ϩ current through the ␣ 1C,IS-IV channel showed almost no decay during the 30-s test pulse (Fig. 4A) indicating that a conducting state is the major favorable stable state of this channel at lasting depolarizations. Up to 55% of the Ba 2ϩ current through the ␣ 1C,IS-IV channel became inactivated with a single-exponential decay characterized by a time constant of 10.8 Ϯ 0.4 ms, which is in fact somewhat faster than the Ca 2ϩ current through the wild-type channel ( f ϭ 18 Ϯ 2 ms, n ϭ 3). Fig. 4B shows the traces of Ba 2ϩ current through the ␣ 1C,IS-IV channel recorded at different voltages. Similar to other S6 mutants tested, the time constant of the faster inactivation did not change more than 1.6-fold with membrane potential (Fig. 4C, f). The current-voltage relationships measured for the sustained and inactivating components of the Ba 2ϩ current (Fig. 4C) were very similar to those for the wild-type channel (cf. Fig. 1B, ⅷ). As expected, the voltage dependence of inactivation of the sustained component of the Ba 2ϩ current was completely inhibited in the ␣ 1C,IS-IV mutant (Fig. 4D, ⅜). The analysis of the voltage dependence of inactivation of the rapidly inactivating Ba 2ϩ current component showed that up to 60% of the ␣ 1C,IS-IV channels remained available at positive potentials (Fig. 4D, ⅷ). The steady-state inactivation curve for the inactivating component of the Ba 2ϩ current through the ␣ 1C,IS-IV channel (V 0.5 ϭ 27.7 Ϯ 1.5 mV, n ϭ 7) showed a negative shift characteristic for the other tested IS6 mutations incorporated into the ␣ 1C,II,III,IV channel.
Taken together, these data are consistent with the idea that FIG. 1. Electrophysiological properties of the mutated ␣ 1C channels. A, representative traces of Ba 2ϩ current through the wild-type ␣ 1C,WT and mutated ␣ 1C,IV , ␣ 1C,III,IV , and ␣ 1C,II,III,IV channels recorded at ϩ10 mV and scaled to the same amplitude. B, averaged currentvoltage relationships for Ba 2ϩ current through the wild-type and mutated channels. Currents were measured in response to 1-s test pulses from V h ϭ Ϫ90 mV applied at 30-s intervals. Peak current amplitudes were normalized to the maximum current. Smooth lines represent fits of the averaged data with equation, where G max is the maximum conductance, E rev is the reversal potential, V 0.5 is the voltage at 50% of I Ba activation, and k I-V is the slope factor. C, averaged steadystate inactivation curves for Ba 2ϩ current through the wild-type and mutated channels. Conditioning pulses (2 s) were applied at 30-s intervals from V h ϭ Ϫ90 mV followed by a 250-ms test pulse to ϩ10 mV. Peak current amplitudes were normalized to the maximum value. slow inactivation in ␣ 1C,IS-IV is completely inhibited, and the respective long-lasting conducting state is preceded by the rapidly inactivating one. We would predict that the recovery of this fast transient fraction from inactivation should also be markedly accelerated. Recovery from inactivation was measured as a time dependence of the ratio of maximum Ba 2ϩ currents elicited by two consecutive test pulses. We have found (Fig. 5) that recovery was substantially accelerated by the introduction of single-or double-threonine mutations in the S6 segments of repeats II-IV. This effect, however, did not progress with additional mutations leading to ␣ 1C,II,III,IV , ␣ 1C,IL-IV , and ␣ 1C,ILS-IV channels as compared with the double mutants. In sharp contrast, the recovery of the ␣ 1C,IS-IV channel from inactivation was complete after 100 ms. Thus, the inhibition of slow inactivation in ␣ 1C,IS-IV correlated with the observed elimination of multiple slow phases of recovery that is characteristic for the ␣ 1C,WT channel (24).
