A Region in IVS5 of the Human Cardiac L-type Calcium Channel Is Required for the Use-dependent Block by Phenylalkylamines and Benzothiazepines*

Mutations in motif IVS5 and IVS6 of the human cardiac calcium channel were made using homologous residues from the rat brain sodium channel 2a. [3H]PN200-110 and allosteric binding assays revealed that the dihydropyridine and benzothiazepine receptor sites maintained normal coupling in the chimeric mutant channels. Whole cell voltage clamp recording fromXenopus oocytes showed a dramatically slowed inactivation and a complete loss of use-dependent block for mutations in the cytoplasmic connecting link to IVS5 (HHT-5371) and in IVS5 transmembrane segment (HHT-5411) with both diltiazem and verapamil. However, the use-dependent block by isradipine was retained by these two mutants. For mutants HHT-5411 and HHT-5371, the residual current appeared associated with a loss of voltage dependence in the rate of inactivation indicating a destabilization of the inactivated state. Furthermore, both HHT-5371 and -5411 recovered from inactivation significantly faster after drug block than that of the wild type channel. Our data demonstrate that accelerated recovery of HHT-5371 and HHT-5411 decreased accumulation of these channels in inactivation during pulse trains and suggest a close link between inactivation gating of the channel and use-dependent block by phenylalkylamines and benzothiazepines and provide evidence of a role for the transmembrane and cytoplasmic regions of IVS5 in the use-dependent block by diltiazem and verapamil.

Mutations in motif IVS5 and IVS6 of the human cardiac calcium channel were made using homologous residues from the rat brain sodium channel 2a. [ 3 H]PN200-110 and allosteric binding assays revealed that the dihydropyridine and benzothiazepine receptor sites maintained normal coupling in the chimeric mutant channels. Whole cell voltage clamp recording from Xenopus oocytes showed a dramatically slowed inactivation and a complete loss of use-dependent block for mutations in the cytoplasmic connecting link to IVS5 (HHT-5371) and in IVS5 transmembrane segment (HHT-5411) with both diltiazem and verapamil. However, the usedependent block by isradipine was retained by these two mutants. For mutants HHT-5411 and HHT-5371, the residual current appeared associated with a loss of voltage dependence in the rate of inactivation indicating a destabilization of the inactivated state. Furthermore, both HHT-5371 and -5411 recovered from inactivation significantly faster after drug block than that of the wild type channel. Our data demonstrate that accelerated recovery of HHT-5371 and HHT-5411 decreased accumulation of these channels in inactivation during pulse trains and suggest a close link between inactivation gating of the channel and use-dependent block by phenylalkylamines and benzothiazepines and provide evidence of a role for the transmembrane and cytoplasmic regions of IVS5 in the use-dependent block by diltiazem and verapamil.
Voltage-dependent calcium channels mediate calcium influx in response to membrane depolarization and play a critical role in cellular activities such as excitation-contraction coupling, neurotransmitter release, and hormone secretion (1,2). L-type voltage-dependent calcium channels are characterized pharmacologically by high sensitivity to the calcium blocking drugs that include dihydropyridines (DHP), 1 phenylalkylamines (PAA), and benzothiazepines (BZT) (3,4). Photolabeling stud-ies of the purified ␣ 1 subunit and identification of the proteolyzed fragments by site-directed antibodies demonstrate that the ␣ 1 subunit harbors all three classes of binding sites (3). The results from these studies provide substantial evidence that motifs IIIS6 and IVS6 contain the major binding regions for these drugs.
Less detailed structural information is known regarding the mechanism of use-dependent block, a feature that is critical to the therapeutically useful calcium antagonists. These drugs preferentially inhibit channels during high electric activity as observed in certain cardiac arrhythmias and diseases such as ischemic heart disease. During repetitive depolarizations, block by PAA and BZT accumulates in a frequency-and voltagedependent manner as a result of additional drug binding during each depolarization that fails to dissociate at the resting potential between depolarizations.
Recent studies have demonstrated the role of inactivation in the mechanism of use-dependent calcium block by phenylalkylamines (21). These results have established that Phe 1164 and Val 1165 contribute to the BZT-binding site resulting in a decrease in block by diltiazem (22). The Ba 2ϩ currents from these channels also exhibited a slowed inactivation rate when these residues were replaced by alanine (22). However, according to the guarded receptor hypothesis (23), the molecular elements responsible for this effect may be distinct from the structures that specify simple or tonic binding.
In order to determine the structural features of the calcium channel responsible for use-dependent block, we introduced mutations in motif IVS5 of the human heart ␣ 1C calcium channel utilizing homologous regions of the rat brain sodium channel 2a. Two of these mutations lost the characteristic use-dependent block by PAA and BZT, whereas the use-dependent block by isradipine was fully retained. These results demonstrate a differential loss of use-dependent block with BZT and PAA but not for isradipine and implicate for the first time a role for motif IVS5 in this phenomenon.

