L-Type Channels

L-type Ca²⁺ channels were initially described in peripheral neurons and cardiac cells, but appear to be present in all excitable as well as many types of non-excitable cells.⁴ In certain cells, L-type channels have been shown to be preferentially localized to specific subcellular regions. For example, the L-type channels responsible for skeletal muscle contraction are concentrated on the transverse tubule membrane,⁵ while neuronal L-type channels are located primarily on cell bodies and proximal dendrites.⁶ The L-type channel is the primary route for Ca²⁺ entry into cardiac, skeletal, and smooth muscles.² The skeletal muscle L-type channel acts as a voltage sensor for excitation-contraction (E-C) coupling in skeletal muscle, presumably linking membrane depolarization to Ca²⁺ release from intracellular stores. While Ca²⁺ entry through this channel is not required for the initiation of contraction in skeletal muscle, it may provide a source of Ca²⁺ to replenish internal stores.^2,4,7,8 There is some evidence that L-type channels are involved in exocytotic release from endocrine cells and some neurons^9-13 and the localization of L-type channels on the cell soma⁶ has also implicated these channels in the regulation of gene expression.^14-16

Much is known about the pharmacological properties of L-type Ca²⁺ channels. The three main classes of organic L-type channel blockers are the phenylalkylamines (verapamil), benzothiazapines (diltiazem), and 1,4-dihydropyridines (DHPs) (e.g., nitrendipine, nifedipine, nimodipine). The DHP antagonists bind preferentially to channels in the active conformation, a state favored by depolarization (producing more potent inhibition at depolarized potentials). A number of DHP agonists have also been developed, the most highly utilized of which is (-)-Bay K 8644 which increases both the open time and the single channel conductance (see chapter by Striessnig for more detail). L-type channels are also blocked by certain native peptide toxins such as ω-agatoxin IIIA (ω-Aga IIIA), isolated from the venom of the funnel web spider Agelenopsis aperta.^17,18 ω-Aga IIIA reduces the current amplitude without affecting the time course and unlike the DHPs, ω-Aga IIIA inhibition is voltage-independent and blocks L-type channels at all potentials² (see chapter by Adams and Lewis for more detail).

L-type channels have a unitary conductance ranging from 20 and 27 pS using 110 mM barium (Ba²⁺) as the charge carrier. L-type channels require large departures from resting potential to become activated and typically begin to open at potentials positive to -10 mV, although they can activate at significantly more negative potentials in chromaffin cells, sensory neurons, and cardiac cells. In the presence of Ba²⁺ as the charge carrier, once open, L-type channels do not inactivate significantly during depolarizations of hundreds of milliseconds.^2,3,19 Compared to Ba²⁺ currents, using Ca²⁺ as the charge carrier L-type currents are smaller and inactivate rapidly. This Ca²⁺-dependent inactivation has a number of characteristic properties and inactivation attributable to Ca²⁺ influx is greatest at depolarizations at which Ca²⁺ entry through the channel is maximal.²⁰ While the degree of inactivation is slowed by the addition of BAPTA and other Ca²⁺ chelators, it is not completely abolished. Ca²⁺-dependent inactivation can however be eliminated by intracellular applications of trypsin, suggesting that the mechanism through which Ca²⁺ acts to inactivate the channel is in close proximity to, if not part of, the channel complex itself.^2,3,21,22 Because the rate of Ca²⁺-dependent inactivation does not change with channel density, Neely et al (1994)22 proposed a “local domain” hypothesis, in which Ca²⁺ affects only the channel through which it enters (see chapter by Lee and Catterall for more detail). Moreover, current models view calmodulin to be constitutively bound to the C-terminus forming part of the Ca²⁺ sensing machinery, ultimately leading to the signal transduction of Ca²⁺-dependent inactivation.²³

N-Type Channels

In addition to L- and T-type Ca²⁺ channels, recordings from chick dorsal root ganglion (DRG) cells revealed a third type of single channel Ca²⁺ conductance of 13 pS (in 110 mM Ba²⁺), intermediate between that of the T- (8 pS) and L- (25 pS) type channels.^1,2,19,24,25 Although this conductance shares some general electrophysiological characteristics with currents through both T- and L-type channels, it could not be attributed to either. Consequently, the corresponding channel was designated as N (neither)-type.

Although first identified in chick DRG neurons, N-type channels have also been detected in mammalian DRG cells,^26-29 mammalian and amphibian sympathetic neurons,^30-33 and other cells of the peripheral and central nervous systems.^34-39 N-type channels appear to be expressed only in neuronal tissues,^31,40 although an N-type current has been reported in rat thyroid C-cell line.⁴¹ Electrophysiologically, N-type channels are most easily distinguished from L-type channels by their inactivation properties. Unlike L-type channels, N-type channels display time-dependent inactivation (with Ba²⁺ as the charge carrier). N-type currents decay with a time constant (τ) ranging from 50 to 110 ms, significantly slower than the rapid (τ = 20-50 ms) inactivation of the LVA T-type channels, but much faster than the non-inactivating L-type channels. In chick DRG neurons, the N-type current decays almost completely during a test depolarization of 140 ms, while L-type current shows little inactivation over the same period of time.¹ However, N-type currents do not always inactivate rapidly. In sympathetic neurons, the decay rate of N-type currents is much slower (τ = 500 to 800 ms) and can be incomplete, even over depolarizations lasting longer than one second.^2,41-42 Thus, there appears to be at least two distinct components to N-type current inactivation. These differences in inactivation kinetics could reflect different subtypes of N-type channel. Alternatively, a single N-type channel could support both currents by switching between the slow- and fast-inactivating states.^40,43

In addition to the time-dependence parameter, there is also a voltage-dependent aspect to N-type channel inactivation.^1,24,25,29 Holding the cell membrane at potentials between -60 and -40 mV results in significant inactivation of the N-type current, and strongly negative potentials are required to reprime the channels. N-type channels are markedly more sensitive to the effects of holding potential on inactivation than are L-type channels. At resting membrane potentials of -20 mV, N-type channels are completely inactivated while L-type channels remain available for opening.

Theoretically, the different inactivation properties of N- and L-type channels provides two parameters that can be used to dissect the relative contributions of the two channel types to the whole cell HVA current.^2,24 One method takes advantage of the different inactivation rates. The component of whole cell current that decays during a prolonged depolarization can be attributed to the inactivating N-type channels, while the non-inactivating portion is identified as L-type current. The second approach exploits the different ranges over which voltage-dependent inactivation takes place. The contribution of each type to the whole cell current may be determined by analyzing the differences in whole cell currents elicited by depolarizations from resting potentials of -40 and -90 mV. Because L-type channels are relatively resistant to the effects of holding potential on inactivation while N-type channels inactivate at depolarized membrane potentials, the difference under these two conditions should reflect the contribution of N-type channels to the whole-cell current. However, neither method may be adequate to properly distinguish these currents. Some N-type channels can inactivate quite slowly and inactivation may not be complete. In addition, voltage-dependent inactivation of N-type channels can be highly variable and takes place over a wide range of holding potentials between -80 to -20 mV.^24,25 If N-type channels predominate in a cell, the residual current through incompletely inactivated N-type channels may be significant. Thus definitions of N- and L-type current based solely on these criteria may not be valid.

Pharmacologically, N-type channels are sensitive to inhibition by a class of native peptide toxins called the ?-conotoxins, which are a family of small (13-29 amino acid) peptides found in the venom of predatory marine snails of the genus Conus.^{44, 45} All known ω-conotoxins inhibit N-type Ca²⁺ channels, although their specificities and blocking affinities for this particular channel vary significantly. To date, ω-conotoxin GVIA (ω-CgTx), a 27-amino acid peptide from Conus geographus⁴⁶ is the most specific ω-conotoxin peptide for N-type channel inhibition. ω-CgTx produces complete and irreversible inhibition of N-type currents in DRG, hippocampal, sympathetic, and sensory neurons at concentrations of approximately 100 nM to 1 μM.^47-49 At higher concentrations (5-15 μM), ω-CgTx also inhibits L- and T-type currents, although unlike N-type channels, the effects are incomplete and reversible^47-52 (see chapter by Adams and Lewis for more detail).