Among the 14 mutants tested, the apparent values of the slower inactivation time constants, s , measured at the peak of current-voltage relationships vary ϳ2.2-fold, whereas the faster inactivation time constants, f , vary over 8-fold (Table  II). In fact, the faster inactivation was accelerated with the reduction of the apparent fraction of the slower inactivating component of the Ba 2ϩ current decay. This dependence becomes evident when f is plotted as a function of I s /(I f ϩ I s ) and fitted by linear regression with r ϭ 0.926 (Fig. 6).
We have noted that this empirical relation is true also for the Ba 2ϩ current through the ␣ 1C,86 channel isoform with disrupted Ca 2ϩ sensors mediating Ca 2ϩ -dependent inactivation (24,29) (Fig. 6, inset). Inactivation of the Ba 2ϩ current through ␣ 1C, 86 is strongly accelerated ( f ϭ 23.5 Ϯ 5.6 ms), whereas the slow component of inactivation accounts for only a small fraction of the decay (I s ϭ 7.2 Ϯ 4.7%; I o ϭ 0 (n ϭ 4)). These data suggest that the slow and fast mechanisms of voltage-dependent inactivation of Ca 2ϩ channels are not entirely independent. Both Ca 2ϩ sensors of the C-terminal tail and the outlined annular determinant appear to be important, possibly in cooperative manner, for the voltage-dependent inactivation.
Ca 2ϩ -induced Inactivation Property Is Missing in the ␣ 1C,IS-IV Channel-It is interesting that Ca 2ϩ current through the ␣ 1C,IS-IV channel exhibited inactivation with a single time constant ( f ϭ 9.5 Ϯ 0.7 ms; n ϭ 3) very close to those found for the Ba 2ϩ current (Fig. 7A). A characteristic 3-fold difference between the peak Ba 2ϩ and Ca 2ϩ currents (29,33) was preserved in the ␣ 1C,IS-IV channel indicating that ion selectivity in this mutant was not altered. Fig. 7 (B and C) shows dependence of the Ca 2ϩ currents through the ␣ 1C,IS-IV channel on membrane potential. The sustained component of Ca 2ϩ current was observed over a wide range of membrane potentials and comprised up to 49% of the total current (Fig. 7B).
The Ca 2ϩ -induced inactivation of the ␣ 1C channels is supported by two calmodulin-binding domains located in the Cterminal tail (34 -36). These Ca 2ϩ sensors remain intact in the ␣ 1C,IS-IV channel, but their function mediating Ca 2ϩ -dependent inactivation appears to be lost. Indeed, simultaneous co-expression of either wild-type calmodulin (CaM WT ) 1 or its Ca 2ϩ -insensitive mutant CaM 1234 (35,37) did not appreciably affect the properties of Ba 2ϩ or Ca 2ϩ currents through the ␣ 1C,IS-IV channel. Even when CaM WT was overexpressed in oocytes, the voltage dependence of the time constant of inactivation of the Ca 2ϩ current (Fig. 7C, ⅜) did not show a U-shape characteristic for the Ca 2ϩ -induced inactivation property (cf., Fig. 2B). Similar observations were made for the ␣ 1C,IS-IV channel co-expressed with CaM 1234 (data not shown). Thus it appears that the ␣ 1C,IS-IV channel lacks the property of Ca 2ϩ -dependent inactivation.
Amino Acid Requirements for the Determinant of Slow Inactivation in S6 Segments of Repeats II-IV-Initial selection of sites for the mutation analysis described in this report was based on the observation that the A752T mutation impaired inactivation in the fibroblast ␣ 1C, 94 channel isoform (25). Having established an annular nature of the determinant of voltagedependent slow inactivation as well as the critical amino acids in the IS6 region, we also evaluated the role of the IIIS6 segment that contribute to inactivation similar to the segments IIS6 and IVS6. To examine the importance of the amino acid in position 1165 of IIIS6, the hydrophilic Thr-1165 residue in the ␣ 1C,IS-IV channel was changed to a charged negative (Asp), positive (Lys), or bulky hydrophobic (Phe) amino acid. Each of these mutations partially restored the slow inactivation of the channel, as can be seen from the comparison of the fractions of the residual Ba 2ϩ current sampled at 200 and 1000 ms (Fig. 8). Thus it appears that the voltage-dependent slow inactivation of the Ca 2ϩ channel critically depends on hydrophobicity rather than charge or size of amino acid residues in the Ϫ2 position of transmembrane segments S6.