EXPERIMENTAL PROCEDURES
Construction of ␣ 1C cDNA Clones in pBluescript-The wild type and mutant human ␣ 1C calcium channel cDNAs were first constructed in pBluescript for expression in Xenopus oocytes and then subsequently placed into pAGS-3 (24) either as the full-length channel or as a channel deleted at the carboxyl-terminal glycine 1633 (25)(26)(27). This cellular strategy was designed to utilize the higher expression level of the carboxyl-terminal truncated form of the channel. Site-directed mutations were introduced by incorporating a two-step polymerase chain reaction (PCR) protocol. Mutant oligonucleotides of IVS5 (except for HHT-5658) were synthesized in the sense direction as follows: HHT-5371, CTGTGACCTCACATGTCCTTGCCGGCCCTGTTCGATATGGG-CCTCCTGAG; HHT-5411, GTGGCCCTGCTCTTCCTGGTGATGTTC-ATCTACGCG; and HHT-5432, TGTTCTTCACTACCGTCTTCGGGAT-GTCGAACTTTGCGTACATTGCCCTGAATG (the mutations are underlined). In the first reaction, mutant primers were amplified using an antisense oligonucleotide that harbors the BclI restriction site (GGTT-GATGATCAGGAAGGCAC) and the wild type ␣ 1C as the template. The resultant 300-base pair PCR product was isolated and purified for use in the second round of PCR. In this second step, the antisense oligonucleotide carrying the mutant sequence (mega-primer) was extended using a forward primer harboring the AatII restriction site (GCATA-ATTGACGTCATTCTCA). Each PCR reaction proceeded for 20 -30 cycles in which the primers were denatured at 94°C for 1 min followed by annealing at 55°C for 30 s and a final extension at 72°C for 30 s. The last cycle consisted of a final 10-min extension at 72°C. After the second round of PCR, the newly synthesized 500-base pair oligonucleotide containing the mutant sequence and harboring both the AatII and BclI restriction sites was gel-purified and subcloned into the EcoRV site of pBluescript. In the construction of HHT-5658, the antisense oligonucleotide GATGATCAGGAAGAACTATGATGACTGACAAGAAGAAGA-TACCAACGCTGCTACCACGGTG that contains the BclI site was used with a sense oligonucleotide primer that harbors the BstbI restriction site (CCACCTTCGAAGGGTGGC) for the first step in the PCR reaction. The 1000 base pair product was then isolated and subcloned into pBluescript as described above. The restriction sites and mutations were sequence-verified by the dideoxynucleotide chain termination method.
The wild type human ␣ 1C channel was prepared by linearization with BclI followed by partial digestion with AatII in which the 8.5-kilobase fragment was isolated. The sequence-verified IVS5 mutants were digested first with BclI followed by AatII to completion. This mutant oligonucleotide was ligated into the partially digested wild type ␣ 1C channel. The full-length channel containing the mutant sequence was again sequence-verified. Conserved residues are boxed and amino acids important for calcium channel blocker binding are boldface. A, comparison of amino acid sequence between the ␣ 1C human cardiac calcium channel and rat sodium channel 2a of regions (IIIS5, IIIS6, and IVS6) known to harbor residues critical for high affinity calcium antagonist binding. B, sequence alignment of the mutant calcium channels (HHT-5371, HHT-5411, HHT-5432, and HHT-5658) with the wild type (WT) human heart ␣ 1C sequence. Mutations were based on non-conserved residues from the rat brain sodium channel 2a. Intervening amino acid residues among the three IVS5 mutations are conserved between the two channels. C, relative orientation of the mutant and wild type amino acids within transmembrane regions IVS5 and IVS6.
Ligation of IVS5 Mutants into pAGS-3-Both the mutant ␣ 1C channel and the full-length channel in pAGS-3 (24) were digested to completion with SalI and HindIII. A 4.8-kilobase band of pAGS-3 was ligated to the 6-kilobase band of the mutant ␣ 1C channel.
Ligation of IVS5 Mutants into pAGS-3 as a ⌬1633 Carboxyl-terminal Deletion-We previously constructed a carboxyl-terminal deletion after glycine 1633 (27). In order to ligate the mutations into the truncated channel, the ⌬1633 pAGS-3 was digested to completion using HindIII and Sse8387I, and the 4.8-kilobase fragment was isolated. The mutant channels in pBluescript were also digested to completion using HindIII and XbaI. The product from this digestion was partially digested using Sse8387I, and the resultant 4.8-kilobase band was isolated. The final step was the ligation of the two 4.8-kilobase bands to complete the ⌬1633 pAGS-3 incorporating the IVS5 mutations.
