Fast Inactivation of Voltage-dependent Calcium Channels

We recently described domains II and III as important determinants of fast, voltage-dependent inactivation of R-type calcium channels (Spaetgens, R. L., and Zamponi, G. W. (1999) J. Biol. Chem. 274, 22428–22438). Here we examine in greater detail the structural determinants of inactivation using a series of chimeras comprising various regions of wild type α1C and α1Ecalcium channels. Substitution of the II S6 and/or III S6 segments of α1E into the α1C backbone resulted in rapid inactivation rates that closely approximated those of wild type α1E channels. However, neither individual or combined substitution of the II S6 and III S6 segments could account for the 60 mV more negative half-inactivation potential seen with wild type α1E channels, indicating that the S6 regions contribute only partially to the voltage dependence of inactivation. Interestingly, the converse replacement of α1E S6 segments of domains II, III, or II+III with those of α1Cwas insufficient to significantly slow inactivation rates. Only when the I-II linker region and the domain II and III S6 regions of α1E were concomitantly replaced with α1Csequence could inactivation be abolished. Conversely, introduction of the α1E domain I-II linker sequence into α1C conferred α1E-like inactivation rates, indicating that the domain I-II linker is a key contributor to calcium channel inactivation. Overall, our data are consistent with a mechanism in which inactivation of voltage-dependent calcium channels may occur via docking of the I-II linker region to a site comprising, at least in part, the domain II and III S6 segments.

Calcium entry through voltage-dependent calcium channels is important for a range of cellular processes, including neurotransmitter release and activation of Ca 2ϩ -dependent enzymes. Molecular cloning has identified the primary structures of at least 9 different neuronal Ca 2ϩ channel ␣ 1 subunits (termed ␣ 1A through ␣ 1I (1)(2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12)(13)(14)(15)) that encode the previously identified native L-, P-, N, -Q-, T-, and R-types (for review, see Refs. 16 and 17). Calcium channels, like many other voltage-dependent ion channels, undergo a series of conformational changes in response to voltage, resulting in their opening, closing, and inactivation. Voltage-dependent inactivation of calcium channels is an important intrinsic process that prevents the breakdown of the calcium gradient as well as excessive calcium entry that is toxic to most cells (18 -20). In addition, many pharmacological agents interact predominantly with inactivated channels (21,22). Unlike sodium (23,24) and potassium (25)(26)(27) channels, the mechanisms that govern calcium channel voltagedependent inactivation are not fully understood. Although a number of structural moieties of the calcium channel ␣ 1 subunit have been implicated in being important in fast calcium channel inactivation (7, 22, 28 -30, 32, 33), the detailed mechanism underlying the inactivation process remain unknown, and there have been few systematic attempts to resolve this issue. By creating a series of chimeras between non-inactivating (L-type) ␣ 1C and rapidly inactivating (R-type) ␣ 1E rat brain calcium channels, we recently demonstrated that multiple structural domains determine the voltage dependence and rates of calcium channel inactivation (34). Here, we present novel evidence implicating the domain II and III S6 segments and the domain I-II linker region as key elements in setting the rate of calcium channel inactivation. Using a number of additional chimeras derived from ␣ 1C and ␣ 1E channels, we demonstrate that insertion of either the domain II S6, III S6, or I-II linker regions of ␣ 1E into ␣ 1C is sufficient to confer ␣ 1E -like inactivation kinetics. Consistent with these data, removal of inactivation from ␣ 1E required the concomitant substitution of all three regions with ␣ 1C sequence. Based on this evidence, we propose a model in which the I-II linker forms a hinged lid that may dock at the domain II and III S6 regions of the channel.