We have also investigated whether mutation of the neighbor hydrophobic amino acids on both sides of the Ϫ2 position of 1 The abbreviations used are: CaM WT , wild-type calmodulin; CaM 1234 , Ca 2ϩ -insensitive calmodulin mutant. transmembrane segment S6 affects voltage-dependent inactivation. To examine this, the conversion of the hydrophobic to a hydrophilic amino acid (Ser or Thr) was shifted from position 1165 in the ␣ 1C,IS-IV channel to the neighbor amino acids of IIIS6 indicated in Table I. Each of the shifted upward mutations F1164S, G1163, or V1162T have partially restored the slow inactivation of the channel (Fig. 8) and slowed down the recovery from inactivation (not shown). These results are compatible with the view that amino acids in the tested positions are not particularly critical for the slow inactivation. However, the I(1166)S conversion in the position Ϫ1 was almost as effective as the mutations in the position Ϫ2, indicating that hydrophobic amino acids at this location of the putative cytoplasmic end of S6 are important for slow inactivation. These data support our initial conclusion on the annular nature of the determinant for slow inactivation that may involve amino acid residues in positions Ϫ2 and Ϫ1.

DISCUSSION
In this study, we have dissected the fast and slow mechanisms of ␣ 1C L-type Ca 2ϩ channel inactivation by introducing a number of mutations that selectively and completely inhibited slow inactivation. Guided by the inactivation-impairing A752T mutation in segment IIS6 (25), we probed amino acid residues situated in the analogous positions of the S6 segments in all four major repeats of the pore-forming ␣ 1C subunit. Molecular interventions leading to complete inhibition of slow inactivation require simultaneous substitution to threonines of Ala-752 in IIS6, Val-1165 in IIIS6, and Ile-1475 in IVS6, each causing an additional increase in the fraction of non-inactivating Ba 2ϩ current. In addition, it requires a S405I mutation in segment IS6 creating a ring of hydrophobic amino acids in the Ϫ1 positions of the S6 segments. The critical role of hydrophobic amino acids in position Ϫ1 was also revealed by the mutation I(1166)S in IIIS6 (Fig. 8). We have found that repeat I contributes to inactivation somewhat differently when compared with repeats II-IV. To explain these differences, one would need data on the molecular architecture of the pore region. However, the fact that all four repeats contribute to the outlined determinant conforms to their arrangement around the central pore and suggests that slow inactivation is supported by an "annular" structure involving critical amino acids in positions Ϫ2 and Ϫ1 of the S6 segments.
Properties of the S6 mutants studied in our work are consistent with the C-type inactivation model (3,8). Similar to C-type inactivation, the rate of inactivation of the S6 mutants is essentially voltage-independent. Therefore, S6-mediated inactivation does not involve a large membrane charge movement, which suggests constriction of the pore as the main mechanism. We speculate that the residues of the annular determinant are oriented away from the ion-conducting pore and play a critical role in stabilizing hydrophobic interactions supporting slow inactivation (38). Mutations to hydrophilic Thr or Ser residues appear to destabilize these interactions so that the conducting state becomes favorable compared with the inactivated state.
