Beta subunit coexpression and the alpha1 subunit domain I-II linker affect piperidine block of neuronal calcium channels

The effects of local anesthetics were examined on a family of transiently expressed neuronal calcium channels. Fomocaine, a local anesthetic containing a morpholine ring, preferentially blocked alpha1E channels (Ki = 100 microM), and had a lower affinity (3- to 15-fold) for alpha1A, alpha1B, and alpha1C channels. Block was incompletely reversible, followed 1:1 kinetics, and did not affect steady-state inactivation properties. Fomocaine block was sensitive to the concentration of permeant ion and enhanced in the presence of external pore blockers, suggesting a site of action in the conducting pathway. Flecainide, which carries a piperidine ring, and the diphenylbutylpiperidine antipsychotic, penfluridol, caused qualitatively similar block, suggesting that morpholine rings are compatible with the piperidine receptor site. In contrast, procaine, which contains an alkyl chain, caused reversible low affinity block of the different calcium channels (Kd values between 2 and 5 mM) and was least effective on alpha1E and did not compete with fomocaine, suggesting that local anesthetics interact with at least two distinct receptor sites. Compared to coexpression with the Ca channel beta1b subunit, block at the piperidine receptor site was significantly weakened with the beta2a subunit suggesting that the nature of the beta subunit contributes to drug binding. Amino acid changes in the cytoplasmic linker between domains I and II resulted in decreased fomocaine and penfluridol blocking affinity. Furthermore, the blocking affinity observed with alpha1B, was conferred on alpha1A by substitution of the domain I-II linker of alpha1B into alpha1A. Taken together, the data suggest that beta subunit binding and the domain I- II linker contribute to the piperidine receptor site on neuronal calcium channels.

Biotechnology Laboratory, University of British Columbia, Vancouver, British Columbia, Canada V6T 123 The effects of local anesthetics were examined on a family of transiently expressed neuronal calcium channels. Fomocaine, a local anesthetic containing a morpholine ring, preferentially blocked (Y,~ channels (Ki = 100 PM), and had a lower affinity (3to 15fold) for (Y,~, (Y,~, and cz,c channels. Block was incompletely reversible, followed I:1 kinetics, and did not affect steady-state inactivation properties. Fomocaine block was sensitive to the concentration of permeant ion and enhanced in the presence of external pore blockers, suggesting a site of action in the conducting pathway. Flecainide, which carries a piperidine ring, and the diphenylbutylpiperidine antipsychotic, penfluridol, caused qualitatively similar block, suggesting that morpholine rings are compatible with the piperidine receptor site. In contrast, procaine, which contains an alkyl chain, caused reversible low affinity block of the different calcium channels (K,, values between 2 and 5 mM) and was least effective on (Y, E and It is well established that local anesthetics are effective blockers of neuronal, cardiac, and skeletal muscle sodium channels (Weidmann, 1956;Strichartz, 1973;Courtney, 1975;Hondeghem and Katzung, 1977;Bean et al., 1983;Butter-worth and Strichartz, 1990;Zamponi et al., 1993a-c). Local anesthetics also affect several members of the potassium channel family (Camm et al., 1992;Yamashita et al., 1995) nicotinic ACh receptors, as well as calcium release channels from the sarcoplasmic reticulum. However, the effects of local anesthetics on voltage-gated calcium channels have been poorly described, and neither the structural requirements for calcium channel block nor the specific site(s) of action of these compounds has been identified. Whereas the archetypal local anesthetics lidocaine and procaine appear to be poor inhibitors of voltage-gated calcium channels (Ki -2-5 mM; Johansen and Kleinhaus, 1985;Sugiyama and Muteki, 1994) there have been several reports of potent block (Ki between 50 and 300 PM) of both cardiac and neuronal calcium channels by the local anesthetic bupivacaine (Sanchez-Chapula, 1988;Guo et al., 1992;Sugiyama and Muteki, 1994;Wulf et al., 1994). Bupivacaine exhibits an obvious structural difference compared to procaine and lidocaine in that the tertiary amine-containing portion-a did not compete with fomocaine, suggesting that local anesthetics interact with at least two distinct receptor sites. Compared to coexpression with the Ca channel filb subunit, block at the piperidine receptor site was significantly weakened with the Pna subunit suggesting that the nature of the p subunit contributes to drug binding. Amino acid changes in the cytoplasmic linker between domains I and II resulted in decreased fomocaine and penfluridol blocking affinity. Furthermore, the blocking affinity observed with 01,~ was conferred on alA by substitution of the domain I-II linker of (~,s into (Y,~. Taken together, the data suggest that p subunit binding and the domain I-II linker contribute to the piperidine receptor site on neuronal calcium channels. Key words: local anesthetics; Xenopus oocytes; functional expression; barium current; fomocaine; penfluridol; procaine critical structure for block of both sodium (Zamponi and French, 1994) and potassium channels-forms part of a piperidine ring rather than an alkyl chain (see Fig. 1). Similar ring structures that include tertiary amines also form the core of a range of calcium channel-blocking antipsychotic agents, including penfluridol (see Fig. 1) (Galizzi et al., 1986;Czuczwar et al., 1992;Enyeart et al., 1992;Sah and Bean, 1994) and the existence of a specific binding site for piperidine-based compounds has been proposed (King et al., 1989) (for review, see Glossmann and Striessnig, 1990;Kaczorowski et al., 1994;Striessnig et al., 1994).