MATERIALS AND METHODS
Molecular Biology-We previously introduced convenient silent restriction enzyme sites (obtained from Life Technologies, Inc. and from New England Biolabs) into the cDNAs encoding for wild-type rat brain ␣ 1E (rbE-II, GenBank TM accession number L15453) and ␣ 1C (rbC-II, GenBank TM accession number M67515). AvrII was inserted at the beginning of the I-II linker, and SalI was inserted at the beginning of the II-III linker (see Ref. 34) to facilitate the creation of a series of chimeras encompassing various combinations of transmembrane domains of the two parent channels. To permit exchange of the domain II and III S6 segments, an additional pair of restriction sites was introduced into several of these chimeras at exactly complimentary positions; an AgeI restriction site was introduced 20 amino acids 5Ј to the beginning of the II S6 segment, and an AatII site was created 5 amino acids 5Ј to the beginning of the III S6 segment. The residues prior to the beginning of II S6 and III S6 are identical in both channels, and hence, the resultant chimeras only differ in their S6 regions. To permit ex-* This work was supported by a grant from the Heart and Stroke Foundation of Alberta and the Northwest Territories, with some additional grant support from the Medical Research Council of Canada. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  change of the domain I-II linker region, a silent NarI site was introduced into the ECCC sequence at the beginning of domain II.
II S6 Chimeras-The EECC and CCEE chimeras (in the pMT2 expression vector) were used as the template for mutagenesis to introduce unique AgeI sites near the beginning of the II S6 segments. Both constructs were first cut with SalI (II-III linker, 3Ј-polycloning site) and recircularized to reduce their length by about 5 kb 1 before proceeding with mutagenesis. Using the QuikChange kit (Stratagene), we created silent mutations at bp 2109 of CCEE and bp 1815 of EECC. Restriction digests confirmed successful addition of the sites, and the coding region was sequenced to confirm the absence of errors. To construct the chimeras, CCEE (ϩAgeI) and EECC (ϩAgeI) were cut with KpnI and AgeI, and the resulting 2-kb fragments from each construct were exchanged via ligation. Finally, the excised 5-kb SalI fragment was reintroduced into both constructs to yield two full-length clones: ␣ 1E (IIS6C)/pMT2 and ␣ 1C (IIS6E)/pMT2. III S6 Chimeras-The EEEC and CCCE chimeras (in the pMT2 expression vector) were used as the templates for mutagenesis to introduce unique AatII sites near the beginning of the III S6 segments. Using the QuikChange kit, a silent mutation at bp 4002 of EEEC and a non-silent mutation at bp 3384 of CCCE were introduced. However, because the non-silent mutation of CCCE involved a substitution to the corresponding residue in the EEEC sequence (serine to valine), the substitution became inconsequential in the completed chimera. Successful addition of the sites was confirmed by restriction digests, and the coding region was sequenced to confirm the absence of errors. To construct the chimeras, CCCE ϩ AatII and EEEC ϩ AatII were cut with NotI and AatII restriction enzymes, and the resulting 4-kb fragments from each construct were exchanged and religated to yield two fulllength clones: ␣ 1E (IIIS6C)/pMT2 and ␣ 1C (IIIS6E)/pMT2.
I-II Linker Chimera-To create CeCCC, a unique non-silent SplI site was generated by site-directed mutagenesis (QuikChange) at the very end of the domain I-II linker regions of CECC and CCCC at exactly complimentary positions in a stretch of residues that is completely conserved between both parent channels (VFYW). A KpnI-SplI fragment (ϳ1.5 kb) was excised from CECC and ligated into likewise digested CCCC to produce CeCCC but still carrying the non-silent SplI site. Finally, another round of site-directed mutagenesis was used to remove the non-silent SplI site, thereby restoring the original amino acid sequence (VFYW).
