Abstract
Voltage-dependent L-type Ca2+ channels are modulated by the binding of Ca2+ channel antagonists and agonists to the pore-forming α1c subunit (CaV 1.2). We recently identified Ser1115 in IIIS5-S6 linker of α1C subunit as a critical determinant of the action of 1,4-dihydropyridine agonists. In this study, we applied alanine-scanning mutational analysis in IIIS5-S6 linker of rat brain α1C subunit (rbCII) to illustrate the role of pore-forming IIIS5-S6 linker in the action of Ca2+ channel modulators. Ca2+ channel currents through wild-type (rbCII) or mutated α1C subunits, transiently expressed in BHK6 cells with β1a and α2/δ subunits, were analyzed. The replacement of Phe1112 by Ala (F1112A) significantly impaired the sensitivity to Ca2+ channel agonists (S)-(-)-Bay k 8644 and FPL-64176, and modestly to 1,4-dihydropyridine (DHP) antagonists. The low sensitivity of F1112A and S1115A to DHP antagonists was consistent with the reduced binding affinity for [3H](+)PN200-110. The replacement of Phe1112 by Tyr, but not by Ala, restored the long openings produced by FPL-64176, thus indicating the critical role of aromatic ring of Phe1112 in the Ca2+ channel agonist action. Interestingly, double-mutant Ca2+ channel (F1112A/S1115A) failed to discriminate between Ca2+ channel agonist (S)-(-)-1,4-dihydro-2,6-dimethyl-5-nitro-4-(2-[trifluoromethyl] phenyl)-3-pyridine carboxylic acid methyl ester (Bay k 8644) and antagonist (R)-(+)-Bay k 8644 and was blocked by the two enantiomers in an identical manner. These results indicate that both Phe1112 and Ser1115 in linker IIIS5-S6 are required for the action of Ca2+ channel agonists. A model of the DHP receptor is proposed to visualize possible interactions of Phe1112, Ser1115, and other DHP-sensing residues with a typical DHP ligand nifedipine.
Voltage-dependent L-type Ca2+ channels play critical roles in shaping action potentials, excitation-contraction coupling, excitation-secretion coupling, and gene expression. L-type Ca2+ channel currents are blocked by classic Ca2+ channel antagonists such as 1,4-dihydropyridines (DHP), benzothiazepines (BTZ), and phenylalkylamines (PAA) (Regulla et al., 1991; Striessnig et al., 1998), whereas other non-L-type Ca2+ channel currents are less sensitive to Ca2+ channel antagonists. Ca2+ channel agonists such as DHP compounds and a benzoylpyrrole derivative, FPL-64176, enhance L-type Ca2+ channel currents through prolongation of the channel open time and enhancement of the open probability (Zheng et al., 1991; Adachi-Akahane and Nagao, 2000). DHP Ca2+ channel agonists have been shown to exert antagonistic effects at high concentrations (Usowicz et al., 1995) or at depolarized membrane potentials. In contrast, FPL-64176 binds to the Ca2+ channel at the binding site distinct from that of DHPs (Rampe and Lacerda, 1991), and produces exclusively Ca2+ channel agonistic action even at very high concentrations without Ca2+ channel antagonistic effects.
Critical determinants of the high-affinity binding sites for Ca2+ channel antagonists and agonists on L-type Ca2+ channel have been identified. However, conformational changes of the Ca2+ channel during modulation by these compounds and the molecular mechanism underlying the differences between Ca2+ channel agonists and antagonists are largely unknown. The elucidation of DHP binding pocket is especially important because only DHP derivatives have both Ca2+ channel agonists and antagonists, whereas neither BTZ nor PAA agonists are known. Amino acid residues involved in the binding pocket of DHP agonists and antagonists have been determined (Fig. 1B) (Grabner et al., 1996; Mitterdorfer et al., 1996; Peterson et al., 1996; Schuster et al., 1996; He et al., 1997; Wappl et al., 2001). The previous studies have mostly depended on differences in amino acid residues between DHP-sensitive α1C subunit and DHP-insensitive subunit such as α1A and α1E. Considering that α1A-DHP (Hockerman et al., 1997; Sinnegger et al., 1997) has completely gained the sensitivity to DHPs by transferring nine amino acids, the DHP binding pocket may involve unidentified amino acid residues common between α1A subunit and α1C subunit. Indeed, some amino acids involved in the binding pocket turned out to be conserved in DHP-insensitive α1 subunits (Peterson et al., 1997).
We recently identified Ser1115, located between IIIS5 and IIIS6 of the pore-forming α1 subunit of L-type Ca2+ channel, as a critical determinant of the action of DHP agonists (Yamaguchi et al., 2000). The facts that high-affinity DHP-binding was stabilized by the binding of Ca2+ to the Ca2+ channel pore (Glossmann and Striessnig, 1990; Peterson and Catterall, 1995; Striessnig et al., 1998) and that IIIS5-S6 linker was photoaffinity labeled by photoreactive DHPs (Striessnig et al., 1991) also support the idea that DHPs interact with the pore-forming region of the Ca2+ channel. In the present study, we aimed at clarifying the role of the pore-forming region of α1C subunit in the modulation of gating kinetics by Ca2+ channel agonists. For this purpose, we applied alanine-scanning mutational analysis in IIIS5-S6 linker of rat brain Ca2+ channel α1C subunit (rbCII, Cav1.2).
