Molecular Determinants of High Affinity Phenylalkylamine Block of l-type Calcium Channels in Transmembrane Segment IIIS6 and the Pore Region of the α1Subunit*

Recent studies of the phenylalkylamine binding site in the α1C subunit of l-type Ca2+ channels have revealed three amino acid residues in transmembrane segment IVS6 that are critical for high affinity block and are unique to l-type channels. We have extended this analysis of the phenylalkylamine binding site to amino acid residues in transmembrane segment IIIS6 and the pore region. Twenty-two consecutive amino acid residues in segment IIIS6 were mutated to alanine and the conserved Glu residues in the pore region of each homologous domain were mutated to Gln. Mutant channels were expressed in tsA-201 cells along with the β1b and α2δ auxiliary subunits. Assay for block of Ba2+ current by (−)-D888 at −60 mV revealed that mutation of five amino acid residues in segment IIIS6 and the pore region that are conserved between l-type and non-l-type channels (Tyr1152, Phe1164, Val1165, Glu1118, and Glu1419) and one l-type-specific amino acid (Ile1153) decreased affinity for (−)-D888 from 10–20-fold. Combination of the four mutations in segment IIIS6 increased the IC50 for block by (−)-D888 to approximately 9 μm, similar to the affinity of non-l-type Ca2+ channels for this drug. These results indicate that there are important determinants of phenylalkylamine binding in both the S6 segments and the pore regions of domains III and IV, some of which are conserved across the different classes of voltage-gated Ca2+ channels. A model of the phenylalkylamine receptor site at the interface between domains III and IV of the α1 subunit is presented.

ers are currently in clinical use, dihydropyridines, benz(othi)azepines, and phenylalkylamines. Drugs from all three classes bind to the pore-forming ␣ 1 subunit of L-type Ca 2ϩ channels in a manner that suggests that their binding sites are closely linked (5,6). Recently, much progress has been made toward the characterization of Ca 2ϩ antagonist receptor sites at the molecular level (for review see Ref. 7). Photoaffinity labeling studies and studies utilizing chimeric and mutant channels have indicated that both transmembrane segments IIIS6 and IVS6 are involved in forming the dihydropyridine (8 -10) and benz(othi)azepine (11) binding sites, and that dihydropyridine binding involves transmembrane segment IIIS5 as well (12).
In contrast, such studies of the phenylalkylamine binding site, to date, have indicated the involvement of only transmembrane segment IVS6. The high affinity photolabel LU49888 was found to derivatize transmembrane segment IVS6 of the skeletal muscle L-type channel exclusively (13). In subsequent studies with mutant channels, three amino acid residues in IVS6, which are unique to L-type channels, have been found to be critical for high affinity phenylalkylamine block (14,15). Simultaneous mutation of these three residues, Tyr 1463 , Ala 1467 , and Ile 1470 , resulted in channels that were blocked by the high affinity phenylalkylamine (Ϫ)-D888 with an affinity similar to that of non-L-type channels. Transfer of the L-type IVS6 sequence in chimeric Ca 2ϩ channels was sufficient to confer L-type sensitivity to (Ϫ)-D888 in a non-L-type channel (16).
Despite the studies implicating only segment IVS6 in phenylalkylamine binding, several lines of evidence suggest that IIIS6 might also be involved in the phenylalkylamine receptor site. The allosteric interactions among all three major classes of L-type Ca 2ϩ channel blockers suggests that the binding sites for these drugs are near each other but not identical. Because evidence from both photolabeling studies and analysis of mutant channels has implicated domain IIIS6 in the binding of both dihydropyridines and benzothiazepines, it is important to examine the role of individual amino acids in domain IIIS6 in block of L-type channels by phenylalkylamines. We have individually mutated 21 amino acids in IIIS6 to Ala and another (Ala 1157 ) to Pro and analyzed the mutant channels for sensitivity to the high affinity phenylalkylamine (Ϫ)-D888. We report here that four amino acid residues in transmembrane domain IIIS6 are required for high affinity block of L-type Ca 2ϩ channels by (Ϫ)-D888.