Additional Chimeras-To create CEEE(IIϩIIIS6C), an AvrII fragment cut from CCCC (900 bp before 5Ј polylinker, I-II linker region, ϳ2 kb) and ligated into likewise-digested ␣ 1E (II/IIIS6C). To create CECC(IIS6C), CEEE(II/IIIS6C) and EECC were cut with SalI, and the fragment from the latter chimera (corresponding to domains III and IV of ␣ 1C ) was ligated into the former construct. The correct orientation of the constructs was confirmed via restriction digests. To create CcEEE(II/IIIS6C), we first introduced a silent NarI site into the EE-EE(II/IIIS6C) sequence ϳ20 amino acid residues before the end of the domain I-II linker region. ECCC contained an endogenous NarI site at an exactly complementary position. A 1900-bp NarI fragment (1000 bp before the 5Ј polylinker, end of I-II linker) was excised from ECCC substituted in the EEEE(II/IIIS6C ϩ NarI) construct to give rise to EcEEE(II/IIIS6C), with the lowercase letter indicating the origin of the I-II linker. This construct did not express functionally in HEK cells but was used to create the chimera CcEEE(II/IIIS6C) by substituting domain I of EcEEE(II/IIIS6C) with that of CCCC using AvrII. Correct orientations of the inserts were confirmed via restriction enzyme digests.
Transient Expression of Calcium Channels and Electrophysiological Recordings-We previously provided a detailed description of the procedures for transient expression of the wild type and chimeric calcium channels in human embryonic kidney tsa-201 cells and their electrophysiological analysis via whole cell patch clamp (34). Unless stated otherwise, the external and internal recording solutions contained, respectively, 20 mM BaCl 2 , 1 mM MgCl 2 , 10 mM HEPES, 40 mM tetraethylammonium chloride, 10 mM glucose, 65 mM CsCl (pH 7.2 with tetraethylammonium hydroxide) and 108 mM cesium methanesulfonate (CsMS), 4 mM MgCl 2 , 9 mM EGTA, 9 mM HEPES (pH 7.2 with tetraethylammonium hydroxide), thus minimizing the possibility of contamination from calcium-dependent inactivation processes. Pipette resistances were typically on the order of 3 to 4 M⍀, and series resistance was compensated by 85% to minimize voltage errors. Currents were typically elicited from holding potentials of Ϫ100 mV (or Ϫ130 mV for ␣ 1E and other chimeras which activated more negatively) to various test potentials using Clampex software (Axon Instruments). However, to obtain steady state inactivation curves, a 5-s conditioning pulse preceded a test depolarization to ϩ10 mV. The rate of inactivation was assessed by considering both the percentage of current that had inactivated over a time course of 125 ms and by mono-exponential fits to the time course of inactivation. Data were analyzed using Clampfit (Axon Instruments) and Sigmaplot 4.0 (Jandel Scientific). Steady state inactivation curves and macroscopic current voltage relations were analyzed using the Boltzmann equation (see Ref. 34). All error bars are standard errors, numbers in parentheses displayed in the figures reflect numbers of experiments, and p values were determined by Student's t tests.

RESULTS
The Cytoplasmic Milieu Affects Inactivation Properties-We previously reported that wild type ␣ 1C and ␣ 1E channels exhibited pronounced differences in their inactivation profiles (34). Since completion of our original study, we switched our internal recording solution (105 mM CsCl, 25 mM tetraethylammonium chloride, 11 mM EGTA, and 10 mM HEPES, pH 7.2) to a solution composed of 108 mM CsMS, 4 mM MgCl 2 , 9 mM EGTA, 9 mM HEPES (pH 7.2), which we found to yield more stable recordings. Hence, it was necessary to reassess the inactivation properties of the two wild type channels under our present experimental conditions. Fig. 1 compares the inactivation profiles of wild type ␣ 1C and ␣ 1E channels, coexpressed with ␣ 2 -␦ and ␤ 1b subunits. As shown in the figure, the two wild type channels exhibit diametrically different inactivation properties such that the half-inactivation potential of ␣ 1E is 60 mV more negative than that of ␣ 1C . Furthermore, the rate of inactivation, expressed either as the time constant for current decay, , or the percentage of current that has inactivated over a time course of 125 ms, is 2-4-fold greater for ␣ 1E than ␣ 1C , depending on the test potential. These differences in the inactivation profiles of the two channels are qualitatively consistent with our previous recordings (34); however, a closer comparison with our previous work reveals three quantitative differences. First, the current densities obtained in the CsMS internal solution were on average twice as large as those observed in CsCl (not shown). Second, the half-inactivation potentials of ␣ 1E and ␣ 1C were shifted, respectively, by 10 to 20 mV in the hyperpolarizing direction, whereas the half-activation potential was not significantly affected (p Ͼ 0.05). Finally, the inactivation rates in internal CsMS were considerably accelerated. Neither the substitution of negative counter ion nor the presence or absence of internal magnesium was found to account for the effects (not shown). However, internal TEA ions significantly affected the inactivation properties of the channel such that the presence of 25 mM internal TEA resulted in a significant slowing of the inactivation rate (Fig. 1, C and D) and an ϳ10-mV rightward shift in the midpoint of the steady state inactivation. Thus, TEA ions, commonly thought to be inert for voltage-dependent calcium channels, exert a pronounced effect on calcium channel gating.