Expression of Calcium Channels in Xenopus Oocytes-Expression of the wild type and mutant calcium channels was done as described previously (26). In vitro synthesized cRNA was made using the mMessage mMachine synthesis kit (Ambion). Xenopus oocyte isolation and cRNA injection was performed as published elsewhere (11). Briefly, female Xenopus laevis (purchased from Xenopus I, Ann Arbor, MI) frogs were anesthetized by exposing them for 15-20 min to 0.15% methanesulfonate salt of 3-aminobenzoic acid ethyl ester (MS-222; Sigma) solution before pieces of the ovary were removed. The follicular layers of the oocytes were digested with 2.0 mg/ml collagenase (Type IA; Sigma) dissolved in OR-2 medium (in mM): 82.5 NaCl, 1 KCl, 1 MgCl 2 , 5 HEPES, pH 7.5. Stage V-VI oocytes were incubated at 19°C in P/S medium (in mM): 96 NaCl, 2.0 KCl, 1.0 MgCl 2 , 1.8 CaCl 2 , 5 HEPES, 2.5 sodium pyruvate, and 0.5 theophylline at pH 7.5. The P/S medium was supplemented with 100 units/ml penicillin and 100 g/ml streptomycin.
The recording medium was a Ca 2ϩ -and Cl Ϫ -free solution composed of the following (in mM): 40 Ba(OH) 2 , 50 N-methyl-D-glucamine, 2 KOH, 5 HEPES, pH adjusted to 7.4 with methanesulfonic acid. Voltage and current electrodes were filled with 3 M KCl and had a resistance of 0.5-1.5 megohms. Currents were recorded using an Axoclamp-2A (Axon Instruments Inc., Foster City, CA) amplifier. Whole cell leakage and capacitive currents were subtracted on-line using the P/4 procedure. Currents were digitized at 1 kHz after being filtered at 1 kHz. The pClamp software (version 5.5 Axon Instruments) was used for data acquisition, and version 6.0.3 was used for analysis.
Ba 2ϩ currents were elicited by a 350-ms-long depolarizing pulse from a holding potential of Ϫ80 mV to test potentials between Ϫ30 mV and ϩ40 mV in 10-mV increments in order to determine the peak potentials of the current voltage relationship of the Ca 2ϩ channel construct. Usedependent block was determined as the inhibition of peak I Ba during trains of 15 test pulses of 80-ms duration applied at 0.5 Hz from a holding potential of Ϫ60 mV to test potentials ϩ10, ϩ20, or/and ϩ30 mV positive to the peak potential of the I-V curves. Identical pulse protocols were used in the presence of drugs. Drugs were perfused in the bath (for a 2.5-min period) at concentrations of 300 and 100 M diltiazem (racemic) and verapamil, respectively. These concentrations were selected since they provided sufficient block for the use-dependent measurements in Xenopus oocytes. Tonic block is defined as the peak I Ba inhibition during the first pulse after a 2.5-min equilibration in the drug containing solution at Ϫ60 mV. The drug block under these conditions (tonic block) reflects steady-state binding to the mixture of closed and closed/inactivated states present at Ϫ60 mV. A double-pulse protocol was used in those experiments measuring the voltage dependence of steady-state inactivation. The protocol consisted of a 5-s prepulse that ranged from Ϫ80 to ϩ50 mV at a holding potential of Ϫ80 mV followed by a 30-ms return pulse to Ϫ80 mV. Finally a 400-ms-long test pulse was applied to ϩ10, ϩ20, and/or ϩ30 mV. The curves were fit to the Boltzmann equation Recovery of I Ba from inactivation was studied after depolarizing the calcium channels during a 3-s prepulse to ϩ10, ϩ20, and/or 30 mV. The time course of I Ba recovery from inactivation was estimated at a holding potential of Ϫ80 mV by applying a 400-ms test pulse to ϩ10, ϩ20, and/or ϩ30 mV at various times after the conditioning prepulse. Peak I Ba values were normalized to the peak current amplitude measured during the prepulse. After the double-pulse protocol, the membrane was hyperpolarized to Ϫ100 mV for 3 min to permit complete recovery from inactivation and block.
Transient Expression in Human Embryonic Kidney (HEK) 293 Cells-Each large 15-cm plate of (90%) confluent HEK 293 cells was transfected with a total of 60 g of cDNA. The ␣ 1C or ⌬1633 (wild type or mutant) cDNA were coinjected with ␣ 2 ␦ (27) and human ␤ 3 (26) in a 1:2:2 molar ratio. All cDNAs were in pAGS-3. Transfection was done according to the calcium phosphate precipitation method by Chen and Okayama (31). Cells were scraped into phosphate-buffered saline, pH Positive allosteric binding assays between labeled DHP and BZTs were done at 37°C for 1 h. In these experiments, varying concentrations of diltiazem hydrochloride and 0.8 nM [ 3 H]PN200-110 were used. The allosteric effect was determined by dividing the specifically bound radioactivity in the presence of diltiazem by that specifically bound in the absence of diltiazem multiplied by 100 for each construct.
and presented as the means Ϯ S.E. The statistical analysis was done using Student's t test (p Ͻ 0.05).