We found that Phe1112 contributes to the binding pockets for DHPs along with Ser1115 in IIIS5-S6 linker and that both Phe1112 and Ser1115 are required for transducing the binding of Ca2+ channel agonists into the Ca2+ channel agonistic action.
Materials and Methods
Point Mutations. Alanine-scanning mutagenesis was performed in 10 amino acids in IIIS5-S6 linker of rat brain α1C subunit (rbCII) that was kindly supplied by Dr. T. P. Snutch (Snutch et al., 1991) using QuikChange method (QIAGEN) (Fig. 1, A and C). Single point mutations in IIIS5-S6 linker were introduced into PstI (3384) to XmnI (3743) fragment of rbCII as described previously (Yamaguchi et al., 2000). Polymerase chain reaction was performed using Pfu-turbo polymerase (Stratagene, La Jolla, CA) and verified by sequence analysis.
Single point mutations of F357A, F700A, and F1413A were introduced into rbCII (full length). Polymerase chain reaction was performed using Pfu-turbo polymerase (Stratagene). After polymerase chain reaction, BamHI (1193) to BamHI (1434) region (F357A), Bsp1407I (1984) to SpeI (3090) region (F700A), and BspHI (3736) to EcoRV (4515) region (F1413A) of PCR products were verified by sequence analysis and introduced into the respective region of rbCII. All α1C mutants were inserted into the expression plasmid pcDNAIII.
Cell Culture and Transfection. Mutated α1C subunits and rb-CII were transiently expressed in BHK6 cells that stably express rabbit β1a and α2/δ subunits (Wakamori et al., 1998) as described previously (Yamaguchi et al., 2000). For electrophysiological experiments, α1C mutants in pcDNAIII were transfected together with GFP (pEGFPC2) using Superfect (Qiagen). Cells were used for experiments 24∼48 h after transfection. For DHP binding assay, α1C mutants in pcDNAIII were transfected using LipofectAMINE Plus (Invitrogen) and membranes were prepared 48 h after transfection.
Whole-Cell Patch-Clamp Recording. The whole-cell L-type Ca2+ channel currents were recorded with Ca2+ (2 mM) or Ba2+ (2 mM; Fig. 6B) as a charge carrier in bath solution containing 137 mM NaCl, 5.4 mM KCl, 1 mM MgCl2, 10 mM glucose, 10 mM HEPES, and 2 mM CaCl2 or BaCl2, pH adjusted to 7.4 with NaOH, at room temperature as described previously (Yamaguchi et al., 2000). The resistance of the heat-polished patch pipettes ranged between 2 and 4 MΩ when filled with the internal solution composed of 120 mM CsCl, 20 mM TEACl, 14 mM EGTA, 5 mM MgATP, 5 mM Na2 creatine phosphate, 0.2 mM GTP, 10 mM HEPES, and 0.2 mM cAMP, pH adjusted to 7.3 with CsOH. Whole-cell currents were measured using a patch/whole-cell clamp amplifier (NIHON KOHDEN, Tokyo, Japan) via analog-to-digital converter (Digidata 1200; Axon Instruments Inc., Union City, CA). Voltage-clamp protocols and data acquisitions were performed using pCLAMP6 software (Axon Instruments, Inc., Union City, CA).
Single-Channel Recording. Cell-attached single-channel recordings were performed with a high-K+ bath solution (5 mM KCl, 112 mM potassium aspartate, 5 mM NaCl, 3 mM MgCl2, 1 mM Mg-ATP, 2 mM EGTA, 10 mM glucose, 10 mM HEPES, pH adjusted to 7.3 with KOH, at room temperature) to cancel membrane potential. The resistance of the Cylgard 184 (Dow Corning, Midland, MI)-coated, heat-polished microelectrodes was between 6 and 10 MΩ when filled with the internal solution composed of 110 mM BaCl2 and 10 mM HEPES, pH adjusted to 7.4 with TEAOH. Single-channel currents were measured using an Axopatch 200B (Axon Instruments) via analog-to-digital converter (Digidata 1200; Axon Instruments). Voltage-clamp protocols, data acquisitions, and analysis of data were performed using pCLAMP7 software (Axon Instruments).
Radioligand Binding. Cell membranes were prepared for DHP binding assay from BHK6 cells. Transfected BHK6 cells were washed two times, scraped, and homogenized using a glass-Teflon homogenizer in binding buffer (50 mM Tris, 1 mM EDTA, 100 mM phenylmethylsulfonyl fluoride, 2 mg/ml aprotinin, 100 mM benzamidine, 1 mM pepstatin A, and 1 mg/ml leupeptin, pH 7.5). The homogenate was centrifuged at 700g for 5 min. The pellet was rehomogenized and centrifuged again. The collected supernatants were centrifuged at 100,000g for 30 min. After removing the supernatant, pellet was homogenized in small volume of binding buffer, frozen in liquid nitrogen, and stored at -70°C until use. Radioligand binding studies were carried out as described previously (Ikeda et al., 1994). Membranes of BHK6 cells (approximately 0.2 mg of protein/ml) were incubated with 0.1 to 2.5 nM [3H](+)-PN200-110 in a total volume (0.2 ml) of 50 mM Tris-HCl, pH 7.5, containing 1.25 mM CaCl2 and 1.25 mM MgCl2. In the experiments with S1115A, 2.5 to 62.5 nM [3H](+)-PN200-110 was used.