The phenylalkylamines are thought to block ion channels by occluding the ion-conducting pore and thereby preventing cation permeation (17)(18)(19). Moreover, at physiological pH, phenylalkylamines are predominantly positively charged due to protonation of the tertiary amino group. In view of these observations, we mutated four conserved Glu residues known to contribute to the ion-selective pore of L-type Ca 2ϩ channels, Glu 363 , Glu 709 , Glu 1118 , and Glu 1419 (20), to Gln and screened the resulting mutant channels for sensitivity to (Ϫ)-D888. We report here that Glu 1118 and Glu 1419 in the putative pore-lining segments of homologous domains III and IV, respectively, are also involved in high affinity block of L-type Ca 2ϩ channels by phenylalkylamines.

EXPERIMENTAL PROCEDURES
Construction of Mutants-All mutations were constructed using oligonucleotide-directed mutagenesis as described previously (21). The IIIS6, E1118Q, and E1419Q mutations were inserted into full-length ␣ 1 subunit constructs in the expression vector Zem229 (Dr. Eileen Mulvihill, Zymogenetics Corp., Seattle) using the 1.5-kilobase SpeI/DraIII fragment and the 272-base pair DraIII/DraIII fragment in a three-way ligation. The E363Q mutation was inserted into the full-length ␣ 1 subunit construct using the 1.4-kilobase NgoMI/BglII fragment. The E709Q mutation was inserted into the full-length ␣1 subunit construct using the 1.3-kilobase SgrAI/SpeI fragment. All mutations were confirmed by cDNA sequencing.
Cell Culture-tsA201 cells, a subclone of the human embryonic kidney cell line HEK293 that expresses SV40 T antigen (a gift of Dr. Robert Dubridge, Cell Genesis, Foster City, CA), were maintained in monolayer culture in Dulbecco's modified Eagle's medium/Ham's F-12 medium (Life Technologies, Inc.), supplemented with 10% fetal bovine serum (Hyclone, Logan, UT), and incubated at 37°C in 10% CO 2 .
Electrophysiology-(Ϫ)-D888 was applied to cells recorded in the whole cell patch-clamp configuration by the addition of 0.2 ml of a 6ϫ stock to a 1-ml bath. The extracellular (bath) saline contained 150 mM Tris, 4 mM MgCl 2 , 10 mM BaCl 2 , and pH adjusted to 7.3 with methanesulfonic acid. Patch electrode saline (intracellular) contained 130 mM N-methyl-D-glucamine, 10 mM EGTA, 60 mM HEPES, 2 mM MgATP, 1 mM MgCl 2 , and pH adjusted to 7.3 with methanesulfonic acid. All experiments were performed at room temperature (20 -23°C). No nonlinear outward currents were detected under these conditions. Patch electrodes were pulled from VWR micropipettes and fire-polished to produce an inner tip diameter of 4 -6 m. Currents were recorded using a List EPC-7 patch clamp amplifier and filtered at 2 kHz (8-pole Bessel filter, Ϫ3 db). Data were acquired using Basic-Fastlab software (Indec Systems). Voltage-dependent currents have been corrected for leak using an on-line P/4 subtraction paradigm.
Effects of Mutations in Transmembrane Segment IIIS6 of the ␣ 1 Subunit-The putative transmembrane segment IIIS6 of the ␣ 1C subunit of L-type Ca 2ϩ channels contains primarily hydrophobic amino acid residues.
␣ 1C ISIFFIIYIIIIAFFMMNIFVGFVI 1169 ␣ 1B L---YVV-FVVFP--FV----ALI-1373 SCHEME 1 Of the 25 amino acid residues predicted to comprise this transmembrane segment, 14 are different in the phenylalkylamineinsensitive (14) ␣ 1B subunits of N-type Ca 2ϩ channels (Scheme 1). To assess the role of the individual amino acid residues in the IIIS6 segment of ␣ 1C in high affinity block by (Ϫ)-D888, we mutated Ala 1157 to Pro as in N-type Ca 2ϩ channels (Scheme 1), mutated the other amino acids in IIIS6 to Ala, and screened the mutant channels for sensitivity to (Ϫ)-D888. Ala was chosen for substitution because it has minimal effects on protein secondary structure when substituted in ␣ helices in the core of proteins (26) and therefore is expected to reduce the hydrophobicity and size of the amino acid residue in each position in this putative ␣ helix without causing global conformational change.