ω-CgTx binding sites (and by extension N-type Ca²⁺ channels) are distributed throughout the PNS and CNS, including the cortex, hippocampus, olfactory bulb, and cerebellar cortex, and appear especially concentrated in regions of high synaptic density.^{2,53- 58} Although N-type channels were first identified by single channel recordings from the cell bodies of DRG neurons,¹ they appear to be more abundantly localized on dendrites and axon terminals. In muscle, ω-CgTx binding occurs at the active zones of presynaptic cells in spatial register with the postsynaptic acetylcholine (ACh) receptors on the muscle. Labeling is rarely found between active zones, nor is it localized to areas of the presynaptic membrane that do not face the muscle. N-type channels have also been observed to cluster in areas of synaptic contact on hippocampal CA1 neurons.⁵⁴

The presence of N-type channels on the presynaptic membrane suggests that Ca²⁺ entry through these channels is responsible for triggering neurotransmitter release. An early study⁵⁹ demonstrated that ω-CgTx blocks electrically-induced release from the frog NMJ and numerous subsequent studies have demonstrated that the application of ω-CgTx inhibits neurotransmitter release in the central and peripheral nervous system.^60-67 Furthermore, biochemical studies indicate that N-type channels are physically associated with proteins such as synaptotagmin and syntaxin which are part of the exocytotic machinery.^68-70 There appear to be species- and cell-specific differences in N-type-channel-regulated neurotransmission. For example, while ω-CgTx completely abolishes neurotransmission at the avian and amphibian NMJ, it has no effect on the mammalian motor nervous system.^59,65,71-73 The ability of this toxin to inhibit neurotransmission also varies depending on the type of synapse within a given species.^52,65-67 For example, inhibitory synaptic transmission in hippocampal CA1 neurons is strongly reduced by the application of ω-CgTx, whereas the toxin blocks excitatory transmissions to a much lesser extent. In addition, while ω-CgTx inhibits release of ACh from both autonomic and central neurons in the rat, release from central neurons is approximately 20-fold less sensitive.⁶⁵

In spite of the complete and irreversible inhibition of N-type channels produced by ω-CgTx, application of the toxin to many types of neurons only partially inhibits neurotransmitter release, suggesting that other types of Ca²⁺ channels contribute to neurotransmitter release from both central and peripheral neurons.^58,74,75 In fact, while regulation of transmitter release from peripheral neurons appears to predominantly involve N-type channels, release in the central nervous system appears to be controlled primarily by other types of Ca²⁺ channels that are insensitive to both ω-CgTx and DHPs.^76,77

The presence of N-type channels in regions other than the synapse indicates that these channels have other functions in addition to neurotransmitter release. N-type channels localized to dendritic branch points may be involved in integration or amplification of neural inputs. ⁵⁶ N-type channels may also play a role in nervous system development as evidenced by the expression of N-type channels on postmitotic cerebellar granule cells. These cells only begin migration after the appearance of N-type channels and ω-CgTx causes a cessation of migration.⁷⁸

Other HVA Ca Channels: P-, Q-, and O-Types

The original classification system of Ca channels, which was expanded from the simple LVA/ HVA dichotomy to encompass T-, L- and N-channels, was subsequently found to be too restrictive to adequately describe all types of Ca²⁺ conductances. The availability of blocking agents that target L- and N-type channels revealed other HVA currents that could not be defined according to this scheme.^{36,37,45,79-81} These novel channel types, variously named P-, Q-, O-, and R-, have primarily been defined on the basis of their distinctive pharmacological properties rather than electrophysiological characteristics.

The P-type current was originally identified as an HVA current in Purkinje cells that is insensitive to the agents typically used to inhibit L- and N-type channels.⁸² These channels are thought to support the Ca²⁺-dependent action potentials in the dendrites of cerebellar Purkinje cells, which are unaffected by DHPs and ω-CgTx, but are potently blocked by components of the venom of the funnel web spider Agelenopsis aperta.^82-85

Whole cell recordings from Purkinje cells reveal an HVA current that peaks at voltages between -30 and -20 mV and inactivates slowly over the duration of the depolarization.^37,52,81,86 Single channel analysis of P-type channels reveals conductances in ranges similar to those of N- and L-type channels. Multiple unitary conductance levels of 9, 14, 19 pS in 110 mM Ba have been reported for P-type channels in the Purkinje cell soma and dendrites,⁸⁷ and a P-type current in hypoglossal motorneurons has a unitary conductance of 20 pS.³⁹

The venom of the funnel web spider, like that of the cone snail, is a cocktail of toxins that target different elements of the synaptic machinery. FTX, a non-peptide component of the venom (arginine polyamine and a synthetic analog of FTX, sFTX), was initially reported to be specific blockers of P-type channels,^3,82,83 but subsequently shown to produce inhibition of other Ca²⁺ currents in conjunction with the P-type block.^3,45

ω-Aga IVA, a 48-amino acid peptide also found in the venom of A. aperta potently inhibits P-type Ca²⁺ channels.^52,84,86 In Purkinje cells, complete inhibition is observed at concentrations below 200 nM, with half-maximal block produced at concentrations between 2 and 10 nM. Inhibition is rapid, occurring within two minutes of application, and while the inhibition is poorly reversible by wash-out, it can be removed by a series of strong depolarizations (i.e., to +70 mV). Block of P-type currents by ?-Aga IVA in other neurons is qualitatively similar to that in Purkinje cells although the kinetics vary slightly. For example, while inhibition of P-type current in spinal cord interneurons and neurons in the visual cortex occurs as rapidly as that in the Purkinje cells, the rate of block is several times slower in CA1 and CA3 hippocampal neurons.⁸⁶

A peptide toxin isolated from the cone snail Conus magus has also been shown to inhibit P-type channels.⁸⁸ This toxin, ω-CgTx MVIIC, blocks P-type channels with an IC₅₀ of 1-10 μM. However, ω-CgTx MVIIC also inhibits N-type channels as well as the Q- and O-type conductances⁸⁸ (see chapter by Adams and Lewis for more detail).

P-type channels do not account for all of the DHP- and ω-CgTx-resistant current in neurons since a substantial fraction of current remains even after exposure to saturating concentrations of DHPs, ω-CgTx, and ω-Aga IVA. Cultured rat cerebellar granule cells express an HVA current that is unaffected by these inhibitors at concentrations which block L-, N- and P-type channels, respectively.^45,89,90 However, the channels supporting this novel current (termed Q-type) are partially blocked by ?-Aga IVA at concentrations 10 to 100 times that required for P-type inhibition and are completely blocked by ω-CgTx MVIIC (IC₅₀= 30 to 300 nM). In addition to differing sensitivities to these toxins, Q-type channels partially recover from ω-Aga IVA-induced inhibition within minutes of toxin washout. Q-type channels also display electrophysiological properties distinct from those of P-type channels. While P-type currents in Purkinje cells and cerebellar granule cells show no inactivation over a 100 ms test depolarization, the Q-type current in granule cells decays to approximately 65% of the peak current over the same time period.

The existence of O-type channels has been inferred solely from pharmacological studies. O-type channels were identified as high affinity ω-CgTx MVIIC binding sites.⁹¹ These channels are significantly more sensitive to the toxin than are other ω-CgTx MVIIC-sensitive channels⁴⁵ It is possible that P-, Q-, and O-type channels are members of the same channel family which possess slightly different pharmacological and electrophysiological properties as a result of alternative splicing of the α₁ subunit gene and/or different complements of auxiliary subunits. While O-type channels have not been localized immunohistochemically, evidence from binding studies suggests that they are widely distributed in the mammalian CNS.⁴⁵ Estimates of O-type bindings sites in rat brain preparations suggest that O-type channels are more prevalent than N-type channels in the CNS, and it has been proposed that O-type channels are localized exclusively to synaptic termini, which would largely prevent their detection through electrophysiological means.

Many studies implicate P-, Q, and O-type channels in neurotransmitter release.⁴⁵ While N-type channels mediate release at some synapses in the mammalian CNS, the ω-Aga IVA-sensitive P- and Q-type channels appear to play a more prominent role.^52,92 ω-Aga IVA potently blocks Ca²⁺ uptake into synaptosomes84 and partially inhibits the release of dopamine and glutamate from synaptosomes^93,94 and at CA1-CA3 synapses in the hippocampus.^58,74,76 In the peripheral nervous system, ω-Aga IVA has little or no effect on the autonomic nervous system,⁹⁵ but P-type channels are probably responsible for neurotransmitter release at the mammalian NMJ.^96,97 As ω-Aga IVA blocks both P- and Q-type channels, it is possible that both channel types are involved in neurotransmission. Wheeler and colleagues found that the pharmacological properties of the ω-Aga IVA-sensitive channels supporting neurotransmission in the hippocampus and for ω-Aga IVA-induced block of inhibitory postsynaptic potentials in the cerebellum were more similar to Q- than P-type channels.^58,98 Finally, O-type channels may also mediate neurotransmission at certain synapses, as norepinephrine release in the hippocampus is inhibited by subnanomolar concentrations of ω-CgTx MVIIC.⁴⁵

R-Type Channels

A component of the HVA current in cerebellar granule cells remains even after the application of nimodipine, ω-CgTx , ω-Aga VIA, and ω-CgTx MVIIC. This current, categorized as R (residual or resistant)- type,⁸⁹ comprises approximately 15% of the HVA current in these cells. R-type current may not necessarily reflect a single channel type, but a family of molecularly distinct channels with similar pharmacological and electrophysiological characteristics.