The dissociation constant (Kd) and the receptor density (Bmax) were estimated by fitting the saturation curve to the equation [B] = (Bmax × F)/(Kd + F), where B is the ligand-receptor complex and F is concentration of ligand. The nonspecific binding was measured in the presence of 10 μM nicardipine and subtracted from the total binding to obtain the specific binding. A high ratio of specific to nonspecific binding of 0.1 to 2.5 nM [3H](+)PN200-110 was observed in binding experiments with rbCII and F1112A (69∼98%).
Modeling the Pore Region of rbCII. The X-ray structure of the KcsA K+ channel (Doyle et al., 1998) was used to build the homology model of the pore region of rbCII. The model involves eight transmembrane segments, S5, S6, and four P-loops (Table 1), which are not linked to each other. Analogous segments of KcsA and rbCII are aligned as shown in Table 1. The DHP receptor model (Zhorov et al., 2001) was taken as the starting approximation for building the rbCII model. The energy of the model was optimized using the Monte Carlo-minimization method as described previously (Zhorov et al., 2001).
In describing the KcsA-based three-dimensional model of rbCII, we use a general scheme of labeling residues in ion channels (Zhorov et al., 2001). In this scheme, residues that correspond to the beginning of transmembrane helices in KcsA are assigned number 1 with a prefix indicating the repeat number and segment name. For example, Thr1038 of rbCII is designated ThrIIIS5.14 because it is located in segment IIIS5 and aligns with 14th residue in M1 of KcsA (Table 1). In P-loops, selectivity-filter glutamates, which are the most conserved residues in Ca2+ channels, are assigned number 50. This marker is used to count residues in the aligned sequences. For example, Phe1112 and Ser1115 of rbCII are designated PheIIIP.44 and SerIIIP.47, respectively (Table 1). In some cases, residues are designated with both their genuine number and general label [e.g., Thr1038(IIIS5.14)].
Materials. Diltiazem (generous gift from Tanabe Seiyaku) and verapamil (purchased from Nacalai Tesque, Kyoto, Japan) were dissolved in distilled water and stored at 4°C as 1 mM stock solutions. Nitrendipine (purchased from Funakoshi Seiyaku, Tokyo, Japan), (S)-(-)-Bay k 8644, (R)-(+)-Bay k 8644, and FPL-64176 (purchased from Sigma Chemical, St. Louis, MO) were dissolved in ethanol and stored at -20°C as 3 or 30 mM stock solutions. Drugs were dissolved in the external solution and applied via concentration-clamp apparatus (Vibraspec, Inc., Philadelphia, PA) in the whole-cell, patch-clamp recording. The concentration-clamp apparatus allowed us to exchange the extracellular solution within 50 ms (Adachi-Akahane et al., 1996).
[3H](+)PN200-110 (82.0 Ci/mmol) was purchased from Amersham Biosciences (Little Chalfont, Buckinghamshire, UK). Nicardipine (purchased from Sigma Chemical) was dissolved in ethanol and stored at -20°C.
Statistical Analysis. Data are expressed as the mean ± S.E.M. Statistical significance was assessed with Student-Welch's t test or Dunnett's test and considered significant when the p value was less than 0.05.
Results
Alanine Scanning Mutagenesis of IIIS5-S6 Linker. We have previously determined Ser1115(IIIP47), located in the pore-forming region of the L-type Ca2+ channel α1C subunit (rbCII; Fig. 1, B and C) (Snutch et al., 1991), as a novel determinant of the interaction with DHPs and Ca2+ agonists (Fig. 1B) (Yamaguchi et al., 2000). Ser1115(IIIP47) is located in the N-terminal part of the P-loop. In bacterial K+ channel KcsA, an analogous part of the P-loop forms an α-helix called the pore helix (Doyle et al., 1998). We therefore examined whether amino acid residues in the pore-forming region other than Ser1115 might contribute to the DHP-action. To search for amino acid residues involved in the action of Ca2+ channel agonists, we employed alanine-scanning mutagenesis in IIIS5-S6 linker (Fig. 1, A and C). Alanine scanning was chosen for this experiment for the following reasons: 1) the N terminus of P-loop was modeled as an α-helix (Lipkind and Fozzard, 2001; Zhorov et al., 2001) and 2) substitution of Ala in the α-helix is expected to remove the amino acid side chain in each position without causing global conformational changes (Blaber et al., 1993; Peterson et al., 1997). Most of the amino acid residues around Ser1115(IIIP47), except for original alanine residues (Ala1106(IIIP38), Ala1107(IIIP39), and Ala1110(IIIP.42)), were systematically replaced by alanine (Fig. 1C). Amino acid residues just around Glu1118(IIIP50) (Fig. 1C, black box), which is supposed to serve as a Ca2+ -selective filter (Tang et al., 1993), were not replaced because their influence on Ca2+ channel kinetics is too large (Williamson and Sather, 1999). Among the alanine-substituted mutant Ca2+ channels, L1111A channel turned out to be nonconducting. The responsiveness of mutant Ca2+ channels to Ca2+ channel agonists were screened as a relative increase of Ca2+ channel currents by Ca2+ agonist FPL-64176.