To test whether local anesthetics containing piperidine-based or similar structural features interact with neuronal calcium chan-  nels, we compared the blocking action of procaine with that of fomocaine (a morpholine-based local anesthetic) on four types of cloned neuronal calcium channels transiently expressed in Xenopus oocytes. The results suggest the existence of two distinct receptor sites for local anesthetics, one that weakly interacts with procaine, and a second higher-affinity receptor that binds piperidine-and morpholine-based compounds. The data are consistent with the notion that the procaine-binding site is directly accessible from the extracellular side, whereas block by compounds such as fomocaine likely occurs at the piperidine receptor site on the intracellular side of the channel. We also present evidence indicating that piperidine binding is affected both by /3 subunit coexpression and by amino acid substitutions in the (or subunit domain I-II linker.
Voltage-clamp and data analysis. Two-electrode voltage-clamp experiments were carried out using an Axoclamp 2A amplifier (Axon Instruments, Foster City, CA) linked to an IBM-compatible PC equipped with pClamp version 5.5 software (Axon Instruments). Microelectrodes were filled with 3 M CsCl and showed typical resistances from 0.5 to 2.5 MR. BAPTA (lo-30 nl from a 100 mM stock) was injected into each oocyte during recording to suppress leak currents carried by endogenous calcium-activated chloride channels (Charnet et al., 1994). The oocytes were bathed in 2 mM BaCl,, 36 mM tetraethylammonium chloride (TEA), 2 IIIM CsCl, 5 mM niflumic acid, 38 IIIM sucrose, 20 PM 5-nitro-2-(3phenylpropylamino) benzoic acid (NPPB), 5 mM HEPES, pH 7.6, or in 10 mM BaCl,, 36 IIIM TEA, 2 mM CsCl, 5 mM niflumic acid, 30 mM sucrose, 20 FM NPPB, 5 mM HEPES, pH 7.6. Drugs were dissolved in the appropriate recording solution before application and perfused into the bath. Fomocaine (Aldrich, Milwaukee, WI) was dissolved in ethanol to make a stock solution of 500 mM fomocaine; procaine (a kind gift from Dr. Robert French), piperazine (Sigma), and diethylcarbamazine (Sigma) were dissolved directly in the recording solution to give a stock of 500 mM; flecainide (a gift from Dr. Robert French) was dissolved in ethanol as a 250 mM stock, penfluridol (a kind gift from Janssen Pharmaceuticals, Berse, Belgium) was dissolved in ethanol at a stock concentration of 200 mM. The chemical structures of the compounds used in this study are depicted in Figure 1. o-Conotoxin GVIA was dissolved in water to make a stock of 200 pM. The final ethanol concentration in the bath during drug application did not cause significant block of the currents as verified by application of ethanol alone (n = 5). Unless otherwise stated, currents were elicited by a 0.033 Hz train of 400 msec pulses from a holding potential of -100 to + 10 mV. In most cases block fully developed within 60 set of drug application. Unless otherwise stated, fomocaine and flecainide were allowed to equilibrate for 90 set, whereas procaine, piperazine, and diethylcarbamazine were applied for 30-60 set before determining their blocking actions. Currents were allowed to stabilize before drug application to minimize contributions from run-up and run-down. Data were filtered at 1 kHz and recorded directly on the hard drive of a personal computer. In most cases, leak subtraction was carried out on line using a p/5 protocol. Peak currents were analyzed using is the reversal potentyal, G is the maximum slope conductance, and S reflects the steepness of the activation curve and is an indication of the apparent gating charge movement. Inactivation curves were constructed by normalizing the currents after a prepulse to those in the absence of a prepulse and then fit to the equation: Z = l/(1 + exp(V, -V&S), where V,,, is the holding potential, V, is the half-inactivation potential, and Z is the normalized peak current. Competition experiments were analyzed as follows. Direct competition between two compounds is represented by the scheme:

RESULTS
Fomocaine preferentially blocks the qE channel The dose-response curves were well fit by simple hyperbolas suggesting a 1:l drug-channel interaction. The IC,, values determined from the fits were 95 wM, 334 FM, and 777 pM for aiE, a,c, and a1A.a channels, respectively. Fomocaine application (up to 600 PM) did not result in any significant changes to the steadystate activation curves of the four calcium channel types examined (not shown).