R-type currents begin to activate around -40 mV and reach a peak amplitude at 0 mV. The current inactivates rapidly, and the increased rate of inactivation with Ca as the charge carrier suggests that the channels supporting the R-type current inactivate in a Ca²⁺-dependent manner. R-type channels are equally sensitive to block by Cd²⁺ and Ni²⁺ ions. The exact nature of the channels supporting this current is currently unknown. See the below section on Class E channels for further discussion.

HVA Ca²⁺ Channels Are Multi-Subunit Complexes

Biochemical studies have established that high threshold voltage-gated Ca²⁺ channels are multi-subunit complexes. Taking advantage of the high-affinity binding of organic antagonists, several groups purified the L-type channel from skeletal muscle. Four distinct polypeptides, designated α₁ (175-kDa), α₂δ (170-kDa), β (52-kDa), and γ (32-kDa), co-migrate with the ligand-binding activity.^99-102 A minor 212-kDa band also co-purified and was shown to represent a larger, much less abundant form of the skeletal muscle α₁ subunit.¹⁰³ Similar approaches have been used to isolate the cardiac L-type¹⁰⁴ and brain N-type channels¹⁰⁵ These complexes also consist of an a1 subunit associated with β and α₂δ subunits. The β and α₂-δ are highly similar, if not identical to, the subunits associated with the skeletal muscle α.^106,107 However, unlike the skeletal muscle L-type channel, no γ subunits appeared as part of either complex. A novel 95-kDa polypeptide was found to comigrate with the N-type channel although it is unclear whether this represents a bona fide channel subunit or a proteolytic fragment.¹⁰⁵

While subunit composition differs slightly depending on channel type, a general model has been proposed for HVA channels in which four to five proteins form a multisubunit complex (fig.1A). In this model, the α₁ subunit forms the channel proper, comprising both the voltage-sensing mechanism and the Ca²⁺ selective pore, and the remaining proteins interact with the α₁ subunit to modulate activity.

Figure 1

Composition of a VGCC complex and structure of the a1 subunit. A) Diagram of a high-voltage-activated VGCC complex, indicating the α₁, α₂/δ, β, and γ subunits. The α₁ subunit forms the channel proper, comprising (more...)

Primary Structure and Properties of Ca²⁺ Channel a1 Subunits

The first cDNAs encoding Ca²⁺ channel α₁ subunits were isolated from rabbit skeletal muscle.^5,108 The α_1s L-type subunit is an 1873-residue protein that bears a high degree of amino acid similarity to the voltage-gated Na⁺ and potassium (K⁺) channels (see fig.1). The α₁ subunit is predicted to consist of four homologous, mainly hydrophobic domains (designated domains I, II, III and IV). Each of the four domains is comprised of six putative membrane-spanning segments (S1-S6). The S4 segment in each domain contains positively-charged residues every third or fourth position and is believed to form part of the voltage-sensing mechanism of the channel. Between the S5 and S6 segments of each domain are two hydrophobic segments, SS1 and SS2, which are predicted to form the channel pore (fig.1B).

Based upon similarity to voltage-gated Na⁺ channels, Tanabe et al (1987)⁵ speculated that the α₁ subunit may form both the Ca²⁺-selective pore and the voltage sensor of the channel complex. This hypothesis was supported by studies demonstrating that expression of the α_1S in myotubes from dysgenic mice restored normal skeletal muscle-type E-C coupling and the slow Ca²⁺ current absent in these cells.⁷ In addition, α_1S expression in dysgenic myotubes restored the charge movement observed in normal myotubes upon membrane depolarization.¹⁰⁹ These results indicated that the skeletal muscle a1S subunit acts both as a voltage-sensor, providing a physical connection between membrane depolarization and Ca²⁺-release from intracellular stores for the initiation of muscle contraction, and is also part of a functional VGCC.

Using the skeletal muscle clone as a probe, cDNAs encoding homologous L-type α₁ subunits have been subsequently cloned from cardiac¹¹⁰ and smooth muscle.^111,112 Injection of the cardiac α₁ subunit into dysgenic myotubes resulted in the expression of a VGCC which differed markedly in terms of activation rate, Ba²⁺ permeability, and E-C coupling from the current conducted through channels formed by the skeletal muscle clone.¹¹³ Expression of cardiac and smooth muscle α₁ subunit clones in Xenopus oocytes^{110,111,114,115} resulted in large inward currents that were sensitive to the organic channel agonists and antagonists, thereby identifying them as L-type channels. Co-expression of skeletal muscle-derived α₂-δ and β subunits, while affecting the amplitude and voltage dependence of the currents, was not required for channel activity or drug binding, suggesting that the α₁ subunit is capable of forming a functional channel in the absence of the other subunits. However, because some VGCC subunits may be endogenously expressed by Xenopus oocytes,^116,117 it is possible that the a1 protein forms a complex with these endogenous auxiliary subunits. This prompted several groups to examine the properties of the a1 subunit in cells lacking these proteins. Murine L-cells^118,119 and Chinese Hamster Ovary (CHO) cells¹¹⁴ stably transformed with α₁ subunits express voltage activated Ca²⁺ currents sensitive to L-type channel blockers. While Ca²⁺ currents in cells expressing the smooth muscle a1 subunit displayed similar drug sensitivities and kinetics to the native currents, the currents supported by α_1S activated considerably more slowly than currents recorded from skeletal muscle cells.

At least nine different α₁ subunit genes are now known to be expressed in the mammalian nervous system (see fig.2, Table 1). Initially, four distinct classes of α₁ subunits were isolated from a rat brain library on the basis of their homology to the rabbit skeletal muscle α_1S subunit.¹²⁰ Each cDNA hybridized to one of four distinct banding patterns on Northern blots of rat brain mRNA, allowing them to be grouped into four classes, designated α_1A, α_1B, α_1C and α_1D. Subsequently, a fifth α₁ subunit (α_1E) was isolated from rat brain.¹²¹ Southern blot analysis and DNA sequencing indicated that the five classes are separate members of a multigene family with the α_1A, α_1B, and α_1E channels being more similar to one another than they are to the class C and D channels (fig.2). The class C clone is almost identical to the cardiac a1 subunit, suggesting that the class C and D clones represent members of the DHP-sensitive L-type channels, while the A, B, and E clones are DHP resistant. Four other VGCC α₁ subunit genes have been identified in the mammalian genome. The α_1F^122,123 shares the most sequence identity with the L-type channels. The α_1G, α_1H, and α_1I clones represent the LVA branch of the VGCC family.^124-126

Figure 2

Similarity tree of mammalian VGCC α₁ subunits. The predicted amino acid sequences of representatives of each class of VGCC α₁ subunit were compared pairwise and the percent similarities plotted. unc-2, egl-19 and cca-1 represent the Caenorhabditis (more...)

Table 1

Voltage-Gated Calcium Channel Subunits.

The individual α₁ subunit clones share the most homology in the transmembrane domains with the majority of sequence divergence occurring in the putative cytoplasmic regions of the channels. The loop between domains II and III, and the cytoplasmic tail vary in size as well as sequence. The DHP-sensitive channels (classes C, D, F, and S) have relatively short (≈130 amino acid) sequences linking domains II and III while the analogous region in the class A and B channels are significantly larger (≈ 430 amino acid). However, despite the size similarity between the linkers, the class A and B clones show little sequence homology in this region.¹²⁷ While the LVA channel clones (classes G, H, and I) share much less sequence identity with the other classes, the voltage-sensing S4 region and the loop that forms the channel pore are well conserved. Other motifs, such as the β-subunit binding site and the E-F hand, which are found in the HVA classes of VGCCs are absent in the LVA channels. Whole cell and single channel electrophysiological techniques have provided information about the functional and pharmacological properties of the cloned channels and allowed researchers to assign the individual clones to channel types (Table 1).