FPL-64176 dramatically altered the gating kinetics of Ca2+ channel currents of rbCII and most of the mutants as has been reported (Fig. 2) (Zheng et al., 1991; Kunze and Rampe, 1992). 1) peak Ca2+ channel currents were augmented; 2) tail current duration was prolonged; 3) both activation and inactivation kinetics of Ca2+ channel currents were slowed down (Fig. 2A). In contrast, FPL-64176 exerted its effects on F1112A channel currents to significantly smaller extent (Fig. 2, A and B). Because effects of Ca2+ agonists depend on test potentials, actions of FPL-64176 on I-V relationships of F1112A channels were examined as shown in Fig. 2C. Agonistic effects of FPL-64176 on F1112A were weaker at all test voltages compared with those on rbCII (Fig. 2C), indicating that F1112A either has low binding affinity for FPL-64176 or lacks the component necessary for agonistic action. Interestingly, when Thr1113(IIIP45) (adjacent to Phe1112(IIIP44)) was replaced by Ala (T1113A), the effect of FPL-64176 (1 μM) was significantly reinforced (Fig. 2, A and B). The enhanced response to FPL-64176 in T1113A was not caused by the change of the voltage-dependence of activation because its I-V relationships were not altered (data not shown). The rest of mutant Ca2+ channels were enhanced by FPL-64176 to the same extent as that of rbCII.
Influence on the Action of Ca2+ Agonists of the Replacement of Phe in Each S5-S6 Linker of Repeats I∼IV. Phe1112(IIIP44) is six amino acids upstream from the selective filter Glu1118(IIIP50) (Figs. 1C and 3A) in the IIIS5-S6 linker. Multiple sequence alignment of different α1 subunits shows that sequence motif '- -F- - -T-E-W- -' is highly conserved among Ca2+ channels (Fig. 1C). Furthermore, this motif is also highly conserved among the four repeats (I∼IV) (Fig. 3A). Therefore, we examined the contribution of each Phe to the action of FPL-64176 by comparing mutants F357A, F700A, F1112A, and F1413A. Among the four conserved Phe residues, Phe1112(IIIP44), but not Phe357(IP44), Phe700(IIP44), or Phe1413(IVP44), turned out to be responsible for the action of FPL-64176 (Fig. 3B).
Contribution of Phe1112 to Pharmacological Properties of the Ca2+ Channel. Ca2+ channel currents of F1112A were slightly enhanced by the DHP agonist (S)-(-) Bay k 8644, but the increment was significantly smaller than that of rbCII (Fig. 3). These results indicate that F1112A is practically insensitive to Ca2+ channel agonists.
Next, we examined effects of the DHP antagonist nitrendipine on F1112A and rbCII. F1112A was significantly less sensitive to nitrendipine than rbCII (Table 2). The low sensitivity to nitrendipine was estimated by the shift of the concentration-response curve. Nitrendipine blocked rbCII and mutant Ca2+ channels in a concentration-dependent manner. As summarized in Table 2, IC50 values for the block of F1112A and S1115A by nitrendipine measured at the holding potential of -70 mV were, respectively, 5.13 and 39.4 times higher than that of rbCII. When the holding potential was depolarized from -70 to -50 mV, the block of F1112A channel by nitrendipine was enhanced. However, it was still significantly weaker than that of rbCII (data not shown). On the other hand, the block of Ca2+ channel currents through F1112A by diltiazem or verapamil was identical to that of rbCII (Fig. 5, B and D). These results indicate that Phe1112 contributes to the sensitivity of rbCII to Ca2+ channel agonists and DHP but not to BTZ or PAA.
Because changes of electrophysiological properties of Ca2+ channels by mutation may affect pharmacological properties, we analyzed kinetics of F1112A channel currents (Fig. 6). The I-V curves were not different between rbCII and F1112A (Fig. 6A). Steady state inactivation curve of F1112A was shifted to the positive voltages compared with that of rbCII (-31.8 ± 1.3 versus -36.7 ± 0.53 mV, p < 0.05; Fig. 6B). The restitution curves and recovery time constants were not significantly different between rbCII and F1112A (Fig. 6C). Because Ca2+ channel antagonists preferentially bind to the Ca2+ channel in the inactivated state, the effects of Ca2+ channel antagonists may be altered by mutations as a consequence of the alteration of gating properties of the Ca2+ channel, such as activation and inactivation. However, at -70 mV, the fraction of the inactivated channel was indistinguishable for the mutant and rbCII. Therefore, it is unlikely that the changes in the steady-state inactivation curve account for the observed loss of DHP sensitivity in F1112A. This was supported by the results that the sensitivity to block by non-DHP Ca2+ channel antagonists was not altered by mutations (Fig. 5). These results indicate that the selective impairment in F1112A of the sensitivity to Ca2+ channel agonists and DHPs were not caused by the alteration of gating properties.