All of the mutant ␣ 1 subunits studied formed functional Ca 2ϩ channels in tsA-201 cells except A1157P. Seventeen of these mutations had no effect on the concentration dependence of block by (Ϫ)-D888, as illustrated for M1160A in Fig. 1 (B and F). In contrast, three of these single amino acid mutations caused marked increases in the IC 50 values for block of Ba 2ϩ currents by (Ϫ)-D888. For example, the mutation F1164A caused a large increase in the concentration of (Ϫ)-D888 required for block of the Ba 2ϩ current (Fig. 1C) and approximately a 10-fold shift to higher concentration of the inhibition curve for (Ϫ)-D888 (Fig. 1F). The IC 50 values for block of Ba 2ϩ currents through the wild type and all of the mutant ␣ 1C subunits are illustrated as a bar graph in (Ϫ)-D888 block (Fig. 2).
In contrast to the other mutations studied, Y1152A reduced the IC 50 for (Ϫ)-D888 approximately 7-fold to 7.5 Ϯ 2.4 nM, suggesting a potentially important role of the amino acid in this position in determining affinity for phenylalkylamines. To assess the effect of a more conservative substitution at this postion, we mutated Tyr 1152 to Phe instead of Ala. The Y1152F mutation, which removes only a hydroxyl group, caused an increase in the concentration of (Ϫ)-D888 required for block of Ba 2ϩ currents (Fig. 1D) and approximately an 18-fold shift to higher concentration in the inhibition curve for (Ϫ)-D888 ( Fig.  1F; IC 50 ϭ 873 Ϯ 219 nM). Thus, our results suggest that a total of four amino acid residues in segment IIIS6 may contribute to formation of the high affinity receptor site for (Ϫ)-D888 (Fig. 2).
Additive Effects of Multiple Mutations in Transmembrane Segment IIIS6 -If these four amino acid residues are determinants of high affinity binding of (Ϫ)-D888, mutation of combinations of them should increase the IC 50 for block of the L-type Ca 2ϩ channels to a value near 5 M, like the N-type Ca 2ϩ channels (14). The quadruple mutation Y1152F,I1153A, F1164A,V1165A (YIFV) caused an even further increase in the IC 50 for block by (Ϫ)-D888 than the single mutations. Concentrations of (Ϫ)-D888 in the range of 5 M were required to observe significant block of Ba 2ϩ currents (Fig. 1E). Analysis of the block of Ba 2ϩ currents by a range of concentrations of (Ϫ)-D888 at equilibrium yielded an IC 50 of 8.7 Ϯ 4.8 M (Fig. 1F).
Each concentration of drug tested in these experiments kcal/mol. Therefore, the decrease in binding energy in the YIFV was only 53% of the sum of the changes caused by the individual mutations, suggesting that the contributions to the binding energy by these four amino acid residues are not independent. Functional Properties of IIIS6 Mutants-To examine the specificity of the effects of the IIIS6 mutations on Ca 2ϩ channel function, we compared their kinetic and voltage-dependent properties with those of Ca 2ϩ channels containing wild type ␣ 1C . Current-voltage relationships were generally similar for wild type and for the mutant ␣ 1C subunits having altered affinity for (Ϫ)-D888 (Fig. 3, A and B) with peak Ba 2ϩ currents observed at ϩ10 to ϩ20 mV in each case. However, closer analysis of conductance-voltage relationships revealed small but significant differences (wild type, V1 ⁄2 ϭ ϩ11.5 mV; Y1152F, V1 ⁄2 ϭ ϩ6 mV; I1153A, V1 ⁄2 ϭ ϩ3.7 mV; F1164A, V1 ⁄2 ϭ ϩ6.8 mV; and V1165A, V1 ⁄2 ϭ ϩ4.0 mV). These voltage shifts were not due to differences in the time from forming the whole cell patch clamp configuration because current-voltage relations were first measured at 5 min after break-in, and no additional shifts were observed after that time. Although the voltage dependence of activation was significantly different for the single mutations, the voltage dependence of channel activation was not significantly different in the combination mutant YIFV compared with wild type (YIFV, V1 ⁄2 ϭ ϩ14.3 mV), suggesting that the effects of the single mutations were compensated in the combined mutant.