α_1A/ Ca_v 2.1

Class A α₁ subunits have been cloned from both rabbit (BI-1, BI-2)¹²⁸ and rat (rbA-I)¹²⁹ brain and Drosophila melanogaster (Dmca1A).¹³⁰ Northern blot analysis identified a single RNA transcript of 9.4 kb in rabbit brain, while two transcripts of 8.3 and 8.8 kb were detected in rat brain. α_1A transcripts are widely distributed throughout the nervous system, as well as being present in the heart and pituitary, but not in skeletal muscle, stomach, or kidney. In brain, the highest levels of class A transcripts were found in the cerebellum, suggesting that this clone might encode a P-type channel and initial expression studies supported this hypothesis. Studies showed that α_1A clones expressed in Xenopus oocytes supported HVA currents which were insensitive to DHPs and ω-CgTx but inhibited by ω-Aga VIA.^128,131,132 However, a number of discrepancies between currents elicited in oocytes expressing class A clones and native P-type currents have called this into question.^131,132 The α_1A currents display prominent time- and voltage-dependent inactivation, yet P-type currents show little time-dependent inactivation and are relatively insensitive to holding potential. Furthermore, the pharmacological sensitivities of α_1A channels are quite different from those of P-type channels. While these currents are blocked by ω-Aga IVA, they are approximately 200-fold less sensitive to the toxin (IC₅₀≈ 200 nM) than are P-type currents (IC₅₀≈ 2-10 nM). In addition, α_1A channels are markedly more sensitive to block by the snail toxin ω-CgTx MVIIC than P-type channels (IC₅₀≈ 150 nM vs. 1-10 μM for P-type channels). Sather et al (1993)¹³¹ noted that the kinetic and electrophysiological features of the α_1A current were more similar to the Q-type current described in cerebellar granule cells by Randall et al (1993).¹³³ Based on these results, some researchers have suggested that the class A clones represent Q-type channels, and that P-type channels are the product of a different gene. However, Stea et al (1994)¹³² noted that the high correlation between the localization of α_1A transcripts and P-type channel immunoreactivity implied a possible structural similarity between P- and Q-type channels and the α_1A gene product. They further proposed that the functional differences between the two channel types may arise as a result of differential post-translational processing of the proteins (which could affect toxin binding), subunit composition of the channel complex, and/or alternative splicing of the α₁ gene.

The auxiliary subunits of the VGCC complex are known to modulate the properties of the α₁ subunit (see below). The inactivation kinetics of the α_1A subunit are dramatically affected by the type of β subunit with which it is associated.¹³² Expression of the α_1A subunit from rat brain in the absence of the β subunit results in a current that inactivates considerably (40% remained after a 400 ms test pulse). Co-expression of either the β_1b or β₃ subunit increased α_1A current inactivation to a rate similar to that of the Q-type current. In contrast, currents recorded from oocytes expressing the α_1A + β_2a combination show significantly slower inactivation kinetics, such that the waveform is more similar to that of native P-type currents. The β subunit also appears to affect voltage-dependent inactivation of the α_1A subunit. The β_2a subunit shifted the steady-state inactivation of the α_1A approximately 15 to 20 mV more depolarized, thus reducing the sensitivity of the channel to holding potential.

Multiple isoforms derived from the alternative splicing of α_1A transcripts have been detected by several groups.^128,134-136 Bourinet et al (1999)¹³⁵ isolated an α_1A variant which possessed an number of sequence differences when compared to the rbA-1 clone and examined the functional implications of these splicing events. A valine insertion in the I-II linker both slowed time-dependent inactivation and altered steady-state inactivation. α_1A variants containing this valine have inactivation properties similar to P-type channels, while valine-less isoforms, such as rbA-1, appeared more Q-like. A second splice site consisted of the insertion of an asparagine-proline (N-P) pair in the IVS3-IVS4 linker. This affects the electrophysiological properties of the α_1A channel by producing a depolarizing shift in the current-voltage relationship. The N-P insertion also had the effect of decreasing the affinity of the channel for ω-Aga IVA by decreasing the on-rate of the toxin and increasing the off-rate. Thus, it is likely that the α_1A gene encodes both P- and Q-type channels, and the distinct channel properties reflect differences both in subunit composition and alternative splicing. P-type channels may be comprised of splice variants that contain the valine insertion in the I-II linker, but not the N-P pair in the IVS3-IVS4 loop. Conversely, Q-type currents may be produced by channels lacking the valine, but containing the N-P insertion. In addition, the association of different β subunits may also be an important determinant of the P- versus Q-type phenotype.

α_1B/ Ca_v 2.2

Class B α₁ subunits have been cloned from rat (rbB-I)¹³⁷ or (α_1B-1),¹³⁸ human (α_1B-1, α_1B-2),¹³⁹ and rabbit brain (BIII),¹⁴⁰ as well as from the forebrain of the marine ray Discopyge ommata (doe-4).¹⁴¹ The various clones encode proteins of 2336 to 2339 amino acids with predicted molecular weights of ≈260 to 262 kDa.¹⁴² The α_1B amino acid sequence is more similar to that of the α_1A, with the majority of the sequence divergence occurring in the cytoplasmic loop between domains II and III and in the cytoplasmic carboxyl tail.

Initial indications that this class of α₁ subunit corresponds to N-type channels came from the work of Dubel et al (1992)¹³⁷ who showed that a polyclonal antibody (CNB-1) raised against the II-III loop region of the rbB-I clone immunoprecipitated almost 50% of the high-affinity ω-CgTx binding sites, but none of the DHP-binding sites from rat brain. Furthermore, Northern blot analysis of experimental cell lines showed that rbB-I expression was correlated with the presence of N-type channels in nerve tissues and cell lines that express N-type channels.^{137,139,140,143} Northern blotting and in situ immunohistochemistry experiments have localized the rbB-I transcripts to the cerebral cortex, hippocampus, forebrain, midbrain, cerebellum, and brainstem. At the subcellular level, rbB-I protein is found on dendrites, at presynaptic terminals and, to a lesser extent, neuronal cell bodies. The localization pattern of the α_1B compares well with that observed with a monoclonal antibody against ω-CgTx,⁵⁶ although the ω-CgTx antibody staining was more widely distributed.

Molecular cloning and biochemical studies have also provided evidence for the existence of multiple isoforms of the α_1B subunit.^{138-140,143-146} At least two of these isoforms represent channels with differentially spliced carboxyl tails, and the inability of CNB-1 to immunoprecipitate all of the ω-CgTx binding sites might suggest the existence of additional isoforms with distinct II-III loop sequences. In addition, α_1B clones with small insertions and deletions scattered throughout the channel have been identified, and expression studies indicate that these sequence variations have a profound influence on the properties of the channel¹³² (see below). These include a variant of the human N-type calcium channel that lacks the synaptic protein interaction site in the domain II-III linker¹⁴⁵ and can therefore not associate with synaptic proteins such as syntaxin 1A and SNAP-25.

Transient expression of both human α_1B-1 in HEK cells¹⁴⁷ and rabbit brain BIII in dysgenic myotubes¹⁴⁰ produced HVA currents that first activated between -10 and -30 mV and reached a maximum between +10 and +30 mV. Currents partially inactivated over the time course of the depolarization and were sensitive to holding potential (showing 50% current inactivation at approximately -60 mV). At a holding potential of -40 mV, the bulk of the current (90%) was inhibited. In agreement with binding studies, the α_1B-induced currents were irreversibly blocked by 1 μM ω-CgTx and were insensitive to DHPs.

The properties of currents generated in Xenopus oocytes by expression of the rbB-I clone agreed well with those seen with the α_1B-1 and BIII clones in terms of pharmacological sensitivities and voltage-dependence of activation. However, there were some notable discrepancies in other properties. For example, the rbB-I channel was less sensitive to holding potential, and the rates of activation and inactivation of the rbB-I clone were markedly slower, resulting in significantly different current waveforms.¹⁴⁸ Co-expression of the β_1b subunit shifted the voltage-dependence of inactivation to more negative potentials similar to those observed with the human and rabbit clones. The β subunit also increased the rate of activation such that the current attained peak magnitude in approximately 120 ms (compared to 150-250 ms for the rbB-I subunit alone), and increased the rate of inactivation of the rbB-I current. After 800 ms, current through rbB-I alone had decreased by 15-20%, whereas co-expression of the β subunit resulted in a 65-70% reduction in peak current. Despite the rate increases produced by β subunit co-expression, these parameters remained dramatically different from those displayed by the other clones. The rate of activation of rbB-I (110 ms to peak) was still significantly slower than that of a1B-1 currents (10 ms). In addition, the α_1B-1 clone showed biphasic inactivation; the first, rapid phase had a τ of 46-105 ms. The τ of the slow phase ranged between 291 and 453 ms. In contrast, decay of rbB-I currents was monophasic and much slower (τ= 700 ms).