RESULTS
The regions of the ␣ 1C calcium channel responsible for high affinity binding of PAA, DHP, and BZT include specific amino acids in IIIS5, IIIS6, and IVS6. However, segments and amino acids responsible for the use-dependent nature of these drugs are found dispersed around the presumed area of calcium blocker-binding sites (20 -22, 33). This provided the impetus to investigate other regions of the calcium channel that may play an important role in use-dependent block. We chose amino acid stretches of the rat brain sodium channel 2a (34) since these channels possess a low affinity pattern of inhibition by some calcium antagonists (35,36). Furthermore, very few amino FIG. 4. Effect of diltiazem inhibition on steady-state inactivation curves. The voltage dependence of inactivation was investigated for the wild type, HHT-5411, HHT-5432, HHT-5371, and HHT-5658 channels at the end of a 5-s prepulse. Open symbols, control; filled symbols in the presence of 300 M diltiazem. A, wild type channel; B, HHT-5432; C, HHT-5411; D, HHT-5371; and E, HHT-5658. The relative peak current at ϩ10 mV or ϩ20 mV was plotted against the different prepulse potentials. Curve fitting was performed by the Boltzmann equation as described under "Experimental Procedures" (n ϭ 5-9).

TABLE I Activation and inactivation rate constants of wild type and mutant calcium channels
The inactivation time constants for V m ϭ ϩ10 or ϩ20 mV were estimated at t ϭ 3 s for wild type and mutant channels in the presence of 40 mM Ba 2ϩ . The inactivation time course of whole cell Ba 2ϩ traces could be fitted well by a single exponential function. The abbreviations used in the table are as follows: V 0.5 , half-maximal voltage for activation; V 0.5 inact., half-maximal voltage for steady-state inactivation; and k inact. , slope factor of the curve at V 0.5 inact. acids are conserved in the membrane spanning regions between the two channels (Fig. 1A), and the majority of the published residues that have been shown to be critical for the action of calcium antagonists are absent in the rat brain sodium channel (Fig. 1A). We investigated motif IVS5 since this segment "opposes" IVS6 much like the faces of IIIS5 and IIIS6 are believed to be oriented for DHP binding in the calcium channel. Three separate mutant constructs, termed HHT-5371, HHT-5411, and HHT 5432, were engineered from the intracellular portion (amino acids 1326) of motif IVS5 to the extracellular loop (amino acids 1355) that show a loss of amino acid conservation. In the three mutations there are a total of nine, five and four amino acids that were changed in HHT-5371, -5411, and -5432, respectively (Fig. 1B). The intervening residues between the three mutations in IVS5 are conserved in the two channels. In addition, in a motif IVS6 mutant (HHT-5658) 11 amino acids were changed to the corresponding sequence of the rat brain 2a sodium channel. In this construct, the first three of the four amino acids of the YMAI sequence that have been demonstrated to form the primary binding site for calcium antagonists have been changed. The relative location of these mutations within the transmembrane regions are shown in Fig. 1C. These mutations were then assayed for their influence on channel function by the different classes of calcium antagonists.
In a previous report from this laboratory, a carboxyl-terminal deletion of the channel tail after glycine 1633 (⌬1633) was found to increase the B max of the ␣ 1C in transient expression systems without affecting the K D (27). Thus, we used this carboxyl-terminal deletion construct in all binding assays of HEK 293 cell membranes to enhance expression. Scatchard analysis showed no difference in affinity (0.1-0.2 nM) but an approximately 2-fold increase (140 versus 250 fmol/mg protein) in B max between the full-length wild type and the ⌬1633 construct (data not shown, see Ref. 27). Some of the mutant constructs expressed poorly in HEK 293 cells (HHT-⌬5432) with minor effects on K D values. Interestingly, some of the mutations within the regions of motif IVS5 also demonstrated a lowered affinity (HHT-⌬5371, Ͼ1.2 nM and HHT-⌬5432 0.4 nM) to [ 3 H]PN200-110 ( Fig. 2A). The mutations in motif IVS5 encompass regions that do not overlap with the published binding sites for calcium channels. The construct HHT-⌬5658 completely lost high affinity binding to [ 3 H]PN200-110. Thus, an estimate for DHP affinity could not be determined. These results are consistent with the published literature that motif IVS6 contains the high affinity binding site for DHP binding (7).