Transfer of ␣ 1E Domain II S6 or III S6 Regions Accelerates Inactivation of ␣ 1C -Our previous work showed that multiple transmembrane domains were involved in the inactivation process of ␣ 1E channels, with domain II and III contributing to the greatest extent (34). We theorized that calcium channel inactivation may occur via a mechanism reminiscent of that underlying C-type inactivation common to many types of voltage-dependent potassium channels, a process that is believed to involve pore constriction mediated by the S6 segments of the channel (35,36). To assess a putative role of the S6 segments in fast calcium channel inactivation, we created chimeras in which the S6 regions of domains II/III of ␣ 1C were replaced by the corresponding regions of ␣ 1E .
As seen in Fig. 2A, replacement of the II S6 or III S6 segments of ␣ 1C with that of ␣ 1E dramatically increased the inactivation rate of ␣ 1C to levels observed with wild type ␣ 1E channels. These data suggest that the individual S6 segments in domains II and III are important determinants of calcium channel inactivation. To test whether the effects of domains II and III were additive, we also examined a double chimera ␣ 1C (II/IIIS6E) in which the S6 segments of domains II and III of ␣ 1E were inserted into ␣ 1C concomitantly (Fig. 2). The double replacement did not result in further speeding of the inactivation kinetics, nor could it enhance the slower, ␣ 1C -like rates that persisted at relatively hyperpolarized test potentials in the two single S6 chimeras. Thus, although the presence of a single "inactivating" S6 segment is sufficient to confer many aspects of the more rapid inactivation of the wild type ␣ 1E channels, even their combination cannot account for all of the voltage dependence associated with the inactivation rates.
The Domain I-II Linker Region of ␣ 1E Maintains Rapid Inactivation Kinetics-If the presence of a single S6 segment in domain II or III of ␣ 1E is sufficient to confer rapid inactivation kinetics onto ␣ 1C , one might expect that replacement of only one of those two S6 segments in ␣ 1E should be ineffective in removing inactivation. To test this hypothesis, we examined two additional chimeras, ␣ 1E (IIS6C) and ␣ 1E (IIIS6C). As seen in Fig. 3, A and B, the two chimeras exhibited inactivation rates that did not differ significantly from those of the wild type ␣ 1E channel, except at relatively hyperpolarized test potentials. However, even simultaneous substitution of the II S6 and III S6 regions of ␣ 1E with those of ␣ 1C (i.e. ␣ 1E (II/IIIS6C)) did not significantly slow inactivation (Fig. 3C), suggesting the presence of an additional region in the ␣ 1E sequence that is independently capable of maintaining inactivation.