The piperidine-based drug flecainide also blocked the neuronal Ca channels in a qualitatively similar manner to that for fomo-Caine (e.g., subtype specificity and degree of reversibility) but with a somewhat lower affinity for (urn. (IC,,: ~lrn + Plb + oZ, 320 PM, n = 7; ale + &, + CQ, 380 PM, n = 4; alA.a + &, + az, 630 /AM, IZ = 4; not shown). Also similar to that for fomocaine, the effect of flecainide was partially reversible. Overall, the relative order of potencies are consistent with previous reports indicating that piperidine-based antipsychotics preferentially block lower threshold calcium channels (Galizzi et al., 1989;Enyeart et al., 1992).
Fomocaine block is sensitive to the concentration of permeant ion As evidenced from Figure 3, raising the external barium concentration from 2 to 10 mM resulted in a substantial (-6-to lo-fold) decrease in the fomocaine blocking affinity for (~r~.~ and (~,n in the absence and presence of fomocaine (in 2 InM barium saline). The records were filtered at 1 kHz and were elicited by stepping from -100 to +lO mV. channels. The ICsc values obtained from the fits were or*+,: 2 mM Ba = 777 PM, 10 mM Ba = 4953 FM; a& 2 mM Ba = 95 pM, 10 mM Ba = 916 PM. These data suggest that fomocaine may act at a region of the channel accessible to external barium ions. Barium ions might act by competitively inhibiting fomocaine binding at the drug site. Alternatively, if fomocaine physically occludes the pore from the intracellular side then permeating barium ions might repel fomocaine accessing its binding site and result in an increase in the IC,, of fomocaine block as has been proposed for local anesthetic action on sodium channels (Cahalan and Almers, 1979;Wang, 1988;Zamponi and French, 1993;Zamponi et al., 1993a).

Blocking effects of procaine
In contrast to that for the piperidine-based fomocaine and flecainide, the local anesthetic procaine only weakly blocked the transiently expressed neuronal calcium currents. For example, Figure 4 shows that procaine at concentrations as high as 2 RIM only resulted in -25% block of the aiAea current. A further distinction was that unlike block by fomocaine, the block by procaine was readily reversible after washout. In addition, the subtype specificity of procaine action was distinct from that of fomocaine and flecainide with oic being the most sensitive channel subtype followed by Olin > (~i*.~ > aiE (Fig. 4). Application of 6 mM diethylamine (pKa = 10.5; Zamponi and French, 1993) or 6 mM TEA did not result in any significant effect on the macroscopic currents suggesting that procaine does not act by screening diffuse surface charges (not shown). This result also suggests that the diethylamine tail of procaine is not the only structural component that determines blocking affinity and is consistent with the more potent action of the procaine analog tetracaine on neuronal calcium channels (Sugiyama and Muteki, 1994).