Evidence obtained by Stea et al (1999)¹³⁸ from a second rat brain clone (rbB-II or α_1B-II) found that the differences in channel kinetics were the result of small amino acid alterations that are most likely the product of alternative splicing and/or cDNA cloning artifacts. While α_1B-II differs from α_1B-I in four regions, they found that the substitution of a glycine for a glutamate in transmembrane segment IS3 was sufficient to speed the activation and inactivation kinetics. It has been noted that, while N-type channels are typically described as having fast kinetics, this is not always the case (see N-type channels, above). It may be that isoforms containing a glutamate in IS3, such as the rbB-I (α_1B-I) clone may account for the slow and incomplete inactivation of N-type current that has been described in sympathetic neurons. As is the case with the α_1A gene, alternative splicing and differential subunit composition may combine to produce slight modifications in channel characteristics with tissue or developmental specificities. Just recently, a newly discovered tissue specific splice isoform variant of the N-type channel has been discovered in dorsal root ganglia (DRG).¹⁴⁹ This splice variant arising from the presence of exon 37a has the unique property of being expressed exclusively in nociceptive neurons of the DRG and ultimately may serve as a novel target for pain management.

α_1C/ Ca_v 1.2

The first complete class C α₁ (α_1C) subunit to be cloned was isolated from cardiac muscle (CARD1).¹¹⁰ The cardiac and skeletal muscle L-type VGCCs arise from separate genes and are approximately 66% identical at the amino acid level. Subsequently, a1C clones were isolated from rabbit lung (pSCaL)¹¹¹ and rat aorta (990: VSMα₁)¹¹² which shared 95% identity with the cardiac clone. Class C clones later isolated from rat (rbC-I, rbC-II)¹⁵⁰ and mouse (mbC)¹⁵¹ brain are also more closely related to the cardiac and smooth muscle α₁ subunits than to the skeletal muscle clone. The α_1C clones code for proteins of 2140 to 2171 amino acids with predicted molecular masses of 235 to 239 kDa. Antibodies directed against the II-III loop of the neuronal class C channel also identify a truncated form with an approximate mass of 195 kDa.⁶

The high degree of similarity amongst these proteins suggest that they are products of a single gene, and this is supported by genomic Southern blotting experiments. However, there are regions of considerable diversity in these clones which are the result of alternative splicing of the primary transcript.^150,152,153 pSCaL, the smooth muscle channel isolated by Biel et al (1990)¹¹¹ differs from the cardiac form in the amino terminus, the IS6 and IVS3 transmembrane segments, and by a 25-amino acid insertion in the I-II linker. In contrast, the VSMα₁ clone, also isolated from smooth muscle, contains a cardiac channel-like IS6 segment and a 68 residue substitution in the carboxyl tail. This sequence is also found in the neuronal clones, rbC-I and —II.¹⁵⁰ Finally, rbC-I and rbC-II contain regions of identity with both the smooth muscle and cardiac clones, but also contain many substitutions, primarily in cytoplasmic regions of the protein and the IIIS2 transmembrane segment. Notably, many of these substitutions are localized to the cytoplasmic linker between domains II and III, and may reflect cell-specific functions of the channels. The truncated form of the neuronal protein may also be the result of alternative splicing, or may be due to post-translational processing as is the case with the skeletal muscle channel.⁶

As would be expected considering the diverse nature of the tissues from which class C cDNAs have been cloned, the α_1C gene has a widespread pattern of expression. Transcripts of 8.9 kb have been detected in heart.¹¹⁰ In smooth muscle and brain, hybridizing transcripts were slightly smaller (8.6 and 8 kb, respectively).^112,150 Additional transcripts of 15.5 kb (cardiac) and 12 kb (smooth muscle and brain) were also detected which were proposed to represent stable processing intermediates.¹¹² Northern blot analysis indicate that the class C gene is expressed in heart, smooth muscle (e.g., uterine, lung, stomach, and intestine), and throughout the CNS.^112,150,151 Within the brain, high expression levels are detected in the olfactory bulb, cerebellum, striatum, thalamus, hypothalamus and cortex, and at much lower levels in the pons/medulla and spinal cord.^6,150 Thus far, there is no evidence for exclusive expression of α_1C splice variants in specific tissues. cDNAs containing both variants of the IVS3 transmembrane segment have been isolated from heart, smooth muscle, and brain^112,150,152 and the alternate carboxyl tail is expressed in both smooth muscle and neuronal tissues.^112,150 However, a more detailed study of the expression pattern of class C variants in rat has revealed tissue-specific differences in expression of the rbC-I and rbC-II proteins.¹⁵⁰ Overall, rbC-II is the more abundant form, and generally more prevalent in any given tissue, although the relative amounts of the two transcripts vary between brain regions and tissue types.

The subcellular localization of class C α₁ subunits was studied using the polyclonal antibody, CNC1.⁶ Immunoprecipitation and Western blotting experiments indicated that class C a1 subunits comprise approximately 75% of the DHP binding sites in rat cerebral cortex and hippocampus. CNC1 immunoreactivity was distributed at low levels on cell bodies and proximal dendrites, with staining diminishing along the length of the dendrite. In addition, clusters of high levels of immunoreactivity were observed on the surface of cells (as opposed to representing a cytoplasmic pool of channels).

Expression of α_1C clones in Xenopus oocytes resulted in currents with electrophysiological and pharmacological properties characteristic of L-type channels.^{110,111,115,154} In Ba²⁺, depolarizations to -10 to -30 mV elicited large inward currents that inactivated slowly, if at all, over the course of a several hundred millisecond the test pulse. The currents peaked between +10 and +30 mV and were inhibited by Cd²⁺ (100-200 μM) and were sensitive to DHPs. Like native L-type currents, the cloned L-type channels showed little sensitivity to holding potential. At holding potentials as high as -20 mV, half of the channels remained available for opening.

The class C channels were shown to be modulated by the auxiliary subunits in much the same manner as the class A and B channels. Co-expression of α_1C with β_1b and α₂-δ significantly increased the magnitude of the whole cell currents. This increase appeared to be mediated primarily through interaction with the β subunit, while the addition of α₂-δ had a slight synergistic effect. In addition, co-expression of rbC-II with the auxiliary subunits β_1b and/or α₂-δ caused a small hyperpolarizing shift in the voltage dependence of activation of the channel and altered channel kinetics.¹⁵⁴ The rate of activation for rbC-II varied substantially among oocytes. Time constants of activation ranged between 4 and 50 ms with an average of approximately 10 ms. Coexpression with β_1b and α₂-δ both increased the rate of activation and reduced the degree of variability. Furthermore, the β_1b and α₂-δ subunits increased the rate of inactivation. Unlike the α_1B rbB-I channel (see above), neither auxiliary subunit had a significant effect on the voltage dependence of inactivation. With the exception of the voltage dependence of channel activation, the modulatory effects on the kinetics and voltage dependent parameters of the channel appear to be mediated primarily through interaction with the β subunit, while the addition of α₂-δ had slight synergistic effects.

Whole cell recording using Ca²⁺ as the charge carrier resulted in current traces with markedly different waveforms. The magnitude of the whole cell current was significantly smaller in Ca²⁺ than Ba²⁺, indicating that the channels are more permeable to Ba²⁺ than to Ca²⁺. In addition, instead of eliciting currents that are essentially non-inactivating, depolarizing pulses produce currents that decay rapidly by more than 50%.^154,155 The increase in inactivation seen with Ca²⁺ as the permeant ion retains all the hallmarks of Ca²⁺-dependent inactivation (see above).