The IIIS6 mutations also affected the apparent reversal potential (E rev ) of the mutant channels, which is a measure of their ion selectivity. Although the E rev of the mutant YIFV was shifted only slightly from wild type (YIFV, E rev ϭ ϩ56.3 Ϯ 2.9 mV, n ϭ 6; wild type, E rev ϭ ϩ61.3 mV Ϯ 4.4, n ϭ 10), both single Tyr 1152 mutations were shifted approximately 15 mV (Y1152A, E rev ϭ ϩ46.9 Ϯ 6.1 mV, n ϭ 4; Y1152F, E rev ϭ ϩ46.9 Ϯ 3.1 mV, n ϭ 4). The E rev values of mutants I1153A and V1165A were also substantially shifted from wild type (I1153A, E rev ϭ 47.4 Ϯ 1.8 mV, n ϭ 5; V1156A, E rev ϭ 42.2 Ϯ 2.3 mV, n ϭ 5), whereas the E rev of mutant F1164A was more moderately affected (F1164A, E rev ϭ 53.1 Ϯ 3.6 mV, n ϭ 5). These results are consistent with the idea that the IIIS6 segment contributes to the lining of the pore of Ca 2ϩ channels, and mutations of amino acid residues in this segment therefore alter the selectivity of ion conductance.
In contrast to their lack of effect on channel activation, phenylalkylamines cause Ca 2ϩ channel inactivation curves to shift in the hyperpolarizing direction, indicating that block by these compounds is more potent at depolarized potentials where inactivation is favored (27)(28)(29). Therefore, mutations that alter voltage-dependent inactivation of Ca 2ϩ channels may affect binding and block by phenylalkylamines indirectly. To avoid such effects, all of our experiments were carried out at a holding potential of Ϫ60 mV, considerably more negative than steady-state inactivation of L-type Ca 2ϩ channels, so further reduction in voltage-dependent inactivation by mutations should not significantly reduce affinity for (Ϫ)-D888. Nevertheless, we have characterized the voltage dependence of inactivation of the mutant Ca 2ϩ channels in detail (Fig. 3C). Muta- . Ca 2ϩ channel current (carried by 10 mM Ba 2ϩ ) was monitored once every 10 s by a 100-ms depolarization to ϩ10 mV from a holding potential of Ϫ60 mV. Bar labeled N-type indicates IC 50 for Ca 2ϩ channel current recorded from NGF-differentiated PC12 (rat pheochromocytoma) cells that express N-type Ca 2ϩ channels containing ␣ 1B . tion of residues Ile 1153 , Phe 1164 , and Val 1165 caused positive shifts in V1 ⁄2 for inactivation (V1165A, V1 ⁄2 ϭ Ϫ14.5 mV; I1153A, V1 ⁄2 ϭ Ϫ11.6 mV; and F1164A, V1 ⁄2 ϭ Ϫ9.1 mV) compared with wild type (V1 ⁄2 ϭ Ϫ17.7 mV). Removal of the hydroxyl group from Tyr 1152 did not affect the voltage dependence of steadystate inactivation significantly (Y1152F, V1 ⁄2 ϭ Ϫ18.8 mV), but removal of the aromatic ring from that position in IIIS6 resulted in a negative shift of approximately 11 mV in V1 ⁄2 (Y1152A, V1 ⁄2 ϭ Ϫ29.0 mV). Inactivation curves for the combination mutant YIFV are approximately 12 mV more positive than wild type (V1 ⁄2 ϭ Ϫ5.7 mV).