To identify the putative region sustaining rapid inactivation of ␣ 1E , we first inserted the ␣ 1C domain I into EEEE(II/IIIS6C), creating the CEEE(II/IIIS6C) chimera, and still, ␣ 1E -like inactivation persisted (Fig. 3D). However, upon substitution of most of the ␣ 1C domain I-II linker region into this construct, inactivation was virtually abolished (Fig. 3E), suggesting that it was the domain I-II linker region of ␣ 1E that maintained ␣ 1E -like inactivation rates in the ␣ 1E (II/IIIS6) construct. To unequivocally show the importance of the domain I-II linker region, we created two additional chimeras (EcEEE and EcEEE (II/IIIS6C)); however, neither construct expressed functionally in tsa-201 cells. Nonetheless, if our hypothesis is correct, then an ␣ 1C channel containing the domain I-II linker of ␣ 1E should exhibit rapid inactivation kinetics. Data obtained with such a chimera (CeCCC) are shown in Fig. 3F. As evident from the figure, the CeCCC construct exhibited inactivation kinetics that were significantly faster than those seen with the wild type ␣ 1C channels, consistent with our hypothesis. We also examined an additional construct that contained the domain I-II linker plus the first five transmembrane segments of domain II of ␣ 1E (thus retaining the "non-inactivating" domain II S6 and III S6 regions of ␣ 1C ), and similar to that of the CeCCC chimera, the CECC(IIS6C) construct exhibited inactivation kinetics that were significantly more rapid than those of the wild type ␣ 1C channel (n ϭ 12, not shown). Taken together, this supports the idea that the presence of either one of three regions in ␣ 1E , the domain I-II linker or II S6 or III S6 regions, is sufficient to preserve rapid inactivation, whereas the concomitant replacement of these three regions with the corre -FIG. 1. A and B, comparison of inactivation properties of the wild type channels. Schematic, proposed transmembrane topology of voltage-dependent calcium channels. Secondary structure of ␣ 1C and ␣ 1E are depicted, respectively, in black and white. A, comparison of the steady state inactivation properties of ␣ 1C and ␣ 1E (coexpressed with ␤ 1b and ␣ 2 -␦). Currents were obtained by stepping from various conditioning potentials (5-s duration) to a test potential of ϩ10 mV. The data were fitted with the Boltzmann relation. B, comparison of the voltage dependence of the inactivation rates (␣ 1C : n ϭ 12; ␣ 1E : n ϭ 8). Currents were elicited by stepping from Ϫ100 mV or Ϫ130 mV (␣ 1E ) to a series of test potentials. Left panel, inactivation rates expressed as the fraction of current inactivated over a time course of a 125-ms test depolarization. Right panel, inactivation rates reflected as the time constant for inactivation obtained from monoexponential fits to the raw current data. C and D, effects of internal TEA ions on inactivation properties of ␣ 1E calcium channels coexpressed with ␤ 1b and ␣ 2 -␦. C, current records elicited by test depolarization to ϩ10 mV in either 108 mM CsMS, 4 mM MgCl 2 , 9 mM EGTA, 9 mM HEPES (pH 7.2) or 83 mM CsMS, 25 mM tetraethylammonium (TEA) chloride, 4 mM MgCl 2 , 9 mM EGTA, 9 mM HEPES (pH 7.2). The currents were arbitrarily scaled to overlap at peak. D, effect of internal TEA (n ϭ 5) on the time constant of inactivation, .
sponding elements of ␣ 1C is required to confer slow inactivation kinetics.
Multiple Structural Elements Control the Voltage Dependence of Inactivation-We have previously shown that domains II and III could account for the majority of differences in halfinactivation potential between the two wild type channels, whereas domains I and IV contributed to a lesser extent (34). To test whether the effects of domains II and III could be attributed to the S6 regions, we compared the half-inactivation potentials of the wild type and chimeric calcium channels (Fig. 4). Replacement of the II S6 region of ␣ 1E did not affect the half-inactivation potential, and the reverse substitution in ␣ 1C resulted in only a small ϳ10-mV hyperpolarizing shift in steady state inactivation kinetics, which was paralleled by a comparable change in half-activation potential (Fig. 4). Thus, it seems unlikely that the domain II S6 region contributes in a meaningful manner to the determination of steady state inactivation kinetics despite its pronounced effects on inactivation rate.