Calcium dependent inactivation (CDI) is predominantly linked to two regions of the Alpha1 subunit C-terminus. The EF hand,¹⁵⁶ located from aa 1526 to1554, is responsible for CDI and may also regulate voltage-dependent inactivation (VDI).¹⁵⁷ The second region (found downstream of the EF hand) is comprised of three distinct binding motifs. Peptide A; (aa 1588 to 1610) and Peptide C (1615 to 1636 aa),¹⁵⁸ and IQ(1649 to 1669),¹⁵⁹ all work together to form a calcium bound Calmodulin binding site. In the absence of calcium, Ca²⁺ free calmodulin (ie Apo Cam) is pre-associated with the channel at a site localized between the EF motif and IQ region.^23,160 Calcium entering through the channel binds to calmodulin, thus inducing a conformational change that relieves an inhibitory action of the calmodulin/C-terminus complex on the voltage-dependent inactivation machinery.¹⁶¹

α_1D/Ca_v1.3

A class of VGCC cDNAs sharing about 70% amino acid identity with the cardiac clones has been cloned from a variety of species including rat (RBα₁;¹⁶² rCACN4A, rCACN4B¹⁶³), human (α_1D;⁵¹ neuroendocrine or β-cell α₁ or hCACN4:¹⁶⁴ HCa3a¹⁶⁵), and chicken.¹⁶⁶ A cDNA encoding an invertebrate ortholog of the class D α₁ subunit has been isolated from Drosophila melanogaster (Dmca 1 D).¹⁶⁷ These clones retained little similarity (~40% amino acid identity) with the non-DHP-sensitive class A clones,^51,164 but were almost identical to the partial rat brain clone designated class D.¹²⁰ In spite of the sequence divergence between cDNAs generated from class C and class D genes, the two channel types are remarkably similar in certain regions. As with all VGCC α₁ subunits discussed thus far, the transmembrane regions tend to be highly conserved, while the intracellular loop sequences are much more divergent. In addition to these regions, the α_1D clones are almost identical to the DHP-sensitive class C and S clones in the segments predicted to form the DHP and phenylalkylamine binding sites, suggesting that the class D α₁ subunit cDNAs also encode members of the DHP-sensitive L-type family of VGCCs. The exception to this lies in the DHP-binding region in the Drosophila Dmca 1D clone which contains a number of non-conserved changes. This finding, however, is consistent with the pharmacology of phenylalkylamine and DHP binding in Drosophila head membranes,¹⁶⁷ and provides further support for the role of these regions in drug binding.

The cloned α_1D subunits range in size from the 1634-amino acid (187-kDa) rat brain isoform to the 2516-amino acid (276-kDa) Drosophila channel clone. The range in protein sizes is due primarily to the truncated carboxyl terminal ends of the RBα₁, HCa3a, and rCACN4B clones.^162,163,165 The rCACN4B clone is a full 535 residues smaller than its rCACN4A counterpart and is proposed to result from the use of an alternative splice acceptor site.¹⁶³ In addition to the truncation of the carboxyl tail, a number of other regions have been identified in which variants have been produced through alternative splicing.^{51,152,162,163} These regions include insertions in the cytoplasmic linker between domains I and II, the extracellular linker connecting IVS3 and IVS4, and the transmembrane segments IS6 and IVS3. In addition, Kollmar et al (1997)¹⁶⁶ have reported that the chicken brain and cochlear α_1D proteins differ in the IIIS2 segment and IVS2-IVS3 loops, as well as in the carboxyl tail. Presumably, these splice variants impart functional differences to the channel. While it is not yet clear what these functional differences may be, Ihara et al (1995)¹⁶³ note that RBa1, HCa3a and rCACN4B are all truncated at different sites. Furthermore, a number of potential PKA sites are eliminated by the truncations, which may result in the differential regulation of these isoforms by phosphorylation.

The class D channels, often termed “neuroendocrine” because of their presence in brain and pancreatic cells, have also been detected in the retina, ovaries, and cochlear hair cells, but not in heart, skeletal muscle, spleen, colon, or liver. Reports differ on whether α_1D transcripts are present in kidney.^164,166 Within the CNS, class D expression is found in the hippocampus, habenula, basal ganglia, and thalamus.⁵¹ The subcellular localization of the class D α₁ subunit was characterized using the polyclonal antisera anti-CND1.⁶ Anti-CND1 was generated against a peptide homologous to the unique II-III loop of the rat brain α_1D clone (Hui et al, 1991).¹⁶² Class D channels appear to be far less abundant in the rat CNS than class C channels. The sera labeled the cell bodies and proximal dendrites of both projection neurons and interneurons throughout the brain. In contrast to the punctate staining pattern seen with the class C antibody, anti-CND1 staining was evenly distributed over the cell body. The staining pattern of anti-CND1 was typical for neurons in all regions of the CNS with the notable exception of the cerebellar Purkinje cells. While the cell bodies of these neurons were labeled, there was a marked absence of staining on the Purkinje cell dendrites.

Transient expression of human α_1D⁵¹ in Xenopus oocytes and stable expression of the rat CACN4A and CACN4B clones (Ihara et al, 1995)¹⁶³ in CHO cells gives rise to DHP-sensitive currents, confirming the notion that class D channels are members of the L-type family. In both systems, functional expression of the α_1D subunit required co-expression of the β subunit. In Xenopus oocytes, transient expression of α_1D with the β and α₂ subunits yielded larger currents than those produced by expression of α_1D plus β alone.51 Ba²⁺ currents in oocytes expressing α_1D+ β + α₂ first activated upon depolarizations positive to -30 mV and peak current attained with depolarizations to 0 mV, thus the current-voltage relationship of the α_1D is somewhat more negative than that of the α_1C¹⁵⁴ (see above). α_1D channels activated rapidly and inactivated little over depolarizations lasting as long as 700 ms. α_1D channels inactivate to a considerably lesser degree over long test pulses than do α_1C channels.¹⁵⁴

As indicated, class D channels fall under the heading of DHP-sensitive L-type channels. Cd²⁺ produces substantial block, while Ni²⁺ has a minimal effect on the current. The DHP agonist Bay K8644 increases current magnitude and shifts the voltage-dependence of activation by approximately -10 mV. In addition, the current is inhibited by the DHP antagonist nifedipine.^51,163 However, the affinity of DHPs for α_1D channels is generally lower than that observed with α_1C.^168,169 Moreover, unlike other DHP-sensitive channels, the cloned α_1D is partially and reversibly blocked by high concentrations of ω-CgTx (10-15 μM).⁵¹

The predominance of the a1D subunit-containing VGCCs in the cochlear hair cells and in the β-cells of the pancreas suggest that these channels may be involved in tonic exocytotic release in these cells^{163,164,166,170,171} Kollmar et al (1997)¹⁶⁶ suggest that the electrophysiological properties of the α_1D subunit, such as its lack of inactivation during depolarizations may render it suitable for mediating tonic release. In addition, as suggested by the localization of α_1D channels on the cell body and at the base of dendrites of neurons in the CNS, these channels may be involved in integrating signals impinging upon the neuron from multiple sources.⁶

α_1E/ Ca_v 2.3

The class E gene encodes a VGCC α₁ subunit (a1E) that does not fall neatly into either the HVA or LVA categories. α_1E cDNAs have been isolated from rabbit (BII-1, BII-2)¹⁷² and rat (rbE-II)¹²¹ brain, and from Dyscopyge ommata.¹⁴¹ These clones code for proteins between 2178 and 2259 amino acids with predicted molecular masses of approximately 250 kDa. Splice variants of the rabbit brain channel, BII-1 and BII-2, differ from one another in their carboxyl tails, resulting in the addition of a putative PKA site.