Reduced Binding Affinity to DHP Antagonist [3H](+)PN200-110 of Mutant Ca2+ Channels. To clarify whether F1112A has low binding affinity for DHPs or has low efficacy for DHP action, we compared the sensitivity of mutant Ca2+ channels with that of DHP antagonist (IC50 values, Table 2) and the binding affinity of mutant Ca2+ channels to DHP antagonist estimated by radioligand binding assay (Table 3). The Kd values for [3H](+)PN200-110 binding of rbCII and mutant Ca2+ channels were estimated by the saturation curve of specific binding. Kd and Bmax of the [3H](+)PN200-110 binding to Ca2+ channels are summarized in Table 3. The relative change of Kd values was consistent with that of IC50 values, thus indicating that the reduced sensitivity of F1112A and S1115A to DHP antagonists resulted from the reduced binding affinity. The IC50 values measured at a holding potential of -70 mV was higher by 1,000∼3,000-fold than Kd values measured in membrane fractions where membrane potential was supposed to be near 0 mV (Table 2, 3). These results are explained by the voltage-dependence of the binding affinity of DHPs to Ca2+ channels (Bean, 1984).
Effects of FPL-64176 on Unitary Ca2+ Channel Currents of rbCII and F1112A. Considering that the decay rate of whole-cell Ca2+ channel currents of F1112A was not slowed by FPL-64176 (Fig. 2A, b), insensitivity of F1112A to Ca2+ channel agonists may be caused by the lack of ability to prolong mean open time. In the presence of FPL-64176, rbCII showed unitary Ca2+ channel currents with long openings (Fig. 7A). However, F1112A currents showed only short openings even in the presence of FPL-64176 at high concentrations (30 μM). In F1112A, Ca2+ channel agonist increased the open probability, but there was little enhancement of the mean open time (Fig. 7B).
Pharmacological Properties of Double Mutant Ca2+ Channel F1112A/S1115A. We have previously shown that S1115A was also less sensitive to DHP agonists and antagonists (Yamaguchi et al., 2000). Accordingly, we studied whether the mutational influences of the two amino acid residues are additive. The ratio of the IC50 value of F1112A/S1115A versus S1115A and F1112A/S1115A versus F1112A were 1.27 and 9.79, respectively (Table 2), suggesting that Ser1115 exerts larger contribution to the sensitivity to DHP than Phe1112 and that their influences are not additive.
Interestingly, effects of Ca2+ channel agonists were quite different between S1115A and F1112A/S1115A. (S)-(-)-Bay k 8644 as well as FPL-64176 decreased Ca2+ channel currents through F1112A/S1115A (Figs. 2B and 4). Therefore, Ca2+ channel agonists behaved as if they were weak Ca2+ channel antagonists in F1112A/S1115A. Then we compared effects on F1112A/S1115A of the stereoisomers (S)-(-)-Bay k 8644 and (R)-(+)-Bay k 8644, with the aim to further investigate the role of Phe1112/Ser1115 in transducing the action of Ca2+ channel agonist and antagonist. Because ethanol used as solvent in DHP stock solution produces a Ca2+ channel blocking action at concentrations higher than 0.03% (Walter and Messing, 1999), Ca2+ channel current amplitudes were normalized to the control value measured in Tyrode's solution containing vehicle before the drug application. In rbCII, Ca2+ channel currents were enhanced by (S)-(-)-Bay k 8644 and blocked by (R)-(+)-Bay k 8644 at holding potentials of -70 and -50 mV (Fig. 8). At -50 mV, the enhancement of Ca2+ channel currents through rbCII by (S)-(-)-Bay k 8644 was smaller than that measured at -70 mV. In sharp contrast, Ca2+ channel currents through F1112A/S1115A were slowly inhibited by agonist (S)-(-)-Bay k 8644 and recovered to the control level on washout (data not shown). (R)-(+)-Bay k 8644 also produced the block of Ca2+ channel currents to the same degree with the same time course and the voltage dependence as those of (S)-(-)-Bay k 8644. The blocking action of (R)-(+)-Bay k 8644 was significantly weaker in F1112A/S1115A compared with that of rbCII (Fig. 8), as was observed with nitrendipine (Table 2). These results indicate that F1112A/S1115A could not distinguish between the stereoisomers of Ca2+ agonist and antagonist.
Interestingly, FPL-64176 also behaved as an antagonist on F1112A/S1115A (Fig. 2B). Unitary currents through F1112A/S1115A were neither enhanced nor prolonged by FPL-64176 (Fig. 7C). These results indicate that F1112A/S1115A has completely lost the ability to be activated by Ca2+ channel agonists. In other words, Ca2+ channel agonists block F1112A/S1115A.