To examine the correlation between shifts in the voltage dependence of inactivation and the affinity for block by (Ϫ)-D888 quantitatively, we plotted the IC 50 values for the IIIS6 mutants against their half-inactivation voltage (Fig. 3D). Halfinactivation values range from Ϫ29 mV to Ϫ4 mV, and no overall correlation with IC 50 is evident. For example, mutant Y1152F has nearly the same V1 ⁄2 value as wild type but substantially increased IC 50 . Many other mutants have substantially increased or decreased values of V1 ⁄2 , but no change in IC 50 . Mutants I1153A, V1165A, and F1164A have different half-inactivation potentials but comparable increases in IC 50 values (Fig. 3D). Thus, because inactivation is minimal at the holding potential used in these experiments (Ϫ60 mV) and no correlation of IC 50 with V1 ⁄2 is observed, the decrease in (Ϫ)-D888 affinity cannot be ascribed to changes in the intrinsic voltage dependence of channel inactivation in the mutants. On the other hand, the correlation plot of Fig. 3D may reveal the reason for the increase in affinity caused by mutation Y1152A. This mutant has by far the most negative V1 ⁄2 value (Ϫ29 mV) and also has the highest affinity for (Ϫ)-D888. Enhanced steady-state inactivation at Ϫ60 mV for this mutant may contribute substantially to its increased affinity for (Ϫ)-D888.
Effects of Mutations in the Pore Region of the ␣ 1 Subunit-The ␣ 1 subunits of voltage-gated Ca 2ϩ channels contain four highly conserved P-loops between the S5 and S6 transmembrane segments of each homologous domain that together form the selectivity filter through which the channel conducts cations. The high selectivity of voltage-gated Ca 2ϩ channels for divalent cations over monovalent ions and for Ca 2ϩ ions over other divalent ions is mediated by four Glu residues, one in each homologous domain, that are conserved across all Ca 2ϩ channels (20,30). Because phenylalkylamines are thought to block ion channels by binding in the pore and the protonated amino group of (Ϫ)-D888 is positively charged at physiological pH and could potentially interact with the pore Glu residues through an electrostatic mechanism, we mutated each of these Glu residues individually to Gln to neutralize the negative charge that the acidic side chains have at physiological pH and tested the affinity for block by (Ϫ)-D888 in the four E 3 Q mutants. The mutant channels E363Q and E709Q having mutations in domains I and II, respectively, were both blocked by (Ϫ)-D888 with IC 50 values that were not significantly different from that of the wild type channel (Fig. 4, A and D). However, the mutant channels E1118Q and E1419Q in domains III and IV, respectively, both had significant increases in the IC 50 for (Ϫ)-D888 block (Fig. 4, B and D). The E1118Q mutant caused approximately a 20-fold rightward shift in the concentration . Apparent reversal potentials were estimated by linear extrapolation of the data between ϩ20 and ϩ40 mV to the abscissa. C, voltage dependence of inactivation. Cells expressing the indicated Ca 2ϩ channels were depolarized from a holding potential of Ϫ60 mV to the indicated prepulse potentials for 5 s and then further depolarized to ϩ10 mV for 100 ms to record Ba 2ϩ currents. Values for V1 ⁄2 for inactivation were determined as described previously (14). D, relationship between V1 ⁄2 for inactivation and IC 50 for (Ϫ)-D888. wt, wild type. dependence of block by (Ϫ)-D888 ( Fig. 4D; IC 50 ϭ 949 Ϯ 275 nM). The E1419Q mutant caused approximately a 15-fold rightward shift in the concentration dependence of block by (Ϫ)-D888 (Fig. 4, B and D; IC 50 ϭ 717 Ϯ 241 nM). These results indicate that the P-loop Glu residues in domains III and IV interact with bound (Ϫ)-D888, but those in domains I and II do not.