As with fomocaine, increasing the concentration of external barium ions from 2 mM to 10 mM resulted in a decreased blocking affinity. In five experiments the effect of 2 mM procaine was studied under 2 mM and 10 mM barium on the same oocyte and the degree of procaine block of (~i~.~ decreased from 24 2 6 to 13 -C 1% at the higher barium concentration (data not shown). Although these values are significantly different (p < 0.0013, paired t test) and reflect an approximately twofold increase in IC,,, the degree of change is much smaller than that observed with fomocaine (Fig. 3). Overall, the qualitative and quantitative differences between block by fomocaine and procaine are suggestive of distinct mechanisms of block on neuronal calcium channels.
Structural requirements for drug action and interactions between fomocaine and procaine A more detailed investigation into the mechanisms of action of fomocaine and procaine was carried out on OI,~.~ because this channel type gives the most robust expression in oocytes (Stea et al., 1994). To define the drug structural requirements for block, we examined the effects of piperazine and diethylcarbamazine, antihelmintic agents used in the treatment of conditions such as filiarisis (Gelband, 1994). Structurally, piperazine closely resembles the amine-containing portions of fomocaine and flecainide (Fig. l), and is essentially permanently uncharged at a pH of 7.6 (pKa = 4.2). Application of piperazine at concentrations between 2 and 6 mM had little effect on oiA+, + (Ye + /3,,, (n = 6), (~,c + t Figure 3. Dose-response curves for fomocaine block. alA.. + p4 + cy2 (4, PIE + &, + a2 (B), and (Y,~ + Plb + (Ye (C) in 2 mM (solid symbols) or 10 mM barium (open symbols). The results indicate that ~l,n channels are more potently blocked than both (pi= and LY~*., and that the external barium concentration significantly reduces the blocking affinity. The data were fitted with the equation Z/Zcdrug freej = l/( 1 + [F]IIC,,), where I and Zcdrug rreej are, respectively, the peak currents measured in the presence and the absence of fomocaine, [F] is the fomocaine concentration, and IC,, is the concentration at 50% current inhibition. The data are well described by a Hill coefficient of 1, suggesting a 1:l interaction between the channel and the drugs. . Current traces obtained from the major types of neuronal calcium channels in the absence and presence of 2 mM procaine. Procaine blocks (or*+ and (tic more effectively than alE channels. Note the complete reversibility of procaine block (A, C, D). The experimental conditions are as described in Figure 2. a2 + &b tn = 3), Or %E + o2 + P,,, (n = 4; data not shown). Although it is possible that other portions of local anesthetics are also required for a high-affinity interaction, these results are consistent with the notion that the protonated species is required for block.
Diethylcarbamazine is essentially a hybrid between the aminecontaining portions of procaine and fomocaine. Diethylcarbamazine contains three amino groups: one is methylated and has a pKa of 7.3 and the other two have permanently protonated groups (pKa > 12; see Fig. 1). This compound would not be expected to permeate cell membranes easily. Application of 2 mM diethylcarbamazine to oocytes expressing oIAMa + 01~ + p4 resulted in a rapidly developing, fully reversible block to -60% of the control level (Fig. 54). Decreasing the pH to 7.1 from 7.6 (which increases the concentration of drug with a protonated methyl-amino group) did not significantly affect block (n = 3; data not shown). This result suggested that the structure crucial for diethylcarbamazine block is more likely to be the diethylated amino group that resembles the head group of procaine, rather than the portion of diethylcarbamazine that resembles the amine-containing portion of fomocaine. If this assumption is correct, diethylcarbamazine and procaine would be expected to compete for binding whereas diethylcarbamazine and fomocaine would not compete. To test this hypothesis, a series of coapplication experiments were performed. Consistent with a direct competition model, Figure 54 shows that coapplication of 2 mM procaine and 2 mM diethylcarbamazine resulted in only a 10% increase in block compared with that for diethylcarbamazine alone. Because diethylcarbamazine is essentially membrane impermeant this result implies that procaine is also likely to act on the extracellular side. In contrast, coapplication of 600 pM fomocaine and 2 mM diethylcarbamazine resulted in a potentiated block that exceeded the predictions for both direct competition and independence (see Materials and Methods; Fig. 5A) and supports the hypothesized existence of separate receptors for procaine and fomocaine. Similarly, coapplication of procaine and fomocaine produced block that exceeded predictions for direct competition between the two compounds (Fig. 5B). The results indicate that procaine and fomocaine do not compete at a common blocking site.