We then posed a question to test whether these sites may have any involvement in establishing the allosteric interaction among calcium channel blockers. The concentration-dependent enhancement of subsaturating (0.8 nM) concentrations of [ 3 H]PN200-110 by increasing concentrations of diltiazem between the wild type and ⌬1633 channel is depicted in Fig. 2B. The maximal allosteric effect of diltiazem between the fulllength wild type channel, the ⌬1633 wild type channel, and the ⌬1633 mutants is shown in Fig. 2C. The full-length wild type construct showed significantly lower (185%) allosteric stimulation by diltiazem compared with the ⌬1633 (385%) construct. Maximal binding for the mutant constructs were not significantly different compared with the wild type channel producing a 300% increase in [ 3 H]PN200-110 binding than in the absence of diltiazem. The exception was with HHT-⌬5432 in which there was a significant decrease (240%) in binding compared with the ⌬1633 control. In all of the constructs the concentrationdependent peak occurred between 30 and 100 M diltiazem and elicited similar curves as the wild type constructs (data not shown). There was no observable specific DHP binding for the HHT-⌬5658 construct even in the presence of diltiazem. These results are consistent with the loss of the high affinity binding site for DHP.
In order to address whether the motif IVS5 mutant constructs maintained normal biophysical properties and pharmacology to calcium channel antagonists, we expressed cRNAs for the full-length channels in Xenopus oocytes. All of the mutants produced voltage-dependent currents with Ba 2ϩ serving as the charge carrier. HHT-5411 produced a substantial positive shift (29.3 Ϯ 0.7 mV), whereas HHT-5658 produced a negative (11.2 Ϯ 1.4 mV) shift in the voltage dependence of the I-V relationship compared with the wild type channel (23.3 Ϯ 1.9 mV). In the case for the constructs HHT-5371 and HHT-5432, there was no significant shift (Fig. 3). As was expected, HHT-5658 produced voltage-dependent Ba 2ϩ currents despite demonstrating no binding of [ 3 H]PN200-110 to transfected HEK 293 membrane preparations.
Analyses of voltage-dependent activation and inactivation for the mutants used in this study are summarized in Table I. Activation rates revealed small but significant differences between the mutants. The half-maximal voltage for activation significantly shifted toward negative potentials (V1 ⁄2 ϭ 3.3 mV) for HHT-5658, whereas HHT-5411 shifted toward positive potentials (V1 ⁄2 ϭ 19.5 mV) compared with the wild type channels (V1 ⁄2 ϭ 13.5 mV). There was no significant difference among the other mutations ( Fig. 4 and Table I). The inactivation time course of Ba 2ϩ traces could be fitted well by a single exponential function (Table I). HHT-5371, HHT-5411, and HHT-5432 mutant channels displayed a significantly slower inactivation rate (Ͼ3 s, Ͼ 3 s, and 1.6 Ϯ 0.08 s, respectively) than the wild type (1.1 Ϯ 0.04 s) channel. The steady-state inactivation properties of wild type and mutant channels exhibited drastic changes in inactivation of HHT-5411 and HHT-5371 (Fig. 4). HHT-5371 and HHT-5411 revealed incomplete inactivation only to approximately 50% of the normalized current. Upon addition of diltiazem the steady-state inactivation was restored to normal (Fig. 4). The V 0.5 values of steady-state inactivation at HHT-5371 and HHT-5411 with diltiazem were significantly different from the one obtained for the wild type; however, the slope factors appeared not significantly different from the wild type (Table I). The wild type and HHT-5658 (V 0.5 ; Ϫ7.6 mV, k v ; 11.9 mV and V 0.5 ; Ϫ12.3 mV, k v ; 11.5 mV, respectively) inactivated completely, and the midpoint voltages of the curves were FIG. 5. Tonic I Ba inhibition by 300 M diltiazem for wild type and motif IV mutants expressed in Xenopus oocytes. Tonic block was defined as the I Ba peak inhibition during the first pulse after a 2.5-min equilibration in the drug containing solution at Ϫ60 mV. Tonic block is presented in the form of peak current ratios I Diltiazem /I Control . A p value smaller than 0.05 is indicated by an asterisk (n ϭ 13-26).
characteristically leftward-shifted in the hyperpolarized direction (V 0.5 , Ϫ22.4 mV; V 0.5 , Ϫ17.9 mV) after diltiazem addition without a significant change in the steepness of the voltage dependence (k v , 9.0 and 10.4 mV). The absence of change either in the V 0.5 or in the slope factor of the steady-state inactivation curve for HHT-5658 is in good agreement with the partial loss of the BZT-binding site. Mutant HHT-5432 behaved very similarly to that of the wild type. This mutant inactivated completely either without or with diltiazem showing only modest changes in the steady-state inactivation parameters (Table I).