Importance of Aromatic Side Group of Phe1112 on the Action of Ca2+ Channel Agonists. We showed that the replacement of Phe1112 in IIIS5-S6 linker of Ca2+ channel by Ala reduced the binding affinity for DHPs and sensitivity to Ca2+ channel agonists. To elucidate the role of aromatic side group of Phe1112, we introduced Tyr in place of Phe1112. Effects of the mutation were evaluated by comparison of effects of FPL-64176 (Fig. 9A) on peak ICa (Fig. 9B) and the amount of Ca2+ entry through the Ca2+ channel (∫ICa) (Fig. 9C) between rbCII and mutant Ca2+ channels. The amount of Ca2+ entry through the Ca2+ channel, ∫ICa, was estimated by integrating ICa during test potential. The increase of ∫ICa represents the prolongation of mean open time as well as the enhancement of open probability. Peak ICa and ∫ICa of rbCII were markedly enhanced by FPL-64176 in a concentration-dependent manner. The relative enhancement of ∫ICa was larger than that of peak amplitude of ICa. In contrast, in F1112A, peak ICa and ∫ICa were only slightly increased by FPL-64176 at 10 μM. Interestingly, the peak ICa through F1112Y was slightly enhanced by FPL-64176 to the extent similar to that of F1112A (Fig. 9B), but the increment of ∫ICa was significantly larger than that of F1112A (Fig. 9C). The enhancement of peak ICa of rbCII by FPL-64176 at 1 μM was almost equivalent to the enhancement of F1112Y and F1112A by FPL-64176 at 10 μM. However, when the three bar graphs were compared (rbCII with FPL-64176 at 1 μM, F1112Y with 10 μM, and F1112A with 10 μM), the enhancement of ∫ICa was significantly larger in rbCII and F1112Y than in F1112A. These results were consistent with results of single-channel experiments, where the prolongation of the mean open time by FPL-64176 was significantly larger in rbCII and F1112Y than in F1112A (Fig. 7). On the other hand, the increment of peak amplitude of ICa and ∫ICa was smaller in F1112Y than in rbCII, thus indicating that the replacement of Phe1112 by Tyr had reduced the sensitivity to Ca2+ channel agonists. Nevertheless, these results suggest that the aromatic ring of Phe1112 plays a critical role in the prolongation of the open time produced by FPL-64176.
Modeling DHP-rbCII Binding. In a three-dimensional model, orientation of residues depends dramatically on the alignment between KcsA and rbCII. The alignment shown in Table 1 yields a model in which several residues whose mutation affects DHP binding (DHP-sensing residues) form a ligand-binding pocket in the interface between repeats III and IV (interface III/IV). A DHP ligand bound in this pocket is partially exposed to the central pore (Fig. 3B in Zhorov et al., 2001). To assess whether residues in the IIIS5-S6 linker can contribute to the ligand binding, a typical DHP antagonist, nifedipine, was docked from a hundred of randomly generated starting positions and orientations, the center of distribution being in the III/IV interface. During each docking, the energy was Monte Carlo-minimization as described previously (Zhorov et al., 2001).
Forty lowest-energy structures collected from these trajectories show that nifedipine can populate large areas in the central pore, in the III/IV interface, and between the pore and the III/IV interface (Fig. 10A). Interestingly, flexible DHP-sensing residues TyrIIIS6.10, MetIIIS6.18, TyrIVS6.11, and MetIVS6.12 can interact with nifedipine bound in the pore and in the III/IV interface. According to calculations, nifedipine binding in the pore is energetically preferable. It should be noted, however, that the current model does not include segments S1 to S4, some of which could enhance DHP binding in the III/IV interface.
Figure 10B shows possible ligand-binding pockets in the III/IV interface formed by 10 DHP-sensing residues: Ser1115(IIIP.47), Phe1112(IIIP.44), ThrIIIS5.14, GlnIIIS5.18, TyrIIIS6.10, IleIIIS6.14, MetIIIS6.18, TyrIVS6.11, MetIVS6.12, and IleIVS6.19. Calculations predict several essentially different orientations of nifedipine in this pocket. No experimental data are currently available to favor one of the orientations. DHP-sensing residues IleIIIS6.11, MetIIIS6.19, IleIVS6.18, and AsnIVS6.20, as well as Ca2+ ion coordinated by the selectivity-filter glutamates, do not contribute to the DHP-binding in the III/IV interface. Some DHP-sensing residues are too far from each other and cannot bind simultaneously to a ligand. For example, the distance between Cβ_PheIIIP.44 and Cβ_IleIVS6.19 (18.3 Å) is significantly larger than the maximal possible distance between nifedipine atoms (11.4 Å) (Zhorov and Ananthanarayanan, 1996).
Discussion
Alanine-Scanning Mutagenesis of IIIS5-S6-Linker. In this study, we present a novel insight into the molecular mechanism underlying modulation of gating kinetics of L-type Ca2+ channels by DHP and non-DHP Ca2+ channel agonists. By use of alanine-scanning mutagenesis, we identified amino acid residue Phe1112(IIIP44), in addition to Ser1115(IIIP47), as an important determinant of DHP binding and action of Ca2+ channel agonist (Fig. 2). The requirement of Phe1112(IIIP44) as well as Ser1115(IIIP47) in the IIIS5-S6 linker for the action of Ca2+ channel agonists highlights the important role of the pore helix in the modulation of the gating kinetics by Ca2+ channel agonists.
Contrary to F1112A, the sensitivity of T1113A to Ca2+ channel agonists was enhanced (Fig. 2), although T1113A channel was inhibited by DHP antagonist in a manner similar to that of rbCII (data not shown). The substitution of Thr1113(IIIP45) by Ala may have turned the side chain of adjacent Phe1112(IIIP44) to the favorable arrangement for Ca2+ channel agonists to exert their action or may have reduced the steric hindrance for the access of Ca2+ channel agonists.