Functional Properties of Pore Mutations-Besides the effects on (Ϫ)-D888 block, mutation of the pore Glu residues affected other parameters of channel function. The current-voltage curves for these four mutants (Fig. 4C) show that the voltage dependence of channel activation was not significantly different from wild type. However, the E rev of the mutant E363Q was shifted approximately 16 mV to more negative potential (E rev ϭ 44.7 Ϯ 6.9 mV, n ϭ 5), whereas the E rev of mutant E1118Q was shifted approximately 8 mV negatively (E rev ϭ 52.9 Ϯ 3.9 mV, n ϭ 5) compared with wild type (E rev ϭ 61.3 Ϯ 4.4 mV, n ϭ 10). Surprisingly, E rev of the corresponding mutations in domains II and IV, E709Q and E1419Q, were not different from wild type (E709Q, E rev ϭ 63.5 Ϯ 4.5, n ϭ 5; E1419Q, E rev ϭ 57.5 Ϯ 4.0, n ϭ 6). The effects of the pore mutations on the voltage dependence of inactivation also varied widely. In mutant E363Q, the V1 ⁄2 for inactivation was not substantially different from wild type (E363Q V1 ⁄2 ϭ Ϫ16.3 mV), whereas V1 ⁄2 values for inactivation of E709Q and E1419Q were shifted to more positive potentials by approximately 10 and 6 mV, respectively (E709Q, V1 ⁄2 ϭ Ϫ7.1 mV; E1419Q, V1 ⁄2 ϭ Ϫ11.6 mV). As with the mutations in segment IIIS6, there was no correlation between the effects of the mutations on IC 50 for (Ϫ)-D888 and V1 ⁄2 for inac-tivation, consistent with the conclusion that these mutations affect the binding of (Ϫ)-D888 directly.

Contribution of Conserved and L-type-specific Amino Acid
Residues in Transmembrane Segment IIIS6 to (Ϫ)-D888 Block of L-type Channels-As summarized in Fig. 2, we have analyzed the contribution of 21 single amino acid side chains in the transmembrane segment IIIS6 to block by (Ϫ)-D888. Of the single amino acid residues mutated in this region, only four had significant effects on the IC 50 for (Ϫ)-D888. One of these four amino acids is unique to L-type channels (Ile 1153 ), whereas the others are conserved across all types of voltage-gated Ca 2ϩ channels (Scheme 1, underlined residues). The involvement of primarily conserved amino acid residues in segment IIIS6 in high affinity (Ϫ)-D888 binding stands in sharp contrast to the results of previous experiments in segment IVS6 where the only three amino acids required for (Ϫ)-D888 binding were unique to L-type channels (14 -16). Our results show that conserved amino acid residues play crucial roles in the action of L-type selective drugs and demonstrate the importance of systematic analysis of all amino acid residues in putative drug binding sites in addition to chimeric approaches that target only isoform-specific residues.
The single amino acid mutation in this region with the largest effect was the conservative substitution of Phe for Tyr (Y1152F), in which only a single hydroxyl group was removed from the native channel structure. This result is similar to our previous observation (29) that mutation of a Tyr in IVS6   FIG. 4. Functional properties and (؊)-D888 block of pore glutamate mutants. L-type Ba 2ϩ currents were recorded from E709Q (domain II) (A) and E1419Q (domain IV) (B). Examples of Ba 2ϩ current records from individual cells are presented in which an ascending series of doses of (Ϫ)-D888 was applied. Currents were recorded during 100-ms depolarizations to ϩ10 mV from a holding potential of Ϫ60 mV. C, current-voltage relations in the absence of (Ϫ)-D888 for control and E363Q (EIQ), E709Q (EIIQ), E1118Q (EIIIQ), and E1419Q (EIVQ). Mean values are shown (n ϭ 5). Apparent reversal potentials were estimated by linear extrapolation of the data between ϩ20 and ϩ40 mV to the abscissa. D, effect of pore glutamate mutations on (Ϫ)-D888 block. The IC 50 for each EQ mutant is shown Ϯ standard error (n ϭ 4 -6). wt, wild type.
(Tyr 1463 ) to Phe caused a large reduction in sensitivity to (Ϫ)-D888. Simultaneous mutation of both Tyr 1152 and Tyr 1463 to Phe resulted in a channel with an approximately 100-fold increase in IC 50 for (Ϫ)-D888 and normal inactivation properties. 1 The importance of these two hydroxyl groups, which are potential hydrogen bond donors, suggests a privileged hydrogen bond between them and the single meta-methoxy group, a potential hydrogen bond acceptor, on the phenethylamine group of D888. This interaction is apparently not accessible for the lower affinity phenylalkylamines verapamil and D600 (methoxyverapamil), which possess an additional para-methoxy group (29).