A possible mechanism of enhanced fomocaine block in the presence of diethylcarbamazine was suggested by the coapplication of nickel and fomocaine. Figure 5C shows that application of Zamponi et al. l  600 PM fomocaine resulted in an -35% reduction in the peak alAma current and that coapplication of 1 mM nickel and 600 PM fomocaine eliminated all of the current. Of particular note, washout of the nickel with a solution containing 600 p,M fomocaine revealed an -100% increase in the degree of fomocaine block (n = 5; Fig. 5C). Because the level of fomocaine block stabilizes rapidly ( Fig. 2A), the enhanced block is unlikely attributable to longer exposure to fomocaine. Furthermore, nickel block is completely reversible (Fig. 5D) and suggests that the enhanced fomo-Caine block is most likely attributable to an effect of nickel on permeation. Because barium ions appear to inhibit fomocaine binding (Fig. 3), one possible explanation is that external nickel ions physically occlude barium permeation resulting in an increased fomocaine block. In this scenario, because fomocaine block is only weakly reversible, the additional fomocaine block developing in the presence of nickel ions persists after nickel removal. This qualitative bahavior might be expected given the high sensitivity of fomocaine block to barium flux through the channel. If our interpretation is correct, these data suggest that fomocaine must act from the cytoplasmic side, consistent with the lack of competition between the membrane impermeant diethylcarbamazine molecule and fomocaine. Furthermore, such a mechanism might account for the potentiation of fomocaine block by diethylcarbamazine.
Overall, the data suggest the presence of at least two separate local anesthetic receptor sites on neuronal calcium channels: a low-affinity site that is accessible directly from the extracellular side and that binds compounds such as procaine, and a second, higher affinity site that interacts with piperidine and morpholinebased compounds on the intracellular side (see also below). In addition, the degree of block by 600 pM fomocaine is reduced compared with that obtained when the channels are coexpressed with &,, &, and p4 subunits (0). Experimental conditions were as described in Figure 2. -Fomocaine block is sensitive to the type of p subunit present To investigate the effects of p subunits on fomocaine block, we coexpressed (Y,~.~ with four different neuronal /3 subunits (p,,, &,, &, and j3J. There were no discernible differences between fomocaine block of a,,+:, channels in the absence of a p subunit or when coexpressed with either P,,,, pa, or p4 (Fig. 6). In contrast, coexpression with &, significantly decreased the blocking affinity of fomocaine. Similar decreases in fomocaine block were also observed when (~,n and (Y,~ subunits were coexpressed with Pza (Fig. 6). The &, subunit has been shown to significantly reduce the degree of calcium channel inactivation (Stea et al., 1994) (Fig.  6). To investigate the possibility that the effect on fomocaine block may arise from secondary affects on inactivation properties, steady-state inactivation profiles were determined in the presence and absence of fomocaine. Fomocaine application did not significantly affect (Vh(contro,j = -61.1 + 2.9 mV, I/hoomocainej = -60 ? 3.6 mV, n = 5) steady-state inactivation (Fig. 7), consistent with previous reports of flunarizine action on native calcium channels (Takahashi and Akaike, 1991). The data suggest that the effect of PZa on fomocaine block does not arise secondarily from removal of voltage-dependent inactivation. This notion is supported by the observation that ollc shows little voltage-dependent inactivation regardless of the type of /3 subunit coexpressed; however, the Pza still reduced the fomocaine blocking affinity (see Fig. 6).

Amino acid substitutions in the (Y, subunit domain I-11 linker affect fomocaine block
Recently, several variants of the OI,*., subunit that result from alternative splicing have been identified (  I  I  I  I  I  I  I  I   alAa, 84, ~2, 2 mM Ba, n=5 -2 E 0.6 -2 $j 0.6a?