Is the insensitivity of Phe1112(IIIP44) (and Ser1115(IIIP47)) to Ca2+ channel agonists attributed to the reduction of the binding affinity? To answer this question, we compared the sensitivity and the binding affinity of mutant Ca2+ channels to DHP antagonists. As summarized in Tables 2 and 3, the relative differences of sensitivity and binding affinity to DHP antagonists between rbCII and F1112A were comparable. From the comparison between two independent experiments, we concluded that the low sensitivity of Ca2+ channel currents of F1112A or S1115A to DHP antagonists is compatible with the reduced binding affinity to them. From these experiments, it could be anticipated that the binding affinity of F1112A or S1115A for DHP agonists was decreased to the same extent as that for DHP antagonists and that the insensitiveness to DHP agonists could have been caused by the reduction of binding affinity.
The prolongation of open time by FPL-64176, assessed by ∫ICa and mean open time (Figs. 7 and 9C), was retained in rbCII and F1112Y but was mostly absent in F1112A, thus indicating that the aromatic side chain of Phe1112(IIIP44) is required for producing full agonistic effects by FPL-64176. If this functional group is missing, Ca2+ agonists may not be able to stabilize the open states.
Synergistic Effects of Phe1112(IIIP44) and Ser1115(IIIP47) to the Action of DHPs. If Phe1112(IIIP44) (or Ser1115(IIIP47)) simply serves as a binding pocket for DHP, Ca2+ currents through the double-mutant Ca2+ channel F1112A/S1115A would be slightly enhanced (although weaker than through the single-mutant channels) or would be unaffected by Ca2+ agonists. However, effects of Ca2+ agonists on F1112A/S1115A were quite distinctive. Unexpectedly, Ca2+ agonists decreased rather than increased Ca2+ currents through F1112A/S1115A (Figs. 2, 4, and 8). The weak blocking action of the Ca2+ channel agonists on F1112A/S1115A conceivably reflects the interaction between Ca2+ agonists and F1112A/S1115A, even though the binding affinity may be somewhat lower. Furthermore, the single channel currents through F1112A/S1115A were not prolonged at all by FPL-64176 (Fig. 7). It is likely that, in F1112A/S1115A, the interaction with Ca2+ agonists is not transmitted to the Ca2+ channel agonistic action. Importantly, in the double mutant F1112A/S1115A, Ca2+ agonists behaved as if they were weak Ca2+ antagonists. Furthermore, F1112A/S1115A was not able to distinguish between (S)-(-)-Bay k 8644 and (R)-(+)-Bay k 8644. Thus, the two amino acid residues seem to serve as a decisive factor for agonist/antagonist actions of Ca2+ channel modulators. We propose that Phe1112(IIIP44) and Ser1115(IIIP47) have two functions: 1) they are part of a binding pocket for DHPs and 2) they are the key residues for transducing the binding of Ca2+ channel agonists into the agonistic action, such as stabilization of Ca2+ channels in the open state.
Under certain conditions (e.g., at very high concentrations or at depolarizing voltages), DHP agonists exert antagonistic effects (Adachi-Akahane and Nagao, 2000). Thus, the conformational state of the α1C subunit seems to be a major factor in determining the agonist/antagonistic action of DHPs (Triggle and Rampe, 1989). In the present study, we demonstrated that substitution of Phe1112(IIIP44) and Ser1115(IIIP47) with Ala results in Ca2+ channel agonists exhibiting effects of weak antagonists. Therefore, in the wild-type channel, the interaction of Ca2+ agonists with Phe1112(IIIP44) and Ser1115(IIIP47) may be disrupted at the depolarized voltages.
Contribution of Phe1112(IIIP44) and Ser1115(IIIP47) to the DHP-Binding Pocket. Most of the binding sites for Ca2+ channel agonists and antagonists have been determined by comparison between α1C and α1A sequences. Therefore, the amino acids highly conserved between DHP- and non-DHP-sensitive Ca2+ channel α1 subunits have been excluded. However, some amino acid residues common between α1C and α1A have been identified as critical binding sites (Bodi et al., 1997; Peterson et al., 1997; Wappl et al., 2001). In fact, Phe1112(IIIP44) and Ser1115(IIIP47) are conserved in high-voltage activated Ca2+ channels. Phe residues in position P44 are highly conserved in the P regions of voltage-dependent Ca2+ and Na+ channels. However, PheIP44, PheIIP44, and PheIVP44 seem unrelated to Ca2+ channel agonist action (Fig. 3).
Mutation of PheIIIP49 preceding the selectivity-filter GluIIIP50 reportedly affects the DHP binding (Peterson and Catterall, 1995). In this study, PheIIIP49 was excluded from targets for alanine-scanning mutagenesis (see Results).
DHP Binding Model. Earlier mutational studies revealed DHP-sensing residues in segments IIIS5, IIIS6, and IVS6 (Table 1). In addition, Ca2+ coordination by the selectivity-filter glutamates is important for DHP binding (Mitterdorfer et al., 1995; Peterson and Catterall, 1995). Present study indicates that mutations of Ser1115(IIIP47) and Phe1112(IIIP.44) also affect DHP binding.
The X-ray structure of a bacterial K+ channel KcsA was used as a template to build homology models of L-type Ca2+ channel that visualize DHPs bound in the III/IV interface (Huber et al., 2000) and in the pore (Zhorov et al., 2001). Finding of DHP-sensing residues in the pore helix of segment IIIP provides additional constraints for modeling the DHP-binding site(s). The alignment shown in Table 1 yields a model in which Ser1115(IIIP47), Phe1112(IIIP.44), and several other DHP-sensing residues can contribute to DHP binding in the III/IV interface.