The class E clones appear to be more closely related to the DHP-insensitive non-L-type channels (54-60% amino acid identity) than to the L-type channels (less than 45% similarity). However, class E channels are less similar to either class A or B channels than these two classes are to one another, suggesting that the class E channels are members of a novel, more distantly related subgroup of DHP-insensitive channel (fig.2).^121,141,172

Northern blotting studies have identified transcripts ranging in size from 10.5 to 12 kb in the mammalian CNS.^121,172 High levels of expression was identified in the cerebral cortex, hippocampus, and striatum, while lower levels were detected in the olfactory bulb, midbrain, and Purkinje and granule cell layers of the cerebellum. While α_1E appears abundant in brain, none was detected in skeletal muscle, heart, stomach or kidney. At the subcellular level, α_1E protein was localized nearly exclusively to the cell body of neurons throughout the CNS. Dendritic staining varies across brain regions. For example, in the cortex and hippocampal formation there is barely perceptible staining of the dendritic branches, while in Purkinje cells, α_1E antibodies labeled the distal dendritic branches, but not the main dendritic trunks.¹⁷³

The α_1E channel was initially reported to be a novel member of the LVA family of Ca²⁺ channels.¹²¹ Expression of rbE-II in Xenopus oocytes produced a channel that activated rapidly at low membrane potentials (threshold≈ -50 mV) and inactivated significantly during the depolarization. Other voltage-dependent parameters of this channel (current-voltage relationship, voltage-dependence of inactivation) were also considerably more negative than those of other cloned HVA channels. The rbE-II current magnitude increased steeply with increasing depolarizations, peaking at around 10 mV, and steady state inactivation analysis indicated that the channels were inactivated near the resting membrane potential of the cell. In addition, rbE-II channels were equally permeable to Ca²⁺ and Ba²⁺, a property reported to be unique to the LVA channels. Another similarity with LVA channels was the high sensitivity of the current to block by Ni. Furthermore, the channel was found to be expressed in many of the cells that have been shown to possess T-type currents. However, Soong et al (1993)¹²¹ noted a number of discrepancies between rbE-II and native T-type currents. For example, although the voltage-dependent properties of rbE-II currents were more negative than those of the other cloned HVA Ca²⁺ channels, the activation and peak current potentials were not as hyperpolarized as for typical T-type channels.¹²¹ Analysis of the electrophysiological properties of other class E channels^174-176 have produced some results that contradict those of Soong et al (1993).¹²¹ In these studies, the α_1E clones formed HVA channels, activating at approximately -10 mV and peaking at +30 mV. The single channel conductance of α_1E channels is also much larger than that of T-type channels (12-14 pS vs. ~8 pS).^141,177 As α_1E channels share properties with LVA as well as HVA channels, the detection of pure α_1E currents in native cells may difficult.

It has been suggested that the class E channels may be one of a group of channels comprising the R-type current.^89,178,179 The two currents share some electrophysiological and pharmacological characteristics, such as strong voltage-dependence of activation and insensitivity to DHPs, ω-CgTx, and ω-Aga IVA. However, the R-type current is smaller in Ca²⁺ than in Ba²⁺, whereas the α_1E channels support the two currents equally.^89,177 Most relevant, mice lacking the α_1E gene entirely still exhibit significant R-type current.¹⁸⁰ Thus, while class E channels may comprise a component of R-type current in cerebellar granule cells, the R-type current may actually result from incomplete block of other Ca²⁺ channels by applied pharmacological agents, the expression of additional splice variants of already identified Ca²⁺ channel subtypes, or as yet to-be-identified α₁ subunits.

α_1F/ Ca_v 1.4

The human class F gene (CACNA1F) was identified through genetic studies in which the X-linked visual disorder Congenital Stationary Night Blindness (CSNB) was mapped to a locus containing a putative VGCC gene.^122,123 The predicted CACNA1F gene product (α_1F) is between 1912 and 1966 amino acids (alternatively spliced forms have been detected) with an estimated molecular mass of 219 kDa. Sequence analysis indicates that α_1F is 55-70% identical at the amino acid level to the L-type channel α₁ subunits, sharing the most similarity with α_1D, and 35% identical to the P- and N-type channels. In addition, the putative DHP-binding domains in IIIS6 and IVS6 appear relatively well conserved. These results suggest that the α_1F is an L-type channel that diverged from the α_1D subunit gene.¹²²

α_1F expression was initially reported to be restricted to the retina in situ hybridization experiments indicate high levels of α_1F transcript in the two retinal layers containing the photoreceptors, and horizontal, bipolar, and amacrine cells, but not the ganglion-cell layers,^122,123 however, recent reports indicate a more global distribution that includes the immune system and skeletal muscle.¹⁸¹ L-type Ca²⁺ channels have been implicated in synaptic release from photoreceptors¹⁸² and the correlation of the hereditary visual disorder CSNB with mutations in the α_1F gene^122,123 suggests that the α_1F channel mediates neurotransmitter release at these synapses. Functional expression of α₁ calcium channels reveals that these channel encode a non-inactivating L-type calcium channel that is DHP sensitive,¹⁸³ and which is not regulated by ancillary subunits.¹⁸¹ Moreover the channel appears to lack CDI, and displays a large window current, thus making this channel ideally suited to support tonic glutamate release from photoreceptors.¹⁸¹

α_1G, α_1H and α_1I

Low stringency library screening strategies such as the ones used to isolate the HVA channels discussed above proved unsuccessful for cloning the T-type Ca²⁺ channels. The first members of the T-type Ca²⁺ channels were identified by screening data banks for sequences with similarity to previously cloned Ca channels.^124,193,200 Subsequently, other classes of T-type Ca²⁺ channels were identified by screening cDNA libraries.^125,126,194

Thus far, T-type clones have been isolated from rat (α_1G;¹²⁴ α_1G, α_1H, α_1I;²⁰⁰ α_1I¹²⁶), mouse (α_1G¹⁹³), human brain (α_1G, α_1I),^194-196 and human heart (α_1H).¹²⁵ The α_1G and α_1H subunits are approximately 65% identical, whereas the α_1I subunit shares only 53% identity with the α_1H and 47% with the α_1G. As expected from the failure to identify T-type channels in the low stringency hybridization screens used to isolate many of the HVA channels, the T-type channels share limited sequence homology with HVA Ca²⁺ channels. The highest level of sequence similarity is found in the four membrane-spanning domains. Most of the amino acid changes in these regions are conservative, thereby maintaining structural elements common to voltage-gated ion channels. The charges located in the fourth transmembrane segment of each domain are conserved, as are the pore-forming loops between the fifth and sixth transmembrane segments. In HVA channels, a glutamate residue located in each of these four loops is believed to determine the ion selectivity of the channels.^197,198 All T-type channels cloned thus far contain aspartate residues instead of glutamates in the domain III and IV P-regions.¹²⁴ This difference may account for the difference in the permeation properties seen between high and low voltage activated channels. The intra- and extracellular linkers joining the transmembrane domains share little homology with either HVA channels or with other T-type channels. Furthermore, the T-type channels do not seem to possess specific functional motifs that have been identified in HVA channels, including the binding site in the I-II linker or the putative EF-hand motif in the carboxyl tail.¹²⁴

The three classes of T-type channel have been localized using Northern blotting, in situ hybridization, and RT-PCR techniques.^{124-126,193,195,200,201} The α_1G subunit appears to be expressed abundantly throughout the brain and to a lesser degree in heart. Low levels have also been detected in placenta, lung, and kidney. High levels of transcript are observed in the cerebellum, hippocampus, thalamus, and olfactory bulb, with lesser amounts localized to the cerebral cortex and septal nuclei. Initially, the α_1H was detected only in cardiac tissue, kidney, and liver, with very little, if any, expression in the brain.¹²⁵ However, a subsequent study¹⁹⁹ suggests that the α_1H subunit may be responsible for a large proportion of the T-type current in sensory neurons, and another study indicates the expression of α_1H in all areas in the rat brain.²⁰¹ α_1I transcripts have only been detected in brain,^126,195 with one study showing specific expression in the striatum of adult rats.²⁰⁰

Expression of these three subunits in Xenopus oocytes^124,126 and HEK-293 cells^{125,193,195,196,200-203} demonstrates that they support currents with most of the characteristics expected of T-type channels. Currents activated upon weak depolarizations from negative holding potentials. In one study, the three T-type channel classes had differing permeability properties.²⁰⁰ As has been noted for classic T-type currents, α_1G channel was more permeable to Ca than to Ba. However, α_1H channels were more permeable to Ba²⁺ than Ca²⁺, while α_1I channels were equally permeable to the two ions. In most cases, currents were inhibited by mibefradil and Ni, although the IC₅₀ of each T-type class varied significantly.^193,200 The activation and inactivation kinetics of the T-type channels are strongly voltage-dependent. While rates of activation and inactivation are slow near threshold potentials, they accelerate as the strength of depolarization increases. Deactivation is also voltage-dependent, increasing at more hyperpolarized potentials. Steady-state inactivation analysis indicates that the majority of the channel population would be inactivated at the resting potential of most cells. However, because all the channels are not inactivated at the resting potential and the threshold of activation is so negative, a small proportion of channels are capable of opening at the resting potential, thus producing a “window” current. The window current refers to a small, but sustained influx of Ca²⁺ that occurs even when the cell is ostensibly at rest. This current can contribute to the overall excitability of the membrane and may contribute to the bursting and pacemaker activities attributed to the T-type channels. Finally, as anticipated for T-type channels, the three exogenously expressed channels have small single channel conductances of 5 (α_1H), 7.5 (α_1G), and 11(α_1I) pS.¹²⁶