We further analyzed motif IVS5 and IVS6 mutant calcium channels with regard to tonic and use-dependent block by diltiazem and verapamil. Resting state-dependent (tonic) I Ba inhibition by diltiazem was 0.69 Ϯ 0.02 (n ϭ 30) in the wild type channel. Only the change in HHT-5432 and HHT-5371 pro-duced a significant increase in tonic block (0.56 Ϯ 0.03, n ϭ 13 and 0.61 Ϯ 0.03, n ϭ 16, respectively). The other mutant calcium channel constructs were not significantly different (HHT-5411, 0.71 Ϯ 0.03, n ϭ 26; HHT-5658, 0.62 Ϯ 0.04, n ϭ 17) from the wild type channel (Fig. 5). Tonic block was less pronounced with verapamil for the wild type channel (0.92 Ϯ 0.04) suggesting that the drug binds poorly to the resting state of the channel. Similarly to the wild type channel, HHT-5411 (0.86 Ϯ 0.03) and HHT-5371 (0.75 Ϯ 0.07) exhibited very little tonic block with verapamil.
We then studied the impact of the amino acid substitutions on use-or frequency-dependent block by calcium antagonists. Representative current traces are shown for diltiazem and verapamil ( Fig. 6 and 7, respectively). Use-dependent block in the wild type channel appeared more efficient with verapamil (30.4 Ϯ 2.1%) than with diltiazem (20.0 Ϯ 2.1%). The IVS6 mutant HHT-5658 displayed less use-dependent block (12.6 Ϯ 1.5%) by diltiazem compared with wild type. The IVS5 mutant HHT-5432 exhibited many similar characteristics to the wild type channel.
A remarkable and unexpected observation was the complete disappearance of use-dependent block of both mutant HHT-5371 and HHT-5411, for either diltiazem (Fig. 6C, 0.9 Ϯ 2.4 and 3.1 Ϯ 3.1%, respectively) or verapamil (Fig. 7, E and G) (3.0 Ϯ 0.7 and 5.8 Ϯ 3.3%, respectively). Previously, it has been demonstrated that DHPs also exert use-dependent block at pulse frequencies of 1 Hz or greater (37). By using the same protocol as described above for BZT, we observed use-dependent block for 10 M isradipine at a frequency of 1 Hz for both the wild type channel (Fig. 7D) and the mutants HHT-5411 and HHT-5371 (Fig. 7, F and H, respectively). Since the mutants tested throughout these studies retained the DHP-and BZTbinding sites (with the exception of the HHT-5658 where the YMA part of the binding site was eliminated), it is logical to assume that the alteration of inactivation properties of the mutant channels is responsible for the loss of use-dependent block by diltiazem and verapamil.
A more complete analysis of the role of channel inactivation in development of block was performed by studying the recovery from depolarization-dependent block (Fig. 8). We measured the I Ba recovery from inactivation by a two-pulse protocol after maintained 3-s depolarizations. The recovery of the channel from inactivation was estimated at different intervals (recovery interval) from 20 ms to 28 s. Data points in the presence of diltiazem (Fig. 8A) were fitted by a two exponential curve with exception of HHT-5371 in which, due to the very fast recovery, a single exponential fit was employed. The mutant HHT-5371 and HHT-5411 displayed characteristically faster recovery kinetics compared with the wild type in the presence of diltiazem (Table II). The recovery from inactivation also was tested for wild type and HHT-5411 in presence of verapamil (Fig. 8B). Upon the addition of verapamil, the time constants of recovery from inactivation were 0.57 and 0.18 s ( fast ) and of 3.1 and 2.1 s ( slow ) for the wild type and the mutant HHT-5411, respectively. In HHT-5371 and HHT-5411 mutants drastic changes in both fast and slow were noted, showing that the drug still binds to the channel and that BZT-and PAA-binding sites are fully retained. Both mutations have a reduced rate of inactivation and an enhanced rate of recovery from inactivation at Ϫ80 mV suggesting that the inactivated state of the calcium channel is destabilized by these two mutations. Therefore, the loss of use dependence is largely due to the decrease of the affinity of the drug to the inactivated state of this mutant compared with the wild type.