DHP agonists and antagonists were suggested to exert their effect from the pore-binding site (Zhorov et al., 2001). Mutational and modeling data of the present study do not rule out this possibility but suggest that the III/IV interface site is also important. Interestingly, several DHP-sensing residues, including TyrIIIS6.10 and TyrIVS6.11, critical determinants of DHP binding, can interact with DHP ligand in the III/IV interface as well as in the pore (Fig. 10A).
The fact that double mutant F1112A/S1115A is blocked rather than activated by agonist (S)-(-)-Bay k 8644 seems inconsistent with the idea that the agonist binds inside the pore. One possibility to resolve this conflict is to suggest that DHPs bind in the III/IV interface on their way from the extracellular space to the pore. In the energetically preferable conformation, the DHP ring has a flattened-boat form. Enantiomers of Bay k 8644 have different substituents at the port side of the boat: the agonist has a hydrophilic NO2 group, whereas antagonist has a hydrophobic COOMe group. The effect of a DHP ligand would depend on how it approaches the hydrophobic gate at the S6 crossing: if near the hydrophobic group, it would act as the antagonist; if by the hydrophilic group, it would act as the agonist (Zhorov et al., 2001). Once inside the channel, the bulky DHP ligand cannot flip-flop. If the DHP agonist (S)-(-)-Bay k 8644 enters the III/IV interface by the port-side hydrophilic group forward, it would approach the gate by this group and would act as the agonist. If the agonist enters the interface with the starboard hydrophobic group forward, it would approach the gate by this group and would act as the antagonist. According to calculations, the port-side-down and starboard-side-down orientations of a DHP ligand in the pore are almost equal in energy (Zhorov et al., 2001). In the wild-type channel, Phe1112(IIIP44) and Ser1115(IIIP47) may stabilize the port-side-down orientation of the Bay k 8644 enantiomers in the III/IV interface, thus predetermining the ligand orientation in the pore and hence the type of its activity. The double mutant would not stabilize the port-side-down orientation of the ligand in the III/IV interface. The agonist (S)-(-)-Bay k 8644 would enter the interface by its starboard hydrophobic group forward simply because Ala1112(IIIP44) and Ala1115(IIIP47) in the double mutant make a more hydrophobic entrance than Phe1112(IIIP44) and Ser1115(IIIP47) in the wild-type channel. As the result, the agonist would approach the S6 crossing by the hydrophobic starboard group and would act as the antagonist.
Thus, the KcsA-based model of rbCII cannot visualize all known DHP-sensing residues in direct contacts with a DHP molecule. This suggests two possibilities: 1) mutations of some DHP-sending residues affect ligand binding indirectly, and 2) the effect of DHPs on the channel gating depends on their binding to more than one site in the pore-forming subunit. Further experimental and theoretical studies are necessary to explore these possibilities.
In conclusion, the present study postulates two possibilities for the selective loss of sensitivity to Ca2+ channel agonists and DHPs: 1) Phe1112(IIIP44) and Ser1115(IIIP47) are binding sites for DHPs or 2) Phe1112(IIIP44) and Ser1115(IIIP47) are required for the transmission of the drug binding into its action as Ca2+ channel agonists (Fig. 10). Elucidation of the role of Phe1112(IIIP44) and Ser1115(IIIP47) in the modulation of Ca2+ channel gating by Ca2+ channel agonists would be helpful for understanding the role of the pore-forming region in the modulation of Ca2+ channel gating by Ca2+ channel modulators and refining the three-dimensional structure model of the pore-forming region of L-type Ca2+ channel α1C subunit.
Acknowledgments
We are grateful to Dr. T. P. Snutch for generous gift of rbCII and Dr. Y. Mori for BHK6 cells. We thank Tanabe Seiyaku Co. Ltd. for generous supply of d-cis-diltiazem.
Footnotes
-
This work was supported by a Grant-in-Aid for Scientific Research from the Japanese Society for the Promotion of Science and by a grant to BSZ from the Canadian Institutes of Health Research (CIHR). B.S.Z. is a recipient of the CIHR Senior Scientist award.
-
This work is part of the Ph.D. thesis of S.Y.
-
This work was presented in part as an abstract [Yamaguchi S and Adachi-Akahane S (2002) Role of the IIIS5-S6 linker of L-type Ca2+ channel α1C subunit in the action of Ca2+ channel agonist. Biophys J 82:103A].
-
ABBREVIATIONS: DHP, dihydropyridine; BTZ, benzothiazepine; PAA, phenylalkylamine; Bay k 8644, 1,4-dihydro-2,6-dimethyl-5-nitro-4-(2- [trifluoromethyl] phenyl)-3-pyridine carboxylic acid methyl ester; FPL-64176,2,5-dimethyl-4-[2- (phenylmethyl) benzoyl]-1H-pyrrole-3-carboxylic acid methyl ester; rbCII, rat brain Ca2+ channel α1C subunit type II; GFP, green fluorescence protein; I-V, current-voltage.
- Received April 7, 2003.
- Accepted April 28, 2003.
- The American Society for Pharmacology and Experimental Therapeutics