Our results support the conclusion that the effects of the mutations of amino acid residues Tyr 1152 , Ile 1153 , Phe 1164 , and Val 1165 on binding of (Ϫ)-D888 results from alteration of the interactions of the side chains of these residues with the bound drug. The effects of mutations of these residues are highly specific; mutations of adjacent residues to Ala have no effect on block by (Ϫ)-D888. There is no correlation between the effects of these mutations on activation or inactivation of Ca 2ϩ channels and their effects on affinity for (Ϫ)-D888, indicating that the mutations do not cause their effects by indirect allosteric changes. Thus, our working hypothesis is that these four amino acid residues interact with (Ϫ)-D888 when it is bound to its receptor site on L-type Ca 2ϩ channels with high affinity.
As suggested previously for amino acid residues in IVS6 that are required for high affinity phenylalkylamine block of L-type Ca 2ϩ channels (14), it is likely that the IIIS6 amino acid residues that affect (Ϫ)-D888 block also project into the ion-conducting pore. Like the IVS6 residue Tyr 1463 , mutation of Tyr 1152 , Ile 1153 , Phe 1164 , and Val 1165 to Ala resulted in significant shifts in apparent reversal potential. As we showed for Y1463A (14), these shifts are likely to be due to increases in channel permeability to the normally impermeant organic cation N-methyl-D-glucamine, which is the principal cation in the intracellular solution. Thus, mutations of Tyr 1152 , Ile 1153 , Phe 1164 , and Val 1165 change the shape or size of the ion-conducting pore.
Contribution of Pore Glutamate Residues to (Ϫ)-D888 Block-Phenylalkylamines are thought to bind in the pore of Ca 2ϩ channels (17)(18)(19), and our results show that amino acid residues in transmembrane segments IIIS6 and IVS6 that are involved in high affinity binding of (Ϫ)-D888 are also involved in maintaining the ion selectivity of the pore (Ref. 14 and this work). We therefore investigated the effects of mutating four highly conserved Glu residues, Glu 363 , Glu 709 , Glu 1118 , and Glu 1419 , in the pore region for two reasons. First, these acidic amino acid side chains apparently project into the pore to form the Ca 2ϩ binding site(s) that confers Ca 2ϩ selectivity to voltage-gated Ca 2ϩ channels (20,30). Second, these amino acid side chains are negatively charged at physiological pH and potentially participate in electrostatic interactions with protonated (i.e. positively charged) phenylalkylamine molecules.
The spatial selectivity of the effects of mutations of these four pore Glu residues is consistent with the large body of data showing domains III and IV to be the site of binding of phenylalkylamines. Mutations of Glu 1118 in domain III and Glu 1419 in domain IV caused major reductions in the affinity for (Ϫ)-D888, whereas mutations of Glu 363 in domain I and Glu 709 in domain II had no appreciable effects. The decreased affinity of E1118Q and E1419Q for (Ϫ)-D888 was likely not caused by shifts in the voltage dependence of inactivation because E709Q, the pore mutant with the largest shift in steady-state inactivation, had no change in affinity for the drug. Although it is possible that the decreased affinity for (Ϫ)-D888 conferred by E1118Q and E1419Q is due to indirect effects, the specificity for these two mutations among the four pore Glu residues and their lack of correlated effects on channel gating argue that they interact directly with the bound drug molecule, most likely through  5. A domain interface model for high affinity phenylalkylamine block. A, amino acid residues in IVS6 that are unique to L-type channels as well as conserved amino acid residues from IIIS6 and the pore region of domains III and IV converge to form the binding site for (Ϫ)-D888. The IIIS6 and IVS6 residues could form a hydrophobic pocket that binds the aromatic moieties of the phenylalkylamines and stabilizes a charge interaction of the alkylamino group of the phenylalkylamines with Glu 1118 and Glu 1419 . White letters inside of black circles represent amino acids that, when mutated, disrupted (Ϫ)-D888 block. Black letters in white circles represent amino acids that, when mutated, did not affect (Ϫ)-D888 block. Gray letters in white circles represent amino acids that were not mutated. B, confluence of amino acids involved in dihydropyridine binding and phenylalkylamine block of L-type Ca 2ϩ channels. The amino acid sequences of transmembrane domains IIIS6 and IVS6 are shown as ␣ helices. White letters inside of black symbols represent amino acids that, when mutated, disrupted binding/block. Black circles represent amino acids involved in DHP binding only. Black diamonds represent amino acids that disrupt PA block only. Black squares represent amino acids that disrupt both DHP binding and PA block. Black letters in white circles represent amino acids that, when mutated, did not affect DHP binding or PA block. electrostatic interactions with the positively charged amino group. This conclusion is consistent with previous studies showing that phenylalkylamines block ion channels in their positively charged, protonated state (18,31) and with the evidence that phenylalkylamines bind within the pore of the Ca 2ϩ channel.