; 5  and voltage-dependent properties essentially identical to those observed for alA.a + (Ye + & (Soong and Snutch, unpublished observations). However, as shown in Figure 8, fomocaine block of CK~*.~ was significantly less pronounced than block of alAea (20 ? 2%, 12 = 7). These data, together with the fact that the region for p subunit binding has been localized to the I-II linker (Pragnell et al., 1994), suggest that a common mechanism may underlie the antagonistic effects of PZa and amino acid substitutions in position 4711472 on fomocaine block. To further investigate this possibility, we tested a chimeric channel in which the I-II loop of alB replaced that of (Y, ++~ (Stea et al., 1995). Figure 8D illustrates that the chimeric channel exhibits only -15% block by 600 PM fomocaine, a concentration that produces almost 40% block of (Y~*.~. The degree of block observed with the chimera did not significantly differ from that for the (Yap complex (Fig. SE), suggesting that insertion of the I-II loop of ~l,n into OI,~., is sufficient to confer the reduced fomo-Caine sensitivity of (~,n onto (Y,~. The observation that a single glycine deletion or valine insertion is sufficient to affect fomocaine block, together with the data obtained from the (Y~~-cQ~ chimeric channel, the /3 subunit dependence of block, and the suggestion for a cytoplasmic location of the fomocaine blocking site, is consistent with the notion of a direct involvement of the I-II loop in fomocaine action. The data are summarized in Figures 60 and 8E. Coexpression with &a shows significant effects on fomocaine block of each a)lA.a, alE, and alcl but the complete absence of a p subunit does not significantly affect fomocaine block. Furthermore, an amino acid change at position 472 reduces the drug affinity by -5O%, with little additional effect of &. In contrast, block by procaine shows a different profile regarding subtype specificity, amino acid substitutions and effects of p subunits. For example, the degree of block observed with (Y 1A.b did not differ significantly from 01,*.~ in combination with any of the p subunits tested (p > 0.15, not shown). Together with the competition experiments and the difference in reversibility of fomocaine and procaine block, these data support the idea of separate receptors for procaine and fomocaine.

DISCUSSION
Morpholine rings are compatible with the piperidine receptor site A range of structurally unrelated compounds block calcium channels. For L-type calcium channels, the presence of at least seven distinct drug binding domains has been suggested, including sites for binding of phenylalkylamines (PAAs), DHPs, benzothiazepines (BTZs), and piperidines (for review, see Glossmann and Striessnig, 1990;Striessnig et al., 1994). The locations of the PAA binding site and the region for DHP binding have been mapped on the primary structure of the channel, with PAAs binding at the C terminal region of the S6 segment of the fourth domain, and DHPs binding near the pore-forming region of domain III and near the extracellular end of the S6 region in domain IV (Striessnig et al., 1991(Striessnig et al., , 1994. Furthermore, there is some evidence that BZTs act at least in part at the S5-S6 region of domain IV (Watanabe et al., 1993). However, the location of the piperidine receptor on the calcium channel primary sequence is not known. Piperidines were originally thought of as highly potent antagonists of the D, receptor, and clinically used to treat various forms of psychosis (Seeman and Lee, 1975;Seeman et al., 1976). However, over the past few years, increasing evidence suggests that piperidine-based antipsychotics are also powerful antagonists of voltage-dependent calcium channels, with affinities ranging from several nanomolar into the micromolar range (Gould et al., 1983;Sah and Bean, 1984;Galizzi et al., 1986;King et al., 1989;Enyeart et al., 1992;Grantham et al., 1994;Xu and Lee, 1994). Several members of the piperidine class have been shown to preferentially block T-type calcium channels in neurons Im et al., 1993). Here, we show that a morpholinebased local anesthetic is also capable of interactions with the Both OI~~.~ and CY~,+~ exhibit a reduced sensitivity to fomocaine block compared with OI~~.~. The chimeric channel exhibits a reduced blocking sensitivity that is comparable with that seen with the intact (~iu channel (0, E). Experimental conditions were as described in Figure 2. Zamponi et al. l Figure 9. Block of the neuronal calcium channels by the piperidine-based antipsychotic penfluridol. As can be seen from the records, ~l,u is the most significantly affected subtype (D), followed by (tic (C), aiA+, (A), and cyin (B). Coexpression of (~i*.~ with Pza (F) results in a reduced blocking affinity, as do amino acid substitutions in the I-II loop (G, H; compare with E). The experimental conditions were as outlined in Figure 2.
piperidine receptor site. Fomocaine qualitatively mimics the action of penfluridol, a compound known to act at the piperidine receptor site, in subtype specificity, lack of reversibility, dependence on ancillary subunits, and sensitivity to mutations in the I-II loop. Furthermore, coapplication of fomocaine and penfluridol results in blockage consistent with direct competition between the two compounds. Overall, these data indicate that morpholine rings are sufficient for interaction with the piperidine receptor. Given the similar blocking behavior seen with flecainide and its structural similarity to bupivacaine, we suggest that the potent calcium channel block reported for bupivacaine might also arise from interactions with the piperidine receptor, thus providing a molecular basis for the observation that some local anesthetics can substantially affect calcium channels, whereas others such as lidocaine or procaine do not.