We further investigated whether this phenomenon holds true with isradipine. In the presence of isradipine, the time course of recovery from inactivation for HHT-5411 did not appear statistically different from the wild type. The I Ba recovery from inactivation of wild type and the mutants HHT-5411 and HHT-5371 in the presence of isradipine are depicted in  Table II. The time course of recovery from inactivation at Ϫ80 mV was determined by subsequent test pulses applied at various times (recovery interval) after the prepulse as described under "Experimental Procedures." In the mutant channels (HHT-5411 and HHT-5371) about 80 -95% of the I Ba recovered within 20 ms.   9A. The recovery from inactivation for HHT-5371 after isradipine treatment could only be fit using a one exponential function. These results are similar to those observed from recovery in the presence of diltiazem and verapamil (Fig. 8) for both HHT-5411 and HHT-5371. Mutant HHT-5371 I Ba recovery appeared to be faster than wild type; however, it should be noted that even after a 5-s conditioning prepulse only 10% of the channels are inactivated (Fig. 9). For HHT-5411 about 30% of the channels were inactivated at the end of the conditioning prepulse. Therefore, determining the time constants for this latter process seemed not to be realistic (Table II). Representative current traces are illustrated in the absence (Fig. 9, B, D, and F) and presence of isradipine ( Fig. 9, C, E, and G) for wild type, HHT-5411, and HHT-5371, respectively. Mean time constants of I Ba recovery from inactivation by diltiazem, isradipine, and verapamil are summarized in Table II. DISCUSSION There is little information on the mechanism of use-dependent block, a feature that is critical to the therapeutically useful calcium antagonists. These drugs preferentially inhibit chan-nels during high electric activity as found in certain cardiac arrhythmias and diseases such as ischemic heart disease. According to the guarded receptor hypothesis (23), the molecular elements may be distinct from the structures that specify simple or tonic binding.
In this study we identified a region proximal to the cytoplasmic connecting link of motif IVS5 that contributes to the usedependent block of BZTs and PAAs. These results are consistent with those reported by Cai et al. (20), in which they demonstrated that a motif I-II replacement with ␣ 1A still maintained use-dependent inhibition by verapamil and diltiazem as well as in a mutant in which the high affinity YAI residues had been also mutated to that of the ␣ 1A channel. In more recent studies (3,21,22), other regions in motifs III and IV that are outside of the high affinity YAI-binding region have been shown to be critical for the use-dependent block observed for these drugs.
We have found that the loss of use-dependent block is closely associated with a decrease in voltage-dependent inactivation. In both HHT-5411 and HHT-5371 diminished inactivation re- FIG. 9. Effect of isradipine on the recovery from inactivation. A, I Ba recovery from inactivation of wild type, HHT-5411, and HHT-5371 mutant channels as measured by a two-pulse protocol in the presence of 200 nM isradipine. The currents were inactivated by a 3-s prepulse to ϩ10 or ϩ20 mV. Recovery from inactivation at Ϫ80 mV was measured by applying a sequence of test pulses at various intervals (between 20 ms and 28 s) after the prepulse. After a two-pulse experiment the membrane was depolarized to Ϫ100 mV for 2 min to permit complete recovery from inactivation and block. Peak currents of the test pulses were normalized to the peak currents of the prepulse and plotted against time. Representative current traces are shown in the absence (B, D, and F) and in presence (C, E and G) of isradipine for wild type, HHT-5411, and HHT-5371, respectively. Displayed currents were recorded after interpulse intervals of 20, 100, 300, 500, 1000, 3000, and 5000 ms (wild type, HHT-5411) and 20, 100, 50, 1000, and 3000 ms (HHT-5371). The traces are superimposed. After the 3-s prepulse (to ϩ10 or ϩ20 mV) the I Ba recovered within 20 ms to 80 -90% of its original value in control (absence of drug) for HHT-5411 and HHT-5371. The data points are fitted by a single exponential function for HHT-5371 and a double exponential function for the wild type and HH-5411. The time constants were calculated from 6 to 11 independent experiments and are summarized in Table II. sulted in the absence of use-dependent block by diltiazem and verapamil. These results are rather specific to these classes of calcium antagonists since use-dependent block by the dihydropyridine antagonist isradipine was not affected. The loss of use-dependent block from the mutations in IVS5 appear to be coupled to a decrease of voltage-dependent inactivation as has been shown for mutants in IIIS6 (22).
It appears that motif IVS5 plays an important role in the stability of channel inactivation. In steady-state inactivation studies, both HHT-5411 and HHT-5371 did not completely inactivate except in the presence of calcium antagonists (4). These results provide evidence that the loss of this aspect of calcium channel block is specific rather than due to a general loss in function attributed to gross changes in tertiary structure resulting from the introduction of foreign amino acid residues. There are several reasons why we support this contention. First, only the use-dependent block of these drugs is affected, and the tonic block by BZTs is spared. These results may be due to the fact that the high affinity binding region in motif IV is still intact. However, one mutation, HHT-5658 which we have described in this report, has three of the four residues in the YMAI sequence altered. Despite these changes the channel when expressed in Xenopus oocytes behaves much like the wild type Ba 2ϩ currents and exhibits tonic and use-dependent block by calcium antagonists. As anticipated, the high affinity binding to [ 3 H]PN200-110 in saturation binding assays was lost. These new data further implicate other regions out-side those areas that have been recently published (IIIS5, IIIS6, and IVS6) to be important for high affinity binding of these classes of drugs (3)(4)(5)(6)(7)(8)(9)(10)(11)(12)(13)(14)(15)(16)(17)(18)(19)(20)(21)(22). Second, the loss of use-dependent block is unique to verapamil and diltiazem. Use-dependent block by DHPs is not affected. These results suggest that the use-dependent block by PAA and BZT operates through a mechanism different from DHPs. Furthermore, the conclusion that there is a functional disparity between PAA/BZT and DHP is reasonable since there is more overlap in the high affinity binding region for PAAs and BZTs than for DHPs. Third, HHT-5411 which demonstrates a loss of use-dependent block to BZT and PAA maintains normal DHP binding affinity and allosteric modulation by diltiazem. HHT-5371 which did not retain usedependent block to BZTs and PAAs shows an approximate 6-fold lowered affinity to DHP binding. A fourth issue in support of our contention that motif IVS5 has a specific structural role for BZT and PAA action is that another mutant, HHT-5432 which is also from the same region, displayed use-dependent block by these drugs and otherwise normal channel currents. When this mutant was expressed in HEK 293 cells, it showed a slightly lowered (2-fold) affinity to DHPs and suggests a role in DHP binding in this region of motif IVS5.