Replacement of the domain III S6 region resulted in more substantial (ϳ15 mV) hyperpolarizing shifts in half-inactivation potential that were not mirrored by activation potential shifts. In view of the 60-mV spread between the wild type channels, the contributions from the S6 regions were relatively minor, suggesting that other regions in domains II and III may determine the voltage dependence of inactivation. Indeed, both the CeCCC and CECC (IIS6C) constructs inactivated 20 mV more negatively than the wild type ␣ 1C channel (Fig. 4) despite sharing a common half-activation potential with ␣ 1C , thereby implicating the domain I-II linker. Thus, although insertion of a single inactivating structure of ␣ 1E is sufficient to confer the rapid inactivation profile in an essentially all or none fashion, multiple substitutions, including some that do not affect inactivation rates (i.e. domain I in Fig. 4D of Ref. 34), are required to account for the differences in voltage dependence of inactivation of the wild type channels.

DISCUSSION
Comparison with Previous Work-We previously presented evidence that the calcium channel domains II and III are critical determinants of both the voltage dependence and the rate of inactivation (34). Each of those two domains contributed to about half of the observed differences in half-inactivation potential between ␣ 1C and ␣ 1E , and insertion of either domains II or III of ␣ 1E into the ␣ 1C sequence conferred all of the rapid inactivation kinetics of ␣ 1E (34). Here, we have more narrowly identified the regions involved in determining the inactivation rate. As seen from Table I, with one exception (ECEE), any construct containing the ␣ 1E sequence in the domain IIS6, III S6, or I-II linker regions exhibited ␣ 1E -like inactivation kinetics. In contrast, only constructs carrying ␣ 1C sequence in each of those regions inactivated slowly, consistent with the idea that the above regions are the central structural elements involved in the control of inactivation rates. reflect, respectively, data obtained from the wild type ␣ 1E and ␣ 1C channels (coexpressed with ␤ 1b and ␣ 2 -␦). A, comparison of the current waveforms of the parent channels and chimeras arbitrarily scaled to overlap at peak. Note that insertion of the II or III S6 segments of ␣ 1E into the ␣ 1C backbone confers rapid inactivation kinetics. The records for the wild type channels are the same in all three panels. Schematic, proposed secondary structure of the constructed chimeras. B, voltage dependence of the inactivation rates of the chimeras. The data for the wild type channels are the same as in Fig. 1 and were included merely to facilitate comparison; the numbers of experiments (n) refer only to the chimeras. The experimental conditions were as described in Fig. 1. The involvement of the I-II linker would be consistent with the observation that two separate point mutations in this region can slow the inactivation of ␣ 1A calcium channels (7,30). In addition, overexpression of the I-II linker regions of ␣ 1A was found to speed inactivation of the ␣ 1A channel (31). Also consistent with our data, amino acid substitutions in the domain III S6 region have been reported to affect inactivation kinetics (32, 37). Zhang et al. (28) implicate exclusively the domain I S6 region in the fast inactivation process by utilizing a series of chimeras between rabbit ␣ 1A and marine ray ␣ 1E calcium channels, which contrasts with our observations that rapid inactivation kinetics do not require the presence of ␣ 1E domain I nor is replacement of this region with ␣ 1C sequence sufficient to abolish inactivation. Both ␣ 1E and ␣ 1A share identical domain II and III S6 regions and differ in their domain I S6 regions in only one position (methionine in ␣ 1A versus valine in ␣ 1E (38)). Interestingly, the ␣ 1C sequence also contains a valine residue in this position, thus perhaps masking any subtle effects of domain I in our experiments. Alternatively, it is possible that an amino acid substitution in the domain I S6 regions second-arily affects inactivation by altering the conformation of the associated I-II linker region.