Approximately 55% of the Ba 2ϩ current through the ␣ 1C,IS-IV FIG. 4. Properties of the ␣ 1C, IS-IV channel lacking slow inactivation. A, typical trace of Ba 2ϩ current through the ␣ 1C,IS-IV channel induced by 30-s depolarization to ϩ10 mV from a holding potential of Ϫ90 mV. B, Ba 2ϩ currents through the ␣ 1C, IS-IV channel evoked by 1-s depolarizations to Ϫ20 with 10-mV increments to ϩ30 mV from a holding potential of Ϫ90 mV. C, current-voltage relationship for the peak (ⅷ; n ϭ 8) and sustained (⅜; n ϭ 7) Ba 2ϩ currents through ␣ 1C,IS-IV as well as voltage dependence of f (f; n ϭ 4). D, steady-state inactivation curves for the peak (ⅷ) and sustained (⅜) Ba 2ϩ currents through the ␣ 1C,IS-IV channel (n ϭ 6).
FIG. 5. Comparison of the recovery from inactivation of the ␣ 1C, IS-IV channel with the wild-type channel and its mutants. 1-s prepulses to ϩ10 mV were applied from V h ϭ Ϫ90 mV followed by increasing recovery intervals at Ϫ90 mV before the application of a test pulse to ϩ10 mV. Fractional recovery was measured as a ratio of maximum Ba 2ϩ currents evoked by the test and prepulse depolarizations, and plotted as a function of interpulse interval. Data are shown for the ␣ 1C,WT (ⅷ), ␣ 1C,II (⅜), ␣ 1C,III,IV (f), ␣ 1C,II,III,IV (Ⅺ), ␣ 1C,ILS-IV (‚), and ␣ 1C,IS-IV (OE) channels. channel inactivates as a single-exponential decay, characterized by the time constant of ϳ11 ms. The steady-state inactivation curve of this current is shifted to negative voltages indicating that the voltage sensors for fast inactivation operate at lower depolarization levels. However, neither the time course of activation nor its voltage dependence was significantly affected by inhibition of the slow inactivation mechanism. Single-channel studies are required to further clarify the microscopic properties and determine the relationship between rapidly inactivating and non-inactivating conducting states of the ␣ 1C,IS-IV channel. Although a molecular determinant for the fast inactivation has not been identified yet, ␣ 1C,IS-IV may be a valuable channel isoform in such investigation.
It appears that the fast and slow inactivation mechanisms are linked. The fractional inhibition of slow inactivation component in the tested mutants caused an acceleration of the fast inactivation (Fig. 6). A similar effect (see asterisk in Fig. 6) was described earlier (29) for the ␣ 1C,86 isoform deprived of Ca 2ϩinduced inactivation by the mutation of Ca 2ϩ sensors in the cytoplasmic C-terminal tail. Although Ca 2ϩ sensors (34 -36) remain intact in the ␣ 1C,IS-IV channel, no acceleration of inactivation was observed when Ba 2ϩ was replaced by Ca 2ϩ as the charge carrier (Fig. 7). This result may suggest that Ca 2ϩinduced inactivation of L-type Ca 2ϩ channels predominantly targets the slow mechanism of inactivation mediated by S6 segments and obstructed in the ␣ 1C,IS-IV channel. Several observations support this hypothesis. The pore region, which is thought to be formed on the cytoplasmic side by the S6 segments (9, 10), can in fact be reached by the distant C-terminal Ca 2ϩ sensors. Indeed, the segmental mutations disrupting the Ca 2ϩ sensors were found to accelerate voltage-dependent inactivation (24,29) possibly via the loss of critical calmodulin interaction (39). These mutations affected the unitary conductance (40) quite likely via the annular determinant in a manner similar to those linking C-type inactivation to the Shaker K ϩ channel ion selectivity (3). Taken together, our data are consistent with the idea that the voltage-dependent slow inactivation mechanism involves the targeting of the annular determinant characterized here by the Ca 2ϩ sensors of inactivation. Additional studies revealing the role of the individual Ca 2ϩ sensors as well as of the ␤ subunit in this mechanism are in progress.