Comparison with previous work and clinical significance Fomocaine preferentially blocks (~,u with an IC,, of 95 PM, whereas o,c, (Y,*.~, and (fin are blocked with, respectively, 3.5 fold, 8-fold, and 15fold lower affinity. A qualitatively similar order of blocking affinities has been observed with penfluridol and flecainide. The o,n channel has been proposed to be a novel type of low to mid threshold calcium channel (Soong et al., 1993) and the subtype-specific action observed in this study is consistent with the subtype-specificity seen with block of native low threshold channels by piperidine antipsychotics and related compounds .
Relatively little information is available concerning the blocking action of fomocaine on native calcium channels. Fomocaine has been previously reported to exhibit pronounced antiarrhythmic properties at concentrations as high as 30 pM (Braeunig et al., 1989), which is similar to clinically used blood plasma concentrations of a range of antiarrhythmic agents (Sheldon et al., 1987). At this concentration, -25% of the calcium current through (Y,~ was blocked by fomocaine, suggesting that calcium channel block by fomocaine occurs at clinically significant concentrations. We know of only two reports describing calcium channel block by flecainide (Scamps et al., 1989;Yamashita et al., 1995). In both of these studies, flecainide weakly blocked cardiac calcium channels (i.e., 15% block at 10 pM concentrations; Yamashita et al., 1995). The structurally related bupivacaine has been reported to block cardiac calcium channels with IC,,s between 100 and 300 PM (Sanchez-Chapula, 1988;Wulf et al., 1994) and N-type calcium channels from bullfrog sensory ganglion cells with an IC,, of -50 PM (Guo et al., 1992). For comparison, between 1 and 25 PM bupivacaine is required to block 50% of the currents through cardiac and neuronal sodium channels (Clarkson and Hondeghem, 1985;Butterworth and Strichartz, 1990;Chernoff, 1990). Hence, the blocking effects on calcium channels seen with fomocaine, flecainide and bupivacaine could be of direct clinical importance, especially in the light of the CAST study (Cardiac Arrhythmia Suppression Trial Investigators, 1989) and in view of reports that bupivacaine is lethal when administered to the cardiovascular system (Clarkson and Hondeghem, 1985).
There are several similarities between the blocking action of fomocaine and what is seen with block of low threshold calcium channels by piperidine/piperazine-based compounds in isolated and cultured neurons. First, the degree of block by flunarizine increases by -lo-fold when the external calcium concentration is dropped from 10 mM to 2.5 mM (Takahashi and Akaike, 1991). Similarly, the efficacy of penfluridol increases with decreasing external calcium concentrations . Second, flunarizine has only negligible effects on the position of the steady-state inactivation curve along the voltage axis (Takahashi and Akaike, 1991). Furthermore, inactivation is not required for the development of use-dependent block of T-type calcium channels ; but see Im et al., 1993). Third, penfluridol block of calcium channels in neuronal C cells is irreversible . Fourth, binding of radiolabeled fluspirilene is enhanced in the presence of external divalent cation blockers (King et al., 1989) (for review, see Kaczorowski et al., 1994). Finally, block of high voltage-activated currents by penfluridol, flunarizine, and fluspirilene is about one order of magnitude weaker than that of low threshold channels (Akaike et al., 1989;Enyeart et al., 1992) (for review, see Peters et al., 1991). Overall, these data support the hypothesis that fomocaine acts at the piperidine receptor site.