(Ϫ)-D888 Binds to the Interface between Domains III and IV-The results of this and previous studies (14 -16) suggest that the receptor site for (Ϫ)-D888 is composed of at least four distinct subsites: IIIS6, the P-loop in the IIIS5-IIIS6 linker, IVS6, and the P-loop in the IVS5-S6 linker. The critical role of amino acid residues from both IIIS6 and IVS6 in (Ϫ)-D888 binding and block strongly suggest that these two transmembrane domains are juxtaposed to form a portion of the intracellular mouth of the ion-conducting pore (Fig. 5). Thus, our results support a "domain interface model" of phenylalkylamine binding and block, as proposed previously for dihydropyridines (32). The YIFV residues are arranged in two clusters of two amino acids each in IIIS6, but these four residues do not align in precisely the same position in consecutive turns of the ␣ helix as the YAI motif in IVS6 does (14). Nevertheless, the deviation from strict cylindrical shape of many bundled ␣ helices in proteins of known structure would allow all four of these residues to contribute to a binding site for phenylalkylamines in the pore of the channel. We propose that the YIFV motif in IIIS6 and the YAI motif in IVS6 act together to form a hydrophobic pocket that stabilizes (Ϫ)-D888 bound in the pore and enhances the electrostatic interactions between the pore Glu residues Glu 1118 and Glu 1419 and the tertiary amino group of (Ϫ)-D888 (Fig. 5A). It will be of interest to determine how other structurally related phenylalkylamines of differing affinity interact with these components of the high affinity phenylalkylamine receptor site.
Dihydropyridine and (Ϫ)-D888 Binding Sites Are Overlapping but Not Identical-As we have shown in the preceding paper (33), mutation of specific amino acid residues in transmembrane segments IIIS6 and IVS6 greatly reduce the affinity of the L-type Ca 2ϩ channel for the dihydropyridine PN200-110 as measured by equilibrium binding. The results of mutations in IIIS6 and IVS6 on block by (Ϫ)-D888 and binding of PN200-110 are summarized in Fig. 5B. It is clear that the binding sites for these structurally distinct molecules are intricately interwoven, because single mutations at three residues in adjacent positions in IIIS6 and IVS6 (Tyr 1152 , Ile 1153 , and Tyr 1463 ) disrupted both DHP 2 binding and (Ϫ)-D888 block. However, mutations at other positions had drug-specific effects. For example, mutation of Phe 1164 , Val 1165 , Ala 1467 , and Ile 1470 affected block by (Ϫ)-D888 but did not significantly affect PN200-110 binding. Conversely, mutations of five amino acids in the central region of IIIS6 (Ile 1156 , Phe 1158 , Phe 1159 , Met 1160 , and Met 1161 ) as well as four amino acids in IVS6 (Phe 1462 , Met 1464 , Ile 1471 , and Asn 1472 ) each had significant effects on DHP binding but not on (Ϫ)-D888 block. We therefore suggest that the phenylalkylamines and DHPs bind to different faces of the IIIS6 and IVS6 transmembrane segments and in some cases bind to opposite sides of the same amino acid residues. In this model, the allosteric interactions between bound phenylalkylamines and DHPs would take place over very short distances, possibly separated by no more than the plane of a phenyl ring or the width of an aliphatic side chain.