Similar to the different classes of channels within the HVA subfamily, the biophysical properties of the three T-type channels vary considerably.^126,200 The α_1G and α_1H possess very similar activation and inactivation potentials, while those of the α_1H appear to be slightly more negative. Rates of activation and inactivation of α_1G and α_1H currents also are quite similar. In contrast, activation and inactivation rates for α_1I currents are significantly slower. In addition, the activation threshold of α_1I channels also differs from the values obtained for α_1G and α_1H channels. However, varying results have been reported concerning the direction of the observed voltage shift. The rat brain α_1I clone studied by J.-H. Lee et al (1999)¹²⁶ activated at more positive potentials than did the α_1G and α_1H channels, while McRory et al (1999)²⁰⁰ reported α_1I current activation at considerably more negative potentials. That the properties of the α_1I channel differ from those of the α_1G and α_1H is not entirely unexpected when one takes into account the degrees of similarity seen amongst the three channels. Furthermore, multiple splice variants of the α_1I have been identified^194,195 and may account for the contrasting results reported for α_1I currents. Finally, the properties of the three LVA channel clones do not account for all of the T-type characteristics in native cells and there may be additional classes of T-type channels and/or a set of as yet unknown auxiliary subunits specific to LVA channels which further modulate the properties of the LVA α₁ subunit. For example, robust alternate splicing for α_1G channels has been reported and shown to result in significantly altered biophysical characteristics.²⁰³

β Subunits

The β subunit is the most extensively studied of the auxiliary subunits and appears to have the most profound effects on the functional properties of the α₁ subunit. In mammals, there are at least four different β subunits (β₁, β₂, β₃, and β₄) which are encoded by distinct genes. The transcripts of at elast two of these genes, the β₁ and β₂, are alternatively spliced to give rise to β_1a β_1b, and β_1c and β_2a and β_2b.

Biochemical and primary sequence analyses indicate that the β subunits are hydrophilic with no transmembrane segments or glycosylation sites.^204-210 The β subunits contain potential phosphorylation sites for both protein kinase C and cAMP- dependent protein kinase. The modulatory effects of these enzymes on VGCC function may, in part, be the result of their actions on this auxiliary subunit.²¹¹

Although the specific effects of channel modulation depend upon the β subunit isoform, all β subunits appear to have the same general impact on the properties of HVA α₁ subunits. Coexpression of the a1 and β subunits in both L-cells and Xenopus oocytes increased whole-cell currents and DHP binding without affecting the level of α₁ message. This suggests that rather than enhancing expression of the α₁ subunit, the β subunit may promote insertion of the α₁ subunit into the membrane and/or stabilize a specific conformation of the protein^212-215 have proposed that the β subunit potentiates coupling of the gating-charge movement caused by changes in membrane potential with the opening of the pore thereby increasing the probability of channel activation and, in turn, increasing the peak current.

In addition to increasing the magnitude of the current through the α₁ subunit, co-expression of the β subunit alters channel kinetics. For most β subunits, the rate of inactivation is increased and there is a shift in the voltage-dependence of activation to more negative potentials. ^{148,154,207,209,216-219} The effect on kinetics of inactivation, however, varies depending upon the class of β subunit expressed. The β₁ and β₃ proteins increase the rate on inactivation, while the β₂ subunit significantly slows inactivation.

These modulatory effects are observed regardless of which α₁ and β subunits are co-expressed, suggesting that the mechanism through which the β subunit acts is common to all HVA VGCCs. The region required for β subunit modulation of the α₁ subunit has been localized to a stretch of 30 amino acids at the amino-terminal side of the second of two conserved domains.²²⁰ This region, known as the BID (β subunit interaction domain) is also responsible for anchoring the β subunit to the α₁. The β subunit has been shown to bind to a conserved motif of 18 amino acids in the intracellular loop between domains I and II of the α₁ subunit (the AID: α₁ subunit interaction domain).²²¹ The observation that the β subunit from skeletal muscle dramatically increases the magnitude of the current through brain α₁ subunits when co-expressed in Xenopus oocytes¹²⁸ further supports the idea of a common mechanism of α₁- β subunit interaction. The β subunits exhibit homology with the Src homology 3-guanylate kinase domain of membrane associated guanylate kinases and this region appears to regulate inactivation of HVA VGCCs.²⁸⁹ A more detailed account of β subunit physiology is provided in the Chapter from the Charnet lab.

α₂-δ Subunits

Purification studies indicate that the VGCC α₂-δ subunit consists of two distinct subunits (α₂ and δ that are disulfide-bonded in the native state.^8,222 The α₂ and δ subunits are derived from a single gene product that is proteolytically cleaved to form 143-kDa α₂ and 27-kDa δ subunits.²²³ There are currently four genes that code for the α₂-δ family; α₂-δ₁, α₂-δ₂, α₂-δ₃, and α₂-δ₄.

Both proteins are heavily glycosylated,²²⁴ supporting the prediction that the α₂-δ subunit is predominantly extracellular and a recent study has determined that no more than five residues comprise the cytoplasmic portion of the protein.²²⁵ The complex is anchored in the membrane by a single transmembrane segment formed by a portion of the δ subunit. The transmembrane domain is thought to interact with other subunit(s) in the VGCC complex while the extracellular domain is responsible for the modulatory effects.^{8,142,223-225}

There are several splice variants of the α₂-δ₁ auxiliary subunit, due primarily to the alternative splicing of three specific regions. The total splicing combination of these three regions reveals 5 unique isoforms which are all expressed in a tissue specific manner.²²⁶ Skeletal muscle and brain express the isoforms α₂-δ_1A and α₂-δ_1B respectively. The heart expresses mainly α₂-δ_1C and α₂-δ_1D, and smooth muscle expresses α₂-δ_1D and α₂-δ_1E. Interestingly the cardiovascular system expresses all five isoforms.²²⁷

The α₂-δ₂ subunit is expressed in many different tissues including heart, brain, pancreas, testis lung, liver, kidney.²²⁷ The α₂-δ₂ subunit has two regions of alternative splicing. The first region is found on the a2 subunit between residues 661/663 and involves the addition of eight amino acid residues. The second is located on the δ protein and is characterized by the addition of three different residues.²²⁸ The resulting three isoforms of the α₂-δ₂ subunit are all expressed in hMTC (human medullary thyroid cells.) Alternately, in the human heart, only α₂-δ_2A is expressed. cDNA cloning of the α₂-δ₃ subunit did not uncover additional splice variants, and this particular α₂-δ gene appears to be expressed exclusively in the brain.

In 2002, a new member of the α₂-δ auxiliary subunit family was described. α₂-δ₄ demonstrated a unique property in that it was shown to be expressed primarily in endocrine tissues. Immunohistochemical studies revealed that the α₂-δ₄ subunit has limited distribution in special cell types of the pituitary, adrenal gland, colon, and fetal liver.²²⁹ Whether the a₂-d₄ subunit plays a physiological role in certain endocrine tissues remains to be seen. Alternative splicing of the α₂-δ₄ gene gives rise to four potential variants (called a-d.²²⁹

The functional effects of the α₂-δ are thought to be more subtle that those of the β subunit and are highly dependent on the class of α₁ subunit and the cell type used for exogenous expression. For example, Singer et al (1991)²¹⁷ found that the Ca²⁺ current in Xenopus oocytes expressing the cardiac α_1C protein was greatly enhanced by co-expression of the α₂-δ subunit from skeletal muscle. In addition, the rates of activation and inactivation were increased and the voltage dependence of inactivation was shifted to more negative potentials. In contrast, Varadi et al (1991)²¹⁸ did not observe these effects when they co-expressed the skeletal muscle α₂ subunit with the α₂-δ subunit in L cells.²³⁰ Unlike Xenopus oocytes, L cells do not express endogenous calcium channel subunits.²¹⁹ Thus, it is possible that the α₂-δ subunit acts synergistically with other auxiliary subunits to modulate the properties of the α₁. Finally, there is some evidence that the α₂-δ subunit is required for efficient expression and/or trafficking of the α₁ subunit to the cell membrane,¹¹⁰ an idea that is supported by recent studies in tsA-201 cells showing that α₂-δ subunits are key modulators of current densities of α_1C, α_1B and α_1E channels.²³¹

γ Subunits