These alterations in function are not likely to be due to the transfer of specific sodium channel sequences into the cardiac calcium channel for several reasons. Motif IVS6 in sodium channels has evolved to harbor specific amino acid residues for local anesthetic action (36). Mutation of these residues reduces not only block but also eliminates the use-dependent characteristics for local anesthetics (36). Thus, unlike the calcium channel, the sodium channel has both its binding site and use-dependent functions contained in motif IVS6.
These results also suggest that the nature of use-dependent block may be different for the calcium antagonists and local anesthetics. Second, it is believed that the intracellular linker between motifs III and IV is responsible for this voltage-dependent inactivation (38).
According to the modulated receptor hypothesis (23), variable affinity channel binding and modification of inactivation kinetics are required for the use-dependent effect of calcium channel drugs. In the case of HHT-5371 and HHT-5411, the very slow inactivation kinetics reflects a slow transition between the open and inactivated state of the channel. Thus, the open state of the channel exerts a lesser affinity to BZT and PAA. The transitions for this mutant to the inactivated state is a less preferred conformation so that repeated depolarizations do not lead to an increased fraction of inactivated channels. However, diltiazem binds to the HHT-5411 and HHT-5371 mutant channels as shown by the change in voltage dependence of inactivation and use dependence. Similar studies performed on these two constructs testing the use-dependence for verapamil confirmed the importance of inactivation in the development of use-dependent block (21,22). It is a reasonable assumption that a slowly inactivating mutant will be less efficiently blocked by PAA. In the case of calcium channels the only published reports of areas responsible for voltage-dependent inactivation implicates motif IS6 (39) and the carboxyl terminus in the case for ␣ 1C alternative splice variants (40). Therefore, the results suggest a possible third region of the calcium channel for use-dependent inactivation.
This report is the first to demonstrate a functional role of motif IVS5 for calcium channel function and the action of calcium channel antagonists, and the first to dissect the molecular regions involved in use-dependent block by BZT and PAA. We suggest a model (Fig. 10) in which inactivation of the channel occurs in two kinetic states. Both states bind to BZT and PAA. However, in the cardiac calcium channel IVS5 mu- FIG. 10. Model illustrating the fast and slow components of inactivation for the cardiac calcium channels in response to BZT and PAA. The channel exists initially in a resting (R) state that does not bind to BZT and PAA. In this state the drug-binding site is inaccessible and/or has a low affinity to the drugs. Upon depolarization the calcium channel opens (O) allowing the passage of calcium through the pore. In the open state the channel still has a low affinity to the drugs. Then the channel inactivates (I). Our results have demonstrated that inactivation can be divided into two kinetic states (I fast and I slow ). Both I fast and I slow bind drug. After the drug dissociates, the channel is left in an intermediate inactivated state (I*) that is no longer bound by drug before returning back to the resting (R) state. We have shown that a mutation in motif IVS5 does not completely inactivate, demonstrating an incomplete interconversion from the open to the inactivated state. This observation can be seen by the incomplete voltage-dependent inactivation shown in the electrophysiological traces of Figs. 5-7. However, upon the addition of BZT or PAA the mutant channels (HHT-5371 and (HHT-5411) completely inactivate. These results demonstrate the coupling between voltage-dependent inactivation and drug binding in the cardiac calcium channel without altering the residues shown to be required for high affinity drug binding. tants HHT-5411 and HHT-5371, the interconversion between the open and the inactivation state is incomplete. Treatment of either BZT or PAA forces the calcium channel toward complete inactivation. The slowed rate of inactivation in these channels can explain the loss of use-dependent block by these drugs in parallel with the observed decrease in voltage-dependent inactivation without affecting any of the residues for high affinity antagonist binding. We believe that motif IVS5 serves as either an accessory binding domain or a conduit for a conformational movement within the membrane to provide the use-dependent nature characterized by these two classes of calcium-blocking drugs.