We know of only a few reports describing procaine block of calcium channels (Johansen and Kleinhaus, 198.5;Sugiyama and Muteki, 1994). Sugiyama and Muteki reported a Ki for procaine block of -2.5 mM whereas Johansen and Kleinhaus reported a 40% inhibition of calcium channels in leech neurons by 5 mM procaine. Based on a pH experiment (raising the pH from 7.4 to 8.5 resulted in a drastic increase in the degree of procaine block), Johansen and Kleinhaus concluded that procaine had to act from the cytoplasmic side. In three experiments, we could not detect a significant increase in procaine block after a pH increase from 7.6 to 8.65 (not shown), despite an associated 12-fold increase in the concentration of uncharged, and thus membrane-permeant, form of the drug (procaine pKa = 9.0; Ritchie and Greengard, 1961). Furthermore, procaine block was mimicked by, and competitively inhibited by membrane-impermeant diethylcarbamazine, again suggesting that the procaine site is directly accessible from the extracellular side. It is possible that the discrepancy between our data and those of Johansen and Kleinhaus might arise from structural differences between leech and mammalian calcium channels.
Is the piperidine receptor site in the domain I-II linker? We have presented four lines of evidence that are consistent with the I-II loop forming part of the piperidine receptor. First, coexpression of (pi subunits with &, significantly reduces the degree of fomocaine and penfluridol block of (Y,*.~, (~,c, and (~,n channels. Second, a single glycine deletion or valine insertion in position 472 in loop I-II also reduced the blocking affinity. Third, inserting the I-II loop of (~,u into (Y,,+;, resulted in block consistent with that seen with (~,u. Finally, fomocaine block was antagonized by external divalent cations and enhanced in the presence of external pore blockers, suggesting a cytoplasmic action close to the narrow region of the pore. Although both coexpression of (Y, subunits with &, and the insertion of a valine residue in position 472 have been shown to reduce the degree of inactivation (Soong et al., 1995) the observation that the steady-state inactivation curve was not affected by the presence of fomocaine suggests that inactivated channels are not blocked (Hille, 1977;Bean et al., 1983). Furthermore, there are no resolvable differences in the electrophysiological properties of oIAma and (Y,*.~ (Soong and Snutch, unpublished observations), arguing against a global conformational change induced by the glycine deletion, and yet the drug affinity is reduced. Although we are unable at this point to demonstrate direct binding to the I-II loop, our data are consistent with a direct involvement of the I-II loop in fomocaine block Zamponi et al. l Block of Ca*+ Channel s by Piperidines and not with a secondary effect arising from altered channel kinetics or a global conformational change. We can only speculate as to a putative mechanism by which ancillary subunits and amino acid substitutions might affect drug affinity. If there is a similarity to sodium channel block, one would expect the charged (note that the permanently uncharged piperazine is ineffective) N-terminal nitrogen atom of fomocaine to be the predominant blocking structure that would physically occlude the pore, whereas the phenyl rings and the morpholine ring would serve to increase the blocking affinity by binding to additional structures near the pore mouth (Zamponi and French, 1994). Ctixpression with &I or small structural changes in the I-II loop might abolish some of these additional interactions, leaving only a weak interaction between the N terminal and the narrow region of the pore. Such a mechanism would be consistent with the observation that 600 PM fomocaine always produces a minimal level of -L-20% block for any of the various combinations of (Ye and /3 subunits. Although far-reaching p subunit effects on calcium channel pharmacology have been reported (i.e., an increased affinity for the external blockers w-conotoxin and some DHPs when (Y, subunits were coexpressed with a j3 subunit compared with expression of 01, alone; Williams et al., 1992a,b;Nishimura et al., 1993), the observation that removal or addition of a single amino acid residue within the I-II loop results in a comparable effect suggests the possibility of a simple structural rearrangement of the I-II loop that occurs after binding of &, but not after binding of any of the other p subunits studied. Although there is a substantial degree of homology between the different p subunits especially in the regions that are involved in binding of the p subunit to the (Y, subunit, (De Waard et al., 1994;Pragnell et al., 1994), there are significant structural differences between &, and other types of /3 subunits. One could envision a scenario in which a unique region on the Pza subunit causes a small displacement of the I-II loop, thereby altering the three-dimensional conformation of the drug receptor. Such a mechanism appears plausible, especially in light of a recent study by Olcese et al. (1994) demonstrating that a short stretch of amino acids at the /3 subunit N terminus region (not directly involved in binding to the a! subunit) was critical for the effects of pza on inactivation.
Although we acknowledge the possibility of alternative interpretations, we view the data as consistent with the idea that the I-II linker is critical for block of calcium channels by piperidine compounds and perhaps forms part of the piperidine receptor.