What Controls the Voltage Dependence of Inactivation?-We previously presented evidence that domains II and III accounted for much of the difference in half inactivation potentials seen with the wild type channels (34). Within domain III, we can attribute a significant effect to the S6 segment, indicating that the III S6 region is involved in controlling both the rate and some of the voltage dependence of inactivation. A similar argument can be made for the domain I-II linker region. In contrast, the II S6 region had little effect on the voltage dependence of inactivation despite being an important feature  The data for the inter-domain chimeras (third through tenth lines) were taken from Spaetgens and Zamponi (34). The terminology "fast" refers to ␣ 1E -like inactivation. The check marks indicate the presence of an ␣ 1E sequence in domain II S6 or III S6 or in the domain I-II linker. The asterisk denotes a chimera which we found to be an outlier in our original paper (34). Construct  II S6  III S6  I-II  E (IIS6 or IIIS6 or I-II) Rate for determining the inactivation rate. Conversely, domain I does not affect inactivation rate but does contribute to voltage dependence (compare ␣ 1E (II/IIS6C) and CEEE (II/IIIS6C)). Thus, despite some overlap, the structural determinants governing the rates and voltage dependence of inactivation appear to be distinct. Whereas the critical determinants of inactivation rate are fairly localized, the mechanism controlling the voltage dependence of inactivation appears to be more globally distributed across the calcium channel ␣ 1 subunit. What Is the Mechanism That Underlies Calcium Channel Fast Voltage-dependent Inactivation?-Our original intent was to gather additional evidence in support of our hypothesis that fast calcium channel inactivation might occur via a mechanism reminiscent of the C-type inactivation process, which is thought to involve a pore collapse mediated by the four S6 segments (i.e. Refs. 35 and 36). Our data implicating the domain II and III S6 regions fit with such a model. However, the critical involvement of the cytoplasmic domain I-II linker region argues against simple pore collapse. As a result, our data are best described by a model in which the I-II linker region forms a cytoplasmic gating particle (39) similar to that proposed for the domain III-IV linker of voltage-dependent sodium channels (23, 24) (see Fig. 5). If so, then the S6 regions might perhaps serve as the docking site for the inactivation gate. The current belief that that S6 segments line the inner vestibule of the pore would be consistent with such a mechanism (40,41,42), but it may well be possible that other regions of the channel could be part of the docking interaction. A putative role of the domain I-II linker as the inactivation gate would fit the previously reported effects of point mutations in the ␣ 1A calcium channel I-II linker (7,30) and the ability of overexpressed ␣ 1A I-II linker to accelerate the inactivation rate of ␣ 1A calcium channels. This model could also account for the effects of the calcium channel ␤ subunits, which are known to physically bind to the I-II linker region, on inactivation rate (10,43,44). The antagonistic effects of the ␤ 2a subunit on inactivation (44) could perhaps arise from a restricted mobility of the domain I-II linker region as a consequence of anchoring the palmitoylated N terminus of this subunit to the plasma membrane (45).
A hinged-lid model could also accommodate our observations that intracellular TEA slows inactivation rates. If a TEA molecule acting as a low affinity blocker were to compete with the I-II linker for its docking site, one would expect to observe a slowing of the macroscopic time course of inactivation. Such a mechanism would not be without precedent, as TEA prevents inactivation gate closure of shaker B potassium channels (35), and open channel block of batrachotoxin-activated cardiac sodium channels by local anesthetics and related compounds prevents fast inactivation (46,47).
How can a hinged-lid mechanism account for the observation that the presence of either the domain IIS6, III S6, or the domain I-II linker region of ␣ 1E was generally sufficient to mediate rapid inactivation? Within the framework of our model, the domain I-II linker of ␣ 1C would have the ability to dock to either the domain II S6 or III S6 regions of ␣ 1E . Conversely, the domain I-II linker of ␣ 1E would have to be capable of interacting with either one of the domain II S6 or III S6 regions of ␣ 1C . In contrast, the relative lack of inactivation of L-type channels would require an inability of the ␣ 1C I-II linker region to dock effectively to the domain II S6 and III S6 regions when of ␣ 1C origin. Biochemical evidence, however, will ultimately be required to prove a putative existence of a physical binding interaction between the domain I-II linker and the S6 segments.
In summary, a hinged-lid model of inactivation can nicely account for our data as well as the key observations reported in the literature. The redundancy of the structural elements that are sufficient to maintain rapid inactivation underlines the fundamental importance of this process for the precise control of calcium entry and, thus, prevention of accumulation of toxic levels of intracellular calcium (18 -20).